{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 7704, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.00025960539979231567, "grad_norm": 17.76244354248047, "learning_rate": 0.0, "loss": 0.40040892362594604, "step": 1 }, { "epoch": 0.0005192107995846313, "grad_norm": 24.69705581665039, "learning_rate": 2.1551724137931035e-08, "loss": 0.4246561825275421, "step": 2 }, { "epoch": 0.000778816199376947, "grad_norm": 28.28066635131836, "learning_rate": 4.310344827586207e-08, "loss": 0.4287872016429901, "step": 3 }, { "epoch": 0.0010384215991692627, "grad_norm": 15.585029602050781, "learning_rate": 6.465517241379311e-08, "loss": 0.39609405398368835, "step": 4 }, { "epoch": 0.0012980269989615785, "grad_norm": 23.996196746826172, "learning_rate": 8.620689655172414e-08, "loss": 0.4229694902896881, "step": 5 }, { "epoch": 0.001557632398753894, "grad_norm": 24.05171775817871, "learning_rate": 1.0775862068965518e-07, "loss": 0.4101257622241974, "step": 6 }, { "epoch": 0.0018172377985462098, "grad_norm": 23.986738204956055, "learning_rate": 1.2931034482758622e-07, "loss": 0.42789047956466675, "step": 7 }, { "epoch": 0.0020768431983385254, "grad_norm": 17.384754180908203, "learning_rate": 1.5086206896551727e-07, "loss": 0.3873000144958496, "step": 8 }, { "epoch": 0.002336448598130841, "grad_norm": 18.249189376831055, "learning_rate": 1.7241379310344828e-07, "loss": 0.38311028480529785, "step": 9 }, { "epoch": 0.002596053997923157, "grad_norm": 16.43436050415039, "learning_rate": 1.9396551724137932e-07, "loss": 0.38997113704681396, "step": 10 }, { "epoch": 0.0028556593977154725, "grad_norm": 15.617770195007324, "learning_rate": 2.1551724137931036e-07, "loss": 0.3789043426513672, "step": 11 }, { "epoch": 0.003115264797507788, "grad_norm": 18.747573852539062, "learning_rate": 2.3706896551724138e-07, "loss": 0.3822753429412842, "step": 12 }, { "epoch": 0.0033748701973001036, "grad_norm": 13.969132423400879, "learning_rate": 2.5862068965517245e-07, "loss": 0.3712037205696106, "step": 13 }, { "epoch": 0.0036344755970924196, "grad_norm": 11.906434059143066, "learning_rate": 2.801724137931035e-07, "loss": 0.3442502021789551, "step": 14 }, { "epoch": 0.003894080996884735, "grad_norm": 12.428316116333008, "learning_rate": 3.0172413793103453e-07, "loss": 0.3609137535095215, "step": 15 }, { "epoch": 0.004153686396677051, "grad_norm": 8.489458084106445, "learning_rate": 3.2327586206896557e-07, "loss": 0.31811970472335815, "step": 16 }, { "epoch": 0.004413291796469367, "grad_norm": 13.103654861450195, "learning_rate": 3.4482758620689656e-07, "loss": 0.3652331233024597, "step": 17 }, { "epoch": 0.004672897196261682, "grad_norm": 15.844515800476074, "learning_rate": 3.663793103448276e-07, "loss": 0.35445016622543335, "step": 18 }, { "epoch": 0.004932502596053998, "grad_norm": 13.7569580078125, "learning_rate": 3.8793103448275865e-07, "loss": 0.34774571657180786, "step": 19 }, { "epoch": 0.005192107995846314, "grad_norm": 10.151107788085938, "learning_rate": 4.094827586206897e-07, "loss": 0.3103325664997101, "step": 20 }, { "epoch": 0.005451713395638629, "grad_norm": 10.302288055419922, "learning_rate": 4.3103448275862073e-07, "loss": 0.29579514265060425, "step": 21 }, { "epoch": 0.005711318795430945, "grad_norm": 13.883718490600586, "learning_rate": 4.5258620689655177e-07, "loss": 0.31388720870018005, "step": 22 }, { "epoch": 0.005970924195223261, "grad_norm": 16.208032608032227, "learning_rate": 4.7413793103448276e-07, "loss": 0.31099939346313477, "step": 23 }, { "epoch": 0.006230529595015576, "grad_norm": 9.975878715515137, "learning_rate": 4.956896551724138e-07, "loss": 0.28410786390304565, "step": 24 }, { "epoch": 0.006490134994807892, "grad_norm": 8.899971961975098, "learning_rate": 5.172413793103449e-07, "loss": 0.31037724018096924, "step": 25 }, { "epoch": 0.006749740394600207, "grad_norm": 5.783668041229248, "learning_rate": 5.387931034482759e-07, "loss": 0.2744828462600708, "step": 26 }, { "epoch": 0.007009345794392523, "grad_norm": 13.521904945373535, "learning_rate": 5.60344827586207e-07, "loss": 0.2683793306350708, "step": 27 }, { "epoch": 0.007268951194184839, "grad_norm": 4.156826972961426, "learning_rate": 5.81896551724138e-07, "loss": 0.25069260597229004, "step": 28 }, { "epoch": 0.007528556593977154, "grad_norm": 6.254832744598389, "learning_rate": 6.034482758620691e-07, "loss": 0.2843886613845825, "step": 29 }, { "epoch": 0.00778816199376947, "grad_norm": 6.3393635749816895, "learning_rate": 6.25e-07, "loss": 0.2864646315574646, "step": 30 }, { "epoch": 0.008047767393561786, "grad_norm": 11.793070793151855, "learning_rate": 6.465517241379311e-07, "loss": 0.2959200143814087, "step": 31 }, { "epoch": 0.008307372793354102, "grad_norm": 7.0782270431518555, "learning_rate": 6.68103448275862e-07, "loss": 0.26904475688934326, "step": 32 }, { "epoch": 0.008566978193146417, "grad_norm": 7.642717361450195, "learning_rate": 6.896551724137931e-07, "loss": 0.2439689338207245, "step": 33 }, { "epoch": 0.008826583592938734, "grad_norm": 9.006231307983398, "learning_rate": 7.112068965517242e-07, "loss": 0.2496529221534729, "step": 34 }, { "epoch": 0.009086188992731049, "grad_norm": 7.655309677124023, "learning_rate": 7.327586206896552e-07, "loss": 0.2773665487766266, "step": 35 }, { "epoch": 0.009345794392523364, "grad_norm": 9.69796085357666, "learning_rate": 7.543103448275863e-07, "loss": 0.312483549118042, "step": 36 }, { "epoch": 0.00960539979231568, "grad_norm": 5.296947479248047, "learning_rate": 7.758620689655173e-07, "loss": 0.28662991523742676, "step": 37 }, { "epoch": 0.009865005192107996, "grad_norm": 5.091712474822998, "learning_rate": 7.974137931034484e-07, "loss": 0.2685718238353729, "step": 38 }, { "epoch": 0.010124610591900311, "grad_norm": 6.756965637207031, "learning_rate": 8.189655172413794e-07, "loss": 0.29394811391830444, "step": 39 }, { "epoch": 0.010384215991692628, "grad_norm": 3.317077875137329, "learning_rate": 8.405172413793105e-07, "loss": 0.21068677306175232, "step": 40 }, { "epoch": 0.010643821391484943, "grad_norm": 9.62589168548584, "learning_rate": 8.620689655172415e-07, "loss": 0.2522183954715729, "step": 41 }, { "epoch": 0.010903426791277258, "grad_norm": 5.092556953430176, "learning_rate": 8.836206896551724e-07, "loss": 0.24760404229164124, "step": 42 }, { "epoch": 0.011163032191069575, "grad_norm": 7.874035835266113, "learning_rate": 9.051724137931035e-07, "loss": 0.2763866186141968, "step": 43 }, { "epoch": 0.01142263759086189, "grad_norm": 4.340612411499023, "learning_rate": 9.267241379310346e-07, "loss": 0.23534998297691345, "step": 44 }, { "epoch": 0.011682242990654205, "grad_norm": 4.73166036605835, "learning_rate": 9.482758620689655e-07, "loss": 0.23033633828163147, "step": 45 }, { "epoch": 0.011941848390446522, "grad_norm": 4.109215259552002, "learning_rate": 9.698275862068966e-07, "loss": 0.23423105478286743, "step": 46 }, { "epoch": 0.012201453790238837, "grad_norm": 7.1240949630737305, "learning_rate": 9.913793103448276e-07, "loss": 0.2511921525001526, "step": 47 }, { "epoch": 0.012461059190031152, "grad_norm": 3.5968129634857178, "learning_rate": 1.0129310344827586e-06, "loss": 0.24115188419818878, "step": 48 }, { "epoch": 0.01272066458982347, "grad_norm": 6.263366222381592, "learning_rate": 1.0344827586206898e-06, "loss": 0.2576523721218109, "step": 49 }, { "epoch": 0.012980269989615784, "grad_norm": 3.2625136375427246, "learning_rate": 1.0560344827586208e-06, "loss": 0.20981857180595398, "step": 50 }, { "epoch": 0.0132398753894081, "grad_norm": 3.771984577178955, "learning_rate": 1.0775862068965518e-06, "loss": 0.21119582653045654, "step": 51 }, { "epoch": 0.013499480789200415, "grad_norm": 1.9114420413970947, "learning_rate": 1.0991379310344828e-06, "loss": 0.1969829499721527, "step": 52 }, { "epoch": 0.013759086188992731, "grad_norm": 4.782560348510742, "learning_rate": 1.120689655172414e-06, "loss": 0.20493212342262268, "step": 53 }, { "epoch": 0.014018691588785047, "grad_norm": 3.090500831604004, "learning_rate": 1.142241379310345e-06, "loss": 0.24320903420448303, "step": 54 }, { "epoch": 0.014278296988577362, "grad_norm": 3.8201966285705566, "learning_rate": 1.163793103448276e-06, "loss": 0.24091142416000366, "step": 55 }, { "epoch": 0.014537902388369679, "grad_norm": 3.0730550289154053, "learning_rate": 1.185344827586207e-06, "loss": 0.21931709349155426, "step": 56 }, { "epoch": 0.014797507788161994, "grad_norm": 3.080374002456665, "learning_rate": 1.2068965517241381e-06, "loss": 0.25993984937667847, "step": 57 }, { "epoch": 0.015057113187954309, "grad_norm": 6.388272285461426, "learning_rate": 1.2284482758620691e-06, "loss": 0.2850598096847534, "step": 58 }, { "epoch": 0.015316718587746626, "grad_norm": 3.370537042617798, "learning_rate": 1.25e-06, "loss": 0.18872398138046265, "step": 59 }, { "epoch": 0.01557632398753894, "grad_norm": 5.399688243865967, "learning_rate": 1.271551724137931e-06, "loss": 0.22195716202259064, "step": 60 }, { "epoch": 0.015835929387331256, "grad_norm": 4.410447120666504, "learning_rate": 1.2931034482758623e-06, "loss": 0.24621836841106415, "step": 61 }, { "epoch": 0.016095534787123573, "grad_norm": 2.8271846771240234, "learning_rate": 1.3146551724137933e-06, "loss": 0.1794200837612152, "step": 62 }, { "epoch": 0.016355140186915886, "grad_norm": 2.8750758171081543, "learning_rate": 1.336206896551724e-06, "loss": 0.2097303718328476, "step": 63 }, { "epoch": 0.016614745586708203, "grad_norm": 5.856748104095459, "learning_rate": 1.3577586206896553e-06, "loss": 0.2488836944103241, "step": 64 }, { "epoch": 0.01687435098650052, "grad_norm": 5.147827625274658, "learning_rate": 1.3793103448275862e-06, "loss": 0.24072429537773132, "step": 65 }, { "epoch": 0.017133956386292833, "grad_norm": 2.811985731124878, "learning_rate": 1.4008620689655172e-06, "loss": 0.18433433771133423, "step": 66 }, { "epoch": 0.01739356178608515, "grad_norm": 2.8571386337280273, "learning_rate": 1.4224137931034484e-06, "loss": 0.2112058848142624, "step": 67 }, { "epoch": 0.017653167185877467, "grad_norm": 3.36450457572937, "learning_rate": 1.4439655172413794e-06, "loss": 0.26114165782928467, "step": 68 }, { "epoch": 0.01791277258566978, "grad_norm": 3.0608627796173096, "learning_rate": 1.4655172413793104e-06, "loss": 0.19026315212249756, "step": 69 }, { "epoch": 0.018172377985462097, "grad_norm": 3.4586384296417236, "learning_rate": 1.4870689655172416e-06, "loss": 0.19382083415985107, "step": 70 }, { "epoch": 0.018431983385254414, "grad_norm": 4.413562774658203, "learning_rate": 1.5086206896551726e-06, "loss": 0.18354101479053497, "step": 71 }, { "epoch": 0.018691588785046728, "grad_norm": 1.9211065769195557, "learning_rate": 1.5301724137931034e-06, "loss": 0.18294459581375122, "step": 72 }, { "epoch": 0.018951194184839044, "grad_norm": 6.141068935394287, "learning_rate": 1.5517241379310346e-06, "loss": 0.21728020906448364, "step": 73 }, { "epoch": 0.01921079958463136, "grad_norm": 6.084436416625977, "learning_rate": 1.5732758620689656e-06, "loss": 0.22853997349739075, "step": 74 }, { "epoch": 0.019470404984423675, "grad_norm": 2.8424758911132812, "learning_rate": 1.5948275862068968e-06, "loss": 0.17517703771591187, "step": 75 }, { "epoch": 0.01973001038421599, "grad_norm": 2.0756947994232178, "learning_rate": 1.6163793103448278e-06, "loss": 0.19106720387935638, "step": 76 }, { "epoch": 0.01998961578400831, "grad_norm": 7.189305782318115, "learning_rate": 1.6379310344827587e-06, "loss": 0.25200608372688293, "step": 77 }, { "epoch": 0.020249221183800622, "grad_norm": 5.6033430099487305, "learning_rate": 1.65948275862069e-06, "loss": 0.19225570559501648, "step": 78 }, { "epoch": 0.02050882658359294, "grad_norm": 3.7203330993652344, "learning_rate": 1.681034482758621e-06, "loss": 0.2240765243768692, "step": 79 }, { "epoch": 0.020768431983385256, "grad_norm": 4.0030131340026855, "learning_rate": 1.7025862068965517e-06, "loss": 0.23276370763778687, "step": 80 }, { "epoch": 0.02102803738317757, "grad_norm": 6.116404056549072, "learning_rate": 1.724137931034483e-06, "loss": 0.2040616273880005, "step": 81 }, { "epoch": 0.021287642782969886, "grad_norm": 6.759488105773926, "learning_rate": 1.745689655172414e-06, "loss": 0.1976706087589264, "step": 82 }, { "epoch": 0.021547248182762203, "grad_norm": 2.898364543914795, "learning_rate": 1.7672413793103449e-06, "loss": 0.2271040380001068, "step": 83 }, { "epoch": 0.021806853582554516, "grad_norm": 3.1605801582336426, "learning_rate": 1.788793103448276e-06, "loss": 0.2191096544265747, "step": 84 }, { "epoch": 0.022066458982346833, "grad_norm": 3.4442529678344727, "learning_rate": 1.810344827586207e-06, "loss": 0.22075694799423218, "step": 85 }, { "epoch": 0.02232606438213915, "grad_norm": 5.875351905822754, "learning_rate": 1.831896551724138e-06, "loss": 0.2434518039226532, "step": 86 }, { "epoch": 0.022585669781931463, "grad_norm": 2.735271692276001, "learning_rate": 1.8534482758620693e-06, "loss": 0.2024746835231781, "step": 87 }, { "epoch": 0.02284527518172378, "grad_norm": 4.973536014556885, "learning_rate": 1.8750000000000003e-06, "loss": 0.23418933153152466, "step": 88 }, { "epoch": 0.023104880581516097, "grad_norm": 2.2572638988494873, "learning_rate": 1.896551724137931e-06, "loss": 0.1837228536605835, "step": 89 }, { "epoch": 0.02336448598130841, "grad_norm": 5.087561130523682, "learning_rate": 1.9181034482758622e-06, "loss": 0.21041950583457947, "step": 90 }, { "epoch": 0.023624091381100727, "grad_norm": 6.0947065353393555, "learning_rate": 1.9396551724137932e-06, "loss": 0.22193646430969238, "step": 91 }, { "epoch": 0.023883696780893044, "grad_norm": 3.2364144325256348, "learning_rate": 1.9612068965517242e-06, "loss": 0.2195255160331726, "step": 92 }, { "epoch": 0.024143302180685357, "grad_norm": 4.876420974731445, "learning_rate": 1.982758620689655e-06, "loss": 0.17967280745506287, "step": 93 }, { "epoch": 0.024402907580477674, "grad_norm": 2.9650182723999023, "learning_rate": 2.004310344827586e-06, "loss": 0.19053241610527039, "step": 94 }, { "epoch": 0.02466251298026999, "grad_norm": 1.2757201194763184, "learning_rate": 2.025862068965517e-06, "loss": 0.17184630036354065, "step": 95 }, { "epoch": 0.024922118380062305, "grad_norm": 1.4904932975769043, "learning_rate": 2.0474137931034486e-06, "loss": 0.17176476120948792, "step": 96 }, { "epoch": 0.02518172377985462, "grad_norm": 4.561407566070557, "learning_rate": 2.0689655172413796e-06, "loss": 0.1999681293964386, "step": 97 }, { "epoch": 0.02544132917964694, "grad_norm": 4.4178147315979, "learning_rate": 2.0905172413793106e-06, "loss": 0.21359455585479736, "step": 98 }, { "epoch": 0.02570093457943925, "grad_norm": 1.1979931592941284, "learning_rate": 2.1120689655172416e-06, "loss": 0.17301452159881592, "step": 99 }, { "epoch": 0.02596053997923157, "grad_norm": 1.241378903388977, "learning_rate": 2.1336206896551725e-06, "loss": 0.17165106534957886, "step": 100 }, { "epoch": 0.026220145379023882, "grad_norm": 1.9226593971252441, "learning_rate": 2.1551724137931035e-06, "loss": 0.18232060968875885, "step": 101 }, { "epoch": 0.0264797507788162, "grad_norm": 3.2651264667510986, "learning_rate": 2.1767241379310345e-06, "loss": 0.20456844568252563, "step": 102 }, { "epoch": 0.026739356178608516, "grad_norm": 3.2870519161224365, "learning_rate": 2.1982758620689655e-06, "loss": 0.19964030385017395, "step": 103 }, { "epoch": 0.02699896157840083, "grad_norm": 2.0126523971557617, "learning_rate": 2.219827586206897e-06, "loss": 0.17268875241279602, "step": 104 }, { "epoch": 0.027258566978193146, "grad_norm": 1.9050925970077515, "learning_rate": 2.241379310344828e-06, "loss": 0.21724924445152283, "step": 105 }, { "epoch": 0.027518172377985463, "grad_norm": 4.3465094566345215, "learning_rate": 2.262931034482759e-06, "loss": 0.19260179996490479, "step": 106 }, { "epoch": 0.027777777777777776, "grad_norm": 2.006563425064087, "learning_rate": 2.28448275862069e-06, "loss": 0.1687426120042801, "step": 107 }, { "epoch": 0.028037383177570093, "grad_norm": 1.710978627204895, "learning_rate": 2.306034482758621e-06, "loss": 0.16151756048202515, "step": 108 }, { "epoch": 0.02829698857736241, "grad_norm": 2.020195722579956, "learning_rate": 2.327586206896552e-06, "loss": 0.16580234467983246, "step": 109 }, { "epoch": 0.028556593977154723, "grad_norm": 3.2942428588867188, "learning_rate": 2.349137931034483e-06, "loss": 0.24338218569755554, "step": 110 }, { "epoch": 0.02881619937694704, "grad_norm": 3.921160936355591, "learning_rate": 2.370689655172414e-06, "loss": 0.2552923858165741, "step": 111 }, { "epoch": 0.029075804776739357, "grad_norm": 4.70251989364624, "learning_rate": 2.392241379310345e-06, "loss": 0.22988329827785492, "step": 112 }, { "epoch": 0.02933541017653167, "grad_norm": 5.694077491760254, "learning_rate": 2.4137931034482762e-06, "loss": 0.24848933517932892, "step": 113 }, { "epoch": 0.029595015576323987, "grad_norm": 2.766247510910034, "learning_rate": 2.4353448275862072e-06, "loss": 0.18685346841812134, "step": 114 }, { "epoch": 0.029854620976116304, "grad_norm": 1.6098958253860474, "learning_rate": 2.4568965517241382e-06, "loss": 0.1965133547782898, "step": 115 }, { "epoch": 0.030114226375908618, "grad_norm": 2.2564351558685303, "learning_rate": 2.4784482758620692e-06, "loss": 0.21073457598686218, "step": 116 }, { "epoch": 0.030373831775700934, "grad_norm": 2.8809213638305664, "learning_rate": 2.5e-06, "loss": 0.16818785667419434, "step": 117 }, { "epoch": 0.03063343717549325, "grad_norm": 4.195605754852295, "learning_rate": 2.521551724137931e-06, "loss": 0.22680827975273132, "step": 118 }, { "epoch": 0.030893042575285565, "grad_norm": 2.791508197784424, "learning_rate": 2.543103448275862e-06, "loss": 0.21383698284626007, "step": 119 }, { "epoch": 0.03115264797507788, "grad_norm": 1.2711834907531738, "learning_rate": 2.5646551724137936e-06, "loss": 0.1648809164762497, "step": 120 }, { "epoch": 0.0314122533748702, "grad_norm": 2.2413330078125, "learning_rate": 2.5862068965517246e-06, "loss": 0.23422105610370636, "step": 121 }, { "epoch": 0.03167185877466251, "grad_norm": 2.2111761569976807, "learning_rate": 2.6077586206896556e-06, "loss": 0.18487215042114258, "step": 122 }, { "epoch": 0.031931464174454825, "grad_norm": 4.019171237945557, "learning_rate": 2.6293103448275866e-06, "loss": 0.17409738898277283, "step": 123 }, { "epoch": 0.032191069574247146, "grad_norm": 1.4829010963439941, "learning_rate": 2.6508620689655175e-06, "loss": 0.17850497364997864, "step": 124 }, { "epoch": 0.03245067497403946, "grad_norm": 6.580495357513428, "learning_rate": 2.672413793103448e-06, "loss": 0.2528199255466461, "step": 125 }, { "epoch": 0.03271028037383177, "grad_norm": 2.037635564804077, "learning_rate": 2.69396551724138e-06, "loss": 0.1594310700893402, "step": 126 }, { "epoch": 0.03296988577362409, "grad_norm": 5.925559997558594, "learning_rate": 2.7155172413793105e-06, "loss": 0.2122388482093811, "step": 127 }, { "epoch": 0.033229491173416406, "grad_norm": 2.0072898864746094, "learning_rate": 2.7370689655172415e-06, "loss": 0.17905761301517487, "step": 128 }, { "epoch": 0.03348909657320872, "grad_norm": 1.4345448017120361, "learning_rate": 2.7586206896551725e-06, "loss": 0.19899123907089233, "step": 129 }, { "epoch": 0.03374870197300104, "grad_norm": 3.9060146808624268, "learning_rate": 2.7801724137931035e-06, "loss": 0.18627023696899414, "step": 130 }, { "epoch": 0.03400830737279335, "grad_norm": 1.8426988124847412, "learning_rate": 2.8017241379310345e-06, "loss": 0.1740626096725464, "step": 131 }, { "epoch": 0.03426791277258567, "grad_norm": 3.463550329208374, "learning_rate": 2.823275862068966e-06, "loss": 0.21081481873989105, "step": 132 }, { "epoch": 0.03452751817237799, "grad_norm": 3.4020581245422363, "learning_rate": 2.844827586206897e-06, "loss": 0.18374650180339813, "step": 133 }, { "epoch": 0.0347871235721703, "grad_norm": 1.7664916515350342, "learning_rate": 2.866379310344828e-06, "loss": 0.167881578207016, "step": 134 }, { "epoch": 0.035046728971962614, "grad_norm": 4.742146968841553, "learning_rate": 2.887931034482759e-06, "loss": 0.20318245887756348, "step": 135 }, { "epoch": 0.035306334371754934, "grad_norm": 3.949084520339966, "learning_rate": 2.90948275862069e-06, "loss": 0.22531503438949585, "step": 136 }, { "epoch": 0.03556593977154725, "grad_norm": 2.8814988136291504, "learning_rate": 2.931034482758621e-06, "loss": 0.20982488989830017, "step": 137 }, { "epoch": 0.03582554517133956, "grad_norm": 3.7385077476501465, "learning_rate": 2.9525862068965522e-06, "loss": 0.17434723675251007, "step": 138 }, { "epoch": 0.03608515057113188, "grad_norm": 2.3230044841766357, "learning_rate": 2.9741379310344832e-06, "loss": 0.17183014750480652, "step": 139 }, { "epoch": 0.036344755970924195, "grad_norm": 4.688577175140381, "learning_rate": 2.9956896551724142e-06, "loss": 0.21502414345741272, "step": 140 }, { "epoch": 0.03660436137071651, "grad_norm": 4.829170227050781, "learning_rate": 3.017241379310345e-06, "loss": 0.26114487648010254, "step": 141 }, { "epoch": 0.03686396677050883, "grad_norm": 2.6513900756835938, "learning_rate": 3.0387931034482758e-06, "loss": 0.19872373342514038, "step": 142 }, { "epoch": 0.03712357217030114, "grad_norm": 2.476109743118286, "learning_rate": 3.0603448275862068e-06, "loss": 0.17572513222694397, "step": 143 }, { "epoch": 0.037383177570093455, "grad_norm": 2.595477819442749, "learning_rate": 3.081896551724138e-06, "loss": 0.1948859989643097, "step": 144 }, { "epoch": 0.037642782969885775, "grad_norm": 2.539123058319092, "learning_rate": 3.103448275862069e-06, "loss": 0.18348285555839539, "step": 145 }, { "epoch": 0.03790238836967809, "grad_norm": 6.364838123321533, "learning_rate": 3.125e-06, "loss": 0.18436114490032196, "step": 146 }, { "epoch": 0.0381619937694704, "grad_norm": 2.7866876125335693, "learning_rate": 3.146551724137931e-06, "loss": 0.22716522216796875, "step": 147 }, { "epoch": 0.03842159916926272, "grad_norm": 4.604449272155762, "learning_rate": 3.168103448275862e-06, "loss": 0.2377164512872696, "step": 148 }, { "epoch": 0.038681204569055036, "grad_norm": 4.9203362464904785, "learning_rate": 3.1896551724137935e-06, "loss": 0.16760745644569397, "step": 149 }, { "epoch": 0.03894080996884735, "grad_norm": 1.6632188558578491, "learning_rate": 3.2112068965517245e-06, "loss": 0.17331033945083618, "step": 150 }, { "epoch": 0.03920041536863967, "grad_norm": 4.9993205070495605, "learning_rate": 3.2327586206896555e-06, "loss": 0.18094709515571594, "step": 151 }, { "epoch": 0.03946002076843198, "grad_norm": 3.521925449371338, "learning_rate": 3.2543103448275865e-06, "loss": 0.22727259993553162, "step": 152 }, { "epoch": 0.0397196261682243, "grad_norm": 2.1458919048309326, "learning_rate": 3.2758620689655175e-06, "loss": 0.17121663689613342, "step": 153 }, { "epoch": 0.03997923156801662, "grad_norm": 2.5441694259643555, "learning_rate": 3.2974137931034485e-06, "loss": 0.1749107986688614, "step": 154 }, { "epoch": 0.04023883696780893, "grad_norm": 3.408388137817383, "learning_rate": 3.31896551724138e-06, "loss": 0.18045544624328613, "step": 155 }, { "epoch": 0.040498442367601244, "grad_norm": 2.1895456314086914, "learning_rate": 3.340517241379311e-06, "loss": 0.16883021593093872, "step": 156 }, { "epoch": 0.040758047767393564, "grad_norm": 4.000824451446533, "learning_rate": 3.362068965517242e-06, "loss": 0.19675695896148682, "step": 157 }, { "epoch": 0.04101765316718588, "grad_norm": 8.38481616973877, "learning_rate": 3.383620689655173e-06, "loss": 0.23398517072200775, "step": 158 }, { "epoch": 0.04127725856697819, "grad_norm": 2.093780517578125, "learning_rate": 3.4051724137931034e-06, "loss": 0.1687224954366684, "step": 159 }, { "epoch": 0.04153686396677051, "grad_norm": 5.233348846435547, "learning_rate": 3.4267241379310344e-06, "loss": 0.18814149498939514, "step": 160 }, { "epoch": 0.041796469366562825, "grad_norm": 2.9181392192840576, "learning_rate": 3.448275862068966e-06, "loss": 0.16316556930541992, "step": 161 }, { "epoch": 0.04205607476635514, "grad_norm": 4.580896854400635, "learning_rate": 3.469827586206897e-06, "loss": 0.2160249650478363, "step": 162 }, { "epoch": 0.04231568016614746, "grad_norm": 6.627445697784424, "learning_rate": 3.491379310344828e-06, "loss": 0.19536840915679932, "step": 163 }, { "epoch": 0.04257528556593977, "grad_norm": 4.890913963317871, "learning_rate": 3.512931034482759e-06, "loss": 0.18500488996505737, "step": 164 }, { "epoch": 0.042834890965732085, "grad_norm": 1.8186455965042114, "learning_rate": 3.5344827586206898e-06, "loss": 0.2155856490135193, "step": 165 }, { "epoch": 0.043094496365524405, "grad_norm": 2.2671046257019043, "learning_rate": 3.5560344827586208e-06, "loss": 0.1994480937719345, "step": 166 }, { "epoch": 0.04335410176531672, "grad_norm": 3.4834201335906982, "learning_rate": 3.577586206896552e-06, "loss": 0.20446015894412994, "step": 167 }, { "epoch": 0.04361370716510903, "grad_norm": 4.226232528686523, "learning_rate": 3.599137931034483e-06, "loss": 0.21279364824295044, "step": 168 }, { "epoch": 0.04387331256490135, "grad_norm": 1.4287127256393433, "learning_rate": 3.620689655172414e-06, "loss": 0.2206575721502304, "step": 169 }, { "epoch": 0.044132917964693666, "grad_norm": 5.1180830001831055, "learning_rate": 3.642241379310345e-06, "loss": 0.2262328565120697, "step": 170 }, { "epoch": 0.04439252336448598, "grad_norm": 3.274852991104126, "learning_rate": 3.663793103448276e-06, "loss": 0.18737316131591797, "step": 171 }, { "epoch": 0.0446521287642783, "grad_norm": 3.09147047996521, "learning_rate": 3.685344827586207e-06, "loss": 0.2473624050617218, "step": 172 }, { "epoch": 0.04491173416407061, "grad_norm": 2.1620044708251953, "learning_rate": 3.7068965517241385e-06, "loss": 0.18144139647483826, "step": 173 }, { "epoch": 0.045171339563862926, "grad_norm": 2.2284724712371826, "learning_rate": 3.7284482758620695e-06, "loss": 0.16838215291500092, "step": 174 }, { "epoch": 0.04543094496365525, "grad_norm": 3.9536876678466797, "learning_rate": 3.7500000000000005e-06, "loss": 0.1978730708360672, "step": 175 }, { "epoch": 0.04569055036344756, "grad_norm": 1.978686809539795, "learning_rate": 3.771551724137931e-06, "loss": 0.16089966893196106, "step": 176 }, { "epoch": 0.045950155763239874, "grad_norm": 2.8891968727111816, "learning_rate": 3.793103448275862e-06, "loss": 0.2205294668674469, "step": 177 }, { "epoch": 0.046209761163032194, "grad_norm": 2.7828853130340576, "learning_rate": 3.8146551724137935e-06, "loss": 0.17377643287181854, "step": 178 }, { "epoch": 0.04646936656282451, "grad_norm": 4.030176639556885, "learning_rate": 3.8362068965517245e-06, "loss": 0.19668154418468475, "step": 179 }, { "epoch": 0.04672897196261682, "grad_norm": 1.9580843448638916, "learning_rate": 3.857758620689656e-06, "loss": 0.19276678562164307, "step": 180 }, { "epoch": 0.04698857736240914, "grad_norm": 3.33959698677063, "learning_rate": 3.8793103448275865e-06, "loss": 0.23855692148208618, "step": 181 }, { "epoch": 0.047248182762201454, "grad_norm": 1.3503426313400269, "learning_rate": 3.900862068965518e-06, "loss": 0.16856056451797485, "step": 182 }, { "epoch": 0.04750778816199377, "grad_norm": 3.164454698562622, "learning_rate": 3.9224137931034484e-06, "loss": 0.20444068312644958, "step": 183 }, { "epoch": 0.04776739356178609, "grad_norm": 1.568030595779419, "learning_rate": 3.94396551724138e-06, "loss": 0.16276738047599792, "step": 184 }, { "epoch": 0.0480269989615784, "grad_norm": 2.0219430923461914, "learning_rate": 3.96551724137931e-06, "loss": 0.19388584792613983, "step": 185 }, { "epoch": 0.048286604361370715, "grad_norm": 1.2395424842834473, "learning_rate": 3.987068965517242e-06, "loss": 0.15929441154003143, "step": 186 }, { "epoch": 0.048546209761163035, "grad_norm": 3.4976437091827393, "learning_rate": 4.008620689655172e-06, "loss": 0.16257381439208984, "step": 187 }, { "epoch": 0.04880581516095535, "grad_norm": 1.2974872589111328, "learning_rate": 4.030172413793104e-06, "loss": 0.16297683119773865, "step": 188 }, { "epoch": 0.04906542056074766, "grad_norm": 0.6613505482673645, "learning_rate": 4.051724137931034e-06, "loss": 0.15624482929706573, "step": 189 }, { "epoch": 0.04932502596053998, "grad_norm": 1.3159602880477905, "learning_rate": 4.073275862068966e-06, "loss": 0.1626216471195221, "step": 190 }, { "epoch": 0.049584631360332296, "grad_norm": 1.5671964883804321, "learning_rate": 4.094827586206897e-06, "loss": 0.14562714099884033, "step": 191 }, { "epoch": 0.04984423676012461, "grad_norm": 4.576347827911377, "learning_rate": 4.116379310344828e-06, "loss": 0.21973174810409546, "step": 192 }, { "epoch": 0.05010384215991693, "grad_norm": 3.416248083114624, "learning_rate": 4.137931034482759e-06, "loss": 0.1864205151796341, "step": 193 }, { "epoch": 0.05036344755970924, "grad_norm": 2.0576677322387695, "learning_rate": 4.15948275862069e-06, "loss": 0.19345176219940186, "step": 194 }, { "epoch": 0.050623052959501556, "grad_norm": 1.3695449829101562, "learning_rate": 4.181034482758621e-06, "loss": 0.16834351420402527, "step": 195 }, { "epoch": 0.05088265835929388, "grad_norm": 2.328423500061035, "learning_rate": 4.2025862068965526e-06, "loss": 0.20497703552246094, "step": 196 }, { "epoch": 0.05114226375908619, "grad_norm": 2.527583360671997, "learning_rate": 4.224137931034483e-06, "loss": 0.1794969141483307, "step": 197 }, { "epoch": 0.0514018691588785, "grad_norm": 1.487028956413269, "learning_rate": 4.2456896551724145e-06, "loss": 0.16057619452476501, "step": 198 }, { "epoch": 0.051661474558670824, "grad_norm": 1.8003592491149902, "learning_rate": 4.267241379310345e-06, "loss": 0.1954765021800995, "step": 199 }, { "epoch": 0.05192107995846314, "grad_norm": 3.867730140686035, "learning_rate": 4.288793103448276e-06, "loss": 0.21561412513256073, "step": 200 }, { "epoch": 0.05218068535825545, "grad_norm": 1.9686810970306396, "learning_rate": 4.310344827586207e-06, "loss": 0.2062413990497589, "step": 201 }, { "epoch": 0.052440290758047764, "grad_norm": 4.321706295013428, "learning_rate": 4.3318965517241385e-06, "loss": 0.1932206153869629, "step": 202 }, { "epoch": 0.052699896157840084, "grad_norm": 1.9081776142120361, "learning_rate": 4.353448275862069e-06, "loss": 0.18990421295166016, "step": 203 }, { "epoch": 0.0529595015576324, "grad_norm": 4.317255973815918, "learning_rate": 4.3750000000000005e-06, "loss": 0.181911438703537, "step": 204 }, { "epoch": 0.05321910695742471, "grad_norm": 4.310423374176025, "learning_rate": 4.396551724137931e-06, "loss": 0.20901861786842346, "step": 205 }, { "epoch": 0.05347871235721703, "grad_norm": 3.9686460494995117, "learning_rate": 4.4181034482758624e-06, "loss": 0.1645030528306961, "step": 206 }, { "epoch": 0.053738317757009345, "grad_norm": 4.166399955749512, "learning_rate": 4.439655172413794e-06, "loss": 0.22936759889125824, "step": 207 }, { "epoch": 0.05399792315680166, "grad_norm": 3.7807960510253906, "learning_rate": 4.461206896551724e-06, "loss": 0.195648193359375, "step": 208 }, { "epoch": 0.05425752855659398, "grad_norm": 3.524062395095825, "learning_rate": 4.482758620689656e-06, "loss": 0.19714024662971497, "step": 209 }, { "epoch": 0.05451713395638629, "grad_norm": 3.3245787620544434, "learning_rate": 4.504310344827586e-06, "loss": 0.1719909906387329, "step": 210 }, { "epoch": 0.054776739356178605, "grad_norm": 1.952414870262146, "learning_rate": 4.525862068965518e-06, "loss": 0.22874316573143005, "step": 211 }, { "epoch": 0.055036344755970926, "grad_norm": 3.7962143421173096, "learning_rate": 4.547413793103448e-06, "loss": 0.20557865500450134, "step": 212 }, { "epoch": 0.05529595015576324, "grad_norm": 1.686001181602478, "learning_rate": 4.56896551724138e-06, "loss": 0.18717244267463684, "step": 213 }, { "epoch": 0.05555555555555555, "grad_norm": 2.1163525581359863, "learning_rate": 4.590517241379311e-06, "loss": 0.15467745065689087, "step": 214 }, { "epoch": 0.05581516095534787, "grad_norm": 1.1937857866287231, "learning_rate": 4.612068965517242e-06, "loss": 0.17140811681747437, "step": 215 }, { "epoch": 0.056074766355140186, "grad_norm": 1.9237440824508667, "learning_rate": 4.633620689655173e-06, "loss": 0.19504371285438538, "step": 216 }, { "epoch": 0.0563343717549325, "grad_norm": 1.9368345737457275, "learning_rate": 4.655172413793104e-06, "loss": 0.1604613959789276, "step": 217 }, { "epoch": 0.05659397715472482, "grad_norm": 1.2884128093719482, "learning_rate": 4.676724137931034e-06, "loss": 0.1613233983516693, "step": 218 }, { "epoch": 0.05685358255451713, "grad_norm": 2.31583833694458, "learning_rate": 4.698275862068966e-06, "loss": 0.23977792263031006, "step": 219 }, { "epoch": 0.05711318795430945, "grad_norm": 0.9721299409866333, "learning_rate": 4.719827586206897e-06, "loss": 0.17529430985450745, "step": 220 }, { "epoch": 0.05737279335410177, "grad_norm": 0.7127910256385803, "learning_rate": 4.741379310344828e-06, "loss": 0.152650848031044, "step": 221 }, { "epoch": 0.05763239875389408, "grad_norm": 1.2793570756912231, "learning_rate": 4.762931034482759e-06, "loss": 0.16592097282409668, "step": 222 }, { "epoch": 0.057892004153686394, "grad_norm": 1.3868319988250732, "learning_rate": 4.78448275862069e-06, "loss": 0.18324071168899536, "step": 223 }, { "epoch": 0.058151609553478714, "grad_norm": 3.039015531539917, "learning_rate": 4.806034482758621e-06, "loss": 0.17593738436698914, "step": 224 }, { "epoch": 0.05841121495327103, "grad_norm": 3.1235618591308594, "learning_rate": 4.8275862068965525e-06, "loss": 0.1859787106513977, "step": 225 }, { "epoch": 0.05867082035306334, "grad_norm": 3.532025098800659, "learning_rate": 4.849137931034483e-06, "loss": 0.21785303950309753, "step": 226 }, { "epoch": 0.05893042575285566, "grad_norm": 4.957466125488281, "learning_rate": 4.8706896551724145e-06, "loss": 0.20628277957439423, "step": 227 }, { "epoch": 0.059190031152647975, "grad_norm": 2.670999765396118, "learning_rate": 4.892241379310345e-06, "loss": 0.19997793436050415, "step": 228 }, { "epoch": 0.05944963655244029, "grad_norm": 1.6332522630691528, "learning_rate": 4.9137931034482765e-06, "loss": 0.19564470648765564, "step": 229 }, { "epoch": 0.05970924195223261, "grad_norm": 3.380540132522583, "learning_rate": 4.935344827586207e-06, "loss": 0.16969847679138184, "step": 230 }, { "epoch": 0.05996884735202492, "grad_norm": 3.067497968673706, "learning_rate": 4.9568965517241384e-06, "loss": 0.15910816192626953, "step": 231 }, { "epoch": 0.060228452751817235, "grad_norm": 1.1694614887237549, "learning_rate": 4.97844827586207e-06, "loss": 0.18652769923210144, "step": 232 }, { "epoch": 0.060488058151609556, "grad_norm": 3.288191556930542, "learning_rate": 5e-06, "loss": 0.19025374948978424, "step": 233 }, { "epoch": 0.06074766355140187, "grad_norm": 2.7786264419555664, "learning_rate": 4.99999980112576e-06, "loss": 0.21750733256340027, "step": 234 }, { "epoch": 0.06100726895119418, "grad_norm": 1.4360765218734741, "learning_rate": 4.999999204503072e-06, "loss": 0.17473462224006653, "step": 235 }, { "epoch": 0.0612668743509865, "grad_norm": 1.8487281799316406, "learning_rate": 4.999998210132044e-06, "loss": 0.16987362504005432, "step": 236 }, { "epoch": 0.061526479750778816, "grad_norm": 2.6921114921569824, "learning_rate": 4.999996818012849e-06, "loss": 0.18400321900844574, "step": 237 }, { "epoch": 0.06178608515057113, "grad_norm": 4.6220574378967285, "learning_rate": 4.999995028145735e-06, "loss": 0.17985257506370544, "step": 238 }, { "epoch": 0.06204569055036345, "grad_norm": 1.788332462310791, "learning_rate": 4.999992840531019e-06, "loss": 0.22297340631484985, "step": 239 }, { "epoch": 0.06230529595015576, "grad_norm": 1.9372018575668335, "learning_rate": 4.999990255169086e-06, "loss": 0.18539103865623474, "step": 240 }, { "epoch": 0.06256490134994808, "grad_norm": 1.6358146667480469, "learning_rate": 4.999987272060394e-06, "loss": 0.19707760214805603, "step": 241 }, { "epoch": 0.0628245067497404, "grad_norm": 2.1858153343200684, "learning_rate": 4.999983891205471e-06, "loss": 0.1700255423784256, "step": 242 }, { "epoch": 0.0630841121495327, "grad_norm": 2.1206791400909424, "learning_rate": 4.999980112604912e-06, "loss": 0.1892598271369934, "step": 243 }, { "epoch": 0.06334371754932502, "grad_norm": 1.3250949382781982, "learning_rate": 4.999975936259389e-06, "loss": 0.18715959787368774, "step": 244 }, { "epoch": 0.06360332294911734, "grad_norm": 3.856212615966797, "learning_rate": 4.9999713621696375e-06, "loss": 0.21653473377227783, "step": 245 }, { "epoch": 0.06386292834890965, "grad_norm": 1.2207112312316895, "learning_rate": 4.999966390336466e-06, "loss": 0.17290866374969482, "step": 246 }, { "epoch": 0.06412253374870197, "grad_norm": 3.0451059341430664, "learning_rate": 4.999961020760754e-06, "loss": 0.18646657466888428, "step": 247 }, { "epoch": 0.06438213914849429, "grad_norm": 2.5277702808380127, "learning_rate": 4.999955253443451e-06, "loss": 0.199119433760643, "step": 248 }, { "epoch": 0.0646417445482866, "grad_norm": 2.3805091381073, "learning_rate": 4.999949088385577e-06, "loss": 0.2273004651069641, "step": 249 }, { "epoch": 0.06490134994807892, "grad_norm": 3.15523362159729, "learning_rate": 4.9999425255882205e-06, "loss": 0.21070589125156403, "step": 250 }, { "epoch": 0.06516095534787124, "grad_norm": 3.147486448287964, "learning_rate": 4.999935565052543e-06, "loss": 0.20378440618515015, "step": 251 }, { "epoch": 0.06542056074766354, "grad_norm": 4.836396217346191, "learning_rate": 4.999928206779774e-06, "loss": 0.22961610555648804, "step": 252 }, { "epoch": 0.06568016614745587, "grad_norm": 2.98244047164917, "learning_rate": 4.999920450771214e-06, "loss": 0.20678475499153137, "step": 253 }, { "epoch": 0.06593977154724819, "grad_norm": 3.418128252029419, "learning_rate": 4.999912297028235e-06, "loss": 0.22309835255146027, "step": 254 }, { "epoch": 0.06619937694704049, "grad_norm": 6.150102138519287, "learning_rate": 4.999903745552278e-06, "loss": 0.20306190848350525, "step": 255 }, { "epoch": 0.06645898234683281, "grad_norm": 5.506738662719727, "learning_rate": 4.9998947963448554e-06, "loss": 0.2523879408836365, "step": 256 }, { "epoch": 0.06671858774662513, "grad_norm": 2.9446592330932617, "learning_rate": 4.999885449407547e-06, "loss": 0.20104053616523743, "step": 257 }, { "epoch": 0.06697819314641744, "grad_norm": 1.4500221014022827, "learning_rate": 4.999875704742009e-06, "loss": 0.202890545129776, "step": 258 }, { "epoch": 0.06723779854620976, "grad_norm": 3.423292636871338, "learning_rate": 4.99986556234996e-06, "loss": 0.2003038227558136, "step": 259 }, { "epoch": 0.06749740394600208, "grad_norm": 5.065746784210205, "learning_rate": 4.999855022233195e-06, "loss": 0.2112542688846588, "step": 260 }, { "epoch": 0.06775700934579439, "grad_norm": 2.7870681285858154, "learning_rate": 4.999844084393577e-06, "loss": 0.21625317633152008, "step": 261 }, { "epoch": 0.0680166147455867, "grad_norm": 5.068099498748779, "learning_rate": 4.99983274883304e-06, "loss": 0.1920754462480545, "step": 262 }, { "epoch": 0.06827622014537903, "grad_norm": 1.194661259651184, "learning_rate": 4.999821015553586e-06, "loss": 0.15646833181381226, "step": 263 }, { "epoch": 0.06853582554517133, "grad_norm": 2.1479997634887695, "learning_rate": 4.999808884557291e-06, "loss": 0.16188585758209229, "step": 264 }, { "epoch": 0.06879543094496365, "grad_norm": 1.5027425289154053, "learning_rate": 4.9997963558463e-06, "loss": 0.18968743085861206, "step": 265 }, { "epoch": 0.06905503634475597, "grad_norm": 4.57689905166626, "learning_rate": 4.999783429422826e-06, "loss": 0.17181605100631714, "step": 266 }, { "epoch": 0.06931464174454828, "grad_norm": 2.0137455463409424, "learning_rate": 4.999770105289155e-06, "loss": 0.19873636960983276, "step": 267 }, { "epoch": 0.0695742471443406, "grad_norm": 2.8453376293182373, "learning_rate": 4.999756383447642e-06, "loss": 0.19875454902648926, "step": 268 }, { "epoch": 0.06983385254413292, "grad_norm": 1.8628369569778442, "learning_rate": 4.999742263900714e-06, "loss": 0.21069923043251038, "step": 269 }, { "epoch": 0.07009345794392523, "grad_norm": 0.8697875142097473, "learning_rate": 4.9997277466508654e-06, "loss": 0.1547846794128418, "step": 270 }, { "epoch": 0.07035306334371755, "grad_norm": 1.8564424514770508, "learning_rate": 4.9997128317006625e-06, "loss": 0.20425237715244293, "step": 271 }, { "epoch": 0.07061266874350987, "grad_norm": 1.4307856559753418, "learning_rate": 4.999697519052743e-06, "loss": 0.16362860798835754, "step": 272 }, { "epoch": 0.07087227414330217, "grad_norm": 1.395220398902893, "learning_rate": 4.999681808709814e-06, "loss": 0.20423370599746704, "step": 273 }, { "epoch": 0.0711318795430945, "grad_norm": 3.0905601978302, "learning_rate": 4.999665700674651e-06, "loss": 0.2038843333721161, "step": 274 }, { "epoch": 0.07139148494288682, "grad_norm": 1.6508548259735107, "learning_rate": 4.999649194950103e-06, "loss": 0.21840661764144897, "step": 275 }, { "epoch": 0.07165109034267912, "grad_norm": 1.9846055507659912, "learning_rate": 4.999632291539088e-06, "loss": 0.18006153404712677, "step": 276 }, { "epoch": 0.07191069574247144, "grad_norm": 2.0074546337127686, "learning_rate": 4.999614990444594e-06, "loss": 0.19426235556602478, "step": 277 }, { "epoch": 0.07217030114226376, "grad_norm": 3.6716082096099854, "learning_rate": 4.999597291669679e-06, "loss": 0.2219502031803131, "step": 278 }, { "epoch": 0.07242990654205607, "grad_norm": 3.0885775089263916, "learning_rate": 4.99957919521747e-06, "loss": 0.16773107647895813, "step": 279 }, { "epoch": 0.07268951194184839, "grad_norm": 1.7581316232681274, "learning_rate": 4.9995607010911696e-06, "loss": 0.20300842821598053, "step": 280 }, { "epoch": 0.07294911734164071, "grad_norm": 0.899101972579956, "learning_rate": 4.999541809294045e-06, "loss": 0.1569625437259674, "step": 281 }, { "epoch": 0.07320872274143302, "grad_norm": 1.9291621446609497, "learning_rate": 4.999522519829435e-06, "loss": 0.16421914100646973, "step": 282 }, { "epoch": 0.07346832814122534, "grad_norm": 2.4294774532318115, "learning_rate": 4.999502832700752e-06, "loss": 0.16958996653556824, "step": 283 }, { "epoch": 0.07372793354101766, "grad_norm": 1.161209225654602, "learning_rate": 4.999482747911475e-06, "loss": 0.17437097430229187, "step": 284 }, { "epoch": 0.07398753894080996, "grad_norm": 1.5347989797592163, "learning_rate": 4.999462265465154e-06, "loss": 0.14994296431541443, "step": 285 }, { "epoch": 0.07424714434060228, "grad_norm": 2.75119948387146, "learning_rate": 4.99944138536541e-06, "loss": 0.1585739701986313, "step": 286 }, { "epoch": 0.0745067497403946, "grad_norm": 0.7582217454910278, "learning_rate": 4.999420107615935e-06, "loss": 0.16042551398277283, "step": 287 }, { "epoch": 0.07476635514018691, "grad_norm": 1.5225114822387695, "learning_rate": 4.9993984322204905e-06, "loss": 0.16898572444915771, "step": 288 }, { "epoch": 0.07502596053997923, "grad_norm": 1.5272427797317505, "learning_rate": 4.999376359182906e-06, "loss": 0.18143364787101746, "step": 289 }, { "epoch": 0.07528556593977155, "grad_norm": 1.3915812969207764, "learning_rate": 4.999353888507086e-06, "loss": 0.15317711234092712, "step": 290 }, { "epoch": 0.07554517133956386, "grad_norm": 0.8917554616928101, "learning_rate": 4.999331020197002e-06, "loss": 0.17853546142578125, "step": 291 }, { "epoch": 0.07580477673935618, "grad_norm": 1.2961158752441406, "learning_rate": 4.999307754256696e-06, "loss": 0.20328715443611145, "step": 292 }, { "epoch": 0.0760643821391485, "grad_norm": 0.8439105749130249, "learning_rate": 4.999284090690282e-06, "loss": 0.15262554585933685, "step": 293 }, { "epoch": 0.0763239875389408, "grad_norm": 2.150160074234009, "learning_rate": 4.999260029501942e-06, "loss": 0.21386441588401794, "step": 294 }, { "epoch": 0.07658359293873312, "grad_norm": 1.708657145500183, "learning_rate": 4.999235570695931e-06, "loss": 0.15788543224334717, "step": 295 }, { "epoch": 0.07684319833852545, "grad_norm": 1.740920066833496, "learning_rate": 4.9992107142765715e-06, "loss": 0.1549152284860611, "step": 296 }, { "epoch": 0.07710280373831775, "grad_norm": 3.39536452293396, "learning_rate": 4.999185460248257e-06, "loss": 0.18172568082809448, "step": 297 }, { "epoch": 0.07736240913811007, "grad_norm": 1.1200584173202515, "learning_rate": 4.999159808615453e-06, "loss": 0.14479994773864746, "step": 298 }, { "epoch": 0.07762201453790239, "grad_norm": 4.164766311645508, "learning_rate": 4.999133759382693e-06, "loss": 0.20624883472919464, "step": 299 }, { "epoch": 0.0778816199376947, "grad_norm": 0.6203001737594604, "learning_rate": 4.999107312554583e-06, "loss": 0.15695782005786896, "step": 300 }, { "epoch": 0.07814122533748702, "grad_norm": 3.0244314670562744, "learning_rate": 4.9990804681358e-06, "loss": 0.1825091540813446, "step": 301 }, { "epoch": 0.07840083073727934, "grad_norm": 3.2826666831970215, "learning_rate": 4.999053226131084e-06, "loss": 0.2306666374206543, "step": 302 }, { "epoch": 0.07866043613707165, "grad_norm": 2.369818687438965, "learning_rate": 4.9990255865452565e-06, "loss": 0.17063847184181213, "step": 303 }, { "epoch": 0.07892004153686397, "grad_norm": 3.233860492706299, "learning_rate": 4.9989975493832005e-06, "loss": 0.17018012702465057, "step": 304 }, { "epoch": 0.07917964693665629, "grad_norm": 2.2156684398651123, "learning_rate": 4.998969114649873e-06, "loss": 0.20863673090934753, "step": 305 }, { "epoch": 0.0794392523364486, "grad_norm": 1.9461263418197632, "learning_rate": 4.9989402823503006e-06, "loss": 0.1911585032939911, "step": 306 }, { "epoch": 0.07969885773624091, "grad_norm": 1.6010572910308838, "learning_rate": 4.9989110524895795e-06, "loss": 0.1844034492969513, "step": 307 }, { "epoch": 0.07995846313603323, "grad_norm": 2.3548054695129395, "learning_rate": 4.998881425072879e-06, "loss": 0.19331872463226318, "step": 308 }, { "epoch": 0.08021806853582554, "grad_norm": 0.46922194957733154, "learning_rate": 4.998851400105433e-06, "loss": 0.15338706970214844, "step": 309 }, { "epoch": 0.08047767393561786, "grad_norm": 2.3520681858062744, "learning_rate": 4.998820977592553e-06, "loss": 0.20604076981544495, "step": 310 }, { "epoch": 0.08073727933541018, "grad_norm": 2.761096954345703, "learning_rate": 4.998790157539615e-06, "loss": 0.18260109424591064, "step": 311 }, { "epoch": 0.08099688473520249, "grad_norm": 2.75929856300354, "learning_rate": 4.998758939952068e-06, "loss": 0.2246493697166443, "step": 312 }, { "epoch": 0.08125649013499481, "grad_norm": 1.7571094036102295, "learning_rate": 4.998727324835429e-06, "loss": 0.20322562754154205, "step": 313 }, { "epoch": 0.08151609553478713, "grad_norm": 1.2335069179534912, "learning_rate": 4.998695312195289e-06, "loss": 0.17093560099601746, "step": 314 }, { "epoch": 0.08177570093457943, "grad_norm": 3.7284271717071533, "learning_rate": 4.998662902037306e-06, "loss": 0.1746000051498413, "step": 315 }, { "epoch": 0.08203530633437175, "grad_norm": 1.9906103610992432, "learning_rate": 4.998630094367209e-06, "loss": 0.17201194167137146, "step": 316 }, { "epoch": 0.08229491173416408, "grad_norm": 4.264441013336182, "learning_rate": 4.998596889190799e-06, "loss": 0.1724981665611267, "step": 317 }, { "epoch": 0.08255451713395638, "grad_norm": 1.555275321006775, "learning_rate": 4.998563286513944e-06, "loss": 0.16188400983810425, "step": 318 }, { "epoch": 0.0828141225337487, "grad_norm": 1.9127650260925293, "learning_rate": 4.998529286342585e-06, "loss": 0.19477254152297974, "step": 319 }, { "epoch": 0.08307372793354102, "grad_norm": 4.789246082305908, "learning_rate": 4.998494888682734e-06, "loss": 0.2411321997642517, "step": 320 }, { "epoch": 0.08333333333333333, "grad_norm": 1.950197696685791, "learning_rate": 4.99846009354047e-06, "loss": 0.17852216958999634, "step": 321 }, { "epoch": 0.08359293873312565, "grad_norm": 1.8324406147003174, "learning_rate": 4.998424900921944e-06, "loss": 0.20578032732009888, "step": 322 }, { "epoch": 0.08385254413291797, "grad_norm": 3.9196090698242188, "learning_rate": 4.998389310833378e-06, "loss": 0.194352924823761, "step": 323 }, { "epoch": 0.08411214953271028, "grad_norm": 3.5510506629943848, "learning_rate": 4.998353323281063e-06, "loss": 0.16696369647979736, "step": 324 }, { "epoch": 0.0843717549325026, "grad_norm": 1.321696400642395, "learning_rate": 4.998316938271361e-06, "loss": 0.17207570374011993, "step": 325 }, { "epoch": 0.08463136033229492, "grad_norm": 2.935328245162964, "learning_rate": 4.998280155810704e-06, "loss": 0.18748261034488678, "step": 326 }, { "epoch": 0.08489096573208722, "grad_norm": 1.973097801208496, "learning_rate": 4.998242975905593e-06, "loss": 0.1999405026435852, "step": 327 }, { "epoch": 0.08515057113187954, "grad_norm": 2.9850378036499023, "learning_rate": 4.998205398562603e-06, "loss": 0.19545945525169373, "step": 328 }, { "epoch": 0.08541017653167186, "grad_norm": 0.98491370677948, "learning_rate": 4.998167423788375e-06, "loss": 0.15914520621299744, "step": 329 }, { "epoch": 0.08566978193146417, "grad_norm": 4.047921180725098, "learning_rate": 4.998129051589624e-06, "loss": 0.22229208052158356, "step": 330 }, { "epoch": 0.08592938733125649, "grad_norm": 0.910455584526062, "learning_rate": 4.998090281973131e-06, "loss": 0.1681692898273468, "step": 331 }, { "epoch": 0.08618899273104881, "grad_norm": 0.7569683194160461, "learning_rate": 4.99805111494575e-06, "loss": 0.16686934232711792, "step": 332 }, { "epoch": 0.08644859813084112, "grad_norm": 2.2894601821899414, "learning_rate": 4.9980115505144065e-06, "loss": 0.18938380479812622, "step": 333 }, { "epoch": 0.08670820353063344, "grad_norm": 1.845862627029419, "learning_rate": 4.997971588686093e-06, "loss": 0.1625329852104187, "step": 334 }, { "epoch": 0.08696780893042576, "grad_norm": 1.7390483617782593, "learning_rate": 4.9979312294678745e-06, "loss": 0.19616277515888214, "step": 335 }, { "epoch": 0.08722741433021806, "grad_norm": 1.8814654350280762, "learning_rate": 4.997890472866885e-06, "loss": 0.18886813521385193, "step": 336 }, { "epoch": 0.08748701973001038, "grad_norm": 3.8286986351013184, "learning_rate": 4.99784931889033e-06, "loss": 0.20531997084617615, "step": 337 }, { "epoch": 0.0877466251298027, "grad_norm": 0.8921045660972595, "learning_rate": 4.997807767545485e-06, "loss": 0.15921656787395477, "step": 338 }, { "epoch": 0.08800623052959501, "grad_norm": 2.5067601203918457, "learning_rate": 4.9977658188396935e-06, "loss": 0.16586771607398987, "step": 339 }, { "epoch": 0.08826583592938733, "grad_norm": 2.933685541152954, "learning_rate": 4.997723472780373e-06, "loss": 0.15885913372039795, "step": 340 }, { "epoch": 0.08852544132917965, "grad_norm": 1.3945153951644897, "learning_rate": 4.997680729375008e-06, "loss": 0.15230810642242432, "step": 341 }, { "epoch": 0.08878504672897196, "grad_norm": 2.4726996421813965, "learning_rate": 4.997637588631156e-06, "loss": 0.20512153208255768, "step": 342 }, { "epoch": 0.08904465212876428, "grad_norm": 4.750267505645752, "learning_rate": 4.997594050556443e-06, "loss": 0.16432812809944153, "step": 343 }, { "epoch": 0.0893042575285566, "grad_norm": 1.0634821653366089, "learning_rate": 4.9975501151585635e-06, "loss": 0.18294969201087952, "step": 344 }, { "epoch": 0.0895638629283489, "grad_norm": 0.7410452961921692, "learning_rate": 4.997505782445286e-06, "loss": 0.1633058786392212, "step": 345 }, { "epoch": 0.08982346832814123, "grad_norm": 3.0469634532928467, "learning_rate": 4.997461052424447e-06, "loss": 0.1907552182674408, "step": 346 }, { "epoch": 0.09008307372793355, "grad_norm": 1.2415013313293457, "learning_rate": 4.997415925103955e-06, "loss": 0.18752774596214294, "step": 347 }, { "epoch": 0.09034267912772585, "grad_norm": 1.6022688150405884, "learning_rate": 4.997370400491786e-06, "loss": 0.1749478578567505, "step": 348 }, { "epoch": 0.09060228452751817, "grad_norm": 1.7512773275375366, "learning_rate": 4.997324478595987e-06, "loss": 0.18295356631278992, "step": 349 }, { "epoch": 0.0908618899273105, "grad_norm": 1.2197771072387695, "learning_rate": 4.997278159424678e-06, "loss": 0.17754769325256348, "step": 350 }, { "epoch": 0.0911214953271028, "grad_norm": 1.9748235940933228, "learning_rate": 4.997231442986047e-06, "loss": 0.19867777824401855, "step": 351 }, { "epoch": 0.09138110072689512, "grad_norm": 1.9493064880371094, "learning_rate": 4.997184329288351e-06, "loss": 0.20339509844779968, "step": 352 }, { "epoch": 0.09164070612668744, "grad_norm": 2.2198758125305176, "learning_rate": 4.99713681833992e-06, "loss": 0.19493775069713593, "step": 353 }, { "epoch": 0.09190031152647975, "grad_norm": 1.1959518194198608, "learning_rate": 4.99708891014915e-06, "loss": 0.18669244647026062, "step": 354 }, { "epoch": 0.09215991692627207, "grad_norm": 2.2815096378326416, "learning_rate": 4.997040604724514e-06, "loss": 0.17067325115203857, "step": 355 }, { "epoch": 0.09241952232606439, "grad_norm": 1.4772506952285767, "learning_rate": 4.9969919020745485e-06, "loss": 0.15304803848266602, "step": 356 }, { "epoch": 0.0926791277258567, "grad_norm": 1.126404047012329, "learning_rate": 4.996942802207866e-06, "loss": 0.15573087334632874, "step": 357 }, { "epoch": 0.09293873312564901, "grad_norm": 1.9615408182144165, "learning_rate": 4.996893305133142e-06, "loss": 0.19558468461036682, "step": 358 }, { "epoch": 0.09319833852544133, "grad_norm": 0.7240015864372253, "learning_rate": 4.996843410859129e-06, "loss": 0.15029317140579224, "step": 359 }, { "epoch": 0.09345794392523364, "grad_norm": 0.4342077076435089, "learning_rate": 4.996793119394648e-06, "loss": 0.15007835626602173, "step": 360 }, { "epoch": 0.09371754932502596, "grad_norm": 0.5805589556694031, "learning_rate": 4.996742430748587e-06, "loss": 0.15383051335811615, "step": 361 }, { "epoch": 0.09397715472481828, "grad_norm": 2.1951215267181396, "learning_rate": 4.996691344929909e-06, "loss": 0.24621374905109406, "step": 362 }, { "epoch": 0.09423676012461059, "grad_norm": 4.971012592315674, "learning_rate": 4.996639861947643e-06, "loss": 0.19408434629440308, "step": 363 }, { "epoch": 0.09449636552440291, "grad_norm": 2.745206832885742, "learning_rate": 4.99658798181089e-06, "loss": 0.1960049271583557, "step": 364 }, { "epoch": 0.09475597092419523, "grad_norm": 2.2209579944610596, "learning_rate": 4.996535704528823e-06, "loss": 0.17903532087802887, "step": 365 }, { "epoch": 0.09501557632398754, "grad_norm": 0.7542111277580261, "learning_rate": 4.996483030110682e-06, "loss": 0.18912142515182495, "step": 366 }, { "epoch": 0.09527518172377986, "grad_norm": 0.9713164567947388, "learning_rate": 4.996429958565779e-06, "loss": 0.15760484337806702, "step": 367 }, { "epoch": 0.09553478712357218, "grad_norm": 1.5352816581726074, "learning_rate": 4.996376489903495e-06, "loss": 0.18304185569286346, "step": 368 }, { "epoch": 0.09579439252336448, "grad_norm": 2.4729881286621094, "learning_rate": 4.996322624133284e-06, "loss": 0.17207801342010498, "step": 369 }, { "epoch": 0.0960539979231568, "grad_norm": 1.9092621803283691, "learning_rate": 4.996268361264666e-06, "loss": 0.16146692633628845, "step": 370 }, { "epoch": 0.09631360332294912, "grad_norm": 0.6623370051383972, "learning_rate": 4.996213701307235e-06, "loss": 0.16024872660636902, "step": 371 }, { "epoch": 0.09657320872274143, "grad_norm": 3.387369155883789, "learning_rate": 4.9961586442706525e-06, "loss": 0.22287172079086304, "step": 372 }, { "epoch": 0.09683281412253375, "grad_norm": 2.6381685733795166, "learning_rate": 4.9961031901646524e-06, "loss": 0.19975891709327698, "step": 373 }, { "epoch": 0.09709241952232607, "grad_norm": 2.5971789360046387, "learning_rate": 4.996047338999037e-06, "loss": 0.18883627653121948, "step": 374 }, { "epoch": 0.09735202492211838, "grad_norm": 3.601688861846924, "learning_rate": 4.99599109078368e-06, "loss": 0.16807964444160461, "step": 375 }, { "epoch": 0.0976116303219107, "grad_norm": 0.8405979871749878, "learning_rate": 4.995934445528524e-06, "loss": 0.1572370082139969, "step": 376 }, { "epoch": 0.09787123572170302, "grad_norm": 1.0406110286712646, "learning_rate": 4.995877403243584e-06, "loss": 0.17433175444602966, "step": 377 }, { "epoch": 0.09813084112149532, "grad_norm": 2.9505534172058105, "learning_rate": 4.995819963938942e-06, "loss": 0.17099004983901978, "step": 378 }, { "epoch": 0.09839044652128764, "grad_norm": 4.094715595245361, "learning_rate": 4.995762127624753e-06, "loss": 0.16198748350143433, "step": 379 }, { "epoch": 0.09865005192107996, "grad_norm": 0.8391990065574646, "learning_rate": 4.995703894311242e-06, "loss": 0.1596495360136032, "step": 380 }, { "epoch": 0.09890965732087227, "grad_norm": 2.4287121295928955, "learning_rate": 4.995645264008701e-06, "loss": 0.20152731239795685, "step": 381 }, { "epoch": 0.09916926272066459, "grad_norm": 3.2036335468292236, "learning_rate": 4.995586236727496e-06, "loss": 0.1866753101348877, "step": 382 }, { "epoch": 0.09942886812045691, "grad_norm": 2.6019299030303955, "learning_rate": 4.995526812478061e-06, "loss": 0.23451721668243408, "step": 383 }, { "epoch": 0.09968847352024922, "grad_norm": 1.4292573928833008, "learning_rate": 4.995466991270903e-06, "loss": 0.15471509099006653, "step": 384 }, { "epoch": 0.09994807892004154, "grad_norm": 1.3146803379058838, "learning_rate": 4.9954067731165945e-06, "loss": 0.15896067023277283, "step": 385 }, { "epoch": 0.10020768431983386, "grad_norm": 3.679997444152832, "learning_rate": 4.995346158025781e-06, "loss": 0.16012118756771088, "step": 386 }, { "epoch": 0.10046728971962617, "grad_norm": 2.5322837829589844, "learning_rate": 4.995285146009179e-06, "loss": 0.16544455289840698, "step": 387 }, { "epoch": 0.10072689511941849, "grad_norm": 1.585534691810608, "learning_rate": 4.995223737077573e-06, "loss": 0.21508359909057617, "step": 388 }, { "epoch": 0.1009865005192108, "grad_norm": 3.8911490440368652, "learning_rate": 4.9951619312418196e-06, "loss": 0.17229801416397095, "step": 389 }, { "epoch": 0.10124610591900311, "grad_norm": 3.186945915222168, "learning_rate": 4.995099728512843e-06, "loss": 0.17726702988147736, "step": 390 }, { "epoch": 0.10150571131879543, "grad_norm": 1.278238296508789, "learning_rate": 4.995037128901642e-06, "loss": 0.1679590493440628, "step": 391 }, { "epoch": 0.10176531671858775, "grad_norm": 1.241910696029663, "learning_rate": 4.99497413241928e-06, "loss": 0.16369350254535675, "step": 392 }, { "epoch": 0.10202492211838006, "grad_norm": 5.098891735076904, "learning_rate": 4.994910739076895e-06, "loss": 0.2340557873249054, "step": 393 }, { "epoch": 0.10228452751817238, "grad_norm": 0.8959225416183472, "learning_rate": 4.994846948885693e-06, "loss": 0.17338421940803528, "step": 394 }, { "epoch": 0.1025441329179647, "grad_norm": 1.947345495223999, "learning_rate": 4.99478276185695e-06, "loss": 0.18999962508678436, "step": 395 }, { "epoch": 0.102803738317757, "grad_norm": 3.426675796508789, "learning_rate": 4.994718178002013e-06, "loss": 0.15233194828033447, "step": 396 }, { "epoch": 0.10306334371754933, "grad_norm": 3.699669599533081, "learning_rate": 4.9946531973323005e-06, "loss": 0.16114100813865662, "step": 397 }, { "epoch": 0.10332294911734165, "grad_norm": 0.40241050720214844, "learning_rate": 4.9945878198592975e-06, "loss": 0.15213997662067413, "step": 398 }, { "epoch": 0.10358255451713395, "grad_norm": 1.7305138111114502, "learning_rate": 4.994522045594563e-06, "loss": 0.18931090831756592, "step": 399 }, { "epoch": 0.10384215991692627, "grad_norm": 3.0147111415863037, "learning_rate": 4.9944558745497236e-06, "loss": 0.17924146354198456, "step": 400 }, { "epoch": 0.1041017653167186, "grad_norm": 3.66682767868042, "learning_rate": 4.9943893067364765e-06, "loss": 0.1814359873533249, "step": 401 }, { "epoch": 0.1043613707165109, "grad_norm": 0.9630410671234131, "learning_rate": 4.99432234216659e-06, "loss": 0.18316501379013062, "step": 402 }, { "epoch": 0.10462097611630322, "grad_norm": 1.3265880346298218, "learning_rate": 4.994254980851901e-06, "loss": 0.16102969646453857, "step": 403 }, { "epoch": 0.10488058151609553, "grad_norm": 3.828901767730713, "learning_rate": 4.994187222804319e-06, "loss": 0.18134915828704834, "step": 404 }, { "epoch": 0.10514018691588785, "grad_norm": 2.737379312515259, "learning_rate": 4.99411906803582e-06, "loss": 0.19231972098350525, "step": 405 }, { "epoch": 0.10539979231568017, "grad_norm": 1.5696314573287964, "learning_rate": 4.994050516558454e-06, "loss": 0.16086545586585999, "step": 406 }, { "epoch": 0.10565939771547248, "grad_norm": 1.9377501010894775, "learning_rate": 4.993981568384339e-06, "loss": 0.16568799316883087, "step": 407 }, { "epoch": 0.1059190031152648, "grad_norm": 1.2523488998413086, "learning_rate": 4.993912223525661e-06, "loss": 0.1814451813697815, "step": 408 }, { "epoch": 0.10617860851505712, "grad_norm": 1.3389956951141357, "learning_rate": 4.993842481994683e-06, "loss": 0.18750634789466858, "step": 409 }, { "epoch": 0.10643821391484942, "grad_norm": 2.5796103477478027, "learning_rate": 4.99377234380373e-06, "loss": 0.190415620803833, "step": 410 }, { "epoch": 0.10669781931464174, "grad_norm": 2.3037898540496826, "learning_rate": 4.9937018089652024e-06, "loss": 0.2186698019504547, "step": 411 }, { "epoch": 0.10695742471443406, "grad_norm": 1.828710913658142, "learning_rate": 4.99363087749157e-06, "loss": 0.1938273012638092, "step": 412 }, { "epoch": 0.10721703011422637, "grad_norm": 3.5210835933685303, "learning_rate": 4.993559549395369e-06, "loss": 0.24912090599536896, "step": 413 }, { "epoch": 0.10747663551401869, "grad_norm": 1.1090534925460815, "learning_rate": 4.993487824689211e-06, "loss": 0.1614549160003662, "step": 414 }, { "epoch": 0.10773624091381101, "grad_norm": 1.260056495666504, "learning_rate": 4.993415703385776e-06, "loss": 0.15153534710407257, "step": 415 }, { "epoch": 0.10799584631360332, "grad_norm": 1.6090062856674194, "learning_rate": 4.99334318549781e-06, "loss": 0.15843307971954346, "step": 416 }, { "epoch": 0.10825545171339564, "grad_norm": 1.6943066120147705, "learning_rate": 4.993270271038136e-06, "loss": 0.15748250484466553, "step": 417 }, { "epoch": 0.10851505711318796, "grad_norm": 0.770469069480896, "learning_rate": 4.993196960019642e-06, "loss": 0.16301998496055603, "step": 418 }, { "epoch": 0.10877466251298026, "grad_norm": 1.5060933828353882, "learning_rate": 4.993123252455288e-06, "loss": 0.19966992735862732, "step": 419 }, { "epoch": 0.10903426791277258, "grad_norm": 1.8146758079528809, "learning_rate": 4.993049148358103e-06, "loss": 0.16177308559417725, "step": 420 }, { "epoch": 0.1092938733125649, "grad_norm": 1.2079815864562988, "learning_rate": 4.992974647741188e-06, "loss": 0.15618294477462769, "step": 421 }, { "epoch": 0.10955347871235721, "grad_norm": 1.7881373167037964, "learning_rate": 4.992899750617712e-06, "loss": 0.2157191038131714, "step": 422 }, { "epoch": 0.10981308411214953, "grad_norm": 1.0063027143478394, "learning_rate": 4.9928244570009165e-06, "loss": 0.1703258603811264, "step": 423 }, { "epoch": 0.11007268951194185, "grad_norm": 2.4637162685394287, "learning_rate": 4.99274876690411e-06, "loss": 0.16744044423103333, "step": 424 }, { "epoch": 0.11033229491173416, "grad_norm": 0.8924534916877747, "learning_rate": 4.992672680340675e-06, "loss": 0.15842998027801514, "step": 425 }, { "epoch": 0.11059190031152648, "grad_norm": 2.2273435592651367, "learning_rate": 4.9925961973240595e-06, "loss": 0.15087661147117615, "step": 426 }, { "epoch": 0.1108515057113188, "grad_norm": 2.3019683361053467, "learning_rate": 4.992519317867784e-06, "loss": 0.163986474275589, "step": 427 }, { "epoch": 0.1111111111111111, "grad_norm": 2.908012866973877, "learning_rate": 4.992442041985442e-06, "loss": 0.18207699060440063, "step": 428 }, { "epoch": 0.11137071651090343, "grad_norm": 0.7466005682945251, "learning_rate": 4.9923643696906915e-06, "loss": 0.14864462614059448, "step": 429 }, { "epoch": 0.11163032191069575, "grad_norm": 2.5377533435821533, "learning_rate": 4.992286300997263e-06, "loss": 0.1674596071243286, "step": 430 }, { "epoch": 0.11188992731048805, "grad_norm": 2.0468080043792725, "learning_rate": 4.992207835918959e-06, "loss": 0.18814846873283386, "step": 431 }, { "epoch": 0.11214953271028037, "grad_norm": 1.0491740703582764, "learning_rate": 4.99212897446965e-06, "loss": 0.15429383516311646, "step": 432 }, { "epoch": 0.11240913811007269, "grad_norm": 3.1393771171569824, "learning_rate": 4.992049716663274e-06, "loss": 0.16078828275203705, "step": 433 }, { "epoch": 0.112668743509865, "grad_norm": 1.5573022365570068, "learning_rate": 4.991970062513845e-06, "loss": 0.15140685439109802, "step": 434 }, { "epoch": 0.11292834890965732, "grad_norm": 0.7577384114265442, "learning_rate": 4.991890012035444e-06, "loss": 0.20291520655155182, "step": 435 }, { "epoch": 0.11318795430944964, "grad_norm": 1.317232370376587, "learning_rate": 4.991809565242221e-06, "loss": 0.19210493564605713, "step": 436 }, { "epoch": 0.11344755970924195, "grad_norm": 5.1076860427856445, "learning_rate": 4.991728722148398e-06, "loss": 0.18398699164390564, "step": 437 }, { "epoch": 0.11370716510903427, "grad_norm": 0.8385660648345947, "learning_rate": 4.991647482768265e-06, "loss": 0.20522961020469666, "step": 438 }, { "epoch": 0.11396677050882659, "grad_norm": 1.1977462768554688, "learning_rate": 4.991565847116184e-06, "loss": 0.18864119052886963, "step": 439 }, { "epoch": 0.1142263759086189, "grad_norm": 1.2893003225326538, "learning_rate": 4.991483815206586e-06, "loss": 0.1481683850288391, "step": 440 }, { "epoch": 0.11448598130841121, "grad_norm": 1.1188980340957642, "learning_rate": 4.991401387053973e-06, "loss": 0.15223604440689087, "step": 441 }, { "epoch": 0.11474558670820353, "grad_norm": 0.7941104769706726, "learning_rate": 4.991318562672917e-06, "loss": 0.1660514771938324, "step": 442 }, { "epoch": 0.11500519210799584, "grad_norm": 1.7618638277053833, "learning_rate": 4.991235342078058e-06, "loss": 0.17179816961288452, "step": 443 }, { "epoch": 0.11526479750778816, "grad_norm": 0.9701156616210938, "learning_rate": 4.9911517252841066e-06, "loss": 0.15271081030368805, "step": 444 }, { "epoch": 0.11552440290758048, "grad_norm": 3.3901991844177246, "learning_rate": 4.991067712305848e-06, "loss": 0.24629855155944824, "step": 445 }, { "epoch": 0.11578400830737279, "grad_norm": 1.1900922060012817, "learning_rate": 4.9909833031581294e-06, "loss": 0.1677289605140686, "step": 446 }, { "epoch": 0.11604361370716511, "grad_norm": 1.6841630935668945, "learning_rate": 4.990898497855876e-06, "loss": 0.1848745346069336, "step": 447 }, { "epoch": 0.11630321910695743, "grad_norm": 0.7385988831520081, "learning_rate": 4.990813296414077e-06, "loss": 0.1940811276435852, "step": 448 }, { "epoch": 0.11656282450674973, "grad_norm": 1.5582243204116821, "learning_rate": 4.990727698847796e-06, "loss": 0.19254335761070251, "step": 449 }, { "epoch": 0.11682242990654206, "grad_norm": 0.7071974873542786, "learning_rate": 4.990641705172163e-06, "loss": 0.1828007996082306, "step": 450 }, { "epoch": 0.11708203530633438, "grad_norm": 1.0246379375457764, "learning_rate": 4.99055531540238e-06, "loss": 0.17876391112804413, "step": 451 }, { "epoch": 0.11734164070612668, "grad_norm": 3.2535572052001953, "learning_rate": 4.99046852955372e-06, "loss": 0.18863128125667572, "step": 452 }, { "epoch": 0.117601246105919, "grad_norm": 0.9514186382293701, "learning_rate": 4.990381347641524e-06, "loss": 0.1500735878944397, "step": 453 }, { "epoch": 0.11786085150571132, "grad_norm": 2.4375290870666504, "learning_rate": 4.9902937696812034e-06, "loss": 0.1709064543247223, "step": 454 }, { "epoch": 0.11812045690550363, "grad_norm": 2.3034040927886963, "learning_rate": 4.99020579568824e-06, "loss": 0.1844632625579834, "step": 455 }, { "epoch": 0.11838006230529595, "grad_norm": 1.9610146284103394, "learning_rate": 4.990117425678186e-06, "loss": 0.1647529900074005, "step": 456 }, { "epoch": 0.11863966770508827, "grad_norm": 0.7715980410575867, "learning_rate": 4.990028659666663e-06, "loss": 0.16200950741767883, "step": 457 }, { "epoch": 0.11889927310488058, "grad_norm": 1.482055902481079, "learning_rate": 4.989939497669363e-06, "loss": 0.16892880201339722, "step": 458 }, { "epoch": 0.1191588785046729, "grad_norm": 4.139158725738525, "learning_rate": 4.989849939702047e-06, "loss": 0.20559494197368622, "step": 459 }, { "epoch": 0.11941848390446522, "grad_norm": 1.1442652940750122, "learning_rate": 4.989759985780548e-06, "loss": 0.15800632536411285, "step": 460 }, { "epoch": 0.11967808930425752, "grad_norm": 2.7751054763793945, "learning_rate": 4.989669635920767e-06, "loss": 0.16937527060508728, "step": 461 }, { "epoch": 0.11993769470404984, "grad_norm": 1.9774348735809326, "learning_rate": 4.989578890138676e-06, "loss": 0.14600136876106262, "step": 462 }, { "epoch": 0.12019730010384216, "grad_norm": 0.5517883896827698, "learning_rate": 4.989487748450318e-06, "loss": 0.15089866518974304, "step": 463 }, { "epoch": 0.12045690550363447, "grad_norm": 1.558395266532898, "learning_rate": 4.9893962108718015e-06, "loss": 0.23237788677215576, "step": 464 }, { "epoch": 0.12071651090342679, "grad_norm": 2.3916635513305664, "learning_rate": 4.989304277419311e-06, "loss": 0.21645551919937134, "step": 465 }, { "epoch": 0.12097611630321911, "grad_norm": 2.8366215229034424, "learning_rate": 4.989211948109099e-06, "loss": 0.21311452984809875, "step": 466 }, { "epoch": 0.12123572170301142, "grad_norm": 1.3344024419784546, "learning_rate": 4.989119222957483e-06, "loss": 0.19026297330856323, "step": 467 }, { "epoch": 0.12149532710280374, "grad_norm": 1.3421872854232788, "learning_rate": 4.98902610198086e-06, "loss": 0.2022741436958313, "step": 468 }, { "epoch": 0.12175493250259606, "grad_norm": 3.2266602516174316, "learning_rate": 4.988932585195688e-06, "loss": 0.21775993704795837, "step": 469 }, { "epoch": 0.12201453790238836, "grad_norm": 2.7871296405792236, "learning_rate": 4.988838672618499e-06, "loss": 0.14962852001190186, "step": 470 }, { "epoch": 0.12227414330218069, "grad_norm": 1.3342427015304565, "learning_rate": 4.988744364265896e-06, "loss": 0.1931169629096985, "step": 471 }, { "epoch": 0.122533748701973, "grad_norm": 1.1260013580322266, "learning_rate": 4.988649660154551e-06, "loss": 0.18294070661067963, "step": 472 }, { "epoch": 0.12279335410176531, "grad_norm": 1.955804467201233, "learning_rate": 4.988554560301203e-06, "loss": 0.1770792007446289, "step": 473 }, { "epoch": 0.12305295950155763, "grad_norm": 1.371018886566162, "learning_rate": 4.988459064722664e-06, "loss": 0.13022300601005554, "step": 474 }, { "epoch": 0.12331256490134995, "grad_norm": 1.5875389575958252, "learning_rate": 4.988363173435818e-06, "loss": 0.17934195697307587, "step": 475 }, { "epoch": 0.12357217030114226, "grad_norm": 1.8112467527389526, "learning_rate": 4.9882668864576135e-06, "loss": 0.20139925181865692, "step": 476 }, { "epoch": 0.12383177570093458, "grad_norm": 0.9807649254798889, "learning_rate": 4.988170203805074e-06, "loss": 0.16676658391952515, "step": 477 }, { "epoch": 0.1240913811007269, "grad_norm": 3.461571455001831, "learning_rate": 4.9880731254952885e-06, "loss": 0.15932786464691162, "step": 478 }, { "epoch": 0.1243509865005192, "grad_norm": 2.129676342010498, "learning_rate": 4.987975651545421e-06, "loss": 0.15221357345581055, "step": 479 }, { "epoch": 0.12461059190031153, "grad_norm": 2.2159178256988525, "learning_rate": 4.9878777819727e-06, "loss": 0.1668568253517151, "step": 480 }, { "epoch": 0.12487019730010385, "grad_norm": 2.119088888168335, "learning_rate": 4.9877795167944285e-06, "loss": 0.15890932083129883, "step": 481 }, { "epoch": 0.12512980269989615, "grad_norm": 1.4601703882217407, "learning_rate": 4.987680856027976e-06, "loss": 0.16679087281227112, "step": 482 }, { "epoch": 0.12538940809968846, "grad_norm": 2.1554372310638428, "learning_rate": 4.987581799690785e-06, "loss": 0.18150001764297485, "step": 483 }, { "epoch": 0.1256490134994808, "grad_norm": 2.2414331436157227, "learning_rate": 4.987482347800365e-06, "loss": 0.1492304652929306, "step": 484 }, { "epoch": 0.1259086188992731, "grad_norm": 1.4478155374526978, "learning_rate": 4.987382500374298e-06, "loss": 0.22232219576835632, "step": 485 }, { "epoch": 0.1261682242990654, "grad_norm": 1.1920336484909058, "learning_rate": 4.987282257430234e-06, "loss": 0.18190577626228333, "step": 486 }, { "epoch": 0.12642782969885774, "grad_norm": 1.3720358610153198, "learning_rate": 4.987181618985893e-06, "loss": 0.18941371142864227, "step": 487 }, { "epoch": 0.12668743509865005, "grad_norm": 1.3482160568237305, "learning_rate": 4.987080585059068e-06, "loss": 0.15337859094142914, "step": 488 }, { "epoch": 0.12694704049844235, "grad_norm": 1.0867912769317627, "learning_rate": 4.9869791556676175e-06, "loss": 0.16872936487197876, "step": 489 }, { "epoch": 0.1272066458982347, "grad_norm": 1.4276787042617798, "learning_rate": 4.986877330829472e-06, "loss": 0.15198448300361633, "step": 490 }, { "epoch": 0.127466251298027, "grad_norm": 0.8938367962837219, "learning_rate": 4.986775110562632e-06, "loss": 0.15170371532440186, "step": 491 }, { "epoch": 0.1277258566978193, "grad_norm": 1.8492263555526733, "learning_rate": 4.9866724948851685e-06, "loss": 0.17725881934165955, "step": 492 }, { "epoch": 0.12798546209761164, "grad_norm": 1.5294370651245117, "learning_rate": 4.98656948381522e-06, "loss": 0.16931486129760742, "step": 493 }, { "epoch": 0.12824506749740394, "grad_norm": 2.7005720138549805, "learning_rate": 4.986466077370998e-06, "loss": 0.15254256129264832, "step": 494 }, { "epoch": 0.12850467289719625, "grad_norm": 1.4808390140533447, "learning_rate": 4.9863622755707815e-06, "loss": 0.19736014306545258, "step": 495 }, { "epoch": 0.12876427829698858, "grad_norm": 2.643636465072632, "learning_rate": 4.986258078432921e-06, "loss": 0.16777893900871277, "step": 496 }, { "epoch": 0.1290238836967809, "grad_norm": 2.1636853218078613, "learning_rate": 4.986153485975835e-06, "loss": 0.20697876811027527, "step": 497 }, { "epoch": 0.1292834890965732, "grad_norm": 0.8944142460823059, "learning_rate": 4.986048498218014e-06, "loss": 0.18024033308029175, "step": 498 }, { "epoch": 0.12954309449636553, "grad_norm": 2.935779094696045, "learning_rate": 4.985943115178018e-06, "loss": 0.22316861152648926, "step": 499 }, { "epoch": 0.12980269989615784, "grad_norm": 1.0678836107254028, "learning_rate": 4.985837336874475e-06, "loss": 0.15641672909259796, "step": 500 }, { "epoch": 0.13006230529595014, "grad_norm": 1.3840616941452026, "learning_rate": 4.985731163326084e-06, "loss": 0.20864388346672058, "step": 501 }, { "epoch": 0.13032191069574248, "grad_norm": 0.6540664434432983, "learning_rate": 4.985624594551616e-06, "loss": 0.17237257957458496, "step": 502 }, { "epoch": 0.13058151609553478, "grad_norm": 3.0782887935638428, "learning_rate": 4.985517630569908e-06, "loss": 0.1497366726398468, "step": 503 }, { "epoch": 0.1308411214953271, "grad_norm": 0.9547432661056519, "learning_rate": 4.98541027139987e-06, "loss": 0.1596902310848236, "step": 504 }, { "epoch": 0.13110072689511942, "grad_norm": 2.5124244689941406, "learning_rate": 4.98530251706048e-06, "loss": 0.19264747202396393, "step": 505 }, { "epoch": 0.13136033229491173, "grad_norm": 2.233949661254883, "learning_rate": 4.9851943675707864e-06, "loss": 0.17076881229877472, "step": 506 }, { "epoch": 0.13161993769470404, "grad_norm": 2.12353515625, "learning_rate": 4.985085822949908e-06, "loss": 0.15542131662368774, "step": 507 }, { "epoch": 0.13187954309449637, "grad_norm": 2.686232328414917, "learning_rate": 4.984976883217032e-06, "loss": 0.18339943885803223, "step": 508 }, { "epoch": 0.13213914849428868, "grad_norm": 1.7528703212738037, "learning_rate": 4.984867548391418e-06, "loss": 0.14993128180503845, "step": 509 }, { "epoch": 0.13239875389408098, "grad_norm": 2.3744986057281494, "learning_rate": 4.984757818492393e-06, "loss": 0.15036332607269287, "step": 510 }, { "epoch": 0.13265835929387332, "grad_norm": 1.0924708843231201, "learning_rate": 4.984647693539355e-06, "loss": 0.1430358588695526, "step": 511 }, { "epoch": 0.13291796469366562, "grad_norm": 1.5519070625305176, "learning_rate": 4.984537173551772e-06, "loss": 0.1489488184452057, "step": 512 }, { "epoch": 0.13317757009345793, "grad_norm": 1.615344762802124, "learning_rate": 4.98442625854918e-06, "loss": 0.1841067373752594, "step": 513 }, { "epoch": 0.13343717549325027, "grad_norm": 1.395182728767395, "learning_rate": 4.984314948551187e-06, "loss": 0.16576829552650452, "step": 514 }, { "epoch": 0.13369678089304257, "grad_norm": 2.255201578140259, "learning_rate": 4.98420324357747e-06, "loss": 0.17672592401504517, "step": 515 }, { "epoch": 0.13395638629283488, "grad_norm": 1.7566978931427002, "learning_rate": 4.984091143647777e-06, "loss": 0.2142219990491867, "step": 516 }, { "epoch": 0.1342159916926272, "grad_norm": 0.7054417133331299, "learning_rate": 4.983978648781923e-06, "loss": 0.16283464431762695, "step": 517 }, { "epoch": 0.13447559709241952, "grad_norm": 1.4237998723983765, "learning_rate": 4.983865758999795e-06, "loss": 0.18693232536315918, "step": 518 }, { "epoch": 0.13473520249221183, "grad_norm": 2.623223304748535, "learning_rate": 4.98375247432135e-06, "loss": 0.14377814531326294, "step": 519 }, { "epoch": 0.13499480789200416, "grad_norm": 1.752745270729065, "learning_rate": 4.983638794766613e-06, "loss": 0.16968940198421478, "step": 520 }, { "epoch": 0.13525441329179647, "grad_norm": 1.8884159326553345, "learning_rate": 4.9835247203556815e-06, "loss": 0.19150611758232117, "step": 521 }, { "epoch": 0.13551401869158877, "grad_norm": 2.4943909645080566, "learning_rate": 4.9834102511087205e-06, "loss": 0.1510632038116455, "step": 522 }, { "epoch": 0.1357736240913811, "grad_norm": 2.126210927963257, "learning_rate": 4.983295387045965e-06, "loss": 0.21344305574893951, "step": 523 }, { "epoch": 0.1360332294911734, "grad_norm": 0.6029767990112305, "learning_rate": 4.98318012818772e-06, "loss": 0.1412392258644104, "step": 524 }, { "epoch": 0.13629283489096572, "grad_norm": 1.72752046585083, "learning_rate": 4.983064474554362e-06, "loss": 0.15502937138080597, "step": 525 }, { "epoch": 0.13655244029075805, "grad_norm": 4.7587761878967285, "learning_rate": 4.9829484261663355e-06, "loss": 0.23057889938354492, "step": 526 }, { "epoch": 0.13681204569055036, "grad_norm": 5.253512382507324, "learning_rate": 4.982831983044155e-06, "loss": 0.16890586912631989, "step": 527 }, { "epoch": 0.13707165109034267, "grad_norm": 2.6000313758850098, "learning_rate": 4.982715145208405e-06, "loss": 0.17273780703544617, "step": 528 }, { "epoch": 0.137331256490135, "grad_norm": 3.380070924758911, "learning_rate": 4.982597912679739e-06, "loss": 0.1768185794353485, "step": 529 }, { "epoch": 0.1375908618899273, "grad_norm": 2.682633399963379, "learning_rate": 4.9824802854788825e-06, "loss": 0.19771867990493774, "step": 530 }, { "epoch": 0.1378504672897196, "grad_norm": 1.9633920192718506, "learning_rate": 4.982362263626629e-06, "loss": 0.13835451006889343, "step": 531 }, { "epoch": 0.13811007268951195, "grad_norm": 1.3901094198226929, "learning_rate": 4.9822438471438404e-06, "loss": 0.15863272547721863, "step": 532 }, { "epoch": 0.13836967808930425, "grad_norm": 2.229185104370117, "learning_rate": 4.982125036051452e-06, "loss": 0.15739703178405762, "step": 533 }, { "epoch": 0.13862928348909656, "grad_norm": 3.6495890617370605, "learning_rate": 4.982005830370466e-06, "loss": 0.19511544704437256, "step": 534 }, { "epoch": 0.1388888888888889, "grad_norm": 1.4584171772003174, "learning_rate": 4.981886230121956e-06, "loss": 0.19155752658843994, "step": 535 }, { "epoch": 0.1391484942886812, "grad_norm": 1.6548500061035156, "learning_rate": 4.981766235327062e-06, "loss": 0.16474860906600952, "step": 536 }, { "epoch": 0.1394080996884735, "grad_norm": 1.9790385961532593, "learning_rate": 4.981645846007e-06, "loss": 0.1495848149061203, "step": 537 }, { "epoch": 0.13966770508826584, "grad_norm": 2.0605082511901855, "learning_rate": 4.98152506218305e-06, "loss": 0.19538608193397522, "step": 538 }, { "epoch": 0.13992731048805815, "grad_norm": 0.8614341616630554, "learning_rate": 4.981403883876564e-06, "loss": 0.18678295612335205, "step": 539 }, { "epoch": 0.14018691588785046, "grad_norm": 0.6708563566207886, "learning_rate": 4.981282311108964e-06, "loss": 0.14425048232078552, "step": 540 }, { "epoch": 0.1404465212876428, "grad_norm": 2.223954916000366, "learning_rate": 4.98116034390174e-06, "loss": 0.1972579061985016, "step": 541 }, { "epoch": 0.1407061266874351, "grad_norm": 1.9004911184310913, "learning_rate": 4.981037982276456e-06, "loss": 0.16927972435951233, "step": 542 }, { "epoch": 0.1409657320872274, "grad_norm": 0.6089682579040527, "learning_rate": 4.980915226254739e-06, "loss": 0.1503254473209381, "step": 543 }, { "epoch": 0.14122533748701974, "grad_norm": 1.7349764108657837, "learning_rate": 4.980792075858291e-06, "loss": 0.15929073095321655, "step": 544 }, { "epoch": 0.14148494288681204, "grad_norm": 1.7705774307250977, "learning_rate": 4.9806685311088834e-06, "loss": 0.165647491812706, "step": 545 }, { "epoch": 0.14174454828660435, "grad_norm": 0.6757422089576721, "learning_rate": 4.980544592028354e-06, "loss": 0.1515163779258728, "step": 546 }, { "epoch": 0.14200415368639668, "grad_norm": 0.7679604291915894, "learning_rate": 4.980420258638615e-06, "loss": 0.1553967446088791, "step": 547 }, { "epoch": 0.142263759086189, "grad_norm": 1.532784342765808, "learning_rate": 4.980295530961643e-06, "loss": 0.17359596490859985, "step": 548 }, { "epoch": 0.1425233644859813, "grad_norm": 0.7515537142753601, "learning_rate": 4.980170409019488e-06, "loss": 0.15910449624061584, "step": 549 }, { "epoch": 0.14278296988577363, "grad_norm": 0.6498897075653076, "learning_rate": 4.980044892834269e-06, "loss": 0.14085766673088074, "step": 550 }, { "epoch": 0.14304257528556594, "grad_norm": 1.8213180303573608, "learning_rate": 4.979918982428174e-06, "loss": 0.17440257966518402, "step": 551 }, { "epoch": 0.14330218068535824, "grad_norm": 0.6420713067054749, "learning_rate": 4.9797926778234615e-06, "loss": 0.1600932776927948, "step": 552 }, { "epoch": 0.14356178608515058, "grad_norm": 0.5839775800704956, "learning_rate": 4.9796659790424595e-06, "loss": 0.14923636615276337, "step": 553 }, { "epoch": 0.14382139148494288, "grad_norm": 0.5377679467201233, "learning_rate": 4.979538886107565e-06, "loss": 0.14824774861335754, "step": 554 }, { "epoch": 0.1440809968847352, "grad_norm": 1.5610138177871704, "learning_rate": 4.979411399041244e-06, "loss": 0.15654608607292175, "step": 555 }, { "epoch": 0.14434060228452752, "grad_norm": 1.9398846626281738, "learning_rate": 4.979283517866036e-06, "loss": 0.1630793660879135, "step": 556 }, { "epoch": 0.14460020768431983, "grad_norm": 1.4422266483306885, "learning_rate": 4.979155242604545e-06, "loss": 0.1796771138906479, "step": 557 }, { "epoch": 0.14485981308411214, "grad_norm": 1.11147141456604, "learning_rate": 4.979026573279447e-06, "loss": 0.1794356107711792, "step": 558 }, { "epoch": 0.14511941848390447, "grad_norm": 1.7719473838806152, "learning_rate": 4.97889750991349e-06, "loss": 0.14767305552959442, "step": 559 }, { "epoch": 0.14537902388369678, "grad_norm": 0.8037489056587219, "learning_rate": 4.978768052529488e-06, "loss": 0.17820659279823303, "step": 560 }, { "epoch": 0.14563862928348908, "grad_norm": 1.9555473327636719, "learning_rate": 4.978638201150327e-06, "loss": 0.16431480646133423, "step": 561 }, { "epoch": 0.14589823468328142, "grad_norm": 0.75835120677948, "learning_rate": 4.9785079557989605e-06, "loss": 0.1678251326084137, "step": 562 }, { "epoch": 0.14615784008307373, "grad_norm": 1.3284589052200317, "learning_rate": 4.978377316498414e-06, "loss": 0.1767311990261078, "step": 563 }, { "epoch": 0.14641744548286603, "grad_norm": 1.1839298009872437, "learning_rate": 4.97824628327178e-06, "loss": 0.1454414576292038, "step": 564 }, { "epoch": 0.14667705088265837, "grad_norm": 0.4038936197757721, "learning_rate": 4.978114856142224e-06, "loss": 0.15228790044784546, "step": 565 }, { "epoch": 0.14693665628245067, "grad_norm": 1.031172513961792, "learning_rate": 4.977983035132977e-06, "loss": 0.15999209880828857, "step": 566 }, { "epoch": 0.14719626168224298, "grad_norm": 1.2716104984283447, "learning_rate": 4.977850820267346e-06, "loss": 0.16585978865623474, "step": 567 }, { "epoch": 0.1474558670820353, "grad_norm": 2.3140926361083984, "learning_rate": 4.9777182115686995e-06, "loss": 0.1493939459323883, "step": 568 }, { "epoch": 0.14771547248182762, "grad_norm": 2.4499082565307617, "learning_rate": 4.977585209060481e-06, "loss": 0.17849931120872498, "step": 569 }, { "epoch": 0.14797507788161993, "grad_norm": 2.520435333251953, "learning_rate": 4.977451812766203e-06, "loss": 0.20266413688659668, "step": 570 }, { "epoch": 0.14823468328141226, "grad_norm": 1.9199259281158447, "learning_rate": 4.977318022709445e-06, "loss": 0.20884713530540466, "step": 571 }, { "epoch": 0.14849428868120457, "grad_norm": 3.6236987113952637, "learning_rate": 4.977183838913862e-06, "loss": 0.15809249877929688, "step": 572 }, { "epoch": 0.14875389408099687, "grad_norm": 3.910932779312134, "learning_rate": 4.97704926140317e-06, "loss": 0.17319542169570923, "step": 573 }, { "epoch": 0.1490134994807892, "grad_norm": 0.8318435549736023, "learning_rate": 4.9769142902011615e-06, "loss": 0.17286643385887146, "step": 574 }, { "epoch": 0.14927310488058151, "grad_norm": 2.4117934703826904, "learning_rate": 4.9767789253316964e-06, "loss": 0.19541087746620178, "step": 575 }, { "epoch": 0.14953271028037382, "grad_norm": 3.995985269546509, "learning_rate": 4.976643166818704e-06, "loss": 0.1625194400548935, "step": 576 }, { "epoch": 0.14979231568016615, "grad_norm": 2.6337244510650635, "learning_rate": 4.976507014686183e-06, "loss": 0.20989376306533813, "step": 577 }, { "epoch": 0.15005192107995846, "grad_norm": 1.102306842803955, "learning_rate": 4.976370468958201e-06, "loss": 0.1507343351840973, "step": 578 }, { "epoch": 0.15031152647975077, "grad_norm": 3.2403523921966553, "learning_rate": 4.976233529658898e-06, "loss": 0.15285080671310425, "step": 579 }, { "epoch": 0.1505711318795431, "grad_norm": 2.604095935821533, "learning_rate": 4.9760961968124815e-06, "loss": 0.17442092299461365, "step": 580 }, { "epoch": 0.1508307372793354, "grad_norm": 0.8524278998374939, "learning_rate": 4.975958470443228e-06, "loss": 0.1448076367378235, "step": 581 }, { "epoch": 0.15109034267912771, "grad_norm": 0.8294644355773926, "learning_rate": 4.9758203505754845e-06, "loss": 0.17929545044898987, "step": 582 }, { "epoch": 0.15134994807892005, "grad_norm": 1.0414146184921265, "learning_rate": 4.975681837233668e-06, "loss": 0.13980203866958618, "step": 583 }, { "epoch": 0.15160955347871236, "grad_norm": 1.1341125965118408, "learning_rate": 4.975542930442263e-06, "loss": 0.17835792899131775, "step": 584 }, { "epoch": 0.15186915887850466, "grad_norm": 1.1226351261138916, "learning_rate": 4.975403630225827e-06, "loss": 0.19428375363349915, "step": 585 }, { "epoch": 0.152128764278297, "grad_norm": 1.2467403411865234, "learning_rate": 4.975263936608984e-06, "loss": 0.13102276623249054, "step": 586 }, { "epoch": 0.1523883696780893, "grad_norm": 0.8722458481788635, "learning_rate": 4.975123849616429e-06, "loss": 0.1512024700641632, "step": 587 }, { "epoch": 0.1526479750778816, "grad_norm": 1.515582799911499, "learning_rate": 4.974983369272925e-06, "loss": 0.1463165581226349, "step": 588 }, { "epoch": 0.15290758047767394, "grad_norm": 3.1456730365753174, "learning_rate": 4.9748424956033076e-06, "loss": 0.1883867383003235, "step": 589 }, { "epoch": 0.15316718587746625, "grad_norm": 1.0460714101791382, "learning_rate": 4.9747012286324784e-06, "loss": 0.1405334770679474, "step": 590 }, { "epoch": 0.15342679127725856, "grad_norm": 0.6841707825660706, "learning_rate": 4.9745595683854115e-06, "loss": 0.14833387732505798, "step": 591 }, { "epoch": 0.1536863966770509, "grad_norm": 2.5054755210876465, "learning_rate": 4.974417514887148e-06, "loss": 0.17717963457107544, "step": 592 }, { "epoch": 0.1539460020768432, "grad_norm": 1.8647609949111938, "learning_rate": 4.9742750681628e-06, "loss": 0.17876973748207092, "step": 593 }, { "epoch": 0.1542056074766355, "grad_norm": 1.0027137994766235, "learning_rate": 4.97413222823755e-06, "loss": 0.14866608381271362, "step": 594 }, { "epoch": 0.15446521287642784, "grad_norm": 2.425513744354248, "learning_rate": 4.973988995136648e-06, "loss": 0.17931903898715973, "step": 595 }, { "epoch": 0.15472481827622014, "grad_norm": 1.2841705083847046, "learning_rate": 4.9738453688854145e-06, "loss": 0.15591634809970856, "step": 596 }, { "epoch": 0.15498442367601245, "grad_norm": 2.211038827896118, "learning_rate": 4.973701349509238e-06, "loss": 0.17172525823116302, "step": 597 }, { "epoch": 0.15524402907580478, "grad_norm": 0.5956721305847168, "learning_rate": 4.973556937033579e-06, "loss": 0.12932434678077698, "step": 598 }, { "epoch": 0.1555036344755971, "grad_norm": 1.174949049949646, "learning_rate": 4.973412131483966e-06, "loss": 0.14718645811080933, "step": 599 }, { "epoch": 0.1557632398753894, "grad_norm": 2.632267951965332, "learning_rate": 4.973266932885999e-06, "loss": 0.1478593796491623, "step": 600 }, { "epoch": 0.15602284527518173, "grad_norm": 1.0874810218811035, "learning_rate": 4.973121341265343e-06, "loss": 0.17707350850105286, "step": 601 }, { "epoch": 0.15628245067497404, "grad_norm": 1.2164853811264038, "learning_rate": 4.9729753566477376e-06, "loss": 0.19254174828529358, "step": 602 }, { "epoch": 0.15654205607476634, "grad_norm": 2.027400493621826, "learning_rate": 4.972828979058988e-06, "loss": 0.20497636497020721, "step": 603 }, { "epoch": 0.15680166147455868, "grad_norm": 2.937351703643799, "learning_rate": 4.972682208524971e-06, "loss": 0.16679608821868896, "step": 604 }, { "epoch": 0.15706126687435099, "grad_norm": 3.444216251373291, "learning_rate": 4.972535045071632e-06, "loss": 0.16020503640174866, "step": 605 }, { "epoch": 0.1573208722741433, "grad_norm": 0.7400794625282288, "learning_rate": 4.972387488724986e-06, "loss": 0.15631183981895447, "step": 606 }, { "epoch": 0.15758047767393563, "grad_norm": 1.386948823928833, "learning_rate": 4.972239539511119e-06, "loss": 0.1737084984779358, "step": 607 }, { "epoch": 0.15784008307372793, "grad_norm": 2.0418174266815186, "learning_rate": 4.9720911974561835e-06, "loss": 0.1508300006389618, "step": 608 }, { "epoch": 0.15809968847352024, "grad_norm": 1.1517281532287598, "learning_rate": 4.971942462586403e-06, "loss": 0.18151938915252686, "step": 609 }, { "epoch": 0.15835929387331257, "grad_norm": 0.5383716821670532, "learning_rate": 4.971793334928071e-06, "loss": 0.1448233723640442, "step": 610 }, { "epoch": 0.15861889927310488, "grad_norm": 1.8620156049728394, "learning_rate": 4.9716438145075505e-06, "loss": 0.19997446238994598, "step": 611 }, { "epoch": 0.1588785046728972, "grad_norm": 0.8875040411949158, "learning_rate": 4.971493901351272e-06, "loss": 0.1690278947353363, "step": 612 }, { "epoch": 0.15913811007268952, "grad_norm": 4.588420867919922, "learning_rate": 4.971343595485736e-06, "loss": 0.21865589916706085, "step": 613 }, { "epoch": 0.15939771547248183, "grad_norm": 2.0321004390716553, "learning_rate": 4.971192896937516e-06, "loss": 0.1801266372203827, "step": 614 }, { "epoch": 0.15965732087227413, "grad_norm": 2.7571165561676025, "learning_rate": 4.97104180573325e-06, "loss": 0.15483474731445312, "step": 615 }, { "epoch": 0.15991692627206647, "grad_norm": 1.515758752822876, "learning_rate": 4.970890321899647e-06, "loss": 0.18296468257904053, "step": 616 }, { "epoch": 0.16017653167185877, "grad_norm": 2.2988340854644775, "learning_rate": 4.970738445463488e-06, "loss": 0.19720003008842468, "step": 617 }, { "epoch": 0.16043613707165108, "grad_norm": 0.9248607754707336, "learning_rate": 4.9705861764516195e-06, "loss": 0.19072097539901733, "step": 618 }, { "epoch": 0.16069574247144341, "grad_norm": 0.8922683000564575, "learning_rate": 4.97043351489096e-06, "loss": 0.144118532538414, "step": 619 }, { "epoch": 0.16095534787123572, "grad_norm": 2.8125011920928955, "learning_rate": 4.970280460808495e-06, "loss": 0.19185808300971985, "step": 620 }, { "epoch": 0.16121495327102803, "grad_norm": 1.4687244892120361, "learning_rate": 4.970127014231283e-06, "loss": 0.15293094515800476, "step": 621 }, { "epoch": 0.16147455867082036, "grad_norm": 1.5851166248321533, "learning_rate": 4.969973175186449e-06, "loss": 0.17232325673103333, "step": 622 }, { "epoch": 0.16173416407061267, "grad_norm": 1.6057004928588867, "learning_rate": 4.969818943701189e-06, "loss": 0.19219970703125, "step": 623 }, { "epoch": 0.16199376947040497, "grad_norm": 4.174839973449707, "learning_rate": 4.969664319802766e-06, "loss": 0.16567838191986084, "step": 624 }, { "epoch": 0.1622533748701973, "grad_norm": 2.7660446166992188, "learning_rate": 4.9695093035185154e-06, "loss": 0.1613733172416687, "step": 625 }, { "epoch": 0.16251298026998962, "grad_norm": 0.8579680919647217, "learning_rate": 4.96935389487584e-06, "loss": 0.13964557647705078, "step": 626 }, { "epoch": 0.16277258566978192, "grad_norm": 2.3079872131347656, "learning_rate": 4.969198093902212e-06, "loss": 0.18008995056152344, "step": 627 }, { "epoch": 0.16303219106957426, "grad_norm": 1.1023926734924316, "learning_rate": 4.969041900625174e-06, "loss": 0.15682421624660492, "step": 628 }, { "epoch": 0.16329179646936656, "grad_norm": 1.4863510131835938, "learning_rate": 4.968885315072338e-06, "loss": 0.16077286005020142, "step": 629 }, { "epoch": 0.16355140186915887, "grad_norm": 2.546595335006714, "learning_rate": 4.968728337271383e-06, "loss": 0.18337076902389526, "step": 630 }, { "epoch": 0.1638110072689512, "grad_norm": 0.8976767063140869, "learning_rate": 4.968570967250061e-06, "loss": 0.1427949070930481, "step": 631 }, { "epoch": 0.1640706126687435, "grad_norm": 2.4721009731292725, "learning_rate": 4.9684132050361896e-06, "loss": 0.1601601541042328, "step": 632 }, { "epoch": 0.16433021806853582, "grad_norm": 1.8669639825820923, "learning_rate": 4.968255050657658e-06, "loss": 0.1659151017665863, "step": 633 }, { "epoch": 0.16458982346832815, "grad_norm": 0.9570918083190918, "learning_rate": 4.968096504142426e-06, "loss": 0.18069233000278473, "step": 634 }, { "epoch": 0.16484942886812046, "grad_norm": 2.996115207672119, "learning_rate": 4.96793756551852e-06, "loss": 0.17166095972061157, "step": 635 }, { "epoch": 0.16510903426791276, "grad_norm": 0.9210231900215149, "learning_rate": 4.967778234814035e-06, "loss": 0.16407272219657898, "step": 636 }, { "epoch": 0.1653686396677051, "grad_norm": 0.9613329768180847, "learning_rate": 4.96761851205714e-06, "loss": 0.1810782104730606, "step": 637 }, { "epoch": 0.1656282450674974, "grad_norm": 1.2179378271102905, "learning_rate": 4.9674583972760676e-06, "loss": 0.1834588646888733, "step": 638 }, { "epoch": 0.1658878504672897, "grad_norm": 1.2703843116760254, "learning_rate": 4.967297890499124e-06, "loss": 0.19999969005584717, "step": 639 }, { "epoch": 0.16614745586708204, "grad_norm": 0.7929041981697083, "learning_rate": 4.967136991754683e-06, "loss": 0.14280837774276733, "step": 640 }, { "epoch": 0.16640706126687435, "grad_norm": 0.9506551623344421, "learning_rate": 4.966975701071188e-06, "loss": 0.14492106437683105, "step": 641 }, { "epoch": 0.16666666666666666, "grad_norm": 1.2350395917892456, "learning_rate": 4.96681401847715e-06, "loss": 0.17259132862091064, "step": 642 }, { "epoch": 0.166926272066459, "grad_norm": 0.9262452125549316, "learning_rate": 4.966651944001153e-06, "loss": 0.19995824992656708, "step": 643 }, { "epoch": 0.1671858774662513, "grad_norm": 0.9393950700759888, "learning_rate": 4.966489477671846e-06, "loss": 0.14873751997947693, "step": 644 }, { "epoch": 0.1674454828660436, "grad_norm": 1.2250738143920898, "learning_rate": 4.966326619517951e-06, "loss": 0.1779806762933731, "step": 645 }, { "epoch": 0.16770508826583594, "grad_norm": 0.9724617600440979, "learning_rate": 4.966163369568256e-06, "loss": 0.14476636052131653, "step": 646 }, { "epoch": 0.16796469366562825, "grad_norm": 2.2918267250061035, "learning_rate": 4.965999727851621e-06, "loss": 0.19331464171409607, "step": 647 }, { "epoch": 0.16822429906542055, "grad_norm": 0.46737140417099, "learning_rate": 4.965835694396973e-06, "loss": 0.1461721807718277, "step": 648 }, { "epoch": 0.16848390446521289, "grad_norm": 1.1035538911819458, "learning_rate": 4.965671269233312e-06, "loss": 0.17177510261535645, "step": 649 }, { "epoch": 0.1687435098650052, "grad_norm": 1.8421696424484253, "learning_rate": 4.9655064523897015e-06, "loss": 0.16707247495651245, "step": 650 }, { "epoch": 0.1690031152647975, "grad_norm": 1.7390724420547485, "learning_rate": 4.965341243895279e-06, "loss": 0.21497440338134766, "step": 651 }, { "epoch": 0.16926272066458983, "grad_norm": 1.9539812803268433, "learning_rate": 4.9651756437792485e-06, "loss": 0.19585376977920532, "step": 652 }, { "epoch": 0.16952232606438214, "grad_norm": 1.6535170078277588, "learning_rate": 4.965009652070885e-06, "loss": 0.19891832768917084, "step": 653 }, { "epoch": 0.16978193146417445, "grad_norm": 0.6929019093513489, "learning_rate": 4.964843268799533e-06, "loss": 0.1441730558872223, "step": 654 }, { "epoch": 0.17004153686396678, "grad_norm": 1.0781141519546509, "learning_rate": 4.964676493994604e-06, "loss": 0.1767147034406662, "step": 655 }, { "epoch": 0.1703011422637591, "grad_norm": 2.797060966491699, "learning_rate": 4.964509327685579e-06, "loss": 0.14675143361091614, "step": 656 }, { "epoch": 0.1705607476635514, "grad_norm": 2.5857365131378174, "learning_rate": 4.964341769902012e-06, "loss": 0.19200876355171204, "step": 657 }, { "epoch": 0.17082035306334373, "grad_norm": 1.8720515966415405, "learning_rate": 4.964173820673521e-06, "loss": 0.14667190611362457, "step": 658 }, { "epoch": 0.17107995846313603, "grad_norm": 2.6366970539093018, "learning_rate": 4.964005480029797e-06, "loss": 0.14772315323352814, "step": 659 }, { "epoch": 0.17133956386292834, "grad_norm": 2.087517738342285, "learning_rate": 4.963836748000597e-06, "loss": 0.1605493724346161, "step": 660 }, { "epoch": 0.17159916926272067, "grad_norm": 0.9202359914779663, "learning_rate": 4.963667624615751e-06, "loss": 0.16838166117668152, "step": 661 }, { "epoch": 0.17185877466251298, "grad_norm": 1.5527070760726929, "learning_rate": 4.9634981099051545e-06, "loss": 0.16467821598052979, "step": 662 }, { "epoch": 0.1721183800623053, "grad_norm": 2.0329718589782715, "learning_rate": 4.963328203898775e-06, "loss": 0.17108629643917084, "step": 663 }, { "epoch": 0.17237798546209762, "grad_norm": 0.828184962272644, "learning_rate": 4.963157906626648e-06, "loss": 0.17297661304473877, "step": 664 }, { "epoch": 0.17263759086188993, "grad_norm": 1.6255948543548584, "learning_rate": 4.9629872181188774e-06, "loss": 0.16939182579517365, "step": 665 }, { "epoch": 0.17289719626168223, "grad_norm": 1.3997122049331665, "learning_rate": 4.962816138405637e-06, "loss": 0.16594083607196808, "step": 666 }, { "epoch": 0.17315680166147457, "grad_norm": 1.2898600101470947, "learning_rate": 4.9626446675171704e-06, "loss": 0.14803436398506165, "step": 667 }, { "epoch": 0.17341640706126688, "grad_norm": 1.0907799005508423, "learning_rate": 4.962472805483789e-06, "loss": 0.15183573961257935, "step": 668 }, { "epoch": 0.17367601246105918, "grad_norm": 1.0392872095108032, "learning_rate": 4.962300552335875e-06, "loss": 0.16841739416122437, "step": 669 }, { "epoch": 0.17393561786085152, "grad_norm": 0.919896125793457, "learning_rate": 4.962127908103878e-06, "loss": 0.18181546032428741, "step": 670 }, { "epoch": 0.17419522326064382, "grad_norm": 1.1809934377670288, "learning_rate": 4.961954872818318e-06, "loss": 0.1752224713563919, "step": 671 }, { "epoch": 0.17445482866043613, "grad_norm": 1.2229024171829224, "learning_rate": 4.961781446509784e-06, "loss": 0.2021273374557495, "step": 672 }, { "epoch": 0.17471443406022846, "grad_norm": 0.4400077760219574, "learning_rate": 4.961607629208933e-06, "loss": 0.13196836411952972, "step": 673 }, { "epoch": 0.17497403946002077, "grad_norm": 1.4407336711883545, "learning_rate": 4.961433420946492e-06, "loss": 0.21533484756946564, "step": 674 }, { "epoch": 0.17523364485981308, "grad_norm": 2.2557950019836426, "learning_rate": 4.961258821753257e-06, "loss": 0.15256774425506592, "step": 675 }, { "epoch": 0.1754932502596054, "grad_norm": 1.715008020401001, "learning_rate": 4.961083831660095e-06, "loss": 0.16376863420009613, "step": 676 }, { "epoch": 0.17575285565939772, "grad_norm": 0.7675665020942688, "learning_rate": 4.960908450697937e-06, "loss": 0.18787875771522522, "step": 677 }, { "epoch": 0.17601246105919002, "grad_norm": 0.5719281435012817, "learning_rate": 4.960732678897788e-06, "loss": 0.1652291715145111, "step": 678 }, { "epoch": 0.17627206645898236, "grad_norm": 2.5497324466705322, "learning_rate": 4.960556516290721e-06, "loss": 0.18529483675956726, "step": 679 }, { "epoch": 0.17653167185877466, "grad_norm": 1.5956840515136719, "learning_rate": 4.960379962907878e-06, "loss": 0.1670048087835312, "step": 680 }, { "epoch": 0.17679127725856697, "grad_norm": 0.7048987746238708, "learning_rate": 4.960203018780467e-06, "loss": 0.15139012038707733, "step": 681 }, { "epoch": 0.1770508826583593, "grad_norm": 3.1906917095184326, "learning_rate": 4.9600256839397695e-06, "loss": 0.1896807849407196, "step": 682 }, { "epoch": 0.1773104880581516, "grad_norm": 2.3884198665618896, "learning_rate": 4.959847958417134e-06, "loss": 0.17432257533073425, "step": 683 }, { "epoch": 0.17757009345794392, "grad_norm": 2.551414728164673, "learning_rate": 4.959669842243979e-06, "loss": 0.18327781558036804, "step": 684 }, { "epoch": 0.17782969885773625, "grad_norm": 1.4593498706817627, "learning_rate": 4.95949133545179e-06, "loss": 0.20115286111831665, "step": 685 }, { "epoch": 0.17808930425752856, "grad_norm": 3.287424325942993, "learning_rate": 4.959312438072123e-06, "loss": 0.18687722086906433, "step": 686 }, { "epoch": 0.17834890965732086, "grad_norm": 2.4829559326171875, "learning_rate": 4.959133150136605e-06, "loss": 0.2081683725118637, "step": 687 }, { "epoch": 0.1786085150571132, "grad_norm": 2.571720600128174, "learning_rate": 4.958953471676927e-06, "loss": 0.17806430160999298, "step": 688 }, { "epoch": 0.1788681204569055, "grad_norm": 2.578777313232422, "learning_rate": 4.958773402724854e-06, "loss": 0.17012082040309906, "step": 689 }, { "epoch": 0.1791277258566978, "grad_norm": 1.6587815284729004, "learning_rate": 4.958592943312218e-06, "loss": 0.15136179327964783, "step": 690 }, { "epoch": 0.17938733125649015, "grad_norm": 2.6289968490600586, "learning_rate": 4.958412093470919e-06, "loss": 0.21072401106357574, "step": 691 }, { "epoch": 0.17964693665628245, "grad_norm": 1.3687199354171753, "learning_rate": 4.958230853232928e-06, "loss": 0.18203851580619812, "step": 692 }, { "epoch": 0.17990654205607476, "grad_norm": 0.8556008338928223, "learning_rate": 4.958049222630285e-06, "loss": 0.1941826343536377, "step": 693 }, { "epoch": 0.1801661474558671, "grad_norm": 4.349623680114746, "learning_rate": 4.957867201695096e-06, "loss": 0.18495455384254456, "step": 694 }, { "epoch": 0.1804257528556594, "grad_norm": 3.268566131591797, "learning_rate": 4.957684790459539e-06, "loss": 0.17037110030651093, "step": 695 }, { "epoch": 0.1806853582554517, "grad_norm": 1.5059996843338013, "learning_rate": 4.957501988955861e-06, "loss": 0.18787305057048798, "step": 696 }, { "epoch": 0.18094496365524404, "grad_norm": 0.5389428734779358, "learning_rate": 4.9573187972163765e-06, "loss": 0.14906352758407593, "step": 697 }, { "epoch": 0.18120456905503635, "grad_norm": 2.3639585971832275, "learning_rate": 4.9571352152734696e-06, "loss": 0.17156758904457092, "step": 698 }, { "epoch": 0.18146417445482865, "grad_norm": 1.8957890272140503, "learning_rate": 4.956951243159593e-06, "loss": 0.158742755651474, "step": 699 }, { "epoch": 0.181723779854621, "grad_norm": 1.2689954042434692, "learning_rate": 4.956766880907269e-06, "loss": 0.15872007608413696, "step": 700 }, { "epoch": 0.1819833852544133, "grad_norm": 0.7325999736785889, "learning_rate": 4.956582128549089e-06, "loss": 0.14297017455101013, "step": 701 }, { "epoch": 0.1822429906542056, "grad_norm": 0.9952068328857422, "learning_rate": 4.9563969861177126e-06, "loss": 0.1744202971458435, "step": 702 }, { "epoch": 0.18250259605399793, "grad_norm": 1.3818058967590332, "learning_rate": 4.9562114536458686e-06, "loss": 0.20506063103675842, "step": 703 }, { "epoch": 0.18276220145379024, "grad_norm": 2.3214683532714844, "learning_rate": 4.956025531166356e-06, "loss": 0.16739334166049957, "step": 704 }, { "epoch": 0.18302180685358255, "grad_norm": 2.480239152908325, "learning_rate": 4.955839218712039e-06, "loss": 0.16488870978355408, "step": 705 }, { "epoch": 0.18328141225337488, "grad_norm": 1.8675527572631836, "learning_rate": 4.955652516315856e-06, "loss": 0.17869243025779724, "step": 706 }, { "epoch": 0.1835410176531672, "grad_norm": 0.6059030294418335, "learning_rate": 4.955465424010812e-06, "loss": 0.143377423286438, "step": 707 }, { "epoch": 0.1838006230529595, "grad_norm": 1.5393089056015015, "learning_rate": 4.955277941829978e-06, "loss": 0.1504325568675995, "step": 708 }, { "epoch": 0.18406022845275183, "grad_norm": 3.608652353286743, "learning_rate": 4.955090069806499e-06, "loss": 0.17577344179153442, "step": 709 }, { "epoch": 0.18431983385254413, "grad_norm": 1.7852610349655151, "learning_rate": 4.954901807973586e-06, "loss": 0.13711290061473846, "step": 710 }, { "epoch": 0.18457943925233644, "grad_norm": 0.9077580571174622, "learning_rate": 4.954713156364517e-06, "loss": 0.14494726061820984, "step": 711 }, { "epoch": 0.18483904465212878, "grad_norm": 1.0199952125549316, "learning_rate": 4.954524115012645e-06, "loss": 0.1512145847082138, "step": 712 }, { "epoch": 0.18509865005192108, "grad_norm": 3.0090532302856445, "learning_rate": 4.9543346839513856e-06, "loss": 0.15638090670108795, "step": 713 }, { "epoch": 0.1853582554517134, "grad_norm": 1.5015649795532227, "learning_rate": 4.954144863214226e-06, "loss": 0.16328243911266327, "step": 714 }, { "epoch": 0.18561786085150572, "grad_norm": 1.5221110582351685, "learning_rate": 4.953954652834724e-06, "loss": 0.1572495698928833, "step": 715 }, { "epoch": 0.18587746625129803, "grad_norm": 0.5040673017501831, "learning_rate": 4.953764052846502e-06, "loss": 0.16409525275230408, "step": 716 }, { "epoch": 0.18613707165109034, "grad_norm": 2.7120931148529053, "learning_rate": 4.953573063283257e-06, "loss": 0.15606534481048584, "step": 717 }, { "epoch": 0.18639667705088267, "grad_norm": 2.3765501976013184, "learning_rate": 4.953381684178748e-06, "loss": 0.14136698842048645, "step": 718 }, { "epoch": 0.18665628245067498, "grad_norm": 3.8693642616271973, "learning_rate": 4.9531899155668076e-06, "loss": 0.177644744515419, "step": 719 }, { "epoch": 0.18691588785046728, "grad_norm": 1.0253064632415771, "learning_rate": 4.952997757481338e-06, "loss": 0.17486947774887085, "step": 720 }, { "epoch": 0.18717549325025962, "grad_norm": 0.6950884461402893, "learning_rate": 4.952805209956306e-06, "loss": 0.18750345706939697, "step": 721 }, { "epoch": 0.18743509865005192, "grad_norm": 3.9729702472686768, "learning_rate": 4.95261227302575e-06, "loss": 0.2228296399116516, "step": 722 }, { "epoch": 0.18769470404984423, "grad_norm": 3.5975003242492676, "learning_rate": 4.952418946723778e-06, "loss": 0.17934557795524597, "step": 723 }, { "epoch": 0.18795430944963656, "grad_norm": 1.736036777496338, "learning_rate": 4.952225231084565e-06, "loss": 0.16648760437965393, "step": 724 }, { "epoch": 0.18821391484942887, "grad_norm": 1.023704171180725, "learning_rate": 4.9520311261423556e-06, "loss": 0.18314531445503235, "step": 725 }, { "epoch": 0.18847352024922118, "grad_norm": 2.348705768585205, "learning_rate": 4.951836631931461e-06, "loss": 0.16922101378440857, "step": 726 }, { "epoch": 0.1887331256490135, "grad_norm": 2.3303604125976562, "learning_rate": 4.951641748486268e-06, "loss": 0.1484547257423401, "step": 727 }, { "epoch": 0.18899273104880582, "grad_norm": 3.3901519775390625, "learning_rate": 4.951446475841223e-06, "loss": 0.18157006800174713, "step": 728 }, { "epoch": 0.18925233644859812, "grad_norm": 1.7963814735412598, "learning_rate": 4.951250814030847e-06, "loss": 0.1987340748310089, "step": 729 }, { "epoch": 0.18951194184839046, "grad_norm": 0.6135780811309814, "learning_rate": 4.951054763089731e-06, "loss": 0.14823734760284424, "step": 730 }, { "epoch": 0.18977154724818276, "grad_norm": 2.7567741870880127, "learning_rate": 4.95085832305253e-06, "loss": 0.2030656784772873, "step": 731 }, { "epoch": 0.19003115264797507, "grad_norm": 1.4581018686294556, "learning_rate": 4.950661493953969e-06, "loss": 0.20543710887432098, "step": 732 }, { "epoch": 0.1902907580477674, "grad_norm": 2.9050498008728027, "learning_rate": 4.950464275828845e-06, "loss": 0.15991829335689545, "step": 733 }, { "epoch": 0.1905503634475597, "grad_norm": 1.3005132675170898, "learning_rate": 4.9502666687120215e-06, "loss": 0.14737483859062195, "step": 734 }, { "epoch": 0.19080996884735202, "grad_norm": 0.8308334350585938, "learning_rate": 4.95006867263843e-06, "loss": 0.15065696835517883, "step": 735 }, { "epoch": 0.19106957424714435, "grad_norm": 0.6689059138298035, "learning_rate": 4.949870287643072e-06, "loss": 0.16499875485897064, "step": 736 }, { "epoch": 0.19132917964693666, "grad_norm": 2.575667381286621, "learning_rate": 4.949671513761018e-06, "loss": 0.1716618686914444, "step": 737 }, { "epoch": 0.19158878504672897, "grad_norm": 1.595591425895691, "learning_rate": 4.949472351027405e-06, "loss": 0.1745269000530243, "step": 738 }, { "epoch": 0.1918483904465213, "grad_norm": 1.9652984142303467, "learning_rate": 4.949272799477442e-06, "loss": 0.15589751303195953, "step": 739 }, { "epoch": 0.1921079958463136, "grad_norm": 1.2318449020385742, "learning_rate": 4.949072859146406e-06, "loss": 0.1538684368133545, "step": 740 }, { "epoch": 0.1923676012461059, "grad_norm": 1.1587445735931396, "learning_rate": 4.948872530069639e-06, "loss": 0.15507040917873383, "step": 741 }, { "epoch": 0.19262720664589825, "grad_norm": 2.0649831295013428, "learning_rate": 4.948671812282557e-06, "loss": 0.16975641250610352, "step": 742 }, { "epoch": 0.19288681204569055, "grad_norm": 1.8744033575057983, "learning_rate": 4.948470705820642e-06, "loss": 0.16172392666339874, "step": 743 }, { "epoch": 0.19314641744548286, "grad_norm": 1.3967070579528809, "learning_rate": 4.948269210719443e-06, "loss": 0.15447208285331726, "step": 744 }, { "epoch": 0.1934060228452752, "grad_norm": 0.8192422389984131, "learning_rate": 4.948067327014583e-06, "loss": 0.14177602529525757, "step": 745 }, { "epoch": 0.1936656282450675, "grad_norm": 0.9616960287094116, "learning_rate": 4.947865054741748e-06, "loss": 0.14841650426387787, "step": 746 }, { "epoch": 0.1939252336448598, "grad_norm": 2.891822576522827, "learning_rate": 4.947662393936695e-06, "loss": 0.20922088623046875, "step": 747 }, { "epoch": 0.19418483904465214, "grad_norm": 0.8668384552001953, "learning_rate": 4.947459344635251e-06, "loss": 0.16401809453964233, "step": 748 }, { "epoch": 0.19444444444444445, "grad_norm": 0.9667485356330872, "learning_rate": 4.9472559068733104e-06, "loss": 0.1680034101009369, "step": 749 }, { "epoch": 0.19470404984423675, "grad_norm": 0.992344319820404, "learning_rate": 4.9470520806868356e-06, "loss": 0.16670766472816467, "step": 750 }, { "epoch": 0.1949636552440291, "grad_norm": 1.4704527854919434, "learning_rate": 4.946847866111859e-06, "loss": 0.1532723605632782, "step": 751 }, { "epoch": 0.1952232606438214, "grad_norm": 0.9106773734092712, "learning_rate": 4.9466432631844805e-06, "loss": 0.13972479104995728, "step": 752 }, { "epoch": 0.1954828660436137, "grad_norm": 0.7380465865135193, "learning_rate": 4.946438271940869e-06, "loss": 0.16488036513328552, "step": 753 }, { "epoch": 0.19574247144340604, "grad_norm": 1.8760056495666504, "learning_rate": 4.946232892417264e-06, "loss": 0.14671048521995544, "step": 754 }, { "epoch": 0.19600207684319834, "grad_norm": 1.0405858755111694, "learning_rate": 4.94602712464997e-06, "loss": 0.1985287368297577, "step": 755 }, { "epoch": 0.19626168224299065, "grad_norm": 0.9870449900627136, "learning_rate": 4.945820968675363e-06, "loss": 0.17467500269412994, "step": 756 }, { "epoch": 0.19652128764278298, "grad_norm": 1.3687822818756104, "learning_rate": 4.945614424529886e-06, "loss": 0.15226681530475616, "step": 757 }, { "epoch": 0.1967808930425753, "grad_norm": 2.5199756622314453, "learning_rate": 4.945407492250052e-06, "loss": 0.2061992585659027, "step": 758 }, { "epoch": 0.1970404984423676, "grad_norm": 0.44093263149261475, "learning_rate": 4.945200171872442e-06, "loss": 0.1291900873184204, "step": 759 }, { "epoch": 0.19730010384215993, "grad_norm": 0.7617220878601074, "learning_rate": 4.944992463433704e-06, "loss": 0.1582765281200409, "step": 760 }, { "epoch": 0.19755970924195224, "grad_norm": 1.6240531206130981, "learning_rate": 4.944784366970558e-06, "loss": 0.15999732911586761, "step": 761 }, { "epoch": 0.19781931464174454, "grad_norm": 1.997636079788208, "learning_rate": 4.944575882519789e-06, "loss": 0.1465025246143341, "step": 762 }, { "epoch": 0.19807892004153688, "grad_norm": 1.3833746910095215, "learning_rate": 4.944367010118254e-06, "loss": 0.17820033431053162, "step": 763 }, { "epoch": 0.19833852544132918, "grad_norm": 0.8825951218605042, "learning_rate": 4.944157749802876e-06, "loss": 0.18184328079223633, "step": 764 }, { "epoch": 0.1985981308411215, "grad_norm": 0.9980071187019348, "learning_rate": 4.9439481016106465e-06, "loss": 0.16489127278327942, "step": 765 }, { "epoch": 0.19885773624091382, "grad_norm": 3.363140106201172, "learning_rate": 4.943738065578628e-06, "loss": 0.2065974920988083, "step": 766 }, { "epoch": 0.19911734164070613, "grad_norm": 1.1597405672073364, "learning_rate": 4.943527641743949e-06, "loss": 0.16904215514659882, "step": 767 }, { "epoch": 0.19937694704049844, "grad_norm": 0.7904525995254517, "learning_rate": 4.9433168301438084e-06, "loss": 0.16282770037651062, "step": 768 }, { "epoch": 0.19963655244029077, "grad_norm": 1.856398582458496, "learning_rate": 4.943105630815472e-06, "loss": 0.18283730745315552, "step": 769 }, { "epoch": 0.19989615784008308, "grad_norm": 1.5863265991210938, "learning_rate": 4.942894043796276e-06, "loss": 0.16454721987247467, "step": 770 }, { "epoch": 0.20015576323987538, "grad_norm": 1.4463210105895996, "learning_rate": 4.942682069123623e-06, "loss": 0.13703900575637817, "step": 771 }, { "epoch": 0.20041536863966772, "grad_norm": 0.5435364246368408, "learning_rate": 4.942469706834987e-06, "loss": 0.1338147073984146, "step": 772 }, { "epoch": 0.20067497403946002, "grad_norm": 1.3717883825302124, "learning_rate": 4.942256956967907e-06, "loss": 0.17564791440963745, "step": 773 }, { "epoch": 0.20093457943925233, "grad_norm": 1.06517493724823, "learning_rate": 4.942043819559993e-06, "loss": 0.18462976813316345, "step": 774 }, { "epoch": 0.20119418483904467, "grad_norm": 1.5689527988433838, "learning_rate": 4.941830294648924e-06, "loss": 0.1486607789993286, "step": 775 }, { "epoch": 0.20145379023883697, "grad_norm": 1.5183995962142944, "learning_rate": 4.941616382272444e-06, "loss": 0.14629259705543518, "step": 776 }, { "epoch": 0.20171339563862928, "grad_norm": 1.5545600652694702, "learning_rate": 4.94140208246837e-06, "loss": 0.16766464710235596, "step": 777 }, { "epoch": 0.2019730010384216, "grad_norm": 1.8622503280639648, "learning_rate": 4.941187395274584e-06, "loss": 0.16811081767082214, "step": 778 }, { "epoch": 0.20223260643821392, "grad_norm": 0.4814753532409668, "learning_rate": 4.940972320729036e-06, "loss": 0.14107225835323334, "step": 779 }, { "epoch": 0.20249221183800623, "grad_norm": 0.46144232153892517, "learning_rate": 4.940756858869751e-06, "loss": 0.13951340317726135, "step": 780 }, { "epoch": 0.20275181723779856, "grad_norm": 1.4274976253509521, "learning_rate": 4.940541009734814e-06, "loss": 0.20416712760925293, "step": 781 }, { "epoch": 0.20301142263759087, "grad_norm": 1.013152003288269, "learning_rate": 4.940324773362383e-06, "loss": 0.15378719568252563, "step": 782 }, { "epoch": 0.20327102803738317, "grad_norm": 0.9661071300506592, "learning_rate": 4.940108149790684e-06, "loss": 0.17651280760765076, "step": 783 }, { "epoch": 0.2035306334371755, "grad_norm": 0.7165428996086121, "learning_rate": 4.939891139058011e-06, "loss": 0.144172802567482, "step": 784 }, { "epoch": 0.2037902388369678, "grad_norm": 1.2842830419540405, "learning_rate": 4.939673741202726e-06, "loss": 0.184444397687912, "step": 785 }, { "epoch": 0.20404984423676012, "grad_norm": 0.8286370635032654, "learning_rate": 4.939455956263262e-06, "loss": 0.13684117794036865, "step": 786 }, { "epoch": 0.20430944963655245, "grad_norm": 2.4812488555908203, "learning_rate": 4.9392377842781156e-06, "loss": 0.1768890619277954, "step": 787 }, { "epoch": 0.20456905503634476, "grad_norm": 0.555282473564148, "learning_rate": 4.939019225285856e-06, "loss": 0.16995057463645935, "step": 788 }, { "epoch": 0.20482866043613707, "grad_norm": 1.721859335899353, "learning_rate": 4.938800279325119e-06, "loss": 0.157206192612648, "step": 789 }, { "epoch": 0.2050882658359294, "grad_norm": 1.5816404819488525, "learning_rate": 4.93858094643461e-06, "loss": 0.1477537304162979, "step": 790 }, { "epoch": 0.2053478712357217, "grad_norm": 1.1964572668075562, "learning_rate": 4.938361226653101e-06, "loss": 0.14996537566184998, "step": 791 }, { "epoch": 0.205607476635514, "grad_norm": 0.8245982527732849, "learning_rate": 4.938141120019435e-06, "loss": 0.16643527150154114, "step": 792 }, { "epoch": 0.20586708203530635, "grad_norm": 3.2168798446655273, "learning_rate": 4.9379206265725196e-06, "loss": 0.14791718125343323, "step": 793 }, { "epoch": 0.20612668743509865, "grad_norm": 3.238724708557129, "learning_rate": 4.937699746351334e-06, "loss": 0.14709194004535675, "step": 794 }, { "epoch": 0.20638629283489096, "grad_norm": 0.8146167397499084, "learning_rate": 4.937478479394925e-06, "loss": 0.16768375039100647, "step": 795 }, { "epoch": 0.2066458982346833, "grad_norm": 2.4719231128692627, "learning_rate": 4.937256825742409e-06, "loss": 0.1474972665309906, "step": 796 }, { "epoch": 0.2069055036344756, "grad_norm": 3.1162688732147217, "learning_rate": 4.937034785432967e-06, "loss": 0.17262201011180878, "step": 797 }, { "epoch": 0.2071651090342679, "grad_norm": 1.376809000968933, "learning_rate": 4.936812358505851e-06, "loss": 0.1525888741016388, "step": 798 }, { "epoch": 0.20742471443406024, "grad_norm": 0.7548062801361084, "learning_rate": 4.936589545000381e-06, "loss": 0.1540077030658722, "step": 799 }, { "epoch": 0.20768431983385255, "grad_norm": 3.742438793182373, "learning_rate": 4.936366344955947e-06, "loss": 0.19573867321014404, "step": 800 }, { "epoch": 0.20794392523364486, "grad_norm": 1.3899879455566406, "learning_rate": 4.936142758412004e-06, "loss": 0.14091134071350098, "step": 801 }, { "epoch": 0.2082035306334372, "grad_norm": 0.6601426005363464, "learning_rate": 4.935918785408076e-06, "loss": 0.15727919340133667, "step": 802 }, { "epoch": 0.2084631360332295, "grad_norm": 1.4593565464019775, "learning_rate": 4.935694425983759e-06, "loss": 0.17819032073020935, "step": 803 }, { "epoch": 0.2087227414330218, "grad_norm": 2.11513090133667, "learning_rate": 4.935469680178713e-06, "loss": 0.20479406416416168, "step": 804 }, { "epoch": 0.2089823468328141, "grad_norm": 1.6440891027450562, "learning_rate": 4.9352445480326675e-06, "loss": 0.17897266149520874, "step": 805 }, { "epoch": 0.20924195223260644, "grad_norm": 1.9139387607574463, "learning_rate": 4.935019029585421e-06, "loss": 0.1805766224861145, "step": 806 }, { "epoch": 0.20950155763239875, "grad_norm": 2.686816453933716, "learning_rate": 4.934793124876842e-06, "loss": 0.1952880322933197, "step": 807 }, { "epoch": 0.20976116303219106, "grad_norm": 1.0191266536712646, "learning_rate": 4.934566833946863e-06, "loss": 0.14853382110595703, "step": 808 }, { "epoch": 0.2100207684319834, "grad_norm": 1.4927254915237427, "learning_rate": 4.934340156835488e-06, "loss": 0.15183578431606293, "step": 809 }, { "epoch": 0.2102803738317757, "grad_norm": 3.247328519821167, "learning_rate": 4.934113093582787e-06, "loss": 0.1650736927986145, "step": 810 }, { "epoch": 0.210539979231568, "grad_norm": 1.1205036640167236, "learning_rate": 4.933885644228902e-06, "loss": 0.12381936609745026, "step": 811 }, { "epoch": 0.21079958463136034, "grad_norm": 1.594746470451355, "learning_rate": 4.933657808814038e-06, "loss": 0.17819292843341827, "step": 812 }, { "epoch": 0.21105919003115264, "grad_norm": 2.271071672439575, "learning_rate": 4.933429587378473e-06, "loss": 0.15450754761695862, "step": 813 }, { "epoch": 0.21131879543094495, "grad_norm": 1.393080234527588, "learning_rate": 4.933200979962551e-06, "loss": 0.13932475447654724, "step": 814 }, { "epoch": 0.21157840083073728, "grad_norm": 0.7440643906593323, "learning_rate": 4.932971986606685e-06, "loss": 0.13512900471687317, "step": 815 }, { "epoch": 0.2118380062305296, "grad_norm": 1.023361325263977, "learning_rate": 4.9327426073513555e-06, "loss": 0.13505828380584717, "step": 816 }, { "epoch": 0.2120976116303219, "grad_norm": 3.2076900005340576, "learning_rate": 4.9325128422371095e-06, "loss": 0.13691145181655884, "step": 817 }, { "epoch": 0.21235721703011423, "grad_norm": 1.9609174728393555, "learning_rate": 4.932282691304567e-06, "loss": 0.1337742954492569, "step": 818 }, { "epoch": 0.21261682242990654, "grad_norm": 2.301546812057495, "learning_rate": 4.932052154594413e-06, "loss": 0.18556763231754303, "step": 819 }, { "epoch": 0.21287642782969884, "grad_norm": 2.109372615814209, "learning_rate": 4.931821232147399e-06, "loss": 0.16978231072425842, "step": 820 }, { "epoch": 0.21313603322949118, "grad_norm": 1.669675588607788, "learning_rate": 4.931589924004349e-06, "loss": 0.19403746724128723, "step": 821 }, { "epoch": 0.21339563862928349, "grad_norm": 4.10573673248291, "learning_rate": 4.931358230206152e-06, "loss": 0.1548784077167511, "step": 822 }, { "epoch": 0.2136552440290758, "grad_norm": 2.8835372924804688, "learning_rate": 4.931126150793767e-06, "loss": 0.19392892718315125, "step": 823 }, { "epoch": 0.21391484942886813, "grad_norm": 1.2267646789550781, "learning_rate": 4.930893685808219e-06, "loss": 0.14509283006191254, "step": 824 }, { "epoch": 0.21417445482866043, "grad_norm": 1.3650946617126465, "learning_rate": 4.930660835290603e-06, "loss": 0.15450166165828705, "step": 825 }, { "epoch": 0.21443406022845274, "grad_norm": 2.9655799865722656, "learning_rate": 4.930427599282082e-06, "loss": 0.18423667550086975, "step": 826 }, { "epoch": 0.21469366562824507, "grad_norm": 2.5984818935394287, "learning_rate": 4.930193977823887e-06, "loss": 0.15259478986263275, "step": 827 }, { "epoch": 0.21495327102803738, "grad_norm": 2.344999313354492, "learning_rate": 4.929959970957316e-06, "loss": 0.17014504969120026, "step": 828 }, { "epoch": 0.21521287642782969, "grad_norm": 0.5130178332328796, "learning_rate": 4.929725578723738e-06, "loss": 0.16165751218795776, "step": 829 }, { "epoch": 0.21547248182762202, "grad_norm": 0.4674050807952881, "learning_rate": 4.929490801164585e-06, "loss": 0.16553744673728943, "step": 830 }, { "epoch": 0.21573208722741433, "grad_norm": 1.6976370811462402, "learning_rate": 4.929255638321363e-06, "loss": 0.15096499025821686, "step": 831 }, { "epoch": 0.21599169262720663, "grad_norm": 2.4875924587249756, "learning_rate": 4.9290200902356424e-06, "loss": 0.14564689993858337, "step": 832 }, { "epoch": 0.21625129802699897, "grad_norm": 0.7840352058410645, "learning_rate": 4.928784156949062e-06, "loss": 0.1500212848186493, "step": 833 }, { "epoch": 0.21651090342679127, "grad_norm": 1.2811545133590698, "learning_rate": 4.928547838503331e-06, "loss": 0.1818496137857437, "step": 834 }, { "epoch": 0.21677050882658358, "grad_norm": 1.3292298316955566, "learning_rate": 4.928311134940224e-06, "loss": 0.1488065868616104, "step": 835 }, { "epoch": 0.21703011422637591, "grad_norm": 2.412620782852173, "learning_rate": 4.928074046301586e-06, "loss": 0.1825704425573349, "step": 836 }, { "epoch": 0.21728971962616822, "grad_norm": 1.0552743673324585, "learning_rate": 4.9278365726293274e-06, "loss": 0.17621132731437683, "step": 837 }, { "epoch": 0.21754932502596053, "grad_norm": 0.5277938842773438, "learning_rate": 4.927598713965428e-06, "loss": 0.1396506279706955, "step": 838 }, { "epoch": 0.21780893042575286, "grad_norm": 1.3932257890701294, "learning_rate": 4.927360470351937e-06, "loss": 0.18063753843307495, "step": 839 }, { "epoch": 0.21806853582554517, "grad_norm": 0.6315699815750122, "learning_rate": 4.927121841830969e-06, "loss": 0.13862556219100952, "step": 840 }, { "epoch": 0.21832814122533747, "grad_norm": 0.7107338905334473, "learning_rate": 4.926882828444709e-06, "loss": 0.13485383987426758, "step": 841 }, { "epoch": 0.2185877466251298, "grad_norm": 0.9064099788665771, "learning_rate": 4.92664343023541e-06, "loss": 0.14622095227241516, "step": 842 }, { "epoch": 0.21884735202492211, "grad_norm": 0.8718262314796448, "learning_rate": 4.926403647245391e-06, "loss": 0.16734805703163147, "step": 843 }, { "epoch": 0.21910695742471442, "grad_norm": 0.6349081993103027, "learning_rate": 4.926163479517039e-06, "loss": 0.1399080753326416, "step": 844 }, { "epoch": 0.21936656282450676, "grad_norm": 1.320677638053894, "learning_rate": 4.9259229270928115e-06, "loss": 0.16776303946971893, "step": 845 }, { "epoch": 0.21962616822429906, "grad_norm": 1.9720760583877563, "learning_rate": 4.925681990015233e-06, "loss": 0.1848866045475006, "step": 846 }, { "epoch": 0.21988577362409137, "grad_norm": 1.419724941253662, "learning_rate": 4.9254406683268955e-06, "loss": 0.19249513745307922, "step": 847 }, { "epoch": 0.2201453790238837, "grad_norm": 1.9187967777252197, "learning_rate": 4.925198962070459e-06, "loss": 0.18928544223308563, "step": 848 }, { "epoch": 0.220404984423676, "grad_norm": 1.0135228633880615, "learning_rate": 4.92495687128865e-06, "loss": 0.1609250009059906, "step": 849 }, { "epoch": 0.22066458982346832, "grad_norm": 2.3981542587280273, "learning_rate": 4.924714396024267e-06, "loss": 0.19439586997032166, "step": 850 }, { "epoch": 0.22092419522326065, "grad_norm": 1.1760437488555908, "learning_rate": 4.9244715363201735e-06, "loss": 0.17178454995155334, "step": 851 }, { "epoch": 0.22118380062305296, "grad_norm": 1.3780285120010376, "learning_rate": 4.924228292219301e-06, "loss": 0.16102690994739532, "step": 852 }, { "epoch": 0.22144340602284526, "grad_norm": 1.4131765365600586, "learning_rate": 4.923984663764649e-06, "loss": 0.13831500709056854, "step": 853 }, { "epoch": 0.2217030114226376, "grad_norm": 1.406933069229126, "learning_rate": 4.923740650999287e-06, "loss": 0.15757374465465546, "step": 854 }, { "epoch": 0.2219626168224299, "grad_norm": 2.3145363330841064, "learning_rate": 4.92349625396635e-06, "loss": 0.14667770266532898, "step": 855 }, { "epoch": 0.2222222222222222, "grad_norm": 2.136075973510742, "learning_rate": 4.923251472709041e-06, "loss": 0.1677953153848648, "step": 856 }, { "epoch": 0.22248182762201454, "grad_norm": 1.3486546277999878, "learning_rate": 4.923006307270632e-06, "loss": 0.1922416388988495, "step": 857 }, { "epoch": 0.22274143302180685, "grad_norm": 2.603461503982544, "learning_rate": 4.922760757694464e-06, "loss": 0.16285143792629242, "step": 858 }, { "epoch": 0.22300103842159916, "grad_norm": 2.296492099761963, "learning_rate": 4.922514824023944e-06, "loss": 0.1480262279510498, "step": 859 }, { "epoch": 0.2232606438213915, "grad_norm": 1.7232643365859985, "learning_rate": 4.922268506302547e-06, "loss": 0.15853428840637207, "step": 860 }, { "epoch": 0.2235202492211838, "grad_norm": 1.509212851524353, "learning_rate": 4.922021804573815e-06, "loss": 0.20186084508895874, "step": 861 }, { "epoch": 0.2237798546209761, "grad_norm": 1.3881230354309082, "learning_rate": 4.921774718881362e-06, "loss": 0.14508450031280518, "step": 862 }, { "epoch": 0.22403946002076844, "grad_norm": 3.990349292755127, "learning_rate": 4.921527249268866e-06, "loss": 0.21824991703033447, "step": 863 }, { "epoch": 0.22429906542056074, "grad_norm": 4.978282451629639, "learning_rate": 4.921279395780073e-06, "loss": 0.17524603009223938, "step": 864 }, { "epoch": 0.22455867082035305, "grad_norm": 1.1465132236480713, "learning_rate": 4.921031158458799e-06, "loss": 0.14214172959327698, "step": 865 }, { "epoch": 0.22481827622014539, "grad_norm": 2.6634774208068848, "learning_rate": 4.920782537348926e-06, "loss": 0.17089441418647766, "step": 866 }, { "epoch": 0.2250778816199377, "grad_norm": 2.116751194000244, "learning_rate": 4.9205335324944045e-06, "loss": 0.15111330151557922, "step": 867 }, { "epoch": 0.22533748701973, "grad_norm": 1.9710478782653809, "learning_rate": 4.9202841439392536e-06, "loss": 0.15985877811908722, "step": 868 }, { "epoch": 0.22559709241952233, "grad_norm": 3.198672294616699, "learning_rate": 4.920034371727559e-06, "loss": 0.1566621959209442, "step": 869 }, { "epoch": 0.22585669781931464, "grad_norm": 2.76448655128479, "learning_rate": 4.919784215903474e-06, "loss": 0.17617973685264587, "step": 870 }, { "epoch": 0.22611630321910695, "grad_norm": 3.4541797637939453, "learning_rate": 4.9195336765112215e-06, "loss": 0.1788228154182434, "step": 871 }, { "epoch": 0.22637590861889928, "grad_norm": 3.376054525375366, "learning_rate": 4.919282753595091e-06, "loss": 0.1480795443058014, "step": 872 }, { "epoch": 0.2266355140186916, "grad_norm": 2.0614752769470215, "learning_rate": 4.919031447199439e-06, "loss": 0.17548851668834686, "step": 873 }, { "epoch": 0.2268951194184839, "grad_norm": 0.6813103556632996, "learning_rate": 4.918779757368692e-06, "loss": 0.15111994743347168, "step": 874 }, { "epoch": 0.22715472481827623, "grad_norm": 0.41513389348983765, "learning_rate": 4.918527684147342e-06, "loss": 0.13536572456359863, "step": 875 }, { "epoch": 0.22741433021806853, "grad_norm": 2.625004529953003, "learning_rate": 4.918275227579951e-06, "loss": 0.14798414707183838, "step": 876 }, { "epoch": 0.22767393561786084, "grad_norm": 2.2326648235321045, "learning_rate": 4.918022387711145e-06, "loss": 0.1979634016752243, "step": 877 }, { "epoch": 0.22793354101765317, "grad_norm": 0.49263322353363037, "learning_rate": 4.917769164585623e-06, "loss": 0.13738134503364563, "step": 878 }, { "epoch": 0.22819314641744548, "grad_norm": 0.8527871966362, "learning_rate": 4.917515558248148e-06, "loss": 0.15004576742649078, "step": 879 }, { "epoch": 0.2284527518172378, "grad_norm": 0.5351042151451111, "learning_rate": 4.9172615687435525e-06, "loss": 0.14554332196712494, "step": 880 }, { "epoch": 0.22871235721703012, "grad_norm": 0.9159504175186157, "learning_rate": 4.917007196116734e-06, "loss": 0.14504140615463257, "step": 881 }, { "epoch": 0.22897196261682243, "grad_norm": 1.5896481275558472, "learning_rate": 4.916752440412661e-06, "loss": 0.18643847107887268, "step": 882 }, { "epoch": 0.22923156801661473, "grad_norm": 0.35685738921165466, "learning_rate": 4.91649730167637e-06, "loss": 0.14225655794143677, "step": 883 }, { "epoch": 0.22949117341640707, "grad_norm": 0.6316930055618286, "learning_rate": 4.916241779952961e-06, "loss": 0.18115606904029846, "step": 884 }, { "epoch": 0.22975077881619937, "grad_norm": 1.863762617111206, "learning_rate": 4.915985875287607e-06, "loss": 0.13812726736068726, "step": 885 }, { "epoch": 0.23001038421599168, "grad_norm": 0.7639143466949463, "learning_rate": 4.915729587725544e-06, "loss": 0.15206700563430786, "step": 886 }, { "epoch": 0.23026998961578402, "grad_norm": 1.4815291166305542, "learning_rate": 4.915472917312079e-06, "loss": 0.17149852216243744, "step": 887 }, { "epoch": 0.23052959501557632, "grad_norm": 1.015082836151123, "learning_rate": 4.915215864092584e-06, "loss": 0.13394126296043396, "step": 888 }, { "epoch": 0.23078920041536863, "grad_norm": 2.629842519760132, "learning_rate": 4.914958428112502e-06, "loss": 0.1866832673549652, "step": 889 }, { "epoch": 0.23104880581516096, "grad_norm": 0.7056467533111572, "learning_rate": 4.91470060941734e-06, "loss": 0.1467040479183197, "step": 890 }, { "epoch": 0.23130841121495327, "grad_norm": 1.3826062679290771, "learning_rate": 4.914442408052676e-06, "loss": 0.13621015846729279, "step": 891 }, { "epoch": 0.23156801661474558, "grad_norm": 0.9611387848854065, "learning_rate": 4.9141838240641535e-06, "loss": 0.1784762293100357, "step": 892 }, { "epoch": 0.2318276220145379, "grad_norm": 0.9753655195236206, "learning_rate": 4.913924857497483e-06, "loss": 0.13825656473636627, "step": 893 }, { "epoch": 0.23208722741433022, "grad_norm": 1.0267157554626465, "learning_rate": 4.913665508398447e-06, "loss": 0.14647556841373444, "step": 894 }, { "epoch": 0.23234683281412252, "grad_norm": 1.5746723413467407, "learning_rate": 4.913405776812888e-06, "loss": 0.16211144626140594, "step": 895 }, { "epoch": 0.23260643821391486, "grad_norm": 1.1142147779464722, "learning_rate": 4.913145662786725e-06, "loss": 0.13012292981147766, "step": 896 }, { "epoch": 0.23286604361370716, "grad_norm": 1.5722033977508545, "learning_rate": 4.912885166365938e-06, "loss": 0.1464678943157196, "step": 897 }, { "epoch": 0.23312564901349947, "grad_norm": 1.1690309047698975, "learning_rate": 4.912624287596577e-06, "loss": 0.1747724413871765, "step": 898 }, { "epoch": 0.2333852544132918, "grad_norm": 1.0004500150680542, "learning_rate": 4.91236302652476e-06, "loss": 0.14171338081359863, "step": 899 }, { "epoch": 0.2336448598130841, "grad_norm": 2.1799423694610596, "learning_rate": 4.9121013831966704e-06, "loss": 0.18714764714241028, "step": 900 }, { "epoch": 0.23390446521287642, "grad_norm": 2.454289436340332, "learning_rate": 4.911839357658562e-06, "loss": 0.14288479089736938, "step": 901 }, { "epoch": 0.23416407061266875, "grad_norm": 3.616914987564087, "learning_rate": 4.911576949956755e-06, "loss": 0.1668398380279541, "step": 902 }, { "epoch": 0.23442367601246106, "grad_norm": 2.130388021469116, "learning_rate": 4.9113141601376365e-06, "loss": 0.18327520787715912, "step": 903 }, { "epoch": 0.23468328141225336, "grad_norm": 1.359930157661438, "learning_rate": 4.911050988247663e-06, "loss": 0.14445428550243378, "step": 904 }, { "epoch": 0.2349428868120457, "grad_norm": 1.9825741052627563, "learning_rate": 4.910787434333355e-06, "loss": 0.14951658248901367, "step": 905 }, { "epoch": 0.235202492211838, "grad_norm": 1.5187861919403076, "learning_rate": 4.910523498441304e-06, "loss": 0.14730995893478394, "step": 906 }, { "epoch": 0.2354620976116303, "grad_norm": 1.071348786354065, "learning_rate": 4.91025918061817e-06, "loss": 0.13769418001174927, "step": 907 }, { "epoch": 0.23572170301142265, "grad_norm": 1.5793229341506958, "learning_rate": 4.909994480910674e-06, "loss": 0.14032980799674988, "step": 908 }, { "epoch": 0.23598130841121495, "grad_norm": 2.2542922496795654, "learning_rate": 4.909729399365613e-06, "loss": 0.13862526416778564, "step": 909 }, { "epoch": 0.23624091381100726, "grad_norm": 1.3901809453964233, "learning_rate": 4.909463936029843e-06, "loss": 0.14729014039039612, "step": 910 }, { "epoch": 0.2365005192107996, "grad_norm": 3.4196383953094482, "learning_rate": 4.909198090950296e-06, "loss": 0.20270639657974243, "step": 911 }, { "epoch": 0.2367601246105919, "grad_norm": 1.6259336471557617, "learning_rate": 4.908931864173965e-06, "loss": 0.14236509799957275, "step": 912 }, { "epoch": 0.2370197300103842, "grad_norm": 2.1467018127441406, "learning_rate": 4.908665255747914e-06, "loss": 0.14371594786643982, "step": 913 }, { "epoch": 0.23727933541017654, "grad_norm": 1.338462233543396, "learning_rate": 4.908398265719273e-06, "loss": 0.13697780668735504, "step": 914 }, { "epoch": 0.23753894080996885, "grad_norm": 2.4177443981170654, "learning_rate": 4.908130894135238e-06, "loss": 0.15808740258216858, "step": 915 }, { "epoch": 0.23779854620976115, "grad_norm": 1.564369797706604, "learning_rate": 4.907863141043076e-06, "loss": 0.14182192087173462, "step": 916 }, { "epoch": 0.2380581516095535, "grad_norm": 1.415053129196167, "learning_rate": 4.90759500649012e-06, "loss": 0.1659221351146698, "step": 917 }, { "epoch": 0.2383177570093458, "grad_norm": 2.0518693923950195, "learning_rate": 4.9073264905237685e-06, "loss": 0.19653625786304474, "step": 918 }, { "epoch": 0.2385773624091381, "grad_norm": 0.5638858675956726, "learning_rate": 4.907057593191489e-06, "loss": 0.15493014454841614, "step": 919 }, { "epoch": 0.23883696780893043, "grad_norm": 2.4973456859588623, "learning_rate": 4.906788314540819e-06, "loss": 0.15576110780239105, "step": 920 }, { "epoch": 0.23909657320872274, "grad_norm": 1.6266127824783325, "learning_rate": 4.906518654619358e-06, "loss": 0.1493910253047943, "step": 921 }, { "epoch": 0.23935617860851505, "grad_norm": 0.7643986940383911, "learning_rate": 4.906248613474776e-06, "loss": 0.1900617480278015, "step": 922 }, { "epoch": 0.23961578400830738, "grad_norm": 1.4025605916976929, "learning_rate": 4.905978191154811e-06, "loss": 0.19739297032356262, "step": 923 }, { "epoch": 0.2398753894080997, "grad_norm": 0.9353542327880859, "learning_rate": 4.905707387707267e-06, "loss": 0.16525757312774658, "step": 924 }, { "epoch": 0.240134994807892, "grad_norm": 0.9269739389419556, "learning_rate": 4.905436203180016e-06, "loss": 0.15365970134735107, "step": 925 }, { "epoch": 0.24039460020768433, "grad_norm": 1.2174197435379028, "learning_rate": 4.9051646376209976e-06, "loss": 0.13566206395626068, "step": 926 }, { "epoch": 0.24065420560747663, "grad_norm": 0.609948992729187, "learning_rate": 4.904892691078218e-06, "loss": 0.14558765292167664, "step": 927 }, { "epoch": 0.24091381100726894, "grad_norm": 0.7669448256492615, "learning_rate": 4.904620363599752e-06, "loss": 0.15251272916793823, "step": 928 }, { "epoch": 0.24117341640706128, "grad_norm": 0.8510509133338928, "learning_rate": 4.904347655233738e-06, "loss": 0.15701495110988617, "step": 929 }, { "epoch": 0.24143302180685358, "grad_norm": 0.7088321447372437, "learning_rate": 4.904074566028388e-06, "loss": 0.14184731245040894, "step": 930 }, { "epoch": 0.2416926272066459, "grad_norm": 1.0291070938110352, "learning_rate": 4.903801096031976e-06, "loss": 0.20367267727851868, "step": 931 }, { "epoch": 0.24195223260643822, "grad_norm": 0.3826223313808441, "learning_rate": 4.903527245292847e-06, "loss": 0.14735159277915955, "step": 932 }, { "epoch": 0.24221183800623053, "grad_norm": 0.9909664392471313, "learning_rate": 4.903253013859408e-06, "loss": 0.14777296781539917, "step": 933 }, { "epoch": 0.24247144340602284, "grad_norm": 0.671820342540741, "learning_rate": 4.9029784017801404e-06, "loss": 0.1351774036884308, "step": 934 }, { "epoch": 0.24273104880581517, "grad_norm": 0.8034611344337463, "learning_rate": 4.902703409103588e-06, "loss": 0.14173367619514465, "step": 935 }, { "epoch": 0.24299065420560748, "grad_norm": 1.0038059949874878, "learning_rate": 4.902428035878363e-06, "loss": 0.15171334147453308, "step": 936 }, { "epoch": 0.24325025960539978, "grad_norm": 2.491426944732666, "learning_rate": 4.902152282153145e-06, "loss": 0.17071136832237244, "step": 937 }, { "epoch": 0.24350986500519212, "grad_norm": 3.056750535964966, "learning_rate": 4.901876147976682e-06, "loss": 0.17919597029685974, "step": 938 }, { "epoch": 0.24376947040498442, "grad_norm": 1.1621748208999634, "learning_rate": 4.901599633397787e-06, "loss": 0.12835197150707245, "step": 939 }, { "epoch": 0.24402907580477673, "grad_norm": 1.6802245378494263, "learning_rate": 4.901322738465341e-06, "loss": 0.1882110834121704, "step": 940 }, { "epoch": 0.24428868120456906, "grad_norm": 1.3099210262298584, "learning_rate": 4.901045463228295e-06, "loss": 0.18541325628757477, "step": 941 }, { "epoch": 0.24454828660436137, "grad_norm": 1.4092720746994019, "learning_rate": 4.900767807735664e-06, "loss": 0.17228634655475616, "step": 942 }, { "epoch": 0.24480789200415368, "grad_norm": 2.8841006755828857, "learning_rate": 4.90048977203653e-06, "loss": 0.16591240465641022, "step": 943 }, { "epoch": 0.245067497403946, "grad_norm": 0.9317487478256226, "learning_rate": 4.900211356180044e-06, "loss": 0.13793838024139404, "step": 944 }, { "epoch": 0.24532710280373832, "grad_norm": 0.9634539484977722, "learning_rate": 4.899932560215423e-06, "loss": 0.15970173478126526, "step": 945 }, { "epoch": 0.24558670820353062, "grad_norm": 1.9143130779266357, "learning_rate": 4.899653384191952e-06, "loss": 0.13931992650032043, "step": 946 }, { "epoch": 0.24584631360332296, "grad_norm": 1.5854843854904175, "learning_rate": 4.899373828158984e-06, "loss": 0.1841861754655838, "step": 947 }, { "epoch": 0.24610591900311526, "grad_norm": 1.9492520093917847, "learning_rate": 4.899093892165938e-06, "loss": 0.15070059895515442, "step": 948 }, { "epoch": 0.24636552440290757, "grad_norm": 0.6940683722496033, "learning_rate": 4.898813576262299e-06, "loss": 0.13698682188987732, "step": 949 }, { "epoch": 0.2466251298026999, "grad_norm": 3.0365312099456787, "learning_rate": 4.8985328804976205e-06, "loss": 0.1814294457435608, "step": 950 }, { "epoch": 0.2468847352024922, "grad_norm": 1.8188533782958984, "learning_rate": 4.898251804921523e-06, "loss": 0.13000187277793884, "step": 951 }, { "epoch": 0.24714434060228452, "grad_norm": 0.870554506778717, "learning_rate": 4.897970349583697e-06, "loss": 0.14729085564613342, "step": 952 }, { "epoch": 0.24740394600207685, "grad_norm": 1.545411467552185, "learning_rate": 4.8976885145338935e-06, "loss": 0.14955571293830872, "step": 953 }, { "epoch": 0.24766355140186916, "grad_norm": 1.9508568048477173, "learning_rate": 4.897406299821936e-06, "loss": 0.19340509176254272, "step": 954 }, { "epoch": 0.24792315680166147, "grad_norm": 2.181702136993408, "learning_rate": 4.897123705497714e-06, "loss": 0.1981184333562851, "step": 955 }, { "epoch": 0.2481827622014538, "grad_norm": 0.9693073630332947, "learning_rate": 4.896840731611185e-06, "loss": 0.16782930493354797, "step": 956 }, { "epoch": 0.2484423676012461, "grad_norm": 1.3175331354141235, "learning_rate": 4.896557378212369e-06, "loss": 0.14356666803359985, "step": 957 }, { "epoch": 0.2487019730010384, "grad_norm": 1.8877848386764526, "learning_rate": 4.8962736453513595e-06, "loss": 0.17393571138381958, "step": 958 }, { "epoch": 0.24896157840083075, "grad_norm": 1.9234546422958374, "learning_rate": 4.895989533078311e-06, "loss": 0.16357913613319397, "step": 959 }, { "epoch": 0.24922118380062305, "grad_norm": 1.1294190883636475, "learning_rate": 4.895705041443452e-06, "loss": 0.14554476737976074, "step": 960 }, { "epoch": 0.24948078920041536, "grad_norm": 0.5450502038002014, "learning_rate": 4.89542017049707e-06, "loss": 0.14529570937156677, "step": 961 }, { "epoch": 0.2497403946002077, "grad_norm": 1.924418568611145, "learning_rate": 4.895134920289526e-06, "loss": 0.1376689076423645, "step": 962 }, { "epoch": 0.25, "grad_norm": 1.1065679788589478, "learning_rate": 4.894849290871245e-06, "loss": 0.15048331022262573, "step": 963 }, { "epoch": 0.2502596053997923, "grad_norm": 1.4948583841323853, "learning_rate": 4.89456328229272e-06, "loss": 0.14783136546611786, "step": 964 }, { "epoch": 0.2505192107995846, "grad_norm": 0.8322339653968811, "learning_rate": 4.894276894604511e-06, "loss": 0.13507916033267975, "step": 965 }, { "epoch": 0.2507788161993769, "grad_norm": 3.124741554260254, "learning_rate": 4.893990127857244e-06, "loss": 0.16489987075328827, "step": 966 }, { "epoch": 0.2510384215991693, "grad_norm": 1.6149213314056396, "learning_rate": 4.893702982101614e-06, "loss": 0.1641269326210022, "step": 967 }, { "epoch": 0.2512980269989616, "grad_norm": 1.1022223234176636, "learning_rate": 4.89341545738838e-06, "loss": 0.15463221073150635, "step": 968 }, { "epoch": 0.2515576323987539, "grad_norm": 2.6485536098480225, "learning_rate": 4.893127553768372e-06, "loss": 0.19172310829162598, "step": 969 }, { "epoch": 0.2518172377985462, "grad_norm": 1.4485723972320557, "learning_rate": 4.892839271292483e-06, "loss": 0.19386301934719086, "step": 970 }, { "epoch": 0.2520768431983385, "grad_norm": 1.9379820823669434, "learning_rate": 4.892550610011675e-06, "loss": 0.13389179110527039, "step": 971 }, { "epoch": 0.2523364485981308, "grad_norm": 2.689387321472168, "learning_rate": 4.8922615699769775e-06, "loss": 0.16044339537620544, "step": 972 }, { "epoch": 0.2525960539979232, "grad_norm": 1.819043517112732, "learning_rate": 4.891972151239486e-06, "loss": 0.1471046507358551, "step": 973 }, { "epoch": 0.2528556593977155, "grad_norm": 0.964270293712616, "learning_rate": 4.891682353850362e-06, "loss": 0.1396956890821457, "step": 974 }, { "epoch": 0.2531152647975078, "grad_norm": 1.321905493736267, "learning_rate": 4.8913921778608355e-06, "loss": 0.137982577085495, "step": 975 }, { "epoch": 0.2533748701973001, "grad_norm": 1.664421558380127, "learning_rate": 4.891101623322205e-06, "loss": 0.15991315245628357, "step": 976 }, { "epoch": 0.2536344755970924, "grad_norm": 1.852707862854004, "learning_rate": 4.890810690285831e-06, "loss": 0.13616126775741577, "step": 977 }, { "epoch": 0.2538940809968847, "grad_norm": 0.8337534666061401, "learning_rate": 4.8905193788031454e-06, "loss": 0.16361995041370392, "step": 978 }, { "epoch": 0.25415368639667707, "grad_norm": 2.2937111854553223, "learning_rate": 4.890227688925645e-06, "loss": 0.1582486778497696, "step": 979 }, { "epoch": 0.2544132917964694, "grad_norm": 1.7392909526824951, "learning_rate": 4.8899356207048945e-06, "loss": 0.1843847632408142, "step": 980 }, { "epoch": 0.2546728971962617, "grad_norm": 1.0264674425125122, "learning_rate": 4.889643174192523e-06, "loss": 0.14382846653461456, "step": 981 }, { "epoch": 0.254932502596054, "grad_norm": 1.5097119808197021, "learning_rate": 4.8893503494402305e-06, "loss": 0.18153443932533264, "step": 982 }, { "epoch": 0.2551921079958463, "grad_norm": 0.9693983793258667, "learning_rate": 4.889057146499781e-06, "loss": 0.20038016140460968, "step": 983 }, { "epoch": 0.2554517133956386, "grad_norm": 1.8553544282913208, "learning_rate": 4.888763565423004e-06, "loss": 0.16542229056358337, "step": 984 }, { "epoch": 0.25571131879543096, "grad_norm": 2.5069360733032227, "learning_rate": 4.888469606261803e-06, "loss": 0.13745853304862976, "step": 985 }, { "epoch": 0.25597092419522327, "grad_norm": 2.1626429557800293, "learning_rate": 4.888175269068138e-06, "loss": 0.14465659856796265, "step": 986 }, { "epoch": 0.2562305295950156, "grad_norm": 0.6637300848960876, "learning_rate": 4.887880553894044e-06, "loss": 0.12930649518966675, "step": 987 }, { "epoch": 0.2564901349948079, "grad_norm": 2.9808197021484375, "learning_rate": 4.887585460791618e-06, "loss": 0.1741601526737213, "step": 988 }, { "epoch": 0.2567497403946002, "grad_norm": 0.6041674017906189, "learning_rate": 4.887289989813029e-06, "loss": 0.12818369269371033, "step": 989 }, { "epoch": 0.2570093457943925, "grad_norm": 2.9273014068603516, "learning_rate": 4.886994141010505e-06, "loss": 0.17196139693260193, "step": 990 }, { "epoch": 0.25726895119418486, "grad_norm": 1.7007828950881958, "learning_rate": 4.886697914436348e-06, "loss": 0.19921070337295532, "step": 991 }, { "epoch": 0.25752855659397716, "grad_norm": 1.6995658874511719, "learning_rate": 4.8864013101429235e-06, "loss": 0.14530307054519653, "step": 992 }, { "epoch": 0.25778816199376947, "grad_norm": 2.2296648025512695, "learning_rate": 4.886104328182665e-06, "loss": 0.1570766270160675, "step": 993 }, { "epoch": 0.2580477673935618, "grad_norm": 1.6330751180648804, "learning_rate": 4.885806968608072e-06, "loss": 0.16394492983818054, "step": 994 }, { "epoch": 0.2583073727933541, "grad_norm": 1.0820616483688354, "learning_rate": 4.8855092314717095e-06, "loss": 0.15605324506759644, "step": 995 }, { "epoch": 0.2585669781931464, "grad_norm": 0.6836034059524536, "learning_rate": 4.885211116826212e-06, "loss": 0.14083543419837952, "step": 996 }, { "epoch": 0.25882658359293875, "grad_norm": 1.9452333450317383, "learning_rate": 4.884912624724279e-06, "loss": 0.17048175632953644, "step": 997 }, { "epoch": 0.25908618899273106, "grad_norm": 3.5826308727264404, "learning_rate": 4.884613755218677e-06, "loss": 0.18759799003601074, "step": 998 }, { "epoch": 0.25934579439252337, "grad_norm": 0.8194742798805237, "learning_rate": 4.884314508362239e-06, "loss": 0.13723066449165344, "step": 999 }, { "epoch": 0.25960539979231567, "grad_norm": 0.6697386503219604, "learning_rate": 4.884014884207865e-06, "loss": 0.15511801838874817, "step": 1000 }, { "epoch": 0.259865005192108, "grad_norm": 2.0789239406585693, "learning_rate": 4.8837148828085225e-06, "loss": 0.18732351064682007, "step": 1001 }, { "epoch": 0.2601246105919003, "grad_norm": 1.8136144876480103, "learning_rate": 4.883414504217244e-06, "loss": 0.14010116457939148, "step": 1002 }, { "epoch": 0.26038421599169265, "grad_norm": 1.5994802713394165, "learning_rate": 4.88311374848713e-06, "loss": 0.2130383849143982, "step": 1003 }, { "epoch": 0.26064382139148495, "grad_norm": 1.8705700635910034, "learning_rate": 4.882812615671347e-06, "loss": 0.1476249098777771, "step": 1004 }, { "epoch": 0.26090342679127726, "grad_norm": 1.9183529615402222, "learning_rate": 4.882511105823128e-06, "loss": 0.16040962934494019, "step": 1005 }, { "epoch": 0.26116303219106957, "grad_norm": 2.2808456420898438, "learning_rate": 4.882209218995775e-06, "loss": 0.13601082563400269, "step": 1006 }, { "epoch": 0.2614226375908619, "grad_norm": 1.5941996574401855, "learning_rate": 4.881906955242652e-06, "loss": 0.1492169201374054, "step": 1007 }, { "epoch": 0.2616822429906542, "grad_norm": 0.9945836067199707, "learning_rate": 4.881604314617194e-06, "loss": 0.13992589712142944, "step": 1008 }, { "epoch": 0.26194184839044654, "grad_norm": 0.8912205696105957, "learning_rate": 4.8813012971729e-06, "loss": 0.17387866973876953, "step": 1009 }, { "epoch": 0.26220145379023885, "grad_norm": 2.7195487022399902, "learning_rate": 4.880997902963337e-06, "loss": 0.16648834943771362, "step": 1010 }, { "epoch": 0.26246105919003115, "grad_norm": 1.8391979932785034, "learning_rate": 4.880694132042138e-06, "loss": 0.16807806491851807, "step": 1011 }, { "epoch": 0.26272066458982346, "grad_norm": 1.168229103088379, "learning_rate": 4.8803899844630034e-06, "loss": 0.14171475172042847, "step": 1012 }, { "epoch": 0.26298026998961577, "grad_norm": 0.9214970469474792, "learning_rate": 4.880085460279699e-06, "loss": 0.13771125674247742, "step": 1013 }, { "epoch": 0.2632398753894081, "grad_norm": 1.3743840456008911, "learning_rate": 4.879780559546058e-06, "loss": 0.1455964744091034, "step": 1014 }, { "epoch": 0.26349948078920044, "grad_norm": 1.8704017400741577, "learning_rate": 4.879475282315978e-06, "loss": 0.1377449333667755, "step": 1015 }, { "epoch": 0.26375908618899274, "grad_norm": 3.3179328441619873, "learning_rate": 4.879169628643428e-06, "loss": 0.15527889132499695, "step": 1016 }, { "epoch": 0.26401869158878505, "grad_norm": 1.3816046714782715, "learning_rate": 4.878863598582439e-06, "loss": 0.17356151342391968, "step": 1017 }, { "epoch": 0.26427829698857735, "grad_norm": 1.774409532546997, "learning_rate": 4.87855719218711e-06, "loss": 0.13637590408325195, "step": 1018 }, { "epoch": 0.26453790238836966, "grad_norm": 0.5008180737495422, "learning_rate": 4.878250409511608e-06, "loss": 0.14232738316059113, "step": 1019 }, { "epoch": 0.26479750778816197, "grad_norm": 1.4409492015838623, "learning_rate": 4.877943250610163e-06, "loss": 0.14993718266487122, "step": 1020 }, { "epoch": 0.26505711318795433, "grad_norm": 2.9080235958099365, "learning_rate": 4.8776357155370745e-06, "loss": 0.18959955871105194, "step": 1021 }, { "epoch": 0.26531671858774664, "grad_norm": 1.6566802263259888, "learning_rate": 4.877327804346709e-06, "loss": 0.17185357213020325, "step": 1022 }, { "epoch": 0.26557632398753894, "grad_norm": 0.6641227006912231, "learning_rate": 4.877019517093497e-06, "loss": 0.13057395815849304, "step": 1023 }, { "epoch": 0.26583592938733125, "grad_norm": 0.8371433019638062, "learning_rate": 4.876710853831935e-06, "loss": 0.1367177963256836, "step": 1024 }, { "epoch": 0.26609553478712356, "grad_norm": 0.9757758378982544, "learning_rate": 4.8764018146165905e-06, "loss": 0.14300376176834106, "step": 1025 }, { "epoch": 0.26635514018691586, "grad_norm": 1.3751721382141113, "learning_rate": 4.876092399502093e-06, "loss": 0.1392664611339569, "step": 1026 }, { "epoch": 0.2666147455867082, "grad_norm": 1.3350005149841309, "learning_rate": 4.87578260854314e-06, "loss": 0.12907196581363678, "step": 1027 }, { "epoch": 0.26687435098650053, "grad_norm": 0.9669610857963562, "learning_rate": 4.875472441794496e-06, "loss": 0.16177397966384888, "step": 1028 }, { "epoch": 0.26713395638629284, "grad_norm": 1.0473089218139648, "learning_rate": 4.875161899310992e-06, "loss": 0.14671733975410461, "step": 1029 }, { "epoch": 0.26739356178608514, "grad_norm": 0.9718663096427917, "learning_rate": 4.874850981147523e-06, "loss": 0.164170041680336, "step": 1030 }, { "epoch": 0.26765316718587745, "grad_norm": 2.3255455493927, "learning_rate": 4.874539687359054e-06, "loss": 0.1895129531621933, "step": 1031 }, { "epoch": 0.26791277258566976, "grad_norm": 1.1727097034454346, "learning_rate": 4.874228018000613e-06, "loss": 0.1927472949028015, "step": 1032 }, { "epoch": 0.2681723779854621, "grad_norm": 1.0520060062408447, "learning_rate": 4.8739159731272975e-06, "loss": 0.13873408734798431, "step": 1033 }, { "epoch": 0.2684319833852544, "grad_norm": 0.7594144940376282, "learning_rate": 4.873603552794268e-06, "loss": 0.16083338856697083, "step": 1034 }, { "epoch": 0.26869158878504673, "grad_norm": 1.056965947151184, "learning_rate": 4.873290757056756e-06, "loss": 0.14387866854667664, "step": 1035 }, { "epoch": 0.26895119418483904, "grad_norm": 2.525404453277588, "learning_rate": 4.872977585970054e-06, "loss": 0.14912661910057068, "step": 1036 }, { "epoch": 0.26921079958463134, "grad_norm": 0.7942218780517578, "learning_rate": 4.872664039589525e-06, "loss": 0.1648775041103363, "step": 1037 }, { "epoch": 0.26947040498442365, "grad_norm": 0.9801016449928284, "learning_rate": 4.872350117970596e-06, "loss": 0.18329991400241852, "step": 1038 }, { "epoch": 0.269730010384216, "grad_norm": 0.6737005710601807, "learning_rate": 4.8720358211687615e-06, "loss": 0.1734044849872589, "step": 1039 }, { "epoch": 0.2699896157840083, "grad_norm": 1.619390606880188, "learning_rate": 4.871721149239583e-06, "loss": 0.17991262674331665, "step": 1040 }, { "epoch": 0.2702492211838006, "grad_norm": 1.425581693649292, "learning_rate": 4.8714061022386845e-06, "loss": 0.1720212996006012, "step": 1041 }, { "epoch": 0.27050882658359293, "grad_norm": 2.4509317874908447, "learning_rate": 4.8710906802217624e-06, "loss": 0.19893011450767517, "step": 1042 }, { "epoch": 0.27076843198338524, "grad_norm": 1.1781842708587646, "learning_rate": 4.8707748832445745e-06, "loss": 0.17685258388519287, "step": 1043 }, { "epoch": 0.27102803738317754, "grad_norm": 0.4437243342399597, "learning_rate": 4.870458711362946e-06, "loss": 0.1425231695175171, "step": 1044 }, { "epoch": 0.2712876427829699, "grad_norm": 0.8560909032821655, "learning_rate": 4.8701421646327694e-06, "loss": 0.1487748771905899, "step": 1045 }, { "epoch": 0.2715472481827622, "grad_norm": 5.497195243835449, "learning_rate": 4.869825243110003e-06, "loss": 0.2042926847934723, "step": 1046 }, { "epoch": 0.2718068535825545, "grad_norm": 1.10425865650177, "learning_rate": 4.869507946850671e-06, "loss": 0.17800132930278778, "step": 1047 }, { "epoch": 0.2720664589823468, "grad_norm": 1.0679246187210083, "learning_rate": 4.869190275910865e-06, "loss": 0.17658880352973938, "step": 1048 }, { "epoch": 0.27232606438213913, "grad_norm": 2.710750102996826, "learning_rate": 4.868872230346741e-06, "loss": 0.1557309329509735, "step": 1049 }, { "epoch": 0.27258566978193144, "grad_norm": 2.8770875930786133, "learning_rate": 4.8685538102145215e-06, "loss": 0.1634281873703003, "step": 1050 }, { "epoch": 0.2728452751817238, "grad_norm": 1.873443603515625, "learning_rate": 4.868235015570497e-06, "loss": 0.130384624004364, "step": 1051 }, { "epoch": 0.2731048805815161, "grad_norm": 0.854224681854248, "learning_rate": 4.867915846471024e-06, "loss": 0.13050352036952972, "step": 1052 }, { "epoch": 0.2733644859813084, "grad_norm": 2.1813087463378906, "learning_rate": 4.867596302972523e-06, "loss": 0.15110306441783905, "step": 1053 }, { "epoch": 0.2736240913811007, "grad_norm": 1.9521028995513916, "learning_rate": 4.867276385131482e-06, "loss": 0.15644711256027222, "step": 1054 }, { "epoch": 0.273883696780893, "grad_norm": 1.392275094985962, "learning_rate": 4.866956093004454e-06, "loss": 0.1409585177898407, "step": 1055 }, { "epoch": 0.27414330218068533, "grad_norm": 2.0848801136016846, "learning_rate": 4.866635426648062e-06, "loss": 0.18296080827713013, "step": 1056 }, { "epoch": 0.2744029075804777, "grad_norm": 1.0447652339935303, "learning_rate": 4.866314386118991e-06, "loss": 0.16476841270923615, "step": 1057 }, { "epoch": 0.27466251298027, "grad_norm": 2.47023344039917, "learning_rate": 4.865992971473993e-06, "loss": 0.1743568331003189, "step": 1058 }, { "epoch": 0.2749221183800623, "grad_norm": 2.5748443603515625, "learning_rate": 4.865671182769888e-06, "loss": 0.16083557903766632, "step": 1059 }, { "epoch": 0.2751817237798546, "grad_norm": 1.344881534576416, "learning_rate": 4.865349020063561e-06, "loss": 0.18373796343803406, "step": 1060 }, { "epoch": 0.2754413291796469, "grad_norm": 1.743715763092041, "learning_rate": 4.865026483411961e-06, "loss": 0.18988138437271118, "step": 1061 }, { "epoch": 0.2757009345794392, "grad_norm": 1.54763662815094, "learning_rate": 4.864703572872107e-06, "loss": 0.1717803031206131, "step": 1062 }, { "epoch": 0.2759605399792316, "grad_norm": 1.089342474937439, "learning_rate": 4.864380288501082e-06, "loss": 0.151065856218338, "step": 1063 }, { "epoch": 0.2762201453790239, "grad_norm": 1.1923902034759521, "learning_rate": 4.864056630356034e-06, "loss": 0.14818570017814636, "step": 1064 }, { "epoch": 0.2764797507788162, "grad_norm": 3.1158695220947266, "learning_rate": 4.86373259849418e-06, "loss": 0.17205025255680084, "step": 1065 }, { "epoch": 0.2767393561786085, "grad_norm": 1.7462143898010254, "learning_rate": 4.8634081929728e-06, "loss": 0.14606422185897827, "step": 1066 }, { "epoch": 0.2769989615784008, "grad_norm": 0.8032808899879456, "learning_rate": 4.863083413849243e-06, "loss": 0.14602352678775787, "step": 1067 }, { "epoch": 0.2772585669781931, "grad_norm": 2.05171275138855, "learning_rate": 4.862758261180921e-06, "loss": 0.20075467228889465, "step": 1068 }, { "epoch": 0.2775181723779855, "grad_norm": 1.4480253458023071, "learning_rate": 4.8624327350253145e-06, "loss": 0.15925566852092743, "step": 1069 }, { "epoch": 0.2777777777777778, "grad_norm": 0.3897758424282074, "learning_rate": 4.862106835439968e-06, "loss": 0.1487690508365631, "step": 1070 }, { "epoch": 0.2780373831775701, "grad_norm": 0.8215789794921875, "learning_rate": 4.861780562482495e-06, "loss": 0.14147615432739258, "step": 1071 }, { "epoch": 0.2782969885773624, "grad_norm": 2.7207000255584717, "learning_rate": 4.861453916210571e-06, "loss": 0.19298720359802246, "step": 1072 }, { "epoch": 0.2785565939771547, "grad_norm": 1.2010146379470825, "learning_rate": 4.861126896681941e-06, "loss": 0.16159102320671082, "step": 1073 }, { "epoch": 0.278816199376947, "grad_norm": 0.6439154744148254, "learning_rate": 4.8607995039544135e-06, "loss": 0.13089123368263245, "step": 1074 }, { "epoch": 0.2790758047767394, "grad_norm": 0.5921880602836609, "learning_rate": 4.8604717380858655e-06, "loss": 0.13300879299640656, "step": 1075 }, { "epoch": 0.2793354101765317, "grad_norm": 1.5396733283996582, "learning_rate": 4.8601435991342375e-06, "loss": 0.16122862696647644, "step": 1076 }, { "epoch": 0.279595015576324, "grad_norm": 1.7361516952514648, "learning_rate": 4.859815087157536e-06, "loss": 0.18353936076164246, "step": 1077 }, { "epoch": 0.2798546209761163, "grad_norm": 0.9734699130058289, "learning_rate": 4.859486202213836e-06, "loss": 0.1924723982810974, "step": 1078 }, { "epoch": 0.2801142263759086, "grad_norm": 1.0797611474990845, "learning_rate": 4.859156944361276e-06, "loss": 0.18339112401008606, "step": 1079 }, { "epoch": 0.2803738317757009, "grad_norm": 2.0898337364196777, "learning_rate": 4.858827313658062e-06, "loss": 0.14428365230560303, "step": 1080 }, { "epoch": 0.28063343717549327, "grad_norm": 3.636979579925537, "learning_rate": 4.858497310162465e-06, "loss": 0.1539073884487152, "step": 1081 }, { "epoch": 0.2808930425752856, "grad_norm": 1.2775094509124756, "learning_rate": 4.858166933932821e-06, "loss": 0.14128880202770233, "step": 1082 }, { "epoch": 0.2811526479750779, "grad_norm": 0.690368115901947, "learning_rate": 4.857836185027534e-06, "loss": 0.1528564989566803, "step": 1083 }, { "epoch": 0.2814122533748702, "grad_norm": 2.6566269397735596, "learning_rate": 4.857505063505073e-06, "loss": 0.20349499583244324, "step": 1084 }, { "epoch": 0.2816718587746625, "grad_norm": 1.7350600957870483, "learning_rate": 4.857173569423971e-06, "loss": 0.18946731090545654, "step": 1085 }, { "epoch": 0.2819314641744548, "grad_norm": 1.1536967754364014, "learning_rate": 4.856841702842831e-06, "loss": 0.17464393377304077, "step": 1086 }, { "epoch": 0.28219106957424717, "grad_norm": 0.8933414816856384, "learning_rate": 4.856509463820318e-06, "loss": 0.1908968687057495, "step": 1087 }, { "epoch": 0.2824506749740395, "grad_norm": 1.6233513355255127, "learning_rate": 4.856176852415164e-06, "loss": 0.1343805193901062, "step": 1088 }, { "epoch": 0.2827102803738318, "grad_norm": 0.8036661744117737, "learning_rate": 4.855843868686168e-06, "loss": 0.1641901731491089, "step": 1089 }, { "epoch": 0.2829698857736241, "grad_norm": 1.109407663345337, "learning_rate": 4.855510512692194e-06, "loss": 0.16002210974693298, "step": 1090 }, { "epoch": 0.2832294911734164, "grad_norm": 4.278115749359131, "learning_rate": 4.8551767844921695e-06, "loss": 0.17144103348255157, "step": 1091 }, { "epoch": 0.2834890965732087, "grad_norm": 0.6545568704605103, "learning_rate": 4.854842684145093e-06, "loss": 0.13552474975585938, "step": 1092 }, { "epoch": 0.28374870197300106, "grad_norm": 0.9572181701660156, "learning_rate": 4.854508211710025e-06, "loss": 0.14720818400382996, "step": 1093 }, { "epoch": 0.28400830737279337, "grad_norm": 1.3410780429840088, "learning_rate": 4.85417336724609e-06, "loss": 0.13695526123046875, "step": 1094 }, { "epoch": 0.2842679127725857, "grad_norm": 1.3540083169937134, "learning_rate": 4.853838150812485e-06, "loss": 0.1397550106048584, "step": 1095 }, { "epoch": 0.284527518172378, "grad_norm": 1.2443681955337524, "learning_rate": 4.853502562468464e-06, "loss": 0.16229699552059174, "step": 1096 }, { "epoch": 0.2847871235721703, "grad_norm": 1.029240369796753, "learning_rate": 4.853166602273356e-06, "loss": 0.15383687615394592, "step": 1097 }, { "epoch": 0.2850467289719626, "grad_norm": 0.7142364978790283, "learning_rate": 4.852830270286548e-06, "loss": 0.17517909407615662, "step": 1098 }, { "epoch": 0.28530633437175495, "grad_norm": 1.9085488319396973, "learning_rate": 4.852493566567496e-06, "loss": 0.17165407538414001, "step": 1099 }, { "epoch": 0.28556593977154726, "grad_norm": 1.545880675315857, "learning_rate": 4.852156491175724e-06, "loss": 0.1441657394170761, "step": 1100 }, { "epoch": 0.28582554517133957, "grad_norm": 1.4200685024261475, "learning_rate": 4.851819044170816e-06, "loss": 0.1630358099937439, "step": 1101 }, { "epoch": 0.2860851505711319, "grad_norm": 0.7125799655914307, "learning_rate": 4.851481225612426e-06, "loss": 0.16098716855049133, "step": 1102 }, { "epoch": 0.2863447559709242, "grad_norm": 1.9995189905166626, "learning_rate": 4.851143035560274e-06, "loss": 0.14858555793762207, "step": 1103 }, { "epoch": 0.2866043613707165, "grad_norm": 0.9386275410652161, "learning_rate": 4.850804474074143e-06, "loss": 0.14757969975471497, "step": 1104 }, { "epoch": 0.28686396677050885, "grad_norm": 1.5498201847076416, "learning_rate": 4.8504655412138816e-06, "loss": 0.14553523063659668, "step": 1105 }, { "epoch": 0.28712357217030116, "grad_norm": 1.697184443473816, "learning_rate": 4.850126237039408e-06, "loss": 0.17127463221549988, "step": 1106 }, { "epoch": 0.28738317757009346, "grad_norm": 0.8095633387565613, "learning_rate": 4.849786561610702e-06, "loss": 0.15744435787200928, "step": 1107 }, { "epoch": 0.28764278296988577, "grad_norm": 0.5121562480926514, "learning_rate": 4.84944651498781e-06, "loss": 0.13821908831596375, "step": 1108 }, { "epoch": 0.2879023883696781, "grad_norm": 0.660726010799408, "learning_rate": 4.849106097230845e-06, "loss": 0.14457234740257263, "step": 1109 }, { "epoch": 0.2881619937694704, "grad_norm": 1.1236283779144287, "learning_rate": 4.848765308399987e-06, "loss": 0.16415366530418396, "step": 1110 }, { "epoch": 0.28842159916926274, "grad_norm": 1.6281708478927612, "learning_rate": 4.848424148555476e-06, "loss": 0.16514816880226135, "step": 1111 }, { "epoch": 0.28868120456905505, "grad_norm": 0.6526694893836975, "learning_rate": 4.848082617757624e-06, "loss": 0.14128315448760986, "step": 1112 }, { "epoch": 0.28894080996884736, "grad_norm": 2.7712669372558594, "learning_rate": 4.847740716066804e-06, "loss": 0.16440758109092712, "step": 1113 }, { "epoch": 0.28920041536863966, "grad_norm": 2.515629529953003, "learning_rate": 4.847398443543459e-06, "loss": 0.14927372336387634, "step": 1114 }, { "epoch": 0.28946002076843197, "grad_norm": 2.306206226348877, "learning_rate": 4.847055800248092e-06, "loss": 0.18249112367630005, "step": 1115 }, { "epoch": 0.2897196261682243, "grad_norm": 1.8624622821807861, "learning_rate": 4.846712786241275e-06, "loss": 0.13866451382637024, "step": 1116 }, { "epoch": 0.28997923156801664, "grad_norm": 0.6877159476280212, "learning_rate": 4.8463694015836465e-06, "loss": 0.1654500663280487, "step": 1117 }, { "epoch": 0.29023883696780894, "grad_norm": 0.681938111782074, "learning_rate": 4.846025646335909e-06, "loss": 0.14666640758514404, "step": 1118 }, { "epoch": 0.29049844236760125, "grad_norm": 1.4356781244277954, "learning_rate": 4.845681520558829e-06, "loss": 0.12916234135627747, "step": 1119 }, { "epoch": 0.29075804776739356, "grad_norm": 2.2375526428222656, "learning_rate": 4.845337024313242e-06, "loss": 0.1484178900718689, "step": 1120 }, { "epoch": 0.29101765316718586, "grad_norm": 0.608214259147644, "learning_rate": 4.844992157660044e-06, "loss": 0.1354941427707672, "step": 1121 }, { "epoch": 0.29127725856697817, "grad_norm": 1.0339287519454956, "learning_rate": 4.844646920660203e-06, "loss": 0.17480728030204773, "step": 1122 }, { "epoch": 0.29153686396677053, "grad_norm": 1.5787616968154907, "learning_rate": 4.844301313374747e-06, "loss": 0.15053080022335052, "step": 1123 }, { "epoch": 0.29179646936656284, "grad_norm": 1.1544181108474731, "learning_rate": 4.843955335864772e-06, "loss": 0.14012984931468964, "step": 1124 }, { "epoch": 0.29205607476635514, "grad_norm": 1.2650578022003174, "learning_rate": 4.843608988191439e-06, "loss": 0.14860576391220093, "step": 1125 }, { "epoch": 0.29231568016614745, "grad_norm": 1.66677725315094, "learning_rate": 4.843262270415974e-06, "loss": 0.15686091780662537, "step": 1126 }, { "epoch": 0.29257528556593976, "grad_norm": 1.9697240591049194, "learning_rate": 4.842915182599669e-06, "loss": 0.13816668093204498, "step": 1127 }, { "epoch": 0.29283489096573206, "grad_norm": 1.414420247077942, "learning_rate": 4.842567724803881e-06, "loss": 0.13961219787597656, "step": 1128 }, { "epoch": 0.2930944963655244, "grad_norm": 2.6066818237304688, "learning_rate": 4.842219897090033e-06, "loss": 0.13493946194648743, "step": 1129 }, { "epoch": 0.29335410176531673, "grad_norm": 1.2377201318740845, "learning_rate": 4.8418716995196125e-06, "loss": 0.1698683500289917, "step": 1130 }, { "epoch": 0.29361370716510904, "grad_norm": 1.6862363815307617, "learning_rate": 4.841523132154173e-06, "loss": 0.16992349922657013, "step": 1131 }, { "epoch": 0.29387331256490135, "grad_norm": 1.5268243551254272, "learning_rate": 4.841174195055333e-06, "loss": 0.13560199737548828, "step": 1132 }, { "epoch": 0.29413291796469365, "grad_norm": 0.5755468010902405, "learning_rate": 4.8408248882847775e-06, "loss": 0.15657006204128265, "step": 1133 }, { "epoch": 0.29439252336448596, "grad_norm": 0.49368804693222046, "learning_rate": 4.8404752119042556e-06, "loss": 0.1397554874420166, "step": 1134 }, { "epoch": 0.2946521287642783, "grad_norm": 0.7641292810440063, "learning_rate": 4.8401251659755805e-06, "loss": 0.18761992454528809, "step": 1135 }, { "epoch": 0.2949117341640706, "grad_norm": 1.1172086000442505, "learning_rate": 4.839774750560635e-06, "loss": 0.18279120326042175, "step": 1136 }, { "epoch": 0.29517133956386293, "grad_norm": 0.6703447103500366, "learning_rate": 4.839423965721362e-06, "loss": 0.1478998064994812, "step": 1137 }, { "epoch": 0.29543094496365524, "grad_norm": 0.9582891464233398, "learning_rate": 4.839072811519773e-06, "loss": 0.17168888449668884, "step": 1138 }, { "epoch": 0.29569055036344755, "grad_norm": 0.45465439558029175, "learning_rate": 4.838721288017945e-06, "loss": 0.13392117619514465, "step": 1139 }, { "epoch": 0.29595015576323985, "grad_norm": 0.7257327437400818, "learning_rate": 4.838369395278019e-06, "loss": 0.1367538422346115, "step": 1140 }, { "epoch": 0.2962097611630322, "grad_norm": 1.1014635562896729, "learning_rate": 4.838017133362201e-06, "loss": 0.12088742852210999, "step": 1141 }, { "epoch": 0.2964693665628245, "grad_norm": 1.167844533920288, "learning_rate": 4.8376645023327625e-06, "loss": 0.13911499083042145, "step": 1142 }, { "epoch": 0.2967289719626168, "grad_norm": 0.6541082262992859, "learning_rate": 4.837311502252042e-06, "loss": 0.1353883296251297, "step": 1143 }, { "epoch": 0.29698857736240913, "grad_norm": 1.0339791774749756, "learning_rate": 4.836958133182441e-06, "loss": 0.19158756732940674, "step": 1144 }, { "epoch": 0.29724818276220144, "grad_norm": 1.5272971391677856, "learning_rate": 4.836604395186426e-06, "loss": 0.18170872330665588, "step": 1145 }, { "epoch": 0.29750778816199375, "grad_norm": 1.167637586593628, "learning_rate": 4.836250288326532e-06, "loss": 0.1467307209968567, "step": 1146 }, { "epoch": 0.2977673935617861, "grad_norm": 1.9985506534576416, "learning_rate": 4.835895812665356e-06, "loss": 0.2041265070438385, "step": 1147 }, { "epoch": 0.2980269989615784, "grad_norm": 0.7498098015785217, "learning_rate": 4.835540968265559e-06, "loss": 0.12732136249542236, "step": 1148 }, { "epoch": 0.2982866043613707, "grad_norm": 1.1050516366958618, "learning_rate": 4.8351857551898734e-06, "loss": 0.140090212225914, "step": 1149 }, { "epoch": 0.29854620976116303, "grad_norm": 1.045189619064331, "learning_rate": 4.83483017350109e-06, "loss": 0.1963970810174942, "step": 1150 }, { "epoch": 0.29880581516095533, "grad_norm": 1.5954177379608154, "learning_rate": 4.834474223262068e-06, "loss": 0.18319997191429138, "step": 1151 }, { "epoch": 0.29906542056074764, "grad_norm": 1.29942786693573, "learning_rate": 4.834117904535732e-06, "loss": 0.1932339370250702, "step": 1152 }, { "epoch": 0.29932502596054, "grad_norm": 2.911543846130371, "learning_rate": 4.83376121738507e-06, "loss": 0.14207351207733154, "step": 1153 }, { "epoch": 0.2995846313603323, "grad_norm": 2.4165523052215576, "learning_rate": 4.833404161873137e-06, "loss": 0.1358581781387329, "step": 1154 }, { "epoch": 0.2998442367601246, "grad_norm": 0.9820058345794678, "learning_rate": 4.8330467380630515e-06, "loss": 0.15116335451602936, "step": 1155 }, { "epoch": 0.3001038421599169, "grad_norm": 2.02492356300354, "learning_rate": 4.8326889460179984e-06, "loss": 0.16554245352745056, "step": 1156 }, { "epoch": 0.30036344755970923, "grad_norm": 3.2318198680877686, "learning_rate": 4.832330785801227e-06, "loss": 0.18369799852371216, "step": 1157 }, { "epoch": 0.30062305295950154, "grad_norm": 1.7569947242736816, "learning_rate": 4.8319722574760504e-06, "loss": 0.16918987035751343, "step": 1158 }, { "epoch": 0.3008826583592939, "grad_norm": 1.7809621095657349, "learning_rate": 4.8316133611058504e-06, "loss": 0.13557523488998413, "step": 1159 }, { "epoch": 0.3011422637590862, "grad_norm": 0.8833993673324585, "learning_rate": 4.831254096754072e-06, "loss": 0.1853547990322113, "step": 1160 }, { "epoch": 0.3014018691588785, "grad_norm": 2.476945638656616, "learning_rate": 4.8308944644842225e-06, "loss": 0.1700230836868286, "step": 1161 }, { "epoch": 0.3016614745586708, "grad_norm": 1.4871004819869995, "learning_rate": 4.8305344643598774e-06, "loss": 0.13859564065933228, "step": 1162 }, { "epoch": 0.3019210799584631, "grad_norm": 1.3441072702407837, "learning_rate": 4.830174096444678e-06, "loss": 0.16515028476715088, "step": 1163 }, { "epoch": 0.30218068535825543, "grad_norm": 1.1483056545257568, "learning_rate": 4.8298133608023275e-06, "loss": 0.18192288279533386, "step": 1164 }, { "epoch": 0.3024402907580478, "grad_norm": 0.49317237734794617, "learning_rate": 4.829452257496596e-06, "loss": 0.136500284075737, "step": 1165 }, { "epoch": 0.3026998961578401, "grad_norm": 1.0747767686843872, "learning_rate": 4.829090786591319e-06, "loss": 0.13874006271362305, "step": 1166 }, { "epoch": 0.3029595015576324, "grad_norm": 0.7988376617431641, "learning_rate": 4.828728948150395e-06, "loss": 0.1316343992948532, "step": 1167 }, { "epoch": 0.3032191069574247, "grad_norm": 1.7374851703643799, "learning_rate": 4.82836674223779e-06, "loss": 0.14658841490745544, "step": 1168 }, { "epoch": 0.303478712357217, "grad_norm": 0.4382416903972626, "learning_rate": 4.828004168917534e-06, "loss": 0.1346709430217743, "step": 1169 }, { "epoch": 0.3037383177570093, "grad_norm": 1.4761098623275757, "learning_rate": 4.8276412282537205e-06, "loss": 0.18326008319854736, "step": 1170 }, { "epoch": 0.3039979231568017, "grad_norm": 1.1887736320495605, "learning_rate": 4.8272779203105095e-06, "loss": 0.12479586899280548, "step": 1171 }, { "epoch": 0.304257528556594, "grad_norm": 1.5381728410720825, "learning_rate": 4.826914245152126e-06, "loss": 0.1350548267364502, "step": 1172 }, { "epoch": 0.3045171339563863, "grad_norm": 1.1818808317184448, "learning_rate": 4.826550202842858e-06, "loss": 0.1574394702911377, "step": 1173 }, { "epoch": 0.3047767393561786, "grad_norm": 0.9172492027282715, "learning_rate": 4.826185793447061e-06, "loss": 0.17559745907783508, "step": 1174 }, { "epoch": 0.3050363447559709, "grad_norm": 1.783239483833313, "learning_rate": 4.825821017029154e-06, "loss": 0.16222497820854187, "step": 1175 }, { "epoch": 0.3052959501557632, "grad_norm": 1.916969895362854, "learning_rate": 4.8254558736536216e-06, "loss": 0.14045464992523193, "step": 1176 }, { "epoch": 0.3055555555555556, "grad_norm": 1.9116647243499756, "learning_rate": 4.8250903633850125e-06, "loss": 0.14996354281902313, "step": 1177 }, { "epoch": 0.3058151609553479, "grad_norm": 0.6555312871932983, "learning_rate": 4.82472448628794e-06, "loss": 0.12686318159103394, "step": 1178 }, { "epoch": 0.3060747663551402, "grad_norm": 1.508439064025879, "learning_rate": 4.824358242427083e-06, "loss": 0.17495104670524597, "step": 1179 }, { "epoch": 0.3063343717549325, "grad_norm": 2.366133689880371, "learning_rate": 4.823991631867186e-06, "loss": 0.13123378157615662, "step": 1180 }, { "epoch": 0.3065939771547248, "grad_norm": 1.9167132377624512, "learning_rate": 4.8236246546730555e-06, "loss": 0.17553821206092834, "step": 1181 }, { "epoch": 0.3068535825545171, "grad_norm": 0.6351718306541443, "learning_rate": 4.823257310909567e-06, "loss": 0.14246457815170288, "step": 1182 }, { "epoch": 0.3071131879543095, "grad_norm": 0.956608772277832, "learning_rate": 4.822889600641657e-06, "loss": 0.13718447089195251, "step": 1183 }, { "epoch": 0.3073727933541018, "grad_norm": 2.6779892444610596, "learning_rate": 4.822521523934328e-06, "loss": 0.18371164798736572, "step": 1184 }, { "epoch": 0.3076323987538941, "grad_norm": 1.2442611455917358, "learning_rate": 4.822153080852648e-06, "loss": 0.12411551177501678, "step": 1185 }, { "epoch": 0.3078920041536864, "grad_norm": 1.3246128559112549, "learning_rate": 4.82178427146175e-06, "loss": 0.18207278847694397, "step": 1186 }, { "epoch": 0.3081516095534787, "grad_norm": 1.2028188705444336, "learning_rate": 4.82141509582683e-06, "loss": 0.13708288967609406, "step": 1187 }, { "epoch": 0.308411214953271, "grad_norm": 1.8726040124893188, "learning_rate": 4.821045554013151e-06, "loss": 0.16583731770515442, "step": 1188 }, { "epoch": 0.30867082035306337, "grad_norm": 1.2099145650863647, "learning_rate": 4.820675646086037e-06, "loss": 0.13515737652778625, "step": 1189 }, { "epoch": 0.3089304257528557, "grad_norm": 0.8098284602165222, "learning_rate": 4.8203053721108815e-06, "loss": 0.13618290424346924, "step": 1190 }, { "epoch": 0.309190031152648, "grad_norm": 1.264261245727539, "learning_rate": 4.819934732153141e-06, "loss": 0.1303674876689911, "step": 1191 }, { "epoch": 0.3094496365524403, "grad_norm": 0.9144351482391357, "learning_rate": 4.819563726278334e-06, "loss": 0.13070812821388245, "step": 1192 }, { "epoch": 0.3097092419522326, "grad_norm": 1.699995994567871, "learning_rate": 4.819192354552047e-06, "loss": 0.13507571816444397, "step": 1193 }, { "epoch": 0.3099688473520249, "grad_norm": 1.6931730508804321, "learning_rate": 4.81882061703993e-06, "loss": 0.1952228993177414, "step": 1194 }, { "epoch": 0.31022845275181726, "grad_norm": 1.4688433408737183, "learning_rate": 4.818448513807698e-06, "loss": 0.1822780817747116, "step": 1195 }, { "epoch": 0.31048805815160957, "grad_norm": 0.5491342544555664, "learning_rate": 4.818076044921129e-06, "loss": 0.12808357179164886, "step": 1196 }, { "epoch": 0.3107476635514019, "grad_norm": 1.144004464149475, "learning_rate": 4.817703210446069e-06, "loss": 0.16710540652275085, "step": 1197 }, { "epoch": 0.3110072689511942, "grad_norm": 0.49662134051322937, "learning_rate": 4.817330010448426e-06, "loss": 0.11646278947591782, "step": 1198 }, { "epoch": 0.3112668743509865, "grad_norm": 0.8552104830741882, "learning_rate": 4.816956444994172e-06, "loss": 0.1612759381532669, "step": 1199 }, { "epoch": 0.3115264797507788, "grad_norm": 1.1694799661636353, "learning_rate": 4.816582514149344e-06, "loss": 0.13421526551246643, "step": 1200 }, { "epoch": 0.31178608515057116, "grad_norm": 0.6900622844696045, "learning_rate": 4.816208217980049e-06, "loss": 0.13492585718631744, "step": 1201 }, { "epoch": 0.31204569055036346, "grad_norm": 1.0705440044403076, "learning_rate": 4.815833556552448e-06, "loss": 0.14058010280132294, "step": 1202 }, { "epoch": 0.31230529595015577, "grad_norm": 0.5777939558029175, "learning_rate": 4.815458529932778e-06, "loss": 0.12638230621814728, "step": 1203 }, { "epoch": 0.3125649013499481, "grad_norm": 1.9334615468978882, "learning_rate": 4.815083138187331e-06, "loss": 0.11735765635967255, "step": 1204 }, { "epoch": 0.3128245067497404, "grad_norm": 0.8715134263038635, "learning_rate": 4.8147073813824695e-06, "loss": 0.14200684428215027, "step": 1205 }, { "epoch": 0.3130841121495327, "grad_norm": 1.2596336603164673, "learning_rate": 4.814331259584619e-06, "loss": 0.14279673993587494, "step": 1206 }, { "epoch": 0.31334371754932505, "grad_norm": 3.047999143600464, "learning_rate": 4.8139547728602685e-06, "loss": 0.16781970858573914, "step": 1207 }, { "epoch": 0.31360332294911736, "grad_norm": 0.5910513401031494, "learning_rate": 4.8135779212759725e-06, "loss": 0.1514284610748291, "step": 1208 }, { "epoch": 0.31386292834890966, "grad_norm": 1.0030078887939453, "learning_rate": 4.81320070489835e-06, "loss": 0.12596392631530762, "step": 1209 }, { "epoch": 0.31412253374870197, "grad_norm": 1.9771549701690674, "learning_rate": 4.812823123794084e-06, "loss": 0.1591602861881256, "step": 1210 }, { "epoch": 0.3143821391484943, "grad_norm": 0.6536117792129517, "learning_rate": 4.8124451780299215e-06, "loss": 0.13097801804542542, "step": 1211 }, { "epoch": 0.3146417445482866, "grad_norm": 1.1841429471969604, "learning_rate": 4.812066867672676e-06, "loss": 0.16435563564300537, "step": 1212 }, { "epoch": 0.31490134994807895, "grad_norm": 0.8176998496055603, "learning_rate": 4.8116881927892225e-06, "loss": 0.1615799069404602, "step": 1213 }, { "epoch": 0.31516095534787125, "grad_norm": 2.897141456604004, "learning_rate": 4.811309153446503e-06, "loss": 0.1822672188282013, "step": 1214 }, { "epoch": 0.31542056074766356, "grad_norm": 0.7446861863136292, "learning_rate": 4.810929749711523e-06, "loss": 0.16736239194869995, "step": 1215 }, { "epoch": 0.31568016614745587, "grad_norm": 1.0260850191116333, "learning_rate": 4.810549981651353e-06, "loss": 0.1807125210762024, "step": 1216 }, { "epoch": 0.31593977154724817, "grad_norm": 2.12046217918396, "learning_rate": 4.810169849333126e-06, "loss": 0.15882466733455658, "step": 1217 }, { "epoch": 0.3161993769470405, "grad_norm": 1.2436250448226929, "learning_rate": 4.8097893528240415e-06, "loss": 0.1317289173603058, "step": 1218 }, { "epoch": 0.31645898234683284, "grad_norm": 1.3400832414627075, "learning_rate": 4.809408492191362e-06, "loss": 0.1849183291196823, "step": 1219 }, { "epoch": 0.31671858774662515, "grad_norm": 0.43000081181526184, "learning_rate": 4.809027267502415e-06, "loss": 0.13111230731010437, "step": 1220 }, { "epoch": 0.31697819314641745, "grad_norm": 0.6867330074310303, "learning_rate": 4.808645678824593e-06, "loss": 0.12172234058380127, "step": 1221 }, { "epoch": 0.31723779854620976, "grad_norm": 2.828310966491699, "learning_rate": 4.8082637262253515e-06, "loss": 0.20266568660736084, "step": 1222 }, { "epoch": 0.31749740394600207, "grad_norm": 0.46733975410461426, "learning_rate": 4.807881409772211e-06, "loss": 0.1300155371427536, "step": 1223 }, { "epoch": 0.3177570093457944, "grad_norm": 1.2946141958236694, "learning_rate": 4.807498729532758e-06, "loss": 0.17392103374004364, "step": 1224 }, { "epoch": 0.31801661474558673, "grad_norm": 1.5619624853134155, "learning_rate": 4.807115685574638e-06, "loss": 0.17220309376716614, "step": 1225 }, { "epoch": 0.31827622014537904, "grad_norm": 0.9387713074684143, "learning_rate": 4.806732277965569e-06, "loss": 0.13682153820991516, "step": 1226 }, { "epoch": 0.31853582554517135, "grad_norm": 1.8898032903671265, "learning_rate": 4.806348506773325e-06, "loss": 0.17863285541534424, "step": 1227 }, { "epoch": 0.31879543094496365, "grad_norm": 1.607904076576233, "learning_rate": 4.805964372065748e-06, "loss": 0.15924519300460815, "step": 1228 }, { "epoch": 0.31905503634475596, "grad_norm": 0.7446286082267761, "learning_rate": 4.805579873910746e-06, "loss": 0.14709565043449402, "step": 1229 }, { "epoch": 0.31931464174454827, "grad_norm": 2.5462536811828613, "learning_rate": 4.805195012376289e-06, "loss": 0.1747829020023346, "step": 1230 }, { "epoch": 0.31957424714434063, "grad_norm": 1.747588038444519, "learning_rate": 4.804809787530411e-06, "loss": 0.19075393676757812, "step": 1231 }, { "epoch": 0.31983385254413293, "grad_norm": 1.158786416053772, "learning_rate": 4.804424199441213e-06, "loss": 0.20255272090435028, "step": 1232 }, { "epoch": 0.32009345794392524, "grad_norm": 1.0258835554122925, "learning_rate": 4.804038248176855e-06, "loss": 0.12738820910453796, "step": 1233 }, { "epoch": 0.32035306334371755, "grad_norm": 1.7888171672821045, "learning_rate": 4.803651933805567e-06, "loss": 0.14000988006591797, "step": 1234 }, { "epoch": 0.32061266874350985, "grad_norm": 1.7757372856140137, "learning_rate": 4.803265256395639e-06, "loss": 0.1348382532596588, "step": 1235 }, { "epoch": 0.32087227414330216, "grad_norm": 0.7122347354888916, "learning_rate": 4.802878216015428e-06, "loss": 0.13561677932739258, "step": 1236 }, { "epoch": 0.3211318795430945, "grad_norm": 1.5217633247375488, "learning_rate": 4.802490812733352e-06, "loss": 0.15147443115711212, "step": 1237 }, { "epoch": 0.32139148494288683, "grad_norm": 1.4579167366027832, "learning_rate": 4.8021030466178965e-06, "loss": 0.16965307295322418, "step": 1238 }, { "epoch": 0.32165109034267914, "grad_norm": 0.6155604124069214, "learning_rate": 4.80171491773761e-06, "loss": 0.14354932308197021, "step": 1239 }, { "epoch": 0.32191069574247144, "grad_norm": 0.7690376043319702, "learning_rate": 4.801326426161103e-06, "loss": 0.16333439946174622, "step": 1240 }, { "epoch": 0.32217030114226375, "grad_norm": 0.8713114261627197, "learning_rate": 4.800937571957053e-06, "loss": 0.13441240787506104, "step": 1241 }, { "epoch": 0.32242990654205606, "grad_norm": 1.2667651176452637, "learning_rate": 4.800548355194202e-06, "loss": 0.1496712863445282, "step": 1242 }, { "epoch": 0.3226895119418484, "grad_norm": 0.6585379242897034, "learning_rate": 4.800158775941351e-06, "loss": 0.13134783506393433, "step": 1243 }, { "epoch": 0.3229491173416407, "grad_norm": 0.7937607765197754, "learning_rate": 4.799768834267372e-06, "loss": 0.14768442511558533, "step": 1244 }, { "epoch": 0.32320872274143303, "grad_norm": 0.6345475316047668, "learning_rate": 4.799378530241197e-06, "loss": 0.13604463636875153, "step": 1245 }, { "epoch": 0.32346832814122534, "grad_norm": 0.8327072262763977, "learning_rate": 4.7989878639318225e-06, "loss": 0.14770300686359406, "step": 1246 }, { "epoch": 0.32372793354101764, "grad_norm": 0.9170403480529785, "learning_rate": 4.798596835408309e-06, "loss": 0.1520552635192871, "step": 1247 }, { "epoch": 0.32398753894080995, "grad_norm": 1.1438686847686768, "learning_rate": 4.79820544473978e-06, "loss": 0.13627998530864716, "step": 1248 }, { "epoch": 0.3242471443406023, "grad_norm": 1.4620471000671387, "learning_rate": 4.797813691995428e-06, "loss": 0.1519857794046402, "step": 1249 }, { "epoch": 0.3245067497403946, "grad_norm": 1.9286487102508545, "learning_rate": 4.7974215772445035e-06, "loss": 0.18364140391349792, "step": 1250 }, { "epoch": 0.3247663551401869, "grad_norm": 1.4169448614120483, "learning_rate": 4.797029100556324e-06, "loss": 0.14569690823554993, "step": 1251 }, { "epoch": 0.32502596053997923, "grad_norm": 0.4861655533313751, "learning_rate": 4.796636262000269e-06, "loss": 0.13384592533111572, "step": 1252 }, { "epoch": 0.32528556593977154, "grad_norm": 1.6958822011947632, "learning_rate": 4.796243061645786e-06, "loss": 0.14436215162277222, "step": 1253 }, { "epoch": 0.32554517133956384, "grad_norm": 3.207772970199585, "learning_rate": 4.7958494995623824e-06, "loss": 0.18410998582839966, "step": 1254 }, { "epoch": 0.3258047767393562, "grad_norm": 0.566855251789093, "learning_rate": 4.79545557581963e-06, "loss": 0.1402546763420105, "step": 1255 }, { "epoch": 0.3260643821391485, "grad_norm": 1.7104243040084839, "learning_rate": 4.795061290487168e-06, "loss": 0.16035044193267822, "step": 1256 }, { "epoch": 0.3263239875389408, "grad_norm": 2.6279752254486084, "learning_rate": 4.794666643634695e-06, "loss": 0.17511925101280212, "step": 1257 }, { "epoch": 0.3265835929387331, "grad_norm": 1.443447232246399, "learning_rate": 4.794271635331975e-06, "loss": 0.13079464435577393, "step": 1258 }, { "epoch": 0.32684319833852543, "grad_norm": 0.8972057104110718, "learning_rate": 4.793876265648839e-06, "loss": 0.14229941368103027, "step": 1259 }, { "epoch": 0.32710280373831774, "grad_norm": 2.4215714931488037, "learning_rate": 4.793480534655177e-06, "loss": 0.16705676913261414, "step": 1260 }, { "epoch": 0.3273624091381101, "grad_norm": 0.5624738931655884, "learning_rate": 4.793084442420946e-06, "loss": 0.1277523636817932, "step": 1261 }, { "epoch": 0.3276220145379024, "grad_norm": 0.9869523644447327, "learning_rate": 4.792687989016166e-06, "loss": 0.16636154055595398, "step": 1262 }, { "epoch": 0.3278816199376947, "grad_norm": 2.1409943103790283, "learning_rate": 4.792291174510921e-06, "loss": 0.13802184164524078, "step": 1263 }, { "epoch": 0.328141225337487, "grad_norm": 0.4728768765926361, "learning_rate": 4.791893998975359e-06, "loss": 0.1294170618057251, "step": 1264 }, { "epoch": 0.3284008307372793, "grad_norm": 0.8682212233543396, "learning_rate": 4.791496462479691e-06, "loss": 0.15446072816848755, "step": 1265 }, { "epoch": 0.32866043613707163, "grad_norm": 2.6666009426116943, "learning_rate": 4.791098565094192e-06, "loss": 0.19114524126052856, "step": 1266 }, { "epoch": 0.328920041536864, "grad_norm": 1.0397381782531738, "learning_rate": 4.790700306889202e-06, "loss": 0.16894958913326263, "step": 1267 }, { "epoch": 0.3291796469366563, "grad_norm": 0.9763180017471313, "learning_rate": 4.790301687935124e-06, "loss": 0.1634388566017151, "step": 1268 }, { "epoch": 0.3294392523364486, "grad_norm": 0.7073900103569031, "learning_rate": 4.789902708302423e-06, "loss": 0.13535642623901367, "step": 1269 }, { "epoch": 0.3296988577362409, "grad_norm": 0.7855846285820007, "learning_rate": 4.789503368061632e-06, "loss": 0.17762985825538635, "step": 1270 }, { "epoch": 0.3299584631360332, "grad_norm": 1.659135341644287, "learning_rate": 4.789103667283344e-06, "loss": 0.1417992115020752, "step": 1271 }, { "epoch": 0.3302180685358255, "grad_norm": 1.0280297994613647, "learning_rate": 4.788703606038216e-06, "loss": 0.1664603352546692, "step": 1272 }, { "epoch": 0.3304776739356179, "grad_norm": 0.4429852366447449, "learning_rate": 4.788303184396971e-06, "loss": 0.1506299376487732, "step": 1273 }, { "epoch": 0.3307372793354102, "grad_norm": 0.5278615951538086, "learning_rate": 4.787902402430395e-06, "loss": 0.13979533314704895, "step": 1274 }, { "epoch": 0.3309968847352025, "grad_norm": 0.4978477656841278, "learning_rate": 4.787501260209334e-06, "loss": 0.136000394821167, "step": 1275 }, { "epoch": 0.3312564901349948, "grad_norm": 0.7485656142234802, "learning_rate": 4.7870997578047035e-06, "loss": 0.13883426785469055, "step": 1276 }, { "epoch": 0.3315160955347871, "grad_norm": 1.5863980054855347, "learning_rate": 4.7866978952874795e-06, "loss": 0.1733720898628235, "step": 1277 }, { "epoch": 0.3317757009345794, "grad_norm": 0.7459287047386169, "learning_rate": 4.786295672728702e-06, "loss": 0.14029262959957123, "step": 1278 }, { "epoch": 0.3320353063343718, "grad_norm": 0.9822279810905457, "learning_rate": 4.785893090199475e-06, "loss": 0.146453857421875, "step": 1279 }, { "epoch": 0.3322949117341641, "grad_norm": 0.9539140462875366, "learning_rate": 4.785490147770964e-06, "loss": 0.16188189387321472, "step": 1280 }, { "epoch": 0.3325545171339564, "grad_norm": 0.5853458642959595, "learning_rate": 4.785086845514403e-06, "loss": 0.16517534852027893, "step": 1281 }, { "epoch": 0.3328141225337487, "grad_norm": 1.425814151763916, "learning_rate": 4.784683183501084e-06, "loss": 0.18667982518672943, "step": 1282 }, { "epoch": 0.333073727933541, "grad_norm": 1.3825892210006714, "learning_rate": 4.7842791618023666e-06, "loss": 0.17594987154006958, "step": 1283 }, { "epoch": 0.3333333333333333, "grad_norm": 1.0627907514572144, "learning_rate": 4.783874780489673e-06, "loss": 0.13892340660095215, "step": 1284 }, { "epoch": 0.3335929387331256, "grad_norm": 0.863532543182373, "learning_rate": 4.783470039634486e-06, "loss": 0.15863776206970215, "step": 1285 }, { "epoch": 0.333852544132918, "grad_norm": 0.8490937948226929, "learning_rate": 4.783064939308357e-06, "loss": 0.1940252184867859, "step": 1286 }, { "epoch": 0.3341121495327103, "grad_norm": 0.40124669671058655, "learning_rate": 4.782659479582898e-06, "loss": 0.12826624512672424, "step": 1287 }, { "epoch": 0.3343717549325026, "grad_norm": 0.45653805136680603, "learning_rate": 4.782253660529785e-06, "loss": 0.12312915921211243, "step": 1288 }, { "epoch": 0.3346313603322949, "grad_norm": 2.0608441829681396, "learning_rate": 4.781847482220757e-06, "loss": 0.1859007328748703, "step": 1289 }, { "epoch": 0.3348909657320872, "grad_norm": 1.0064475536346436, "learning_rate": 4.781440944727617e-06, "loss": 0.16104656457901, "step": 1290 }, { "epoch": 0.3351505711318795, "grad_norm": 1.3740633726119995, "learning_rate": 4.781034048122232e-06, "loss": 0.1318732053041458, "step": 1291 }, { "epoch": 0.3354101765316719, "grad_norm": 2.049422025680542, "learning_rate": 4.780626792476532e-06, "loss": 0.19168996810913086, "step": 1292 }, { "epoch": 0.3356697819314642, "grad_norm": 2.9633989334106445, "learning_rate": 4.78021917786251e-06, "loss": 0.15936219692230225, "step": 1293 }, { "epoch": 0.3359293873312565, "grad_norm": 0.7102957963943481, "learning_rate": 4.779811204352223e-06, "loss": 0.14062711596488953, "step": 1294 }, { "epoch": 0.3361889927310488, "grad_norm": 0.6308387517929077, "learning_rate": 4.779402872017792e-06, "loss": 0.12829014658927917, "step": 1295 }, { "epoch": 0.3364485981308411, "grad_norm": 0.8654884696006775, "learning_rate": 4.778994180931401e-06, "loss": 0.14708483219146729, "step": 1296 }, { "epoch": 0.3367082035306334, "grad_norm": 0.5844686031341553, "learning_rate": 4.778585131165295e-06, "loss": 0.15988841652870178, "step": 1297 }, { "epoch": 0.33696780893042577, "grad_norm": 3.2721526622772217, "learning_rate": 4.778175722791788e-06, "loss": 0.1104959174990654, "step": 1298 }, { "epoch": 0.3372274143302181, "grad_norm": 1.2151135206222534, "learning_rate": 4.777765955883251e-06, "loss": 0.15533077716827393, "step": 1299 }, { "epoch": 0.3374870197300104, "grad_norm": 1.2589294910430908, "learning_rate": 4.777355830512122e-06, "loss": 0.20175188779830933, "step": 1300 }, { "epoch": 0.3377466251298027, "grad_norm": 2.95419979095459, "learning_rate": 4.776945346750905e-06, "loss": 0.1508932262659073, "step": 1301 }, { "epoch": 0.338006230529595, "grad_norm": 2.967531442642212, "learning_rate": 4.776534504672159e-06, "loss": 0.15352703630924225, "step": 1302 }, { "epoch": 0.3382658359293873, "grad_norm": 1.910301923751831, "learning_rate": 4.7761233043485165e-06, "loss": 0.1455647051334381, "step": 1303 }, { "epoch": 0.33852544132917967, "grad_norm": 1.0516918897628784, "learning_rate": 4.7757117458526645e-06, "loss": 0.16224738955497742, "step": 1304 }, { "epoch": 0.338785046728972, "grad_norm": 1.448189377784729, "learning_rate": 4.775299829257358e-06, "loss": 0.18459929525852203, "step": 1305 }, { "epoch": 0.3390446521287643, "grad_norm": 2.8877477645874023, "learning_rate": 4.7748875546354155e-06, "loss": 0.19607552886009216, "step": 1306 }, { "epoch": 0.3393042575285566, "grad_norm": 0.714919924736023, "learning_rate": 4.774474922059718e-06, "loss": 0.13741549849510193, "step": 1307 }, { "epoch": 0.3395638629283489, "grad_norm": 0.6675977110862732, "learning_rate": 4.774061931603207e-06, "loss": 0.12792521715164185, "step": 1308 }, { "epoch": 0.3398234683281412, "grad_norm": 0.7172074317932129, "learning_rate": 4.773648583338892e-06, "loss": 0.1377151906490326, "step": 1309 }, { "epoch": 0.34008307372793356, "grad_norm": 0.7614622116088867, "learning_rate": 4.773234877339843e-06, "loss": 0.13814181089401245, "step": 1310 }, { "epoch": 0.34034267912772587, "grad_norm": 1.085028052330017, "learning_rate": 4.772820813679194e-06, "loss": 0.14606651663780212, "step": 1311 }, { "epoch": 0.3406022845275182, "grad_norm": 0.5786448121070862, "learning_rate": 4.772406392430141e-06, "loss": 0.1745041161775589, "step": 1312 }, { "epoch": 0.3408618899273105, "grad_norm": 0.8311803340911865, "learning_rate": 4.771991613665945e-06, "loss": 0.1721210777759552, "step": 1313 }, { "epoch": 0.3411214953271028, "grad_norm": 0.8934260606765747, "learning_rate": 4.771576477459929e-06, "loss": 0.14493323862552643, "step": 1314 }, { "epoch": 0.3413811007268951, "grad_norm": 1.4259806871414185, "learning_rate": 4.77116098388548e-06, "loss": 0.15307831764221191, "step": 1315 }, { "epoch": 0.34164070612668745, "grad_norm": 1.4968363046646118, "learning_rate": 4.770745133016048e-06, "loss": 0.17768289148807526, "step": 1316 }, { "epoch": 0.34190031152647976, "grad_norm": 0.6076454520225525, "learning_rate": 4.770328924925145e-06, "loss": 0.12032482028007507, "step": 1317 }, { "epoch": 0.34215991692627207, "grad_norm": 0.5408424735069275, "learning_rate": 4.769912359686346e-06, "loss": 0.16125792264938354, "step": 1318 }, { "epoch": 0.3424195223260644, "grad_norm": 1.55613112449646, "learning_rate": 4.7694954373732935e-06, "loss": 0.13113349676132202, "step": 1319 }, { "epoch": 0.3426791277258567, "grad_norm": 0.4462912082672119, "learning_rate": 4.769078158059687e-06, "loss": 0.14352986216545105, "step": 1320 }, { "epoch": 0.342938733125649, "grad_norm": 0.7809919118881226, "learning_rate": 4.7686605218192935e-06, "loss": 0.17777389287948608, "step": 1321 }, { "epoch": 0.34319833852544135, "grad_norm": 1.0490319728851318, "learning_rate": 4.768242528725941e-06, "loss": 0.17890939116477966, "step": 1322 }, { "epoch": 0.34345794392523366, "grad_norm": 0.9000142216682434, "learning_rate": 4.767824178853521e-06, "loss": 0.14205223321914673, "step": 1323 }, { "epoch": 0.34371754932502596, "grad_norm": 1.913751482963562, "learning_rate": 4.767405472275988e-06, "loss": 0.16148176789283752, "step": 1324 }, { "epoch": 0.34397715472481827, "grad_norm": 0.8513673543930054, "learning_rate": 4.76698640906736e-06, "loss": 0.13412144780158997, "step": 1325 }, { "epoch": 0.3442367601246106, "grad_norm": 0.8516746163368225, "learning_rate": 4.766566989301718e-06, "loss": 0.1750883162021637, "step": 1326 }, { "epoch": 0.3444963655244029, "grad_norm": 1.0712980031967163, "learning_rate": 4.7661472130532056e-06, "loss": 0.15639731287956238, "step": 1327 }, { "epoch": 0.34475597092419524, "grad_norm": 1.0911283493041992, "learning_rate": 4.7657270803960285e-06, "loss": 0.14698168635368347, "step": 1328 }, { "epoch": 0.34501557632398755, "grad_norm": 0.7639453411102295, "learning_rate": 4.765306591404459e-06, "loss": 0.17833887040615082, "step": 1329 }, { "epoch": 0.34527518172377986, "grad_norm": 1.4660342931747437, "learning_rate": 4.764885746152828e-06, "loss": 0.1280405968427658, "step": 1330 }, { "epoch": 0.34553478712357216, "grad_norm": 1.0507904291152954, "learning_rate": 4.764464544715532e-06, "loss": 0.14281363785266876, "step": 1331 }, { "epoch": 0.34579439252336447, "grad_norm": 0.9103289842605591, "learning_rate": 4.764042987167029e-06, "loss": 0.15439197421073914, "step": 1332 }, { "epoch": 0.3460539979231568, "grad_norm": 1.1471476554870605, "learning_rate": 4.763621073581842e-06, "loss": 0.14923009276390076, "step": 1333 }, { "epoch": 0.34631360332294914, "grad_norm": 0.680395245552063, "learning_rate": 4.7631988040345554e-06, "loss": 0.13847826421260834, "step": 1334 }, { "epoch": 0.34657320872274144, "grad_norm": 0.5429221987724304, "learning_rate": 4.762776178599815e-06, "loss": 0.1355435699224472, "step": 1335 }, { "epoch": 0.34683281412253375, "grad_norm": 1.1850481033325195, "learning_rate": 4.762353197352333e-06, "loss": 0.16767184436321259, "step": 1336 }, { "epoch": 0.34709241952232606, "grad_norm": 0.5057375431060791, "learning_rate": 4.761929860366883e-06, "loss": 0.1401139497756958, "step": 1337 }, { "epoch": 0.34735202492211836, "grad_norm": 0.6782482266426086, "learning_rate": 4.761506167718302e-06, "loss": 0.12489967793226242, "step": 1338 }, { "epoch": 0.34761163032191067, "grad_norm": 0.9127222299575806, "learning_rate": 4.761082119481486e-06, "loss": 0.13682124018669128, "step": 1339 }, { "epoch": 0.34787123572170303, "grad_norm": 1.3756327629089355, "learning_rate": 4.7606577157314e-06, "loss": 0.15532994270324707, "step": 1340 }, { "epoch": 0.34813084112149534, "grad_norm": 0.6735143661499023, "learning_rate": 4.760232956543067e-06, "loss": 0.1375035047531128, "step": 1341 }, { "epoch": 0.34839044652128764, "grad_norm": 1.3001412153244019, "learning_rate": 4.759807841991576e-06, "loss": 0.1486371010541916, "step": 1342 }, { "epoch": 0.34865005192107995, "grad_norm": 1.908765196800232, "learning_rate": 4.759382372152077e-06, "loss": 0.1739533394575119, "step": 1343 }, { "epoch": 0.34890965732087226, "grad_norm": 0.6199779510498047, "learning_rate": 4.7589565470997836e-06, "loss": 0.13811033964157104, "step": 1344 }, { "epoch": 0.34916926272066456, "grad_norm": 1.8893630504608154, "learning_rate": 4.758530366909971e-06, "loss": 0.14388909935951233, "step": 1345 }, { "epoch": 0.3494288681204569, "grad_norm": 1.1992987394332886, "learning_rate": 4.758103831657979e-06, "loss": 0.15304023027420044, "step": 1346 }, { "epoch": 0.34968847352024923, "grad_norm": 1.1625206470489502, "learning_rate": 4.75767694141921e-06, "loss": 0.1804961860179901, "step": 1347 }, { "epoch": 0.34994807892004154, "grad_norm": 1.5193520784378052, "learning_rate": 4.757249696269126e-06, "loss": 0.17206597328186035, "step": 1348 }, { "epoch": 0.35020768431983385, "grad_norm": 1.7872798442840576, "learning_rate": 4.756822096283257e-06, "loss": 0.1881038248538971, "step": 1349 }, { "epoch": 0.35046728971962615, "grad_norm": 1.6727529764175415, "learning_rate": 4.7563941415371905e-06, "loss": 0.1500054895877838, "step": 1350 }, { "epoch": 0.35072689511941846, "grad_norm": 1.9057284593582153, "learning_rate": 4.755965832106581e-06, "loss": 0.18854732811450958, "step": 1351 }, { "epoch": 0.3509865005192108, "grad_norm": 0.6123793721199036, "learning_rate": 4.755537168067142e-06, "loss": 0.17447441816329956, "step": 1352 }, { "epoch": 0.3512461059190031, "grad_norm": 1.1371536254882812, "learning_rate": 4.755108149494653e-06, "loss": 0.1403551697731018, "step": 1353 }, { "epoch": 0.35150571131879543, "grad_norm": 1.7405625581741333, "learning_rate": 4.754678776464954e-06, "loss": 0.13971015810966492, "step": 1354 }, { "epoch": 0.35176531671858774, "grad_norm": 0.6587004065513611, "learning_rate": 4.754249049053948e-06, "loss": 0.17371171712875366, "step": 1355 }, { "epoch": 0.35202492211838005, "grad_norm": 1.0755815505981445, "learning_rate": 4.753818967337602e-06, "loss": 0.17246368527412415, "step": 1356 }, { "epoch": 0.35228452751817235, "grad_norm": 0.9343434572219849, "learning_rate": 4.753388531391943e-06, "loss": 0.1927688717842102, "step": 1357 }, { "epoch": 0.3525441329179647, "grad_norm": 1.5483529567718506, "learning_rate": 4.752957741293064e-06, "loss": 0.16005036234855652, "step": 1358 }, { "epoch": 0.352803738317757, "grad_norm": 0.8788819313049316, "learning_rate": 4.752526597117117e-06, "loss": 0.15012013912200928, "step": 1359 }, { "epoch": 0.3530633437175493, "grad_norm": 1.7149635553359985, "learning_rate": 4.7520950989403205e-06, "loss": 0.1588546633720398, "step": 1360 }, { "epoch": 0.35332294911734163, "grad_norm": 2.104280948638916, "learning_rate": 4.751663246838953e-06, "loss": 0.15096509456634521, "step": 1361 }, { "epoch": 0.35358255451713394, "grad_norm": 1.1374595165252686, "learning_rate": 4.751231040889354e-06, "loss": 0.1330614686012268, "step": 1362 }, { "epoch": 0.35384215991692625, "grad_norm": 2.9140162467956543, "learning_rate": 4.7507984811679296e-06, "loss": 0.1783280074596405, "step": 1363 }, { "epoch": 0.3541017653167186, "grad_norm": 3.2862768173217773, "learning_rate": 4.750365567751146e-06, "loss": 0.18591755628585815, "step": 1364 }, { "epoch": 0.3543613707165109, "grad_norm": 1.1923397779464722, "learning_rate": 4.749932300715532e-06, "loss": 0.12650083005428314, "step": 1365 }, { "epoch": 0.3546209761163032, "grad_norm": 0.6687277555465698, "learning_rate": 4.74949868013768e-06, "loss": 0.14000046253204346, "step": 1366 }, { "epoch": 0.35488058151609553, "grad_norm": 1.7264763116836548, "learning_rate": 4.749064706094244e-06, "loss": 0.1491914838552475, "step": 1367 }, { "epoch": 0.35514018691588783, "grad_norm": 1.9226351976394653, "learning_rate": 4.74863037866194e-06, "loss": 0.13464730978012085, "step": 1368 }, { "epoch": 0.35539979231568014, "grad_norm": 3.4508421421051025, "learning_rate": 4.748195697917547e-06, "loss": 0.17989182472229004, "step": 1369 }, { "epoch": 0.3556593977154725, "grad_norm": 2.853571653366089, "learning_rate": 4.747760663937908e-06, "loss": 0.17103983461856842, "step": 1370 }, { "epoch": 0.3559190031152648, "grad_norm": 1.6659013032913208, "learning_rate": 4.747325276799927e-06, "loss": 0.13344745337963104, "step": 1371 }, { "epoch": 0.3561786085150571, "grad_norm": 3.2854063510894775, "learning_rate": 4.746889536580569e-06, "loss": 0.18229719996452332, "step": 1372 }, { "epoch": 0.3564382139148494, "grad_norm": 3.566643238067627, "learning_rate": 4.7464534433568634e-06, "loss": 0.15610219538211823, "step": 1373 }, { "epoch": 0.35669781931464173, "grad_norm": 1.6007627248764038, "learning_rate": 4.7460169972059015e-06, "loss": 0.13282883167266846, "step": 1374 }, { "epoch": 0.35695742471443404, "grad_norm": 0.6078305840492249, "learning_rate": 4.7455801982048374e-06, "loss": 0.14909088611602783, "step": 1375 }, { "epoch": 0.3572170301142264, "grad_norm": 0.6727648377418518, "learning_rate": 4.7451430464308875e-06, "loss": 0.12780094146728516, "step": 1376 }, { "epoch": 0.3574766355140187, "grad_norm": 2.0464909076690674, "learning_rate": 4.74470554196133e-06, "loss": 0.17224770784378052, "step": 1377 }, { "epoch": 0.357736240913811, "grad_norm": 2.4937453269958496, "learning_rate": 4.744267684873505e-06, "loss": 0.15296369791030884, "step": 1378 }, { "epoch": 0.3579958463136033, "grad_norm": 1.5340887308120728, "learning_rate": 4.743829475244816e-06, "loss": 0.1341835856437683, "step": 1379 }, { "epoch": 0.3582554517133956, "grad_norm": 0.5707199573516846, "learning_rate": 4.743390913152728e-06, "loss": 0.13142593204975128, "step": 1380 }, { "epoch": 0.35851505711318793, "grad_norm": 0.751836895942688, "learning_rate": 4.74295199867477e-06, "loss": 0.15158843994140625, "step": 1381 }, { "epoch": 0.3587746625129803, "grad_norm": 2.648228168487549, "learning_rate": 4.74251273188853e-06, "loss": 0.17131248116493225, "step": 1382 }, { "epoch": 0.3590342679127726, "grad_norm": 1.9406723976135254, "learning_rate": 4.742073112871663e-06, "loss": 0.1394500881433487, "step": 1383 }, { "epoch": 0.3592938733125649, "grad_norm": 1.4884428977966309, "learning_rate": 4.741633141701881e-06, "loss": 0.16417816281318665, "step": 1384 }, { "epoch": 0.3595534787123572, "grad_norm": 1.146889567375183, "learning_rate": 4.741192818456961e-06, "loss": 0.13254579901695251, "step": 1385 }, { "epoch": 0.3598130841121495, "grad_norm": 1.0445529222488403, "learning_rate": 4.740752143214744e-06, "loss": 0.15464389324188232, "step": 1386 }, { "epoch": 0.3600726895119418, "grad_norm": 0.9591387510299683, "learning_rate": 4.74031111605313e-06, "loss": 0.16428007185459137, "step": 1387 }, { "epoch": 0.3603322949117342, "grad_norm": 1.519633412361145, "learning_rate": 4.739869737050084e-06, "loss": 0.1377807855606079, "step": 1388 }, { "epoch": 0.3605919003115265, "grad_norm": 1.9757423400878906, "learning_rate": 4.73942800628363e-06, "loss": 0.15462222695350647, "step": 1389 }, { "epoch": 0.3608515057113188, "grad_norm": 1.9731391668319702, "learning_rate": 4.7389859238318555e-06, "loss": 0.15206831693649292, "step": 1390 }, { "epoch": 0.3611111111111111, "grad_norm": 1.1847347021102905, "learning_rate": 4.738543489772912e-06, "loss": 0.19017250835895538, "step": 1391 }, { "epoch": 0.3613707165109034, "grad_norm": 2.161623001098633, "learning_rate": 4.738100704185011e-06, "loss": 0.1624441295862198, "step": 1392 }, { "epoch": 0.3616303219106957, "grad_norm": 0.7427332401275635, "learning_rate": 4.737657567146427e-06, "loss": 0.143422469496727, "step": 1393 }, { "epoch": 0.3618899273104881, "grad_norm": 0.9858644604682922, "learning_rate": 4.737214078735496e-06, "loss": 0.15638718008995056, "step": 1394 }, { "epoch": 0.3621495327102804, "grad_norm": 0.916603684425354, "learning_rate": 4.736770239030618e-06, "loss": 0.15419936180114746, "step": 1395 }, { "epoch": 0.3624091381100727, "grad_norm": 0.8351823687553406, "learning_rate": 4.736326048110252e-06, "loss": 0.16394414007663727, "step": 1396 }, { "epoch": 0.362668743509865, "grad_norm": 0.8772950172424316, "learning_rate": 4.735881506052922e-06, "loss": 0.18078210949897766, "step": 1397 }, { "epoch": 0.3629283489096573, "grad_norm": 0.6445213556289673, "learning_rate": 4.735436612937211e-06, "loss": 0.12563708424568176, "step": 1398 }, { "epoch": 0.3631879543094496, "grad_norm": 1.3797622919082642, "learning_rate": 4.734991368841769e-06, "loss": 0.1352149397134781, "step": 1399 }, { "epoch": 0.363447559709242, "grad_norm": 1.0146727561950684, "learning_rate": 4.7345457738453025e-06, "loss": 0.1526431143283844, "step": 1400 }, { "epoch": 0.3637071651090343, "grad_norm": 1.39897882938385, "learning_rate": 4.734099828026584e-06, "loss": 0.15208099782466888, "step": 1401 }, { "epoch": 0.3639667705088266, "grad_norm": 0.9093642830848694, "learning_rate": 4.733653531464445e-06, "loss": 0.1433805227279663, "step": 1402 }, { "epoch": 0.3642263759086189, "grad_norm": 1.384414792060852, "learning_rate": 4.7332068842377815e-06, "loss": 0.13865086436271667, "step": 1403 }, { "epoch": 0.3644859813084112, "grad_norm": 1.0792979001998901, "learning_rate": 4.73275988642555e-06, "loss": 0.16430261731147766, "step": 1404 }, { "epoch": 0.3647455867082035, "grad_norm": 0.7917479276657104, "learning_rate": 4.73231253810677e-06, "loss": 0.12924744188785553, "step": 1405 }, { "epoch": 0.36500519210799587, "grad_norm": 1.6499814987182617, "learning_rate": 4.731864839360523e-06, "loss": 0.15581631660461426, "step": 1406 }, { "epoch": 0.3652647975077882, "grad_norm": 0.8145169019699097, "learning_rate": 4.73141679026595e-06, "loss": 0.14733800292015076, "step": 1407 }, { "epoch": 0.3655244029075805, "grad_norm": 1.1472148895263672, "learning_rate": 4.730968390902257e-06, "loss": 0.17571969330310822, "step": 1408 }, { "epoch": 0.3657840083073728, "grad_norm": 1.597164273262024, "learning_rate": 4.730519641348711e-06, "loss": 0.1902889609336853, "step": 1409 }, { "epoch": 0.3660436137071651, "grad_norm": 1.5235178470611572, "learning_rate": 4.73007054168464e-06, "loss": 0.14786924421787262, "step": 1410 }, { "epoch": 0.3663032191069574, "grad_norm": 0.6915741562843323, "learning_rate": 4.729621091989434e-06, "loss": 0.12788957357406616, "step": 1411 }, { "epoch": 0.36656282450674976, "grad_norm": 1.4754328727722168, "learning_rate": 4.729171292342549e-06, "loss": 0.14334893226623535, "step": 1412 }, { "epoch": 0.36682242990654207, "grad_norm": 1.0146420001983643, "learning_rate": 4.728721142823495e-06, "loss": 0.1331711709499359, "step": 1413 }, { "epoch": 0.3670820353063344, "grad_norm": 1.415614128112793, "learning_rate": 4.728270643511849e-06, "loss": 0.16686339676380157, "step": 1414 }, { "epoch": 0.3673416407061267, "grad_norm": 0.4931534230709076, "learning_rate": 4.72781979448725e-06, "loss": 0.12941549718379974, "step": 1415 }, { "epoch": 0.367601246105919, "grad_norm": 1.072163462638855, "learning_rate": 4.727368595829397e-06, "loss": 0.15018987655639648, "step": 1416 }, { "epoch": 0.3678608515057113, "grad_norm": 0.6225241422653198, "learning_rate": 4.726917047618053e-06, "loss": 0.15408512949943542, "step": 1417 }, { "epoch": 0.36812045690550366, "grad_norm": 1.1997487545013428, "learning_rate": 4.72646514993304e-06, "loss": 0.14936485886573792, "step": 1418 }, { "epoch": 0.36838006230529596, "grad_norm": 1.3411604166030884, "learning_rate": 4.726012902854243e-06, "loss": 0.1580609679222107, "step": 1419 }, { "epoch": 0.36863966770508827, "grad_norm": 2.6280791759490967, "learning_rate": 4.72556030646161e-06, "loss": 0.17608773708343506, "step": 1420 }, { "epoch": 0.3688992731048806, "grad_norm": 0.9696816802024841, "learning_rate": 4.7251073608351485e-06, "loss": 0.13819792866706848, "step": 1421 }, { "epoch": 0.3691588785046729, "grad_norm": 1.347024917602539, "learning_rate": 4.724654066054931e-06, "loss": 0.12801039218902588, "step": 1422 }, { "epoch": 0.3694184839044652, "grad_norm": 2.0716607570648193, "learning_rate": 4.724200422201087e-06, "loss": 0.17658330500125885, "step": 1423 }, { "epoch": 0.36967808930425755, "grad_norm": 2.2434465885162354, "learning_rate": 4.7237464293538115e-06, "loss": 0.13697132468223572, "step": 1424 }, { "epoch": 0.36993769470404986, "grad_norm": 0.6106371283531189, "learning_rate": 4.72329208759336e-06, "loss": 0.16869306564331055, "step": 1425 }, { "epoch": 0.37019730010384216, "grad_norm": 0.5471572875976562, "learning_rate": 4.722837397000052e-06, "loss": 0.14065170288085938, "step": 1426 }, { "epoch": 0.37045690550363447, "grad_norm": 0.6492620706558228, "learning_rate": 4.722382357654262e-06, "loss": 0.13363829255104065, "step": 1427 }, { "epoch": 0.3707165109034268, "grad_norm": 1.4539406299591064, "learning_rate": 4.721926969636433e-06, "loss": 0.17856809496879578, "step": 1428 }, { "epoch": 0.3709761163032191, "grad_norm": 1.5595436096191406, "learning_rate": 4.721471233027067e-06, "loss": 0.16323186457157135, "step": 1429 }, { "epoch": 0.37123572170301145, "grad_norm": 1.6123615503311157, "learning_rate": 4.721015147906727e-06, "loss": 0.1372678130865097, "step": 1430 }, { "epoch": 0.37149532710280375, "grad_norm": 0.9590815305709839, "learning_rate": 4.720558714356041e-06, "loss": 0.1556023508310318, "step": 1431 }, { "epoch": 0.37175493250259606, "grad_norm": 1.4893758296966553, "learning_rate": 4.720101932455693e-06, "loss": 0.1395808756351471, "step": 1432 }, { "epoch": 0.37201453790238836, "grad_norm": 1.4239075183868408, "learning_rate": 4.719644802286433e-06, "loss": 0.18386611342430115, "step": 1433 }, { "epoch": 0.37227414330218067, "grad_norm": 1.4775761365890503, "learning_rate": 4.71918732392907e-06, "loss": 0.15051518380641937, "step": 1434 }, { "epoch": 0.372533748701973, "grad_norm": 0.43242916464805603, "learning_rate": 4.718729497464477e-06, "loss": 0.13167965412139893, "step": 1435 }, { "epoch": 0.37279335410176534, "grad_norm": 0.8950396776199341, "learning_rate": 4.7182713229735874e-06, "loss": 0.1347874104976654, "step": 1436 }, { "epoch": 0.37305295950155765, "grad_norm": 1.0013177394866943, "learning_rate": 4.717812800537394e-06, "loss": 0.18236202001571655, "step": 1437 }, { "epoch": 0.37331256490134995, "grad_norm": 1.0308747291564941, "learning_rate": 4.717353930236956e-06, "loss": 0.1371552050113678, "step": 1438 }, { "epoch": 0.37357217030114226, "grad_norm": 2.382816791534424, "learning_rate": 4.716894712153388e-06, "loss": 0.20247149467468262, "step": 1439 }, { "epoch": 0.37383177570093457, "grad_norm": 0.8919193744659424, "learning_rate": 4.716435146367872e-06, "loss": 0.15714147686958313, "step": 1440 }, { "epoch": 0.37409138110072687, "grad_norm": 2.792567014694214, "learning_rate": 4.715975232961647e-06, "loss": 0.16721418499946594, "step": 1441 }, { "epoch": 0.37435098650051923, "grad_norm": 0.9804661273956299, "learning_rate": 4.715514972016016e-06, "loss": 0.1415787637233734, "step": 1442 }, { "epoch": 0.37461059190031154, "grad_norm": 1.3679276704788208, "learning_rate": 4.715054363612343e-06, "loss": 0.15700221061706543, "step": 1443 }, { "epoch": 0.37487019730010385, "grad_norm": 1.112379550933838, "learning_rate": 4.714593407832051e-06, "loss": 0.1468597650527954, "step": 1444 }, { "epoch": 0.37512980269989615, "grad_norm": 1.68025541305542, "learning_rate": 4.714132104756628e-06, "loss": 0.15552449226379395, "step": 1445 }, { "epoch": 0.37538940809968846, "grad_norm": 1.2244476079940796, "learning_rate": 4.7136704544676215e-06, "loss": 0.16698387265205383, "step": 1446 }, { "epoch": 0.37564901349948077, "grad_norm": 0.7769119143486023, "learning_rate": 4.713208457046641e-06, "loss": 0.12628039717674255, "step": 1447 }, { "epoch": 0.37590861889927313, "grad_norm": 1.51413893699646, "learning_rate": 4.7127461125753575e-06, "loss": 0.19139879941940308, "step": 1448 }, { "epoch": 0.37616822429906543, "grad_norm": 1.328454613685608, "learning_rate": 4.7122834211355e-06, "loss": 0.12344683706760406, "step": 1449 }, { "epoch": 0.37642782969885774, "grad_norm": 1.4142968654632568, "learning_rate": 4.711820382808866e-06, "loss": 0.17816977202892303, "step": 1450 }, { "epoch": 0.37668743509865005, "grad_norm": 2.178015947341919, "learning_rate": 4.711356997677308e-06, "loss": 0.1614900678396225, "step": 1451 }, { "epoch": 0.37694704049844235, "grad_norm": 0.5517861247062683, "learning_rate": 4.710893265822741e-06, "loss": 0.10960031300783157, "step": 1452 }, { "epoch": 0.37720664589823466, "grad_norm": 0.7914402484893799, "learning_rate": 4.710429187327144e-06, "loss": 0.1494309902191162, "step": 1453 }, { "epoch": 0.377466251298027, "grad_norm": 1.483963966369629, "learning_rate": 4.709964762272554e-06, "loss": 0.145544171333313, "step": 1454 }, { "epoch": 0.37772585669781933, "grad_norm": 1.3192347288131714, "learning_rate": 4.709499990741071e-06, "loss": 0.136140376329422, "step": 1455 }, { "epoch": 0.37798546209761164, "grad_norm": 1.0180808305740356, "learning_rate": 4.709034872814856e-06, "loss": 0.12861742079257965, "step": 1456 }, { "epoch": 0.37824506749740394, "grad_norm": 1.2136595249176025, "learning_rate": 4.708569408576133e-06, "loss": 0.1815188229084015, "step": 1457 }, { "epoch": 0.37850467289719625, "grad_norm": 1.4457619190216064, "learning_rate": 4.708103598107182e-06, "loss": 0.13035912811756134, "step": 1458 }, { "epoch": 0.37876427829698855, "grad_norm": 1.3193939924240112, "learning_rate": 4.707637441490351e-06, "loss": 0.12925317883491516, "step": 1459 }, { "epoch": 0.3790238836967809, "grad_norm": 1.9721046686172485, "learning_rate": 4.707170938808044e-06, "loss": 0.177663654088974, "step": 1460 }, { "epoch": 0.3792834890965732, "grad_norm": 0.6200783848762512, "learning_rate": 4.706704090142728e-06, "loss": 0.11877068132162094, "step": 1461 }, { "epoch": 0.37954309449636553, "grad_norm": 1.2406692504882812, "learning_rate": 4.706236895576931e-06, "loss": 0.13363923132419586, "step": 1462 }, { "epoch": 0.37980269989615784, "grad_norm": 1.22507905960083, "learning_rate": 4.705769355193244e-06, "loss": 0.13449376821517944, "step": 1463 }, { "epoch": 0.38006230529595014, "grad_norm": 0.7286678552627563, "learning_rate": 4.705301469074314e-06, "loss": 0.15503916144371033, "step": 1464 }, { "epoch": 0.38032191069574245, "grad_norm": 1.2342314720153809, "learning_rate": 4.704833237302857e-06, "loss": 0.1303499937057495, "step": 1465 }, { "epoch": 0.3805815160955348, "grad_norm": 1.403685450553894, "learning_rate": 4.7043646599616425e-06, "loss": 0.151693657040596, "step": 1466 }, { "epoch": 0.3808411214953271, "grad_norm": 0.7136582136154175, "learning_rate": 4.703895737133505e-06, "loss": 0.1766369640827179, "step": 1467 }, { "epoch": 0.3811007268951194, "grad_norm": 1.588955283164978, "learning_rate": 4.703426468901339e-06, "loss": 0.18643967807292938, "step": 1468 }, { "epoch": 0.38136033229491173, "grad_norm": 1.7456690073013306, "learning_rate": 4.702956855348102e-06, "loss": 0.15909704566001892, "step": 1469 }, { "epoch": 0.38161993769470404, "grad_norm": 1.5388065576553345, "learning_rate": 4.702486896556809e-06, "loss": 0.1503351926803589, "step": 1470 }, { "epoch": 0.38187954309449634, "grad_norm": 0.829245388507843, "learning_rate": 4.702016592610539e-06, "loss": 0.15604425966739655, "step": 1471 }, { "epoch": 0.3821391484942887, "grad_norm": 1.2163132429122925, "learning_rate": 4.70154594359243e-06, "loss": 0.17706410586833954, "step": 1472 }, { "epoch": 0.382398753894081, "grad_norm": 1.2458575963974, "learning_rate": 4.701074949585684e-06, "loss": 0.1578412652015686, "step": 1473 }, { "epoch": 0.3826583592938733, "grad_norm": 1.2809102535247803, "learning_rate": 4.7006036106735594e-06, "loss": 0.13331767916679382, "step": 1474 }, { "epoch": 0.3829179646936656, "grad_norm": 0.39731743931770325, "learning_rate": 4.700131926939379e-06, "loss": 0.13782256841659546, "step": 1475 }, { "epoch": 0.38317757009345793, "grad_norm": 2.3231611251831055, "learning_rate": 4.699659898466527e-06, "loss": 0.17783460021018982, "step": 1476 }, { "epoch": 0.38343717549325024, "grad_norm": 0.9867796897888184, "learning_rate": 4.699187525338447e-06, "loss": 0.1603970229625702, "step": 1477 }, { "epoch": 0.3836967808930426, "grad_norm": 1.6447802782058716, "learning_rate": 4.698714807638641e-06, "loss": 0.1613849401473999, "step": 1478 }, { "epoch": 0.3839563862928349, "grad_norm": 0.9194654822349548, "learning_rate": 4.698241745450678e-06, "loss": 0.15598605573177338, "step": 1479 }, { "epoch": 0.3842159916926272, "grad_norm": 1.817332148551941, "learning_rate": 4.697768338858183e-06, "loss": 0.13058951497077942, "step": 1480 }, { "epoch": 0.3844755970924195, "grad_norm": 0.9876196980476379, "learning_rate": 4.697294587944844e-06, "loss": 0.16914960741996765, "step": 1481 }, { "epoch": 0.3847352024922118, "grad_norm": 3.52862286567688, "learning_rate": 4.696820492794408e-06, "loss": 0.15640082955360413, "step": 1482 }, { "epoch": 0.38499480789200413, "grad_norm": 0.40970009565353394, "learning_rate": 4.6963460534906865e-06, "loss": 0.13563212752342224, "step": 1483 }, { "epoch": 0.3852544132917965, "grad_norm": 1.8771617412567139, "learning_rate": 4.6958712701175465e-06, "loss": 0.13497436046600342, "step": 1484 }, { "epoch": 0.3855140186915888, "grad_norm": 0.7591804265975952, "learning_rate": 4.695396142758922e-06, "loss": 0.14346207678318024, "step": 1485 }, { "epoch": 0.3857736240913811, "grad_norm": 1.942435622215271, "learning_rate": 4.694920671498802e-06, "loss": 0.15291547775268555, "step": 1486 }, { "epoch": 0.3860332294911734, "grad_norm": 1.3351961374282837, "learning_rate": 4.694444856421242e-06, "loss": 0.14861437678337097, "step": 1487 }, { "epoch": 0.3862928348909657, "grad_norm": 1.1277055740356445, "learning_rate": 4.693968697610351e-06, "loss": 0.1888297200202942, "step": 1488 }, { "epoch": 0.386552440290758, "grad_norm": 1.6556798219680786, "learning_rate": 4.693492195150306e-06, "loss": 0.1647450029850006, "step": 1489 }, { "epoch": 0.3868120456905504, "grad_norm": 3.052086114883423, "learning_rate": 4.693015349125341e-06, "loss": 0.15372014045715332, "step": 1490 }, { "epoch": 0.3870716510903427, "grad_norm": 1.2682162523269653, "learning_rate": 4.692538159619751e-06, "loss": 0.16024187207221985, "step": 1491 }, { "epoch": 0.387331256490135, "grad_norm": 1.7497435808181763, "learning_rate": 4.692060626717894e-06, "loss": 0.13084964454174042, "step": 1492 }, { "epoch": 0.3875908618899273, "grad_norm": 0.7820917963981628, "learning_rate": 4.691582750504184e-06, "loss": 0.1332128643989563, "step": 1493 }, { "epoch": 0.3878504672897196, "grad_norm": 3.0337486267089844, "learning_rate": 4.691104531063101e-06, "loss": 0.17444400489330292, "step": 1494 }, { "epoch": 0.3881100726895119, "grad_norm": 0.5446790456771851, "learning_rate": 4.6906259684791816e-06, "loss": 0.13431572914123535, "step": 1495 }, { "epoch": 0.3883696780893043, "grad_norm": 0.8867394924163818, "learning_rate": 4.690147062837026e-06, "loss": 0.1331116408109665, "step": 1496 }, { "epoch": 0.3886292834890966, "grad_norm": 0.8451734185218811, "learning_rate": 4.689667814221293e-06, "loss": 0.1328561156988144, "step": 1497 }, { "epoch": 0.3888888888888889, "grad_norm": 1.2995920181274414, "learning_rate": 4.689188222716703e-06, "loss": 0.14129146933555603, "step": 1498 }, { "epoch": 0.3891484942886812, "grad_norm": 1.0196330547332764, "learning_rate": 4.688708288408036e-06, "loss": 0.16275203227996826, "step": 1499 }, { "epoch": 0.3894080996884735, "grad_norm": 1.1882455348968506, "learning_rate": 4.688228011380135e-06, "loss": 0.1502024084329605, "step": 1500 }, { "epoch": 0.3896677050882658, "grad_norm": 0.8873152136802673, "learning_rate": 4.687747391717901e-06, "loss": 0.14724291861057281, "step": 1501 }, { "epoch": 0.3899273104880582, "grad_norm": 0.9247148633003235, "learning_rate": 4.687266429506296e-06, "loss": 0.13803225755691528, "step": 1502 }, { "epoch": 0.3901869158878505, "grad_norm": 0.9519082903862, "learning_rate": 4.686785124830345e-06, "loss": 0.1319287121295929, "step": 1503 }, { "epoch": 0.3904465212876428, "grad_norm": 1.6848118305206299, "learning_rate": 4.686303477775129e-06, "loss": 0.15646964311599731, "step": 1504 }, { "epoch": 0.3907061266874351, "grad_norm": 0.7959449887275696, "learning_rate": 4.685821488425795e-06, "loss": 0.1412193775177002, "step": 1505 }, { "epoch": 0.3909657320872274, "grad_norm": 1.0111733675003052, "learning_rate": 4.6853391568675456e-06, "loss": 0.14509868621826172, "step": 1506 }, { "epoch": 0.3912253374870197, "grad_norm": 0.9311089515686035, "learning_rate": 4.684856483185647e-06, "loss": 0.12968629598617554, "step": 1507 }, { "epoch": 0.39148494288681207, "grad_norm": 2.1710641384124756, "learning_rate": 4.684373467465424e-06, "loss": 0.1612618863582611, "step": 1508 }, { "epoch": 0.3917445482866044, "grad_norm": 0.836155116558075, "learning_rate": 4.683890109792264e-06, "loss": 0.17391636967658997, "step": 1509 }, { "epoch": 0.3920041536863967, "grad_norm": 1.1164872646331787, "learning_rate": 4.683406410251612e-06, "loss": 0.16961835324764252, "step": 1510 }, { "epoch": 0.392263759086189, "grad_norm": 0.6873112916946411, "learning_rate": 4.682922368928977e-06, "loss": 0.12835076451301575, "step": 1511 }, { "epoch": 0.3925233644859813, "grad_norm": 1.3869231939315796, "learning_rate": 4.682437985909924e-06, "loss": 0.14813420176506042, "step": 1512 }, { "epoch": 0.3927829698857736, "grad_norm": 1.5145564079284668, "learning_rate": 4.6819532612800824e-06, "loss": 0.16683810949325562, "step": 1513 }, { "epoch": 0.39304257528556596, "grad_norm": 2.738891124725342, "learning_rate": 4.68146819512514e-06, "loss": 0.20122961699962616, "step": 1514 }, { "epoch": 0.39330218068535827, "grad_norm": 0.7786145210266113, "learning_rate": 4.680982787530845e-06, "loss": 0.1611742079257965, "step": 1515 }, { "epoch": 0.3935617860851506, "grad_norm": 1.5390934944152832, "learning_rate": 4.680497038583008e-06, "loss": 0.1697726845741272, "step": 1516 }, { "epoch": 0.3938213914849429, "grad_norm": 1.2910408973693848, "learning_rate": 4.680010948367496e-06, "loss": 0.14787325263023376, "step": 1517 }, { "epoch": 0.3940809968847352, "grad_norm": 0.42705971002578735, "learning_rate": 4.67952451697024e-06, "loss": 0.1335919201374054, "step": 1518 }, { "epoch": 0.3943406022845275, "grad_norm": 1.5555458068847656, "learning_rate": 4.67903774447723e-06, "loss": 0.15973392128944397, "step": 1519 }, { "epoch": 0.39460020768431986, "grad_norm": 1.3210484981536865, "learning_rate": 4.678550630974515e-06, "loss": 0.20394469797611237, "step": 1520 }, { "epoch": 0.39485981308411217, "grad_norm": 0.6268264055252075, "learning_rate": 4.678063176548209e-06, "loss": 0.1378369927406311, "step": 1521 }, { "epoch": 0.39511941848390447, "grad_norm": 0.3770848214626312, "learning_rate": 4.677575381284478e-06, "loss": 0.13903997838497162, "step": 1522 }, { "epoch": 0.3953790238836968, "grad_norm": 1.0715868473052979, "learning_rate": 4.6770872452695546e-06, "loss": 0.13497236371040344, "step": 1523 }, { "epoch": 0.3956386292834891, "grad_norm": 1.8251368999481201, "learning_rate": 4.6765987685897315e-06, "loss": 0.1731567680835724, "step": 1524 }, { "epoch": 0.3958982346832814, "grad_norm": 1.4299705028533936, "learning_rate": 4.6761099513313595e-06, "loss": 0.1744992733001709, "step": 1525 }, { "epoch": 0.39615784008307375, "grad_norm": 1.1484472751617432, "learning_rate": 4.67562079358085e-06, "loss": 0.1652020812034607, "step": 1526 }, { "epoch": 0.39641744548286606, "grad_norm": 0.4331541955471039, "learning_rate": 4.675131295424675e-06, "loss": 0.13348960876464844, "step": 1527 }, { "epoch": 0.39667705088265837, "grad_norm": 0.8704463839530945, "learning_rate": 4.674641456949368e-06, "loss": 0.15901100635528564, "step": 1528 }, { "epoch": 0.3969366562824507, "grad_norm": 1.6115771532058716, "learning_rate": 4.674151278241518e-06, "loss": 0.16367676854133606, "step": 1529 }, { "epoch": 0.397196261682243, "grad_norm": 0.6941859126091003, "learning_rate": 4.673660759387779e-06, "loss": 0.14474952220916748, "step": 1530 }, { "epoch": 0.3974558670820353, "grad_norm": 3.7740561962127686, "learning_rate": 4.673169900474865e-06, "loss": 0.1853797733783722, "step": 1531 }, { "epoch": 0.39771547248182765, "grad_norm": 0.7824135422706604, "learning_rate": 4.672678701589547e-06, "loss": 0.14229251444339752, "step": 1532 }, { "epoch": 0.39797507788161995, "grad_norm": 1.301961898803711, "learning_rate": 4.672187162818658e-06, "loss": 0.1292920857667923, "step": 1533 }, { "epoch": 0.39823468328141226, "grad_norm": 1.6618423461914062, "learning_rate": 4.671695284249091e-06, "loss": 0.12738589942455292, "step": 1534 }, { "epoch": 0.39849428868120457, "grad_norm": 0.6321597099304199, "learning_rate": 4.671203065967799e-06, "loss": 0.11932814866304398, "step": 1535 }, { "epoch": 0.3987538940809969, "grad_norm": 2.4011154174804688, "learning_rate": 4.670710508061794e-06, "loss": 0.18149429559707642, "step": 1536 }, { "epoch": 0.3990134994807892, "grad_norm": 1.7599443197250366, "learning_rate": 4.67021761061815e-06, "loss": 0.15720638632774353, "step": 1537 }, { "epoch": 0.39927310488058154, "grad_norm": 1.235481858253479, "learning_rate": 4.669724373724e-06, "loss": 0.1770581305027008, "step": 1538 }, { "epoch": 0.39953271028037385, "grad_norm": 1.2423055171966553, "learning_rate": 4.6692307974665355e-06, "loss": 0.179568350315094, "step": 1539 }, { "epoch": 0.39979231568016615, "grad_norm": 1.131093978881836, "learning_rate": 4.668736881933012e-06, "loss": 0.13509704172611237, "step": 1540 }, { "epoch": 0.40005192107995846, "grad_norm": 2.1220614910125732, "learning_rate": 4.66824262721074e-06, "loss": 0.18129834532737732, "step": 1541 }, { "epoch": 0.40031152647975077, "grad_norm": 0.964339554309845, "learning_rate": 4.6677480333870935e-06, "loss": 0.15733908116817474, "step": 1542 }, { "epoch": 0.4005711318795431, "grad_norm": 0.9195988178253174, "learning_rate": 4.667253100549506e-06, "loss": 0.1457013189792633, "step": 1543 }, { "epoch": 0.40083073727933544, "grad_norm": 0.4887255132198334, "learning_rate": 4.666757828785469e-06, "loss": 0.14161741733551025, "step": 1544 }, { "epoch": 0.40109034267912774, "grad_norm": 1.1497405767440796, "learning_rate": 4.666262218182535e-06, "loss": 0.14011338353157043, "step": 1545 }, { "epoch": 0.40134994807892005, "grad_norm": 1.2371922731399536, "learning_rate": 4.665766268828319e-06, "loss": 0.13215480744838715, "step": 1546 }, { "epoch": 0.40160955347871236, "grad_norm": 1.8883112668991089, "learning_rate": 4.66526998081049e-06, "loss": 0.14158892631530762, "step": 1547 }, { "epoch": 0.40186915887850466, "grad_norm": 0.9280254244804382, "learning_rate": 4.664773354216785e-06, "loss": 0.1304088532924652, "step": 1548 }, { "epoch": 0.40212876427829697, "grad_norm": 2.4456334114074707, "learning_rate": 4.664276389134991e-06, "loss": 0.14419789612293243, "step": 1549 }, { "epoch": 0.40238836967808933, "grad_norm": 3.2109272480010986, "learning_rate": 4.663779085652964e-06, "loss": 0.16457664966583252, "step": 1550 }, { "epoch": 0.40264797507788164, "grad_norm": 1.5362805128097534, "learning_rate": 4.663281443858614e-06, "loss": 0.16717132925987244, "step": 1551 }, { "epoch": 0.40290758047767394, "grad_norm": 1.0331073999404907, "learning_rate": 4.662783463839913e-06, "loss": 0.13546475768089294, "step": 1552 }, { "epoch": 0.40316718587746625, "grad_norm": 2.4418487548828125, "learning_rate": 4.662285145684892e-06, "loss": 0.17511290311813354, "step": 1553 }, { "epoch": 0.40342679127725856, "grad_norm": 3.4899353981018066, "learning_rate": 4.661786489481644e-06, "loss": 0.19403982162475586, "step": 1554 }, { "epoch": 0.40368639667705086, "grad_norm": 0.9144020080566406, "learning_rate": 4.661287495318317e-06, "loss": 0.1685030311346054, "step": 1555 }, { "epoch": 0.4039460020768432, "grad_norm": 1.3298816680908203, "learning_rate": 4.660788163283125e-06, "loss": 0.19072601199150085, "step": 1556 }, { "epoch": 0.40420560747663553, "grad_norm": 1.3101601600646973, "learning_rate": 4.6602884934643364e-06, "loss": 0.18083252012729645, "step": 1557 }, { "epoch": 0.40446521287642784, "grad_norm": 1.4980857372283936, "learning_rate": 4.659788485950282e-06, "loss": 0.16682901978492737, "step": 1558 }, { "epoch": 0.40472481827622014, "grad_norm": 0.6488878726959229, "learning_rate": 4.659288140829351e-06, "loss": 0.1407448798418045, "step": 1559 }, { "epoch": 0.40498442367601245, "grad_norm": 0.6661688089370728, "learning_rate": 4.658787458189995e-06, "loss": 0.14830699563026428, "step": 1560 }, { "epoch": 0.40524402907580476, "grad_norm": 1.2328866720199585, "learning_rate": 4.658286438120721e-06, "loss": 0.17142997682094574, "step": 1561 }, { "epoch": 0.4055036344755971, "grad_norm": 0.48988866806030273, "learning_rate": 4.657785080710099e-06, "loss": 0.1712334007024765, "step": 1562 }, { "epoch": 0.4057632398753894, "grad_norm": 0.5301328301429749, "learning_rate": 4.657283386046757e-06, "loss": 0.13259285688400269, "step": 1563 }, { "epoch": 0.40602284527518173, "grad_norm": 0.6876408457756042, "learning_rate": 4.656781354219383e-06, "loss": 0.17346766591072083, "step": 1564 }, { "epoch": 0.40628245067497404, "grad_norm": 0.367214173078537, "learning_rate": 4.656278985316725e-06, "loss": 0.13449488580226898, "step": 1565 }, { "epoch": 0.40654205607476634, "grad_norm": 0.6601355671882629, "learning_rate": 4.655776279427591e-06, "loss": 0.1365434229373932, "step": 1566 }, { "epoch": 0.40680166147455865, "grad_norm": 0.8429990410804749, "learning_rate": 4.655273236640847e-06, "loss": 0.13453826308250427, "step": 1567 }, { "epoch": 0.407061266874351, "grad_norm": 0.8959377408027649, "learning_rate": 4.6547698570454206e-06, "loss": 0.16507157683372498, "step": 1568 }, { "epoch": 0.4073208722741433, "grad_norm": 1.103081464767456, "learning_rate": 4.654266140730297e-06, "loss": 0.17045235633850098, "step": 1569 }, { "epoch": 0.4075804776739356, "grad_norm": 0.4880412817001343, "learning_rate": 4.653762087784521e-06, "loss": 0.12414804100990295, "step": 1570 }, { "epoch": 0.40784008307372793, "grad_norm": 1.5001251697540283, "learning_rate": 4.653257698297198e-06, "loss": 0.12884286046028137, "step": 1571 }, { "epoch": 0.40809968847352024, "grad_norm": 0.6880617737770081, "learning_rate": 4.652752972357492e-06, "loss": 0.17392542958259583, "step": 1572 }, { "epoch": 0.40835929387331255, "grad_norm": 1.0094590187072754, "learning_rate": 4.652247910054629e-06, "loss": 0.14792871475219727, "step": 1573 }, { "epoch": 0.4086188992731049, "grad_norm": 1.573278784751892, "learning_rate": 4.6517425114778905e-06, "loss": 0.16084638237953186, "step": 1574 }, { "epoch": 0.4088785046728972, "grad_norm": 1.5208933353424072, "learning_rate": 4.65123677671662e-06, "loss": 0.12744355201721191, "step": 1575 }, { "epoch": 0.4091381100726895, "grad_norm": 1.8674077987670898, "learning_rate": 4.6507307058602205e-06, "loss": 0.13546645641326904, "step": 1576 }, { "epoch": 0.4093977154724818, "grad_norm": 0.9354351758956909, "learning_rate": 4.650224298998154e-06, "loss": 0.13740739226341248, "step": 1577 }, { "epoch": 0.40965732087227413, "grad_norm": 1.3202701807022095, "learning_rate": 4.649717556219939e-06, "loss": 0.1627960503101349, "step": 1578 }, { "epoch": 0.40991692627206644, "grad_norm": 1.092005729675293, "learning_rate": 4.649210477615159e-06, "loss": 0.14599072933197021, "step": 1579 }, { "epoch": 0.4101765316718588, "grad_norm": 1.2170952558517456, "learning_rate": 4.648703063273452e-06, "loss": 0.146516352891922, "step": 1580 }, { "epoch": 0.4104361370716511, "grad_norm": 1.1856191158294678, "learning_rate": 4.6481953132845184e-06, "loss": 0.18800124526023865, "step": 1581 }, { "epoch": 0.4106957424714434, "grad_norm": 1.410839319229126, "learning_rate": 4.647687227738116e-06, "loss": 0.18299609422683716, "step": 1582 }, { "epoch": 0.4109553478712357, "grad_norm": 0.7865191102027893, "learning_rate": 4.647178806724064e-06, "loss": 0.13167861104011536, "step": 1583 }, { "epoch": 0.411214953271028, "grad_norm": 0.5359923243522644, "learning_rate": 4.646670050332238e-06, "loss": 0.13323849439620972, "step": 1584 }, { "epoch": 0.41147455867082033, "grad_norm": 4.145393371582031, "learning_rate": 4.646160958652575e-06, "loss": 0.17173835635185242, "step": 1585 }, { "epoch": 0.4117341640706127, "grad_norm": 0.576059103012085, "learning_rate": 4.645651531775071e-06, "loss": 0.12786099314689636, "step": 1586 }, { "epoch": 0.411993769470405, "grad_norm": 0.833230197429657, "learning_rate": 4.64514176978978e-06, "loss": 0.16207656264305115, "step": 1587 }, { "epoch": 0.4122533748701973, "grad_norm": 1.2251582145690918, "learning_rate": 4.6446316727868175e-06, "loss": 0.17889656126499176, "step": 1588 }, { "epoch": 0.4125129802699896, "grad_norm": 1.2748032808303833, "learning_rate": 4.644121240856357e-06, "loss": 0.16233229637145996, "step": 1589 }, { "epoch": 0.4127725856697819, "grad_norm": 1.3470220565795898, "learning_rate": 4.6436104740886305e-06, "loss": 0.13039903342723846, "step": 1590 }, { "epoch": 0.41303219106957423, "grad_norm": 0.6405133605003357, "learning_rate": 4.64309937257393e-06, "loss": 0.17288869619369507, "step": 1591 }, { "epoch": 0.4132917964693666, "grad_norm": 2.9446780681610107, "learning_rate": 4.642587936402607e-06, "loss": 0.16228951513767242, "step": 1592 }, { "epoch": 0.4135514018691589, "grad_norm": 0.9198150634765625, "learning_rate": 4.642076165665071e-06, "loss": 0.1423719823360443, "step": 1593 }, { "epoch": 0.4138110072689512, "grad_norm": 0.6359437108039856, "learning_rate": 4.641564060451792e-06, "loss": 0.1496790051460266, "step": 1594 }, { "epoch": 0.4140706126687435, "grad_norm": 0.6721006035804749, "learning_rate": 4.641051620853298e-06, "loss": 0.13992910087108612, "step": 1595 }, { "epoch": 0.4143302180685358, "grad_norm": 1.6325494050979614, "learning_rate": 4.640538846960177e-06, "loss": 0.1428951621055603, "step": 1596 }, { "epoch": 0.4145898234683281, "grad_norm": 5.423976898193359, "learning_rate": 4.640025738863076e-06, "loss": 0.18432486057281494, "step": 1597 }, { "epoch": 0.4148494288681205, "grad_norm": 1.83668053150177, "learning_rate": 4.639512296652699e-06, "loss": 0.19806718826293945, "step": 1598 }, { "epoch": 0.4151090342679128, "grad_norm": 0.39371559023857117, "learning_rate": 4.638998520419812e-06, "loss": 0.11587470769882202, "step": 1599 }, { "epoch": 0.4153686396677051, "grad_norm": 1.4022338390350342, "learning_rate": 4.63848441025524e-06, "loss": 0.1233137771487236, "step": 1600 }, { "epoch": 0.4156282450674974, "grad_norm": 1.6156644821166992, "learning_rate": 4.6379699662498646e-06, "loss": 0.17328093945980072, "step": 1601 }, { "epoch": 0.4158878504672897, "grad_norm": 2.6533567905426025, "learning_rate": 4.637455188494627e-06, "loss": 0.1939099133014679, "step": 1602 }, { "epoch": 0.416147455867082, "grad_norm": 0.637993335723877, "learning_rate": 4.636940077080531e-06, "loss": 0.13586542010307312, "step": 1603 }, { "epoch": 0.4164070612668744, "grad_norm": 1.2970858812332153, "learning_rate": 4.6364246320986325e-06, "loss": 0.14073729515075684, "step": 1604 }, { "epoch": 0.4166666666666667, "grad_norm": 0.8848934173583984, "learning_rate": 4.6359088536400534e-06, "loss": 0.13462215662002563, "step": 1605 }, { "epoch": 0.416926272066459, "grad_norm": 1.7168787717819214, "learning_rate": 4.635392741795971e-06, "loss": 0.14987812936306, "step": 1606 }, { "epoch": 0.4171858774662513, "grad_norm": 1.2917581796646118, "learning_rate": 4.63487629665762e-06, "loss": 0.17237871885299683, "step": 1607 }, { "epoch": 0.4174454828660436, "grad_norm": 2.046635389328003, "learning_rate": 4.634359518316299e-06, "loss": 0.1941310316324234, "step": 1608 }, { "epoch": 0.4177050882658359, "grad_norm": 1.3207143545150757, "learning_rate": 4.6338424068633615e-06, "loss": 0.15079393982887268, "step": 1609 }, { "epoch": 0.4179646936656282, "grad_norm": 2.0914604663848877, "learning_rate": 4.6333249623902195e-06, "loss": 0.13451164960861206, "step": 1610 }, { "epoch": 0.4182242990654206, "grad_norm": 1.3538421392440796, "learning_rate": 4.632807184988348e-06, "loss": 0.12968912720680237, "step": 1611 }, { "epoch": 0.4184839044652129, "grad_norm": 0.7330348491668701, "learning_rate": 4.6322890747492765e-06, "loss": 0.15952475368976593, "step": 1612 }, { "epoch": 0.4187435098650052, "grad_norm": 0.6634077429771423, "learning_rate": 4.631770631764595e-06, "loss": 0.1332464963197708, "step": 1613 }, { "epoch": 0.4190031152647975, "grad_norm": 1.5263385772705078, "learning_rate": 4.631251856125952e-06, "loss": 0.18358483910560608, "step": 1614 }, { "epoch": 0.4192627206645898, "grad_norm": 2.0318357944488525, "learning_rate": 4.630732747925057e-06, "loss": 0.1662733256816864, "step": 1615 }, { "epoch": 0.4195223260643821, "grad_norm": 0.8023936152458191, "learning_rate": 4.630213307253674e-06, "loss": 0.14277903735637665, "step": 1616 }, { "epoch": 0.4197819314641745, "grad_norm": 0.6411685943603516, "learning_rate": 4.6296935342036305e-06, "loss": 0.17403413355350494, "step": 1617 }, { "epoch": 0.4200415368639668, "grad_norm": 1.7959051132202148, "learning_rate": 4.629173428866808e-06, "loss": 0.1415567547082901, "step": 1618 }, { "epoch": 0.4203011422637591, "grad_norm": 1.782177448272705, "learning_rate": 4.628652991335153e-06, "loss": 0.16606424748897552, "step": 1619 }, { "epoch": 0.4205607476635514, "grad_norm": 0.5777544379234314, "learning_rate": 4.628132221700663e-06, "loss": 0.12887181341648102, "step": 1620 }, { "epoch": 0.4208203530633437, "grad_norm": 0.30373096466064453, "learning_rate": 4.627611120055399e-06, "loss": 0.12429970502853394, "step": 1621 }, { "epoch": 0.421079958463136, "grad_norm": 1.1561537981033325, "learning_rate": 4.627089686491482e-06, "loss": 0.1287899911403656, "step": 1622 }, { "epoch": 0.42133956386292837, "grad_norm": 1.2928836345672607, "learning_rate": 4.626567921101087e-06, "loss": 0.153412863612175, "step": 1623 }, { "epoch": 0.4215991692627207, "grad_norm": 0.6758120059967041, "learning_rate": 4.626045823976453e-06, "loss": 0.13084357976913452, "step": 1624 }, { "epoch": 0.421858774662513, "grad_norm": 0.3793712556362152, "learning_rate": 4.625523395209871e-06, "loss": 0.11863264441490173, "step": 1625 }, { "epoch": 0.4221183800623053, "grad_norm": 0.5575692653656006, "learning_rate": 4.625000634893698e-06, "loss": 0.1416250318288803, "step": 1626 }, { "epoch": 0.4223779854620976, "grad_norm": 0.9998924136161804, "learning_rate": 4.624477543120344e-06, "loss": 0.16649802029132843, "step": 1627 }, { "epoch": 0.4226375908618899, "grad_norm": 0.8630498051643372, "learning_rate": 4.62395411998228e-06, "loss": 0.1400529146194458, "step": 1628 }, { "epoch": 0.42289719626168226, "grad_norm": 1.5385007858276367, "learning_rate": 4.623430365572036e-06, "loss": 0.1336202323436737, "step": 1629 }, { "epoch": 0.42315680166147457, "grad_norm": 1.9885880947113037, "learning_rate": 4.622906279982199e-06, "loss": 0.19756491482257843, "step": 1630 }, { "epoch": 0.4234164070612669, "grad_norm": 0.8268086910247803, "learning_rate": 4.6223818633054155e-06, "loss": 0.1414109766483307, "step": 1631 }, { "epoch": 0.4236760124610592, "grad_norm": 0.3698302209377289, "learning_rate": 4.62185711563439e-06, "loss": 0.13444489240646362, "step": 1632 }, { "epoch": 0.4239356178608515, "grad_norm": 0.43688273429870605, "learning_rate": 4.621332037061887e-06, "loss": 0.1258346438407898, "step": 1633 }, { "epoch": 0.4241952232606438, "grad_norm": 1.5690762996673584, "learning_rate": 4.620806627680728e-06, "loss": 0.1699170172214508, "step": 1634 }, { "epoch": 0.42445482866043616, "grad_norm": 0.7503250241279602, "learning_rate": 4.6202808875837915e-06, "loss": 0.16885992884635925, "step": 1635 }, { "epoch": 0.42471443406022846, "grad_norm": 0.7251030206680298, "learning_rate": 4.619754816864019e-06, "loss": 0.17381741106510162, "step": 1636 }, { "epoch": 0.42497403946002077, "grad_norm": 0.6124511957168579, "learning_rate": 4.619228415614406e-06, "loss": 0.15460234880447388, "step": 1637 }, { "epoch": 0.4252336448598131, "grad_norm": 1.0978144407272339, "learning_rate": 4.61870168392801e-06, "loss": 0.14465171098709106, "step": 1638 }, { "epoch": 0.4254932502596054, "grad_norm": 1.6433937549591064, "learning_rate": 4.6181746218979424e-06, "loss": 0.1281154751777649, "step": 1639 }, { "epoch": 0.4257528556593977, "grad_norm": 2.7609481811523438, "learning_rate": 4.6176472296173775e-06, "loss": 0.16428840160369873, "step": 1640 }, { "epoch": 0.42601246105919005, "grad_norm": 1.0497987270355225, "learning_rate": 4.617119507179546e-06, "loss": 0.13646328449249268, "step": 1641 }, { "epoch": 0.42627206645898236, "grad_norm": 1.2291896343231201, "learning_rate": 4.616591454677737e-06, "loss": 0.12930969893932343, "step": 1642 }, { "epoch": 0.42653167185877466, "grad_norm": 0.4373193681240082, "learning_rate": 4.616063072205298e-06, "loss": 0.13445904850959778, "step": 1643 }, { "epoch": 0.42679127725856697, "grad_norm": 0.8763113617897034, "learning_rate": 4.615534359855636e-06, "loss": 0.16310299932956696, "step": 1644 }, { "epoch": 0.4270508826583593, "grad_norm": 1.836077332496643, "learning_rate": 4.615005317722213e-06, "loss": 0.15805235505104065, "step": 1645 }, { "epoch": 0.4273104880581516, "grad_norm": 1.6918624639511108, "learning_rate": 4.614475945898554e-06, "loss": 0.1332392394542694, "step": 1646 }, { "epoch": 0.42757009345794394, "grad_norm": 0.38597360253334045, "learning_rate": 4.613946244478238e-06, "loss": 0.1299980878829956, "step": 1647 }, { "epoch": 0.42782969885773625, "grad_norm": 0.6709108352661133, "learning_rate": 4.613416213554904e-06, "loss": 0.12143685668706894, "step": 1648 }, { "epoch": 0.42808930425752856, "grad_norm": 0.6332637071609497, "learning_rate": 4.61288585322225e-06, "loss": 0.11876001954078674, "step": 1649 }, { "epoch": 0.42834890965732086, "grad_norm": 1.1269888877868652, "learning_rate": 4.6123551635740336e-06, "loss": 0.1316554844379425, "step": 1650 }, { "epoch": 0.42860851505711317, "grad_norm": 0.760967493057251, "learning_rate": 4.611824144704064e-06, "loss": 0.1450500786304474, "step": 1651 }, { "epoch": 0.4288681204569055, "grad_norm": 1.1742219924926758, "learning_rate": 4.611292796706218e-06, "loss": 0.15188172459602356, "step": 1652 }, { "epoch": 0.42912772585669784, "grad_norm": 0.4416950047016144, "learning_rate": 4.610761119674422e-06, "loss": 0.12869085371494293, "step": 1653 }, { "epoch": 0.42938733125649015, "grad_norm": 0.6286911964416504, "learning_rate": 4.610229113702666e-06, "loss": 0.13834135234355927, "step": 1654 }, { "epoch": 0.42964693665628245, "grad_norm": 1.1902716159820557, "learning_rate": 4.609696778884997e-06, "loss": 0.13256415724754333, "step": 1655 }, { "epoch": 0.42990654205607476, "grad_norm": 1.7379165887832642, "learning_rate": 4.6091641153155175e-06, "loss": 0.14221517741680145, "step": 1656 }, { "epoch": 0.43016614745586707, "grad_norm": 6.300755977630615, "learning_rate": 4.608631123088393e-06, "loss": 0.16881400346755981, "step": 1657 }, { "epoch": 0.43042575285565937, "grad_norm": 1.216483473777771, "learning_rate": 4.6080978022978425e-06, "loss": 0.1400473415851593, "step": 1658 }, { "epoch": 0.43068535825545173, "grad_norm": 0.7103630304336548, "learning_rate": 4.607564153038146e-06, "loss": 0.14067482948303223, "step": 1659 }, { "epoch": 0.43094496365524404, "grad_norm": 0.7678994536399841, "learning_rate": 4.607030175403638e-06, "loss": 0.1478848159313202, "step": 1660 }, { "epoch": 0.43120456905503635, "grad_norm": 1.999974250793457, "learning_rate": 4.606495869488716e-06, "loss": 0.16728365421295166, "step": 1661 }, { "epoch": 0.43146417445482865, "grad_norm": 1.183729648590088, "learning_rate": 4.605961235387834e-06, "loss": 0.16643503308296204, "step": 1662 }, { "epoch": 0.43172377985462096, "grad_norm": 2.1812009811401367, "learning_rate": 4.605426273195499e-06, "loss": 0.15462610125541687, "step": 1663 }, { "epoch": 0.43198338525441327, "grad_norm": 1.771714687347412, "learning_rate": 4.604890983006285e-06, "loss": 0.1439315527677536, "step": 1664 }, { "epoch": 0.4322429906542056, "grad_norm": 2.0076005458831787, "learning_rate": 4.604355364914816e-06, "loss": 0.17372268438339233, "step": 1665 }, { "epoch": 0.43250259605399793, "grad_norm": 1.954439401626587, "learning_rate": 4.603819419015776e-06, "loss": 0.1633705198764801, "step": 1666 }, { "epoch": 0.43276220145379024, "grad_norm": 1.4161335229873657, "learning_rate": 4.603283145403911e-06, "loss": 0.13667839765548706, "step": 1667 }, { "epoch": 0.43302180685358255, "grad_norm": 2.1409714221954346, "learning_rate": 4.602746544174021e-06, "loss": 0.13789242506027222, "step": 1668 }, { "epoch": 0.43328141225337485, "grad_norm": 2.0848162174224854, "learning_rate": 4.602209615420963e-06, "loss": 0.1673193722963333, "step": 1669 }, { "epoch": 0.43354101765316716, "grad_norm": 0.7337987422943115, "learning_rate": 4.601672359239654e-06, "loss": 0.128505676984787, "step": 1670 }, { "epoch": 0.4338006230529595, "grad_norm": 1.3298476934432983, "learning_rate": 4.6011347757250716e-06, "loss": 0.15750530362129211, "step": 1671 }, { "epoch": 0.43406022845275183, "grad_norm": 3.4022810459136963, "learning_rate": 4.600596864972246e-06, "loss": 0.18595927953720093, "step": 1672 }, { "epoch": 0.43431983385254413, "grad_norm": 1.2145408391952515, "learning_rate": 4.600058627076266e-06, "loss": 0.13097234070301056, "step": 1673 }, { "epoch": 0.43457943925233644, "grad_norm": 1.134260654449463, "learning_rate": 4.599520062132283e-06, "loss": 0.15433043241500854, "step": 1674 }, { "epoch": 0.43483904465212875, "grad_norm": 1.20473051071167, "learning_rate": 4.598981170235501e-06, "loss": 0.16148236393928528, "step": 1675 }, { "epoch": 0.43509865005192105, "grad_norm": 0.4596278667449951, "learning_rate": 4.598441951481184e-06, "loss": 0.12514427304267883, "step": 1676 }, { "epoch": 0.4353582554517134, "grad_norm": 2.737642288208008, "learning_rate": 4.597902405964654e-06, "loss": 0.13931241631507874, "step": 1677 }, { "epoch": 0.4356178608515057, "grad_norm": 0.9223387837409973, "learning_rate": 4.59736253378129e-06, "loss": 0.15324196219444275, "step": 1678 }, { "epoch": 0.43587746625129803, "grad_norm": 0.6365410685539246, "learning_rate": 4.596822335026529e-06, "loss": 0.13211208581924438, "step": 1679 }, { "epoch": 0.43613707165109034, "grad_norm": 1.0692932605743408, "learning_rate": 4.596281809795866e-06, "loss": 0.16556736826896667, "step": 1680 }, { "epoch": 0.43639667705088264, "grad_norm": 1.2459734678268433, "learning_rate": 4.5957409581848535e-06, "loss": 0.14360295236110687, "step": 1681 }, { "epoch": 0.43665628245067495, "grad_norm": 2.430497407913208, "learning_rate": 4.595199780289101e-06, "loss": 0.19826209545135498, "step": 1682 }, { "epoch": 0.4369158878504673, "grad_norm": 0.4663613438606262, "learning_rate": 4.5946582762042765e-06, "loss": 0.1485881507396698, "step": 1683 }, { "epoch": 0.4371754932502596, "grad_norm": 1.1873013973236084, "learning_rate": 4.594116446026106e-06, "loss": 0.15806373953819275, "step": 1684 }, { "epoch": 0.4374350986500519, "grad_norm": 0.4844723045825958, "learning_rate": 4.5935742898503725e-06, "loss": 0.13309478759765625, "step": 1685 }, { "epoch": 0.43769470404984423, "grad_norm": 0.5499377846717834, "learning_rate": 4.5930318077729165e-06, "loss": 0.1287618726491928, "step": 1686 }, { "epoch": 0.43795430944963654, "grad_norm": 0.5916170477867126, "learning_rate": 4.592488999889638e-06, "loss": 0.1333182156085968, "step": 1687 }, { "epoch": 0.43821391484942884, "grad_norm": 0.5112435221672058, "learning_rate": 4.5919458662964905e-06, "loss": 0.12089111655950546, "step": 1688 }, { "epoch": 0.4384735202492212, "grad_norm": 1.481573462486267, "learning_rate": 4.591402407089489e-06, "loss": 0.14779949188232422, "step": 1689 }, { "epoch": 0.4387331256490135, "grad_norm": 0.868942379951477, "learning_rate": 4.590858622364705e-06, "loss": 0.1375557780265808, "step": 1690 }, { "epoch": 0.4389927310488058, "grad_norm": 0.8002480864524841, "learning_rate": 4.590314512218266e-06, "loss": 0.1569346934556961, "step": 1691 }, { "epoch": 0.4392523364485981, "grad_norm": 0.38387808203697205, "learning_rate": 4.58977007674636e-06, "loss": 0.13489025831222534, "step": 1692 }, { "epoch": 0.43951194184839043, "grad_norm": 1.0727436542510986, "learning_rate": 4.589225316045228e-06, "loss": 0.13296723365783691, "step": 1693 }, { "epoch": 0.43977154724818274, "grad_norm": 0.5544653534889221, "learning_rate": 4.588680230211174e-06, "loss": 0.13181917369365692, "step": 1694 }, { "epoch": 0.4400311526479751, "grad_norm": 1.032894253730774, "learning_rate": 4.588134819340554e-06, "loss": 0.12377907335758209, "step": 1695 }, { "epoch": 0.4402907580477674, "grad_norm": 1.1359468698501587, "learning_rate": 4.587589083529788e-06, "loss": 0.1849934309720993, "step": 1696 }, { "epoch": 0.4405503634475597, "grad_norm": 0.7564195394515991, "learning_rate": 4.587043022875345e-06, "loss": 0.1433834284543991, "step": 1697 }, { "epoch": 0.440809968847352, "grad_norm": 0.3415316641330719, "learning_rate": 4.586496637473759e-06, "loss": 0.12880997359752655, "step": 1698 }, { "epoch": 0.4410695742471443, "grad_norm": 1.3897037506103516, "learning_rate": 4.585949927421618e-06, "loss": 0.14896005392074585, "step": 1699 }, { "epoch": 0.44132917964693663, "grad_norm": 0.6462336182594299, "learning_rate": 4.5854028928155674e-06, "loss": 0.12004268169403076, "step": 1700 }, { "epoch": 0.441588785046729, "grad_norm": 0.5070260763168335, "learning_rate": 4.584855533752311e-06, "loss": 0.12877009809017181, "step": 1701 }, { "epoch": 0.4418483904465213, "grad_norm": 0.6711042523384094, "learning_rate": 4.584307850328608e-06, "loss": 0.12900391221046448, "step": 1702 }, { "epoch": 0.4421079958463136, "grad_norm": 0.7232217788696289, "learning_rate": 4.5837598426412775e-06, "loss": 0.12029361724853516, "step": 1703 }, { "epoch": 0.4423676012461059, "grad_norm": 0.8111364841461182, "learning_rate": 4.583211510787194e-06, "loss": 0.12851779162883759, "step": 1704 }, { "epoch": 0.4426272066458982, "grad_norm": 1.264696478843689, "learning_rate": 4.582662854863291e-06, "loss": 0.12518493831157684, "step": 1705 }, { "epoch": 0.4428868120456905, "grad_norm": 0.9412873983383179, "learning_rate": 4.582113874966558e-06, "loss": 0.18633264303207397, "step": 1706 }, { "epoch": 0.4431464174454829, "grad_norm": 1.7028133869171143, "learning_rate": 4.581564571194041e-06, "loss": 0.1529483050107956, "step": 1707 }, { "epoch": 0.4434060228452752, "grad_norm": 2.5227885246276855, "learning_rate": 4.581014943642845e-06, "loss": 0.1854090392589569, "step": 1708 }, { "epoch": 0.4436656282450675, "grad_norm": 0.6514282822608948, "learning_rate": 4.580464992410132e-06, "loss": 0.1609516441822052, "step": 1709 }, { "epoch": 0.4439252336448598, "grad_norm": 2.423785924911499, "learning_rate": 4.57991471759312e-06, "loss": 0.17067766189575195, "step": 1710 }, { "epoch": 0.4441848390446521, "grad_norm": 1.8695087432861328, "learning_rate": 4.579364119289086e-06, "loss": 0.16581782698631287, "step": 1711 }, { "epoch": 0.4444444444444444, "grad_norm": 0.6979555487632751, "learning_rate": 4.578813197595363e-06, "loss": 0.16430744528770447, "step": 1712 }, { "epoch": 0.4447040498442368, "grad_norm": 0.4715621769428253, "learning_rate": 4.578261952609339e-06, "loss": 0.1396649330854416, "step": 1713 }, { "epoch": 0.4449636552440291, "grad_norm": 1.038863182067871, "learning_rate": 4.577710384428466e-06, "loss": 0.1396864652633667, "step": 1714 }, { "epoch": 0.4452232606438214, "grad_norm": 2.8936426639556885, "learning_rate": 4.577158493150244e-06, "loss": 0.17207050323486328, "step": 1715 }, { "epoch": 0.4454828660436137, "grad_norm": 3.1613988876342773, "learning_rate": 4.576606278872238e-06, "loss": 0.15004119277000427, "step": 1716 }, { "epoch": 0.445742471443406, "grad_norm": 0.873537003993988, "learning_rate": 4.576053741692065e-06, "loss": 0.1631045639514923, "step": 1717 }, { "epoch": 0.4460020768431983, "grad_norm": 2.507040500640869, "learning_rate": 4.575500881707402e-06, "loss": 0.17703057825565338, "step": 1718 }, { "epoch": 0.4462616822429907, "grad_norm": 1.867425799369812, "learning_rate": 4.574947699015981e-06, "loss": 0.1667645275592804, "step": 1719 }, { "epoch": 0.446521287642783, "grad_norm": 2.080010414123535, "learning_rate": 4.574394193715593e-06, "loss": 0.1863013058900833, "step": 1720 }, { "epoch": 0.4467808930425753, "grad_norm": 1.9750624895095825, "learning_rate": 4.573840365904084e-06, "loss": 0.13791269063949585, "step": 1721 }, { "epoch": 0.4470404984423676, "grad_norm": 0.7592383623123169, "learning_rate": 4.57328621567936e-06, "loss": 0.16959847509860992, "step": 1722 }, { "epoch": 0.4473001038421599, "grad_norm": 1.5382407903671265, "learning_rate": 4.57273174313938e-06, "loss": 0.15564945340156555, "step": 1723 }, { "epoch": 0.4475597092419522, "grad_norm": 1.0695828199386597, "learning_rate": 4.5721769483821625e-06, "loss": 0.18897652626037598, "step": 1724 }, { "epoch": 0.44781931464174457, "grad_norm": 2.620821952819824, "learning_rate": 4.571621831505783e-06, "loss": 0.17365501821041107, "step": 1725 }, { "epoch": 0.4480789200415369, "grad_norm": 2.095560312271118, "learning_rate": 4.571066392608374e-06, "loss": 0.18155476450920105, "step": 1726 }, { "epoch": 0.4483385254413292, "grad_norm": 1.3464607000350952, "learning_rate": 4.5705106317881225e-06, "loss": 0.17188197374343872, "step": 1727 }, { "epoch": 0.4485981308411215, "grad_norm": 1.174898386001587, "learning_rate": 4.569954549143276e-06, "loss": 0.16650764644145966, "step": 1728 }, { "epoch": 0.4488577362409138, "grad_norm": 1.1942949295043945, "learning_rate": 4.569398144772137e-06, "loss": 0.13004808127880096, "step": 1729 }, { "epoch": 0.4491173416407061, "grad_norm": 1.1411466598510742, "learning_rate": 4.568841418773064e-06, "loss": 0.15005457401275635, "step": 1730 }, { "epoch": 0.44937694704049846, "grad_norm": 1.422704815864563, "learning_rate": 4.568284371244475e-06, "loss": 0.1510956585407257, "step": 1731 }, { "epoch": 0.44963655244029077, "grad_norm": 0.9065611958503723, "learning_rate": 4.567727002284842e-06, "loss": 0.18431641161441803, "step": 1732 }, { "epoch": 0.4498961578400831, "grad_norm": 1.458550214767456, "learning_rate": 4.567169311992696e-06, "loss": 0.1524624228477478, "step": 1733 }, { "epoch": 0.4501557632398754, "grad_norm": 1.4796394109725952, "learning_rate": 4.566611300466622e-06, "loss": 0.1661050021648407, "step": 1734 }, { "epoch": 0.4504153686396677, "grad_norm": 0.4189717471599579, "learning_rate": 4.566052967805266e-06, "loss": 0.12595830857753754, "step": 1735 }, { "epoch": 0.45067497403946, "grad_norm": 0.6831382513046265, "learning_rate": 4.565494314107327e-06, "loss": 0.13309112191200256, "step": 1736 }, { "epoch": 0.45093457943925236, "grad_norm": 1.8855352401733398, "learning_rate": 4.564935339471563e-06, "loss": 0.14525583386421204, "step": 1737 }, { "epoch": 0.45119418483904467, "grad_norm": 0.5363000631332397, "learning_rate": 4.564376043996787e-06, "loss": 0.15453097224235535, "step": 1738 }, { "epoch": 0.45145379023883697, "grad_norm": 1.306569218635559, "learning_rate": 4.5638164277818705e-06, "loss": 0.1473064422607422, "step": 1739 }, { "epoch": 0.4517133956386293, "grad_norm": 1.1546275615692139, "learning_rate": 4.56325649092574e-06, "loss": 0.1525249034166336, "step": 1740 }, { "epoch": 0.4519730010384216, "grad_norm": 0.6874066591262817, "learning_rate": 4.56269623352738e-06, "loss": 0.15301012992858887, "step": 1741 }, { "epoch": 0.4522326064382139, "grad_norm": 1.0629138946533203, "learning_rate": 4.562135655685831e-06, "loss": 0.153615802526474, "step": 1742 }, { "epoch": 0.45249221183800625, "grad_norm": 0.5311412215232849, "learning_rate": 4.561574757500192e-06, "loss": 0.13673219084739685, "step": 1743 }, { "epoch": 0.45275181723779856, "grad_norm": 0.5026026964187622, "learning_rate": 4.5610135390696135e-06, "loss": 0.13998916745185852, "step": 1744 }, { "epoch": 0.45301142263759087, "grad_norm": 0.6126778721809387, "learning_rate": 4.560452000493309e-06, "loss": 0.11906172335147858, "step": 1745 }, { "epoch": 0.4532710280373832, "grad_norm": 0.6620154976844788, "learning_rate": 4.559890141870545e-06, "loss": 0.15813155472278595, "step": 1746 }, { "epoch": 0.4535306334371755, "grad_norm": 0.9103249311447144, "learning_rate": 4.559327963300643e-06, "loss": 0.1300731897354126, "step": 1747 }, { "epoch": 0.4537902388369678, "grad_norm": 0.8234556317329407, "learning_rate": 4.558765464882987e-06, "loss": 0.19664335250854492, "step": 1748 }, { "epoch": 0.45404984423676015, "grad_norm": 1.2184115648269653, "learning_rate": 4.558202646717011e-06, "loss": 0.13521192967891693, "step": 1749 }, { "epoch": 0.45430944963655245, "grad_norm": 0.7259323000907898, "learning_rate": 4.557639508902211e-06, "loss": 0.13619527220726013, "step": 1750 }, { "epoch": 0.45456905503634476, "grad_norm": 0.7915078997612, "learning_rate": 4.557076051538134e-06, "loss": 0.13062763214111328, "step": 1751 }, { "epoch": 0.45482866043613707, "grad_norm": 0.8794267177581787, "learning_rate": 4.556512274724387e-06, "loss": 0.16452684998512268, "step": 1752 }, { "epoch": 0.4550882658359294, "grad_norm": 2.305474281311035, "learning_rate": 4.555948178560635e-06, "loss": 0.13216876983642578, "step": 1753 }, { "epoch": 0.4553478712357217, "grad_norm": 0.5669839382171631, "learning_rate": 4.555383763146596e-06, "loss": 0.13988257944583893, "step": 1754 }, { "epoch": 0.45560747663551404, "grad_norm": 1.1039726734161377, "learning_rate": 4.554819028582044e-06, "loss": 0.14305803179740906, "step": 1755 }, { "epoch": 0.45586708203530635, "grad_norm": 1.1674286127090454, "learning_rate": 4.5542539749668134e-06, "loss": 0.13311076164245605, "step": 1756 }, { "epoch": 0.45612668743509865, "grad_norm": 1.961518406867981, "learning_rate": 4.553688602400792e-06, "loss": 0.15270018577575684, "step": 1757 }, { "epoch": 0.45638629283489096, "grad_norm": 0.9383907914161682, "learning_rate": 4.5531229109839244e-06, "loss": 0.1680382788181305, "step": 1758 }, { "epoch": 0.45664589823468327, "grad_norm": 0.6887012720108032, "learning_rate": 4.552556900816212e-06, "loss": 0.12861208617687225, "step": 1759 }, { "epoch": 0.4569055036344756, "grad_norm": 0.6508644223213196, "learning_rate": 4.551990571997713e-06, "loss": 0.126919686794281, "step": 1760 }, { "epoch": 0.45716510903426794, "grad_norm": 0.9653517603874207, "learning_rate": 4.551423924628541e-06, "loss": 0.12417528033256531, "step": 1761 }, { "epoch": 0.45742471443406024, "grad_norm": 0.6722694039344788, "learning_rate": 4.550856958808867e-06, "loss": 0.1452588140964508, "step": 1762 }, { "epoch": 0.45768431983385255, "grad_norm": 1.0387054681777954, "learning_rate": 4.550289674638915e-06, "loss": 0.1629820466041565, "step": 1763 }, { "epoch": 0.45794392523364486, "grad_norm": 0.8792378306388855, "learning_rate": 4.549722072218971e-06, "loss": 0.14787133038043976, "step": 1764 }, { "epoch": 0.45820353063343716, "grad_norm": 3.0418195724487305, "learning_rate": 4.5491541516493725e-06, "loss": 0.15600457787513733, "step": 1765 }, { "epoch": 0.45846313603322947, "grad_norm": 0.9283748269081116, "learning_rate": 4.548585913030515e-06, "loss": 0.14193879067897797, "step": 1766 }, { "epoch": 0.45872274143302183, "grad_norm": 0.8629863858222961, "learning_rate": 4.548017356462851e-06, "loss": 0.1440736949443817, "step": 1767 }, { "epoch": 0.45898234683281414, "grad_norm": 1.662719488143921, "learning_rate": 4.547448482046887e-06, "loss": 0.18081134557724, "step": 1768 }, { "epoch": 0.45924195223260644, "grad_norm": 0.7584439516067505, "learning_rate": 4.546879289883187e-06, "loss": 0.1708371490240097, "step": 1769 }, { "epoch": 0.45950155763239875, "grad_norm": 1.791077733039856, "learning_rate": 4.5463097800723725e-06, "loss": 0.17961524426937103, "step": 1770 }, { "epoch": 0.45976116303219106, "grad_norm": 1.0120834112167358, "learning_rate": 4.545739952715118e-06, "loss": 0.1430525779724121, "step": 1771 }, { "epoch": 0.46002076843198336, "grad_norm": 0.7021933794021606, "learning_rate": 4.5451698079121575e-06, "loss": 0.14051823318004608, "step": 1772 }, { "epoch": 0.4602803738317757, "grad_norm": 0.7792288064956665, "learning_rate": 4.544599345764279e-06, "loss": 0.16688519716262817, "step": 1773 }, { "epoch": 0.46053997923156803, "grad_norm": 0.9359555840492249, "learning_rate": 4.544028566372326e-06, "loss": 0.16328038275241852, "step": 1774 }, { "epoch": 0.46079958463136034, "grad_norm": 0.7734531164169312, "learning_rate": 4.543457469837202e-06, "loss": 0.14698278903961182, "step": 1775 }, { "epoch": 0.46105919003115264, "grad_norm": 0.9816541075706482, "learning_rate": 4.542886056259861e-06, "loss": 0.14516761898994446, "step": 1776 }, { "epoch": 0.46131879543094495, "grad_norm": 1.1109129190444946, "learning_rate": 4.542314325741316e-06, "loss": 0.140770822763443, "step": 1777 }, { "epoch": 0.46157840083073726, "grad_norm": 0.6529984474182129, "learning_rate": 4.541742278382638e-06, "loss": 0.14318060874938965, "step": 1778 }, { "epoch": 0.4618380062305296, "grad_norm": 0.7549343705177307, "learning_rate": 4.541169914284951e-06, "loss": 0.13971948623657227, "step": 1779 }, { "epoch": 0.4620976116303219, "grad_norm": 1.4766440391540527, "learning_rate": 4.5405972335494345e-06, "loss": 0.153525710105896, "step": 1780 }, { "epoch": 0.46235721703011423, "grad_norm": 0.5131231546401978, "learning_rate": 4.540024236277327e-06, "loss": 0.15236878395080566, "step": 1781 }, { "epoch": 0.46261682242990654, "grad_norm": 1.238099455833435, "learning_rate": 4.539450922569921e-06, "loss": 0.16844072937965393, "step": 1782 }, { "epoch": 0.46287642782969884, "grad_norm": 1.0083136558532715, "learning_rate": 4.538877292528564e-06, "loss": 0.15740585327148438, "step": 1783 }, { "epoch": 0.46313603322949115, "grad_norm": 0.7503997087478638, "learning_rate": 4.538303346254662e-06, "loss": 0.11896078288555145, "step": 1784 }, { "epoch": 0.4633956386292835, "grad_norm": 0.6431311368942261, "learning_rate": 4.537729083849675e-06, "loss": 0.14462898671627045, "step": 1785 }, { "epoch": 0.4636552440290758, "grad_norm": 0.6126083731651306, "learning_rate": 4.53715450541512e-06, "loss": 0.13951259851455688, "step": 1786 }, { "epoch": 0.4639148494288681, "grad_norm": 0.856364905834198, "learning_rate": 4.5365796110525674e-06, "loss": 0.14114677906036377, "step": 1787 }, { "epoch": 0.46417445482866043, "grad_norm": 0.8778282999992371, "learning_rate": 4.5360044008636485e-06, "loss": 0.16514742374420166, "step": 1788 }, { "epoch": 0.46443406022845274, "grad_norm": 1.1182301044464111, "learning_rate": 4.535428874950045e-06, "loss": 0.16740542650222778, "step": 1789 }, { "epoch": 0.46469366562824505, "grad_norm": 0.7285099625587463, "learning_rate": 4.5348530334134975e-06, "loss": 0.14137813448905945, "step": 1790 }, { "epoch": 0.4649532710280374, "grad_norm": 1.8969851732254028, "learning_rate": 4.534276876355801e-06, "loss": 0.13462838530540466, "step": 1791 }, { "epoch": 0.4652128764278297, "grad_norm": 0.6323307752609253, "learning_rate": 4.533700403878806e-06, "loss": 0.18376129865646362, "step": 1792 }, { "epoch": 0.465472481827622, "grad_norm": 1.2957600355148315, "learning_rate": 4.533123616084423e-06, "loss": 0.1410655826330185, "step": 1793 }, { "epoch": 0.4657320872274143, "grad_norm": 2.06538987159729, "learning_rate": 4.5325465130746125e-06, "loss": 0.15869396924972534, "step": 1794 }, { "epoch": 0.46599169262720663, "grad_norm": 1.0907012224197388, "learning_rate": 4.531969094951393e-06, "loss": 0.17571726441383362, "step": 1795 }, { "epoch": 0.46625129802699894, "grad_norm": 1.055612325668335, "learning_rate": 4.5313913618168394e-06, "loss": 0.13771812617778778, "step": 1796 }, { "epoch": 0.4665109034267913, "grad_norm": 0.8030140995979309, "learning_rate": 4.530813313773081e-06, "loss": 0.13552309572696686, "step": 1797 }, { "epoch": 0.4667705088265836, "grad_norm": 0.5806055665016174, "learning_rate": 4.530234950922304e-06, "loss": 0.13031616806983948, "step": 1798 }, { "epoch": 0.4670301142263759, "grad_norm": 2.428091049194336, "learning_rate": 4.52965627336675e-06, "loss": 0.1635275036096573, "step": 1799 }, { "epoch": 0.4672897196261682, "grad_norm": 0.5711771249771118, "learning_rate": 4.529077281208716e-06, "loss": 0.13316284120082855, "step": 1800 }, { "epoch": 0.4675493250259605, "grad_norm": 1.2326396703720093, "learning_rate": 4.528497974550553e-06, "loss": 0.13943856954574585, "step": 1801 }, { "epoch": 0.46780893042575283, "grad_norm": 1.880886197090149, "learning_rate": 4.527918353494672e-06, "loss": 0.12932872772216797, "step": 1802 }, { "epoch": 0.4680685358255452, "grad_norm": 0.4558084011077881, "learning_rate": 4.527338418143533e-06, "loss": 0.15396885573863983, "step": 1803 }, { "epoch": 0.4683281412253375, "grad_norm": 0.6643089056015015, "learning_rate": 4.526758168599658e-06, "loss": 0.13467130064964294, "step": 1804 }, { "epoch": 0.4685877466251298, "grad_norm": 1.487993597984314, "learning_rate": 4.526177604965622e-06, "loss": 0.15787458419799805, "step": 1805 }, { "epoch": 0.4688473520249221, "grad_norm": 0.8805105686187744, "learning_rate": 4.525596727344054e-06, "loss": 0.13032078742980957, "step": 1806 }, { "epoch": 0.4691069574247144, "grad_norm": 2.3862063884735107, "learning_rate": 4.52501553583764e-06, "loss": 0.1486000418663025, "step": 1807 }, { "epoch": 0.46936656282450673, "grad_norm": 1.6497834920883179, "learning_rate": 4.524434030549122e-06, "loss": 0.13760121166706085, "step": 1808 }, { "epoch": 0.4696261682242991, "grad_norm": 3.8456029891967773, "learning_rate": 4.523852211581297e-06, "loss": 0.1830214411020279, "step": 1809 }, { "epoch": 0.4698857736240914, "grad_norm": 0.451435387134552, "learning_rate": 4.523270079037016e-06, "loss": 0.12344275414943695, "step": 1810 }, { "epoch": 0.4701453790238837, "grad_norm": 1.0491763353347778, "learning_rate": 4.522687633019189e-06, "loss": 0.12661457061767578, "step": 1811 }, { "epoch": 0.470404984423676, "grad_norm": 2.5516164302825928, "learning_rate": 4.522104873630776e-06, "loss": 0.14656192064285278, "step": 1812 }, { "epoch": 0.4706645898234683, "grad_norm": 0.9058260321617126, "learning_rate": 4.521521800974798e-06, "loss": 0.18448421359062195, "step": 1813 }, { "epoch": 0.4709241952232606, "grad_norm": 1.3172669410705566, "learning_rate": 4.520938415154328e-06, "loss": 0.17000871896743774, "step": 1814 }, { "epoch": 0.471183800623053, "grad_norm": 0.5087918043136597, "learning_rate": 4.520354716272495e-06, "loss": 0.1260063350200653, "step": 1815 }, { "epoch": 0.4714434060228453, "grad_norm": 1.5031405687332153, "learning_rate": 4.519770704432484e-06, "loss": 0.159598708152771, "step": 1816 }, { "epoch": 0.4717030114226376, "grad_norm": 1.5998045206069946, "learning_rate": 4.519186379737535e-06, "loss": 0.16559620201587677, "step": 1817 }, { "epoch": 0.4719626168224299, "grad_norm": 2.532949924468994, "learning_rate": 4.518601742290943e-06, "loss": 0.15985286235809326, "step": 1818 }, { "epoch": 0.4722222222222222, "grad_norm": 1.2155563831329346, "learning_rate": 4.518016792196059e-06, "loss": 0.1689949929714203, "step": 1819 }, { "epoch": 0.4724818276220145, "grad_norm": 0.7085964679718018, "learning_rate": 4.5174315295562885e-06, "loss": 0.1684882640838623, "step": 1820 }, { "epoch": 0.4727414330218069, "grad_norm": 0.3491671085357666, "learning_rate": 4.516845954475091e-06, "loss": 0.13522422313690186, "step": 1821 }, { "epoch": 0.4730010384215992, "grad_norm": 0.7636972665786743, "learning_rate": 4.516260067055985e-06, "loss": 0.14071807265281677, "step": 1822 }, { "epoch": 0.4732606438213915, "grad_norm": 0.7609314322471619, "learning_rate": 4.515673867402541e-06, "loss": 0.15042875707149506, "step": 1823 }, { "epoch": 0.4735202492211838, "grad_norm": 1.6939172744750977, "learning_rate": 4.515087355618387e-06, "loss": 0.18125471472740173, "step": 1824 }, { "epoch": 0.4737798546209761, "grad_norm": 0.4424884021282196, "learning_rate": 4.514500531807201e-06, "loss": 0.12772944569587708, "step": 1825 }, { "epoch": 0.4740394600207684, "grad_norm": 0.9541527628898621, "learning_rate": 4.513913396072725e-06, "loss": 0.15524336695671082, "step": 1826 }, { "epoch": 0.4742990654205608, "grad_norm": 0.8754070997238159, "learning_rate": 4.513325948518748e-06, "loss": 0.15154848992824554, "step": 1827 }, { "epoch": 0.4745586708203531, "grad_norm": 1.2619742155075073, "learning_rate": 4.512738189249119e-06, "loss": 0.1380329728126526, "step": 1828 }, { "epoch": 0.4748182762201454, "grad_norm": 0.8172751665115356, "learning_rate": 4.512150118367739e-06, "loss": 0.17034882307052612, "step": 1829 }, { "epoch": 0.4750778816199377, "grad_norm": 1.7121918201446533, "learning_rate": 4.511561735978564e-06, "loss": 0.1568497121334076, "step": 1830 }, { "epoch": 0.47533748701973, "grad_norm": 0.5144473314285278, "learning_rate": 4.510973042185611e-06, "loss": 0.13593259453773499, "step": 1831 }, { "epoch": 0.4755970924195223, "grad_norm": 0.7231699824333191, "learning_rate": 4.5103840370929446e-06, "loss": 0.14711128175258636, "step": 1832 }, { "epoch": 0.47585669781931467, "grad_norm": 0.6376970410346985, "learning_rate": 4.509794720804688e-06, "loss": 0.14443878829479218, "step": 1833 }, { "epoch": 0.476116303219107, "grad_norm": 0.7274124622344971, "learning_rate": 4.509205093425018e-06, "loss": 0.12288062274456024, "step": 1834 }, { "epoch": 0.4763759086188993, "grad_norm": 0.3695403039455414, "learning_rate": 4.508615155058168e-06, "loss": 0.1406925767660141, "step": 1835 }, { "epoch": 0.4766355140186916, "grad_norm": 0.7025565505027771, "learning_rate": 4.508024905808426e-06, "loss": 0.14899587631225586, "step": 1836 }, { "epoch": 0.4768951194184839, "grad_norm": 0.6960121989250183, "learning_rate": 4.507434345780133e-06, "loss": 0.13807374238967896, "step": 1837 }, { "epoch": 0.4771547248182762, "grad_norm": 1.8511677980422974, "learning_rate": 4.506843475077689e-06, "loss": 0.17004388570785522, "step": 1838 }, { "epoch": 0.47741433021806856, "grad_norm": 0.7942731380462646, "learning_rate": 4.5062522938055444e-06, "loss": 0.16932719945907593, "step": 1839 }, { "epoch": 0.47767393561786087, "grad_norm": 0.6321853995323181, "learning_rate": 4.505660802068207e-06, "loss": 0.1430903822183609, "step": 1840 }, { "epoch": 0.4779335410176532, "grad_norm": 0.7925999760627747, "learning_rate": 4.50506899997024e-06, "loss": 0.15346446633338928, "step": 1841 }, { "epoch": 0.4781931464174455, "grad_norm": 0.9765720367431641, "learning_rate": 4.504476887616258e-06, "loss": 0.15910837054252625, "step": 1842 }, { "epoch": 0.4784527518172378, "grad_norm": 1.3728338479995728, "learning_rate": 4.503884465110935e-06, "loss": 0.16755008697509766, "step": 1843 }, { "epoch": 0.4787123572170301, "grad_norm": 0.8927087783813477, "learning_rate": 4.503291732558998e-06, "loss": 0.16394630074501038, "step": 1844 }, { "epoch": 0.47897196261682246, "grad_norm": 1.8155956268310547, "learning_rate": 4.502698690065227e-06, "loss": 0.15145204961299896, "step": 1845 }, { "epoch": 0.47923156801661476, "grad_norm": 0.6230736374855042, "learning_rate": 4.5021053377344596e-06, "loss": 0.1355191171169281, "step": 1846 }, { "epoch": 0.47949117341640707, "grad_norm": 1.6112709045410156, "learning_rate": 4.501511675671586e-06, "loss": 0.13345620036125183, "step": 1847 }, { "epoch": 0.4797507788161994, "grad_norm": 1.4537910223007202, "learning_rate": 4.500917703981553e-06, "loss": 0.19531098008155823, "step": 1848 }, { "epoch": 0.4800103842159917, "grad_norm": 0.7562597990036011, "learning_rate": 4.5003234227693596e-06, "loss": 0.1542602777481079, "step": 1849 }, { "epoch": 0.480269989615784, "grad_norm": 1.476495385169983, "learning_rate": 4.499728832140063e-06, "loss": 0.16326698660850525, "step": 1850 }, { "epoch": 0.48052959501557635, "grad_norm": 1.4570616483688354, "learning_rate": 4.499133932198772e-06, "loss": 0.15966126322746277, "step": 1851 }, { "epoch": 0.48078920041536866, "grad_norm": 1.2011737823486328, "learning_rate": 4.498538723050651e-06, "loss": 0.17223399877548218, "step": 1852 }, { "epoch": 0.48104880581516096, "grad_norm": 1.3280278444290161, "learning_rate": 4.4979432048009215e-06, "loss": 0.13241079449653625, "step": 1853 }, { "epoch": 0.48130841121495327, "grad_norm": 2.390017509460449, "learning_rate": 4.497347377554855e-06, "loss": 0.14267411828041077, "step": 1854 }, { "epoch": 0.4815680166147456, "grad_norm": 1.9973353147506714, "learning_rate": 4.4967512414177816e-06, "loss": 0.12766532599925995, "step": 1855 }, { "epoch": 0.4818276220145379, "grad_norm": 0.5750787854194641, "learning_rate": 4.496154796495083e-06, "loss": 0.13434746861457825, "step": 1856 }, { "epoch": 0.48208722741433024, "grad_norm": 1.5843507051467896, "learning_rate": 4.495558042892199e-06, "loss": 0.1473964899778366, "step": 1857 }, { "epoch": 0.48234683281412255, "grad_norm": 1.0021826028823853, "learning_rate": 4.49496098071462e-06, "loss": 0.13776561617851257, "step": 1858 }, { "epoch": 0.48260643821391486, "grad_norm": 0.7466667294502258, "learning_rate": 4.494363610067895e-06, "loss": 0.15860453248023987, "step": 1859 }, { "epoch": 0.48286604361370716, "grad_norm": 1.6375030279159546, "learning_rate": 4.493765931057624e-06, "loss": 0.12863662838935852, "step": 1860 }, { "epoch": 0.48312564901349947, "grad_norm": 2.155425786972046, "learning_rate": 4.493167943789463e-06, "loss": 0.17827776074409485, "step": 1861 }, { "epoch": 0.4833852544132918, "grad_norm": 0.686149001121521, "learning_rate": 4.492569648369124e-06, "loss": 0.1336689591407776, "step": 1862 }, { "epoch": 0.48364485981308414, "grad_norm": 1.85280179977417, "learning_rate": 4.4919710449023695e-06, "loss": 0.15019099414348602, "step": 1863 }, { "epoch": 0.48390446521287644, "grad_norm": 3.765214204788208, "learning_rate": 4.491372133495021e-06, "loss": 0.18736600875854492, "step": 1864 }, { "epoch": 0.48416407061266875, "grad_norm": 2.216399908065796, "learning_rate": 4.49077291425295e-06, "loss": 0.17291170358657837, "step": 1865 }, { "epoch": 0.48442367601246106, "grad_norm": 1.2102274894714355, "learning_rate": 4.490173387282087e-06, "loss": 0.15585842728614807, "step": 1866 }, { "epoch": 0.48468328141225336, "grad_norm": 2.5701284408569336, "learning_rate": 4.489573552688414e-06, "loss": 0.14001020789146423, "step": 1867 }, { "epoch": 0.48494288681204567, "grad_norm": 1.5948126316070557, "learning_rate": 4.488973410577968e-06, "loss": 0.14200568199157715, "step": 1868 }, { "epoch": 0.48520249221183803, "grad_norm": 1.5482375621795654, "learning_rate": 4.48837296105684e-06, "loss": 0.13441161811351776, "step": 1869 }, { "epoch": 0.48546209761163034, "grad_norm": 0.7012434005737305, "learning_rate": 4.4877722042311765e-06, "loss": 0.16493940353393555, "step": 1870 }, { "epoch": 0.48572170301142265, "grad_norm": 1.0791317224502563, "learning_rate": 4.487171140207176e-06, "loss": 0.15689431130886078, "step": 1871 }, { "epoch": 0.48598130841121495, "grad_norm": 0.9806047677993774, "learning_rate": 4.4865697690910945e-06, "loss": 0.16137878596782684, "step": 1872 }, { "epoch": 0.48624091381100726, "grad_norm": 1.4050166606903076, "learning_rate": 4.48596809098924e-06, "loss": 0.15736524760723114, "step": 1873 }, { "epoch": 0.48650051921079956, "grad_norm": 1.0695523023605347, "learning_rate": 4.485366106007975e-06, "loss": 0.14396396279335022, "step": 1874 }, { "epoch": 0.4867601246105919, "grad_norm": 2.233203172683716, "learning_rate": 4.484763814253717e-06, "loss": 0.17239779233932495, "step": 1875 }, { "epoch": 0.48701973001038423, "grad_norm": 0.9763746857643127, "learning_rate": 4.484161215832938e-06, "loss": 0.13036954402923584, "step": 1876 }, { "epoch": 0.48727933541017654, "grad_norm": 1.7484551668167114, "learning_rate": 4.483558310852163e-06, "loss": 0.16732537746429443, "step": 1877 }, { "epoch": 0.48753894080996885, "grad_norm": 2.9814860820770264, "learning_rate": 4.482955099417971e-06, "loss": 0.13538965582847595, "step": 1878 }, { "epoch": 0.48779854620976115, "grad_norm": 1.7053593397140503, "learning_rate": 4.482351581636998e-06, "loss": 0.128360778093338, "step": 1879 }, { "epoch": 0.48805815160955346, "grad_norm": 0.7358659505844116, "learning_rate": 4.48174775761593e-06, "loss": 0.15166111290454865, "step": 1880 }, { "epoch": 0.4883177570093458, "grad_norm": 1.7273131608963013, "learning_rate": 4.48114362746151e-06, "loss": 0.15282674133777618, "step": 1881 }, { "epoch": 0.4885773624091381, "grad_norm": 2.025639772415161, "learning_rate": 4.4805391912805366e-06, "loss": 0.16055354475975037, "step": 1882 }, { "epoch": 0.48883696780893043, "grad_norm": 1.566045880317688, "learning_rate": 4.4799344491798564e-06, "loss": 0.13497477769851685, "step": 1883 }, { "epoch": 0.48909657320872274, "grad_norm": 1.1126652956008911, "learning_rate": 4.4793294012663765e-06, "loss": 0.1793002486228943, "step": 1884 }, { "epoch": 0.48935617860851505, "grad_norm": 2.3207476139068604, "learning_rate": 4.478724047647054e-06, "loss": 0.16514305770397186, "step": 1885 }, { "epoch": 0.48961578400830735, "grad_norm": 1.0066355466842651, "learning_rate": 4.478118388428904e-06, "loss": 0.15256404876708984, "step": 1886 }, { "epoch": 0.4898753894080997, "grad_norm": 0.8768764734268188, "learning_rate": 4.47751242371899e-06, "loss": 0.13333813846111298, "step": 1887 }, { "epoch": 0.490134994807892, "grad_norm": 1.9877166748046875, "learning_rate": 4.476906153624435e-06, "loss": 0.12206098437309265, "step": 1888 }, { "epoch": 0.49039460020768433, "grad_norm": 3.3972103595733643, "learning_rate": 4.476299578252414e-06, "loss": 0.17048822343349457, "step": 1889 }, { "epoch": 0.49065420560747663, "grad_norm": 1.104264497756958, "learning_rate": 4.475692697710153e-06, "loss": 0.1584184765815735, "step": 1890 }, { "epoch": 0.49091381100726894, "grad_norm": 0.971032977104187, "learning_rate": 4.475085512104937e-06, "loss": 0.13662943243980408, "step": 1891 }, { "epoch": 0.49117341640706125, "grad_norm": 1.6025724411010742, "learning_rate": 4.474478021544102e-06, "loss": 0.1616520881652832, "step": 1892 }, { "epoch": 0.4914330218068536, "grad_norm": 1.3861117362976074, "learning_rate": 4.473870226135036e-06, "loss": 0.15754538774490356, "step": 1893 }, { "epoch": 0.4916926272066459, "grad_norm": 2.47055983543396, "learning_rate": 4.473262125985188e-06, "loss": 0.17058536410331726, "step": 1894 }, { "epoch": 0.4919522326064382, "grad_norm": 1.931892991065979, "learning_rate": 4.472653721202052e-06, "loss": 0.1456853449344635, "step": 1895 }, { "epoch": 0.49221183800623053, "grad_norm": 1.8183095455169678, "learning_rate": 4.472045011893182e-06, "loss": 0.1654357612133026, "step": 1896 }, { "epoch": 0.49247144340602284, "grad_norm": 1.553460955619812, "learning_rate": 4.471435998166185e-06, "loss": 0.14012910425662994, "step": 1897 }, { "epoch": 0.49273104880581514, "grad_norm": 0.29984986782073975, "learning_rate": 4.470826680128717e-06, "loss": 0.13853995501995087, "step": 1898 }, { "epoch": 0.4929906542056075, "grad_norm": 1.1730802059173584, "learning_rate": 4.470217057888495e-06, "loss": 0.14229808747768402, "step": 1899 }, { "epoch": 0.4932502596053998, "grad_norm": 1.6725828647613525, "learning_rate": 4.469607131553284e-06, "loss": 0.14171727001667023, "step": 1900 }, { "epoch": 0.4935098650051921, "grad_norm": 1.5909446477890015, "learning_rate": 4.468996901230907e-06, "loss": 0.14291983842849731, "step": 1901 }, { "epoch": 0.4937694704049844, "grad_norm": 1.9741160869598389, "learning_rate": 4.468386367029236e-06, "loss": 0.18397459387779236, "step": 1902 }, { "epoch": 0.49402907580477673, "grad_norm": 2.064985513687134, "learning_rate": 4.467775529056203e-06, "loss": 0.1510852575302124, "step": 1903 }, { "epoch": 0.49428868120456904, "grad_norm": 1.395674467086792, "learning_rate": 4.467164387419787e-06, "loss": 0.1421617865562439, "step": 1904 }, { "epoch": 0.4945482866043614, "grad_norm": 0.7952483296394348, "learning_rate": 4.466552942228026e-06, "loss": 0.12781879305839539, "step": 1905 }, { "epoch": 0.4948078920041537, "grad_norm": 1.8834797143936157, "learning_rate": 4.465941193589008e-06, "loss": 0.12063492834568024, "step": 1906 }, { "epoch": 0.495067497403946, "grad_norm": 1.4881246089935303, "learning_rate": 4.465329141610877e-06, "loss": 0.1650175154209137, "step": 1907 }, { "epoch": 0.4953271028037383, "grad_norm": 3.45229172706604, "learning_rate": 4.46471678640183e-06, "loss": 0.15470880270004272, "step": 1908 }, { "epoch": 0.4955867082035306, "grad_norm": 3.0872304439544678, "learning_rate": 4.4641041280701164e-06, "loss": 0.1486656814813614, "step": 1909 }, { "epoch": 0.49584631360332293, "grad_norm": 0.8903875946998596, "learning_rate": 4.463491166724041e-06, "loss": 0.140774205327034, "step": 1910 }, { "epoch": 0.4961059190031153, "grad_norm": 2.73543643951416, "learning_rate": 4.462877902471961e-06, "loss": 0.14488300681114197, "step": 1911 }, { "epoch": 0.4963655244029076, "grad_norm": 2.732766628265381, "learning_rate": 4.462264335422288e-06, "loss": 0.13180387020111084, "step": 1912 }, { "epoch": 0.4966251298026999, "grad_norm": 3.370340347290039, "learning_rate": 4.4616504656834855e-06, "loss": 0.14292950928211212, "step": 1913 }, { "epoch": 0.4968847352024922, "grad_norm": 1.7153234481811523, "learning_rate": 4.461036293364073e-06, "loss": 0.13590219616889954, "step": 1914 }, { "epoch": 0.4971443406022845, "grad_norm": 0.8361161351203918, "learning_rate": 4.46042181857262e-06, "loss": 0.1238330602645874, "step": 1915 }, { "epoch": 0.4974039460020768, "grad_norm": 1.0398122072219849, "learning_rate": 4.459807041417755e-06, "loss": 0.1556941270828247, "step": 1916 }, { "epoch": 0.4976635514018692, "grad_norm": 1.612070918083191, "learning_rate": 4.4591919620081526e-06, "loss": 0.15103581547737122, "step": 1917 }, { "epoch": 0.4979231568016615, "grad_norm": 1.9871364831924438, "learning_rate": 4.458576580452547e-06, "loss": 0.15081268548965454, "step": 1918 }, { "epoch": 0.4981827622014538, "grad_norm": 1.9882599115371704, "learning_rate": 4.457960896859724e-06, "loss": 0.13769665360450745, "step": 1919 }, { "epoch": 0.4984423676012461, "grad_norm": 1.5576859712600708, "learning_rate": 4.4573449113385214e-06, "loss": 0.1560300886631012, "step": 1920 }, { "epoch": 0.4987019730010384, "grad_norm": 1.3468265533447266, "learning_rate": 4.456728623997831e-06, "loss": 0.1626220941543579, "step": 1921 }, { "epoch": 0.4989615784008307, "grad_norm": 1.3737510442733765, "learning_rate": 4.456112034946599e-06, "loss": 0.1376369297504425, "step": 1922 }, { "epoch": 0.4992211838006231, "grad_norm": 1.5688668489456177, "learning_rate": 4.455495144293824e-06, "loss": 0.1643165946006775, "step": 1923 }, { "epoch": 0.4994807892004154, "grad_norm": 1.744381070137024, "learning_rate": 4.454877952148558e-06, "loss": 0.1580972671508789, "step": 1924 }, { "epoch": 0.4997403946002077, "grad_norm": 2.246840715408325, "learning_rate": 4.454260458619907e-06, "loss": 0.1635170876979828, "step": 1925 }, { "epoch": 0.5, "grad_norm": 0.5351076126098633, "learning_rate": 4.45364266381703e-06, "loss": 0.1284310221672058, "step": 1926 }, { "epoch": 0.5002596053997923, "grad_norm": 1.7704684734344482, "learning_rate": 4.453024567849136e-06, "loss": 0.15143883228302002, "step": 1927 }, { "epoch": 0.5005192107995846, "grad_norm": 1.2027407884597778, "learning_rate": 4.452406170825494e-06, "loss": 0.16405171155929565, "step": 1928 }, { "epoch": 0.5007788161993769, "grad_norm": 1.0062814950942993, "learning_rate": 4.451787472855421e-06, "loss": 0.12616774439811707, "step": 1929 }, { "epoch": 0.5010384215991692, "grad_norm": 1.1963666677474976, "learning_rate": 4.451168474048288e-06, "loss": 0.16658179461956024, "step": 1930 }, { "epoch": 0.5012980269989615, "grad_norm": 1.5002154111862183, "learning_rate": 4.450549174513521e-06, "loss": 0.18909952044487, "step": 1931 }, { "epoch": 0.5015576323987538, "grad_norm": 0.8425201177597046, "learning_rate": 4.449929574360597e-06, "loss": 0.19523072242736816, "step": 1932 }, { "epoch": 0.5018172377985463, "grad_norm": 1.1584380865097046, "learning_rate": 4.4493096736990476e-06, "loss": 0.1441173106431961, "step": 1933 }, { "epoch": 0.5020768431983386, "grad_norm": 0.5195335149765015, "learning_rate": 4.448689472638457e-06, "loss": 0.12883591651916504, "step": 1934 }, { "epoch": 0.5023364485981309, "grad_norm": 0.6367800831794739, "learning_rate": 4.448068971288462e-06, "loss": 0.1410442590713501, "step": 1935 }, { "epoch": 0.5025960539979232, "grad_norm": 0.5574820637702942, "learning_rate": 4.447448169758754e-06, "loss": 0.1258518248796463, "step": 1936 }, { "epoch": 0.5028556593977155, "grad_norm": 0.9766342639923096, "learning_rate": 4.446827068159076e-06, "loss": 0.13626278936862946, "step": 1937 }, { "epoch": 0.5031152647975078, "grad_norm": 1.3701618909835815, "learning_rate": 4.446205666599224e-06, "loss": 0.15544211864471436, "step": 1938 }, { "epoch": 0.5033748701973001, "grad_norm": 0.6208629608154297, "learning_rate": 4.445583965189049e-06, "loss": 0.15246328711509705, "step": 1939 }, { "epoch": 0.5036344755970924, "grad_norm": 0.45977848768234253, "learning_rate": 4.444961964038452e-06, "loss": 0.1257430613040924, "step": 1940 }, { "epoch": 0.5038940809968847, "grad_norm": 0.6709108948707581, "learning_rate": 4.444339663257389e-06, "loss": 0.16712048649787903, "step": 1941 }, { "epoch": 0.504153686396677, "grad_norm": 1.3577615022659302, "learning_rate": 4.443717062955869e-06, "loss": 0.16928143799304962, "step": 1942 }, { "epoch": 0.5044132917964693, "grad_norm": 0.5447478890419006, "learning_rate": 4.443094163243954e-06, "loss": 0.12569214403629303, "step": 1943 }, { "epoch": 0.5046728971962616, "grad_norm": 2.108279228210449, "learning_rate": 4.442470964231758e-06, "loss": 0.15994693338871002, "step": 1944 }, { "epoch": 0.504932502596054, "grad_norm": 1.2480028867721558, "learning_rate": 4.441847466029447e-06, "loss": 0.1265966147184372, "step": 1945 }, { "epoch": 0.5051921079958464, "grad_norm": 1.7484453916549683, "learning_rate": 4.441223668747241e-06, "loss": 0.13841480016708374, "step": 1946 }, { "epoch": 0.5054517133956387, "grad_norm": 0.6347739100456238, "learning_rate": 4.440599572495416e-06, "loss": 0.1257619708776474, "step": 1947 }, { "epoch": 0.505711318795431, "grad_norm": 1.0582869052886963, "learning_rate": 4.439975177384296e-06, "loss": 0.1284586787223816, "step": 1948 }, { "epoch": 0.5059709241952233, "grad_norm": 0.5248618125915527, "learning_rate": 4.439350483524259e-06, "loss": 0.12343094497919083, "step": 1949 }, { "epoch": 0.5062305295950156, "grad_norm": 1.1440343856811523, "learning_rate": 4.4387254910257386e-06, "loss": 0.11800433695316315, "step": 1950 }, { "epoch": 0.5064901349948079, "grad_norm": 1.1437067985534668, "learning_rate": 4.438100199999217e-06, "loss": 0.1631302833557129, "step": 1951 }, { "epoch": 0.5067497403946002, "grad_norm": 0.8650118112564087, "learning_rate": 4.437474610555234e-06, "loss": 0.1703287959098816, "step": 1952 }, { "epoch": 0.5070093457943925, "grad_norm": 0.9369866847991943, "learning_rate": 4.436848722804376e-06, "loss": 0.14968082308769226, "step": 1953 }, { "epoch": 0.5072689511941848, "grad_norm": 1.1868622303009033, "learning_rate": 4.436222536857289e-06, "loss": 0.1454724371433258, "step": 1954 }, { "epoch": 0.5075285565939771, "grad_norm": 0.8224453330039978, "learning_rate": 4.435596052824666e-06, "loss": 0.14663778245449066, "step": 1955 }, { "epoch": 0.5077881619937694, "grad_norm": 0.87957364320755, "learning_rate": 4.434969270817257e-06, "loss": 0.14330115914344788, "step": 1956 }, { "epoch": 0.5080477673935618, "grad_norm": 0.9910774230957031, "learning_rate": 4.4343421909458605e-06, "loss": 0.16997499763965607, "step": 1957 }, { "epoch": 0.5083073727933541, "grad_norm": 2.481688976287842, "learning_rate": 4.433714813321331e-06, "loss": 0.1930268555879593, "step": 1958 }, { "epoch": 0.5085669781931464, "grad_norm": 0.5221814513206482, "learning_rate": 4.433087138054575e-06, "loss": 0.11532025039196014, "step": 1959 }, { "epoch": 0.5088265835929388, "grad_norm": 1.1301251649856567, "learning_rate": 4.4324591652565505e-06, "loss": 0.1630367785692215, "step": 1960 }, { "epoch": 0.5090861889927311, "grad_norm": 1.3348597288131714, "learning_rate": 4.4318308950382685e-06, "loss": 0.16025668382644653, "step": 1961 }, { "epoch": 0.5093457943925234, "grad_norm": 0.5907612442970276, "learning_rate": 4.431202327510793e-06, "loss": 0.12896105647087097, "step": 1962 }, { "epoch": 0.5096053997923157, "grad_norm": 0.7231733798980713, "learning_rate": 4.43057346278524e-06, "loss": 0.1315690279006958, "step": 1963 }, { "epoch": 0.509865005192108, "grad_norm": 1.2962779998779297, "learning_rate": 4.42994430097278e-06, "loss": 0.1532338708639145, "step": 1964 }, { "epoch": 0.5101246105919003, "grad_norm": 0.678250789642334, "learning_rate": 4.429314842184632e-06, "loss": 0.11047132313251495, "step": 1965 }, { "epoch": 0.5103842159916926, "grad_norm": 1.1329309940338135, "learning_rate": 4.428685086532072e-06, "loss": 0.15701839327812195, "step": 1966 }, { "epoch": 0.5106438213914849, "grad_norm": 1.4838930368423462, "learning_rate": 4.428055034126426e-06, "loss": 0.14962346851825714, "step": 1967 }, { "epoch": 0.5109034267912772, "grad_norm": 1.2045648097991943, "learning_rate": 4.427424685079071e-06, "loss": 0.15970554947853088, "step": 1968 }, { "epoch": 0.5111630321910696, "grad_norm": 1.0895664691925049, "learning_rate": 4.42679403950144e-06, "loss": 0.1468680500984192, "step": 1969 }, { "epoch": 0.5114226375908619, "grad_norm": 1.7288503646850586, "learning_rate": 4.426163097505017e-06, "loss": 0.16378776729106903, "step": 1970 }, { "epoch": 0.5116822429906542, "grad_norm": 0.8310821652412415, "learning_rate": 4.425531859201337e-06, "loss": 0.1420811414718628, "step": 1971 }, { "epoch": 0.5119418483904465, "grad_norm": 1.205039143562317, "learning_rate": 4.424900324701989e-06, "loss": 0.12550552189350128, "step": 1972 }, { "epoch": 0.5122014537902388, "grad_norm": 2.047318696975708, "learning_rate": 4.424268494118613e-06, "loss": 0.14037427306175232, "step": 1973 }, { "epoch": 0.5124610591900312, "grad_norm": 1.3758331537246704, "learning_rate": 4.423636367562903e-06, "loss": 0.14943456649780273, "step": 1974 }, { "epoch": 0.5127206645898235, "grad_norm": 1.4054844379425049, "learning_rate": 4.423003945146605e-06, "loss": 0.13284888863563538, "step": 1975 }, { "epoch": 0.5129802699896158, "grad_norm": 0.5684958696365356, "learning_rate": 4.422371226981517e-06, "loss": 0.13464172184467316, "step": 1976 }, { "epoch": 0.5132398753894081, "grad_norm": 1.56656813621521, "learning_rate": 4.421738213179487e-06, "loss": 0.15312492847442627, "step": 1977 }, { "epoch": 0.5134994807892004, "grad_norm": 1.9046435356140137, "learning_rate": 4.421104903852419e-06, "loss": 0.14756083488464355, "step": 1978 }, { "epoch": 0.5137590861889927, "grad_norm": 1.7534153461456299, "learning_rate": 4.420471299112267e-06, "loss": 0.14740732312202454, "step": 1979 }, { "epoch": 0.514018691588785, "grad_norm": 1.686735987663269, "learning_rate": 4.419837399071038e-06, "loss": 0.12592029571533203, "step": 1980 }, { "epoch": 0.5142782969885774, "grad_norm": 0.7243223190307617, "learning_rate": 4.419203203840792e-06, "loss": 0.11711182445287704, "step": 1981 }, { "epoch": 0.5145379023883697, "grad_norm": 1.1278984546661377, "learning_rate": 4.418568713533638e-06, "loss": 0.12440955638885498, "step": 1982 }, { "epoch": 0.514797507788162, "grad_norm": 1.7632256746292114, "learning_rate": 4.4179339282617416e-06, "loss": 0.1653175950050354, "step": 1983 }, { "epoch": 0.5150571131879543, "grad_norm": 1.9851975440979004, "learning_rate": 4.417298848137317e-06, "loss": 0.12930700182914734, "step": 1984 }, { "epoch": 0.5153167185877466, "grad_norm": 1.4241069555282593, "learning_rate": 4.416663473272632e-06, "loss": 0.15501686930656433, "step": 1985 }, { "epoch": 0.5155763239875389, "grad_norm": 2.3555381298065186, "learning_rate": 4.416027803780006e-06, "loss": 0.15824246406555176, "step": 1986 }, { "epoch": 0.5158359293873312, "grad_norm": 0.4148210883140564, "learning_rate": 4.415391839771812e-06, "loss": 0.14549192786216736, "step": 1987 }, { "epoch": 0.5160955347871236, "grad_norm": 0.5204850435256958, "learning_rate": 4.414755581360473e-06, "loss": 0.1327250897884369, "step": 1988 }, { "epoch": 0.5163551401869159, "grad_norm": 0.704142153263092, "learning_rate": 4.414119028658465e-06, "loss": 0.13486286997795105, "step": 1989 }, { "epoch": 0.5166147455867082, "grad_norm": 0.5558380484580994, "learning_rate": 4.4134821817783155e-06, "loss": 0.14061857759952545, "step": 1990 }, { "epoch": 0.5168743509865005, "grad_norm": 1.2690600156784058, "learning_rate": 4.4128450408326065e-06, "loss": 0.1601475477218628, "step": 1991 }, { "epoch": 0.5171339563862928, "grad_norm": 1.1115272045135498, "learning_rate": 4.412207605933967e-06, "loss": 0.13885246217250824, "step": 1992 }, { "epoch": 0.5173935617860852, "grad_norm": 1.1833372116088867, "learning_rate": 4.411569877195082e-06, "loss": 0.16858027875423431, "step": 1993 }, { "epoch": 0.5176531671858775, "grad_norm": 0.841456413269043, "learning_rate": 4.410931854728689e-06, "loss": 0.1312878280878067, "step": 1994 }, { "epoch": 0.5179127725856698, "grad_norm": 1.5370588302612305, "learning_rate": 4.410293538647573e-06, "loss": 0.14734914898872375, "step": 1995 }, { "epoch": 0.5181723779854621, "grad_norm": 1.072848916053772, "learning_rate": 4.409654929064575e-06, "loss": 0.12482823431491852, "step": 1996 }, { "epoch": 0.5184319833852544, "grad_norm": 1.2979779243469238, "learning_rate": 4.4090160260925874e-06, "loss": 0.12506386637687683, "step": 1997 }, { "epoch": 0.5186915887850467, "grad_norm": 0.9825807809829712, "learning_rate": 4.408376829844553e-06, "loss": 0.14105042815208435, "step": 1998 }, { "epoch": 0.518951194184839, "grad_norm": 1.8966188430786133, "learning_rate": 4.407737340433466e-06, "loss": 0.16187556087970734, "step": 1999 }, { "epoch": 0.5192107995846313, "grad_norm": 1.1132196187973022, "learning_rate": 4.407097557972375e-06, "loss": 0.18024598062038422, "step": 2000 }, { "epoch": 0.5194704049844237, "grad_norm": 1.583038568496704, "learning_rate": 4.406457482574379e-06, "loss": 0.14221373200416565, "step": 2001 }, { "epoch": 0.519730010384216, "grad_norm": 1.8978338241577148, "learning_rate": 4.405817114352627e-06, "loss": 0.1833895444869995, "step": 2002 }, { "epoch": 0.5199896157840083, "grad_norm": 1.0877127647399902, "learning_rate": 4.405176453420322e-06, "loss": 0.17558136582374573, "step": 2003 }, { "epoch": 0.5202492211838006, "grad_norm": 1.4778594970703125, "learning_rate": 4.4045354998907196e-06, "loss": 0.14711278676986694, "step": 2004 }, { "epoch": 0.520508826583593, "grad_norm": 0.45350611209869385, "learning_rate": 4.4038942538771235e-06, "loss": 0.10937255620956421, "step": 2005 }, { "epoch": 0.5207684319833853, "grad_norm": 1.0021461248397827, "learning_rate": 4.403252715492893e-06, "loss": 0.12839581072330475, "step": 2006 }, { "epoch": 0.5210280373831776, "grad_norm": 1.1630841493606567, "learning_rate": 4.402610884851437e-06, "loss": 0.11798729002475739, "step": 2007 }, { "epoch": 0.5212876427829699, "grad_norm": 0.929119348526001, "learning_rate": 4.401968762066216e-06, "loss": 0.12557044625282288, "step": 2008 }, { "epoch": 0.5215472481827622, "grad_norm": 0.5402151942253113, "learning_rate": 4.4013263472507434e-06, "loss": 0.12035427987575531, "step": 2009 }, { "epoch": 0.5218068535825545, "grad_norm": 0.44382819533348083, "learning_rate": 4.400683640518583e-06, "loss": 0.13142666220664978, "step": 2010 }, { "epoch": 0.5220664589823468, "grad_norm": 0.8593860268592834, "learning_rate": 4.40004064198335e-06, "loss": 0.12147949635982513, "step": 2011 }, { "epoch": 0.5223260643821391, "grad_norm": 0.389128714799881, "learning_rate": 4.399397351758714e-06, "loss": 0.1517072468996048, "step": 2012 }, { "epoch": 0.5225856697819314, "grad_norm": 1.4527332782745361, "learning_rate": 4.39875376995839e-06, "loss": 0.1709626466035843, "step": 2013 }, { "epoch": 0.5228452751817237, "grad_norm": 1.7025507688522339, "learning_rate": 4.3981098966961524e-06, "loss": 0.13751161098480225, "step": 2014 }, { "epoch": 0.523104880581516, "grad_norm": 0.899504542350769, "learning_rate": 4.397465732085822e-06, "loss": 0.1580716371536255, "step": 2015 }, { "epoch": 0.5233644859813084, "grad_norm": 1.1115144491195679, "learning_rate": 4.396821276241272e-06, "loss": 0.15751394629478455, "step": 2016 }, { "epoch": 0.5236240913811008, "grad_norm": 0.7342708110809326, "learning_rate": 4.396176529276426e-06, "loss": 0.1524086594581604, "step": 2017 }, { "epoch": 0.5238836967808931, "grad_norm": 1.6716454029083252, "learning_rate": 4.3955314913052636e-06, "loss": 0.14066506922245026, "step": 2018 }, { "epoch": 0.5241433021806854, "grad_norm": 1.4698081016540527, "learning_rate": 4.394886162441811e-06, "loss": 0.13704144954681396, "step": 2019 }, { "epoch": 0.5244029075804777, "grad_norm": 1.1873071193695068, "learning_rate": 4.394240542800148e-06, "loss": 0.1538647711277008, "step": 2020 }, { "epoch": 0.52466251298027, "grad_norm": 0.7092640399932861, "learning_rate": 4.393594632494404e-06, "loss": 0.13199767470359802, "step": 2021 }, { "epoch": 0.5249221183800623, "grad_norm": 1.1736514568328857, "learning_rate": 4.3929484316387636e-06, "loss": 0.1412409543991089, "step": 2022 }, { "epoch": 0.5251817237798546, "grad_norm": 0.7807546854019165, "learning_rate": 4.392301940347458e-06, "loss": 0.13921523094177246, "step": 2023 }, { "epoch": 0.5254413291796469, "grad_norm": 2.3608434200286865, "learning_rate": 4.391655158734774e-06, "loss": 0.17245197296142578, "step": 2024 }, { "epoch": 0.5257009345794392, "grad_norm": 1.3086106777191162, "learning_rate": 4.391008086915046e-06, "loss": 0.18113285303115845, "step": 2025 }, { "epoch": 0.5259605399792315, "grad_norm": 0.5857496857643127, "learning_rate": 4.390360725002663e-06, "loss": 0.12653139233589172, "step": 2026 }, { "epoch": 0.5262201453790238, "grad_norm": 0.8584783673286438, "learning_rate": 4.389713073112063e-06, "loss": 0.1310880482196808, "step": 2027 }, { "epoch": 0.5264797507788161, "grad_norm": 0.990122377872467, "learning_rate": 4.389065131357736e-06, "loss": 0.1583786904811859, "step": 2028 }, { "epoch": 0.5267393561786086, "grad_norm": 0.8961661458015442, "learning_rate": 4.388416899854224e-06, "loss": 0.13574925065040588, "step": 2029 }, { "epoch": 0.5269989615784009, "grad_norm": 1.1528769731521606, "learning_rate": 4.387768378716117e-06, "loss": 0.15001118183135986, "step": 2030 }, { "epoch": 0.5272585669781932, "grad_norm": 0.5987032055854797, "learning_rate": 4.387119568058063e-06, "loss": 0.1316699981689453, "step": 2031 }, { "epoch": 0.5275181723779855, "grad_norm": 1.6862382888793945, "learning_rate": 4.386470467994753e-06, "loss": 0.16867920756340027, "step": 2032 }, { "epoch": 0.5277777777777778, "grad_norm": 1.575526475906372, "learning_rate": 4.385821078640935e-06, "loss": 0.16640791296958923, "step": 2033 }, { "epoch": 0.5280373831775701, "grad_norm": 1.3080592155456543, "learning_rate": 4.385171400111405e-06, "loss": 0.1283121556043625, "step": 2034 }, { "epoch": 0.5282969885773624, "grad_norm": 0.5440056920051575, "learning_rate": 4.3845214325210124e-06, "loss": 0.1290937066078186, "step": 2035 }, { "epoch": 0.5285565939771547, "grad_norm": 0.907330334186554, "learning_rate": 4.383871175984656e-06, "loss": 0.12591315805912018, "step": 2036 }, { "epoch": 0.528816199376947, "grad_norm": 0.8768210411071777, "learning_rate": 4.383220630617287e-06, "loss": 0.14065434038639069, "step": 2037 }, { "epoch": 0.5290758047767393, "grad_norm": 1.160094141960144, "learning_rate": 4.3825697965339055e-06, "loss": 0.14245101809501648, "step": 2038 }, { "epoch": 0.5293354101765316, "grad_norm": 1.4292898178100586, "learning_rate": 4.3819186738495645e-06, "loss": 0.13666588068008423, "step": 2039 }, { "epoch": 0.5295950155763239, "grad_norm": 0.5576953291893005, "learning_rate": 4.381267262679369e-06, "loss": 0.13198280334472656, "step": 2040 }, { "epoch": 0.5298546209761164, "grad_norm": 0.42416810989379883, "learning_rate": 4.380615563138472e-06, "loss": 0.14966051280498505, "step": 2041 }, { "epoch": 0.5301142263759087, "grad_norm": 0.9963811039924622, "learning_rate": 4.3799635753420794e-06, "loss": 0.13857918977737427, "step": 2042 }, { "epoch": 0.530373831775701, "grad_norm": 1.1891634464263916, "learning_rate": 4.379311299405449e-06, "loss": 0.13946415483951569, "step": 2043 }, { "epoch": 0.5306334371754933, "grad_norm": 1.410388469696045, "learning_rate": 4.378658735443886e-06, "loss": 0.1405634582042694, "step": 2044 }, { "epoch": 0.5308930425752856, "grad_norm": 1.3995375633239746, "learning_rate": 4.37800588357275e-06, "loss": 0.13755875825881958, "step": 2045 }, { "epoch": 0.5311526479750779, "grad_norm": 0.8999095559120178, "learning_rate": 4.37735274390745e-06, "loss": 0.153894305229187, "step": 2046 }, { "epoch": 0.5314122533748702, "grad_norm": 2.4560129642486572, "learning_rate": 4.376699316563447e-06, "loss": 0.15859657526016235, "step": 2047 }, { "epoch": 0.5316718587746625, "grad_norm": 0.5375726222991943, "learning_rate": 4.3760456016562505e-06, "loss": 0.14220589399337769, "step": 2048 }, { "epoch": 0.5319314641744548, "grad_norm": 0.5684874057769775, "learning_rate": 4.375391599301424e-06, "loss": 0.11924712359905243, "step": 2049 }, { "epoch": 0.5321910695742471, "grad_norm": 0.6462218165397644, "learning_rate": 4.374737309614579e-06, "loss": 0.13588321208953857, "step": 2050 }, { "epoch": 0.5324506749740394, "grad_norm": 0.6736932396888733, "learning_rate": 4.374082732711379e-06, "loss": 0.13803759217262268, "step": 2051 }, { "epoch": 0.5327102803738317, "grad_norm": 0.8363766670227051, "learning_rate": 4.373427868707539e-06, "loss": 0.11727050691843033, "step": 2052 }, { "epoch": 0.5329698857736241, "grad_norm": 1.3230594396591187, "learning_rate": 4.3727727177188226e-06, "loss": 0.13195054233074188, "step": 2053 }, { "epoch": 0.5332294911734164, "grad_norm": 1.5755587816238403, "learning_rate": 4.372117279861047e-06, "loss": 0.12000700831413269, "step": 2054 }, { "epoch": 0.5334890965732088, "grad_norm": 1.1850287914276123, "learning_rate": 4.371461555250077e-06, "loss": 0.14321866631507874, "step": 2055 }, { "epoch": 0.5337487019730011, "grad_norm": 0.8607345223426819, "learning_rate": 4.370805544001831e-06, "loss": 0.14667044579982758, "step": 2056 }, { "epoch": 0.5340083073727934, "grad_norm": 0.616509735584259, "learning_rate": 4.370149246232277e-06, "loss": 0.124856136739254, "step": 2057 }, { "epoch": 0.5342679127725857, "grad_norm": 1.2115514278411865, "learning_rate": 4.3694926620574315e-06, "loss": 0.133662611246109, "step": 2058 }, { "epoch": 0.534527518172378, "grad_norm": 2.616938829421997, "learning_rate": 4.3688357915933656e-06, "loss": 0.17491166293621063, "step": 2059 }, { "epoch": 0.5347871235721703, "grad_norm": 1.6099979877471924, "learning_rate": 4.368178634956199e-06, "loss": 0.17824164032936096, "step": 2060 }, { "epoch": 0.5350467289719626, "grad_norm": 0.9003065228462219, "learning_rate": 4.3675211922621e-06, "loss": 0.1639494001865387, "step": 2061 }, { "epoch": 0.5353063343717549, "grad_norm": 2.866642713546753, "learning_rate": 4.36686346362729e-06, "loss": 0.14983788132667542, "step": 2062 }, { "epoch": 0.5355659397715472, "grad_norm": 0.5982717871665955, "learning_rate": 4.366205449168042e-06, "loss": 0.12300603836774826, "step": 2063 }, { "epoch": 0.5358255451713395, "grad_norm": 0.4749871790409088, "learning_rate": 4.3655471490006766e-06, "loss": 0.11778729408979416, "step": 2064 }, { "epoch": 0.5360851505711319, "grad_norm": 1.286927342414856, "learning_rate": 4.364888563241566e-06, "loss": 0.13144326210021973, "step": 2065 }, { "epoch": 0.5363447559709242, "grad_norm": 1.0259122848510742, "learning_rate": 4.364229692007134e-06, "loss": 0.14625120162963867, "step": 2066 }, { "epoch": 0.5366043613707165, "grad_norm": 1.508382797241211, "learning_rate": 4.363570535413852e-06, "loss": 0.1435680389404297, "step": 2067 }, { "epoch": 0.5368639667705088, "grad_norm": 0.8771193623542786, "learning_rate": 4.362911093578246e-06, "loss": 0.16274556517601013, "step": 2068 }, { "epoch": 0.5371235721703012, "grad_norm": 1.3566398620605469, "learning_rate": 4.362251366616891e-06, "loss": 0.17882457375526428, "step": 2069 }, { "epoch": 0.5373831775700935, "grad_norm": 1.34805166721344, "learning_rate": 4.361591354646409e-06, "loss": 0.1609060913324356, "step": 2070 }, { "epoch": 0.5376427829698858, "grad_norm": 0.8784144520759583, "learning_rate": 4.360931057783475e-06, "loss": 0.15549778938293457, "step": 2071 }, { "epoch": 0.5379023883696781, "grad_norm": 1.3743157386779785, "learning_rate": 4.360270476144817e-06, "loss": 0.13979728519916534, "step": 2072 }, { "epoch": 0.5381619937694704, "grad_norm": 0.9996192455291748, "learning_rate": 4.359609609847208e-06, "loss": 0.13845384120941162, "step": 2073 }, { "epoch": 0.5384215991692627, "grad_norm": 0.5752806663513184, "learning_rate": 4.358948459007475e-06, "loss": 0.1295846849679947, "step": 2074 }, { "epoch": 0.538681204569055, "grad_norm": 1.4374723434448242, "learning_rate": 4.358287023742495e-06, "loss": 0.15286894142627716, "step": 2075 }, { "epoch": 0.5389408099688473, "grad_norm": 1.2336862087249756, "learning_rate": 4.357625304169194e-06, "loss": 0.14124369621276855, "step": 2076 }, { "epoch": 0.5392004153686397, "grad_norm": 0.6081708073616028, "learning_rate": 4.356963300404549e-06, "loss": 0.1250338852405548, "step": 2077 }, { "epoch": 0.539460020768432, "grad_norm": 1.0118521451950073, "learning_rate": 4.356301012565588e-06, "loss": 0.13292011618614197, "step": 2078 }, { "epoch": 0.5397196261682243, "grad_norm": 1.2198342084884644, "learning_rate": 4.355638440769385e-06, "loss": 0.18758420646190643, "step": 2079 }, { "epoch": 0.5399792315680166, "grad_norm": 1.6892664432525635, "learning_rate": 4.354975585133072e-06, "loss": 0.15074780583381653, "step": 2080 }, { "epoch": 0.5402388369678089, "grad_norm": 1.450956106185913, "learning_rate": 4.354312445773825e-06, "loss": 0.160169780254364, "step": 2081 }, { "epoch": 0.5404984423676013, "grad_norm": 1.0982112884521484, "learning_rate": 4.353649022808869e-06, "loss": 0.16365794837474823, "step": 2082 }, { "epoch": 0.5407580477673936, "grad_norm": 1.3341407775878906, "learning_rate": 4.352985316355487e-06, "loss": 0.16518187522888184, "step": 2083 }, { "epoch": 0.5410176531671859, "grad_norm": 0.5959885716438293, "learning_rate": 4.3523213265310036e-06, "loss": 0.12867125868797302, "step": 2084 }, { "epoch": 0.5412772585669782, "grad_norm": 1.1227858066558838, "learning_rate": 4.351657053452798e-06, "loss": 0.14274339377880096, "step": 2085 }, { "epoch": 0.5415368639667705, "grad_norm": 2.7419118881225586, "learning_rate": 4.350992497238298e-06, "loss": 0.15305671095848083, "step": 2086 }, { "epoch": 0.5417964693665628, "grad_norm": 1.2818440198898315, "learning_rate": 4.350327658004983e-06, "loss": 0.14636161923408508, "step": 2087 }, { "epoch": 0.5420560747663551, "grad_norm": 0.35028648376464844, "learning_rate": 4.349662535870381e-06, "loss": 0.13075634837150574, "step": 2088 }, { "epoch": 0.5423156801661475, "grad_norm": 1.4436008930206299, "learning_rate": 4.348997130952069e-06, "loss": 0.1751553863286972, "step": 2089 }, { "epoch": 0.5425752855659398, "grad_norm": 1.1844605207443237, "learning_rate": 4.348331443367677e-06, "loss": 0.16974905133247375, "step": 2090 }, { "epoch": 0.5428348909657321, "grad_norm": 1.4659194946289062, "learning_rate": 4.3476654732348826e-06, "loss": 0.11819124966859818, "step": 2091 }, { "epoch": 0.5430944963655244, "grad_norm": 1.4138152599334717, "learning_rate": 4.346999220671414e-06, "loss": 0.16054914891719818, "step": 2092 }, { "epoch": 0.5433541017653167, "grad_norm": 0.42056599259376526, "learning_rate": 4.34633268579505e-06, "loss": 0.12193632125854492, "step": 2093 }, { "epoch": 0.543613707165109, "grad_norm": 1.0305970907211304, "learning_rate": 4.345665868723618e-06, "loss": 0.14213410019874573, "step": 2094 }, { "epoch": 0.5438733125649013, "grad_norm": 1.5455312728881836, "learning_rate": 4.3449987695749964e-06, "loss": 0.16635309159755707, "step": 2095 }, { "epoch": 0.5441329179646937, "grad_norm": 2.010435104370117, "learning_rate": 4.344331388467113e-06, "loss": 0.16786406934261322, "step": 2096 }, { "epoch": 0.544392523364486, "grad_norm": 0.8525503277778625, "learning_rate": 4.343663725517946e-06, "loss": 0.13089311122894287, "step": 2097 }, { "epoch": 0.5446521287642783, "grad_norm": 1.0414780378341675, "learning_rate": 4.342995780845522e-06, "loss": 0.18874210119247437, "step": 2098 }, { "epoch": 0.5449117341640706, "grad_norm": 2.547698736190796, "learning_rate": 4.342327554567918e-06, "loss": 0.17531758546829224, "step": 2099 }, { "epoch": 0.5451713395638629, "grad_norm": 0.6923290491104126, "learning_rate": 4.341659046803263e-06, "loss": 0.1340133398771286, "step": 2100 }, { "epoch": 0.5454309449636553, "grad_norm": 1.0073119401931763, "learning_rate": 4.340990257669732e-06, "loss": 0.13726383447647095, "step": 2101 }, { "epoch": 0.5456905503634476, "grad_norm": 2.1828670501708984, "learning_rate": 4.340321187285553e-06, "loss": 0.16463567316532135, "step": 2102 }, { "epoch": 0.5459501557632399, "grad_norm": 2.496532678604126, "learning_rate": 4.339651835769e-06, "loss": 0.16038313508033752, "step": 2103 }, { "epoch": 0.5462097611630322, "grad_norm": 1.1729450225830078, "learning_rate": 4.338982203238403e-06, "loss": 0.14472347497940063, "step": 2104 }, { "epoch": 0.5464693665628245, "grad_norm": 0.6566655039787292, "learning_rate": 4.338312289812134e-06, "loss": 0.1639712154865265, "step": 2105 }, { "epoch": 0.5467289719626168, "grad_norm": 1.3029483556747437, "learning_rate": 4.33764209560862e-06, "loss": 0.16278135776519775, "step": 2106 }, { "epoch": 0.5469885773624091, "grad_norm": 2.2933762073516846, "learning_rate": 4.336971620746334e-06, "loss": 0.12491636723279953, "step": 2107 }, { "epoch": 0.5472481827622014, "grad_norm": 2.1884429454803467, "learning_rate": 4.336300865343805e-06, "loss": 0.15442585945129395, "step": 2108 }, { "epoch": 0.5475077881619937, "grad_norm": 1.3581138849258423, "learning_rate": 4.335629829519602e-06, "loss": 0.15046410262584686, "step": 2109 }, { "epoch": 0.547767393561786, "grad_norm": 0.7693226337432861, "learning_rate": 4.334958513392353e-06, "loss": 0.13053488731384277, "step": 2110 }, { "epoch": 0.5480269989615784, "grad_norm": 0.7611463665962219, "learning_rate": 4.33428691708073e-06, "loss": 0.128693088889122, "step": 2111 }, { "epoch": 0.5482866043613707, "grad_norm": 0.5923795700073242, "learning_rate": 4.333615040703454e-06, "loss": 0.12585175037384033, "step": 2112 }, { "epoch": 0.5485462097611631, "grad_norm": 2.3049135208129883, "learning_rate": 4.332942884379299e-06, "loss": 0.13894134759902954, "step": 2113 }, { "epoch": 0.5488058151609554, "grad_norm": 1.1824980974197388, "learning_rate": 4.332270448227087e-06, "loss": 0.12789055705070496, "step": 2114 }, { "epoch": 0.5490654205607477, "grad_norm": 0.879863977432251, "learning_rate": 4.331597732365689e-06, "loss": 0.13624924421310425, "step": 2115 }, { "epoch": 0.54932502596054, "grad_norm": 1.5637046098709106, "learning_rate": 4.330924736914026e-06, "loss": 0.1565348505973816, "step": 2116 }, { "epoch": 0.5495846313603323, "grad_norm": 0.8270760774612427, "learning_rate": 4.3302514619910665e-06, "loss": 0.12331441044807434, "step": 2117 }, { "epoch": 0.5498442367601246, "grad_norm": 1.2273591756820679, "learning_rate": 4.329577907715833e-06, "loss": 0.15879419445991516, "step": 2118 }, { "epoch": 0.5501038421599169, "grad_norm": 2.0905797481536865, "learning_rate": 4.328904074207392e-06, "loss": 0.16745401918888092, "step": 2119 }, { "epoch": 0.5503634475597092, "grad_norm": 3.2156620025634766, "learning_rate": 4.328229961584864e-06, "loss": 0.18997198343276978, "step": 2120 }, { "epoch": 0.5506230529595015, "grad_norm": 1.221010684967041, "learning_rate": 4.327555569967414e-06, "loss": 0.15718486905097961, "step": 2121 }, { "epoch": 0.5508826583592938, "grad_norm": 0.4310303330421448, "learning_rate": 4.326880899474261e-06, "loss": 0.14140579104423523, "step": 2122 }, { "epoch": 0.5511422637590861, "grad_norm": 1.1746593713760376, "learning_rate": 4.32620595022467e-06, "loss": 0.13158902525901794, "step": 2123 }, { "epoch": 0.5514018691588785, "grad_norm": 2.2494680881500244, "learning_rate": 4.325530722337959e-06, "loss": 0.16848215460777283, "step": 2124 }, { "epoch": 0.5516614745586709, "grad_norm": 1.5868723392486572, "learning_rate": 4.32485521593349e-06, "loss": 0.12811459600925446, "step": 2125 }, { "epoch": 0.5519210799584632, "grad_norm": 1.0790104866027832, "learning_rate": 4.324179431130679e-06, "loss": 0.12625053524971008, "step": 2126 }, { "epoch": 0.5521806853582555, "grad_norm": 1.7707915306091309, "learning_rate": 4.323503368048987e-06, "loss": 0.15223428606987, "step": 2127 }, { "epoch": 0.5524402907580478, "grad_norm": 1.562003254890442, "learning_rate": 4.32282702680793e-06, "loss": 0.12785258889198303, "step": 2128 }, { "epoch": 0.5526998961578401, "grad_norm": 1.459416151046753, "learning_rate": 4.322150407527067e-06, "loss": 0.1549133062362671, "step": 2129 }, { "epoch": 0.5529595015576324, "grad_norm": 1.7126916646957397, "learning_rate": 4.321473510326009e-06, "loss": 0.14127947390079498, "step": 2130 }, { "epoch": 0.5532191069574247, "grad_norm": 0.9374878406524658, "learning_rate": 4.320796335324416e-06, "loss": 0.17070040106773376, "step": 2131 }, { "epoch": 0.553478712357217, "grad_norm": 0.5554973483085632, "learning_rate": 4.3201188826419985e-06, "loss": 0.14117518067359924, "step": 2132 }, { "epoch": 0.5537383177570093, "grad_norm": 1.408199429512024, "learning_rate": 4.319441152398513e-06, "loss": 0.12370751798152924, "step": 2133 }, { "epoch": 0.5539979231568016, "grad_norm": 0.6483302712440491, "learning_rate": 4.318763144713767e-06, "loss": 0.1465035229921341, "step": 2134 }, { "epoch": 0.5542575285565939, "grad_norm": 0.8814477324485779, "learning_rate": 4.318084859707617e-06, "loss": 0.13342447578907013, "step": 2135 }, { "epoch": 0.5545171339563862, "grad_norm": 0.4677409529685974, "learning_rate": 4.317406297499968e-06, "loss": 0.1254015862941742, "step": 2136 }, { "epoch": 0.5547767393561787, "grad_norm": 1.9379161596298218, "learning_rate": 4.316727458210774e-06, "loss": 0.1441696584224701, "step": 2137 }, { "epoch": 0.555036344755971, "grad_norm": 0.8811458349227905, "learning_rate": 4.31604834196004e-06, "loss": 0.1317538172006607, "step": 2138 }, { "epoch": 0.5552959501557633, "grad_norm": 1.224442958831787, "learning_rate": 4.315368948867816e-06, "loss": 0.15204253792762756, "step": 2139 }, { "epoch": 0.5555555555555556, "grad_norm": 1.3064621686935425, "learning_rate": 4.3146892790542035e-06, "loss": 0.1632775366306305, "step": 2140 }, { "epoch": 0.5558151609553479, "grad_norm": 0.698509156703949, "learning_rate": 4.314009332639355e-06, "loss": 0.1483507752418518, "step": 2141 }, { "epoch": 0.5560747663551402, "grad_norm": 1.212509274482727, "learning_rate": 4.313329109743466e-06, "loss": 0.1341300904750824, "step": 2142 }, { "epoch": 0.5563343717549325, "grad_norm": 0.9859030842781067, "learning_rate": 4.312648610486786e-06, "loss": 0.13280680775642395, "step": 2143 }, { "epoch": 0.5565939771547248, "grad_norm": 1.2259515523910522, "learning_rate": 4.311967834989612e-06, "loss": 0.16925713419914246, "step": 2144 }, { "epoch": 0.5568535825545171, "grad_norm": 0.8953250050544739, "learning_rate": 4.311286783372289e-06, "loss": 0.1425480842590332, "step": 2145 }, { "epoch": 0.5571131879543094, "grad_norm": 0.8011786341667175, "learning_rate": 4.310605455755211e-06, "loss": 0.13758322596549988, "step": 2146 }, { "epoch": 0.5573727933541017, "grad_norm": 0.4695817828178406, "learning_rate": 4.3099238522588225e-06, "loss": 0.132910817861557, "step": 2147 }, { "epoch": 0.557632398753894, "grad_norm": 0.7648541927337646, "learning_rate": 4.309241973003614e-06, "loss": 0.11924894899129868, "step": 2148 }, { "epoch": 0.5578920041536864, "grad_norm": 0.872509777545929, "learning_rate": 4.3085598181101256e-06, "loss": 0.17583368718624115, "step": 2149 }, { "epoch": 0.5581516095534788, "grad_norm": 1.3153228759765625, "learning_rate": 4.307877387698949e-06, "loss": 0.13956871628761292, "step": 2150 }, { "epoch": 0.5584112149532711, "grad_norm": 1.1751517057418823, "learning_rate": 4.30719468189072e-06, "loss": 0.154902845621109, "step": 2151 }, { "epoch": 0.5586708203530634, "grad_norm": 0.8956025838851929, "learning_rate": 4.306511700806127e-06, "loss": 0.12491951137781143, "step": 2152 }, { "epoch": 0.5589304257528557, "grad_norm": 1.5569437742233276, "learning_rate": 4.305828444565904e-06, "loss": 0.13498555123806, "step": 2153 }, { "epoch": 0.559190031152648, "grad_norm": 1.0736583471298218, "learning_rate": 4.305144913290837e-06, "loss": 0.1298777163028717, "step": 2154 }, { "epoch": 0.5594496365524403, "grad_norm": 3.433091402053833, "learning_rate": 4.304461107101756e-06, "loss": 0.17167623341083527, "step": 2155 }, { "epoch": 0.5597092419522326, "grad_norm": 0.765629231929779, "learning_rate": 4.303777026119545e-06, "loss": 0.15894079208374023, "step": 2156 }, { "epoch": 0.5599688473520249, "grad_norm": 1.9295969009399414, "learning_rate": 4.3030926704651314e-06, "loss": 0.16929194331169128, "step": 2157 }, { "epoch": 0.5602284527518172, "grad_norm": 0.5996467471122742, "learning_rate": 4.302408040259495e-06, "loss": 0.14456358551979065, "step": 2158 }, { "epoch": 0.5604880581516095, "grad_norm": 1.5874388217926025, "learning_rate": 4.301723135623662e-06, "loss": 0.1676119863986969, "step": 2159 }, { "epoch": 0.5607476635514018, "grad_norm": 1.1257277727127075, "learning_rate": 4.30103795667871e-06, "loss": 0.13293232023715973, "step": 2160 }, { "epoch": 0.5610072689511942, "grad_norm": 1.8176615238189697, "learning_rate": 4.300352503545761e-06, "loss": 0.14383424818515778, "step": 2161 }, { "epoch": 0.5612668743509865, "grad_norm": 2.266650915145874, "learning_rate": 4.299666776345987e-06, "loss": 0.17376676201820374, "step": 2162 }, { "epoch": 0.5615264797507789, "grad_norm": 0.5792590975761414, "learning_rate": 4.298980775200609e-06, "loss": 0.13912424445152283, "step": 2163 }, { "epoch": 0.5617860851505712, "grad_norm": 1.3648930788040161, "learning_rate": 4.298294500230898e-06, "loss": 0.12999987602233887, "step": 2164 }, { "epoch": 0.5620456905503635, "grad_norm": 1.6013997793197632, "learning_rate": 4.29760795155817e-06, "loss": 0.1499829888343811, "step": 2165 }, { "epoch": 0.5623052959501558, "grad_norm": 1.591899037361145, "learning_rate": 4.296921129303792e-06, "loss": 0.15002956986427307, "step": 2166 }, { "epoch": 0.5625649013499481, "grad_norm": 1.089423656463623, "learning_rate": 4.296234033589178e-06, "loss": 0.14751307666301727, "step": 2167 }, { "epoch": 0.5628245067497404, "grad_norm": 1.3160361051559448, "learning_rate": 4.295546664535791e-06, "loss": 0.15664774179458618, "step": 2168 }, { "epoch": 0.5630841121495327, "grad_norm": 1.089060664176941, "learning_rate": 4.2948590222651424e-06, "loss": 0.13762149214744568, "step": 2169 }, { "epoch": 0.563343717549325, "grad_norm": 0.800458550453186, "learning_rate": 4.294171106898791e-06, "loss": 0.1515216827392578, "step": 2170 }, { "epoch": 0.5636033229491173, "grad_norm": 0.5633087158203125, "learning_rate": 4.293482918558344e-06, "loss": 0.13268117606639862, "step": 2171 }, { "epoch": 0.5638629283489096, "grad_norm": 1.548930287361145, "learning_rate": 4.2927944573654595e-06, "loss": 0.1496484875679016, "step": 2172 }, { "epoch": 0.564122533748702, "grad_norm": 1.9444133043289185, "learning_rate": 4.292105723441839e-06, "loss": 0.1512450873851776, "step": 2173 }, { "epoch": 0.5643821391484943, "grad_norm": 1.4145973920822144, "learning_rate": 4.291416716909237e-06, "loss": 0.14328570663928986, "step": 2174 }, { "epoch": 0.5646417445482866, "grad_norm": 0.6859724521636963, "learning_rate": 4.290727437889453e-06, "loss": 0.1280905306339264, "step": 2175 }, { "epoch": 0.564901349948079, "grad_norm": 0.35122114419937134, "learning_rate": 4.290037886504335e-06, "loss": 0.11795367300510406, "step": 2176 }, { "epoch": 0.5651609553478713, "grad_norm": 0.6748156547546387, "learning_rate": 4.289348062875782e-06, "loss": 0.1469731479883194, "step": 2177 }, { "epoch": 0.5654205607476636, "grad_norm": 1.529887318611145, "learning_rate": 4.288657967125738e-06, "loss": 0.15771859884262085, "step": 2178 }, { "epoch": 0.5656801661474559, "grad_norm": 2.258462905883789, "learning_rate": 4.2879675993761954e-06, "loss": 0.16884522140026093, "step": 2179 }, { "epoch": 0.5659397715472482, "grad_norm": 1.6096688508987427, "learning_rate": 4.287276959749196e-06, "loss": 0.12439864873886108, "step": 2180 }, { "epoch": 0.5661993769470405, "grad_norm": 1.3519511222839355, "learning_rate": 4.286586048366829e-06, "loss": 0.11922718584537506, "step": 2181 }, { "epoch": 0.5664589823468328, "grad_norm": 1.4625903367996216, "learning_rate": 4.285894865351233e-06, "loss": 0.17085392773151398, "step": 2182 }, { "epoch": 0.5667185877466251, "grad_norm": 1.1445099115371704, "learning_rate": 4.2852034108245895e-06, "loss": 0.14169737696647644, "step": 2183 }, { "epoch": 0.5669781931464174, "grad_norm": 1.6349550485610962, "learning_rate": 4.284511684909136e-06, "loss": 0.13192304968833923, "step": 2184 }, { "epoch": 0.5672377985462098, "grad_norm": 1.7558975219726562, "learning_rate": 4.2838196877271536e-06, "loss": 0.14318935573101044, "step": 2185 }, { "epoch": 0.5674974039460021, "grad_norm": 0.9661133289337158, "learning_rate": 4.283127419400969e-06, "loss": 0.14467504620552063, "step": 2186 }, { "epoch": 0.5677570093457944, "grad_norm": 0.9093027710914612, "learning_rate": 4.28243488005296e-06, "loss": 0.1293337047100067, "step": 2187 }, { "epoch": 0.5680166147455867, "grad_norm": 1.077207088470459, "learning_rate": 4.281742069805553e-06, "loss": 0.1313164085149765, "step": 2188 }, { "epoch": 0.568276220145379, "grad_norm": 0.4229333698749542, "learning_rate": 4.281048988781221e-06, "loss": 0.13534891605377197, "step": 2189 }, { "epoch": 0.5685358255451713, "grad_norm": 0.8727226853370667, "learning_rate": 4.280355637102483e-06, "loss": 0.10711170732975006, "step": 2190 }, { "epoch": 0.5687954309449637, "grad_norm": 1.3826498985290527, "learning_rate": 4.279662014891909e-06, "loss": 0.12936684489250183, "step": 2191 }, { "epoch": 0.569055036344756, "grad_norm": 3.3055644035339355, "learning_rate": 4.2789681222721155e-06, "loss": 0.17132937908172607, "step": 2192 }, { "epoch": 0.5693146417445483, "grad_norm": 1.5493476390838623, "learning_rate": 4.278273959365768e-06, "loss": 0.13380032777786255, "step": 2193 }, { "epoch": 0.5695742471443406, "grad_norm": 1.0912903547286987, "learning_rate": 4.277579526295575e-06, "loss": 0.1257137656211853, "step": 2194 }, { "epoch": 0.5698338525441329, "grad_norm": 1.1170724630355835, "learning_rate": 4.2768848231843e-06, "loss": 0.16309235990047455, "step": 2195 }, { "epoch": 0.5700934579439252, "grad_norm": 0.4980577528476715, "learning_rate": 4.27618985015475e-06, "loss": 0.14452283084392548, "step": 2196 }, { "epoch": 0.5703530633437176, "grad_norm": 1.2892794609069824, "learning_rate": 4.275494607329778e-06, "loss": 0.1484300047159195, "step": 2197 }, { "epoch": 0.5706126687435099, "grad_norm": 0.8269965648651123, "learning_rate": 4.27479909483229e-06, "loss": 0.13300426304340363, "step": 2198 }, { "epoch": 0.5708722741433022, "grad_norm": 0.7830909490585327, "learning_rate": 4.274103312785235e-06, "loss": 0.15134841203689575, "step": 2199 }, { "epoch": 0.5711318795430945, "grad_norm": 1.1842620372772217, "learning_rate": 4.27340726131161e-06, "loss": 0.16099253296852112, "step": 2200 }, { "epoch": 0.5713914849428868, "grad_norm": 1.2201128005981445, "learning_rate": 4.272710940534465e-06, "loss": 0.15309080481529236, "step": 2201 }, { "epoch": 0.5716510903426791, "grad_norm": 0.8669965863227844, "learning_rate": 4.27201435057689e-06, "loss": 0.1328464150428772, "step": 2202 }, { "epoch": 0.5719106957424714, "grad_norm": 0.939900279045105, "learning_rate": 4.271317491562028e-06, "loss": 0.12055699527263641, "step": 2203 }, { "epoch": 0.5721703011422637, "grad_norm": 0.9723026752471924, "learning_rate": 4.270620363613066e-06, "loss": 0.13324546813964844, "step": 2204 }, { "epoch": 0.572429906542056, "grad_norm": 1.5940167903900146, "learning_rate": 4.269922966853242e-06, "loss": 0.14967064559459686, "step": 2205 }, { "epoch": 0.5726895119418484, "grad_norm": 0.4683036208152771, "learning_rate": 4.269225301405838e-06, "loss": 0.13066264986991882, "step": 2206 }, { "epoch": 0.5729491173416407, "grad_norm": 1.9738019704818726, "learning_rate": 4.268527367394188e-06, "loss": 0.18178030848503113, "step": 2207 }, { "epoch": 0.573208722741433, "grad_norm": 1.189687728881836, "learning_rate": 4.267829164941669e-06, "loss": 0.1419353187084198, "step": 2208 }, { "epoch": 0.5734683281412254, "grad_norm": 0.5594265460968018, "learning_rate": 4.267130694171706e-06, "loss": 0.1310182809829712, "step": 2209 }, { "epoch": 0.5737279335410177, "grad_norm": 0.6179895997047424, "learning_rate": 4.266431955207774e-06, "loss": 0.1675294041633606, "step": 2210 }, { "epoch": 0.57398753894081, "grad_norm": 1.043413519859314, "learning_rate": 4.265732948173395e-06, "loss": 0.15660183131694794, "step": 2211 }, { "epoch": 0.5742471443406023, "grad_norm": 0.7484740614891052, "learning_rate": 4.265033673192136e-06, "loss": 0.13695743680000305, "step": 2212 }, { "epoch": 0.5745067497403946, "grad_norm": 0.7406579256057739, "learning_rate": 4.264334130387614e-06, "loss": 0.16828250885009766, "step": 2213 }, { "epoch": 0.5747663551401869, "grad_norm": 2.7745096683502197, "learning_rate": 4.26363431988349e-06, "loss": 0.1713849902153015, "step": 2214 }, { "epoch": 0.5750259605399792, "grad_norm": 1.2922271490097046, "learning_rate": 4.262934241803477e-06, "loss": 0.15023668110370636, "step": 2215 }, { "epoch": 0.5752855659397715, "grad_norm": 0.7837429046630859, "learning_rate": 4.2622338962713315e-06, "loss": 0.12317696958780289, "step": 2216 }, { "epoch": 0.5755451713395638, "grad_norm": 0.6875805854797363, "learning_rate": 4.261533283410858e-06, "loss": 0.15823142230510712, "step": 2217 }, { "epoch": 0.5758047767393562, "grad_norm": 1.708173394203186, "learning_rate": 4.26083240334591e-06, "loss": 0.1442362368106842, "step": 2218 }, { "epoch": 0.5760643821391485, "grad_norm": 0.9522128701210022, "learning_rate": 4.260131256200387e-06, "loss": 0.13462968170642853, "step": 2219 }, { "epoch": 0.5763239875389408, "grad_norm": 1.3197602033615112, "learning_rate": 4.259429842098235e-06, "loss": 0.1395062506198883, "step": 2220 }, { "epoch": 0.5765835929387332, "grad_norm": 0.9885492920875549, "learning_rate": 4.258728161163448e-06, "loss": 0.12320567667484283, "step": 2221 }, { "epoch": 0.5768431983385255, "grad_norm": 0.7654373645782471, "learning_rate": 4.258026213520067e-06, "loss": 0.13010457158088684, "step": 2222 }, { "epoch": 0.5771028037383178, "grad_norm": 0.7320343255996704, "learning_rate": 4.257323999292182e-06, "loss": 0.16303765773773193, "step": 2223 }, { "epoch": 0.5773624091381101, "grad_norm": 1.2722511291503906, "learning_rate": 4.256621518603926e-06, "loss": 0.12455020099878311, "step": 2224 }, { "epoch": 0.5776220145379024, "grad_norm": 1.1728321313858032, "learning_rate": 4.255918771579482e-06, "loss": 0.17042791843414307, "step": 2225 }, { "epoch": 0.5778816199376947, "grad_norm": 1.9902153015136719, "learning_rate": 4.255215758343082e-06, "loss": 0.14805909991264343, "step": 2226 }, { "epoch": 0.578141225337487, "grad_norm": 0.9607979655265808, "learning_rate": 4.2545124790189995e-06, "loss": 0.127325177192688, "step": 2227 }, { "epoch": 0.5784008307372793, "grad_norm": 0.5636826157569885, "learning_rate": 4.2538089337315594e-06, "loss": 0.1449686586856842, "step": 2228 }, { "epoch": 0.5786604361370716, "grad_norm": 0.607964038848877, "learning_rate": 4.253105122605134e-06, "loss": 0.11601535230875015, "step": 2229 }, { "epoch": 0.5789200415368639, "grad_norm": 0.9971367120742798, "learning_rate": 4.252401045764138e-06, "loss": 0.14506377279758453, "step": 2230 }, { "epoch": 0.5791796469366562, "grad_norm": 1.6546710729599, "learning_rate": 4.251696703333037e-06, "loss": 0.1350918710231781, "step": 2231 }, { "epoch": 0.5794392523364486, "grad_norm": 0.5777446031570435, "learning_rate": 4.250992095436345e-06, "loss": 0.12550590932369232, "step": 2232 }, { "epoch": 0.579698857736241, "grad_norm": 0.8886168003082275, "learning_rate": 4.250287222198618e-06, "loss": 0.1154436394572258, "step": 2233 }, { "epoch": 0.5799584631360333, "grad_norm": 0.5563759803771973, "learning_rate": 4.249582083744461e-06, "loss": 0.12805479764938354, "step": 2234 }, { "epoch": 0.5802180685358256, "grad_norm": 3.838718891143799, "learning_rate": 4.248876680198529e-06, "loss": 0.15458905696868896, "step": 2235 }, { "epoch": 0.5804776739356179, "grad_norm": 0.49812495708465576, "learning_rate": 4.248171011685519e-06, "loss": 0.13039326667785645, "step": 2236 }, { "epoch": 0.5807372793354102, "grad_norm": 0.7976740598678589, "learning_rate": 4.247465078330177e-06, "loss": 0.13807369768619537, "step": 2237 }, { "epoch": 0.5809968847352025, "grad_norm": 1.0168839693069458, "learning_rate": 4.246758880257299e-06, "loss": 0.15039704740047455, "step": 2238 }, { "epoch": 0.5812564901349948, "grad_norm": 0.7996509075164795, "learning_rate": 4.24605241759172e-06, "loss": 0.1385539323091507, "step": 2239 }, { "epoch": 0.5815160955347871, "grad_norm": 0.9554955959320068, "learning_rate": 4.245345690458329e-06, "loss": 0.11792933940887451, "step": 2240 }, { "epoch": 0.5817757009345794, "grad_norm": 2.0399389266967773, "learning_rate": 4.244638698982059e-06, "loss": 0.15396463871002197, "step": 2241 }, { "epoch": 0.5820353063343717, "grad_norm": 1.6898654699325562, "learning_rate": 4.243931443287889e-06, "loss": 0.12861531972885132, "step": 2242 }, { "epoch": 0.582294911734164, "grad_norm": 0.6715939044952393, "learning_rate": 4.243223923500846e-06, "loss": 0.12669500708580017, "step": 2243 }, { "epoch": 0.5825545171339563, "grad_norm": 0.9264000654220581, "learning_rate": 4.2425161397460055e-06, "loss": 0.12242238223552704, "step": 2244 }, { "epoch": 0.5828141225337488, "grad_norm": 1.0984553098678589, "learning_rate": 4.241808092148485e-06, "loss": 0.15966880321502686, "step": 2245 }, { "epoch": 0.5830737279335411, "grad_norm": 1.4158947467803955, "learning_rate": 4.241099780833451e-06, "loss": 0.13803543150424957, "step": 2246 }, { "epoch": 0.5833333333333334, "grad_norm": 1.0771379470825195, "learning_rate": 4.240391205926118e-06, "loss": 0.13035908341407776, "step": 2247 }, { "epoch": 0.5835929387331257, "grad_norm": 0.4804001450538635, "learning_rate": 4.239682367551745e-06, "loss": 0.1302664875984192, "step": 2248 }, { "epoch": 0.583852544132918, "grad_norm": 0.6805130243301392, "learning_rate": 4.238973265835639e-06, "loss": 0.11880309879779816, "step": 2249 }, { "epoch": 0.5841121495327103, "grad_norm": 0.6841399669647217, "learning_rate": 4.238263900903152e-06, "loss": 0.13439232110977173, "step": 2250 }, { "epoch": 0.5843717549325026, "grad_norm": 1.566993236541748, "learning_rate": 4.2375542728796835e-06, "loss": 0.13131317496299744, "step": 2251 }, { "epoch": 0.5846313603322949, "grad_norm": 0.8833286166191101, "learning_rate": 4.236844381890681e-06, "loss": 0.13651561737060547, "step": 2252 }, { "epoch": 0.5848909657320872, "grad_norm": 0.5520321130752563, "learning_rate": 4.236134228061636e-06, "loss": 0.14803805947303772, "step": 2253 }, { "epoch": 0.5851505711318795, "grad_norm": 0.9788375496864319, "learning_rate": 4.235423811518087e-06, "loss": 0.1273384988307953, "step": 2254 }, { "epoch": 0.5854101765316718, "grad_norm": 1.847069501876831, "learning_rate": 4.2347131323856195e-06, "loss": 0.17851579189300537, "step": 2255 }, { "epoch": 0.5856697819314641, "grad_norm": 0.9802236557006836, "learning_rate": 4.234002190789867e-06, "loss": 0.1530790627002716, "step": 2256 }, { "epoch": 0.5859293873312564, "grad_norm": 1.4136950969696045, "learning_rate": 4.233290986856504e-06, "loss": 0.12046264857053757, "step": 2257 }, { "epoch": 0.5861889927310489, "grad_norm": 2.388747453689575, "learning_rate": 4.23257952071126e-06, "loss": 0.131452277302742, "step": 2258 }, { "epoch": 0.5864485981308412, "grad_norm": 3.7876763343811035, "learning_rate": 4.231867792479902e-06, "loss": 0.15764811635017395, "step": 2259 }, { "epoch": 0.5867082035306335, "grad_norm": 0.6489400863647461, "learning_rate": 4.2311558022882485e-06, "loss": 0.1299557387828827, "step": 2260 }, { "epoch": 0.5869678089304258, "grad_norm": 0.7358956336975098, "learning_rate": 4.2304435502621635e-06, "loss": 0.14383837580680847, "step": 2261 }, { "epoch": 0.5872274143302181, "grad_norm": 0.8988000154495239, "learning_rate": 4.229731036527556e-06, "loss": 0.12811490893363953, "step": 2262 }, { "epoch": 0.5874870197300104, "grad_norm": 2.1063284873962402, "learning_rate": 4.229018261210383e-06, "loss": 0.14225494861602783, "step": 2263 }, { "epoch": 0.5877466251298027, "grad_norm": 1.2560406923294067, "learning_rate": 4.2283052244366465e-06, "loss": 0.1515149176120758, "step": 2264 }, { "epoch": 0.588006230529595, "grad_norm": 2.6018166542053223, "learning_rate": 4.227591926332394e-06, "loss": 0.16990751028060913, "step": 2265 }, { "epoch": 0.5882658359293873, "grad_norm": 1.6254899501800537, "learning_rate": 4.226878367023724e-06, "loss": 0.1604834496974945, "step": 2266 }, { "epoch": 0.5885254413291796, "grad_norm": 0.40035563707351685, "learning_rate": 4.226164546636772e-06, "loss": 0.13261745870113373, "step": 2267 }, { "epoch": 0.5887850467289719, "grad_norm": 0.7627634406089783, "learning_rate": 4.2254504652977285e-06, "loss": 0.13511265814304352, "step": 2268 }, { "epoch": 0.5890446521287642, "grad_norm": 0.6826353073120117, "learning_rate": 4.224736123132826e-06, "loss": 0.11303715407848358, "step": 2269 }, { "epoch": 0.5893042575285566, "grad_norm": 3.3403968811035156, "learning_rate": 4.224021520268345e-06, "loss": 0.15396183729171753, "step": 2270 }, { "epoch": 0.589563862928349, "grad_norm": 1.6062710285186768, "learning_rate": 4.2233066568306084e-06, "loss": 0.13388791680335999, "step": 2271 }, { "epoch": 0.5898234683281413, "grad_norm": 1.905274748802185, "learning_rate": 4.22259153294599e-06, "loss": 0.14282605051994324, "step": 2272 }, { "epoch": 0.5900830737279336, "grad_norm": 0.9853888750076294, "learning_rate": 4.221876148740906e-06, "loss": 0.13358530402183533, "step": 2273 }, { "epoch": 0.5903426791277259, "grad_norm": 0.5619341135025024, "learning_rate": 4.221160504341821e-06, "loss": 0.1515040248632431, "step": 2274 }, { "epoch": 0.5906022845275182, "grad_norm": 1.0949472188949585, "learning_rate": 4.220444599875243e-06, "loss": 0.13024254143238068, "step": 2275 }, { "epoch": 0.5908618899273105, "grad_norm": 1.740714192390442, "learning_rate": 4.219728435467729e-06, "loss": 0.1334589123725891, "step": 2276 }, { "epoch": 0.5911214953271028, "grad_norm": 2.024717330932617, "learning_rate": 4.21901201124588e-06, "loss": 0.15233725309371948, "step": 2277 }, { "epoch": 0.5913811007268951, "grad_norm": 1.041885495185852, "learning_rate": 4.2182953273363445e-06, "loss": 0.14881527423858643, "step": 2278 }, { "epoch": 0.5916407061266874, "grad_norm": 1.0380756855010986, "learning_rate": 4.217578383865813e-06, "loss": 0.1404321938753128, "step": 2279 }, { "epoch": 0.5919003115264797, "grad_norm": 1.4251625537872314, "learning_rate": 4.216861180961026e-06, "loss": 0.17093564569950104, "step": 2280 }, { "epoch": 0.592159916926272, "grad_norm": 1.3630878925323486, "learning_rate": 4.21614371874877e-06, "loss": 0.16024619340896606, "step": 2281 }, { "epoch": 0.5924195223260644, "grad_norm": 1.761519193649292, "learning_rate": 4.215425997355875e-06, "loss": 0.14228205382823944, "step": 2282 }, { "epoch": 0.5926791277258567, "grad_norm": 1.172609806060791, "learning_rate": 4.214708016909216e-06, "loss": 0.12654493749141693, "step": 2283 }, { "epoch": 0.592938733125649, "grad_norm": 4.01507043838501, "learning_rate": 4.213989777535719e-06, "loss": 0.15819427371025085, "step": 2284 }, { "epoch": 0.5931983385254413, "grad_norm": 0.416380912065506, "learning_rate": 4.2132712793623485e-06, "loss": 0.11725105345249176, "step": 2285 }, { "epoch": 0.5934579439252337, "grad_norm": 1.5555216073989868, "learning_rate": 4.2125525225161205e-06, "loss": 0.12039388716220856, "step": 2286 }, { "epoch": 0.593717549325026, "grad_norm": 2.5207936763763428, "learning_rate": 4.211833507124095e-06, "loss": 0.1465930938720703, "step": 2287 }, { "epoch": 0.5939771547248183, "grad_norm": 3.815614700317383, "learning_rate": 4.211114233313377e-06, "loss": 0.16588225960731506, "step": 2288 }, { "epoch": 0.5942367601246106, "grad_norm": 1.3989272117614746, "learning_rate": 4.2103947012111176e-06, "loss": 0.13192197680473328, "step": 2289 }, { "epoch": 0.5944963655244029, "grad_norm": 0.3830917179584503, "learning_rate": 4.209674910944513e-06, "loss": 0.1292702555656433, "step": 2290 }, { "epoch": 0.5947559709241952, "grad_norm": 1.1282906532287598, "learning_rate": 4.208954862640807e-06, "loss": 0.12129200994968414, "step": 2291 }, { "epoch": 0.5950155763239875, "grad_norm": 1.0504801273345947, "learning_rate": 4.208234556427288e-06, "loss": 0.12338007986545563, "step": 2292 }, { "epoch": 0.5952751817237798, "grad_norm": 2.0590476989746094, "learning_rate": 4.207513992431286e-06, "loss": 0.15411067008972168, "step": 2293 }, { "epoch": 0.5955347871235722, "grad_norm": 1.9890685081481934, "learning_rate": 4.206793170780185e-06, "loss": 0.12689971923828125, "step": 2294 }, { "epoch": 0.5957943925233645, "grad_norm": 1.2593350410461426, "learning_rate": 4.206072091601406e-06, "loss": 0.17661991715431213, "step": 2295 }, { "epoch": 0.5960539979231568, "grad_norm": 0.6509336233139038, "learning_rate": 4.2053507550224224e-06, "loss": 0.15092894434928894, "step": 2296 }, { "epoch": 0.5963136033229491, "grad_norm": 1.0459539890289307, "learning_rate": 4.204629161170747e-06, "loss": 0.1542167067527771, "step": 2297 }, { "epoch": 0.5965732087227414, "grad_norm": 0.9828463196754456, "learning_rate": 4.2039073101739434e-06, "loss": 0.11261782795190811, "step": 2298 }, { "epoch": 0.5968328141225338, "grad_norm": 1.6083457469940186, "learning_rate": 4.203185202159617e-06, "loss": 0.1349659264087677, "step": 2299 }, { "epoch": 0.5970924195223261, "grad_norm": 1.1662931442260742, "learning_rate": 4.202462837255421e-06, "loss": 0.1392998844385147, "step": 2300 }, { "epoch": 0.5973520249221184, "grad_norm": 0.7593648433685303, "learning_rate": 4.201740215589051e-06, "loss": 0.1212771087884903, "step": 2301 }, { "epoch": 0.5976116303219107, "grad_norm": 0.751245379447937, "learning_rate": 4.201017337288253e-06, "loss": 0.1387275606393814, "step": 2302 }, { "epoch": 0.597871235721703, "grad_norm": 0.571230947971344, "learning_rate": 4.2002942024808125e-06, "loss": 0.12925970554351807, "step": 2303 }, { "epoch": 0.5981308411214953, "grad_norm": 1.0395832061767578, "learning_rate": 4.199570811294564e-06, "loss": 0.11400146782398224, "step": 2304 }, { "epoch": 0.5983904465212876, "grad_norm": 0.4179803431034088, "learning_rate": 4.198847163857388e-06, "loss": 0.13665533065795898, "step": 2305 }, { "epoch": 0.59865005192108, "grad_norm": 1.1577154397964478, "learning_rate": 4.1981232602972075e-06, "loss": 0.1411379873752594, "step": 2306 }, { "epoch": 0.5989096573208723, "grad_norm": 0.5581679940223694, "learning_rate": 4.197399100741991e-06, "loss": 0.1408814787864685, "step": 2307 }, { "epoch": 0.5991692627206646, "grad_norm": 1.6731364727020264, "learning_rate": 4.196674685319755e-06, "loss": 0.18932420015335083, "step": 2308 }, { "epoch": 0.5994288681204569, "grad_norm": 1.7413967847824097, "learning_rate": 4.19595001415856e-06, "loss": 0.13866844773292542, "step": 2309 }, { "epoch": 0.5996884735202492, "grad_norm": 2.3275973796844482, "learning_rate": 4.19522508738651e-06, "loss": 0.1363133043050766, "step": 2310 }, { "epoch": 0.5999480789200415, "grad_norm": 0.6687864661216736, "learning_rate": 4.194499905131755e-06, "loss": 0.1265631765127182, "step": 2311 }, { "epoch": 0.6002076843198338, "grad_norm": 0.5303484201431274, "learning_rate": 4.1937744675224925e-06, "loss": 0.12797121703624725, "step": 2312 }, { "epoch": 0.6004672897196262, "grad_norm": 0.8195741772651672, "learning_rate": 4.193048774686961e-06, "loss": 0.11903519183397293, "step": 2313 }, { "epoch": 0.6007268951194185, "grad_norm": 2.069246292114258, "learning_rate": 4.192322826753449e-06, "loss": 0.1528368443250656, "step": 2314 }, { "epoch": 0.6009865005192108, "grad_norm": 1.744726300239563, "learning_rate": 4.1915966238502855e-06, "loss": 0.13414865732192993, "step": 2315 }, { "epoch": 0.6012461059190031, "grad_norm": 1.6453677415847778, "learning_rate": 4.1908701661058465e-06, "loss": 0.14194709062576294, "step": 2316 }, { "epoch": 0.6015057113187954, "grad_norm": 1.52095365524292, "learning_rate": 4.190143453648554e-06, "loss": 0.1339641511440277, "step": 2317 }, { "epoch": 0.6017653167185878, "grad_norm": 1.2910147905349731, "learning_rate": 4.1894164866068745e-06, "loss": 0.14184482395648956, "step": 2318 }, { "epoch": 0.6020249221183801, "grad_norm": 1.199120283126831, "learning_rate": 4.188689265109318e-06, "loss": 0.13224747776985168, "step": 2319 }, { "epoch": 0.6022845275181724, "grad_norm": 1.2671904563903809, "learning_rate": 4.187961789284441e-06, "loss": 0.1814723014831543, "step": 2320 }, { "epoch": 0.6025441329179647, "grad_norm": 0.6910039186477661, "learning_rate": 4.187234059260845e-06, "loss": 0.1342199742794037, "step": 2321 }, { "epoch": 0.602803738317757, "grad_norm": 2.223318099975586, "learning_rate": 4.186506075167174e-06, "loss": 0.16969284415245056, "step": 2322 }, { "epoch": 0.6030633437175493, "grad_norm": 0.5534135699272156, "learning_rate": 4.1857778371321225e-06, "loss": 0.13063329458236694, "step": 2323 }, { "epoch": 0.6033229491173416, "grad_norm": 1.213074803352356, "learning_rate": 4.185049345284423e-06, "loss": 0.18141406774520874, "step": 2324 }, { "epoch": 0.6035825545171339, "grad_norm": 0.5695230960845947, "learning_rate": 4.184320599752858e-06, "loss": 0.1228247657418251, "step": 2325 }, { "epoch": 0.6038421599169262, "grad_norm": 0.7984333038330078, "learning_rate": 4.183591600666253e-06, "loss": 0.13165545463562012, "step": 2326 }, { "epoch": 0.6041017653167186, "grad_norm": 0.8579720854759216, "learning_rate": 4.182862348153478e-06, "loss": 0.13445739448070526, "step": 2327 }, { "epoch": 0.6043613707165109, "grad_norm": 1.140733003616333, "learning_rate": 4.182132842343448e-06, "loss": 0.1852092742919922, "step": 2328 }, { "epoch": 0.6046209761163032, "grad_norm": 1.6692205667495728, "learning_rate": 4.181403083365122e-06, "loss": 0.14847174286842346, "step": 2329 }, { "epoch": 0.6048805815160956, "grad_norm": 0.6482698321342468, "learning_rate": 4.180673071347506e-06, "loss": 0.1608787477016449, "step": 2330 }, { "epoch": 0.6051401869158879, "grad_norm": 0.85514897108078, "learning_rate": 4.17994280641965e-06, "loss": 0.14532923698425293, "step": 2331 }, { "epoch": 0.6053997923156802, "grad_norm": 0.9341305494308472, "learning_rate": 4.179212288710646e-06, "loss": 0.1361994445323944, "step": 2332 }, { "epoch": 0.6056593977154725, "grad_norm": 1.0442733764648438, "learning_rate": 4.178481518349634e-06, "loss": 0.13547557592391968, "step": 2333 }, { "epoch": 0.6059190031152648, "grad_norm": 0.8463013172149658, "learning_rate": 4.1777504954657975e-06, "loss": 0.14075447618961334, "step": 2334 }, { "epoch": 0.6061786085150571, "grad_norm": 1.6493498086929321, "learning_rate": 4.1770192201883656e-06, "loss": 0.18773004412651062, "step": 2335 }, { "epoch": 0.6064382139148494, "grad_norm": 0.7091905474662781, "learning_rate": 4.17628769264661e-06, "loss": 0.1743507981300354, "step": 2336 }, { "epoch": 0.6066978193146417, "grad_norm": 1.4010348320007324, "learning_rate": 4.175555912969846e-06, "loss": 0.17192965745925903, "step": 2337 }, { "epoch": 0.606957424714434, "grad_norm": 1.0922569036483765, "learning_rate": 4.17482388128744e-06, "loss": 0.11987000703811646, "step": 2338 }, { "epoch": 0.6072170301142263, "grad_norm": 1.193038821220398, "learning_rate": 4.174091597728793e-06, "loss": 0.15422162413597107, "step": 2339 }, { "epoch": 0.6074766355140186, "grad_norm": 2.175330638885498, "learning_rate": 4.173359062423361e-06, "loss": 0.1406227946281433, "step": 2340 }, { "epoch": 0.607736240913811, "grad_norm": 0.6325914859771729, "learning_rate": 4.172626275500637e-06, "loss": 0.12461137771606445, "step": 2341 }, { "epoch": 0.6079958463136034, "grad_norm": 0.6513791680335999, "learning_rate": 4.171893237090161e-06, "loss": 0.12717165052890778, "step": 2342 }, { "epoch": 0.6082554517133957, "grad_norm": 0.5746421813964844, "learning_rate": 4.171159947321519e-06, "loss": 0.14948579668998718, "step": 2343 }, { "epoch": 0.608515057113188, "grad_norm": 0.7215250730514526, "learning_rate": 4.170426406324337e-06, "loss": 0.12913388013839722, "step": 2344 }, { "epoch": 0.6087746625129803, "grad_norm": 0.676729679107666, "learning_rate": 4.16969261422829e-06, "loss": 0.11616098880767822, "step": 2345 }, { "epoch": 0.6090342679127726, "grad_norm": 0.5968407988548279, "learning_rate": 4.168958571163097e-06, "loss": 0.13468413054943085, "step": 2346 }, { "epoch": 0.6092938733125649, "grad_norm": 2.327138900756836, "learning_rate": 4.168224277258517e-06, "loss": 0.14600585401058197, "step": 2347 }, { "epoch": 0.6095534787123572, "grad_norm": 1.2054603099822998, "learning_rate": 4.167489732644358e-06, "loss": 0.16322070360183716, "step": 2348 }, { "epoch": 0.6098130841121495, "grad_norm": 0.7364179491996765, "learning_rate": 4.166754937450471e-06, "loss": 0.13659924268722534, "step": 2349 }, { "epoch": 0.6100726895119418, "grad_norm": 1.1874959468841553, "learning_rate": 4.1660198918067504e-06, "loss": 0.13236874341964722, "step": 2350 }, { "epoch": 0.6103322949117341, "grad_norm": 0.6618941426277161, "learning_rate": 4.165284595843135e-06, "loss": 0.12346188724040985, "step": 2351 }, { "epoch": 0.6105919003115264, "grad_norm": 1.4600178003311157, "learning_rate": 4.1645490496896085e-06, "loss": 0.11833281815052032, "step": 2352 }, { "epoch": 0.6108515057113187, "grad_norm": 1.5540369749069214, "learning_rate": 4.1638132534762e-06, "loss": 0.16452091932296753, "step": 2353 }, { "epoch": 0.6111111111111112, "grad_norm": 0.647371232509613, "learning_rate": 4.163077207332979e-06, "loss": 0.15669944882392883, "step": 2354 }, { "epoch": 0.6113707165109035, "grad_norm": 0.8195186853408813, "learning_rate": 4.1623409113900636e-06, "loss": 0.1465873420238495, "step": 2355 }, { "epoch": 0.6116303219106958, "grad_norm": 1.2577441930770874, "learning_rate": 4.161604365777613e-06, "loss": 0.12416893243789673, "step": 2356 }, { "epoch": 0.6118899273104881, "grad_norm": 1.0548344850540161, "learning_rate": 4.160867570625833e-06, "loss": 0.1265028864145279, "step": 2357 }, { "epoch": 0.6121495327102804, "grad_norm": 0.5219338536262512, "learning_rate": 4.16013052606497e-06, "loss": 0.12745779752731323, "step": 2358 }, { "epoch": 0.6124091381100727, "grad_norm": 0.8616567850112915, "learning_rate": 4.159393232225318e-06, "loss": 0.14264105260372162, "step": 2359 }, { "epoch": 0.612668743509865, "grad_norm": 1.0486904382705688, "learning_rate": 4.1586556892372135e-06, "loss": 0.1593325436115265, "step": 2360 }, { "epoch": 0.6129283489096573, "grad_norm": 1.6920626163482666, "learning_rate": 4.157917897231037e-06, "loss": 0.12447264045476913, "step": 2361 }, { "epoch": 0.6131879543094496, "grad_norm": 1.0277791023254395, "learning_rate": 4.157179856337214e-06, "loss": 0.16721796989440918, "step": 2362 }, { "epoch": 0.6134475597092419, "grad_norm": 1.1292859315872192, "learning_rate": 4.156441566686212e-06, "loss": 0.14915917813777924, "step": 2363 }, { "epoch": 0.6137071651090342, "grad_norm": 0.8155049085617065, "learning_rate": 4.155703028408544e-06, "loss": 0.17529632151126862, "step": 2364 }, { "epoch": 0.6139667705088265, "grad_norm": 0.7785475254058838, "learning_rate": 4.1549642416347676e-06, "loss": 0.15110820531845093, "step": 2365 }, { "epoch": 0.614226375908619, "grad_norm": 0.4692411720752716, "learning_rate": 4.154225206495482e-06, "loss": 0.13158972561359406, "step": 2366 }, { "epoch": 0.6144859813084113, "grad_norm": 0.7551090121269226, "learning_rate": 4.153485923121332e-06, "loss": 0.14081710577011108, "step": 2367 }, { "epoch": 0.6147455867082036, "grad_norm": 2.3930606842041016, "learning_rate": 4.152746391643008e-06, "loss": 0.17286145687103271, "step": 2368 }, { "epoch": 0.6150051921079959, "grad_norm": 1.3359582424163818, "learning_rate": 4.152006612191239e-06, "loss": 0.14626725018024445, "step": 2369 }, { "epoch": 0.6152647975077882, "grad_norm": 1.280137538909912, "learning_rate": 4.151266584896803e-06, "loss": 0.15118691325187683, "step": 2370 }, { "epoch": 0.6155244029075805, "grad_norm": 0.7643393278121948, "learning_rate": 4.1505263098905196e-06, "loss": 0.12614652514457703, "step": 2371 }, { "epoch": 0.6157840083073728, "grad_norm": 1.1543556451797485, "learning_rate": 4.149785787303253e-06, "loss": 0.13121403753757477, "step": 2372 }, { "epoch": 0.6160436137071651, "grad_norm": 1.6693905591964722, "learning_rate": 4.14904501726591e-06, "loss": 0.1425570547580719, "step": 2373 }, { "epoch": 0.6163032191069574, "grad_norm": 2.154991626739502, "learning_rate": 4.148303999909441e-06, "loss": 0.14928743243217468, "step": 2374 }, { "epoch": 0.6165628245067497, "grad_norm": 1.547340750694275, "learning_rate": 4.147562735364842e-06, "loss": 0.13991546630859375, "step": 2375 }, { "epoch": 0.616822429906542, "grad_norm": 1.2738457918167114, "learning_rate": 4.146821223763151e-06, "loss": 0.13487675786018372, "step": 2376 }, { "epoch": 0.6170820353063343, "grad_norm": 1.1942700147628784, "learning_rate": 4.146079465235452e-06, "loss": 0.11645746231079102, "step": 2377 }, { "epoch": 0.6173416407061267, "grad_norm": 1.0279765129089355, "learning_rate": 4.145337459912868e-06, "loss": 0.13689181208610535, "step": 2378 }, { "epoch": 0.617601246105919, "grad_norm": 0.7047700881958008, "learning_rate": 4.144595207926571e-06, "loss": 0.11190681904554367, "step": 2379 }, { "epoch": 0.6178608515057114, "grad_norm": 0.8530855178833008, "learning_rate": 4.143852709407773e-06, "loss": 0.11272202432155609, "step": 2380 }, { "epoch": 0.6181204569055037, "grad_norm": 1.0260018110275269, "learning_rate": 4.143109964487731e-06, "loss": 0.13121581077575684, "step": 2381 }, { "epoch": 0.618380062305296, "grad_norm": 0.8776349425315857, "learning_rate": 4.1423669732977455e-06, "loss": 0.11349792778491974, "step": 2382 }, { "epoch": 0.6186396677050883, "grad_norm": 1.182934284210205, "learning_rate": 4.1416237359691595e-06, "loss": 0.1374327838420868, "step": 2383 }, { "epoch": 0.6188992731048806, "grad_norm": 1.289018154144287, "learning_rate": 4.140880252633362e-06, "loss": 0.16151070594787598, "step": 2384 }, { "epoch": 0.6191588785046729, "grad_norm": 3.4829916954040527, "learning_rate": 4.140136523421782e-06, "loss": 0.18052734434604645, "step": 2385 }, { "epoch": 0.6194184839044652, "grad_norm": 0.9971153140068054, "learning_rate": 4.139392548465895e-06, "loss": 0.1265297383069992, "step": 2386 }, { "epoch": 0.6196780893042575, "grad_norm": 1.6930943727493286, "learning_rate": 4.138648327897216e-06, "loss": 0.12377292662858963, "step": 2387 }, { "epoch": 0.6199376947040498, "grad_norm": 0.8604514598846436, "learning_rate": 4.137903861847311e-06, "loss": 0.1183759942650795, "step": 2388 }, { "epoch": 0.6201973001038421, "grad_norm": 1.0200510025024414, "learning_rate": 4.137159150447782e-06, "loss": 0.12720713019371033, "step": 2389 }, { "epoch": 0.6204569055036345, "grad_norm": 1.597255825996399, "learning_rate": 4.136414193830276e-06, "loss": 0.16829553246498108, "step": 2390 }, { "epoch": 0.6207165109034268, "grad_norm": 2.1428635120391846, "learning_rate": 4.135668992126485e-06, "loss": 0.1605052500963211, "step": 2391 }, { "epoch": 0.6209761163032191, "grad_norm": 1.278432011604309, "learning_rate": 4.134923545468144e-06, "loss": 0.15857918560504913, "step": 2392 }, { "epoch": 0.6212357217030114, "grad_norm": 1.1815792322158813, "learning_rate": 4.134177853987032e-06, "loss": 0.1283862143754959, "step": 2393 }, { "epoch": 0.6214953271028038, "grad_norm": 1.0022767782211304, "learning_rate": 4.133431917814967e-06, "loss": 0.12799450755119324, "step": 2394 }, { "epoch": 0.6217549325025961, "grad_norm": 1.218098521232605, "learning_rate": 4.132685737083817e-06, "loss": 0.1239372193813324, "step": 2395 }, { "epoch": 0.6220145379023884, "grad_norm": 1.019478440284729, "learning_rate": 4.131939311925486e-06, "loss": 0.13214674592018127, "step": 2396 }, { "epoch": 0.6222741433021807, "grad_norm": 0.7960788011550903, "learning_rate": 4.131192642471929e-06, "loss": 0.1369401216506958, "step": 2397 }, { "epoch": 0.622533748701973, "grad_norm": 1.0412098169326782, "learning_rate": 4.130445728855137e-06, "loss": 0.1283097267150879, "step": 2398 }, { "epoch": 0.6227933541017653, "grad_norm": 1.782652735710144, "learning_rate": 4.129698571207148e-06, "loss": 0.13377927243709564, "step": 2399 }, { "epoch": 0.6230529595015576, "grad_norm": 1.3065255880355835, "learning_rate": 4.128951169660043e-06, "loss": 0.13514584302902222, "step": 2400 }, { "epoch": 0.6233125649013499, "grad_norm": 1.1144713163375854, "learning_rate": 4.1282035243459446e-06, "loss": 0.14977344870567322, "step": 2401 }, { "epoch": 0.6235721703011423, "grad_norm": 1.4855449199676514, "learning_rate": 4.1274556353970196e-06, "loss": 0.1451391875743866, "step": 2402 }, { "epoch": 0.6238317757009346, "grad_norm": 1.3382623195648193, "learning_rate": 4.126707502945478e-06, "loss": 0.1437869668006897, "step": 2403 }, { "epoch": 0.6240913811007269, "grad_norm": 2.607598304748535, "learning_rate": 4.125959127123572e-06, "loss": 0.14138878881931305, "step": 2404 }, { "epoch": 0.6243509865005192, "grad_norm": 1.8267618417739868, "learning_rate": 4.125210508063599e-06, "loss": 0.1291561722755432, "step": 2405 }, { "epoch": 0.6246105919003115, "grad_norm": 0.711517333984375, "learning_rate": 4.1244616458978945e-06, "loss": 0.11853916943073273, "step": 2406 }, { "epoch": 0.6248701973001038, "grad_norm": 0.6620626449584961, "learning_rate": 4.1237125407588434e-06, "loss": 0.11830247938632965, "step": 2407 }, { "epoch": 0.6251298026998962, "grad_norm": 1.6307579278945923, "learning_rate": 4.1229631927788684e-06, "loss": 0.15684479475021362, "step": 2408 }, { "epoch": 0.6253894080996885, "grad_norm": 1.7517414093017578, "learning_rate": 4.122213602090438e-06, "loss": 0.16057616472244263, "step": 2409 }, { "epoch": 0.6256490134994808, "grad_norm": 1.048628330230713, "learning_rate": 4.121463768826061e-06, "loss": 0.1261781007051468, "step": 2410 }, { "epoch": 0.6259086188992731, "grad_norm": 2.696665048599243, "learning_rate": 4.120713693118293e-06, "loss": 0.17046961188316345, "step": 2411 }, { "epoch": 0.6261682242990654, "grad_norm": 0.8574861884117126, "learning_rate": 4.119963375099729e-06, "loss": 0.16725145280361176, "step": 2412 }, { "epoch": 0.6264278296988577, "grad_norm": 0.8462238907814026, "learning_rate": 4.119212814903009e-06, "loss": 0.15076497197151184, "step": 2413 }, { "epoch": 0.6266874350986501, "grad_norm": 0.5224396586418152, "learning_rate": 4.118462012660813e-06, "loss": 0.12132604420185089, "step": 2414 }, { "epoch": 0.6269470404984424, "grad_norm": 1.740088701248169, "learning_rate": 4.117710968505867e-06, "loss": 0.16681379079818726, "step": 2415 }, { "epoch": 0.6272066458982347, "grad_norm": 1.6184334754943848, "learning_rate": 4.116959682570938e-06, "loss": 0.15407058596611023, "step": 2416 }, { "epoch": 0.627466251298027, "grad_norm": 1.5199368000030518, "learning_rate": 4.1162081549888375e-06, "loss": 0.15501669049263, "step": 2417 }, { "epoch": 0.6277258566978193, "grad_norm": 1.3327633142471313, "learning_rate": 4.1154563858924164e-06, "loss": 0.12454606592655182, "step": 2418 }, { "epoch": 0.6279854620976116, "grad_norm": 2.2896993160247803, "learning_rate": 4.114704375414572e-06, "loss": 0.15691159665584564, "step": 2419 }, { "epoch": 0.6282450674974039, "grad_norm": 1.3544197082519531, "learning_rate": 4.1139521236882405e-06, "loss": 0.14904794096946716, "step": 2420 }, { "epoch": 0.6285046728971962, "grad_norm": 1.47916579246521, "learning_rate": 4.113199630846405e-06, "loss": 0.14229825139045715, "step": 2421 }, { "epoch": 0.6287642782969886, "grad_norm": 0.9589843153953552, "learning_rate": 4.112446897022086e-06, "loss": 0.12647366523742676, "step": 2422 }, { "epoch": 0.6290238836967809, "grad_norm": 3.0045387744903564, "learning_rate": 4.111693922348353e-06, "loss": 0.1673937737941742, "step": 2423 }, { "epoch": 0.6292834890965732, "grad_norm": 1.3983818292617798, "learning_rate": 4.110940706958313e-06, "loss": 0.1611328423023224, "step": 2424 }, { "epoch": 0.6295430944963655, "grad_norm": 0.8691693544387817, "learning_rate": 4.110187250985117e-06, "loss": 0.12124001979827881, "step": 2425 }, { "epoch": 0.6298026998961579, "grad_norm": 0.68727707862854, "learning_rate": 4.10943355456196e-06, "loss": 0.14165039360523224, "step": 2426 }, { "epoch": 0.6300623052959502, "grad_norm": 1.0786601305007935, "learning_rate": 4.108679617822077e-06, "loss": 0.13455963134765625, "step": 2427 }, { "epoch": 0.6303219106957425, "grad_norm": 1.5668730735778809, "learning_rate": 4.107925440898747e-06, "loss": 0.1314702033996582, "step": 2428 }, { "epoch": 0.6305815160955348, "grad_norm": 1.6533297300338745, "learning_rate": 4.107171023925292e-06, "loss": 0.1679115742444992, "step": 2429 }, { "epoch": 0.6308411214953271, "grad_norm": 1.1302902698516846, "learning_rate": 4.106416367035075e-06, "loss": 0.13754862546920776, "step": 2430 }, { "epoch": 0.6311007268951194, "grad_norm": 1.322767734527588, "learning_rate": 4.1056614703615015e-06, "loss": 0.17336797714233398, "step": 2431 }, { "epoch": 0.6313603322949117, "grad_norm": 1.4013197422027588, "learning_rate": 4.1049063340380205e-06, "loss": 0.16835367679595947, "step": 2432 }, { "epoch": 0.631619937694704, "grad_norm": 1.7652779817581177, "learning_rate": 4.104150958198123e-06, "loss": 0.12751322984695435, "step": 2433 }, { "epoch": 0.6318795430944963, "grad_norm": 1.0656960010528564, "learning_rate": 4.103395342975343e-06, "loss": 0.1585533767938614, "step": 2434 }, { "epoch": 0.6321391484942886, "grad_norm": 1.13642418384552, "learning_rate": 4.102639488503254e-06, "loss": 0.1225346177816391, "step": 2435 }, { "epoch": 0.632398753894081, "grad_norm": 1.032214879989624, "learning_rate": 4.101883394915475e-06, "loss": 0.12619081139564514, "step": 2436 }, { "epoch": 0.6326583592938733, "grad_norm": 1.2905927896499634, "learning_rate": 4.101127062345666e-06, "loss": 0.16597962379455566, "step": 2437 }, { "epoch": 0.6329179646936657, "grad_norm": 0.923974871635437, "learning_rate": 4.100370490927529e-06, "loss": 0.15122823417186737, "step": 2438 }, { "epoch": 0.633177570093458, "grad_norm": 0.7862182855606079, "learning_rate": 4.0996136807948085e-06, "loss": 0.12382400035858154, "step": 2439 }, { "epoch": 0.6334371754932503, "grad_norm": 0.7608806490898132, "learning_rate": 4.0988566320812915e-06, "loss": 0.14456629753112793, "step": 2440 }, { "epoch": 0.6336967808930426, "grad_norm": 0.4471298158168793, "learning_rate": 4.098099344920806e-06, "loss": 0.11829574406147003, "step": 2441 }, { "epoch": 0.6339563862928349, "grad_norm": 1.0193829536437988, "learning_rate": 4.097341819447224e-06, "loss": 0.15543395280838013, "step": 2442 }, { "epoch": 0.6342159916926272, "grad_norm": 1.058858036994934, "learning_rate": 4.096584055794459e-06, "loss": 0.1355583667755127, "step": 2443 }, { "epoch": 0.6344755970924195, "grad_norm": 1.3999849557876587, "learning_rate": 4.095826054096465e-06, "loss": 0.16799567639827728, "step": 2444 }, { "epoch": 0.6347352024922118, "grad_norm": 1.7614291906356812, "learning_rate": 4.09506781448724e-06, "loss": 0.17571821808815002, "step": 2445 }, { "epoch": 0.6349948078920041, "grad_norm": 1.292887806892395, "learning_rate": 4.094309337100823e-06, "loss": 0.15030863881111145, "step": 2446 }, { "epoch": 0.6352544132917964, "grad_norm": 1.09231436252594, "learning_rate": 4.093550622071296e-06, "loss": 0.11805649846792221, "step": 2447 }, { "epoch": 0.6355140186915887, "grad_norm": 1.3985978364944458, "learning_rate": 4.092791669532784e-06, "loss": 0.14873480796813965, "step": 2448 }, { "epoch": 0.635773624091381, "grad_norm": 0.7091554403305054, "learning_rate": 4.0920324796194485e-06, "loss": 0.13252893090248108, "step": 2449 }, { "epoch": 0.6360332294911735, "grad_norm": 0.9374380111694336, "learning_rate": 4.091273052465501e-06, "loss": 0.14215010404586792, "step": 2450 }, { "epoch": 0.6362928348909658, "grad_norm": 0.9474545121192932, "learning_rate": 4.090513388205187e-06, "loss": 0.12809813022613525, "step": 2451 }, { "epoch": 0.6365524402907581, "grad_norm": 1.080925703048706, "learning_rate": 4.0897534869728e-06, "loss": 0.13813915848731995, "step": 2452 }, { "epoch": 0.6368120456905504, "grad_norm": 1.2393920421600342, "learning_rate": 4.088993348902673e-06, "loss": 0.16167035698890686, "step": 2453 }, { "epoch": 0.6370716510903427, "grad_norm": 1.2565844058990479, "learning_rate": 4.0882329741291815e-06, "loss": 0.16580846905708313, "step": 2454 }, { "epoch": 0.637331256490135, "grad_norm": 1.3302838802337646, "learning_rate": 4.087472362786741e-06, "loss": 0.13702887296676636, "step": 2455 }, { "epoch": 0.6375908618899273, "grad_norm": 0.6065617203712463, "learning_rate": 4.086711515009811e-06, "loss": 0.13366535305976868, "step": 2456 }, { "epoch": 0.6378504672897196, "grad_norm": 0.4548128545284271, "learning_rate": 4.085950430932892e-06, "loss": 0.1446312665939331, "step": 2457 }, { "epoch": 0.6381100726895119, "grad_norm": 0.7890312075614929, "learning_rate": 4.085189110690526e-06, "loss": 0.1259925812482834, "step": 2458 }, { "epoch": 0.6383696780893042, "grad_norm": 1.0109062194824219, "learning_rate": 4.084427554417296e-06, "loss": 0.16015520691871643, "step": 2459 }, { "epoch": 0.6386292834890965, "grad_norm": 0.6085334420204163, "learning_rate": 4.08366576224783e-06, "loss": 0.12732966244220734, "step": 2460 }, { "epoch": 0.6388888888888888, "grad_norm": 1.2553062438964844, "learning_rate": 4.082903734316793e-06, "loss": 0.16045457124710083, "step": 2461 }, { "epoch": 0.6391484942886813, "grad_norm": 1.0774438381195068, "learning_rate": 4.082141470758896e-06, "loss": 0.1590103805065155, "step": 2462 }, { "epoch": 0.6394080996884736, "grad_norm": 1.3018451929092407, "learning_rate": 4.081378971708889e-06, "loss": 0.1655876785516739, "step": 2463 }, { "epoch": 0.6396677050882659, "grad_norm": 1.63238525390625, "learning_rate": 4.080616237301563e-06, "loss": 0.15468689799308777, "step": 2464 }, { "epoch": 0.6399273104880582, "grad_norm": 0.7043606638908386, "learning_rate": 4.079853267671755e-06, "loss": 0.11445482075214386, "step": 2465 }, { "epoch": 0.6401869158878505, "grad_norm": 1.8864723443984985, "learning_rate": 4.079090062954339e-06, "loss": 0.14223287999629974, "step": 2466 }, { "epoch": 0.6404465212876428, "grad_norm": 1.303972840309143, "learning_rate": 4.078326623284231e-06, "loss": 0.1384502351284027, "step": 2467 }, { "epoch": 0.6407061266874351, "grad_norm": 1.0610820055007935, "learning_rate": 4.077562948796391e-06, "loss": 0.1374143809080124, "step": 2468 }, { "epoch": 0.6409657320872274, "grad_norm": 2.1016042232513428, "learning_rate": 4.076799039625819e-06, "loss": 0.12748490273952484, "step": 2469 }, { "epoch": 0.6412253374870197, "grad_norm": 1.5806612968444824, "learning_rate": 4.076034895907556e-06, "loss": 0.15686015784740448, "step": 2470 }, { "epoch": 0.641484942886812, "grad_norm": 1.5548059940338135, "learning_rate": 4.075270517776686e-06, "loss": 0.13262799382209778, "step": 2471 }, { "epoch": 0.6417445482866043, "grad_norm": 1.4689847230911255, "learning_rate": 4.0745059053683325e-06, "loss": 0.13061299920082092, "step": 2472 }, { "epoch": 0.6420041536863966, "grad_norm": 0.5433465838432312, "learning_rate": 4.073741058817663e-06, "loss": 0.12582439184188843, "step": 2473 }, { "epoch": 0.642263759086189, "grad_norm": 0.620755136013031, "learning_rate": 4.072975978259883e-06, "loss": 0.12043818831443787, "step": 2474 }, { "epoch": 0.6425233644859814, "grad_norm": 1.6254082918167114, "learning_rate": 4.072210663830243e-06, "loss": 0.1292869597673416, "step": 2475 }, { "epoch": 0.6427829698857737, "grad_norm": 0.8437191843986511, "learning_rate": 4.0714451156640314e-06, "loss": 0.1521489918231964, "step": 2476 }, { "epoch": 0.643042575285566, "grad_norm": 0.4940001666545868, "learning_rate": 4.070679333896582e-06, "loss": 0.1302942931652069, "step": 2477 }, { "epoch": 0.6433021806853583, "grad_norm": 1.1406540870666504, "learning_rate": 4.069913318663264e-06, "loss": 0.14440876245498657, "step": 2478 }, { "epoch": 0.6435617860851506, "grad_norm": 0.9811059236526489, "learning_rate": 4.069147070099495e-06, "loss": 0.13156822323799133, "step": 2479 }, { "epoch": 0.6438213914849429, "grad_norm": 1.311179757118225, "learning_rate": 4.068380588340728e-06, "loss": 0.12119528651237488, "step": 2480 }, { "epoch": 0.6440809968847352, "grad_norm": 0.5286730527877808, "learning_rate": 4.067613873522459e-06, "loss": 0.12554678320884705, "step": 2481 }, { "epoch": 0.6443406022845275, "grad_norm": 2.276003837585449, "learning_rate": 4.066846925780226e-06, "loss": 0.15075460076332092, "step": 2482 }, { "epoch": 0.6446002076843198, "grad_norm": 1.0044487714767456, "learning_rate": 4.06607974524961e-06, "loss": 0.1281050145626068, "step": 2483 }, { "epoch": 0.6448598130841121, "grad_norm": 0.6408442854881287, "learning_rate": 4.065312332066229e-06, "loss": 0.14362095296382904, "step": 2484 }, { "epoch": 0.6451194184839044, "grad_norm": 0.586118221282959, "learning_rate": 4.064544686365744e-06, "loss": 0.12315832823514938, "step": 2485 }, { "epoch": 0.6453790238836968, "grad_norm": 1.2916768789291382, "learning_rate": 4.063776808283857e-06, "loss": 0.14601752161979675, "step": 2486 }, { "epoch": 0.6456386292834891, "grad_norm": 0.6053633093833923, "learning_rate": 4.0630086979563125e-06, "loss": 0.11539329588413239, "step": 2487 }, { "epoch": 0.6458982346832814, "grad_norm": 0.35566526651382446, "learning_rate": 4.062240355518895e-06, "loss": 0.12083128094673157, "step": 2488 }, { "epoch": 0.6461578400830738, "grad_norm": 0.9589380621910095, "learning_rate": 4.061471781107427e-06, "loss": 0.12567664682865143, "step": 2489 }, { "epoch": 0.6464174454828661, "grad_norm": 1.7754923105239868, "learning_rate": 4.060702974857778e-06, "loss": 0.17268884181976318, "step": 2490 }, { "epoch": 0.6466770508826584, "grad_norm": 1.7723568677902222, "learning_rate": 4.059933936905854e-06, "loss": 0.13307419419288635, "step": 2491 }, { "epoch": 0.6469366562824507, "grad_norm": 0.7687461972236633, "learning_rate": 4.059164667387603e-06, "loss": 0.13331621885299683, "step": 2492 }, { "epoch": 0.647196261682243, "grad_norm": 1.4264761209487915, "learning_rate": 4.058395166439015e-06, "loss": 0.1375953108072281, "step": 2493 }, { "epoch": 0.6474558670820353, "grad_norm": 0.8462773561477661, "learning_rate": 4.05762543419612e-06, "loss": 0.16933803260326385, "step": 2494 }, { "epoch": 0.6477154724818276, "grad_norm": 1.7627233266830444, "learning_rate": 4.056855470794988e-06, "loss": 0.14213871955871582, "step": 2495 }, { "epoch": 0.6479750778816199, "grad_norm": 1.2804900407791138, "learning_rate": 4.0560852763717335e-06, "loss": 0.1182442158460617, "step": 2496 }, { "epoch": 0.6482346832814122, "grad_norm": 0.6187120676040649, "learning_rate": 4.0553148510625065e-06, "loss": 0.11204996705055237, "step": 2497 }, { "epoch": 0.6484942886812046, "grad_norm": 1.1962614059448242, "learning_rate": 4.054544195003502e-06, "loss": 0.11903320997953415, "step": 2498 }, { "epoch": 0.6487538940809969, "grad_norm": 0.848578691482544, "learning_rate": 4.053773308330953e-06, "loss": 0.14100457727909088, "step": 2499 }, { "epoch": 0.6490134994807892, "grad_norm": 0.8900125622749329, "learning_rate": 4.053002191181137e-06, "loss": 0.1191290020942688, "step": 2500 }, { "epoch": 0.6492731048805815, "grad_norm": 1.1675684452056885, "learning_rate": 4.052230843690367e-06, "loss": 0.15693721175193787, "step": 2501 }, { "epoch": 0.6495327102803738, "grad_norm": 1.539891242980957, "learning_rate": 4.051459265995001e-06, "loss": 0.17080152034759521, "step": 2502 }, { "epoch": 0.6497923156801662, "grad_norm": 1.0870559215545654, "learning_rate": 4.050687458231437e-06, "loss": 0.1120421513915062, "step": 2503 }, { "epoch": 0.6500519210799585, "grad_norm": 0.6832218170166016, "learning_rate": 4.049915420536112e-06, "loss": 0.11320368200540543, "step": 2504 }, { "epoch": 0.6503115264797508, "grad_norm": 0.6968546509742737, "learning_rate": 4.049143153045505e-06, "loss": 0.11858789622783661, "step": 2505 }, { "epoch": 0.6505711318795431, "grad_norm": 0.650597870349884, "learning_rate": 4.048370655896135e-06, "loss": 0.11334012448787689, "step": 2506 }, { "epoch": 0.6508307372793354, "grad_norm": 0.4953792095184326, "learning_rate": 4.047597929224561e-06, "loss": 0.1303212195634842, "step": 2507 }, { "epoch": 0.6510903426791277, "grad_norm": 0.8071627020835876, "learning_rate": 4.0468249731673845e-06, "loss": 0.1453877091407776, "step": 2508 }, { "epoch": 0.65134994807892, "grad_norm": 1.1724870204925537, "learning_rate": 4.046051787861247e-06, "loss": 0.17132678627967834, "step": 2509 }, { "epoch": 0.6516095534787124, "grad_norm": 0.4639494717121124, "learning_rate": 4.045278373442828e-06, "loss": 0.12605267763137817, "step": 2510 }, { "epoch": 0.6518691588785047, "grad_norm": 1.1190593242645264, "learning_rate": 4.04450473004885e-06, "loss": 0.1191617101430893, "step": 2511 }, { "epoch": 0.652128764278297, "grad_norm": 1.2831356525421143, "learning_rate": 4.043730857816077e-06, "loss": 0.17143994569778442, "step": 2512 }, { "epoch": 0.6523883696780893, "grad_norm": 0.6122549176216125, "learning_rate": 4.042956756881311e-06, "loss": 0.1270160973072052, "step": 2513 }, { "epoch": 0.6526479750778816, "grad_norm": 1.306437373161316, "learning_rate": 4.042182427381395e-06, "loss": 0.15049664676189423, "step": 2514 }, { "epoch": 0.6529075804776739, "grad_norm": 0.9033039808273315, "learning_rate": 4.041407869453212e-06, "loss": 0.14901390671730042, "step": 2515 }, { "epoch": 0.6531671858774662, "grad_norm": 0.968703031539917, "learning_rate": 4.0406330832336885e-06, "loss": 0.1403176486492157, "step": 2516 }, { "epoch": 0.6534267912772586, "grad_norm": 1.301203966140747, "learning_rate": 4.039858068859787e-06, "loss": 0.14344990253448486, "step": 2517 }, { "epoch": 0.6536863966770509, "grad_norm": 0.5045216083526611, "learning_rate": 4.039082826468511e-06, "loss": 0.12506048381328583, "step": 2518 }, { "epoch": 0.6539460020768432, "grad_norm": 0.5407527685165405, "learning_rate": 4.03830735619691e-06, "loss": 0.1237538605928421, "step": 2519 }, { "epoch": 0.6542056074766355, "grad_norm": 1.1670596599578857, "learning_rate": 4.037531658182065e-06, "loss": 0.13022716343402863, "step": 2520 }, { "epoch": 0.6544652128764278, "grad_norm": 1.644808292388916, "learning_rate": 4.036755732561103e-06, "loss": 0.12961120903491974, "step": 2521 }, { "epoch": 0.6547248182762202, "grad_norm": 0.9018443822860718, "learning_rate": 4.035979579471192e-06, "loss": 0.14222615957260132, "step": 2522 }, { "epoch": 0.6549844236760125, "grad_norm": 0.8138481974601746, "learning_rate": 4.035203199049536e-06, "loss": 0.1113336980342865, "step": 2523 }, { "epoch": 0.6552440290758048, "grad_norm": 0.7388895153999329, "learning_rate": 4.0344265914333805e-06, "loss": 0.10836488008499146, "step": 2524 }, { "epoch": 0.6555036344755971, "grad_norm": 4.146419048309326, "learning_rate": 4.033649756760014e-06, "loss": 0.13306546211242676, "step": 2525 }, { "epoch": 0.6557632398753894, "grad_norm": 0.8807663917541504, "learning_rate": 4.032872695166763e-06, "loss": 0.15916089713573456, "step": 2526 }, { "epoch": 0.6560228452751817, "grad_norm": 0.3658001124858856, "learning_rate": 4.032095406790992e-06, "loss": 0.1192704290151596, "step": 2527 }, { "epoch": 0.656282450674974, "grad_norm": 1.7006670236587524, "learning_rate": 4.031317891770109e-06, "loss": 0.16737323999404907, "step": 2528 }, { "epoch": 0.6565420560747663, "grad_norm": 0.8724948763847351, "learning_rate": 4.030540150241563e-06, "loss": 0.1469338834285736, "step": 2529 }, { "epoch": 0.6568016614745587, "grad_norm": 0.42785391211509705, "learning_rate": 4.029762182342836e-06, "loss": 0.12078619748353958, "step": 2530 }, { "epoch": 0.657061266874351, "grad_norm": 0.858973503112793, "learning_rate": 4.02898398821146e-06, "loss": 0.12957845628261566, "step": 2531 }, { "epoch": 0.6573208722741433, "grad_norm": 0.9466069340705872, "learning_rate": 4.028205567984999e-06, "loss": 0.14932388067245483, "step": 2532 }, { "epoch": 0.6575804776739356, "grad_norm": 1.01719331741333, "learning_rate": 4.0274269218010606e-06, "loss": 0.13749441504478455, "step": 2533 }, { "epoch": 0.657840083073728, "grad_norm": 1.2087637186050415, "learning_rate": 4.026648049797292e-06, "loss": 0.12141121923923492, "step": 2534 }, { "epoch": 0.6580996884735203, "grad_norm": 1.3893554210662842, "learning_rate": 4.025868952111379e-06, "loss": 0.16356316208839417, "step": 2535 }, { "epoch": 0.6583592938733126, "grad_norm": 1.211998462677002, "learning_rate": 4.025089628881049e-06, "loss": 0.1312839686870575, "step": 2536 }, { "epoch": 0.6586188992731049, "grad_norm": 0.7694472074508667, "learning_rate": 4.024310080244069e-06, "loss": 0.14894434809684753, "step": 2537 }, { "epoch": 0.6588785046728972, "grad_norm": 2.203134059906006, "learning_rate": 4.023530306338244e-06, "loss": 0.17795145511627197, "step": 2538 }, { "epoch": 0.6591381100726895, "grad_norm": 2.249614715576172, "learning_rate": 4.02275030730142e-06, "loss": 0.16513916850090027, "step": 2539 }, { "epoch": 0.6593977154724818, "grad_norm": 1.431992530822754, "learning_rate": 4.021970083271486e-06, "loss": 0.12272194027900696, "step": 2540 }, { "epoch": 0.6596573208722741, "grad_norm": 0.8785675764083862, "learning_rate": 4.021189634386363e-06, "loss": 0.11953286826610565, "step": 2541 }, { "epoch": 0.6599169262720664, "grad_norm": 1.6956549882888794, "learning_rate": 4.0204089607840204e-06, "loss": 0.15927693247795105, "step": 2542 }, { "epoch": 0.6601765316718587, "grad_norm": 0.8102462887763977, "learning_rate": 4.019628062602462e-06, "loss": 0.12899526953697205, "step": 2543 }, { "epoch": 0.660436137071651, "grad_norm": 0.8229552507400513, "learning_rate": 4.018846939979732e-06, "loss": 0.12573440372943878, "step": 2544 }, { "epoch": 0.6606957424714434, "grad_norm": 0.5108199119567871, "learning_rate": 4.018065593053917e-06, "loss": 0.1211048811674118, "step": 2545 }, { "epoch": 0.6609553478712358, "grad_norm": 1.4099476337432861, "learning_rate": 4.01728402196314e-06, "loss": 0.13538768887519836, "step": 2546 }, { "epoch": 0.6612149532710281, "grad_norm": 1.0252217054367065, "learning_rate": 4.016502226845564e-06, "loss": 0.14658866822719574, "step": 2547 }, { "epoch": 0.6614745586708204, "grad_norm": 0.679469883441925, "learning_rate": 4.015720207839394e-06, "loss": 0.11294043064117432, "step": 2548 }, { "epoch": 0.6617341640706127, "grad_norm": 0.6988166570663452, "learning_rate": 4.014937965082871e-06, "loss": 0.12915009260177612, "step": 2549 }, { "epoch": 0.661993769470405, "grad_norm": 1.225891351699829, "learning_rate": 4.014155498714281e-06, "loss": 0.12096970528364182, "step": 2550 }, { "epoch": 0.6622533748701973, "grad_norm": 1.1918444633483887, "learning_rate": 4.0133728088719425e-06, "loss": 0.1199110895395279, "step": 2551 }, { "epoch": 0.6625129802699896, "grad_norm": 1.037353754043579, "learning_rate": 4.0125898956942195e-06, "loss": 0.12478721141815186, "step": 2552 }, { "epoch": 0.6627725856697819, "grad_norm": 0.9685472846031189, "learning_rate": 4.011806759319513e-06, "loss": 0.13105037808418274, "step": 2553 }, { "epoch": 0.6630321910695742, "grad_norm": 0.724343478679657, "learning_rate": 4.011023399886262e-06, "loss": 0.13038352131843567, "step": 2554 }, { "epoch": 0.6632917964693665, "grad_norm": 0.8705444931983948, "learning_rate": 4.010239817532947e-06, "loss": 0.14851710200309753, "step": 2555 }, { "epoch": 0.6635514018691588, "grad_norm": 1.1843385696411133, "learning_rate": 4.009456012398089e-06, "loss": 0.1364021748304367, "step": 2556 }, { "epoch": 0.6638110072689511, "grad_norm": 1.141239047050476, "learning_rate": 4.008671984620246e-06, "loss": 0.1646416038274765, "step": 2557 }, { "epoch": 0.6640706126687436, "grad_norm": 1.1201528310775757, "learning_rate": 4.007887734338016e-06, "loss": 0.17547151446342468, "step": 2558 }, { "epoch": 0.6643302180685359, "grad_norm": 0.8861656188964844, "learning_rate": 4.007103261690035e-06, "loss": 0.14692482352256775, "step": 2559 }, { "epoch": 0.6645898234683282, "grad_norm": 0.9532508850097656, "learning_rate": 4.006318566814982e-06, "loss": 0.13035881519317627, "step": 2560 }, { "epoch": 0.6648494288681205, "grad_norm": 0.6614839434623718, "learning_rate": 4.005533649851573e-06, "loss": 0.10943343490362167, "step": 2561 }, { "epoch": 0.6651090342679128, "grad_norm": 1.7592698335647583, "learning_rate": 4.004748510938562e-06, "loss": 0.1688317060470581, "step": 2562 }, { "epoch": 0.6653686396677051, "grad_norm": 0.9341519474983215, "learning_rate": 4.003963150214743e-06, "loss": 0.16431596875190735, "step": 2563 }, { "epoch": 0.6656282450674974, "grad_norm": 0.8118551969528198, "learning_rate": 4.003177567818952e-06, "loss": 0.11566506326198578, "step": 2564 }, { "epoch": 0.6658878504672897, "grad_norm": 2.1150548458099365, "learning_rate": 4.002391763890062e-06, "loss": 0.14388948678970337, "step": 2565 }, { "epoch": 0.666147455867082, "grad_norm": 1.2519360780715942, "learning_rate": 4.001605738566983e-06, "loss": 0.14724557101726532, "step": 2566 }, { "epoch": 0.6664070612668743, "grad_norm": 0.9508949518203735, "learning_rate": 4.000819491988667e-06, "loss": 0.13985636830329895, "step": 2567 }, { "epoch": 0.6666666666666666, "grad_norm": 0.6038898825645447, "learning_rate": 4.000033024294105e-06, "loss": 0.11393657326698303, "step": 2568 }, { "epoch": 0.6669262720664589, "grad_norm": 0.6966007351875305, "learning_rate": 3.999246335622327e-06, "loss": 0.1307174563407898, "step": 2569 }, { "epoch": 0.6671858774662512, "grad_norm": 0.5729826092720032, "learning_rate": 3.9984594261124e-06, "loss": 0.12196571379899979, "step": 2570 }, { "epoch": 0.6674454828660437, "grad_norm": 0.6726158261299133, "learning_rate": 3.997672295903433e-06, "loss": 0.12470430135726929, "step": 2571 }, { "epoch": 0.667705088265836, "grad_norm": 0.9696387052536011, "learning_rate": 3.996884945134572e-06, "loss": 0.11993852257728577, "step": 2572 }, { "epoch": 0.6679646936656283, "grad_norm": 1.8777228593826294, "learning_rate": 3.996097373945003e-06, "loss": 0.14984369277954102, "step": 2573 }, { "epoch": 0.6682242990654206, "grad_norm": 1.2946815490722656, "learning_rate": 3.995309582473949e-06, "loss": 0.12614618241786957, "step": 2574 }, { "epoch": 0.6684839044652129, "grad_norm": 1.7375248670578003, "learning_rate": 3.994521570860676e-06, "loss": 0.15257570147514343, "step": 2575 }, { "epoch": 0.6687435098650052, "grad_norm": 0.4536024332046509, "learning_rate": 3.993733339244484e-06, "loss": 0.12384437024593353, "step": 2576 }, { "epoch": 0.6690031152647975, "grad_norm": 0.581375777721405, "learning_rate": 3.992944887764715e-06, "loss": 0.14819854497909546, "step": 2577 }, { "epoch": 0.6692627206645898, "grad_norm": 0.38886815309524536, "learning_rate": 3.9921562165607494e-06, "loss": 0.13254912197589874, "step": 2578 }, { "epoch": 0.6695223260643821, "grad_norm": 1.2359960079193115, "learning_rate": 3.991367325772007e-06, "loss": 0.11872677505016327, "step": 2579 }, { "epoch": 0.6697819314641744, "grad_norm": 1.2284737825393677, "learning_rate": 3.990578215537945e-06, "loss": 0.14034253358840942, "step": 2580 }, { "epoch": 0.6700415368639667, "grad_norm": 1.288191795349121, "learning_rate": 3.98978888599806e-06, "loss": 0.16564220190048218, "step": 2581 }, { "epoch": 0.670301142263759, "grad_norm": 0.9679012298583984, "learning_rate": 3.988999337291887e-06, "loss": 0.1319967806339264, "step": 2582 }, { "epoch": 0.6705607476635514, "grad_norm": 1.496110200881958, "learning_rate": 3.988209569559001e-06, "loss": 0.1374078243970871, "step": 2583 }, { "epoch": 0.6708203530633438, "grad_norm": 1.4122626781463623, "learning_rate": 3.987419582939014e-06, "loss": 0.12088017165660858, "step": 2584 }, { "epoch": 0.6710799584631361, "grad_norm": 1.233674168586731, "learning_rate": 3.986629377571579e-06, "loss": 0.14359894394874573, "step": 2585 }, { "epoch": 0.6713395638629284, "grad_norm": 1.9095158576965332, "learning_rate": 3.985838953596383e-06, "loss": 0.14996521174907684, "step": 2586 }, { "epoch": 0.6715991692627207, "grad_norm": 1.234321117401123, "learning_rate": 3.985048311153158e-06, "loss": 0.1707027554512024, "step": 2587 }, { "epoch": 0.671858774662513, "grad_norm": 1.096054196357727, "learning_rate": 3.98425745038167e-06, "loss": 0.16726823151111603, "step": 2588 }, { "epoch": 0.6721183800623053, "grad_norm": 1.3099734783172607, "learning_rate": 3.983466371421727e-06, "loss": 0.12841853499412537, "step": 2589 }, { "epoch": 0.6723779854620976, "grad_norm": 1.0783772468566895, "learning_rate": 3.982675074413171e-06, "loss": 0.11808285117149353, "step": 2590 }, { "epoch": 0.6726375908618899, "grad_norm": 1.3818403482437134, "learning_rate": 3.981883559495886e-06, "loss": 0.14956636726856232, "step": 2591 }, { "epoch": 0.6728971962616822, "grad_norm": 0.8392860293388367, "learning_rate": 3.9810918268097945e-06, "loss": 0.14635373651981354, "step": 2592 }, { "epoch": 0.6731568016614745, "grad_norm": 1.5611672401428223, "learning_rate": 3.980299876494857e-06, "loss": 0.15965503454208374, "step": 2593 }, { "epoch": 0.6734164070612668, "grad_norm": 0.9948474168777466, "learning_rate": 3.979507708691071e-06, "loss": 0.12245458364486694, "step": 2594 }, { "epoch": 0.6736760124610592, "grad_norm": 1.0900425910949707, "learning_rate": 3.978715323538474e-06, "loss": 0.12387025356292725, "step": 2595 }, { "epoch": 0.6739356178608515, "grad_norm": 0.9704186916351318, "learning_rate": 3.977922721177141e-06, "loss": 0.14529508352279663, "step": 2596 }, { "epoch": 0.6741952232606438, "grad_norm": 1.4637722969055176, "learning_rate": 3.977129901747187e-06, "loss": 0.14783445000648499, "step": 2597 }, { "epoch": 0.6744548286604362, "grad_norm": 0.4961097538471222, "learning_rate": 3.976336865388765e-06, "loss": 0.1335369497537613, "step": 2598 }, { "epoch": 0.6747144340602285, "grad_norm": 1.1583958864212036, "learning_rate": 3.975543612242063e-06, "loss": 0.12998397648334503, "step": 2599 }, { "epoch": 0.6749740394600208, "grad_norm": 0.9353237152099609, "learning_rate": 3.974750142447312e-06, "loss": 0.12450702488422394, "step": 2600 }, { "epoch": 0.6752336448598131, "grad_norm": 0.5624980330467224, "learning_rate": 3.97395645614478e-06, "loss": 0.10268670320510864, "step": 2601 }, { "epoch": 0.6754932502596054, "grad_norm": 0.8212575912475586, "learning_rate": 3.97316255347477e-06, "loss": 0.15542256832122803, "step": 2602 }, { "epoch": 0.6757528556593977, "grad_norm": 0.7621381878852844, "learning_rate": 3.972368434577628e-06, "loss": 0.14197662472724915, "step": 2603 }, { "epoch": 0.67601246105919, "grad_norm": 0.6412274837493896, "learning_rate": 3.971574099593735e-06, "loss": 0.11617112159729004, "step": 2604 }, { "epoch": 0.6762720664589823, "grad_norm": 0.542973518371582, "learning_rate": 3.970779548663512e-06, "loss": 0.1215914636850357, "step": 2605 }, { "epoch": 0.6765316718587746, "grad_norm": 1.3010973930358887, "learning_rate": 3.969984781927417e-06, "loss": 0.16432900726795197, "step": 2606 }, { "epoch": 0.676791277258567, "grad_norm": 1.2478528022766113, "learning_rate": 3.969189799525946e-06, "loss": 0.16481779515743256, "step": 2607 }, { "epoch": 0.6770508826583593, "grad_norm": 1.709912896156311, "learning_rate": 3.9683946015996335e-06, "loss": 0.1932959109544754, "step": 2608 }, { "epoch": 0.6773104880581516, "grad_norm": 1.0154411792755127, "learning_rate": 3.967599188289054e-06, "loss": 0.12273392826318741, "step": 2609 }, { "epoch": 0.677570093457944, "grad_norm": 0.6645529866218567, "learning_rate": 3.966803559734818e-06, "loss": 0.1443484127521515, "step": 2610 }, { "epoch": 0.6778296988577363, "grad_norm": 0.5865403413772583, "learning_rate": 3.966007716077572e-06, "loss": 0.14066338539123535, "step": 2611 }, { "epoch": 0.6780893042575286, "grad_norm": 2.056795597076416, "learning_rate": 3.965211657458006e-06, "loss": 0.15071231126785278, "step": 2612 }, { "epoch": 0.6783489096573209, "grad_norm": 0.7812544703483582, "learning_rate": 3.964415384016843e-06, "loss": 0.1529235541820526, "step": 2613 }, { "epoch": 0.6786085150571132, "grad_norm": 0.8567522764205933, "learning_rate": 3.963618895894847e-06, "loss": 0.12873393297195435, "step": 2614 }, { "epoch": 0.6788681204569055, "grad_norm": 0.8234334588050842, "learning_rate": 3.962822193232818e-06, "loss": 0.13768935203552246, "step": 2615 }, { "epoch": 0.6791277258566978, "grad_norm": 1.6348873376846313, "learning_rate": 3.9620252761715955e-06, "loss": 0.1650448441505432, "step": 2616 }, { "epoch": 0.6793873312564901, "grad_norm": 0.8116151690483093, "learning_rate": 3.961228144852056e-06, "loss": 0.1370278000831604, "step": 2617 }, { "epoch": 0.6796469366562824, "grad_norm": 0.5886943936347961, "learning_rate": 3.960430799415114e-06, "loss": 0.11847664415836334, "step": 2618 }, { "epoch": 0.6799065420560748, "grad_norm": 0.789974570274353, "learning_rate": 3.959633240001721e-06, "loss": 0.14605405926704407, "step": 2619 }, { "epoch": 0.6801661474558671, "grad_norm": 0.7287637591362, "learning_rate": 3.958835466752868e-06, "loss": 0.11813671886920929, "step": 2620 }, { "epoch": 0.6804257528556594, "grad_norm": 1.2326722145080566, "learning_rate": 3.958037479809583e-06, "loss": 0.14537525177001953, "step": 2621 }, { "epoch": 0.6806853582554517, "grad_norm": 1.8939851522445679, "learning_rate": 3.957239279312934e-06, "loss": 0.13202324509620667, "step": 2622 }, { "epoch": 0.680944963655244, "grad_norm": 0.6472837328910828, "learning_rate": 3.9564408654040206e-06, "loss": 0.11386342346668243, "step": 2623 }, { "epoch": 0.6812045690550363, "grad_norm": 0.4259259104728699, "learning_rate": 3.955642238223987e-06, "loss": 0.1229158341884613, "step": 2624 }, { "epoch": 0.6814641744548287, "grad_norm": 1.4155397415161133, "learning_rate": 3.954843397914011e-06, "loss": 0.11050251126289368, "step": 2625 }, { "epoch": 0.681723779854621, "grad_norm": 1.3715920448303223, "learning_rate": 3.954044344615309e-06, "loss": 0.15294508635997772, "step": 2626 }, { "epoch": 0.6819833852544133, "grad_norm": 0.6566516757011414, "learning_rate": 3.953245078469137e-06, "loss": 0.13553568720817566, "step": 2627 }, { "epoch": 0.6822429906542056, "grad_norm": 0.7692194581031799, "learning_rate": 3.952445599616784e-06, "loss": 0.11797890067100525, "step": 2628 }, { "epoch": 0.6825025960539979, "grad_norm": 0.5127258896827698, "learning_rate": 3.951645908199583e-06, "loss": 0.119974285364151, "step": 2629 }, { "epoch": 0.6827622014537902, "grad_norm": 0.665611207485199, "learning_rate": 3.950846004358899e-06, "loss": 0.12132702022790909, "step": 2630 }, { "epoch": 0.6830218068535826, "grad_norm": 0.8317548036575317, "learning_rate": 3.950045888236137e-06, "loss": 0.14942902326583862, "step": 2631 }, { "epoch": 0.6832814122533749, "grad_norm": 0.49711883068084717, "learning_rate": 3.949245559972739e-06, "loss": 0.13377593457698822, "step": 2632 }, { "epoch": 0.6835410176531672, "grad_norm": 0.46066856384277344, "learning_rate": 3.9484450197101856e-06, "loss": 0.11956271529197693, "step": 2633 }, { "epoch": 0.6838006230529595, "grad_norm": 1.3076937198638916, "learning_rate": 3.947644267589993e-06, "loss": 0.13033974170684814, "step": 2634 }, { "epoch": 0.6840602284527518, "grad_norm": 0.42160218954086304, "learning_rate": 3.946843303753716e-06, "loss": 0.12211185693740845, "step": 2635 }, { "epoch": 0.6843198338525441, "grad_norm": 0.5987022519111633, "learning_rate": 3.946042128342948e-06, "loss": 0.14501163363456726, "step": 2636 }, { "epoch": 0.6845794392523364, "grad_norm": 0.7386796474456787, "learning_rate": 3.945240741499317e-06, "loss": 0.1317647099494934, "step": 2637 }, { "epoch": 0.6848390446521287, "grad_norm": 1.316279649734497, "learning_rate": 3.94443914336449e-06, "loss": 0.1286362111568451, "step": 2638 }, { "epoch": 0.685098650051921, "grad_norm": 1.1500729322433472, "learning_rate": 3.9436373340801724e-06, "loss": 0.14314866065979004, "step": 2639 }, { "epoch": 0.6853582554517134, "grad_norm": 0.7913724780082703, "learning_rate": 3.9428353137881034e-06, "loss": 0.17588722705841064, "step": 2640 }, { "epoch": 0.6856178608515057, "grad_norm": 0.7022826671600342, "learning_rate": 3.942033082630064e-06, "loss": 0.12032610177993774, "step": 2641 }, { "epoch": 0.685877466251298, "grad_norm": 1.2523983716964722, "learning_rate": 3.94123064074787e-06, "loss": 0.12557952105998993, "step": 2642 }, { "epoch": 0.6861370716510904, "grad_norm": 1.0687021017074585, "learning_rate": 3.940427988283374e-06, "loss": 0.1340150386095047, "step": 2643 }, { "epoch": 0.6863966770508827, "grad_norm": 0.6986905336380005, "learning_rate": 3.939625125378467e-06, "loss": 0.123805932700634, "step": 2644 }, { "epoch": 0.686656282450675, "grad_norm": 1.3526945114135742, "learning_rate": 3.938822052175076e-06, "loss": 0.1585797369480133, "step": 2645 }, { "epoch": 0.6869158878504673, "grad_norm": 1.3989742994308472, "learning_rate": 3.9380187688151664e-06, "loss": 0.1335345357656479, "step": 2646 }, { "epoch": 0.6871754932502596, "grad_norm": 1.6272341012954712, "learning_rate": 3.9372152754407415e-06, "loss": 0.15640035271644592, "step": 2647 }, { "epoch": 0.6874350986500519, "grad_norm": 1.1170251369476318, "learning_rate": 3.936411572193839e-06, "loss": 0.13828812539577484, "step": 2648 }, { "epoch": 0.6876947040498442, "grad_norm": 0.8658260703086853, "learning_rate": 3.9356076592165355e-06, "loss": 0.13760152459144592, "step": 2649 }, { "epoch": 0.6879543094496365, "grad_norm": 0.9643164277076721, "learning_rate": 3.934803536650945e-06, "loss": 0.11977019906044006, "step": 2650 }, { "epoch": 0.6882139148494288, "grad_norm": 4.002286911010742, "learning_rate": 3.933999204639218e-06, "loss": 0.1781056821346283, "step": 2651 }, { "epoch": 0.6884735202492211, "grad_norm": 2.1172423362731934, "learning_rate": 3.933194663323541e-06, "loss": 0.14716258645057678, "step": 2652 }, { "epoch": 0.6887331256490135, "grad_norm": 0.6261714696884155, "learning_rate": 3.93238991284614e-06, "loss": 0.14652395248413086, "step": 2653 }, { "epoch": 0.6889927310488058, "grad_norm": 1.1292786598205566, "learning_rate": 3.931584953349274e-06, "loss": 0.12897640466690063, "step": 2654 }, { "epoch": 0.6892523364485982, "grad_norm": 1.4567211866378784, "learning_rate": 3.930779784975244e-06, "loss": 0.12476596236228943, "step": 2655 }, { "epoch": 0.6895119418483905, "grad_norm": 1.2353005409240723, "learning_rate": 3.929974407866384e-06, "loss": 0.13947704434394836, "step": 2656 }, { "epoch": 0.6897715472481828, "grad_norm": 2.6134960651397705, "learning_rate": 3.929168822165067e-06, "loss": 0.15762168169021606, "step": 2657 }, { "epoch": 0.6900311526479751, "grad_norm": 1.3813608884811401, "learning_rate": 3.9283630280137e-06, "loss": 0.15188530087471008, "step": 2658 }, { "epoch": 0.6902907580477674, "grad_norm": 1.2853279113769531, "learning_rate": 3.927557025554733e-06, "loss": 0.12630826234817505, "step": 2659 }, { "epoch": 0.6905503634475597, "grad_norm": 1.4772642850875854, "learning_rate": 3.926750814930645e-06, "loss": 0.15414634346961975, "step": 2660 }, { "epoch": 0.690809968847352, "grad_norm": 0.8046848177909851, "learning_rate": 3.925944396283957e-06, "loss": 0.13188527524471283, "step": 2661 }, { "epoch": 0.6910695742471443, "grad_norm": 1.1540498733520508, "learning_rate": 3.9251377697572254e-06, "loss": 0.16602681577205658, "step": 2662 }, { "epoch": 0.6913291796469366, "grad_norm": 1.6039890050888062, "learning_rate": 3.9243309354930426e-06, "loss": 0.12801870703697205, "step": 2663 }, { "epoch": 0.6915887850467289, "grad_norm": 1.441299319267273, "learning_rate": 3.9235238936340405e-06, "loss": 0.13301268219947815, "step": 2664 }, { "epoch": 0.6918483904465212, "grad_norm": 0.7660601139068604, "learning_rate": 3.922716644322883e-06, "loss": 0.1165248230099678, "step": 2665 }, { "epoch": 0.6921079958463136, "grad_norm": 0.9008677005767822, "learning_rate": 3.921909187702275e-06, "loss": 0.1527191400527954, "step": 2666 }, { "epoch": 0.692367601246106, "grad_norm": 1.4980016946792603, "learning_rate": 3.921101523914956e-06, "loss": 0.15415911376476288, "step": 2667 }, { "epoch": 0.6926272066458983, "grad_norm": 0.8007984757423401, "learning_rate": 3.920293653103701e-06, "loss": 0.12194646894931793, "step": 2668 }, { "epoch": 0.6928868120456906, "grad_norm": 1.0609618425369263, "learning_rate": 3.9194855754113245e-06, "loss": 0.1489497572183609, "step": 2669 }, { "epoch": 0.6931464174454829, "grad_norm": 2.4102282524108887, "learning_rate": 3.918677290980678e-06, "loss": 0.16867510974407196, "step": 2670 }, { "epoch": 0.6934060228452752, "grad_norm": 1.0330746173858643, "learning_rate": 3.917868799954644e-06, "loss": 0.12098532170057297, "step": 2671 }, { "epoch": 0.6936656282450675, "grad_norm": 0.44134747982025146, "learning_rate": 3.9170601024761465e-06, "loss": 0.11975489556789398, "step": 2672 }, { "epoch": 0.6939252336448598, "grad_norm": 1.3653572797775269, "learning_rate": 3.916251198688145e-06, "loss": 0.1688917875289917, "step": 2673 }, { "epoch": 0.6941848390446521, "grad_norm": 0.977958083152771, "learning_rate": 3.915442088733636e-06, "loss": 0.11752933263778687, "step": 2674 }, { "epoch": 0.6944444444444444, "grad_norm": 0.7769039273262024, "learning_rate": 3.9146327727556504e-06, "loss": 0.15949001908302307, "step": 2675 }, { "epoch": 0.6947040498442367, "grad_norm": 1.5013408660888672, "learning_rate": 3.913823250897257e-06, "loss": 0.15170517563819885, "step": 2676 }, { "epoch": 0.694963655244029, "grad_norm": 2.2489590644836426, "learning_rate": 3.913013523301561e-06, "loss": 0.15182030200958252, "step": 2677 }, { "epoch": 0.6952232606438213, "grad_norm": 1.5830079317092896, "learning_rate": 3.912203590111703e-06, "loss": 0.17300832271575928, "step": 2678 }, { "epoch": 0.6954828660436138, "grad_norm": 0.488534152507782, "learning_rate": 3.911393451470863e-06, "loss": 0.1212650015950203, "step": 2679 }, { "epoch": 0.6957424714434061, "grad_norm": 0.9180216193199158, "learning_rate": 3.910583107522251e-06, "loss": 0.11843477189540863, "step": 2680 }, { "epoch": 0.6960020768431984, "grad_norm": 1.487025260925293, "learning_rate": 3.90977255840912e-06, "loss": 0.1340441107749939, "step": 2681 }, { "epoch": 0.6962616822429907, "grad_norm": 2.048521041870117, "learning_rate": 3.9089618042747564e-06, "loss": 0.16427630186080933, "step": 2682 }, { "epoch": 0.696521287642783, "grad_norm": 0.6944494843482971, "learning_rate": 3.908150845262483e-06, "loss": 0.12096599489450455, "step": 2683 }, { "epoch": 0.6967808930425753, "grad_norm": 1.1771323680877686, "learning_rate": 3.907339681515657e-06, "loss": 0.16427673399448395, "step": 2684 }, { "epoch": 0.6970404984423676, "grad_norm": 0.6091452240943909, "learning_rate": 3.906528313177676e-06, "loss": 0.12589877843856812, "step": 2685 }, { "epoch": 0.6973001038421599, "grad_norm": 2.199723720550537, "learning_rate": 3.90571674039197e-06, "loss": 0.13730016350746155, "step": 2686 }, { "epoch": 0.6975597092419522, "grad_norm": 1.9166476726531982, "learning_rate": 3.904904963302007e-06, "loss": 0.11600879579782486, "step": 2687 }, { "epoch": 0.6978193146417445, "grad_norm": 1.478198528289795, "learning_rate": 3.904092982051289e-06, "loss": 0.11840878427028656, "step": 2688 }, { "epoch": 0.6980789200415368, "grad_norm": 1.1836535930633545, "learning_rate": 3.903280796783358e-06, "loss": 0.13731303811073303, "step": 2689 }, { "epoch": 0.6983385254413291, "grad_norm": 1.733557939529419, "learning_rate": 3.902468407641788e-06, "loss": 0.12340482324361801, "step": 2690 }, { "epoch": 0.6985981308411215, "grad_norm": 2.3528547286987305, "learning_rate": 3.901655814770193e-06, "loss": 0.14999794960021973, "step": 2691 }, { "epoch": 0.6988577362409139, "grad_norm": 1.3591969013214111, "learning_rate": 3.900843018312218e-06, "loss": 0.16391998529434204, "step": 2692 }, { "epoch": 0.6991173416407062, "grad_norm": 1.6162880659103394, "learning_rate": 3.90003001841155e-06, "loss": 0.1417486071586609, "step": 2693 }, { "epoch": 0.6993769470404985, "grad_norm": 1.1196354627609253, "learning_rate": 3.899216815211905e-06, "loss": 0.15132436156272888, "step": 2694 }, { "epoch": 0.6996365524402908, "grad_norm": 1.021953821182251, "learning_rate": 3.898403408857042e-06, "loss": 0.12025286257266998, "step": 2695 }, { "epoch": 0.6998961578400831, "grad_norm": 0.7398144006729126, "learning_rate": 3.897589799490749e-06, "loss": 0.13215580582618713, "step": 2696 }, { "epoch": 0.7001557632398754, "grad_norm": 0.6624515056610107, "learning_rate": 3.8967759872568584e-06, "loss": 0.11477941274642944, "step": 2697 }, { "epoch": 0.7004153686396677, "grad_norm": 1.5301231145858765, "learning_rate": 3.89596197229923e-06, "loss": 0.13406914472579956, "step": 2698 }, { "epoch": 0.70067497403946, "grad_norm": 0.7730684876441956, "learning_rate": 3.895147754761763e-06, "loss": 0.12083166837692261, "step": 2699 }, { "epoch": 0.7009345794392523, "grad_norm": 0.6845230460166931, "learning_rate": 3.894333334788395e-06, "loss": 0.12784399092197418, "step": 2700 }, { "epoch": 0.7011941848390446, "grad_norm": 1.335761308670044, "learning_rate": 3.893518712523094e-06, "loss": 0.15664558112621307, "step": 2701 }, { "epoch": 0.7014537902388369, "grad_norm": 1.1168969869613647, "learning_rate": 3.8927038881098675e-06, "loss": 0.12056680023670197, "step": 2702 }, { "epoch": 0.7017133956386293, "grad_norm": 1.6671009063720703, "learning_rate": 3.891888861692759e-06, "loss": 0.15134066343307495, "step": 2703 }, { "epoch": 0.7019730010384216, "grad_norm": 0.9193665981292725, "learning_rate": 3.891073633415845e-06, "loss": 0.11588171124458313, "step": 2704 }, { "epoch": 0.702232606438214, "grad_norm": 2.2170357704162598, "learning_rate": 3.89025820342324e-06, "loss": 0.1717088222503662, "step": 2705 }, { "epoch": 0.7024922118380063, "grad_norm": 0.9484986662864685, "learning_rate": 3.889442571859093e-06, "loss": 0.14745715260505676, "step": 2706 }, { "epoch": 0.7027518172377986, "grad_norm": 0.5595765709877014, "learning_rate": 3.8886267388675885e-06, "loss": 0.13523998856544495, "step": 2707 }, { "epoch": 0.7030114226375909, "grad_norm": 0.6920851469039917, "learning_rate": 3.887810704592949e-06, "loss": 0.1315462738275528, "step": 2708 }, { "epoch": 0.7032710280373832, "grad_norm": 1.725612998008728, "learning_rate": 3.886994469179429e-06, "loss": 0.12133760750293732, "step": 2709 }, { "epoch": 0.7035306334371755, "grad_norm": 1.0939775705337524, "learning_rate": 3.886178032771319e-06, "loss": 0.13669922947883606, "step": 2710 }, { "epoch": 0.7037902388369678, "grad_norm": 0.7772513628005981, "learning_rate": 3.8853613955129495e-06, "loss": 0.10867948830127716, "step": 2711 }, { "epoch": 0.7040498442367601, "grad_norm": 0.6405813694000244, "learning_rate": 3.88454455754868e-06, "loss": 0.1122007817029953, "step": 2712 }, { "epoch": 0.7043094496365524, "grad_norm": 0.6309729218482971, "learning_rate": 3.883727519022912e-06, "loss": 0.11538343131542206, "step": 2713 }, { "epoch": 0.7045690550363447, "grad_norm": 0.8903509974479675, "learning_rate": 3.882910280080077e-06, "loss": 0.1441054493188858, "step": 2714 }, { "epoch": 0.7048286604361371, "grad_norm": 0.7241120934486389, "learning_rate": 3.882092840864645e-06, "loss": 0.11274515092372894, "step": 2715 }, { "epoch": 0.7050882658359294, "grad_norm": 1.2914685010910034, "learning_rate": 3.88127520152112e-06, "loss": 0.13414841890335083, "step": 2716 }, { "epoch": 0.7053478712357217, "grad_norm": 2.363971471786499, "learning_rate": 3.880457362194041e-06, "loss": 0.16696779429912567, "step": 2717 }, { "epoch": 0.705607476635514, "grad_norm": 1.6480512619018555, "learning_rate": 3.879639323027986e-06, "loss": 0.12914445996284485, "step": 2718 }, { "epoch": 0.7058670820353063, "grad_norm": 0.6351020932197571, "learning_rate": 3.878821084167563e-06, "loss": 0.13848167657852173, "step": 2719 }, { "epoch": 0.7061266874350987, "grad_norm": 1.744951605796814, "learning_rate": 3.8780026457574195e-06, "loss": 0.13843105733394623, "step": 2720 }, { "epoch": 0.706386292834891, "grad_norm": 0.5927450060844421, "learning_rate": 3.877184007942236e-06, "loss": 0.11298786848783493, "step": 2721 }, { "epoch": 0.7066458982346833, "grad_norm": 0.963493287563324, "learning_rate": 3.8763651708667285e-06, "loss": 0.1251424252986908, "step": 2722 }, { "epoch": 0.7069055036344756, "grad_norm": 0.49081477522850037, "learning_rate": 3.8755461346756505e-06, "loss": 0.12377649545669556, "step": 2723 }, { "epoch": 0.7071651090342679, "grad_norm": 0.7451759576797485, "learning_rate": 3.874726899513786e-06, "loss": 0.09276250749826431, "step": 2724 }, { "epoch": 0.7074247144340602, "grad_norm": 1.4717811346054077, "learning_rate": 3.873907465525959e-06, "loss": 0.12519985437393188, "step": 2725 }, { "epoch": 0.7076843198338525, "grad_norm": 0.4215551018714905, "learning_rate": 3.873087832857026e-06, "loss": 0.11494424939155579, "step": 2726 }, { "epoch": 0.7079439252336449, "grad_norm": 1.3300447463989258, "learning_rate": 3.8722680016518796e-06, "loss": 0.17024369537830353, "step": 2727 }, { "epoch": 0.7082035306334372, "grad_norm": 0.6798640489578247, "learning_rate": 3.8714479720554475e-06, "loss": 0.11192208528518677, "step": 2728 }, { "epoch": 0.7084631360332295, "grad_norm": 1.0323894023895264, "learning_rate": 3.870627744212691e-06, "loss": 0.11244663596153259, "step": 2729 }, { "epoch": 0.7087227414330218, "grad_norm": 1.5052570104599, "learning_rate": 3.869807318268609e-06, "loss": 0.12383602559566498, "step": 2730 }, { "epoch": 0.7089823468328141, "grad_norm": 0.5972686409950256, "learning_rate": 3.868986694368232e-06, "loss": 0.12290462851524353, "step": 2731 }, { "epoch": 0.7092419522326064, "grad_norm": 1.147426962852478, "learning_rate": 3.8681658726566305e-06, "loss": 0.15947361290454865, "step": 2732 }, { "epoch": 0.7095015576323987, "grad_norm": 1.135428786277771, "learning_rate": 3.8673448532789044e-06, "loss": 0.1139197051525116, "step": 2733 }, { "epoch": 0.7097611630321911, "grad_norm": 0.6401082277297974, "learning_rate": 3.8665236363801915e-06, "loss": 0.12627054750919342, "step": 2734 }, { "epoch": 0.7100207684319834, "grad_norm": 0.885895848274231, "learning_rate": 3.865702222105666e-06, "loss": 0.1305391937494278, "step": 2735 }, { "epoch": 0.7102803738317757, "grad_norm": 1.05241858959198, "learning_rate": 3.8648806106005335e-06, "loss": 0.14441275596618652, "step": 2736 }, { "epoch": 0.710539979231568, "grad_norm": 0.6668108701705933, "learning_rate": 3.864058802010036e-06, "loss": 0.14682596921920776, "step": 2737 }, { "epoch": 0.7107995846313603, "grad_norm": 1.4367130994796753, "learning_rate": 3.86323679647945e-06, "loss": 0.13643601536750793, "step": 2738 }, { "epoch": 0.7110591900311527, "grad_norm": 0.6266149878501892, "learning_rate": 3.8624145941540896e-06, "loss": 0.1173168495297432, "step": 2739 }, { "epoch": 0.711318795430945, "grad_norm": 1.0331377983093262, "learning_rate": 3.861592195179299e-06, "loss": 0.1683410108089447, "step": 2740 }, { "epoch": 0.7115784008307373, "grad_norm": 0.8318489193916321, "learning_rate": 3.860769599700461e-06, "loss": 0.11449376493692398, "step": 2741 }, { "epoch": 0.7118380062305296, "grad_norm": 2.4502005577087402, "learning_rate": 3.859946807862991e-06, "loss": 0.16896210610866547, "step": 2742 }, { "epoch": 0.7120976116303219, "grad_norm": 1.1212551593780518, "learning_rate": 3.8591238198123395e-06, "loss": 0.12221307307481766, "step": 2743 }, { "epoch": 0.7123572170301142, "grad_norm": 1.5640788078308105, "learning_rate": 3.8583006356939926e-06, "loss": 0.12757344543933868, "step": 2744 }, { "epoch": 0.7126168224299065, "grad_norm": 0.9260667562484741, "learning_rate": 3.85747725565347e-06, "loss": 0.1389389932155609, "step": 2745 }, { "epoch": 0.7128764278296988, "grad_norm": 0.4726833701133728, "learning_rate": 3.856653679836326e-06, "loss": 0.12854893505573273, "step": 2746 }, { "epoch": 0.7131360332294912, "grad_norm": 0.744814932346344, "learning_rate": 3.85582990838815e-06, "loss": 0.13028958439826965, "step": 2747 }, { "epoch": 0.7133956386292835, "grad_norm": 1.3812538385391235, "learning_rate": 3.855005941454567e-06, "loss": 0.17156609892845154, "step": 2748 }, { "epoch": 0.7136552440290758, "grad_norm": 0.7274197936058044, "learning_rate": 3.854181779181235e-06, "loss": 0.12309277057647705, "step": 2749 }, { "epoch": 0.7139148494288681, "grad_norm": 1.6298927068710327, "learning_rate": 3.853357421713847e-06, "loss": 0.16265834867954254, "step": 2750 }, { "epoch": 0.7141744548286605, "grad_norm": 0.7537944912910461, "learning_rate": 3.852532869198129e-06, "loss": 0.12705764174461365, "step": 2751 }, { "epoch": 0.7144340602284528, "grad_norm": 0.4322092533111572, "learning_rate": 3.851708121779846e-06, "loss": 0.12641726434230804, "step": 2752 }, { "epoch": 0.7146936656282451, "grad_norm": 0.4570847749710083, "learning_rate": 3.8508831796047925e-06, "loss": 0.1260416954755783, "step": 2753 }, { "epoch": 0.7149532710280374, "grad_norm": 0.609411358833313, "learning_rate": 3.850058042818799e-06, "loss": 0.11893366277217865, "step": 2754 }, { "epoch": 0.7152128764278297, "grad_norm": 0.5679219961166382, "learning_rate": 3.849232711567732e-06, "loss": 0.12071148306131363, "step": 2755 }, { "epoch": 0.715472481827622, "grad_norm": 0.9716575145721436, "learning_rate": 3.848407185997491e-06, "loss": 0.15424534678459167, "step": 2756 }, { "epoch": 0.7157320872274143, "grad_norm": 1.0984269380569458, "learning_rate": 3.847581466254009e-06, "loss": 0.12852905690670013, "step": 2757 }, { "epoch": 0.7159916926272066, "grad_norm": 0.7864635586738586, "learning_rate": 3.846755552483256e-06, "loss": 0.13038021326065063, "step": 2758 }, { "epoch": 0.7162512980269989, "grad_norm": 0.48035383224487305, "learning_rate": 3.845929444831234e-06, "loss": 0.11785197257995605, "step": 2759 }, { "epoch": 0.7165109034267912, "grad_norm": 1.3445066213607788, "learning_rate": 3.845103143443978e-06, "loss": 0.15350893139839172, "step": 2760 }, { "epoch": 0.7167705088265836, "grad_norm": 1.962375521659851, "learning_rate": 3.844276648467561e-06, "loss": 0.14856582880020142, "step": 2761 }, { "epoch": 0.7170301142263759, "grad_norm": 1.1435043811798096, "learning_rate": 3.843449960048089e-06, "loss": 0.1338689625263214, "step": 2762 }, { "epoch": 0.7172897196261683, "grad_norm": 0.6042613983154297, "learning_rate": 3.8426230783317e-06, "loss": 0.12792669236660004, "step": 2763 }, { "epoch": 0.7175493250259606, "grad_norm": 1.330314040184021, "learning_rate": 3.841796003464569e-06, "loss": 0.1395300179719925, "step": 2764 }, { "epoch": 0.7178089304257529, "grad_norm": 2.6982831954956055, "learning_rate": 3.8409687355929035e-06, "loss": 0.13027368485927582, "step": 2765 }, { "epoch": 0.7180685358255452, "grad_norm": 3.148270845413208, "learning_rate": 3.840141274862945e-06, "loss": 0.17565923929214478, "step": 2766 }, { "epoch": 0.7183281412253375, "grad_norm": 1.06434166431427, "learning_rate": 3.83931362142097e-06, "loss": 0.12603500485420227, "step": 2767 }, { "epoch": 0.7185877466251298, "grad_norm": 1.6227219104766846, "learning_rate": 3.83848577541329e-06, "loss": 0.13750961422920227, "step": 2768 }, { "epoch": 0.7188473520249221, "grad_norm": 2.516838550567627, "learning_rate": 3.837657736986246e-06, "loss": 0.13825151324272156, "step": 2769 }, { "epoch": 0.7191069574247144, "grad_norm": 1.918134331703186, "learning_rate": 3.8368295062862196e-06, "loss": 0.12701627612113953, "step": 2770 }, { "epoch": 0.7193665628245067, "grad_norm": 1.7034828662872314, "learning_rate": 3.836001083459621e-06, "loss": 0.15248899161815643, "step": 2771 }, { "epoch": 0.719626168224299, "grad_norm": 2.1658544540405273, "learning_rate": 3.835172468652897e-06, "loss": 0.1550672948360443, "step": 2772 }, { "epoch": 0.7198857736240913, "grad_norm": 1.6034868955612183, "learning_rate": 3.834343662012528e-06, "loss": 0.13505269587039948, "step": 2773 }, { "epoch": 0.7201453790238836, "grad_norm": 0.7937694787979126, "learning_rate": 3.833514663685027e-06, "loss": 0.13681000471115112, "step": 2774 }, { "epoch": 0.7204049844236761, "grad_norm": 1.389899492263794, "learning_rate": 3.832685473816944e-06, "loss": 0.14111429452896118, "step": 2775 }, { "epoch": 0.7206645898234684, "grad_norm": 1.965455412864685, "learning_rate": 3.831856092554858e-06, "loss": 0.15871822834014893, "step": 2776 }, { "epoch": 0.7209241952232607, "grad_norm": 2.5478336811065674, "learning_rate": 3.8310265200453865e-06, "loss": 0.1232709214091301, "step": 2777 }, { "epoch": 0.721183800623053, "grad_norm": 1.7953499555587769, "learning_rate": 3.830196756435179e-06, "loss": 0.12740828096866608, "step": 2778 }, { "epoch": 0.7214434060228453, "grad_norm": 0.7226132154464722, "learning_rate": 3.829366801870917e-06, "loss": 0.13609279692173004, "step": 2779 }, { "epoch": 0.7217030114226376, "grad_norm": 0.6104289889335632, "learning_rate": 3.828536656499321e-06, "loss": 0.10883583128452301, "step": 2780 }, { "epoch": 0.7219626168224299, "grad_norm": 1.3355519771575928, "learning_rate": 3.827706320467137e-06, "loss": 0.11680570244789124, "step": 2781 }, { "epoch": 0.7222222222222222, "grad_norm": 1.2243701219558716, "learning_rate": 3.826875793921152e-06, "loss": 0.11882743239402771, "step": 2782 }, { "epoch": 0.7224818276220145, "grad_norm": 0.6511145830154419, "learning_rate": 3.826045077008185e-06, "loss": 0.14139384031295776, "step": 2783 }, { "epoch": 0.7227414330218068, "grad_norm": 1.4273730516433716, "learning_rate": 3.825214169875085e-06, "loss": 0.16254860162734985, "step": 2784 }, { "epoch": 0.7230010384215991, "grad_norm": 1.2401480674743652, "learning_rate": 3.82438307266874e-06, "loss": 0.1295381486415863, "step": 2785 }, { "epoch": 0.7232606438213914, "grad_norm": 1.1287235021591187, "learning_rate": 3.8235517855360685e-06, "loss": 0.13322359323501587, "step": 2786 }, { "epoch": 0.7235202492211839, "grad_norm": 0.49368226528167725, "learning_rate": 3.822720308624021e-06, "loss": 0.12408016622066498, "step": 2787 }, { "epoch": 0.7237798546209762, "grad_norm": 0.9186971783638, "learning_rate": 3.821888642079586e-06, "loss": 0.11202734708786011, "step": 2788 }, { "epoch": 0.7240394600207685, "grad_norm": 1.3221033811569214, "learning_rate": 3.821056786049781e-06, "loss": 0.12449444830417633, "step": 2789 }, { "epoch": 0.7242990654205608, "grad_norm": 1.6310933828353882, "learning_rate": 3.8202247406816614e-06, "loss": 0.14789196848869324, "step": 2790 }, { "epoch": 0.7245586708203531, "grad_norm": 1.9150031805038452, "learning_rate": 3.819392506122312e-06, "loss": 0.12723536789417267, "step": 2791 }, { "epoch": 0.7248182762201454, "grad_norm": 1.3080676794052124, "learning_rate": 3.8185600825188526e-06, "loss": 0.15041911602020264, "step": 2792 }, { "epoch": 0.7250778816199377, "grad_norm": 0.9461440443992615, "learning_rate": 3.817727470018439e-06, "loss": 0.13338765501976013, "step": 2793 }, { "epoch": 0.72533748701973, "grad_norm": 0.45392051339149475, "learning_rate": 3.816894668768255e-06, "loss": 0.1163177490234375, "step": 2794 }, { "epoch": 0.7255970924195223, "grad_norm": 0.8828633427619934, "learning_rate": 3.816061678915523e-06, "loss": 0.11521786451339722, "step": 2795 }, { "epoch": 0.7258566978193146, "grad_norm": 1.3877187967300415, "learning_rate": 3.815228500607495e-06, "loss": 0.1376389116048813, "step": 2796 }, { "epoch": 0.7261163032191069, "grad_norm": 0.9679870009422302, "learning_rate": 3.8143951339914596e-06, "loss": 0.16121646761894226, "step": 2797 }, { "epoch": 0.7263759086188992, "grad_norm": 1.391513466835022, "learning_rate": 3.813561579214736e-06, "loss": 0.13051170110702515, "step": 2798 }, { "epoch": 0.7266355140186916, "grad_norm": 1.7177648544311523, "learning_rate": 3.8127278364246766e-06, "loss": 0.14456556737422943, "step": 2799 }, { "epoch": 0.726895119418484, "grad_norm": 0.7918587327003479, "learning_rate": 3.811893905768669e-06, "loss": 0.12357808649539948, "step": 2800 }, { "epoch": 0.7271547248182763, "grad_norm": 0.9380884170532227, "learning_rate": 3.811059787394133e-06, "loss": 0.11497846990823746, "step": 2801 }, { "epoch": 0.7274143302180686, "grad_norm": 1.2028698921203613, "learning_rate": 3.8102254814485225e-06, "loss": 0.16700103878974915, "step": 2802 }, { "epoch": 0.7276739356178609, "grad_norm": 1.0384427309036255, "learning_rate": 3.8093909880793223e-06, "loss": 0.11812679469585419, "step": 2803 }, { "epoch": 0.7279335410176532, "grad_norm": 0.41458380222320557, "learning_rate": 3.8085563074340513e-06, "loss": 0.12998820841312408, "step": 2804 }, { "epoch": 0.7281931464174455, "grad_norm": 2.036444664001465, "learning_rate": 3.8077214396602642e-06, "loss": 0.17522040009498596, "step": 2805 }, { "epoch": 0.7284527518172378, "grad_norm": 1.3060671091079712, "learning_rate": 3.806886384905545e-06, "loss": 0.13344909250736237, "step": 2806 }, { "epoch": 0.7287123572170301, "grad_norm": 2.1206040382385254, "learning_rate": 3.806051143317512e-06, "loss": 0.14820241928100586, "step": 2807 }, { "epoch": 0.7289719626168224, "grad_norm": 1.1556628942489624, "learning_rate": 3.805215715043818e-06, "loss": 0.1464838683605194, "step": 2808 }, { "epoch": 0.7292315680166147, "grad_norm": 1.1674237251281738, "learning_rate": 3.8043801002321457e-06, "loss": 0.12689077854156494, "step": 2809 }, { "epoch": 0.729491173416407, "grad_norm": 1.21588933467865, "learning_rate": 3.8035442990302145e-06, "loss": 0.12056069076061249, "step": 2810 }, { "epoch": 0.7297507788161994, "grad_norm": 1.9966316223144531, "learning_rate": 3.802708311585774e-06, "loss": 0.13981524109840393, "step": 2811 }, { "epoch": 0.7300103842159917, "grad_norm": 0.9644107818603516, "learning_rate": 3.801872138046608e-06, "loss": 0.14802703261375427, "step": 2812 }, { "epoch": 0.730269989615784, "grad_norm": 1.630061388015747, "learning_rate": 3.801035778560532e-06, "loss": 0.1254170835018158, "step": 2813 }, { "epoch": 0.7305295950155763, "grad_norm": 1.4875568151474, "learning_rate": 3.800199233275397e-06, "loss": 0.12671396136283875, "step": 2814 }, { "epoch": 0.7307892004153687, "grad_norm": 0.9002212285995483, "learning_rate": 3.7993625023390835e-06, "loss": 0.11681786924600601, "step": 2815 }, { "epoch": 0.731048805815161, "grad_norm": 1.5203810930252075, "learning_rate": 3.7985255858995067e-06, "loss": 0.16512572765350342, "step": 2816 }, { "epoch": 0.7313084112149533, "grad_norm": 1.0548685789108276, "learning_rate": 3.7976884841046135e-06, "loss": 0.14511653780937195, "step": 2817 }, { "epoch": 0.7315680166147456, "grad_norm": 1.2581241130828857, "learning_rate": 3.7968511971023863e-06, "loss": 0.1187683716416359, "step": 2818 }, { "epoch": 0.7318276220145379, "grad_norm": 1.2136250734329224, "learning_rate": 3.7960137250408374e-06, "loss": 0.12384340167045593, "step": 2819 }, { "epoch": 0.7320872274143302, "grad_norm": 0.8849804401397705, "learning_rate": 3.795176068068012e-06, "loss": 0.14297860860824585, "step": 2820 }, { "epoch": 0.7323468328141225, "grad_norm": 1.8636010885238647, "learning_rate": 3.794338226331989e-06, "loss": 0.12339852750301361, "step": 2821 }, { "epoch": 0.7326064382139148, "grad_norm": 2.319173812866211, "learning_rate": 3.7935001999808796e-06, "loss": 0.15197256207466125, "step": 2822 }, { "epoch": 0.7328660436137072, "grad_norm": 0.729279637336731, "learning_rate": 3.792661989162828e-06, "loss": 0.1316017508506775, "step": 2823 }, { "epoch": 0.7331256490134995, "grad_norm": 1.3422569036483765, "learning_rate": 3.7918235940260108e-06, "loss": 0.12898258864879608, "step": 2824 }, { "epoch": 0.7333852544132918, "grad_norm": 1.1166818141937256, "learning_rate": 3.790985014718636e-06, "loss": 0.12068963795900345, "step": 2825 }, { "epoch": 0.7336448598130841, "grad_norm": 1.0301692485809326, "learning_rate": 3.790146251388947e-06, "loss": 0.11535390466451645, "step": 2826 }, { "epoch": 0.7339044652128764, "grad_norm": 1.4617846012115479, "learning_rate": 3.789307304185216e-06, "loss": 0.15290267765522003, "step": 2827 }, { "epoch": 0.7341640706126688, "grad_norm": 1.5088692903518677, "learning_rate": 3.7884681732557517e-06, "loss": 0.1455363929271698, "step": 2828 }, { "epoch": 0.7344236760124611, "grad_norm": 1.6994929313659668, "learning_rate": 3.7876288587488913e-06, "loss": 0.1316034197807312, "step": 2829 }, { "epoch": 0.7346832814122534, "grad_norm": 2.1727867126464844, "learning_rate": 3.786789360813007e-06, "loss": 0.14933088421821594, "step": 2830 }, { "epoch": 0.7349428868120457, "grad_norm": 1.277083396911621, "learning_rate": 3.785949679596504e-06, "loss": 0.14916184544563293, "step": 2831 }, { "epoch": 0.735202492211838, "grad_norm": 0.5715277194976807, "learning_rate": 3.785109815247817e-06, "loss": 0.1219492107629776, "step": 2832 }, { "epoch": 0.7354620976116303, "grad_norm": 0.9311897158622742, "learning_rate": 3.784269767915416e-06, "loss": 0.13132907450199127, "step": 2833 }, { "epoch": 0.7357217030114226, "grad_norm": 0.39967912435531616, "learning_rate": 3.7834295377478013e-06, "loss": 0.12367407232522964, "step": 2834 }, { "epoch": 0.735981308411215, "grad_norm": 0.45859894156455994, "learning_rate": 3.7825891248935066e-06, "loss": 0.10889606177806854, "step": 2835 }, { "epoch": 0.7362409138110073, "grad_norm": 0.6576430201530457, "learning_rate": 3.7817485295010982e-06, "loss": 0.11087368428707123, "step": 2836 }, { "epoch": 0.7365005192107996, "grad_norm": 2.0576746463775635, "learning_rate": 3.7809077517191733e-06, "loss": 0.14081436395645142, "step": 2837 }, { "epoch": 0.7367601246105919, "grad_norm": 0.65217524766922, "learning_rate": 3.780066791696363e-06, "loss": 0.11904457211494446, "step": 2838 }, { "epoch": 0.7370197300103842, "grad_norm": 1.3146755695343018, "learning_rate": 3.779225649581329e-06, "loss": 0.1498395800590515, "step": 2839 }, { "epoch": 0.7372793354101765, "grad_norm": 1.4385634660720825, "learning_rate": 3.778384325522767e-06, "loss": 0.1281532198190689, "step": 2840 }, { "epoch": 0.7375389408099688, "grad_norm": 0.6107858419418335, "learning_rate": 3.7775428196694023e-06, "loss": 0.12470649182796478, "step": 2841 }, { "epoch": 0.7377985462097612, "grad_norm": 2.3210880756378174, "learning_rate": 3.776701132169995e-06, "loss": 0.13522379100322723, "step": 2842 }, { "epoch": 0.7380581516095535, "grad_norm": 1.598090648651123, "learning_rate": 3.7758592631733354e-06, "loss": 0.15540042519569397, "step": 2843 }, { "epoch": 0.7383177570093458, "grad_norm": 0.6512052416801453, "learning_rate": 3.775017212828248e-06, "loss": 0.1392170786857605, "step": 2844 }, { "epoch": 0.7385773624091381, "grad_norm": 0.9935952425003052, "learning_rate": 3.774174981283587e-06, "loss": 0.16057512164115906, "step": 2845 }, { "epoch": 0.7388369678089304, "grad_norm": 0.6656081676483154, "learning_rate": 3.773332568688239e-06, "loss": 0.11980396509170532, "step": 2846 }, { "epoch": 0.7390965732087228, "grad_norm": 1.4191821813583374, "learning_rate": 3.7724899751911254e-06, "loss": 0.11658395826816559, "step": 2847 }, { "epoch": 0.7393561786085151, "grad_norm": 1.1159294843673706, "learning_rate": 3.7716472009411944e-06, "loss": 0.12282313406467438, "step": 2848 }, { "epoch": 0.7396157840083074, "grad_norm": 0.616184413433075, "learning_rate": 3.7708042460874315e-06, "loss": 0.12359504401683807, "step": 2849 }, { "epoch": 0.7398753894080997, "grad_norm": 0.38761699199676514, "learning_rate": 3.769961110778852e-06, "loss": 0.1209191158413887, "step": 2850 }, { "epoch": 0.740134994807892, "grad_norm": 1.6509764194488525, "learning_rate": 3.7691177951645018e-06, "loss": 0.15713152289390564, "step": 2851 }, { "epoch": 0.7403946002076843, "grad_norm": 0.5869132280349731, "learning_rate": 3.7682742993934595e-06, "loss": 0.13655199110507965, "step": 2852 }, { "epoch": 0.7406542056074766, "grad_norm": 0.39876309037208557, "learning_rate": 3.7674306236148374e-06, "loss": 0.12562882900238037, "step": 2853 }, { "epoch": 0.7409138110072689, "grad_norm": 0.7036820650100708, "learning_rate": 3.7665867679777763e-06, "loss": 0.11870056390762329, "step": 2854 }, { "epoch": 0.7411734164070612, "grad_norm": 1.0119805335998535, "learning_rate": 3.7657427326314523e-06, "loss": 0.13731953501701355, "step": 2855 }, { "epoch": 0.7414330218068536, "grad_norm": 0.5159564018249512, "learning_rate": 3.7648985177250697e-06, "loss": 0.1299196183681488, "step": 2856 }, { "epoch": 0.7416926272066459, "grad_norm": 1.6662778854370117, "learning_rate": 3.764054123407867e-06, "loss": 0.13048160076141357, "step": 2857 }, { "epoch": 0.7419522326064382, "grad_norm": 1.0844345092773438, "learning_rate": 3.763209549829115e-06, "loss": 0.12988053262233734, "step": 2858 }, { "epoch": 0.7422118380062306, "grad_norm": 0.6602047085762024, "learning_rate": 3.7623647971381132e-06, "loss": 0.13066084682941437, "step": 2859 }, { "epoch": 0.7424714434060229, "grad_norm": 0.7996280789375305, "learning_rate": 3.7615198654841957e-06, "loss": 0.15612152218818665, "step": 2860 }, { "epoch": 0.7427310488058152, "grad_norm": 1.4275537729263306, "learning_rate": 3.760674755016727e-06, "loss": 0.15018950402736664, "step": 2861 }, { "epoch": 0.7429906542056075, "grad_norm": 0.8661129474639893, "learning_rate": 3.7598294658851025e-06, "loss": 0.13048090040683746, "step": 2862 }, { "epoch": 0.7432502596053998, "grad_norm": 2.0595593452453613, "learning_rate": 3.758983998238751e-06, "loss": 0.15836772322654724, "step": 2863 }, { "epoch": 0.7435098650051921, "grad_norm": 1.03414785861969, "learning_rate": 3.758138352227131e-06, "loss": 0.14659710228443146, "step": 2864 }, { "epoch": 0.7437694704049844, "grad_norm": 0.8945192694664001, "learning_rate": 3.757292527999734e-06, "loss": 0.10721272230148315, "step": 2865 }, { "epoch": 0.7440290758047767, "grad_norm": 1.5768104791641235, "learning_rate": 3.756446525706081e-06, "loss": 0.16853147745132446, "step": 2866 }, { "epoch": 0.744288681204569, "grad_norm": 0.9750483632087708, "learning_rate": 3.7556003454957273e-06, "loss": 0.1574917733669281, "step": 2867 }, { "epoch": 0.7445482866043613, "grad_norm": 0.7810298800468445, "learning_rate": 3.754753987518259e-06, "loss": 0.12087561190128326, "step": 2868 }, { "epoch": 0.7448078920041536, "grad_norm": 0.7806090712547302, "learning_rate": 3.75390745192329e-06, "loss": 0.18045440316200256, "step": 2869 }, { "epoch": 0.745067497403946, "grad_norm": 0.6414445638656616, "learning_rate": 3.75306073886047e-06, "loss": 0.11867902427911758, "step": 2870 }, { "epoch": 0.7453271028037384, "grad_norm": 0.9316613078117371, "learning_rate": 3.752213848479479e-06, "loss": 0.1687156856060028, "step": 2871 }, { "epoch": 0.7455867082035307, "grad_norm": 1.2755911350250244, "learning_rate": 3.751366780930027e-06, "loss": 0.10856010764837265, "step": 2872 }, { "epoch": 0.745846313603323, "grad_norm": 0.9770328998565674, "learning_rate": 3.7505195363618563e-06, "loss": 0.15054859220981598, "step": 2873 }, { "epoch": 0.7461059190031153, "grad_norm": 0.8606960773468018, "learning_rate": 3.7496721149247407e-06, "loss": 0.11677353084087372, "step": 2874 }, { "epoch": 0.7463655244029076, "grad_norm": 0.8263111114501953, "learning_rate": 3.7488245167684846e-06, "loss": 0.14358526468276978, "step": 2875 }, { "epoch": 0.7466251298026999, "grad_norm": 2.101653575897217, "learning_rate": 3.7479767420429235e-06, "loss": 0.1458864063024521, "step": 2876 }, { "epoch": 0.7468847352024922, "grad_norm": 0.43834128975868225, "learning_rate": 3.7471287908979254e-06, "loss": 0.1267164945602417, "step": 2877 }, { "epoch": 0.7471443406022845, "grad_norm": 1.5879014730453491, "learning_rate": 3.746280663483389e-06, "loss": 0.13494816422462463, "step": 2878 }, { "epoch": 0.7474039460020768, "grad_norm": 0.6269850134849548, "learning_rate": 3.7454323599492414e-06, "loss": 0.1251240074634552, "step": 2879 }, { "epoch": 0.7476635514018691, "grad_norm": 1.2615723609924316, "learning_rate": 3.744583880445446e-06, "loss": 0.1525464653968811, "step": 2880 }, { "epoch": 0.7479231568016614, "grad_norm": 1.5566779375076294, "learning_rate": 3.743735225121994e-06, "loss": 0.15866237878799438, "step": 2881 }, { "epoch": 0.7481827622014537, "grad_norm": 0.8218278288841248, "learning_rate": 3.7428863941289067e-06, "loss": 0.11739620566368103, "step": 2882 }, { "epoch": 0.7484423676012462, "grad_norm": 0.6242524981498718, "learning_rate": 3.742037387616239e-06, "loss": 0.1187063604593277, "step": 2883 }, { "epoch": 0.7487019730010385, "grad_norm": 1.3642946481704712, "learning_rate": 3.7411882057340766e-06, "loss": 0.15240907669067383, "step": 2884 }, { "epoch": 0.7489615784008308, "grad_norm": 0.585020899772644, "learning_rate": 3.7403388486325345e-06, "loss": 0.13522449135780334, "step": 2885 }, { "epoch": 0.7492211838006231, "grad_norm": 1.5749956369400024, "learning_rate": 3.739489316461759e-06, "loss": 0.13751806318759918, "step": 2886 }, { "epoch": 0.7494807892004154, "grad_norm": 1.4554438591003418, "learning_rate": 3.7386396093719297e-06, "loss": 0.13034158945083618, "step": 2887 }, { "epoch": 0.7497403946002077, "grad_norm": 0.8753548264503479, "learning_rate": 3.7377897275132534e-06, "loss": 0.11938848346471786, "step": 2888 }, { "epoch": 0.75, "grad_norm": 1.2108615636825562, "learning_rate": 3.736939671035971e-06, "loss": 0.14090624451637268, "step": 2889 }, { "epoch": 0.7502596053997923, "grad_norm": 2.25524640083313, "learning_rate": 3.736089440090353e-06, "loss": 0.12534740567207336, "step": 2890 }, { "epoch": 0.7505192107995846, "grad_norm": 0.88081294298172, "learning_rate": 3.7352390348266997e-06, "loss": 0.14918889105319977, "step": 2891 }, { "epoch": 0.7507788161993769, "grad_norm": 1.4003503322601318, "learning_rate": 3.734388455395344e-06, "loss": 0.15067210793495178, "step": 2892 }, { "epoch": 0.7510384215991692, "grad_norm": 1.1831380128860474, "learning_rate": 3.7335377019466497e-06, "loss": 0.12560227513313293, "step": 2893 }, { "epoch": 0.7512980269989615, "grad_norm": 1.5355585813522339, "learning_rate": 3.73268677463101e-06, "loss": 0.1457262933254242, "step": 2894 }, { "epoch": 0.7515576323987538, "grad_norm": 1.8538466691970825, "learning_rate": 3.731835673598848e-06, "loss": 0.13592791557312012, "step": 2895 }, { "epoch": 0.7518172377985463, "grad_norm": 1.4546033143997192, "learning_rate": 3.7309843990006207e-06, "loss": 0.1161670982837677, "step": 2896 }, { "epoch": 0.7520768431983386, "grad_norm": 1.7629292011260986, "learning_rate": 3.7301329509868124e-06, "loss": 0.1327940821647644, "step": 2897 }, { "epoch": 0.7523364485981309, "grad_norm": 0.7776616811752319, "learning_rate": 3.7292813297079416e-06, "loss": 0.12049609422683716, "step": 2898 }, { "epoch": 0.7525960539979232, "grad_norm": 0.8135874271392822, "learning_rate": 3.728429535314554e-06, "loss": 0.12373442947864532, "step": 2899 }, { "epoch": 0.7528556593977155, "grad_norm": 1.3835173845291138, "learning_rate": 3.727577567957227e-06, "loss": 0.12878704071044922, "step": 2900 }, { "epoch": 0.7531152647975078, "grad_norm": 0.9724093079566956, "learning_rate": 3.72672542778657e-06, "loss": 0.13539062440395355, "step": 2901 }, { "epoch": 0.7533748701973001, "grad_norm": 1.3075278997421265, "learning_rate": 3.7258731149532207e-06, "loss": 0.11064336448907852, "step": 2902 }, { "epoch": 0.7536344755970924, "grad_norm": 1.0098317861557007, "learning_rate": 3.72502062960785e-06, "loss": 0.16081754863262177, "step": 2903 }, { "epoch": 0.7538940809968847, "grad_norm": 0.3789827823638916, "learning_rate": 3.724167971901157e-06, "loss": 0.12113432586193085, "step": 2904 }, { "epoch": 0.754153686396677, "grad_norm": 0.7352511882781982, "learning_rate": 3.7233151419838713e-06, "loss": 0.1254417598247528, "step": 2905 }, { "epoch": 0.7544132917964693, "grad_norm": 0.7946515679359436, "learning_rate": 3.722462140006755e-06, "loss": 0.12450249493122101, "step": 2906 }, { "epoch": 0.7546728971962616, "grad_norm": 1.526239275932312, "learning_rate": 3.721608966120599e-06, "loss": 0.12210990488529205, "step": 2907 }, { "epoch": 0.754932502596054, "grad_norm": 0.6140344738960266, "learning_rate": 3.7207556204762248e-06, "loss": 0.13552415370941162, "step": 2908 }, { "epoch": 0.7551921079958464, "grad_norm": 0.6423452496528625, "learning_rate": 3.7199021032244843e-06, "loss": 0.12720945477485657, "step": 2909 }, { "epoch": 0.7554517133956387, "grad_norm": 0.8180195689201355, "learning_rate": 3.7190484145162585e-06, "loss": 0.1340494453907013, "step": 2910 }, { "epoch": 0.755711318795431, "grad_norm": 1.369163990020752, "learning_rate": 3.7181945545024622e-06, "loss": 0.12772975862026215, "step": 2911 }, { "epoch": 0.7559709241952233, "grad_norm": 1.3261581659317017, "learning_rate": 3.717340523334037e-06, "loss": 0.16720786690711975, "step": 2912 }, { "epoch": 0.7562305295950156, "grad_norm": 1.0930678844451904, "learning_rate": 3.716486321161957e-06, "loss": 0.13847950100898743, "step": 2913 }, { "epoch": 0.7564901349948079, "grad_norm": 0.5485044121742249, "learning_rate": 3.7156319481372247e-06, "loss": 0.11880974471569061, "step": 2914 }, { "epoch": 0.7567497403946002, "grad_norm": 0.827796995639801, "learning_rate": 3.7147774044108743e-06, "loss": 0.12300625443458557, "step": 2915 }, { "epoch": 0.7570093457943925, "grad_norm": 1.0070844888687134, "learning_rate": 3.7139226901339685e-06, "loss": 0.14401543140411377, "step": 2916 }, { "epoch": 0.7572689511941848, "grad_norm": 1.3226070404052734, "learning_rate": 3.713067805457603e-06, "loss": 0.13629847764968872, "step": 2917 }, { "epoch": 0.7575285565939771, "grad_norm": 0.4787199795246124, "learning_rate": 3.7122127505329003e-06, "loss": 0.10056959837675095, "step": 2918 }, { "epoch": 0.7577881619937694, "grad_norm": 2.017669677734375, "learning_rate": 3.7113575255110155e-06, "loss": 0.14453737437725067, "step": 2919 }, { "epoch": 0.7580477673935618, "grad_norm": 0.6154787540435791, "learning_rate": 3.7105021305431317e-06, "loss": 0.13139910995960236, "step": 2920 }, { "epoch": 0.7583073727933541, "grad_norm": 1.0606563091278076, "learning_rate": 3.709646565780464e-06, "loss": 0.13323664665222168, "step": 2921 }, { "epoch": 0.7585669781931464, "grad_norm": 1.2618813514709473, "learning_rate": 3.708790831374257e-06, "loss": 0.15910395979881287, "step": 2922 }, { "epoch": 0.7588265835929388, "grad_norm": 0.7178090214729309, "learning_rate": 3.7079349274757838e-06, "loss": 0.11739201098680496, "step": 2923 }, { "epoch": 0.7590861889927311, "grad_norm": 0.929623544216156, "learning_rate": 3.7070788542363495e-06, "loss": 0.15225309133529663, "step": 2924 }, { "epoch": 0.7593457943925234, "grad_norm": 2.908353567123413, "learning_rate": 3.706222611807288e-06, "loss": 0.12159212678670883, "step": 2925 }, { "epoch": 0.7596053997923157, "grad_norm": 0.9185747504234314, "learning_rate": 3.705366200339964e-06, "loss": 0.15304329991340637, "step": 2926 }, { "epoch": 0.759865005192108, "grad_norm": 0.8913516402244568, "learning_rate": 3.7045096199857694e-06, "loss": 0.1318478286266327, "step": 2927 }, { "epoch": 0.7601246105919003, "grad_norm": 0.696664035320282, "learning_rate": 3.7036528708961297e-06, "loss": 0.12616363167762756, "step": 2928 }, { "epoch": 0.7603842159916926, "grad_norm": 0.8118481040000916, "learning_rate": 3.702795953222498e-06, "loss": 0.12915974855422974, "step": 2929 }, { "epoch": 0.7606438213914849, "grad_norm": 0.8601959943771362, "learning_rate": 3.7019388671163588e-06, "loss": 0.13160920143127441, "step": 2930 }, { "epoch": 0.7609034267912772, "grad_norm": 0.6762826442718506, "learning_rate": 3.7010816127292245e-06, "loss": 0.12078243494033813, "step": 2931 }, { "epoch": 0.7611630321910696, "grad_norm": 0.7594860792160034, "learning_rate": 3.7002241902126366e-06, "loss": 0.11182296276092529, "step": 2932 }, { "epoch": 0.7614226375908619, "grad_norm": 0.6539742946624756, "learning_rate": 3.6993665997181698e-06, "loss": 0.13055451214313507, "step": 2933 }, { "epoch": 0.7616822429906542, "grad_norm": 0.4792957007884979, "learning_rate": 3.698508841397426e-06, "loss": 0.13024982810020447, "step": 2934 }, { "epoch": 0.7619418483904465, "grad_norm": 0.7374566793441772, "learning_rate": 3.6976509154020375e-06, "loss": 0.1327759325504303, "step": 2935 }, { "epoch": 0.7622014537902388, "grad_norm": 0.727636992931366, "learning_rate": 3.6967928218836646e-06, "loss": 0.13072967529296875, "step": 2936 }, { "epoch": 0.7624610591900312, "grad_norm": 0.5232887864112854, "learning_rate": 3.695934560993999e-06, "loss": 0.10837307572364807, "step": 2937 }, { "epoch": 0.7627206645898235, "grad_norm": 1.18186354637146, "learning_rate": 3.6950761328847634e-06, "loss": 0.14934468269348145, "step": 2938 }, { "epoch": 0.7629802699896158, "grad_norm": 0.9433581233024597, "learning_rate": 3.694217537707706e-06, "loss": 0.14650774002075195, "step": 2939 }, { "epoch": 0.7632398753894081, "grad_norm": 1.3038249015808105, "learning_rate": 3.6933587756146085e-06, "loss": 0.14831475913524628, "step": 2940 }, { "epoch": 0.7634994807892004, "grad_norm": 0.6303321719169617, "learning_rate": 3.6924998467572782e-06, "loss": 0.13856348395347595, "step": 2941 }, { "epoch": 0.7637590861889927, "grad_norm": 0.5738450288772583, "learning_rate": 3.6916407512875567e-06, "loss": 0.1339329481124878, "step": 2942 }, { "epoch": 0.764018691588785, "grad_norm": 1.6080589294433594, "learning_rate": 3.6907814893573102e-06, "loss": 0.13708917796611786, "step": 2943 }, { "epoch": 0.7642782969885774, "grad_norm": 1.5134367942810059, "learning_rate": 3.6899220611184384e-06, "loss": 0.1622796505689621, "step": 2944 }, { "epoch": 0.7645379023883697, "grad_norm": 0.9671915173530579, "learning_rate": 3.6890624667228664e-06, "loss": 0.1534830629825592, "step": 2945 }, { "epoch": 0.764797507788162, "grad_norm": 1.1862012147903442, "learning_rate": 3.6882027063225527e-06, "loss": 0.1220996081829071, "step": 2946 }, { "epoch": 0.7650571131879543, "grad_norm": 1.579140067100525, "learning_rate": 3.687342780069482e-06, "loss": 0.11295185983181, "step": 2947 }, { "epoch": 0.7653167185877466, "grad_norm": 1.347744345664978, "learning_rate": 3.68648268811567e-06, "loss": 0.12977510690689087, "step": 2948 }, { "epoch": 0.7655763239875389, "grad_norm": 2.1818926334381104, "learning_rate": 3.6856224306131622e-06, "loss": 0.13041800260543823, "step": 2949 }, { "epoch": 0.7658359293873312, "grad_norm": 1.6817002296447754, "learning_rate": 3.684762007714031e-06, "loss": 0.1485755741596222, "step": 2950 }, { "epoch": 0.7660955347871236, "grad_norm": 1.4088993072509766, "learning_rate": 3.68390141957038e-06, "loss": 0.13829351961612701, "step": 2951 }, { "epoch": 0.7663551401869159, "grad_norm": 0.6310020685195923, "learning_rate": 3.683040666334342e-06, "loss": 0.12002574652433395, "step": 2952 }, { "epoch": 0.7666147455867082, "grad_norm": 1.2549110651016235, "learning_rate": 3.682179748158078e-06, "loss": 0.12563183903694153, "step": 2953 }, { "epoch": 0.7668743509865005, "grad_norm": 0.9790667295455933, "learning_rate": 3.6813186651937775e-06, "loss": 0.14301177859306335, "step": 2954 }, { "epoch": 0.7671339563862928, "grad_norm": 1.5135159492492676, "learning_rate": 3.6804574175936623e-06, "loss": 0.15535272657871246, "step": 2955 }, { "epoch": 0.7673935617860852, "grad_norm": 2.8924381732940674, "learning_rate": 3.6795960055099804e-06, "loss": 0.14965657889842987, "step": 2956 }, { "epoch": 0.7676531671858775, "grad_norm": 1.4449771642684937, "learning_rate": 3.6787344290950094e-06, "loss": 0.12135956436395645, "step": 2957 }, { "epoch": 0.7679127725856698, "grad_norm": 1.109570026397705, "learning_rate": 3.677872688501057e-06, "loss": 0.15458238124847412, "step": 2958 }, { "epoch": 0.7681723779854621, "grad_norm": 1.7396775484085083, "learning_rate": 3.6770107838804577e-06, "loss": 0.16418105363845825, "step": 2959 }, { "epoch": 0.7684319833852544, "grad_norm": 1.1419793367385864, "learning_rate": 3.676148715385578e-06, "loss": 0.12123428285121918, "step": 2960 }, { "epoch": 0.7686915887850467, "grad_norm": 1.5248329639434814, "learning_rate": 3.6752864831688124e-06, "loss": 0.15027856826782227, "step": 2961 }, { "epoch": 0.768951194184839, "grad_norm": 1.2380425930023193, "learning_rate": 3.6744240873825816e-06, "loss": 0.14444123208522797, "step": 2962 }, { "epoch": 0.7692107995846313, "grad_norm": 1.9100834131240845, "learning_rate": 3.673561528179339e-06, "loss": 0.11066551506519318, "step": 2963 }, { "epoch": 0.7694704049844237, "grad_norm": 0.8918291926383972, "learning_rate": 3.6726988057115646e-06, "loss": 0.13518081605434418, "step": 2964 }, { "epoch": 0.769730010384216, "grad_norm": 0.8016351461410522, "learning_rate": 3.6718359201317695e-06, "loss": 0.13608884811401367, "step": 2965 }, { "epoch": 0.7699896157840083, "grad_norm": 2.431972026824951, "learning_rate": 3.67097287159249e-06, "loss": 0.17196349799633026, "step": 2966 }, { "epoch": 0.7702492211838006, "grad_norm": 0.8118324279785156, "learning_rate": 3.6701096602462944e-06, "loss": 0.11433656513690948, "step": 2967 }, { "epoch": 0.770508826583593, "grad_norm": 1.272602915763855, "learning_rate": 3.6692462862457784e-06, "loss": 0.17330867052078247, "step": 2968 }, { "epoch": 0.7707684319833853, "grad_norm": 0.47152623534202576, "learning_rate": 3.6683827497435678e-06, "loss": 0.12390555441379547, "step": 2969 }, { "epoch": 0.7710280373831776, "grad_norm": 0.430052250623703, "learning_rate": 3.667519050892315e-06, "loss": 0.11767039448022842, "step": 2970 }, { "epoch": 0.7712876427829699, "grad_norm": 0.9827051758766174, "learning_rate": 3.666655189844701e-06, "loss": 0.11618689447641373, "step": 2971 }, { "epoch": 0.7715472481827622, "grad_norm": 1.1080501079559326, "learning_rate": 3.6657911667534397e-06, "loss": 0.1412663757801056, "step": 2972 }, { "epoch": 0.7718068535825545, "grad_norm": 1.1572580337524414, "learning_rate": 3.6649269817712686e-06, "loss": 0.11297032237052917, "step": 2973 }, { "epoch": 0.7720664589823468, "grad_norm": 0.8001288175582886, "learning_rate": 3.6640626350509557e-06, "loss": 0.11485598981380463, "step": 2974 }, { "epoch": 0.7723260643821391, "grad_norm": 1.0771698951721191, "learning_rate": 3.663198126745299e-06, "loss": 0.11658136546611786, "step": 2975 }, { "epoch": 0.7725856697819314, "grad_norm": 1.2454218864440918, "learning_rate": 3.6623334570071216e-06, "loss": 0.13524755835533142, "step": 2976 }, { "epoch": 0.7728452751817237, "grad_norm": 1.6609680652618408, "learning_rate": 3.6614686259892794e-06, "loss": 0.12783509492874146, "step": 2977 }, { "epoch": 0.773104880581516, "grad_norm": 1.0987070798873901, "learning_rate": 3.660603633844654e-06, "loss": 0.13382494449615479, "step": 2978 }, { "epoch": 0.7733644859813084, "grad_norm": 0.7514723539352417, "learning_rate": 3.6597384807261564e-06, "loss": 0.14251133799552917, "step": 2979 }, { "epoch": 0.7736240913811008, "grad_norm": 0.4745015501976013, "learning_rate": 3.658873166786726e-06, "loss": 0.11798955500125885, "step": 2980 }, { "epoch": 0.7738836967808931, "grad_norm": 1.0699173212051392, "learning_rate": 3.6580076921793295e-06, "loss": 0.1578500121831894, "step": 2981 }, { "epoch": 0.7741433021806854, "grad_norm": 0.7577852606773376, "learning_rate": 3.6571420570569643e-06, "loss": 0.12330951541662216, "step": 2982 }, { "epoch": 0.7744029075804777, "grad_norm": 1.2175461053848267, "learning_rate": 3.6562762615726538e-06, "loss": 0.12724213302135468, "step": 2983 }, { "epoch": 0.77466251298027, "grad_norm": 1.1206190586090088, "learning_rate": 3.6554103058794504e-06, "loss": 0.14205250144004822, "step": 2984 }, { "epoch": 0.7749221183800623, "grad_norm": 0.8334574699401855, "learning_rate": 3.6545441901304366e-06, "loss": 0.14015167951583862, "step": 2985 }, { "epoch": 0.7751817237798546, "grad_norm": 0.3650757670402527, "learning_rate": 3.6536779144787223e-06, "loss": 0.1209346204996109, "step": 2986 }, { "epoch": 0.7754413291796469, "grad_norm": 0.5055058598518372, "learning_rate": 3.652811479077443e-06, "loss": 0.11570637673139572, "step": 2987 }, { "epoch": 0.7757009345794392, "grad_norm": 0.736051619052887, "learning_rate": 3.6519448840797667e-06, "loss": 0.11990070343017578, "step": 2988 }, { "epoch": 0.7759605399792315, "grad_norm": 1.2463539838790894, "learning_rate": 3.651078129638886e-06, "loss": 0.1425105333328247, "step": 2989 }, { "epoch": 0.7762201453790238, "grad_norm": 2.3241217136383057, "learning_rate": 3.6502112159080243e-06, "loss": 0.16570842266082764, "step": 2990 }, { "epoch": 0.7764797507788161, "grad_norm": 0.6101783514022827, "learning_rate": 3.6493441430404315e-06, "loss": 0.1333724856376648, "step": 2991 }, { "epoch": 0.7767393561786086, "grad_norm": 1.9372745752334595, "learning_rate": 3.648476911189387e-06, "loss": 0.14312134683132172, "step": 2992 }, { "epoch": 0.7769989615784009, "grad_norm": 1.7479950189590454, "learning_rate": 3.6476095205081963e-06, "loss": 0.14795014262199402, "step": 2993 }, { "epoch": 0.7772585669781932, "grad_norm": 1.0676580667495728, "learning_rate": 3.646741971150195e-06, "loss": 0.11960157752037048, "step": 2994 }, { "epoch": 0.7775181723779855, "grad_norm": 0.6776900291442871, "learning_rate": 3.6458742632687462e-06, "loss": 0.1331315040588379, "step": 2995 }, { "epoch": 0.7777777777777778, "grad_norm": 0.4445134699344635, "learning_rate": 3.64500639701724e-06, "loss": 0.1384040117263794, "step": 2996 }, { "epoch": 0.7780373831775701, "grad_norm": 0.9642837047576904, "learning_rate": 3.6441383725490957e-06, "loss": 0.1332297921180725, "step": 2997 }, { "epoch": 0.7782969885773624, "grad_norm": 0.6753078699111938, "learning_rate": 3.643270190017761e-06, "loss": 0.16078606247901917, "step": 2998 }, { "epoch": 0.7785565939771547, "grad_norm": 1.1077461242675781, "learning_rate": 3.642401849576709e-06, "loss": 0.15305593609809875, "step": 2999 }, { "epoch": 0.778816199376947, "grad_norm": 0.9844903945922852, "learning_rate": 3.6415333513794442e-06, "loss": 0.13248944282531738, "step": 3000 }, { "epoch": 0.7790758047767393, "grad_norm": 0.524440586566925, "learning_rate": 3.640664695579496e-06, "loss": 0.12445004284381866, "step": 3001 }, { "epoch": 0.7793354101765316, "grad_norm": 0.5426634550094604, "learning_rate": 3.639795882330423e-06, "loss": 0.1302594393491745, "step": 3002 }, { "epoch": 0.7795950155763239, "grad_norm": 0.7941511273384094, "learning_rate": 3.638926911785811e-06, "loss": 0.11667034029960632, "step": 3003 }, { "epoch": 0.7798546209761164, "grad_norm": 1.3479667901992798, "learning_rate": 3.6380577840992758e-06, "loss": 0.14484421908855438, "step": 3004 }, { "epoch": 0.7801142263759087, "grad_norm": 0.5748679041862488, "learning_rate": 3.637188499424457e-06, "loss": 0.10682033002376556, "step": 3005 }, { "epoch": 0.780373831775701, "grad_norm": 1.277795433998108, "learning_rate": 3.636319057915027e-06, "loss": 0.15339645743370056, "step": 3006 }, { "epoch": 0.7806334371754933, "grad_norm": 0.4855096638202667, "learning_rate": 3.63544945972468e-06, "loss": 0.11138072609901428, "step": 3007 }, { "epoch": 0.7808930425752856, "grad_norm": 0.9529732465744019, "learning_rate": 3.634579705007144e-06, "loss": 0.14332032203674316, "step": 3008 }, { "epoch": 0.7811526479750779, "grad_norm": 1.068101406097412, "learning_rate": 3.6337097939161698e-06, "loss": 0.1326829344034195, "step": 3009 }, { "epoch": 0.7814122533748702, "grad_norm": 0.6469199657440186, "learning_rate": 3.632839726605538e-06, "loss": 0.12049874663352966, "step": 3010 }, { "epoch": 0.7816718587746625, "grad_norm": 0.9762584567070007, "learning_rate": 3.6319695032290565e-06, "loss": 0.13574150204658508, "step": 3011 }, { "epoch": 0.7819314641744548, "grad_norm": 2.2029366493225098, "learning_rate": 3.6310991239405625e-06, "loss": 0.1379753053188324, "step": 3012 }, { "epoch": 0.7821910695742471, "grad_norm": 0.7643151879310608, "learning_rate": 3.630228588893917e-06, "loss": 0.13108986616134644, "step": 3013 }, { "epoch": 0.7824506749740394, "grad_norm": 1.6213968992233276, "learning_rate": 3.629357898243011e-06, "loss": 0.1631002128124237, "step": 3014 }, { "epoch": 0.7827102803738317, "grad_norm": 1.2264796495437622, "learning_rate": 3.6284870521417646e-06, "loss": 0.14566084742546082, "step": 3015 }, { "epoch": 0.7829698857736241, "grad_norm": 0.9135777354240417, "learning_rate": 3.627616050744121e-06, "loss": 0.13079944252967834, "step": 3016 }, { "epoch": 0.7832294911734164, "grad_norm": 1.115212082862854, "learning_rate": 3.6267448942040553e-06, "loss": 0.15841513872146606, "step": 3017 }, { "epoch": 0.7834890965732088, "grad_norm": 1.2186405658721924, "learning_rate": 3.6258735826755663e-06, "loss": 0.12332269549369812, "step": 3018 }, { "epoch": 0.7837487019730011, "grad_norm": 1.0196696519851685, "learning_rate": 3.625002116312683e-06, "loss": 0.14499354362487793, "step": 3019 }, { "epoch": 0.7840083073727934, "grad_norm": 1.2057218551635742, "learning_rate": 3.6241304952694606e-06, "loss": 0.13345563411712646, "step": 3020 }, { "epoch": 0.7842679127725857, "grad_norm": 0.9268214702606201, "learning_rate": 3.623258719699982e-06, "loss": 0.13701850175857544, "step": 3021 }, { "epoch": 0.784527518172378, "grad_norm": 1.0249310731887817, "learning_rate": 3.6223867897583555e-06, "loss": 0.12330830842256546, "step": 3022 }, { "epoch": 0.7847871235721703, "grad_norm": 0.9731420278549194, "learning_rate": 3.6215147055987204e-06, "loss": 0.15972991287708282, "step": 3023 }, { "epoch": 0.7850467289719626, "grad_norm": 1.1104564666748047, "learning_rate": 3.62064246737524e-06, "loss": 0.11601535230875015, "step": 3024 }, { "epoch": 0.7853063343717549, "grad_norm": 0.5247167348861694, "learning_rate": 3.6197700752421066e-06, "loss": 0.1204422116279602, "step": 3025 }, { "epoch": 0.7855659397715472, "grad_norm": 1.03827965259552, "learning_rate": 3.6188975293535393e-06, "loss": 0.14855170249938965, "step": 3026 }, { "epoch": 0.7858255451713395, "grad_norm": 1.5412085056304932, "learning_rate": 3.618024829863784e-06, "loss": 0.12437313795089722, "step": 3027 }, { "epoch": 0.7860851505711319, "grad_norm": 1.1319475173950195, "learning_rate": 3.617151976927113e-06, "loss": 0.1453539878129959, "step": 3028 }, { "epoch": 0.7863447559709242, "grad_norm": 1.7717111110687256, "learning_rate": 3.6162789706978287e-06, "loss": 0.13320666551589966, "step": 3029 }, { "epoch": 0.7866043613707165, "grad_norm": 2.578726291656494, "learning_rate": 3.6154058113302566e-06, "loss": 0.16903601586818695, "step": 3030 }, { "epoch": 0.7868639667705088, "grad_norm": 1.3588800430297852, "learning_rate": 3.614532498978753e-06, "loss": 0.12031905353069305, "step": 3031 }, { "epoch": 0.7871235721703012, "grad_norm": 2.0405778884887695, "learning_rate": 3.6136590337976985e-06, "loss": 0.12039274722337723, "step": 3032 }, { "epoch": 0.7873831775700935, "grad_norm": 2.317875385284424, "learning_rate": 3.6127854159415014e-06, "loss": 0.12618502974510193, "step": 3033 }, { "epoch": 0.7876427829698858, "grad_norm": 2.379819869995117, "learning_rate": 3.611911645564599e-06, "loss": 0.14488865435123444, "step": 3034 }, { "epoch": 0.7879023883696781, "grad_norm": 0.8825043439865112, "learning_rate": 3.611037722821452e-06, "loss": 0.11534763127565384, "step": 3035 }, { "epoch": 0.7881619937694704, "grad_norm": 1.0657567977905273, "learning_rate": 3.610163647866552e-06, "loss": 0.13759344816207886, "step": 3036 }, { "epoch": 0.7884215991692627, "grad_norm": 0.908158540725708, "learning_rate": 3.6092894208544128e-06, "loss": 0.12318301200866699, "step": 3037 }, { "epoch": 0.788681204569055, "grad_norm": 1.1055400371551514, "learning_rate": 3.608415041939579e-06, "loss": 0.11945571005344391, "step": 3038 }, { "epoch": 0.7889408099688473, "grad_norm": 2.499997138977051, "learning_rate": 3.6075405112766213e-06, "loss": 0.1299126148223877, "step": 3039 }, { "epoch": 0.7892004153686397, "grad_norm": 1.9304134845733643, "learning_rate": 3.606665829020137e-06, "loss": 0.15110941231250763, "step": 3040 }, { "epoch": 0.789460020768432, "grad_norm": 1.3195987939834595, "learning_rate": 3.605790995324749e-06, "loss": 0.1309405118227005, "step": 3041 }, { "epoch": 0.7897196261682243, "grad_norm": 1.34890878200531, "learning_rate": 3.604916010345107e-06, "loss": 0.10845392942428589, "step": 3042 }, { "epoch": 0.7899792315680166, "grad_norm": 0.6136237978935242, "learning_rate": 3.60404087423589e-06, "loss": 0.11903539299964905, "step": 3043 }, { "epoch": 0.7902388369678089, "grad_norm": 2.2406747341156006, "learning_rate": 3.6031655871518017e-06, "loss": 0.14419510960578918, "step": 3044 }, { "epoch": 0.7904984423676013, "grad_norm": 1.3919496536254883, "learning_rate": 3.602290149247572e-06, "loss": 0.12765900790691376, "step": 3045 }, { "epoch": 0.7907580477673936, "grad_norm": 1.1999568939208984, "learning_rate": 3.6014145606779586e-06, "loss": 0.132629856467247, "step": 3046 }, { "epoch": 0.7910176531671859, "grad_norm": 1.6609636545181274, "learning_rate": 3.600538821597746e-06, "loss": 0.10953126102685928, "step": 3047 }, { "epoch": 0.7912772585669782, "grad_norm": 1.2001516819000244, "learning_rate": 3.599662932161745e-06, "loss": 0.15210509300231934, "step": 3048 }, { "epoch": 0.7915368639667705, "grad_norm": 1.7942441701889038, "learning_rate": 3.5987868925247916e-06, "loss": 0.1654585301876068, "step": 3049 }, { "epoch": 0.7917964693665628, "grad_norm": 1.020098328590393, "learning_rate": 3.597910702841751e-06, "loss": 0.12332376092672348, "step": 3050 }, { "epoch": 0.7920560747663551, "grad_norm": 1.0081992149353027, "learning_rate": 3.5970343632675115e-06, "loss": 0.12016786634922028, "step": 3051 }, { "epoch": 0.7923156801661475, "grad_norm": 1.6053615808486938, "learning_rate": 3.5961578739569926e-06, "loss": 0.13017304241657257, "step": 3052 }, { "epoch": 0.7925752855659398, "grad_norm": 0.9558069705963135, "learning_rate": 3.5952812350651346e-06, "loss": 0.12459734827280045, "step": 3053 }, { "epoch": 0.7928348909657321, "grad_norm": 1.0048869848251343, "learning_rate": 3.5944044467469095e-06, "loss": 0.11861534416675568, "step": 3054 }, { "epoch": 0.7930944963655244, "grad_norm": 1.1815505027770996, "learning_rate": 3.5935275091573124e-06, "loss": 0.12736593186855316, "step": 3055 }, { "epoch": 0.7933541017653167, "grad_norm": 0.46272292733192444, "learning_rate": 3.5926504224513655e-06, "loss": 0.1254248470067978, "step": 3056 }, { "epoch": 0.793613707165109, "grad_norm": 0.7250261902809143, "learning_rate": 3.591773186784118e-06, "loss": 0.1348804533481598, "step": 3057 }, { "epoch": 0.7938733125649013, "grad_norm": 1.2084600925445557, "learning_rate": 3.590895802310646e-06, "loss": 0.1457294225692749, "step": 3058 }, { "epoch": 0.7941329179646937, "grad_norm": 1.1095939874649048, "learning_rate": 3.590018269186049e-06, "loss": 0.12040579319000244, "step": 3059 }, { "epoch": 0.794392523364486, "grad_norm": 0.9814741611480713, "learning_rate": 3.589140587565456e-06, "loss": 0.12969136238098145, "step": 3060 }, { "epoch": 0.7946521287642783, "grad_norm": 0.9152584075927734, "learning_rate": 3.5882627576040212e-06, "loss": 0.11368738859891891, "step": 3061 }, { "epoch": 0.7949117341640706, "grad_norm": 1.2114953994750977, "learning_rate": 3.587384779456925e-06, "loss": 0.15552879869937897, "step": 3062 }, { "epoch": 0.7951713395638629, "grad_norm": 0.41996264457702637, "learning_rate": 3.5865066532793737e-06, "loss": 0.10865435004234314, "step": 3063 }, { "epoch": 0.7954309449636553, "grad_norm": 0.8362895250320435, "learning_rate": 3.5856283792265984e-06, "loss": 0.15626606345176697, "step": 3064 }, { "epoch": 0.7956905503634476, "grad_norm": 1.204785704612732, "learning_rate": 3.5847499574538598e-06, "loss": 0.1371845155954361, "step": 3065 }, { "epoch": 0.7959501557632399, "grad_norm": 1.2757693529129028, "learning_rate": 3.5838713881164413e-06, "loss": 0.1662396341562271, "step": 3066 }, { "epoch": 0.7962097611630322, "grad_norm": 0.37812209129333496, "learning_rate": 3.582992671369655e-06, "loss": 0.12465011328458786, "step": 3067 }, { "epoch": 0.7964693665628245, "grad_norm": 2.7801952362060547, "learning_rate": 3.5821138073688373e-06, "loss": 0.13555127382278442, "step": 3068 }, { "epoch": 0.7967289719626168, "grad_norm": 0.49104365706443787, "learning_rate": 3.581234796269351e-06, "loss": 0.11452454328536987, "step": 3069 }, { "epoch": 0.7969885773624091, "grad_norm": 0.8573936223983765, "learning_rate": 3.5803556382265854e-06, "loss": 0.1326587200164795, "step": 3070 }, { "epoch": 0.7972481827622014, "grad_norm": 1.9692020416259766, "learning_rate": 3.5794763333959564e-06, "loss": 0.14809197187423706, "step": 3071 }, { "epoch": 0.7975077881619937, "grad_norm": 1.1293410062789917, "learning_rate": 3.578596881932904e-06, "loss": 0.1391453742980957, "step": 3072 }, { "epoch": 0.797767393561786, "grad_norm": 1.6681861877441406, "learning_rate": 3.5777172839928943e-06, "loss": 0.16061510145664215, "step": 3073 }, { "epoch": 0.7980269989615784, "grad_norm": 0.7192931175231934, "learning_rate": 3.5768375397314215e-06, "loss": 0.1254766434431076, "step": 3074 }, { "epoch": 0.7982866043613707, "grad_norm": 0.8185360431671143, "learning_rate": 3.5759576493040045e-06, "loss": 0.12421536445617676, "step": 3075 }, { "epoch": 0.7985462097611631, "grad_norm": 0.7551009654998779, "learning_rate": 3.5750776128661865e-06, "loss": 0.11586335301399231, "step": 3076 }, { "epoch": 0.7988058151609554, "grad_norm": 1.1147953271865845, "learning_rate": 3.5741974305735377e-06, "loss": 0.14193010330200195, "step": 3077 }, { "epoch": 0.7990654205607477, "grad_norm": 0.9815282225608826, "learning_rate": 3.5733171025816547e-06, "loss": 0.1446722149848938, "step": 3078 }, { "epoch": 0.79932502596054, "grad_norm": 1.2914173603057861, "learning_rate": 3.5724366290461604e-06, "loss": 0.12240670621395111, "step": 3079 }, { "epoch": 0.7995846313603323, "grad_norm": 1.0760459899902344, "learning_rate": 3.5715560101227013e-06, "loss": 0.1763843297958374, "step": 3080 }, { "epoch": 0.7998442367601246, "grad_norm": 0.9005177617073059, "learning_rate": 3.5706752459669506e-06, "loss": 0.13733884692192078, "step": 3081 }, { "epoch": 0.8001038421599169, "grad_norm": 1.0331199169158936, "learning_rate": 3.5697943367346066e-06, "loss": 0.15240302681922913, "step": 3082 }, { "epoch": 0.8003634475597092, "grad_norm": 0.838158130645752, "learning_rate": 3.5689132825813953e-06, "loss": 0.12587249279022217, "step": 3083 }, { "epoch": 0.8006230529595015, "grad_norm": 1.1210190057754517, "learning_rate": 3.5680320836630667e-06, "loss": 0.13428866863250732, "step": 3084 }, { "epoch": 0.8008826583592938, "grad_norm": 0.8704268932342529, "learning_rate": 3.567150740135396e-06, "loss": 0.11718801409006119, "step": 3085 }, { "epoch": 0.8011422637590861, "grad_norm": 0.6199086904525757, "learning_rate": 3.566269252154184e-06, "loss": 0.11716147512197495, "step": 3086 }, { "epoch": 0.8014018691588785, "grad_norm": 1.207219123840332, "learning_rate": 3.565387619875259e-06, "loss": 0.11287416517734528, "step": 3087 }, { "epoch": 0.8016614745586709, "grad_norm": 2.697087049484253, "learning_rate": 3.564505843454472e-06, "loss": 0.17416992783546448, "step": 3088 }, { "epoch": 0.8019210799584632, "grad_norm": 0.8484975695610046, "learning_rate": 3.5636239230477025e-06, "loss": 0.13005681335926056, "step": 3089 }, { "epoch": 0.8021806853582555, "grad_norm": 0.811375081539154, "learning_rate": 3.562741858810852e-06, "loss": 0.14145523309707642, "step": 3090 }, { "epoch": 0.8024402907580478, "grad_norm": 0.9452233910560608, "learning_rate": 3.561859650899851e-06, "loss": 0.16019853949546814, "step": 3091 }, { "epoch": 0.8026998961578401, "grad_norm": 2.3500592708587646, "learning_rate": 3.560977299470653e-06, "loss": 0.11202605068683624, "step": 3092 }, { "epoch": 0.8029595015576324, "grad_norm": 1.0548012256622314, "learning_rate": 3.560094804679237e-06, "loss": 0.12872177362442017, "step": 3093 }, { "epoch": 0.8032191069574247, "grad_norm": 1.0440280437469482, "learning_rate": 3.5592121666816083e-06, "loss": 0.12520132958889008, "step": 3094 }, { "epoch": 0.803478712357217, "grad_norm": 0.6418920159339905, "learning_rate": 3.5583293856337963e-06, "loss": 0.11613583564758301, "step": 3095 }, { "epoch": 0.8037383177570093, "grad_norm": 1.2600762844085693, "learning_rate": 3.557446461691858e-06, "loss": 0.14444145560264587, "step": 3096 }, { "epoch": 0.8039979231568016, "grad_norm": 1.033321738243103, "learning_rate": 3.5565633950118727e-06, "loss": 0.1250264197587967, "step": 3097 }, { "epoch": 0.8042575285565939, "grad_norm": 1.1866713762283325, "learning_rate": 3.5556801857499474e-06, "loss": 0.14299434423446655, "step": 3098 }, { "epoch": 0.8045171339563862, "grad_norm": 1.5829029083251953, "learning_rate": 3.554796834062213e-06, "loss": 0.1313110888004303, "step": 3099 }, { "epoch": 0.8047767393561787, "grad_norm": 1.3741346597671509, "learning_rate": 3.553913340104826e-06, "loss": 0.10961742699146271, "step": 3100 }, { "epoch": 0.805036344755971, "grad_norm": 1.5932445526123047, "learning_rate": 3.5530297040339667e-06, "loss": 0.12675036489963531, "step": 3101 }, { "epoch": 0.8052959501557633, "grad_norm": 0.8509587049484253, "learning_rate": 3.5521459260058435e-06, "loss": 0.10136590898036957, "step": 3102 }, { "epoch": 0.8055555555555556, "grad_norm": 0.5014936923980713, "learning_rate": 3.5512620061766866e-06, "loss": 0.12179064005613327, "step": 3103 }, { "epoch": 0.8058151609553479, "grad_norm": 0.7391080856323242, "learning_rate": 3.550377944702755e-06, "loss": 0.11362196505069733, "step": 3104 }, { "epoch": 0.8060747663551402, "grad_norm": 1.4421719312667847, "learning_rate": 3.549493741740327e-06, "loss": 0.12854351103305817, "step": 3105 }, { "epoch": 0.8063343717549325, "grad_norm": 0.9022886157035828, "learning_rate": 3.5486093974457133e-06, "loss": 0.13056877255439758, "step": 3106 }, { "epoch": 0.8065939771547248, "grad_norm": 1.495040774345398, "learning_rate": 3.5477249119752426e-06, "loss": 0.13603505492210388, "step": 3107 }, { "epoch": 0.8068535825545171, "grad_norm": 1.3045531511306763, "learning_rate": 3.546840285485273e-06, "loss": 0.14498814940452576, "step": 3108 }, { "epoch": 0.8071131879543094, "grad_norm": 0.7827392220497131, "learning_rate": 3.5459555181321856e-06, "loss": 0.1223001480102539, "step": 3109 }, { "epoch": 0.8073727933541017, "grad_norm": 0.5169654488563538, "learning_rate": 3.545070610072389e-06, "loss": 0.1134672686457634, "step": 3110 }, { "epoch": 0.807632398753894, "grad_norm": 0.8241338729858398, "learning_rate": 3.5441855614623126e-06, "loss": 0.1328059583902359, "step": 3111 }, { "epoch": 0.8078920041536864, "grad_norm": 1.800590991973877, "learning_rate": 3.5433003724584143e-06, "loss": 0.16485148668289185, "step": 3112 }, { "epoch": 0.8081516095534788, "grad_norm": 0.969745397567749, "learning_rate": 3.5424150432171727e-06, "loss": 0.1597498208284378, "step": 3113 }, { "epoch": 0.8084112149532711, "grad_norm": 1.176658272743225, "learning_rate": 3.5415295738950965e-06, "loss": 0.1254729926586151, "step": 3114 }, { "epoch": 0.8086708203530634, "grad_norm": 0.9667266607284546, "learning_rate": 3.5406439646487144e-06, "loss": 0.14237073063850403, "step": 3115 }, { "epoch": 0.8089304257528557, "grad_norm": 0.9535409212112427, "learning_rate": 3.539758215634584e-06, "loss": 0.13666635751724243, "step": 3116 }, { "epoch": 0.809190031152648, "grad_norm": 0.43555259704589844, "learning_rate": 3.5388723270092824e-06, "loss": 0.11568447947502136, "step": 3117 }, { "epoch": 0.8094496365524403, "grad_norm": 1.1769143342971802, "learning_rate": 3.5379862989294174e-06, "loss": 0.16222837567329407, "step": 3118 }, { "epoch": 0.8097092419522326, "grad_norm": 1.1328140497207642, "learning_rate": 3.537100131551617e-06, "loss": 0.1379791796207428, "step": 3119 }, { "epoch": 0.8099688473520249, "grad_norm": 1.559218168258667, "learning_rate": 3.536213825032535e-06, "loss": 0.15050454437732697, "step": 3120 }, { "epoch": 0.8102284527518172, "grad_norm": 1.2047795057296753, "learning_rate": 3.535327379528851e-06, "loss": 0.13872915506362915, "step": 3121 }, { "epoch": 0.8104880581516095, "grad_norm": 0.9716189503669739, "learning_rate": 3.5344407951972687e-06, "loss": 0.11117549985647202, "step": 3122 }, { "epoch": 0.8107476635514018, "grad_norm": 1.1762542724609375, "learning_rate": 3.5335540721945145e-06, "loss": 0.15822476148605347, "step": 3123 }, { "epoch": 0.8110072689511942, "grad_norm": 1.3227410316467285, "learning_rate": 3.532667210677341e-06, "loss": 0.15588319301605225, "step": 3124 }, { "epoch": 0.8112668743509865, "grad_norm": 1.0124759674072266, "learning_rate": 3.531780210802526e-06, "loss": 0.13713738322257996, "step": 3125 }, { "epoch": 0.8115264797507789, "grad_norm": 1.053591012954712, "learning_rate": 3.5308930727268693e-06, "loss": 0.1255495697259903, "step": 3126 }, { "epoch": 0.8117860851505712, "grad_norm": 1.2858771085739136, "learning_rate": 3.530005796607198e-06, "loss": 0.14645442366600037, "step": 3127 }, { "epoch": 0.8120456905503635, "grad_norm": 3.2934679985046387, "learning_rate": 3.529118382600361e-06, "loss": 0.15403886139392853, "step": 3128 }, { "epoch": 0.8123052959501558, "grad_norm": 0.458841472864151, "learning_rate": 3.5282308308632348e-06, "loss": 0.12121519446372986, "step": 3129 }, { "epoch": 0.8125649013499481, "grad_norm": 1.9653829336166382, "learning_rate": 3.527343141552715e-06, "loss": 0.12970200181007385, "step": 3130 }, { "epoch": 0.8128245067497404, "grad_norm": 0.5408825874328613, "learning_rate": 3.526455314825728e-06, "loss": 0.13059160113334656, "step": 3131 }, { "epoch": 0.8130841121495327, "grad_norm": 0.5114349722862244, "learning_rate": 3.525567350839219e-06, "loss": 0.11618991196155548, "step": 3132 }, { "epoch": 0.813343717549325, "grad_norm": 1.1135517358779907, "learning_rate": 3.5246792497501615e-06, "loss": 0.15336257219314575, "step": 3133 }, { "epoch": 0.8136033229491173, "grad_norm": 1.3338738679885864, "learning_rate": 3.5237910117155494e-06, "loss": 0.12280502170324326, "step": 3134 }, { "epoch": 0.8138629283489096, "grad_norm": 1.086597204208374, "learning_rate": 3.5229026368924043e-06, "loss": 0.12526321411132812, "step": 3135 }, { "epoch": 0.814122533748702, "grad_norm": 1.0041590929031372, "learning_rate": 3.5220141254377697e-06, "loss": 0.1299581676721573, "step": 3136 }, { "epoch": 0.8143821391484943, "grad_norm": 1.163253664970398, "learning_rate": 3.521125477508715e-06, "loss": 0.12361900508403778, "step": 3137 }, { "epoch": 0.8146417445482866, "grad_norm": 1.079283595085144, "learning_rate": 3.5202366932623327e-06, "loss": 0.12449367344379425, "step": 3138 }, { "epoch": 0.814901349948079, "grad_norm": 0.5517918467521667, "learning_rate": 3.519347772855738e-06, "loss": 0.11641529947519302, "step": 3139 }, { "epoch": 0.8151609553478713, "grad_norm": 1.4822416305541992, "learning_rate": 3.518458716446074e-06, "loss": 0.14783824980258942, "step": 3140 }, { "epoch": 0.8154205607476636, "grad_norm": 1.0522634983062744, "learning_rate": 3.517569524190503e-06, "loss": 0.11564965546131134, "step": 3141 }, { "epoch": 0.8156801661474559, "grad_norm": 0.9341430068016052, "learning_rate": 3.516680196246216e-06, "loss": 0.13992008566856384, "step": 3142 }, { "epoch": 0.8159397715472482, "grad_norm": 0.7859828472137451, "learning_rate": 3.5157907327704244e-06, "loss": 0.13001307845115662, "step": 3143 }, { "epoch": 0.8161993769470405, "grad_norm": 1.786096215248108, "learning_rate": 3.514901133920366e-06, "loss": 0.13114574551582336, "step": 3144 }, { "epoch": 0.8164589823468328, "grad_norm": 0.6815663576126099, "learning_rate": 3.5140113998533015e-06, "loss": 0.13043150305747986, "step": 3145 }, { "epoch": 0.8167185877466251, "grad_norm": 0.8178061842918396, "learning_rate": 3.513121530726515e-06, "loss": 0.1217510774731636, "step": 3146 }, { "epoch": 0.8169781931464174, "grad_norm": 0.6290716528892517, "learning_rate": 3.5122315266973146e-06, "loss": 0.13081182539463043, "step": 3147 }, { "epoch": 0.8172377985462098, "grad_norm": 1.346726894378662, "learning_rate": 3.511341387923032e-06, "loss": 0.13838928937911987, "step": 3148 }, { "epoch": 0.8174974039460021, "grad_norm": 2.1755640506744385, "learning_rate": 3.5104511145610255e-06, "loss": 0.16773946583271027, "step": 3149 }, { "epoch": 0.8177570093457944, "grad_norm": 1.8072065114974976, "learning_rate": 3.509560706768674e-06, "loss": 0.15764570236206055, "step": 3150 }, { "epoch": 0.8180166147455867, "grad_norm": 1.0396183729171753, "learning_rate": 3.508670164703382e-06, "loss": 0.12778720259666443, "step": 3151 }, { "epoch": 0.818276220145379, "grad_norm": 0.8658239841461182, "learning_rate": 3.5077794885225747e-06, "loss": 0.11208626627922058, "step": 3152 }, { "epoch": 0.8185358255451713, "grad_norm": 0.9023687839508057, "learning_rate": 3.5068886783837047e-06, "loss": 0.11767293512821198, "step": 3153 }, { "epoch": 0.8187954309449637, "grad_norm": 1.2312670946121216, "learning_rate": 3.505997734444248e-06, "loss": 0.12272192537784576, "step": 3154 }, { "epoch": 0.819055036344756, "grad_norm": 1.2021063566207886, "learning_rate": 3.505106656861701e-06, "loss": 0.13456059992313385, "step": 3155 }, { "epoch": 0.8193146417445483, "grad_norm": 0.6731691956520081, "learning_rate": 3.5042154457935867e-06, "loss": 0.12614132463932037, "step": 3156 }, { "epoch": 0.8195742471443406, "grad_norm": 1.135693073272705, "learning_rate": 3.5033241013974512e-06, "loss": 0.12439552694559097, "step": 3157 }, { "epoch": 0.8198338525441329, "grad_norm": 1.2233082056045532, "learning_rate": 3.502432623830863e-06, "loss": 0.14633630216121674, "step": 3158 }, { "epoch": 0.8200934579439252, "grad_norm": 0.6798512935638428, "learning_rate": 3.5015410132514153e-06, "loss": 0.11948370188474655, "step": 3159 }, { "epoch": 0.8203530633437176, "grad_norm": 0.4656713306903839, "learning_rate": 3.5006492698167256e-06, "loss": 0.11884568631649017, "step": 3160 }, { "epoch": 0.8206126687435099, "grad_norm": 0.632023811340332, "learning_rate": 3.499757393684431e-06, "loss": 0.11647961288690567, "step": 3161 }, { "epoch": 0.8208722741433022, "grad_norm": 0.4411171078681946, "learning_rate": 3.4988653850121975e-06, "loss": 0.12160660326480865, "step": 3162 }, { "epoch": 0.8211318795430945, "grad_norm": 0.555399477481842, "learning_rate": 3.4979732439577103e-06, "loss": 0.12307199835777283, "step": 3163 }, { "epoch": 0.8213914849428868, "grad_norm": 0.9403476119041443, "learning_rate": 3.4970809706786802e-06, "loss": 0.15212678909301758, "step": 3164 }, { "epoch": 0.8216510903426791, "grad_norm": 0.5645343661308289, "learning_rate": 3.49618856533284e-06, "loss": 0.11662258207798004, "step": 3165 }, { "epoch": 0.8219106957424714, "grad_norm": 1.4279370307922363, "learning_rate": 3.4952960280779463e-06, "loss": 0.12169738858938217, "step": 3166 }, { "epoch": 0.8221703011422637, "grad_norm": 1.194346308708191, "learning_rate": 3.494403359071781e-06, "loss": 0.12310844659805298, "step": 3167 }, { "epoch": 0.822429906542056, "grad_norm": 1.0832116603851318, "learning_rate": 3.4935105584721455e-06, "loss": 0.15876993536949158, "step": 3168 }, { "epoch": 0.8226895119418484, "grad_norm": 0.8213955760002136, "learning_rate": 3.4926176264368683e-06, "loss": 0.13186302781105042, "step": 3169 }, { "epoch": 0.8229491173416407, "grad_norm": 0.9754164218902588, "learning_rate": 3.491724563123798e-06, "loss": 0.10968123376369476, "step": 3170 }, { "epoch": 0.823208722741433, "grad_norm": 0.7596884965896606, "learning_rate": 3.4908313686908074e-06, "loss": 0.11976739019155502, "step": 3171 }, { "epoch": 0.8234683281412254, "grad_norm": 0.6953774690628052, "learning_rate": 3.4899380432957948e-06, "loss": 0.11199498176574707, "step": 3172 }, { "epoch": 0.8237279335410177, "grad_norm": 0.5495917201042175, "learning_rate": 3.4890445870966783e-06, "loss": 0.12346706539392471, "step": 3173 }, { "epoch": 0.82398753894081, "grad_norm": 1.263890027999878, "learning_rate": 3.4881510002514e-06, "loss": 0.12269724905490875, "step": 3174 }, { "epoch": 0.8242471443406023, "grad_norm": 0.6002413630485535, "learning_rate": 3.487257282917927e-06, "loss": 0.1188378632068634, "step": 3175 }, { "epoch": 0.8245067497403946, "grad_norm": 0.5680797696113586, "learning_rate": 3.4863634352542465e-06, "loss": 0.1131473183631897, "step": 3176 }, { "epoch": 0.8247663551401869, "grad_norm": 1.1112943887710571, "learning_rate": 3.4854694574183716e-06, "loss": 0.12847942113876343, "step": 3177 }, { "epoch": 0.8250259605399792, "grad_norm": 0.5867354869842529, "learning_rate": 3.4845753495683366e-06, "loss": 0.10250163078308105, "step": 3178 }, { "epoch": 0.8252855659397715, "grad_norm": 0.8389600515365601, "learning_rate": 3.4836811118621995e-06, "loss": 0.12986060976982117, "step": 3179 }, { "epoch": 0.8255451713395638, "grad_norm": 1.2460232973098755, "learning_rate": 3.482786744458041e-06, "loss": 0.13326452672481537, "step": 3180 }, { "epoch": 0.8258047767393562, "grad_norm": 0.611691951751709, "learning_rate": 3.4818922475139643e-06, "loss": 0.10813353210687637, "step": 3181 }, { "epoch": 0.8260643821391485, "grad_norm": 1.0969302654266357, "learning_rate": 3.4809976211880963e-06, "loss": 0.1575036346912384, "step": 3182 }, { "epoch": 0.8263239875389408, "grad_norm": 0.9110082983970642, "learning_rate": 3.4801028656385867e-06, "loss": 0.12094863504171371, "step": 3183 }, { "epoch": 0.8265835929387332, "grad_norm": 0.9121484160423279, "learning_rate": 3.4792079810236078e-06, "loss": 0.124091237783432, "step": 3184 }, { "epoch": 0.8268431983385255, "grad_norm": 0.823767364025116, "learning_rate": 3.4783129675013545e-06, "loss": 0.11310813575983047, "step": 3185 }, { "epoch": 0.8271028037383178, "grad_norm": 0.7640917897224426, "learning_rate": 3.4774178252300445e-06, "loss": 0.13589513301849365, "step": 3186 }, { "epoch": 0.8273624091381101, "grad_norm": 1.041934847831726, "learning_rate": 3.476522554367918e-06, "loss": 0.1278075873851776, "step": 3187 }, { "epoch": 0.8276220145379024, "grad_norm": 1.0280208587646484, "learning_rate": 3.4756271550732397e-06, "loss": 0.12600909173488617, "step": 3188 }, { "epoch": 0.8278816199376947, "grad_norm": 0.6748073697090149, "learning_rate": 3.474731627504296e-06, "loss": 0.12056758254766464, "step": 3189 }, { "epoch": 0.828141225337487, "grad_norm": 1.1098926067352295, "learning_rate": 3.4738359718193932e-06, "loss": 0.1461021602153778, "step": 3190 }, { "epoch": 0.8284008307372793, "grad_norm": 0.4327656924724579, "learning_rate": 3.4729401881768655e-06, "loss": 0.10877739638090134, "step": 3191 }, { "epoch": 0.8286604361370716, "grad_norm": 0.5205568075180054, "learning_rate": 3.4720442767350637e-06, "loss": 0.11918169260025024, "step": 3192 }, { "epoch": 0.8289200415368639, "grad_norm": 1.8090565204620361, "learning_rate": 3.4711482376523687e-06, "loss": 0.15546561777591705, "step": 3193 }, { "epoch": 0.8291796469366562, "grad_norm": 0.7410904169082642, "learning_rate": 3.4702520710871757e-06, "loss": 0.11675897240638733, "step": 3194 }, { "epoch": 0.8294392523364486, "grad_norm": 1.583964467048645, "learning_rate": 3.469355777197909e-06, "loss": 0.12047266960144043, "step": 3195 }, { "epoch": 0.829698857736241, "grad_norm": 1.2715284824371338, "learning_rate": 3.468459356143012e-06, "loss": 0.11899659037590027, "step": 3196 }, { "epoch": 0.8299584631360333, "grad_norm": 0.8225136995315552, "learning_rate": 3.4675628080809508e-06, "loss": 0.1284695267677307, "step": 3197 }, { "epoch": 0.8302180685358256, "grad_norm": 1.0102434158325195, "learning_rate": 3.466666133170215e-06, "loss": 0.144975483417511, "step": 3198 }, { "epoch": 0.8304776739356179, "grad_norm": 1.0206055641174316, "learning_rate": 3.4657693315693177e-06, "loss": 0.1348123848438263, "step": 3199 }, { "epoch": 0.8307372793354102, "grad_norm": 1.2478829622268677, "learning_rate": 3.4648724034367903e-06, "loss": 0.13453230261802673, "step": 3200 }, { "epoch": 0.8309968847352025, "grad_norm": 1.55693519115448, "learning_rate": 3.4639753489311905e-06, "loss": 0.13896825909614563, "step": 3201 }, { "epoch": 0.8312564901349948, "grad_norm": 1.8472673892974854, "learning_rate": 3.4630781682110975e-06, "loss": 0.14649273455142975, "step": 3202 }, { "epoch": 0.8315160955347871, "grad_norm": 2.133795738220215, "learning_rate": 3.462180861435111e-06, "loss": 0.12643253803253174, "step": 3203 }, { "epoch": 0.8317757009345794, "grad_norm": 1.0578960180282593, "learning_rate": 3.4612834287618556e-06, "loss": 0.13869234919548035, "step": 3204 }, { "epoch": 0.8320353063343717, "grad_norm": 0.6341173648834229, "learning_rate": 3.4603858703499747e-06, "loss": 0.1232677474617958, "step": 3205 }, { "epoch": 0.832294911734164, "grad_norm": 1.1772186756134033, "learning_rate": 3.4594881863581397e-06, "loss": 0.1299850195646286, "step": 3206 }, { "epoch": 0.8325545171339563, "grad_norm": 0.657999575138092, "learning_rate": 3.4585903769450372e-06, "loss": 0.12591218948364258, "step": 3207 }, { "epoch": 0.8328141225337488, "grad_norm": 1.144580364227295, "learning_rate": 3.4576924422693813e-06, "loss": 0.13884121179580688, "step": 3208 }, { "epoch": 0.8330737279335411, "grad_norm": 1.3635059595108032, "learning_rate": 3.4567943824899055e-06, "loss": 0.1327594518661499, "step": 3209 }, { "epoch": 0.8333333333333334, "grad_norm": 0.7845833897590637, "learning_rate": 3.455896197765366e-06, "loss": 0.11387283354997635, "step": 3210 }, { "epoch": 0.8335929387331257, "grad_norm": 1.5286105871200562, "learning_rate": 3.454997888254541e-06, "loss": 0.17617452144622803, "step": 3211 }, { "epoch": 0.833852544132918, "grad_norm": 0.919597327709198, "learning_rate": 3.4540994541162333e-06, "loss": 0.12209280580282211, "step": 3212 }, { "epoch": 0.8341121495327103, "grad_norm": 1.2062058448791504, "learning_rate": 3.4532008955092627e-06, "loss": 0.12771117687225342, "step": 3213 }, { "epoch": 0.8343717549325026, "grad_norm": 1.2208456993103027, "learning_rate": 3.452302212592475e-06, "loss": 0.1242537796497345, "step": 3214 }, { "epoch": 0.8346313603322949, "grad_norm": 0.9995679259300232, "learning_rate": 3.4514034055247366e-06, "loss": 0.13190002739429474, "step": 3215 }, { "epoch": 0.8348909657320872, "grad_norm": 1.0464842319488525, "learning_rate": 3.450504474464937e-06, "loss": 0.1181735023856163, "step": 3216 }, { "epoch": 0.8351505711318795, "grad_norm": 1.6150388717651367, "learning_rate": 3.4496054195719847e-06, "loss": 0.14064431190490723, "step": 3217 }, { "epoch": 0.8354101765316718, "grad_norm": 1.130375862121582, "learning_rate": 3.4487062410048135e-06, "loss": 0.12814854085445404, "step": 3218 }, { "epoch": 0.8356697819314641, "grad_norm": 1.2709565162658691, "learning_rate": 3.447806938922377e-06, "loss": 0.1329893171787262, "step": 3219 }, { "epoch": 0.8359293873312564, "grad_norm": 0.7438920140266418, "learning_rate": 3.4469075134836516e-06, "loss": 0.16214948892593384, "step": 3220 }, { "epoch": 0.8361889927310489, "grad_norm": 1.1053370237350464, "learning_rate": 3.4460079648476354e-06, "loss": 0.14160913228988647, "step": 3221 }, { "epoch": 0.8364485981308412, "grad_norm": 1.4605096578598022, "learning_rate": 3.4451082931733463e-06, "loss": 0.12222766131162643, "step": 3222 }, { "epoch": 0.8367082035306335, "grad_norm": 1.1245269775390625, "learning_rate": 3.444208498619827e-06, "loss": 0.124769426882267, "step": 3223 }, { "epoch": 0.8369678089304258, "grad_norm": 0.5297252535820007, "learning_rate": 3.4433085813461407e-06, "loss": 0.12249191105365753, "step": 3224 }, { "epoch": 0.8372274143302181, "grad_norm": 0.7034416198730469, "learning_rate": 3.4424085415113722e-06, "loss": 0.10644441843032837, "step": 3225 }, { "epoch": 0.8374870197300104, "grad_norm": 0.5493385195732117, "learning_rate": 3.4415083792746284e-06, "loss": 0.11908281594514847, "step": 3226 }, { "epoch": 0.8377466251298027, "grad_norm": 0.6008524894714355, "learning_rate": 3.440608094795035e-06, "loss": 0.11254777759313583, "step": 3227 }, { "epoch": 0.838006230529595, "grad_norm": 1.0974371433258057, "learning_rate": 3.4397076882317438e-06, "loss": 0.12448026239871979, "step": 3228 }, { "epoch": 0.8382658359293873, "grad_norm": 0.623895525932312, "learning_rate": 3.438807159743926e-06, "loss": 0.1043727844953537, "step": 3229 }, { "epoch": 0.8385254413291796, "grad_norm": 0.47930464148521423, "learning_rate": 3.4379065094907734e-06, "loss": 0.11517924070358276, "step": 3230 }, { "epoch": 0.8387850467289719, "grad_norm": 1.2181111574172974, "learning_rate": 3.4370057376315e-06, "loss": 0.1366053819656372, "step": 3231 }, { "epoch": 0.8390446521287642, "grad_norm": 1.1408894062042236, "learning_rate": 3.4361048443253443e-06, "loss": 0.13100609183311462, "step": 3232 }, { "epoch": 0.8393042575285566, "grad_norm": 1.1757296323776245, "learning_rate": 3.4352038297315616e-06, "loss": 0.1344655454158783, "step": 3233 }, { "epoch": 0.839563862928349, "grad_norm": 0.9047799110412598, "learning_rate": 3.4343026940094297e-06, "loss": 0.1152629628777504, "step": 3234 }, { "epoch": 0.8398234683281413, "grad_norm": 1.4750471115112305, "learning_rate": 3.433401437318251e-06, "loss": 0.16212666034698486, "step": 3235 }, { "epoch": 0.8400830737279336, "grad_norm": 0.5390756726264954, "learning_rate": 3.4325000598173454e-06, "loss": 0.11317233741283417, "step": 3236 }, { "epoch": 0.8403426791277259, "grad_norm": 0.37078580260276794, "learning_rate": 3.4315985616660572e-06, "loss": 0.12118791043758392, "step": 3237 }, { "epoch": 0.8406022845275182, "grad_norm": 1.0477676391601562, "learning_rate": 3.430696943023749e-06, "loss": 0.14118549227714539, "step": 3238 }, { "epoch": 0.8408618899273105, "grad_norm": 1.028696060180664, "learning_rate": 3.429795204049808e-06, "loss": 0.14759337902069092, "step": 3239 }, { "epoch": 0.8411214953271028, "grad_norm": 1.9480197429656982, "learning_rate": 3.428893344903639e-06, "loss": 0.15694445371627808, "step": 3240 }, { "epoch": 0.8413811007268951, "grad_norm": 1.3975900411605835, "learning_rate": 3.427991365744672e-06, "loss": 0.10839153826236725, "step": 3241 }, { "epoch": 0.8416407061266874, "grad_norm": 1.2520310878753662, "learning_rate": 3.4270892667323553e-06, "loss": 0.147682785987854, "step": 3242 }, { "epoch": 0.8419003115264797, "grad_norm": 1.2176687717437744, "learning_rate": 3.42618704802616e-06, "loss": 0.1519448459148407, "step": 3243 }, { "epoch": 0.842159916926272, "grad_norm": 2.2425246238708496, "learning_rate": 3.4252847097855764e-06, "loss": 0.13789863884449005, "step": 3244 }, { "epoch": 0.8424195223260644, "grad_norm": 0.8929219841957092, "learning_rate": 3.4243822521701183e-06, "loss": 0.12061379104852676, "step": 3245 }, { "epoch": 0.8426791277258567, "grad_norm": 0.7586475610733032, "learning_rate": 3.4234796753393197e-06, "loss": 0.11784620583057404, "step": 3246 }, { "epoch": 0.842938733125649, "grad_norm": 0.6565606594085693, "learning_rate": 3.4225769794527354e-06, "loss": 0.12441635131835938, "step": 3247 }, { "epoch": 0.8431983385254413, "grad_norm": 1.0750863552093506, "learning_rate": 3.4216741646699396e-06, "loss": 0.10976943373680115, "step": 3248 }, { "epoch": 0.8434579439252337, "grad_norm": 1.275319218635559, "learning_rate": 3.420771231150532e-06, "loss": 0.10209093987941742, "step": 3249 }, { "epoch": 0.843717549325026, "grad_norm": 1.7571392059326172, "learning_rate": 3.4198681790541296e-06, "loss": 0.12393023073673248, "step": 3250 }, { "epoch": 0.8439771547248183, "grad_norm": 1.47515070438385, "learning_rate": 3.418965008540371e-06, "loss": 0.15682625770568848, "step": 3251 }, { "epoch": 0.8442367601246106, "grad_norm": 0.5697008967399597, "learning_rate": 3.418061719768916e-06, "loss": 0.12029551714658737, "step": 3252 }, { "epoch": 0.8444963655244029, "grad_norm": 1.4389498233795166, "learning_rate": 3.4171583128994453e-06, "loss": 0.144536554813385, "step": 3253 }, { "epoch": 0.8447559709241952, "grad_norm": 0.4840291440486908, "learning_rate": 3.41625478809166e-06, "loss": 0.12016084045171738, "step": 3254 }, { "epoch": 0.8450155763239875, "grad_norm": 0.8190675377845764, "learning_rate": 3.4153511455052844e-06, "loss": 0.1205066367983818, "step": 3255 }, { "epoch": 0.8452751817237798, "grad_norm": 0.5838197469711304, "learning_rate": 3.414447385300061e-06, "loss": 0.12687824666500092, "step": 3256 }, { "epoch": 0.8455347871235722, "grad_norm": 1.0275698900222778, "learning_rate": 3.4135435076357537e-06, "loss": 0.14038710296154022, "step": 3257 }, { "epoch": 0.8457943925233645, "grad_norm": 1.1043179035186768, "learning_rate": 3.412639512672147e-06, "loss": 0.1124916821718216, "step": 3258 }, { "epoch": 0.8460539979231568, "grad_norm": 1.0888681411743164, "learning_rate": 3.411735400569046e-06, "loss": 0.12448568642139435, "step": 3259 }, { "epoch": 0.8463136033229491, "grad_norm": 0.7995297312736511, "learning_rate": 3.41083117148628e-06, "loss": 0.1290043145418167, "step": 3260 }, { "epoch": 0.8465732087227414, "grad_norm": 1.063839077949524, "learning_rate": 3.409926825583692e-06, "loss": 0.14339667558670044, "step": 3261 }, { "epoch": 0.8468328141225338, "grad_norm": 1.3066468238830566, "learning_rate": 3.4090223630211523e-06, "loss": 0.08710834383964539, "step": 3262 }, { "epoch": 0.8470924195223261, "grad_norm": 0.8633002042770386, "learning_rate": 3.4081177839585485e-06, "loss": 0.12289538234472275, "step": 3263 }, { "epoch": 0.8473520249221184, "grad_norm": 0.9349656701087952, "learning_rate": 3.407213088555789e-06, "loss": 0.122114397585392, "step": 3264 }, { "epoch": 0.8476116303219107, "grad_norm": 1.2093687057495117, "learning_rate": 3.4063082769728045e-06, "loss": 0.12121149897575378, "step": 3265 }, { "epoch": 0.847871235721703, "grad_norm": 1.433965802192688, "learning_rate": 3.405403349369544e-06, "loss": 0.11649468541145325, "step": 3266 }, { "epoch": 0.8481308411214953, "grad_norm": 1.6414896249771118, "learning_rate": 3.4044983059059766e-06, "loss": 0.1438460648059845, "step": 3267 }, { "epoch": 0.8483904465212876, "grad_norm": 0.700115442276001, "learning_rate": 3.4035931467420964e-06, "loss": 0.14234259724617004, "step": 3268 }, { "epoch": 0.84865005192108, "grad_norm": 0.94774329662323, "learning_rate": 3.402687872037912e-06, "loss": 0.1282399594783783, "step": 3269 }, { "epoch": 0.8489096573208723, "grad_norm": 2.7171716690063477, "learning_rate": 3.401782481953457e-06, "loss": 0.14247441291809082, "step": 3270 }, { "epoch": 0.8491692627206646, "grad_norm": 0.8624356389045715, "learning_rate": 3.4008769766487826e-06, "loss": 0.12781044840812683, "step": 3271 }, { "epoch": 0.8494288681204569, "grad_norm": 0.8061146140098572, "learning_rate": 3.399971356283963e-06, "loss": 0.11994931846857071, "step": 3272 }, { "epoch": 0.8496884735202492, "grad_norm": 1.4761756658554077, "learning_rate": 3.399065621019089e-06, "loss": 0.13860592246055603, "step": 3273 }, { "epoch": 0.8499480789200415, "grad_norm": 1.3344932794570923, "learning_rate": 3.3981597710142757e-06, "loss": 0.13761265575885773, "step": 3274 }, { "epoch": 0.8502076843198338, "grad_norm": 0.4900318682193756, "learning_rate": 3.3972538064296545e-06, "loss": 0.11342832446098328, "step": 3275 }, { "epoch": 0.8504672897196262, "grad_norm": 0.4988614320755005, "learning_rate": 3.396347727425381e-06, "loss": 0.12028244137763977, "step": 3276 }, { "epoch": 0.8507268951194185, "grad_norm": 0.7784753441810608, "learning_rate": 3.3954415341616285e-06, "loss": 0.13081295788288116, "step": 3277 }, { "epoch": 0.8509865005192108, "grad_norm": 1.2207999229431152, "learning_rate": 3.394535226798592e-06, "loss": 0.09603434801101685, "step": 3278 }, { "epoch": 0.8512461059190031, "grad_norm": 1.116332769393921, "learning_rate": 3.393628805496485e-06, "loss": 0.11340773850679398, "step": 3279 }, { "epoch": 0.8515057113187954, "grad_norm": 0.50827956199646, "learning_rate": 3.3927222704155415e-06, "loss": 0.12205158919095993, "step": 3280 }, { "epoch": 0.8517653167185878, "grad_norm": 0.7226197719573975, "learning_rate": 3.3918156217160182e-06, "loss": 0.1153540164232254, "step": 3281 }, { "epoch": 0.8520249221183801, "grad_norm": 0.7372469902038574, "learning_rate": 3.3909088595581884e-06, "loss": 0.1252022385597229, "step": 3282 }, { "epoch": 0.8522845275181724, "grad_norm": 0.9349359273910522, "learning_rate": 3.3900019841023476e-06, "loss": 0.11568127572536469, "step": 3283 }, { "epoch": 0.8525441329179647, "grad_norm": 1.747983694076538, "learning_rate": 3.389094995508809e-06, "loss": 0.10644505172967911, "step": 3284 }, { "epoch": 0.852803738317757, "grad_norm": 1.8002885580062866, "learning_rate": 3.3881878939379086e-06, "loss": 0.12764602899551392, "step": 3285 }, { "epoch": 0.8530633437175493, "grad_norm": 1.900203824043274, "learning_rate": 3.3872806795500014e-06, "loss": 0.1245010495185852, "step": 3286 }, { "epoch": 0.8533229491173416, "grad_norm": 1.4222685098648071, "learning_rate": 3.386373352505462e-06, "loss": 0.14054271578788757, "step": 3287 }, { "epoch": 0.8535825545171339, "grad_norm": 0.6755129098892212, "learning_rate": 3.385465912964685e-06, "loss": 0.1329725682735443, "step": 3288 }, { "epoch": 0.8538421599169262, "grad_norm": 1.459843635559082, "learning_rate": 3.3845583610880844e-06, "loss": 0.18358682096004486, "step": 3289 }, { "epoch": 0.8541017653167186, "grad_norm": 1.5636603832244873, "learning_rate": 3.383650697036095e-06, "loss": 0.11088328063488007, "step": 3290 }, { "epoch": 0.8543613707165109, "grad_norm": 0.6336997151374817, "learning_rate": 3.3827429209691722e-06, "loss": 0.11370953917503357, "step": 3291 }, { "epoch": 0.8546209761163032, "grad_norm": 0.43373197317123413, "learning_rate": 3.3818350330477874e-06, "loss": 0.12474130094051361, "step": 3292 }, { "epoch": 0.8548805815160956, "grad_norm": 1.5050041675567627, "learning_rate": 3.3809270334324364e-06, "loss": 0.14024382829666138, "step": 3293 }, { "epoch": 0.8551401869158879, "grad_norm": 0.756847620010376, "learning_rate": 3.3800189222836316e-06, "loss": 0.11662571877241135, "step": 3294 }, { "epoch": 0.8553997923156802, "grad_norm": 0.8584235310554504, "learning_rate": 3.3791106997619078e-06, "loss": 0.10516568273305893, "step": 3295 }, { "epoch": 0.8556593977154725, "grad_norm": 0.8668588995933533, "learning_rate": 3.378202366027817e-06, "loss": 0.13765181601047516, "step": 3296 }, { "epoch": 0.8559190031152648, "grad_norm": 1.3760254383087158, "learning_rate": 3.3772939212419314e-06, "loss": 0.1367536336183548, "step": 3297 }, { "epoch": 0.8561786085150571, "grad_norm": 0.864259660243988, "learning_rate": 3.3763853655648436e-06, "loss": 0.1226017102599144, "step": 3298 }, { "epoch": 0.8564382139148494, "grad_norm": 1.1935184001922607, "learning_rate": 3.3754766991571657e-06, "loss": 0.10600963234901428, "step": 3299 }, { "epoch": 0.8566978193146417, "grad_norm": 1.449960470199585, "learning_rate": 3.3745679221795284e-06, "loss": 0.13893643021583557, "step": 3300 }, { "epoch": 0.856957424714434, "grad_norm": 0.4811531901359558, "learning_rate": 3.3736590347925842e-06, "loss": 0.117829829454422, "step": 3301 }, { "epoch": 0.8572170301142263, "grad_norm": 0.6361599564552307, "learning_rate": 3.3727500371570014e-06, "loss": 0.10995806753635406, "step": 3302 }, { "epoch": 0.8574766355140186, "grad_norm": 0.4670996069908142, "learning_rate": 3.371840929433472e-06, "loss": 0.14082878828048706, "step": 3303 }, { "epoch": 0.857736240913811, "grad_norm": 0.8077273368835449, "learning_rate": 3.370931711782704e-06, "loss": 0.12527501583099365, "step": 3304 }, { "epoch": 0.8579958463136034, "grad_norm": 0.8762412071228027, "learning_rate": 3.370022384365427e-06, "loss": 0.11853480339050293, "step": 3305 }, { "epoch": 0.8582554517133957, "grad_norm": 0.8890673518180847, "learning_rate": 3.3691129473423895e-06, "loss": 0.11991079151630402, "step": 3306 }, { "epoch": 0.858515057113188, "grad_norm": 1.1972544193267822, "learning_rate": 3.368203400874358e-06, "loss": 0.1543772965669632, "step": 3307 }, { "epoch": 0.8587746625129803, "grad_norm": 0.7511789798736572, "learning_rate": 3.3672937451221206e-06, "loss": 0.10770942270755768, "step": 3308 }, { "epoch": 0.8590342679127726, "grad_norm": 1.4628037214279175, "learning_rate": 3.366383980246483e-06, "loss": 0.1546071171760559, "step": 3309 }, { "epoch": 0.8592938733125649, "grad_norm": 1.6413733959197998, "learning_rate": 3.3654741064082712e-06, "loss": 0.13933098316192627, "step": 3310 }, { "epoch": 0.8595534787123572, "grad_norm": 1.2488373517990112, "learning_rate": 3.3645641237683295e-06, "loss": 0.12139400839805603, "step": 3311 }, { "epoch": 0.8598130841121495, "grad_norm": 1.3037617206573486, "learning_rate": 3.363654032487523e-06, "loss": 0.14559593796730042, "step": 3312 }, { "epoch": 0.8600726895119418, "grad_norm": 0.5080331563949585, "learning_rate": 3.362743832726735e-06, "loss": 0.1119336485862732, "step": 3313 }, { "epoch": 0.8603322949117341, "grad_norm": 1.706135630607605, "learning_rate": 3.3618335246468657e-06, "loss": 0.12423667311668396, "step": 3314 }, { "epoch": 0.8605919003115264, "grad_norm": 0.9481654763221741, "learning_rate": 3.360923108408839e-06, "loss": 0.12832501530647278, "step": 3315 }, { "epoch": 0.8608515057113187, "grad_norm": 0.6592422723770142, "learning_rate": 3.3600125841735955e-06, "loss": 0.12019260227680206, "step": 3316 }, { "epoch": 0.8611111111111112, "grad_norm": 0.954469621181488, "learning_rate": 3.359101952102094e-06, "loss": 0.11830579489469528, "step": 3317 }, { "epoch": 0.8613707165109035, "grad_norm": 1.161431908607483, "learning_rate": 3.3581912123553144e-06, "loss": 0.1114446222782135, "step": 3318 }, { "epoch": 0.8616303219106958, "grad_norm": 1.1424560546875, "learning_rate": 3.357280365094254e-06, "loss": 0.16546133160591125, "step": 3319 }, { "epoch": 0.8618899273104881, "grad_norm": 0.8929593563079834, "learning_rate": 3.3563694104799292e-06, "loss": 0.11667852848768234, "step": 3320 }, { "epoch": 0.8621495327102804, "grad_norm": 0.9919658899307251, "learning_rate": 3.3554583486733776e-06, "loss": 0.14950521290302277, "step": 3321 }, { "epoch": 0.8624091381100727, "grad_norm": 0.872340738773346, "learning_rate": 3.3545471798356526e-06, "loss": 0.10437881946563721, "step": 3322 }, { "epoch": 0.862668743509865, "grad_norm": 0.8963647484779358, "learning_rate": 3.353635904127829e-06, "loss": 0.13067525625228882, "step": 3323 }, { "epoch": 0.8629283489096573, "grad_norm": 0.5242427587509155, "learning_rate": 3.3527245217109986e-06, "loss": 0.11395171284675598, "step": 3324 }, { "epoch": 0.8631879543094496, "grad_norm": 0.8616454601287842, "learning_rate": 3.3518130327462734e-06, "loss": 0.1232360303401947, "step": 3325 }, { "epoch": 0.8634475597092419, "grad_norm": 0.4292108118534088, "learning_rate": 3.3509014373947834e-06, "loss": 0.11976774036884308, "step": 3326 }, { "epoch": 0.8637071651090342, "grad_norm": 0.5484509468078613, "learning_rate": 3.3499897358176793e-06, "loss": 0.11520132422447205, "step": 3327 }, { "epoch": 0.8639667705088265, "grad_norm": 1.1393741369247437, "learning_rate": 3.349077928176126e-06, "loss": 0.138082355260849, "step": 3328 }, { "epoch": 0.864226375908619, "grad_norm": 1.513007402420044, "learning_rate": 3.3481660146313133e-06, "loss": 0.11913828551769257, "step": 3329 }, { "epoch": 0.8644859813084113, "grad_norm": 1.2766673564910889, "learning_rate": 3.3472539953444453e-06, "loss": 0.13289004564285278, "step": 3330 }, { "epoch": 0.8647455867082036, "grad_norm": 1.6772098541259766, "learning_rate": 3.346341870476746e-06, "loss": 0.12334419786930084, "step": 3331 }, { "epoch": 0.8650051921079959, "grad_norm": 1.2145487070083618, "learning_rate": 3.3454296401894585e-06, "loss": 0.1183912456035614, "step": 3332 }, { "epoch": 0.8652647975077882, "grad_norm": 1.1761682033538818, "learning_rate": 3.344517304643844e-06, "loss": 0.112546905875206, "step": 3333 }, { "epoch": 0.8655244029075805, "grad_norm": 1.414306640625, "learning_rate": 3.3436048640011832e-06, "loss": 0.1442907154560089, "step": 3334 }, { "epoch": 0.8657840083073728, "grad_norm": 0.5912219285964966, "learning_rate": 3.342692318422773e-06, "loss": 0.11767653375864029, "step": 3335 }, { "epoch": 0.8660436137071651, "grad_norm": 0.6417497992515564, "learning_rate": 3.3417796680699334e-06, "loss": 0.11576254665851593, "step": 3336 }, { "epoch": 0.8663032191069574, "grad_norm": 1.540064811706543, "learning_rate": 3.3408669131039972e-06, "loss": 0.1281229555606842, "step": 3337 }, { "epoch": 0.8665628245067497, "grad_norm": 0.9546684622764587, "learning_rate": 3.33995405368632e-06, "loss": 0.14062507450580597, "step": 3338 }, { "epoch": 0.866822429906542, "grad_norm": 1.4580118656158447, "learning_rate": 3.3390410899782748e-06, "loss": 0.11716768145561218, "step": 3339 }, { "epoch": 0.8670820353063343, "grad_norm": 1.482161045074463, "learning_rate": 3.3381280221412517e-06, "loss": 0.12551036477088928, "step": 3340 }, { "epoch": 0.8673416407061267, "grad_norm": 2.142117500305176, "learning_rate": 3.3372148503366607e-06, "loss": 0.08618998527526855, "step": 3341 }, { "epoch": 0.867601246105919, "grad_norm": 0.8270699381828308, "learning_rate": 3.3363015747259294e-06, "loss": 0.12091463059186935, "step": 3342 }, { "epoch": 0.8678608515057114, "grad_norm": 1.170172095298767, "learning_rate": 3.3353881954705047e-06, "loss": 0.13523682951927185, "step": 3343 }, { "epoch": 0.8681204569055037, "grad_norm": 1.2763288021087646, "learning_rate": 3.3344747127318506e-06, "loss": 0.13447390496730804, "step": 3344 }, { "epoch": 0.868380062305296, "grad_norm": 1.210585355758667, "learning_rate": 3.33356112667145e-06, "loss": 0.1417675018310547, "step": 3345 }, { "epoch": 0.8686396677050883, "grad_norm": 1.0763131380081177, "learning_rate": 3.332647437450803e-06, "loss": 0.11178214102983475, "step": 3346 }, { "epoch": 0.8688992731048806, "grad_norm": 2.6089510917663574, "learning_rate": 3.3317336452314308e-06, "loss": 0.16416385769844055, "step": 3347 }, { "epoch": 0.8691588785046729, "grad_norm": 1.6873430013656616, "learning_rate": 3.33081975017487e-06, "loss": 0.11520731449127197, "step": 3348 }, { "epoch": 0.8694184839044652, "grad_norm": 0.5625249743461609, "learning_rate": 3.329905752442677e-06, "loss": 0.12111778557300568, "step": 3349 }, { "epoch": 0.8696780893042575, "grad_norm": 1.074654459953308, "learning_rate": 3.328991652196424e-06, "loss": 0.13760948181152344, "step": 3350 }, { "epoch": 0.8699376947040498, "grad_norm": 0.7701892256736755, "learning_rate": 3.3280774495977048e-06, "loss": 0.11858459562063217, "step": 3351 }, { "epoch": 0.8701973001038421, "grad_norm": 1.0855282545089722, "learning_rate": 3.3271631448081277e-06, "loss": 0.1193389892578125, "step": 3352 }, { "epoch": 0.8704569055036345, "grad_norm": 1.4131397008895874, "learning_rate": 3.3262487379893227e-06, "loss": 0.1132965236902237, "step": 3353 }, { "epoch": 0.8707165109034268, "grad_norm": 0.64585280418396, "learning_rate": 3.325334229302935e-06, "loss": 0.12865899503231049, "step": 3354 }, { "epoch": 0.8709761163032191, "grad_norm": 1.2060697078704834, "learning_rate": 3.3244196189106287e-06, "loss": 0.1265856772661209, "step": 3355 }, { "epoch": 0.8712357217030114, "grad_norm": 1.2101367712020874, "learning_rate": 3.3235049069740864e-06, "loss": 0.1340877264738083, "step": 3356 }, { "epoch": 0.8714953271028038, "grad_norm": 0.9130421876907349, "learning_rate": 3.3225900936550094e-06, "loss": 0.10882620513439178, "step": 3357 }, { "epoch": 0.8717549325025961, "grad_norm": 0.8444953560829163, "learning_rate": 3.3216751791151124e-06, "loss": 0.1218133196234703, "step": 3358 }, { "epoch": 0.8720145379023884, "grad_norm": 1.28744637966156, "learning_rate": 3.320760163516134e-06, "loss": 0.12915043532848358, "step": 3359 }, { "epoch": 0.8722741433021807, "grad_norm": 1.3197901248931885, "learning_rate": 3.3198450470198273e-06, "loss": 0.1256876438856125, "step": 3360 }, { "epoch": 0.872533748701973, "grad_norm": 1.1082565784454346, "learning_rate": 3.3189298297879653e-06, "loss": 0.11769422888755798, "step": 3361 }, { "epoch": 0.8727933541017653, "grad_norm": 0.855627715587616, "learning_rate": 3.3180145119823347e-06, "loss": 0.12508787214756012, "step": 3362 }, { "epoch": 0.8730529595015576, "grad_norm": 1.0931695699691772, "learning_rate": 3.3170990937647445e-06, "loss": 0.13183218240737915, "step": 3363 }, { "epoch": 0.8733125649013499, "grad_norm": 0.4344463646411896, "learning_rate": 3.316183575297019e-06, "loss": 0.10964487493038177, "step": 3364 }, { "epoch": 0.8735721703011423, "grad_norm": 1.241247296333313, "learning_rate": 3.3152679567410023e-06, "loss": 0.12472331523895264, "step": 3365 }, { "epoch": 0.8738317757009346, "grad_norm": 1.5288268327713013, "learning_rate": 3.3143522382585536e-06, "loss": 0.13726294040679932, "step": 3366 }, { "epoch": 0.8740913811007269, "grad_norm": 1.2002038955688477, "learning_rate": 3.3134364200115504e-06, "loss": 0.12254093587398529, "step": 3367 }, { "epoch": 0.8743509865005192, "grad_norm": 0.49064353108406067, "learning_rate": 3.3125205021618893e-06, "loss": 0.11638201773166656, "step": 3368 }, { "epoch": 0.8746105919003115, "grad_norm": 1.3964263200759888, "learning_rate": 3.3116044848714833e-06, "loss": 0.1335245668888092, "step": 3369 }, { "epoch": 0.8748701973001038, "grad_norm": 0.84981769323349, "learning_rate": 3.3106883683022646e-06, "loss": 0.14185285568237305, "step": 3370 }, { "epoch": 0.8751298026998962, "grad_norm": 1.28457510471344, "learning_rate": 3.309772152616179e-06, "loss": 0.12019296735525131, "step": 3371 }, { "epoch": 0.8753894080996885, "grad_norm": 1.1704912185668945, "learning_rate": 3.3088558379751945e-06, "loss": 0.12404002249240875, "step": 3372 }, { "epoch": 0.8756490134994808, "grad_norm": 0.9550379514694214, "learning_rate": 3.307939424541294e-06, "loss": 0.12231802195310593, "step": 3373 }, { "epoch": 0.8759086188992731, "grad_norm": 1.79567289352417, "learning_rate": 3.3070229124764784e-06, "loss": 0.14765870571136475, "step": 3374 }, { "epoch": 0.8761682242990654, "grad_norm": 1.366347312927246, "learning_rate": 3.3061063019427653e-06, "loss": 0.17360730469226837, "step": 3375 }, { "epoch": 0.8764278296988577, "grad_norm": 0.9168893694877625, "learning_rate": 3.305189593102192e-06, "loss": 0.11434248089790344, "step": 3376 }, { "epoch": 0.8766874350986501, "grad_norm": 1.1893569231033325, "learning_rate": 3.30427278611681e-06, "loss": 0.11493487656116486, "step": 3377 }, { "epoch": 0.8769470404984424, "grad_norm": 1.2023190259933472, "learning_rate": 3.3033558811486915e-06, "loss": 0.13539570569992065, "step": 3378 }, { "epoch": 0.8772066458982347, "grad_norm": 1.7590584754943848, "learning_rate": 3.302438878359923e-06, "loss": 0.1411457657814026, "step": 3379 }, { "epoch": 0.877466251298027, "grad_norm": 1.6475090980529785, "learning_rate": 3.30152177791261e-06, "loss": 0.11710011959075928, "step": 3380 }, { "epoch": 0.8777258566978193, "grad_norm": 1.1642985343933105, "learning_rate": 3.300604579968875e-06, "loss": 0.11093273758888245, "step": 3381 }, { "epoch": 0.8779854620976116, "grad_norm": 0.4847330152988434, "learning_rate": 3.2996872846908574e-06, "loss": 0.11354966461658478, "step": 3382 }, { "epoch": 0.8782450674974039, "grad_norm": 1.002138614654541, "learning_rate": 3.298769892240713e-06, "loss": 0.10146275162696838, "step": 3383 }, { "epoch": 0.8785046728971962, "grad_norm": 1.8354620933532715, "learning_rate": 3.2978524027806185e-06, "loss": 0.12991386651992798, "step": 3384 }, { "epoch": 0.8787642782969886, "grad_norm": 0.8995708227157593, "learning_rate": 3.2969348164727624e-06, "loss": 0.10623802244663239, "step": 3385 }, { "epoch": 0.8790238836967809, "grad_norm": 2.048926830291748, "learning_rate": 3.296017133479354e-06, "loss": 0.10718551278114319, "step": 3386 }, { "epoch": 0.8792834890965732, "grad_norm": 1.185653567314148, "learning_rate": 3.295099353962618e-06, "loss": 0.12985000014305115, "step": 3387 }, { "epoch": 0.8795430944963655, "grad_norm": 1.007273554801941, "learning_rate": 3.2941814780847987e-06, "loss": 0.11549647152423859, "step": 3388 }, { "epoch": 0.8798026998961579, "grad_norm": 2.743014097213745, "learning_rate": 3.293263506008153e-06, "loss": 0.1128731518983841, "step": 3389 }, { "epoch": 0.8800623052959502, "grad_norm": 1.5479575395584106, "learning_rate": 3.2923454378949587e-06, "loss": 0.14749380946159363, "step": 3390 }, { "epoch": 0.8803219106957425, "grad_norm": 0.8271548748016357, "learning_rate": 3.291427273907509e-06, "loss": 0.11421199142932892, "step": 3391 }, { "epoch": 0.8805815160955348, "grad_norm": 2.3799397945404053, "learning_rate": 3.2905090142081147e-06, "loss": 0.1488475501537323, "step": 3392 }, { "epoch": 0.8808411214953271, "grad_norm": 0.7401686906814575, "learning_rate": 3.289590658959103e-06, "loss": 0.10764801502227783, "step": 3393 }, { "epoch": 0.8811007268951194, "grad_norm": 1.4563748836517334, "learning_rate": 3.2886722083228177e-06, "loss": 0.11585129797458649, "step": 3394 }, { "epoch": 0.8813603322949117, "grad_norm": 1.922060251235962, "learning_rate": 3.2877536624616198e-06, "loss": 0.1396317183971405, "step": 3395 }, { "epoch": 0.881619937694704, "grad_norm": 2.852698564529419, "learning_rate": 3.286835021537886e-06, "loss": 0.14463382959365845, "step": 3396 }, { "epoch": 0.8818795430944963, "grad_norm": 0.8032242059707642, "learning_rate": 3.2859162857140143e-06, "loss": 0.12649652361869812, "step": 3397 }, { "epoch": 0.8821391484942886, "grad_norm": 1.885541319847107, "learning_rate": 3.2849974551524135e-06, "loss": 0.12940286099910736, "step": 3398 }, { "epoch": 0.882398753894081, "grad_norm": 0.4016580879688263, "learning_rate": 3.284078530015512e-06, "loss": 0.11221565306186676, "step": 3399 }, { "epoch": 0.8826583592938733, "grad_norm": 0.9229528903961182, "learning_rate": 3.2831595104657547e-06, "loss": 0.1203397735953331, "step": 3400 }, { "epoch": 0.8829179646936657, "grad_norm": 1.3634024858474731, "learning_rate": 3.2822403966656045e-06, "loss": 0.11470779776573181, "step": 3401 }, { "epoch": 0.883177570093458, "grad_norm": 0.397066593170166, "learning_rate": 3.2813211887775386e-06, "loss": 0.11673960089683533, "step": 3402 }, { "epoch": 0.8834371754932503, "grad_norm": 0.8468995094299316, "learning_rate": 3.2804018869640525e-06, "loss": 0.12268547713756561, "step": 3403 }, { "epoch": 0.8836967808930426, "grad_norm": 0.7708533406257629, "learning_rate": 3.279482491387657e-06, "loss": 0.12708918750286102, "step": 3404 }, { "epoch": 0.8839563862928349, "grad_norm": 1.5099620819091797, "learning_rate": 3.2785630022108804e-06, "loss": 0.1896066963672638, "step": 3405 }, { "epoch": 0.8842159916926272, "grad_norm": 1.1849615573883057, "learning_rate": 3.2776434195962682e-06, "loss": 0.12816110253334045, "step": 3406 }, { "epoch": 0.8844755970924195, "grad_norm": 0.7539719343185425, "learning_rate": 3.2767237437063803e-06, "loss": 0.12615254521369934, "step": 3407 }, { "epoch": 0.8847352024922118, "grad_norm": 1.2847942113876343, "learning_rate": 3.2758039747037944e-06, "loss": 0.12448417395353317, "step": 3408 }, { "epoch": 0.8849948078920041, "grad_norm": 0.6192044615745544, "learning_rate": 3.274884112751106e-06, "loss": 0.11741824448108673, "step": 3409 }, { "epoch": 0.8852544132917964, "grad_norm": 1.4272300004959106, "learning_rate": 3.2739641580109243e-06, "loss": 0.1560003161430359, "step": 3410 }, { "epoch": 0.8855140186915887, "grad_norm": 1.343165397644043, "learning_rate": 3.273044110645878e-06, "loss": 0.16865332424640656, "step": 3411 }, { "epoch": 0.885773624091381, "grad_norm": 1.0370516777038574, "learning_rate": 3.2721239708186077e-06, "loss": 0.12880058586597443, "step": 3412 }, { "epoch": 0.8860332294911735, "grad_norm": 1.204537034034729, "learning_rate": 3.271203738691774e-06, "loss": 0.1291445940732956, "step": 3413 }, { "epoch": 0.8862928348909658, "grad_norm": 0.47942033410072327, "learning_rate": 3.2702834144280536e-06, "loss": 0.11967524886131287, "step": 3414 }, { "epoch": 0.8865524402907581, "grad_norm": 0.5192655324935913, "learning_rate": 3.2693629981901392e-06, "loss": 0.11495418846607208, "step": 3415 }, { "epoch": 0.8868120456905504, "grad_norm": 0.46123865246772766, "learning_rate": 3.268442490140738e-06, "loss": 0.10623094439506531, "step": 3416 }, { "epoch": 0.8870716510903427, "grad_norm": 0.6229047179222107, "learning_rate": 3.2675218904425748e-06, "loss": 0.11514849960803986, "step": 3417 }, { "epoch": 0.887331256490135, "grad_norm": 0.6756922006607056, "learning_rate": 3.266601199258391e-06, "loss": 0.13145330548286438, "step": 3418 }, { "epoch": 0.8875908618899273, "grad_norm": 0.5215660333633423, "learning_rate": 3.2656804167509456e-06, "loss": 0.12338656187057495, "step": 3419 }, { "epoch": 0.8878504672897196, "grad_norm": 0.3552406430244446, "learning_rate": 3.264759543083009e-06, "loss": 0.10579342395067215, "step": 3420 }, { "epoch": 0.8881100726895119, "grad_norm": 0.8808271884918213, "learning_rate": 3.2638385784173704e-06, "loss": 0.1256992220878601, "step": 3421 }, { "epoch": 0.8883696780893042, "grad_norm": 0.6130419373512268, "learning_rate": 3.262917522916838e-06, "loss": 0.1365487426519394, "step": 3422 }, { "epoch": 0.8886292834890965, "grad_norm": 1.410222053527832, "learning_rate": 3.2619963767442314e-06, "loss": 0.13214533030986786, "step": 3423 }, { "epoch": 0.8888888888888888, "grad_norm": 1.6325103044509888, "learning_rate": 3.2610751400623887e-06, "loss": 0.11584435403347015, "step": 3424 }, { "epoch": 0.8891484942886813, "grad_norm": 1.6992695331573486, "learning_rate": 3.2601538130341627e-06, "loss": 0.15157076716423035, "step": 3425 }, { "epoch": 0.8894080996884736, "grad_norm": 1.1623566150665283, "learning_rate": 3.2592323958224236e-06, "loss": 0.10671891272068024, "step": 3426 }, { "epoch": 0.8896677050882659, "grad_norm": 1.2560001611709595, "learning_rate": 3.258310888590058e-06, "loss": 0.1269027292728424, "step": 3427 }, { "epoch": 0.8899273104880582, "grad_norm": 1.3522136211395264, "learning_rate": 3.257389291499965e-06, "loss": 0.12287075817584991, "step": 3428 }, { "epoch": 0.8901869158878505, "grad_norm": 2.0355069637298584, "learning_rate": 3.256467604715062e-06, "loss": 0.14777448773384094, "step": 3429 }, { "epoch": 0.8904465212876428, "grad_norm": 1.540557861328125, "learning_rate": 3.255545828398285e-06, "loss": 0.111782006919384, "step": 3430 }, { "epoch": 0.8907061266874351, "grad_norm": 2.129040241241455, "learning_rate": 3.2546239627125796e-06, "loss": 0.14490796625614166, "step": 3431 }, { "epoch": 0.8909657320872274, "grad_norm": 2.241398811340332, "learning_rate": 3.2537020078209126e-06, "loss": 0.13164353370666504, "step": 3432 }, { "epoch": 0.8912253374870197, "grad_norm": 1.3379074335098267, "learning_rate": 3.252779963886264e-06, "loss": 0.11302664875984192, "step": 3433 }, { "epoch": 0.891484942886812, "grad_norm": 1.6258865594863892, "learning_rate": 3.25185783107163e-06, "loss": 0.12837789952754974, "step": 3434 }, { "epoch": 0.8917445482866043, "grad_norm": 1.6746984720230103, "learning_rate": 3.250935609540021e-06, "loss": 0.15862277150154114, "step": 3435 }, { "epoch": 0.8920041536863966, "grad_norm": 1.0317391157150269, "learning_rate": 3.250013299454468e-06, "loss": 0.10624176263809204, "step": 3436 }, { "epoch": 0.892263759086189, "grad_norm": 0.40851008892059326, "learning_rate": 3.2490909009780117e-06, "loss": 0.11180723458528519, "step": 3437 }, { "epoch": 0.8925233644859814, "grad_norm": 0.6551343202590942, "learning_rate": 3.248168414273712e-06, "loss": 0.11616137623786926, "step": 3438 }, { "epoch": 0.8927829698857737, "grad_norm": 1.528543472290039, "learning_rate": 3.2472458395046423e-06, "loss": 0.10741507261991501, "step": 3439 }, { "epoch": 0.893042575285566, "grad_norm": 1.5158805847167969, "learning_rate": 3.2463231768338954e-06, "loss": 0.1559123396873474, "step": 3440 }, { "epoch": 0.8933021806853583, "grad_norm": 0.7417702078819275, "learning_rate": 3.2454004264245735e-06, "loss": 0.11436013132333755, "step": 3441 }, { "epoch": 0.8935617860851506, "grad_norm": 1.6808043718338013, "learning_rate": 3.2444775884398e-06, "loss": 0.13207271695137024, "step": 3442 }, { "epoch": 0.8938213914849429, "grad_norm": 0.5640512108802795, "learning_rate": 3.2435546630427105e-06, "loss": 0.11609925329685211, "step": 3443 }, { "epoch": 0.8940809968847352, "grad_norm": 1.7479569911956787, "learning_rate": 3.2426316503964583e-06, "loss": 0.13882020115852356, "step": 3444 }, { "epoch": 0.8943406022845275, "grad_norm": 0.6938169002532959, "learning_rate": 3.2417085506642095e-06, "loss": 0.11468670517206192, "step": 3445 }, { "epoch": 0.8946002076843198, "grad_norm": 2.180403709411621, "learning_rate": 3.240785364009148e-06, "loss": 0.13297392427921295, "step": 3446 }, { "epoch": 0.8948598130841121, "grad_norm": 0.8283228278160095, "learning_rate": 3.239862090594471e-06, "loss": 0.126366525888443, "step": 3447 }, { "epoch": 0.8951194184839044, "grad_norm": 1.0389578342437744, "learning_rate": 3.2389387305833932e-06, "loss": 0.11906616389751434, "step": 3448 }, { "epoch": 0.8953790238836968, "grad_norm": 1.9306174516677856, "learning_rate": 3.238015284139143e-06, "loss": 0.11606601625680923, "step": 3449 }, { "epoch": 0.8956386292834891, "grad_norm": 0.9125521183013916, "learning_rate": 3.2370917514249646e-06, "loss": 0.14011281728744507, "step": 3450 }, { "epoch": 0.8958982346832814, "grad_norm": 1.5772980451583862, "learning_rate": 3.2361681326041167e-06, "loss": 0.10818463563919067, "step": 3451 }, { "epoch": 0.8961578400830738, "grad_norm": 1.4318110942840576, "learning_rate": 3.2352444278398754e-06, "loss": 0.13675451278686523, "step": 3452 }, { "epoch": 0.8964174454828661, "grad_norm": 0.7508276700973511, "learning_rate": 3.2343206372955305e-06, "loss": 0.1184215098619461, "step": 3453 }, { "epoch": 0.8966770508826584, "grad_norm": 0.5904049873352051, "learning_rate": 3.2333967611343854e-06, "loss": 0.13024258613586426, "step": 3454 }, { "epoch": 0.8969366562824507, "grad_norm": 0.7640577554702759, "learning_rate": 3.232472799519762e-06, "loss": 0.13565737009048462, "step": 3455 }, { "epoch": 0.897196261682243, "grad_norm": 1.886267066001892, "learning_rate": 3.231548752614994e-06, "loss": 0.13015012443065643, "step": 3456 }, { "epoch": 0.8974558670820353, "grad_norm": 2.007755756378174, "learning_rate": 3.2306246205834333e-06, "loss": 0.14979109168052673, "step": 3457 }, { "epoch": 0.8977154724818276, "grad_norm": 0.9863632321357727, "learning_rate": 3.229700403588444e-06, "loss": 0.1218753233551979, "step": 3458 }, { "epoch": 0.8979750778816199, "grad_norm": 1.008504033088684, "learning_rate": 3.2287761017934076e-06, "loss": 0.1597210168838501, "step": 3459 }, { "epoch": 0.8982346832814122, "grad_norm": 1.854435920715332, "learning_rate": 3.2278517153617186e-06, "loss": 0.1413886547088623, "step": 3460 }, { "epoch": 0.8984942886812046, "grad_norm": 1.0565381050109863, "learning_rate": 3.226927244456788e-06, "loss": 0.11728417873382568, "step": 3461 }, { "epoch": 0.8987538940809969, "grad_norm": 0.7559845447540283, "learning_rate": 3.2260026892420397e-06, "loss": 0.11718688905239105, "step": 3462 }, { "epoch": 0.8990134994807892, "grad_norm": 1.2220808267593384, "learning_rate": 3.2250780498809157e-06, "loss": 0.13335831463336945, "step": 3463 }, { "epoch": 0.8992731048805815, "grad_norm": 0.6327401399612427, "learning_rate": 3.22415332653687e-06, "loss": 0.11093008518218994, "step": 3464 }, { "epoch": 0.8995327102803738, "grad_norm": 1.2687636613845825, "learning_rate": 3.2232285193733735e-06, "loss": 0.11598409712314606, "step": 3465 }, { "epoch": 0.8997923156801662, "grad_norm": 1.593436598777771, "learning_rate": 3.2223036285539092e-06, "loss": 0.12183905392885208, "step": 3466 }, { "epoch": 0.9000519210799585, "grad_norm": 1.5669276714324951, "learning_rate": 3.2213786542419796e-06, "loss": 0.12438543140888214, "step": 3467 }, { "epoch": 0.9003115264797508, "grad_norm": 0.48986753821372986, "learning_rate": 3.220453596601096e-06, "loss": 0.10763151943683624, "step": 3468 }, { "epoch": 0.9005711318795431, "grad_norm": 0.5560157895088196, "learning_rate": 3.219528455794788e-06, "loss": 0.11262976378202438, "step": 3469 }, { "epoch": 0.9008307372793354, "grad_norm": 1.063844919204712, "learning_rate": 3.2186032319866e-06, "loss": 0.12750640511512756, "step": 3470 }, { "epoch": 0.9010903426791277, "grad_norm": 1.4789026975631714, "learning_rate": 3.217677925340091e-06, "loss": 0.12759089469909668, "step": 3471 }, { "epoch": 0.90134994807892, "grad_norm": 0.6418126225471497, "learning_rate": 3.2167525360188317e-06, "loss": 0.10813739150762558, "step": 3472 }, { "epoch": 0.9016095534787124, "grad_norm": 0.8404300808906555, "learning_rate": 3.2158270641864126e-06, "loss": 0.11899193376302719, "step": 3473 }, { "epoch": 0.9018691588785047, "grad_norm": 0.8138826489448547, "learning_rate": 3.2149015100064333e-06, "loss": 0.13041242957115173, "step": 3474 }, { "epoch": 0.902128764278297, "grad_norm": 1.0375580787658691, "learning_rate": 3.213975873642513e-06, "loss": 0.12627878785133362, "step": 3475 }, { "epoch": 0.9023883696780893, "grad_norm": 2.0209648609161377, "learning_rate": 3.2130501552582823e-06, "loss": 0.1298895925283432, "step": 3476 }, { "epoch": 0.9026479750778816, "grad_norm": 0.9199742674827576, "learning_rate": 3.2121243550173854e-06, "loss": 0.14160852134227753, "step": 3477 }, { "epoch": 0.9029075804776739, "grad_norm": 0.8358656764030457, "learning_rate": 3.2111984730834837e-06, "loss": 0.10323600471019745, "step": 3478 }, { "epoch": 0.9031671858774662, "grad_norm": 0.3953405022621155, "learning_rate": 3.2102725096202527e-06, "loss": 0.10887280106544495, "step": 3479 }, { "epoch": 0.9034267912772586, "grad_norm": 1.5322421789169312, "learning_rate": 3.2093464647913815e-06, "loss": 0.1139141395688057, "step": 3480 }, { "epoch": 0.9036863966770509, "grad_norm": 0.45890969038009644, "learning_rate": 3.2084203387605716e-06, "loss": 0.123210608959198, "step": 3481 }, { "epoch": 0.9039460020768432, "grad_norm": 0.9990977048873901, "learning_rate": 3.207494131691542e-06, "loss": 0.12367308139801025, "step": 3482 }, { "epoch": 0.9042056074766355, "grad_norm": 1.645093321800232, "learning_rate": 3.2065678437480265e-06, "loss": 0.13306990265846252, "step": 3483 }, { "epoch": 0.9044652128764278, "grad_norm": 1.5934196710586548, "learning_rate": 3.2056414750937696e-06, "loss": 0.12837152183055878, "step": 3484 }, { "epoch": 0.9047248182762202, "grad_norm": 1.8137973546981812, "learning_rate": 3.2047150258925324e-06, "loss": 0.1538674533367157, "step": 3485 }, { "epoch": 0.9049844236760125, "grad_norm": 1.0778216123580933, "learning_rate": 3.203788496308091e-06, "loss": 0.12313719093799591, "step": 3486 }, { "epoch": 0.9052440290758048, "grad_norm": 0.6024559736251831, "learning_rate": 3.2028618865042327e-06, "loss": 0.11591902375221252, "step": 3487 }, { "epoch": 0.9055036344755971, "grad_norm": 1.020709753036499, "learning_rate": 3.201935196644763e-06, "loss": 0.12566614151000977, "step": 3488 }, { "epoch": 0.9057632398753894, "grad_norm": 0.48630109429359436, "learning_rate": 3.2010084268934984e-06, "loss": 0.11631686240434647, "step": 3489 }, { "epoch": 0.9060228452751817, "grad_norm": 0.8670169115066528, "learning_rate": 3.2000815774142703e-06, "loss": 0.11554788053035736, "step": 3490 }, { "epoch": 0.906282450674974, "grad_norm": 2.260155439376831, "learning_rate": 3.199154648370925e-06, "loss": 0.1176566556096077, "step": 3491 }, { "epoch": 0.9065420560747663, "grad_norm": 1.5392565727233887, "learning_rate": 3.198227639927322e-06, "loss": 0.11881051957607269, "step": 3492 }, { "epoch": 0.9068016614745587, "grad_norm": 1.7910863161087036, "learning_rate": 3.1973005522473355e-06, "loss": 0.13007083535194397, "step": 3493 }, { "epoch": 0.907061266874351, "grad_norm": 0.8639808893203735, "learning_rate": 3.196373385494853e-06, "loss": 0.11738935858011246, "step": 3494 }, { "epoch": 0.9073208722741433, "grad_norm": 0.7360579371452332, "learning_rate": 3.1954461398337766e-06, "loss": 0.12366420775651932, "step": 3495 }, { "epoch": 0.9075804776739356, "grad_norm": 1.4422926902770996, "learning_rate": 3.194518815428022e-06, "loss": 0.1019461452960968, "step": 3496 }, { "epoch": 0.907840083073728, "grad_norm": 1.1742252111434937, "learning_rate": 3.193591412441519e-06, "loss": 0.13813209533691406, "step": 3497 }, { "epoch": 0.9080996884735203, "grad_norm": 0.4275050163269043, "learning_rate": 3.1926639310382114e-06, "loss": 0.11049474030733109, "step": 3498 }, { "epoch": 0.9083592938733126, "grad_norm": 0.8992841243743896, "learning_rate": 3.1917363713820565e-06, "loss": 0.1310219168663025, "step": 3499 }, { "epoch": 0.9086188992731049, "grad_norm": 0.48350846767425537, "learning_rate": 3.1908087336370245e-06, "loss": 0.12004135549068451, "step": 3500 }, { "epoch": 0.9088785046728972, "grad_norm": 1.198820948600769, "learning_rate": 3.189881017967103e-06, "loss": 0.1262073665857315, "step": 3501 }, { "epoch": 0.9091381100726895, "grad_norm": 1.1706057786941528, "learning_rate": 3.188953224536289e-06, "loss": 0.14398261904716492, "step": 3502 }, { "epoch": 0.9093977154724818, "grad_norm": 1.4901248216629028, "learning_rate": 3.188025353508595e-06, "loss": 0.13831928372383118, "step": 3503 }, { "epoch": 0.9096573208722741, "grad_norm": 1.6340577602386475, "learning_rate": 3.187097405048049e-06, "loss": 0.12826892733573914, "step": 3504 }, { "epoch": 0.9099169262720664, "grad_norm": 1.3066093921661377, "learning_rate": 3.18616937931869e-06, "loss": 0.1138419583439827, "step": 3505 }, { "epoch": 0.9101765316718587, "grad_norm": 0.8380355834960938, "learning_rate": 3.1852412764845704e-06, "loss": 0.13095611333847046, "step": 3506 }, { "epoch": 0.910436137071651, "grad_norm": 1.0247035026550293, "learning_rate": 3.1843130967097596e-06, "loss": 0.09979502856731415, "step": 3507 }, { "epoch": 0.9106957424714434, "grad_norm": 2.478783130645752, "learning_rate": 3.183384840158338e-06, "loss": 0.11898026615381241, "step": 3508 }, { "epoch": 0.9109553478712358, "grad_norm": 1.1819692850112915, "learning_rate": 3.1824565069943993e-06, "loss": 0.13614130020141602, "step": 3509 }, { "epoch": 0.9112149532710281, "grad_norm": 1.566826343536377, "learning_rate": 3.181528097382052e-06, "loss": 0.13198858499526978, "step": 3510 }, { "epoch": 0.9114745586708204, "grad_norm": 1.3598636388778687, "learning_rate": 3.180599611485418e-06, "loss": 0.10785463452339172, "step": 3511 }, { "epoch": 0.9117341640706127, "grad_norm": 1.147301435470581, "learning_rate": 3.179671049468632e-06, "loss": 0.10685621947050095, "step": 3512 }, { "epoch": 0.911993769470405, "grad_norm": 1.6122421026229858, "learning_rate": 3.1787424114958433e-06, "loss": 0.1136908009648323, "step": 3513 }, { "epoch": 0.9122533748701973, "grad_norm": 1.5572853088378906, "learning_rate": 3.1778136977312125e-06, "loss": 0.12337710708379745, "step": 3514 }, { "epoch": 0.9125129802699896, "grad_norm": 0.9512377977371216, "learning_rate": 3.1768849083389164e-06, "loss": 0.10916805267333984, "step": 3515 }, { "epoch": 0.9127725856697819, "grad_norm": 0.827459454536438, "learning_rate": 3.175956043483142e-06, "loss": 0.11735579371452332, "step": 3516 }, { "epoch": 0.9130321910695742, "grad_norm": 0.9566985964775085, "learning_rate": 3.1750271033280925e-06, "loss": 0.12825821340084076, "step": 3517 }, { "epoch": 0.9132917964693665, "grad_norm": 0.5086838603019714, "learning_rate": 3.1740980880379823e-06, "loss": 0.10427124798297882, "step": 3518 }, { "epoch": 0.9135514018691588, "grad_norm": 0.9598219990730286, "learning_rate": 3.1731689977770414e-06, "loss": 0.14320528507232666, "step": 3519 }, { "epoch": 0.9138110072689511, "grad_norm": 0.8021107912063599, "learning_rate": 3.172239832709511e-06, "loss": 0.12548120319843292, "step": 3520 }, { "epoch": 0.9140706126687436, "grad_norm": 1.0257784128189087, "learning_rate": 3.171310592999646e-06, "loss": 0.11594288796186447, "step": 3521 }, { "epoch": 0.9143302180685359, "grad_norm": 1.8216280937194824, "learning_rate": 3.1703812788117157e-06, "loss": 0.16367287933826447, "step": 3522 }, { "epoch": 0.9145898234683282, "grad_norm": 2.0906214714050293, "learning_rate": 3.1694518903099993e-06, "loss": 0.14254003763198853, "step": 3523 }, { "epoch": 0.9148494288681205, "grad_norm": 2.3137080669403076, "learning_rate": 3.168522427658794e-06, "loss": 0.13861748576164246, "step": 3524 }, { "epoch": 0.9151090342679128, "grad_norm": 1.1509182453155518, "learning_rate": 3.1675928910224063e-06, "loss": 0.11559456586837769, "step": 3525 }, { "epoch": 0.9153686396677051, "grad_norm": 1.5863486528396606, "learning_rate": 3.1666632805651564e-06, "loss": 0.13320699334144592, "step": 3526 }, { "epoch": 0.9156282450674974, "grad_norm": 2.492680549621582, "learning_rate": 3.165733596451379e-06, "loss": 0.1453624665737152, "step": 3527 }, { "epoch": 0.9158878504672897, "grad_norm": 0.8627732992172241, "learning_rate": 3.164803838845421e-06, "loss": 0.11790665239095688, "step": 3528 }, { "epoch": 0.916147455867082, "grad_norm": 0.8584651947021484, "learning_rate": 3.1638740079116427e-06, "loss": 0.11443829536437988, "step": 3529 }, { "epoch": 0.9164070612668743, "grad_norm": 0.8665988445281982, "learning_rate": 3.1629441038144152e-06, "loss": 0.1142067089676857, "step": 3530 }, { "epoch": 0.9166666666666666, "grad_norm": 0.866764485836029, "learning_rate": 3.1620141267181253e-06, "loss": 0.1299867331981659, "step": 3531 }, { "epoch": 0.9169262720664589, "grad_norm": 1.2223788499832153, "learning_rate": 3.1610840767871726e-06, "loss": 0.12429695576429367, "step": 3532 }, { "epoch": 0.9171858774662512, "grad_norm": 0.7431601881980896, "learning_rate": 3.160153954185967e-06, "loss": 0.11257565021514893, "step": 3533 }, { "epoch": 0.9174454828660437, "grad_norm": 0.5863739848136902, "learning_rate": 3.159223759078934e-06, "loss": 0.11122748255729675, "step": 3534 }, { "epoch": 0.917705088265836, "grad_norm": 1.0438300371170044, "learning_rate": 3.15829349163051e-06, "loss": 0.10346940159797668, "step": 3535 }, { "epoch": 0.9179646936656283, "grad_norm": 0.7089246511459351, "learning_rate": 3.157363152005145e-06, "loss": 0.13435642421245575, "step": 3536 }, { "epoch": 0.9182242990654206, "grad_norm": 1.3733673095703125, "learning_rate": 3.1564327403673024e-06, "loss": 0.15854544937610626, "step": 3537 }, { "epoch": 0.9184839044652129, "grad_norm": 1.3861737251281738, "learning_rate": 3.155502256881458e-06, "loss": 0.13887907564640045, "step": 3538 }, { "epoch": 0.9187435098650052, "grad_norm": 0.48711997270584106, "learning_rate": 3.1545717017120993e-06, "loss": 0.11305584013462067, "step": 3539 }, { "epoch": 0.9190031152647975, "grad_norm": 1.503480076789856, "learning_rate": 3.153641075023726e-06, "loss": 0.1371222287416458, "step": 3540 }, { "epoch": 0.9192627206645898, "grad_norm": 1.0057867765426636, "learning_rate": 3.152710376980854e-06, "loss": 0.12892839312553406, "step": 3541 }, { "epoch": 0.9195223260643821, "grad_norm": 0.47729507088661194, "learning_rate": 3.151779607748008e-06, "loss": 0.11542834341526031, "step": 3542 }, { "epoch": 0.9197819314641744, "grad_norm": 1.1809033155441284, "learning_rate": 3.150848767489726e-06, "loss": 0.11253613233566284, "step": 3543 }, { "epoch": 0.9200415368639667, "grad_norm": 1.236161470413208, "learning_rate": 3.1499178563705605e-06, "loss": 0.11644509434700012, "step": 3544 }, { "epoch": 0.920301142263759, "grad_norm": 0.85579514503479, "learning_rate": 3.1489868745550754e-06, "loss": 0.11441677063703537, "step": 3545 }, { "epoch": 0.9205607476635514, "grad_norm": 0.8560343384742737, "learning_rate": 3.1480558222078467e-06, "loss": 0.122810959815979, "step": 3546 }, { "epoch": 0.9208203530633438, "grad_norm": 1.055882215499878, "learning_rate": 3.147124699493462e-06, "loss": 0.11729389429092407, "step": 3547 }, { "epoch": 0.9210799584631361, "grad_norm": 0.5787237882614136, "learning_rate": 3.146193506576524e-06, "loss": 0.10993855446577072, "step": 3548 }, { "epoch": 0.9213395638629284, "grad_norm": 1.714371681213379, "learning_rate": 3.1452622436216452e-06, "loss": 0.11814598739147186, "step": 3549 }, { "epoch": 0.9215991692627207, "grad_norm": 1.3930333852767944, "learning_rate": 3.144330910793453e-06, "loss": 0.12493197619915009, "step": 3550 }, { "epoch": 0.921858774662513, "grad_norm": 1.824010968208313, "learning_rate": 3.143399508256584e-06, "loss": 0.11623306572437286, "step": 3551 }, { "epoch": 0.9221183800623053, "grad_norm": 1.8061192035675049, "learning_rate": 3.14246803617569e-06, "loss": 0.134272962808609, "step": 3552 }, { "epoch": 0.9223779854620976, "grad_norm": 1.3669084310531616, "learning_rate": 3.1415364947154323e-06, "loss": 0.12070365250110626, "step": 3553 }, { "epoch": 0.9226375908618899, "grad_norm": 0.96274334192276, "learning_rate": 3.1406048840404886e-06, "loss": 0.1158745288848877, "step": 3554 }, { "epoch": 0.9228971962616822, "grad_norm": 1.6195945739746094, "learning_rate": 3.1396732043155447e-06, "loss": 0.15038833022117615, "step": 3555 }, { "epoch": 0.9231568016614745, "grad_norm": 2.486295461654663, "learning_rate": 3.1387414557053013e-06, "loss": 0.12469583749771118, "step": 3556 }, { "epoch": 0.9234164070612668, "grad_norm": 0.5513415932655334, "learning_rate": 3.1378096383744685e-06, "loss": 0.11736800521612167, "step": 3557 }, { "epoch": 0.9236760124610592, "grad_norm": 1.2029774188995361, "learning_rate": 3.1368777524877713e-06, "loss": 0.1505933701992035, "step": 3558 }, { "epoch": 0.9239356178608515, "grad_norm": 0.641311764717102, "learning_rate": 3.1359457982099467e-06, "loss": 0.14805348217487335, "step": 3559 }, { "epoch": 0.9241952232606438, "grad_norm": 1.0385562181472778, "learning_rate": 3.1350137757057412e-06, "loss": 0.11807854473590851, "step": 3560 }, { "epoch": 0.9244548286604362, "grad_norm": 0.44112277030944824, "learning_rate": 3.1340816851399164e-06, "loss": 0.11872999370098114, "step": 3561 }, { "epoch": 0.9247144340602285, "grad_norm": 0.7723562121391296, "learning_rate": 3.133149526677243e-06, "loss": 0.1202070415019989, "step": 3562 }, { "epoch": 0.9249740394600208, "grad_norm": 0.6129085421562195, "learning_rate": 3.1322173004825076e-06, "loss": 0.12031146883964539, "step": 3563 }, { "epoch": 0.9252336448598131, "grad_norm": 1.0358550548553467, "learning_rate": 3.1312850067205036e-06, "loss": 0.14276424050331116, "step": 3564 }, { "epoch": 0.9254932502596054, "grad_norm": 1.579879879951477, "learning_rate": 3.130352645556042e-06, "loss": 0.1062278300523758, "step": 3565 }, { "epoch": 0.9257528556593977, "grad_norm": 0.5542742013931274, "learning_rate": 3.1294202171539406e-06, "loss": 0.09723533689975739, "step": 3566 }, { "epoch": 0.92601246105919, "grad_norm": 0.6278021931648254, "learning_rate": 3.128487721679033e-06, "loss": 0.11852622032165527, "step": 3567 }, { "epoch": 0.9262720664589823, "grad_norm": 1.1785306930541992, "learning_rate": 3.1275551592961617e-06, "loss": 0.10847573727369308, "step": 3568 }, { "epoch": 0.9265316718587746, "grad_norm": 0.7806999087333679, "learning_rate": 3.1266225301701842e-06, "loss": 0.11127156019210815, "step": 3569 }, { "epoch": 0.926791277258567, "grad_norm": 2.144697666168213, "learning_rate": 3.1256898344659657e-06, "loss": 0.13282164931297302, "step": 3570 }, { "epoch": 0.9270508826583593, "grad_norm": 0.9566797018051147, "learning_rate": 3.124757072348388e-06, "loss": 0.12307196855545044, "step": 3571 }, { "epoch": 0.9273104880581516, "grad_norm": 1.5490225553512573, "learning_rate": 3.1238242439823393e-06, "loss": 0.12056483328342438, "step": 3572 }, { "epoch": 0.927570093457944, "grad_norm": 1.5148762464523315, "learning_rate": 3.1228913495327245e-06, "loss": 0.1521565318107605, "step": 3573 }, { "epoch": 0.9278296988577363, "grad_norm": 1.0783988237380981, "learning_rate": 3.1219583891644574e-06, "loss": 0.1303655058145523, "step": 3574 }, { "epoch": 0.9280893042575286, "grad_norm": 1.0804669857025146, "learning_rate": 3.121025363042463e-06, "loss": 0.11063717305660248, "step": 3575 }, { "epoch": 0.9283489096573209, "grad_norm": 1.2562669515609741, "learning_rate": 3.12009227133168e-06, "loss": 0.14358346164226532, "step": 3576 }, { "epoch": 0.9286085150571132, "grad_norm": 2.353130578994751, "learning_rate": 3.119159114197058e-06, "loss": 0.1262306421995163, "step": 3577 }, { "epoch": 0.9288681204569055, "grad_norm": 2.781403064727783, "learning_rate": 3.118225891803556e-06, "loss": 0.12123338878154755, "step": 3578 }, { "epoch": 0.9291277258566978, "grad_norm": 0.6287989020347595, "learning_rate": 3.1172926043161496e-06, "loss": 0.10246845334768295, "step": 3579 }, { "epoch": 0.9293873312564901, "grad_norm": 1.2994978427886963, "learning_rate": 3.116359251899819e-06, "loss": 0.14054644107818604, "step": 3580 }, { "epoch": 0.9296469366562824, "grad_norm": 0.7998136878013611, "learning_rate": 3.1154258347195614e-06, "loss": 0.12248586118221283, "step": 3581 }, { "epoch": 0.9299065420560748, "grad_norm": 1.2949305772781372, "learning_rate": 3.114492352940384e-06, "loss": 0.12116221338510513, "step": 3582 }, { "epoch": 0.9301661474558671, "grad_norm": 1.243203043937683, "learning_rate": 3.1135588067273044e-06, "loss": 0.15810100734233856, "step": 3583 }, { "epoch": 0.9304257528556594, "grad_norm": 1.1111582517623901, "learning_rate": 3.1126251962453507e-06, "loss": 0.1100996881723404, "step": 3584 }, { "epoch": 0.9306853582554517, "grad_norm": 1.3917697668075562, "learning_rate": 3.111691521659567e-06, "loss": 0.11891303956508636, "step": 3585 }, { "epoch": 0.930944963655244, "grad_norm": 0.6709244847297668, "learning_rate": 3.1107577831350033e-06, "loss": 0.11808512359857559, "step": 3586 }, { "epoch": 0.9312045690550363, "grad_norm": 1.6574596166610718, "learning_rate": 3.1098239808367237e-06, "loss": 0.1656171679496765, "step": 3587 }, { "epoch": 0.9314641744548287, "grad_norm": 0.8429265022277832, "learning_rate": 3.108890114929803e-06, "loss": 0.12843212485313416, "step": 3588 }, { "epoch": 0.931723779854621, "grad_norm": 2.0915231704711914, "learning_rate": 3.1079561855793277e-06, "loss": 0.14978668093681335, "step": 3589 }, { "epoch": 0.9319833852544133, "grad_norm": 1.4549126625061035, "learning_rate": 3.107022192950395e-06, "loss": 0.11876359581947327, "step": 3590 }, { "epoch": 0.9322429906542056, "grad_norm": 1.1403717994689941, "learning_rate": 3.106088137208113e-06, "loss": 0.10944711416959763, "step": 3591 }, { "epoch": 0.9325025960539979, "grad_norm": 1.0507532358169556, "learning_rate": 3.105154018517602e-06, "loss": 0.13455699384212494, "step": 3592 }, { "epoch": 0.9327622014537902, "grad_norm": 1.0623929500579834, "learning_rate": 3.1042198370439925e-06, "loss": 0.12397171556949615, "step": 3593 }, { "epoch": 0.9330218068535826, "grad_norm": 0.5971724987030029, "learning_rate": 3.103285592952427e-06, "loss": 0.11550848186016083, "step": 3594 }, { "epoch": 0.9332814122533749, "grad_norm": 0.9256536960601807, "learning_rate": 3.1023512864080563e-06, "loss": 0.0965586006641388, "step": 3595 }, { "epoch": 0.9335410176531672, "grad_norm": 1.4198139905929565, "learning_rate": 3.101416917576047e-06, "loss": 0.12174518406391144, "step": 3596 }, { "epoch": 0.9338006230529595, "grad_norm": 0.9930853843688965, "learning_rate": 3.1004824866215723e-06, "loss": 0.11695271730422974, "step": 3597 }, { "epoch": 0.9340602284527518, "grad_norm": 0.6840318441390991, "learning_rate": 3.09954799370982e-06, "loss": 0.11869680881500244, "step": 3598 }, { "epoch": 0.9343198338525441, "grad_norm": 1.1521919965744019, "learning_rate": 3.0986134390059847e-06, "loss": 0.1186465471982956, "step": 3599 }, { "epoch": 0.9345794392523364, "grad_norm": 0.7965601682662964, "learning_rate": 3.0976788226752767e-06, "loss": 0.12169674783945084, "step": 3600 }, { "epoch": 0.9348390446521287, "grad_norm": 0.7523419857025146, "learning_rate": 3.0967441448829132e-06, "loss": 0.12999382615089417, "step": 3601 }, { "epoch": 0.935098650051921, "grad_norm": 1.354883074760437, "learning_rate": 3.0958094057941245e-06, "loss": 0.13338041305541992, "step": 3602 }, { "epoch": 0.9353582554517134, "grad_norm": 0.48337915539741516, "learning_rate": 3.09487460557415e-06, "loss": 0.10301221907138824, "step": 3603 }, { "epoch": 0.9356178608515057, "grad_norm": 1.5710617303848267, "learning_rate": 3.093939744388243e-06, "loss": 0.11008461564779282, "step": 3604 }, { "epoch": 0.935877466251298, "grad_norm": 1.1011738777160645, "learning_rate": 3.093004822401664e-06, "loss": 0.14145292341709137, "step": 3605 }, { "epoch": 0.9361370716510904, "grad_norm": 0.564085841178894, "learning_rate": 3.092069839779686e-06, "loss": 0.10156415402889252, "step": 3606 }, { "epoch": 0.9363966770508827, "grad_norm": 1.4238866567611694, "learning_rate": 3.0911347966875925e-06, "loss": 0.13175445795059204, "step": 3607 }, { "epoch": 0.936656282450675, "grad_norm": 1.7116376161575317, "learning_rate": 3.090199693290678e-06, "loss": 0.14062738418579102, "step": 3608 }, { "epoch": 0.9369158878504673, "grad_norm": 1.399938702583313, "learning_rate": 3.0892645297542475e-06, "loss": 0.10215212404727936, "step": 3609 }, { "epoch": 0.9371754932502596, "grad_norm": 0.6083171963691711, "learning_rate": 3.088329306243616e-06, "loss": 0.10367842018604279, "step": 3610 }, { "epoch": 0.9374350986500519, "grad_norm": 0.9393215179443359, "learning_rate": 3.0873940229241095e-06, "loss": 0.11767024546861649, "step": 3611 }, { "epoch": 0.9376947040498442, "grad_norm": 1.5680803060531616, "learning_rate": 3.086458679961066e-06, "loss": 0.1087254136800766, "step": 3612 }, { "epoch": 0.9379543094496365, "grad_norm": 3.9665961265563965, "learning_rate": 3.0855232775198317e-06, "loss": 0.11858923733234406, "step": 3613 }, { "epoch": 0.9382139148494288, "grad_norm": 0.9419423937797546, "learning_rate": 3.084587815765764e-06, "loss": 0.14095017313957214, "step": 3614 }, { "epoch": 0.9384735202492211, "grad_norm": 1.0077632665634155, "learning_rate": 3.0836522948642317e-06, "loss": 0.14758875966072083, "step": 3615 }, { "epoch": 0.9387331256490135, "grad_norm": 1.170576810836792, "learning_rate": 3.0827167149806126e-06, "loss": 0.12544915080070496, "step": 3616 }, { "epoch": 0.9389927310488058, "grad_norm": 0.704822301864624, "learning_rate": 3.0817810762802974e-06, "loss": 0.11706969141960144, "step": 3617 }, { "epoch": 0.9392523364485982, "grad_norm": 0.9360300898551941, "learning_rate": 3.0808453789286842e-06, "loss": 0.12424904853105545, "step": 3618 }, { "epoch": 0.9395119418483905, "grad_norm": 0.9030289649963379, "learning_rate": 3.079909623091183e-06, "loss": 0.10867402702569962, "step": 3619 }, { "epoch": 0.9397715472481828, "grad_norm": 1.9389694929122925, "learning_rate": 3.078973808933215e-06, "loss": 0.14152616262435913, "step": 3620 }, { "epoch": 0.9400311526479751, "grad_norm": 0.9448820352554321, "learning_rate": 3.0780379366202106e-06, "loss": 0.13502687215805054, "step": 3621 }, { "epoch": 0.9402907580477674, "grad_norm": 0.8978136777877808, "learning_rate": 3.077102006317609e-06, "loss": 0.11596864461898804, "step": 3622 }, { "epoch": 0.9405503634475597, "grad_norm": 0.8051126003265381, "learning_rate": 3.0761660181908624e-06, "loss": 0.1418246030807495, "step": 3623 }, { "epoch": 0.940809968847352, "grad_norm": 1.2295987606048584, "learning_rate": 3.0752299724054324e-06, "loss": 0.16210369765758514, "step": 3624 }, { "epoch": 0.9410695742471443, "grad_norm": 1.328837275505066, "learning_rate": 3.0742938691267903e-06, "loss": 0.12894348800182343, "step": 3625 }, { "epoch": 0.9413291796469366, "grad_norm": 1.5187098979949951, "learning_rate": 3.073357708520417e-06, "loss": 0.12358137965202332, "step": 3626 }, { "epoch": 0.9415887850467289, "grad_norm": 0.6070700883865356, "learning_rate": 3.072421490751805e-06, "loss": 0.11419252306222916, "step": 3627 }, { "epoch": 0.9418483904465212, "grad_norm": 0.605068564414978, "learning_rate": 3.071485215986455e-06, "loss": 0.1080532819032669, "step": 3628 }, { "epoch": 0.9421079958463136, "grad_norm": 0.8893508911132812, "learning_rate": 3.0705488843898816e-06, "loss": 0.11787326633930206, "step": 3629 }, { "epoch": 0.942367601246106, "grad_norm": 0.8063585162162781, "learning_rate": 3.0696124961276036e-06, "loss": 0.1266651153564453, "step": 3630 }, { "epoch": 0.9426272066458983, "grad_norm": 1.343308448791504, "learning_rate": 3.0686760513651555e-06, "loss": 0.11769391596317291, "step": 3631 }, { "epoch": 0.9428868120456906, "grad_norm": 1.1381233930587769, "learning_rate": 3.0677395502680774e-06, "loss": 0.10990001261234283, "step": 3632 }, { "epoch": 0.9431464174454829, "grad_norm": 1.0886204242706299, "learning_rate": 3.066802993001922e-06, "loss": 0.10543236881494522, "step": 3633 }, { "epoch": 0.9434060228452752, "grad_norm": 1.1038905382156372, "learning_rate": 3.0658663797322518e-06, "loss": 0.13584816455841064, "step": 3634 }, { "epoch": 0.9436656282450675, "grad_norm": 0.8202618956565857, "learning_rate": 3.0649297106246386e-06, "loss": 0.12480396032333374, "step": 3635 }, { "epoch": 0.9439252336448598, "grad_norm": 0.41739949584007263, "learning_rate": 3.0639929858446626e-06, "loss": 0.10506580770015717, "step": 3636 }, { "epoch": 0.9441848390446521, "grad_norm": 0.4616498649120331, "learning_rate": 3.0630562055579154e-06, "loss": 0.1102537214756012, "step": 3637 }, { "epoch": 0.9444444444444444, "grad_norm": 0.5062544345855713, "learning_rate": 3.0621193699300004e-06, "loss": 0.11649258434772491, "step": 3638 }, { "epoch": 0.9447040498442367, "grad_norm": 1.1121011972427368, "learning_rate": 3.0611824791265278e-06, "loss": 0.11079806089401245, "step": 3639 }, { "epoch": 0.944963655244029, "grad_norm": 0.9625909328460693, "learning_rate": 3.060245533313117e-06, "loss": 0.13540561497211456, "step": 3640 }, { "epoch": 0.9452232606438213, "grad_norm": 1.1674225330352783, "learning_rate": 3.0593085326553995e-06, "loss": 0.134206160902977, "step": 3641 }, { "epoch": 0.9454828660436138, "grad_norm": 0.551033616065979, "learning_rate": 3.058371477319016e-06, "loss": 0.11165263503789902, "step": 3642 }, { "epoch": 0.9457424714434061, "grad_norm": 1.5879707336425781, "learning_rate": 3.0574343674696162e-06, "loss": 0.15153953433036804, "step": 3643 }, { "epoch": 0.9460020768431984, "grad_norm": 1.3152589797973633, "learning_rate": 3.05649720327286e-06, "loss": 0.14913102984428406, "step": 3644 }, { "epoch": 0.9462616822429907, "grad_norm": 1.3330570459365845, "learning_rate": 3.0555599848944152e-06, "loss": 0.12890908122062683, "step": 3645 }, { "epoch": 0.946521287642783, "grad_norm": 1.1075340509414673, "learning_rate": 3.054622712499962e-06, "loss": 0.1509559005498886, "step": 3646 }, { "epoch": 0.9467808930425753, "grad_norm": 0.8979610204696655, "learning_rate": 3.0536853862551875e-06, "loss": 0.1256677657365799, "step": 3647 }, { "epoch": 0.9470404984423676, "grad_norm": 1.294724702835083, "learning_rate": 3.0527480063257908e-06, "loss": 0.12028826773166656, "step": 3648 }, { "epoch": 0.9473001038421599, "grad_norm": 0.6716082096099854, "learning_rate": 3.0518105728774782e-06, "loss": 0.1040494441986084, "step": 3649 }, { "epoch": 0.9475597092419522, "grad_norm": 1.0279473066329956, "learning_rate": 3.0508730860759673e-06, "loss": 0.1190531998872757, "step": 3650 }, { "epoch": 0.9478193146417445, "grad_norm": 1.4481174945831299, "learning_rate": 3.0499355460869833e-06, "loss": 0.11207647621631622, "step": 3651 }, { "epoch": 0.9480789200415368, "grad_norm": 1.7348761558532715, "learning_rate": 3.048997953076263e-06, "loss": 0.09936198592185974, "step": 3652 }, { "epoch": 0.9483385254413291, "grad_norm": 2.005300283432007, "learning_rate": 3.0480603072095507e-06, "loss": 0.11458542197942734, "step": 3653 }, { "epoch": 0.9485981308411215, "grad_norm": 0.4545767605304718, "learning_rate": 3.0471226086526e-06, "loss": 0.11905691772699356, "step": 3654 }, { "epoch": 0.9488577362409139, "grad_norm": 2.6835989952087402, "learning_rate": 3.046184857571176e-06, "loss": 0.10934188961982727, "step": 3655 }, { "epoch": 0.9491173416407062, "grad_norm": 0.6915906071662903, "learning_rate": 3.0452470541310507e-06, "loss": 0.1017162948846817, "step": 3656 }, { "epoch": 0.9493769470404985, "grad_norm": 2.0352208614349365, "learning_rate": 3.0443091984980066e-06, "loss": 0.11972920596599579, "step": 3657 }, { "epoch": 0.9496365524402908, "grad_norm": 0.9071739315986633, "learning_rate": 3.0433712908378356e-06, "loss": 0.12709695100784302, "step": 3658 }, { "epoch": 0.9498961578400831, "grad_norm": 2.259082794189453, "learning_rate": 3.042433331316336e-06, "loss": 0.1154792457818985, "step": 3659 }, { "epoch": 0.9501557632398754, "grad_norm": 1.7085058689117432, "learning_rate": 3.04149532009932e-06, "loss": 0.14000196754932404, "step": 3660 }, { "epoch": 0.9504153686396677, "grad_norm": 1.5746783018112183, "learning_rate": 3.0405572573526067e-06, "loss": 0.11789582669734955, "step": 3661 }, { "epoch": 0.95067497403946, "grad_norm": 1.094720482826233, "learning_rate": 3.0396191432420225e-06, "loss": 0.13073763251304626, "step": 3662 }, { "epoch": 0.9509345794392523, "grad_norm": 0.582131564617157, "learning_rate": 3.0386809779334055e-06, "loss": 0.10843811184167862, "step": 3663 }, { "epoch": 0.9511941848390446, "grad_norm": 0.9942604899406433, "learning_rate": 3.037742761592601e-06, "loss": 0.11030279099941254, "step": 3664 }, { "epoch": 0.9514537902388369, "grad_norm": 1.3689028024673462, "learning_rate": 3.0368044943854647e-06, "loss": 0.14405052363872528, "step": 3665 }, { "epoch": 0.9517133956386293, "grad_norm": 2.1073670387268066, "learning_rate": 3.0358661764778614e-06, "loss": 0.13610011339187622, "step": 3666 }, { "epoch": 0.9519730010384216, "grad_norm": 0.6489287614822388, "learning_rate": 3.0349278080356626e-06, "loss": 0.12258259952068329, "step": 3667 }, { "epoch": 0.952232606438214, "grad_norm": 1.2131247520446777, "learning_rate": 3.0339893892247518e-06, "loss": 0.11172857880592346, "step": 3668 }, { "epoch": 0.9524922118380063, "grad_norm": 0.6340915560722351, "learning_rate": 3.0330509202110197e-06, "loss": 0.10316307842731476, "step": 3669 }, { "epoch": 0.9527518172377986, "grad_norm": 0.6923078894615173, "learning_rate": 3.032112401160366e-06, "loss": 0.11099937558174133, "step": 3670 }, { "epoch": 0.9530114226375909, "grad_norm": 3.4434428215026855, "learning_rate": 3.0311738322386996e-06, "loss": 0.13220252096652985, "step": 3671 }, { "epoch": 0.9532710280373832, "grad_norm": 0.4410495460033417, "learning_rate": 3.0302352136119364e-06, "loss": 0.12685658037662506, "step": 3672 }, { "epoch": 0.9535306334371755, "grad_norm": 0.8921287059783936, "learning_rate": 3.029296545446005e-06, "loss": 0.11882296204566956, "step": 3673 }, { "epoch": 0.9537902388369678, "grad_norm": 0.6809223890304565, "learning_rate": 3.0283578279068387e-06, "loss": 0.10141302645206451, "step": 3674 }, { "epoch": 0.9540498442367601, "grad_norm": 1.4803098440170288, "learning_rate": 3.027419061160383e-06, "loss": 0.15593159198760986, "step": 3675 }, { "epoch": 0.9543094496365524, "grad_norm": 1.608384370803833, "learning_rate": 3.026480245372588e-06, "loss": 0.138657346367836, "step": 3676 }, { "epoch": 0.9545690550363447, "grad_norm": 2.366518497467041, "learning_rate": 3.025541380709417e-06, "loss": 0.13319168984889984, "step": 3677 }, { "epoch": 0.9548286604361371, "grad_norm": 1.5027389526367188, "learning_rate": 3.024602467336839e-06, "loss": 0.13693471252918243, "step": 3678 }, { "epoch": 0.9550882658359294, "grad_norm": 0.9855160117149353, "learning_rate": 3.0236635054208322e-06, "loss": 0.11421099305152893, "step": 3679 }, { "epoch": 0.9553478712357217, "grad_norm": 1.8612451553344727, "learning_rate": 3.022724495127384e-06, "loss": 0.1398756206035614, "step": 3680 }, { "epoch": 0.955607476635514, "grad_norm": 1.32582426071167, "learning_rate": 3.0217854366224885e-06, "loss": 0.11584751307964325, "step": 3681 }, { "epoch": 0.9558670820353063, "grad_norm": 1.0809946060180664, "learning_rate": 3.020846330072152e-06, "loss": 0.133887380361557, "step": 3682 }, { "epoch": 0.9561266874350987, "grad_norm": 1.5275052785873413, "learning_rate": 3.019907175642386e-06, "loss": 0.12276848405599594, "step": 3683 }, { "epoch": 0.956386292834891, "grad_norm": 3.5838167667388916, "learning_rate": 3.0189679734992107e-06, "loss": 0.16039544343948364, "step": 3684 }, { "epoch": 0.9566458982346833, "grad_norm": 1.976257562637329, "learning_rate": 3.0180287238086563e-06, "loss": 0.13781867921352386, "step": 3685 }, { "epoch": 0.9569055036344756, "grad_norm": 0.5298016667366028, "learning_rate": 3.0170894267367605e-06, "loss": 0.12257799506187439, "step": 3686 }, { "epoch": 0.9571651090342679, "grad_norm": 1.722537875175476, "learning_rate": 3.0161500824495704e-06, "loss": 0.12189963459968567, "step": 3687 }, { "epoch": 0.9574247144340602, "grad_norm": 1.4069560766220093, "learning_rate": 3.015210691113139e-06, "loss": 0.11567997932434082, "step": 3688 }, { "epoch": 0.9576843198338525, "grad_norm": 0.9407320618629456, "learning_rate": 3.014271252893531e-06, "loss": 0.11584492027759552, "step": 3689 }, { "epoch": 0.9579439252336449, "grad_norm": 1.455040693283081, "learning_rate": 3.0133317679568145e-06, "loss": 0.11850917339324951, "step": 3690 }, { "epoch": 0.9582035306334372, "grad_norm": 0.8647380471229553, "learning_rate": 3.012392236469073e-06, "loss": 0.10654205083847046, "step": 3691 }, { "epoch": 0.9584631360332295, "grad_norm": 0.7222638726234436, "learning_rate": 3.0114526585963903e-06, "loss": 0.1049661785364151, "step": 3692 }, { "epoch": 0.9587227414330218, "grad_norm": 1.35196852684021, "learning_rate": 3.0105130345048654e-06, "loss": 0.110152468085289, "step": 3693 }, { "epoch": 0.9589823468328141, "grad_norm": 0.5427117347717285, "learning_rate": 3.0095733643606005e-06, "loss": 0.10056132078170776, "step": 3694 }, { "epoch": 0.9592419522326064, "grad_norm": 0.7672409415245056, "learning_rate": 3.0086336483297084e-06, "loss": 0.11649681627750397, "step": 3695 }, { "epoch": 0.9595015576323987, "grad_norm": 0.7768580913543701, "learning_rate": 3.007693886578309e-06, "loss": 0.12169135361909866, "step": 3696 }, { "epoch": 0.9597611630321911, "grad_norm": 1.3771610260009766, "learning_rate": 3.0067540792725304e-06, "loss": 0.14866074919700623, "step": 3697 }, { "epoch": 0.9600207684319834, "grad_norm": 1.9178318977355957, "learning_rate": 3.00581422657851e-06, "loss": 0.14198409020900726, "step": 3698 }, { "epoch": 0.9602803738317757, "grad_norm": 0.9173719882965088, "learning_rate": 3.004874328662391e-06, "loss": 0.13111254572868347, "step": 3699 }, { "epoch": 0.960539979231568, "grad_norm": 1.2896473407745361, "learning_rate": 3.003934385690328e-06, "loss": 0.10060685873031616, "step": 3700 }, { "epoch": 0.9607995846313603, "grad_norm": 2.500861406326294, "learning_rate": 3.002994397828479e-06, "loss": 0.16615456342697144, "step": 3701 }, { "epoch": 0.9610591900311527, "grad_norm": 1.5886404514312744, "learning_rate": 3.002054365243014e-06, "loss": 0.09866541624069214, "step": 3702 }, { "epoch": 0.961318795430945, "grad_norm": 0.9247984886169434, "learning_rate": 3.001114288100107e-06, "loss": 0.11787340044975281, "step": 3703 }, { "epoch": 0.9615784008307373, "grad_norm": 1.7316577434539795, "learning_rate": 3.0001741665659444e-06, "loss": 0.09457825869321823, "step": 3704 }, { "epoch": 0.9618380062305296, "grad_norm": 0.8907904624938965, "learning_rate": 2.9992340008067177e-06, "loss": 0.12012028694152832, "step": 3705 }, { "epoch": 0.9620976116303219, "grad_norm": 0.6182599067687988, "learning_rate": 2.9982937909886267e-06, "loss": 0.10888855904340744, "step": 3706 }, { "epoch": 0.9623572170301142, "grad_norm": 2.0272979736328125, "learning_rate": 2.997353537277878e-06, "loss": 0.13138067722320557, "step": 3707 }, { "epoch": 0.9626168224299065, "grad_norm": 0.8788102865219116, "learning_rate": 2.996413239840687e-06, "loss": 0.1489306539297104, "step": 3708 }, { "epoch": 0.9628764278296988, "grad_norm": 0.5109236240386963, "learning_rate": 2.995472898843278e-06, "loss": 0.11016827076673508, "step": 3709 }, { "epoch": 0.9631360332294912, "grad_norm": 1.167911410331726, "learning_rate": 2.994532514451881e-06, "loss": 0.12834641337394714, "step": 3710 }, { "epoch": 0.9633956386292835, "grad_norm": 1.0967859029769897, "learning_rate": 2.9935920868327345e-06, "loss": 0.13065756857395172, "step": 3711 }, { "epoch": 0.9636552440290758, "grad_norm": 1.2013965845108032, "learning_rate": 2.9926516161520848e-06, "loss": 0.1269702911376953, "step": 3712 }, { "epoch": 0.9639148494288681, "grad_norm": 0.8535230159759521, "learning_rate": 2.9917111025761848e-06, "loss": 0.12120828032493591, "step": 3713 }, { "epoch": 0.9641744548286605, "grad_norm": 0.900721549987793, "learning_rate": 2.9907705462712974e-06, "loss": 0.13999660313129425, "step": 3714 }, { "epoch": 0.9644340602284528, "grad_norm": 1.495367169380188, "learning_rate": 2.989829947403689e-06, "loss": 0.11327357590198517, "step": 3715 }, { "epoch": 0.9646936656282451, "grad_norm": 0.7511675357818604, "learning_rate": 2.9888893061396375e-06, "loss": 0.11963999271392822, "step": 3716 }, { "epoch": 0.9649532710280374, "grad_norm": 2.1479907035827637, "learning_rate": 2.987948622645427e-06, "loss": 0.08747071027755737, "step": 3717 }, { "epoch": 0.9652128764278297, "grad_norm": 0.6529593467712402, "learning_rate": 2.9870078970873483e-06, "loss": 0.11328043788671494, "step": 3718 }, { "epoch": 0.965472481827622, "grad_norm": 1.2235214710235596, "learning_rate": 2.9860671296317e-06, "loss": 0.1196330338716507, "step": 3719 }, { "epoch": 0.9657320872274143, "grad_norm": 0.6153112053871155, "learning_rate": 2.985126320444789e-06, "loss": 0.11735290288925171, "step": 3720 }, { "epoch": 0.9659916926272066, "grad_norm": 0.5406326055526733, "learning_rate": 2.9841854696929266e-06, "loss": 0.12039783596992493, "step": 3721 }, { "epoch": 0.9662512980269989, "grad_norm": 1.255975365638733, "learning_rate": 2.983244577542437e-06, "loss": 0.12994559109210968, "step": 3722 }, { "epoch": 0.9665109034267912, "grad_norm": 1.5021666288375854, "learning_rate": 2.9823036441596465e-06, "loss": 0.10695203393697739, "step": 3723 }, { "epoch": 0.9667705088265836, "grad_norm": 0.7586162090301514, "learning_rate": 2.98136266971089e-06, "loss": 0.11617104709148407, "step": 3724 }, { "epoch": 0.9670301142263759, "grad_norm": 0.5209779143333435, "learning_rate": 2.980421654362511e-06, "loss": 0.13101866841316223, "step": 3725 }, { "epoch": 0.9672897196261683, "grad_norm": 1.2993080615997314, "learning_rate": 2.97948059828086e-06, "loss": 0.14483709633350372, "step": 3726 }, { "epoch": 0.9675493250259606, "grad_norm": 1.1443195343017578, "learning_rate": 2.978539501632293e-06, "loss": 0.11546622216701508, "step": 3727 }, { "epoch": 0.9678089304257529, "grad_norm": 1.8374396562576294, "learning_rate": 2.977598364583176e-06, "loss": 0.10749003291130066, "step": 3728 }, { "epoch": 0.9680685358255452, "grad_norm": 0.7468070983886719, "learning_rate": 2.9766571872998777e-06, "loss": 0.12728282809257507, "step": 3729 }, { "epoch": 0.9683281412253375, "grad_norm": 1.1276944875717163, "learning_rate": 2.9757159699487798e-06, "loss": 0.13099007308483124, "step": 3730 }, { "epoch": 0.9685877466251298, "grad_norm": 1.1307423114776611, "learning_rate": 2.9747747126962666e-06, "loss": 0.13215936720371246, "step": 3731 }, { "epoch": 0.9688473520249221, "grad_norm": 0.6328394412994385, "learning_rate": 2.9738334157087307e-06, "loss": 0.12193648517131805, "step": 3732 }, { "epoch": 0.9691069574247144, "grad_norm": 1.1123926639556885, "learning_rate": 2.9728920791525724e-06, "loss": 0.13951602578163147, "step": 3733 }, { "epoch": 0.9693665628245067, "grad_norm": 1.0152240991592407, "learning_rate": 2.971950703194197e-06, "loss": 0.11171071231365204, "step": 3734 }, { "epoch": 0.969626168224299, "grad_norm": 1.658255934715271, "learning_rate": 2.97100928800002e-06, "loss": 0.10963352024555206, "step": 3735 }, { "epoch": 0.9698857736240913, "grad_norm": 1.088563084602356, "learning_rate": 2.9700678337364615e-06, "loss": 0.1525009572505951, "step": 3736 }, { "epoch": 0.9701453790238836, "grad_norm": 1.067818522453308, "learning_rate": 2.9691263405699485e-06, "loss": 0.125337615609169, "step": 3737 }, { "epoch": 0.9704049844236761, "grad_norm": 1.3864532709121704, "learning_rate": 2.968184808666916e-06, "loss": 0.13962005078792572, "step": 3738 }, { "epoch": 0.9706645898234684, "grad_norm": 1.7341859340667725, "learning_rate": 2.967243238193805e-06, "loss": 0.11394108831882477, "step": 3739 }, { "epoch": 0.9709241952232607, "grad_norm": 0.8348596692085266, "learning_rate": 2.9663016293170632e-06, "loss": 0.10358917713165283, "step": 3740 }, { "epoch": 0.971183800623053, "grad_norm": 1.5448195934295654, "learning_rate": 2.965359982203147e-06, "loss": 0.12664848566055298, "step": 3741 }, { "epoch": 0.9714434060228453, "grad_norm": 1.3824775218963623, "learning_rate": 2.9644182970185163e-06, "loss": 0.09953880310058594, "step": 3742 }, { "epoch": 0.9717030114226376, "grad_norm": 1.1806915998458862, "learning_rate": 2.96347657392964e-06, "loss": 0.09468217194080353, "step": 3743 }, { "epoch": 0.9719626168224299, "grad_norm": 0.833479106426239, "learning_rate": 2.962534813102994e-06, "loss": 0.13852664828300476, "step": 3744 }, { "epoch": 0.9722222222222222, "grad_norm": 0.9444780945777893, "learning_rate": 2.9615930147050597e-06, "loss": 0.11415073275566101, "step": 3745 }, { "epoch": 0.9724818276220145, "grad_norm": 0.712580680847168, "learning_rate": 2.9606511789023247e-06, "loss": 0.11141888797283173, "step": 3746 }, { "epoch": 0.9727414330218068, "grad_norm": 1.4863907098770142, "learning_rate": 2.9597093058612843e-06, "loss": 0.12805208563804626, "step": 3747 }, { "epoch": 0.9730010384215991, "grad_norm": 0.928066074848175, "learning_rate": 2.958767395748441e-06, "loss": 0.12672141194343567, "step": 3748 }, { "epoch": 0.9732606438213914, "grad_norm": 1.3152936697006226, "learning_rate": 2.957825448730302e-06, "loss": 0.13488127291202545, "step": 3749 }, { "epoch": 0.9735202492211839, "grad_norm": 0.6730455756187439, "learning_rate": 2.9568834649733825e-06, "loss": 0.11885574460029602, "step": 3750 }, { "epoch": 0.9737798546209762, "grad_norm": 1.6487421989440918, "learning_rate": 2.9559414446442024e-06, "loss": 0.1068938821554184, "step": 3751 }, { "epoch": 0.9740394600207685, "grad_norm": 1.3264007568359375, "learning_rate": 2.9549993879092912e-06, "loss": 0.11423000693321228, "step": 3752 }, { "epoch": 0.9742990654205608, "grad_norm": 0.5239525437355042, "learning_rate": 2.954057294935182e-06, "loss": 0.10969862341880798, "step": 3753 }, { "epoch": 0.9745586708203531, "grad_norm": 0.8229908347129822, "learning_rate": 2.9531151658884156e-06, "loss": 0.13012072443962097, "step": 3754 }, { "epoch": 0.9748182762201454, "grad_norm": 1.5475282669067383, "learning_rate": 2.9521730009355383e-06, "loss": 0.1327424943447113, "step": 3755 }, { "epoch": 0.9750778816199377, "grad_norm": 1.454093337059021, "learning_rate": 2.951230800243103e-06, "loss": 0.10438282787799835, "step": 3756 }, { "epoch": 0.97533748701973, "grad_norm": 1.4827369451522827, "learning_rate": 2.95028856397767e-06, "loss": 0.14752933382987976, "step": 3757 }, { "epoch": 0.9755970924195223, "grad_norm": 1.5650290250778198, "learning_rate": 2.9493462923058057e-06, "loss": 0.11983222514390945, "step": 3758 }, { "epoch": 0.9758566978193146, "grad_norm": 1.1529080867767334, "learning_rate": 2.948403985394081e-06, "loss": 0.13661114871501923, "step": 3759 }, { "epoch": 0.9761163032191069, "grad_norm": 0.9986563920974731, "learning_rate": 2.9474616434090737e-06, "loss": 0.1098552718758583, "step": 3760 }, { "epoch": 0.9763759086188992, "grad_norm": 0.9294162392616272, "learning_rate": 2.9465192665173697e-06, "loss": 0.1393057107925415, "step": 3761 }, { "epoch": 0.9766355140186916, "grad_norm": 0.790172815322876, "learning_rate": 2.9455768548855597e-06, "loss": 0.13261984288692474, "step": 3762 }, { "epoch": 0.976895119418484, "grad_norm": 0.9690536856651306, "learning_rate": 2.944634408680239e-06, "loss": 0.12220969796180725, "step": 3763 }, { "epoch": 0.9771547248182763, "grad_norm": 0.6556792855262756, "learning_rate": 2.9436919280680115e-06, "loss": 0.12011049687862396, "step": 3764 }, { "epoch": 0.9774143302180686, "grad_norm": 0.7164608836174011, "learning_rate": 2.942749413215486e-06, "loss": 0.11715318262577057, "step": 3765 }, { "epoch": 0.9776739356178609, "grad_norm": 0.7095233798027039, "learning_rate": 2.941806864289279e-06, "loss": 0.11863058805465698, "step": 3766 }, { "epoch": 0.9779335410176532, "grad_norm": 1.0901761054992676, "learning_rate": 2.9408642814560082e-06, "loss": 0.12310856580734253, "step": 3767 }, { "epoch": 0.9781931464174455, "grad_norm": 1.0520524978637695, "learning_rate": 2.939921664882304e-06, "loss": 0.11435052752494812, "step": 3768 }, { "epoch": 0.9784527518172378, "grad_norm": 2.0011510848999023, "learning_rate": 2.9389790147347973e-06, "loss": 0.12460924685001373, "step": 3769 }, { "epoch": 0.9787123572170301, "grad_norm": 2.3292574882507324, "learning_rate": 2.9380363311801286e-06, "loss": 0.1479063630104065, "step": 3770 }, { "epoch": 0.9789719626168224, "grad_norm": 1.1696958541870117, "learning_rate": 2.9370936143849415e-06, "loss": 0.10385873913764954, "step": 3771 }, { "epoch": 0.9792315680166147, "grad_norm": 1.2346131801605225, "learning_rate": 2.9361508645158875e-06, "loss": 0.11256121098995209, "step": 3772 }, { "epoch": 0.979491173416407, "grad_norm": 1.1628338098526, "learning_rate": 2.9352080817396227e-06, "loss": 0.13049431145191193, "step": 3773 }, { "epoch": 0.9797507788161994, "grad_norm": 1.1699522733688354, "learning_rate": 2.93426526622281e-06, "loss": 0.11946050822734833, "step": 3774 }, { "epoch": 0.9800103842159917, "grad_norm": 1.1001901626586914, "learning_rate": 2.933322418132117e-06, "loss": 0.13529247045516968, "step": 3775 }, { "epoch": 0.980269989615784, "grad_norm": 1.1291133165359497, "learning_rate": 2.932379537634218e-06, "loss": 0.10036106407642365, "step": 3776 }, { "epoch": 0.9805295950155763, "grad_norm": 1.824533462524414, "learning_rate": 2.9314366248957927e-06, "loss": 0.09415817260742188, "step": 3777 }, { "epoch": 0.9807892004153687, "grad_norm": 1.0265719890594482, "learning_rate": 2.9304936800835264e-06, "loss": 0.11673387885093689, "step": 3778 }, { "epoch": 0.981048805815161, "grad_norm": 1.2400529384613037, "learning_rate": 2.9295507033641105e-06, "loss": 0.12604692578315735, "step": 3779 }, { "epoch": 0.9813084112149533, "grad_norm": 1.3453766107559204, "learning_rate": 2.9286076949042415e-06, "loss": 0.117560014128685, "step": 3780 }, { "epoch": 0.9815680166147456, "grad_norm": 1.357344388961792, "learning_rate": 2.9276646548706222e-06, "loss": 0.12200293689966202, "step": 3781 }, { "epoch": 0.9818276220145379, "grad_norm": 1.4366551637649536, "learning_rate": 2.9267215834299585e-06, "loss": 0.11060323566198349, "step": 3782 }, { "epoch": 0.9820872274143302, "grad_norm": 1.6809842586517334, "learning_rate": 2.9257784807489663e-06, "loss": 0.10128430277109146, "step": 3783 }, { "epoch": 0.9823468328141225, "grad_norm": 1.9291126728057861, "learning_rate": 2.924835346994363e-06, "loss": 0.11620382964611053, "step": 3784 }, { "epoch": 0.9826064382139148, "grad_norm": 0.6821800470352173, "learning_rate": 2.9238921823328737e-06, "loss": 0.11302037537097931, "step": 3785 }, { "epoch": 0.9828660436137072, "grad_norm": 0.5118109583854675, "learning_rate": 2.9229489869312276e-06, "loss": 0.11178793013095856, "step": 3786 }, { "epoch": 0.9831256490134995, "grad_norm": 1.1469089984893799, "learning_rate": 2.9220057609561616e-06, "loss": 0.10663264244794846, "step": 3787 }, { "epoch": 0.9833852544132918, "grad_norm": 1.565632939338684, "learning_rate": 2.9210625045744143e-06, "loss": 0.11297101527452469, "step": 3788 }, { "epoch": 0.9836448598130841, "grad_norm": 0.9689202904701233, "learning_rate": 2.920119217952735e-06, "loss": 0.11225748062133789, "step": 3789 }, { "epoch": 0.9839044652128764, "grad_norm": 1.2722679376602173, "learning_rate": 2.9191759012578712e-06, "loss": 0.10092244297266006, "step": 3790 }, { "epoch": 0.9841640706126688, "grad_norm": 0.7790635228157043, "learning_rate": 2.9182325546565822e-06, "loss": 0.11490773409605026, "step": 3791 }, { "epoch": 0.9844236760124611, "grad_norm": 0.4782704710960388, "learning_rate": 2.91728917831563e-06, "loss": 0.11703354120254517, "step": 3792 }, { "epoch": 0.9846832814122534, "grad_norm": 2.2768499851226807, "learning_rate": 2.916345772401781e-06, "loss": 0.1356249749660492, "step": 3793 }, { "epoch": 0.9849428868120457, "grad_norm": 1.1301714181900024, "learning_rate": 2.9154023370818085e-06, "loss": 0.12422753125429153, "step": 3794 }, { "epoch": 0.985202492211838, "grad_norm": 0.781060516834259, "learning_rate": 2.91445887252249e-06, "loss": 0.10895030945539474, "step": 3795 }, { "epoch": 0.9854620976116303, "grad_norm": 0.9027605652809143, "learning_rate": 2.9135153788906084e-06, "loss": 0.13653923571109772, "step": 3796 }, { "epoch": 0.9857217030114226, "grad_norm": 0.5311713218688965, "learning_rate": 2.912571856352952e-06, "loss": 0.11123089492321014, "step": 3797 }, { "epoch": 0.985981308411215, "grad_norm": 0.8519848585128784, "learning_rate": 2.9116283050763137e-06, "loss": 0.1362316906452179, "step": 3798 }, { "epoch": 0.9862409138110073, "grad_norm": 1.3050018548965454, "learning_rate": 2.910684725227492e-06, "loss": 0.13664112985134125, "step": 3799 }, { "epoch": 0.9865005192107996, "grad_norm": 1.0136055946350098, "learning_rate": 2.909741116973289e-06, "loss": 0.10877678543329239, "step": 3800 }, { "epoch": 0.9867601246105919, "grad_norm": 1.3284776210784912, "learning_rate": 2.908797480480515e-06, "loss": 0.11275890469551086, "step": 3801 }, { "epoch": 0.9870197300103842, "grad_norm": 1.7243030071258545, "learning_rate": 2.9078538159159823e-06, "loss": 0.1221725270152092, "step": 3802 }, { "epoch": 0.9872793354101765, "grad_norm": 2.3629608154296875, "learning_rate": 2.90691012344651e-06, "loss": 0.13006871938705444, "step": 3803 }, { "epoch": 0.9875389408099688, "grad_norm": 1.5319974422454834, "learning_rate": 2.90596640323892e-06, "loss": 0.14080245792865753, "step": 3804 }, { "epoch": 0.9877985462097612, "grad_norm": 1.2289119958877563, "learning_rate": 2.905022655460041e-06, "loss": 0.12064732611179352, "step": 3805 }, { "epoch": 0.9880581516095535, "grad_norm": 0.5641564130783081, "learning_rate": 2.9040788802767065e-06, "loss": 0.11112260073423386, "step": 3806 }, { "epoch": 0.9883177570093458, "grad_norm": 0.6224297285079956, "learning_rate": 2.903135077855754e-06, "loss": 0.12083721160888672, "step": 3807 }, { "epoch": 0.9885773624091381, "grad_norm": 1.1779615879058838, "learning_rate": 2.902191248364025e-06, "loss": 0.10865093767642975, "step": 3808 }, { "epoch": 0.9888369678089304, "grad_norm": 1.396714448928833, "learning_rate": 2.9012473919683686e-06, "loss": 0.13749399781227112, "step": 3809 }, { "epoch": 0.9890965732087228, "grad_norm": 0.47647303342819214, "learning_rate": 2.900303508835637e-06, "loss": 0.11756657063961029, "step": 3810 }, { "epoch": 0.9893561786085151, "grad_norm": 1.1228222846984863, "learning_rate": 2.8993595991326854e-06, "loss": 0.1311112493276596, "step": 3811 }, { "epoch": 0.9896157840083074, "grad_norm": 1.3236984014511108, "learning_rate": 2.8984156630263777e-06, "loss": 0.10601944476366043, "step": 3812 }, { "epoch": 0.9898753894080997, "grad_norm": 1.5266011953353882, "learning_rate": 2.8974717006835785e-06, "loss": 0.15981478989124298, "step": 3813 }, { "epoch": 0.990134994807892, "grad_norm": 1.2129250764846802, "learning_rate": 2.8965277122711593e-06, "loss": 0.12019418179988861, "step": 3814 }, { "epoch": 0.9903946002076843, "grad_norm": 1.5269582271575928, "learning_rate": 2.895583697955995e-06, "loss": 0.15467655658721924, "step": 3815 }, { "epoch": 0.9906542056074766, "grad_norm": 1.3170939683914185, "learning_rate": 2.8946396579049674e-06, "loss": 0.1233711689710617, "step": 3816 }, { "epoch": 0.9909138110072689, "grad_norm": 0.966560959815979, "learning_rate": 2.8936955922849597e-06, "loss": 0.1082422062754631, "step": 3817 }, { "epoch": 0.9911734164070612, "grad_norm": 1.3856091499328613, "learning_rate": 2.8927515012628616e-06, "loss": 0.13195134699344635, "step": 3818 }, { "epoch": 0.9914330218068536, "grad_norm": 1.0272787809371948, "learning_rate": 2.8918073850055663e-06, "loss": 0.11316933482885361, "step": 3819 }, { "epoch": 0.9916926272066459, "grad_norm": 0.7880271077156067, "learning_rate": 2.8908632436799725e-06, "loss": 0.13361522555351257, "step": 3820 }, { "epoch": 0.9919522326064382, "grad_norm": 1.4239951372146606, "learning_rate": 2.8899190774529823e-06, "loss": 0.12118130922317505, "step": 3821 }, { "epoch": 0.9922118380062306, "grad_norm": 2.1592652797698975, "learning_rate": 2.888974886491503e-06, "loss": 0.15197879076004028, "step": 3822 }, { "epoch": 0.9924714434060229, "grad_norm": 1.6915431022644043, "learning_rate": 2.888030670962446e-06, "loss": 0.1065332442522049, "step": 3823 }, { "epoch": 0.9927310488058152, "grad_norm": 1.2033659219741821, "learning_rate": 2.8870864310327267e-06, "loss": 0.09213057160377502, "step": 3824 }, { "epoch": 0.9929906542056075, "grad_norm": 0.8666830062866211, "learning_rate": 2.886142166869265e-06, "loss": 0.12178542464971542, "step": 3825 }, { "epoch": 0.9932502596053998, "grad_norm": 0.771435558795929, "learning_rate": 2.885197878638986e-06, "loss": 0.1053818017244339, "step": 3826 }, { "epoch": 0.9935098650051921, "grad_norm": 1.569062352180481, "learning_rate": 2.8842535665088177e-06, "loss": 0.11586170643568039, "step": 3827 }, { "epoch": 0.9937694704049844, "grad_norm": 1.749344825744629, "learning_rate": 2.8833092306456934e-06, "loss": 0.1173168271780014, "step": 3828 }, { "epoch": 0.9940290758047767, "grad_norm": 1.5364100933074951, "learning_rate": 2.882364871216548e-06, "loss": 0.12096813321113586, "step": 3829 }, { "epoch": 0.994288681204569, "grad_norm": 1.0976427793502808, "learning_rate": 2.881420488388325e-06, "loss": 0.12412351369857788, "step": 3830 }, { "epoch": 0.9945482866043613, "grad_norm": 0.49669149518013, "learning_rate": 2.8804760823279687e-06, "loss": 0.11065071821212769, "step": 3831 }, { "epoch": 0.9948078920041536, "grad_norm": 1.955161213874817, "learning_rate": 2.8795316532024287e-06, "loss": 0.14792019128799438, "step": 3832 }, { "epoch": 0.995067497403946, "grad_norm": 0.8339216709136963, "learning_rate": 2.878587201178658e-06, "loss": 0.1245688870549202, "step": 3833 }, { "epoch": 0.9953271028037384, "grad_norm": 1.03313148021698, "learning_rate": 2.8776427264236147e-06, "loss": 0.13762280344963074, "step": 3834 }, { "epoch": 0.9955867082035307, "grad_norm": 1.5212956666946411, "learning_rate": 2.8766982291042593e-06, "loss": 0.13161908090114594, "step": 3835 }, { "epoch": 0.995846313603323, "grad_norm": 0.8366340398788452, "learning_rate": 2.875753709387559e-06, "loss": 0.1054236888885498, "step": 3836 }, { "epoch": 0.9961059190031153, "grad_norm": 1.246438980102539, "learning_rate": 2.8748091674404816e-06, "loss": 0.11956246197223663, "step": 3837 }, { "epoch": 0.9963655244029076, "grad_norm": 2.187342882156372, "learning_rate": 2.8738646034300015e-06, "loss": 0.10728167742490768, "step": 3838 }, { "epoch": 0.9966251298026999, "grad_norm": 1.7081471681594849, "learning_rate": 2.8729200175230955e-06, "loss": 0.1444810926914215, "step": 3839 }, { "epoch": 0.9968847352024922, "grad_norm": 1.397011160850525, "learning_rate": 2.8719754098867446e-06, "loss": 0.12355025857686996, "step": 3840 }, { "epoch": 0.9971443406022845, "grad_norm": 0.7868269085884094, "learning_rate": 2.8710307806879343e-06, "loss": 0.11257244646549225, "step": 3841 }, { "epoch": 0.9974039460020768, "grad_norm": 1.8437389135360718, "learning_rate": 2.870086130093653e-06, "loss": 0.13506615161895752, "step": 3842 }, { "epoch": 0.9976635514018691, "grad_norm": 1.2914085388183594, "learning_rate": 2.8691414582708943e-06, "loss": 0.11636663228273392, "step": 3843 }, { "epoch": 0.9979231568016614, "grad_norm": 1.880534291267395, "learning_rate": 2.8681967653866527e-06, "loss": 0.08052729815244675, "step": 3844 }, { "epoch": 0.9981827622014537, "grad_norm": 0.6480333805084229, "learning_rate": 2.86725205160793e-06, "loss": 0.12132562696933746, "step": 3845 }, { "epoch": 0.9984423676012462, "grad_norm": 1.1808347702026367, "learning_rate": 2.8663073171017284e-06, "loss": 0.11190977692604065, "step": 3846 }, { "epoch": 0.9987019730010385, "grad_norm": 1.3156135082244873, "learning_rate": 2.865362562035058e-06, "loss": 0.13003727793693542, "step": 3847 }, { "epoch": 0.9989615784008308, "grad_norm": 1.6033433675765991, "learning_rate": 2.8644177865749266e-06, "loss": 0.13632966578006744, "step": 3848 }, { "epoch": 0.9992211838006231, "grad_norm": 1.7059190273284912, "learning_rate": 2.8634729908883506e-06, "loss": 0.12513668835163116, "step": 3849 }, { "epoch": 0.9994807892004154, "grad_norm": 1.4860705137252808, "learning_rate": 2.8625281751423485e-06, "loss": 0.14996124804019928, "step": 3850 }, { "epoch": 0.9997403946002077, "grad_norm": 0.866325855255127, "learning_rate": 2.861583339503941e-06, "loss": 0.11126700043678284, "step": 3851 }, { "epoch": 1.0, "grad_norm": 1.884236454963684, "learning_rate": 2.8606384841401548e-06, "loss": 0.11065177619457245, "step": 3852 }, { "epoch": 1.0002596053997923, "grad_norm": 1.2337764501571655, "learning_rate": 2.8596936092180173e-06, "loss": 0.14602336287498474, "step": 3853 }, { "epoch": 1.0005192107995846, "grad_norm": 1.251979112625122, "learning_rate": 2.858748714904561e-06, "loss": 0.11600523442029953, "step": 3854 }, { "epoch": 1.000778816199377, "grad_norm": 0.7514260411262512, "learning_rate": 2.857803801366823e-06, "loss": 0.11505185067653656, "step": 3855 }, { "epoch": 1.0010384215991692, "grad_norm": 1.4043197631835938, "learning_rate": 2.85685886877184e-06, "loss": 0.09492863714694977, "step": 3856 }, { "epoch": 1.0012980269989615, "grad_norm": 1.184584140777588, "learning_rate": 2.8559139172866563e-06, "loss": 0.1114528626203537, "step": 3857 }, { "epoch": 1.0015576323987538, "grad_norm": 0.9406325817108154, "learning_rate": 2.854968947078317e-06, "loss": 0.11497830599546432, "step": 3858 }, { "epoch": 1.0018172377985461, "grad_norm": 1.5625959634780884, "learning_rate": 2.854023958313872e-06, "loss": 0.1255756914615631, "step": 3859 }, { "epoch": 1.0020768431983385, "grad_norm": 1.4970570802688599, "learning_rate": 2.853078951160373e-06, "loss": 0.09053562581539154, "step": 3860 }, { "epoch": 1.0023364485981308, "grad_norm": 0.49850893020629883, "learning_rate": 2.852133925784875e-06, "loss": 0.11220589280128479, "step": 3861 }, { "epoch": 1.002596053997923, "grad_norm": 1.2297990322113037, "learning_rate": 2.8511888823544376e-06, "loss": 0.11031800508499146, "step": 3862 }, { "epoch": 1.0028556593977154, "grad_norm": 2.129093885421753, "learning_rate": 2.850243821036123e-06, "loss": 0.11054401099681854, "step": 3863 }, { "epoch": 1.0031152647975077, "grad_norm": 0.6866240501403809, "learning_rate": 2.849298741996996e-06, "loss": 0.1046447828412056, "step": 3864 }, { "epoch": 1.0033748701973002, "grad_norm": 1.6080220937728882, "learning_rate": 2.8483536454041245e-06, "loss": 0.11307187378406525, "step": 3865 }, { "epoch": 1.0036344755970925, "grad_norm": 0.9103686213493347, "learning_rate": 2.847408531424581e-06, "loss": 0.1069146990776062, "step": 3866 }, { "epoch": 1.0038940809968848, "grad_norm": 1.4332008361816406, "learning_rate": 2.8464634002254387e-06, "loss": 0.08589915931224823, "step": 3867 }, { "epoch": 1.0041536863966771, "grad_norm": 2.167208194732666, "learning_rate": 2.8455182519737774e-06, "loss": 0.14323171973228455, "step": 3868 }, { "epoch": 1.0044132917964694, "grad_norm": 1.298835277557373, "learning_rate": 2.844573086836675e-06, "loss": 0.13718847930431366, "step": 3869 }, { "epoch": 1.0046728971962617, "grad_norm": 1.2289190292358398, "learning_rate": 2.8436279049812167e-06, "loss": 0.09378133714199066, "step": 3870 }, { "epoch": 1.004932502596054, "grad_norm": 1.2546080350875854, "learning_rate": 2.8426827065744887e-06, "loss": 0.13711796700954437, "step": 3871 }, { "epoch": 1.0051921079958464, "grad_norm": 1.4289569854736328, "learning_rate": 2.8417374917835804e-06, "loss": 0.10649722814559937, "step": 3872 }, { "epoch": 1.0054517133956387, "grad_norm": 1.0974849462509155, "learning_rate": 2.8407922607755843e-06, "loss": 0.10097813606262207, "step": 3873 }, { "epoch": 1.005711318795431, "grad_norm": 0.7555848360061646, "learning_rate": 2.839847013717595e-06, "loss": 0.116892509162426, "step": 3874 }, { "epoch": 1.0059709241952233, "grad_norm": 1.4964104890823364, "learning_rate": 2.838901750776711e-06, "loss": 0.10749995708465576, "step": 3875 }, { "epoch": 1.0062305295950156, "grad_norm": 1.2448718547821045, "learning_rate": 2.8379564721200333e-06, "loss": 0.13785725831985474, "step": 3876 }, { "epoch": 1.0064901349948079, "grad_norm": 0.6032921075820923, "learning_rate": 2.8370111779146646e-06, "loss": 0.13108806312084198, "step": 3877 }, { "epoch": 1.0067497403946002, "grad_norm": 0.6492344737052917, "learning_rate": 2.8360658683277125e-06, "loss": 0.1055123507976532, "step": 3878 }, { "epoch": 1.0070093457943925, "grad_norm": 0.8508695960044861, "learning_rate": 2.835120543526285e-06, "loss": 0.09314541518688202, "step": 3879 }, { "epoch": 1.0072689511941848, "grad_norm": 1.5589301586151123, "learning_rate": 2.8341752036774945e-06, "loss": 0.10530807077884674, "step": 3880 }, { "epoch": 1.007528556593977, "grad_norm": 1.021140217781067, "learning_rate": 2.8332298489484555e-06, "loss": 0.08535963296890259, "step": 3881 }, { "epoch": 1.0077881619937694, "grad_norm": 1.031917929649353, "learning_rate": 2.832284479506286e-06, "loss": 0.11378469318151474, "step": 3882 }, { "epoch": 1.0080477673935617, "grad_norm": 1.482170820236206, "learning_rate": 2.8313390955181025e-06, "loss": 0.1337224841117859, "step": 3883 }, { "epoch": 1.008307372793354, "grad_norm": 0.792189359664917, "learning_rate": 2.830393697151031e-06, "loss": 0.13208450376987457, "step": 3884 }, { "epoch": 1.0085669781931463, "grad_norm": 0.8739804625511169, "learning_rate": 2.8294482845721934e-06, "loss": 0.09814782440662384, "step": 3885 }, { "epoch": 1.0088265835929386, "grad_norm": 0.8358056545257568, "learning_rate": 2.828502857948719e-06, "loss": 0.10171101987361908, "step": 3886 }, { "epoch": 1.009086188992731, "grad_norm": 0.7243061661720276, "learning_rate": 2.8275574174477372e-06, "loss": 0.09695728123188019, "step": 3887 }, { "epoch": 1.0093457943925233, "grad_norm": 0.8347339034080505, "learning_rate": 2.8266119632363786e-06, "loss": 0.10302324593067169, "step": 3888 }, { "epoch": 1.0096053997923158, "grad_norm": 0.9327459931373596, "learning_rate": 2.8256664954817805e-06, "loss": 0.11117678880691528, "step": 3889 }, { "epoch": 1.009865005192108, "grad_norm": 0.9924494624137878, "learning_rate": 2.8247210143510774e-06, "loss": 0.10716325789690018, "step": 3890 }, { "epoch": 1.0101246105919004, "grad_norm": 1.2301084995269775, "learning_rate": 2.8237755200114105e-06, "loss": 0.11177676916122437, "step": 3891 }, { "epoch": 1.0103842159916927, "grad_norm": 1.1846929788589478, "learning_rate": 2.82283001262992e-06, "loss": 0.09734989702701569, "step": 3892 }, { "epoch": 1.010643821391485, "grad_norm": 1.312569260597229, "learning_rate": 2.8218844923737517e-06, "loss": 0.11038421839475632, "step": 3893 }, { "epoch": 1.0109034267912773, "grad_norm": 2.14351224899292, "learning_rate": 2.8209389594100507e-06, "loss": 0.10822346061468124, "step": 3894 }, { "epoch": 1.0111630321910696, "grad_norm": 0.8807772994041443, "learning_rate": 2.8199934139059664e-06, "loss": 0.09112060815095901, "step": 3895 }, { "epoch": 1.011422637590862, "grad_norm": 1.654892086982727, "learning_rate": 2.8190478560286488e-06, "loss": 0.10853829979896545, "step": 3896 }, { "epoch": 1.0116822429906542, "grad_norm": 1.2953203916549683, "learning_rate": 2.818102285945251e-06, "loss": 0.13166169822216034, "step": 3897 }, { "epoch": 1.0119418483904465, "grad_norm": 0.9773269891738892, "learning_rate": 2.8171567038229286e-06, "loss": 0.10041586309671402, "step": 3898 }, { "epoch": 1.0122014537902388, "grad_norm": 2.464078903198242, "learning_rate": 2.8162111098288393e-06, "loss": 0.10841977596282959, "step": 3899 }, { "epoch": 1.0124610591900312, "grad_norm": 1.5682332515716553, "learning_rate": 2.8152655041301413e-06, "loss": 0.14779983460903168, "step": 3900 }, { "epoch": 1.0127206645898235, "grad_norm": 2.2886619567871094, "learning_rate": 2.8143198868939957e-06, "loss": 0.12152053415775299, "step": 3901 }, { "epoch": 1.0129802699896158, "grad_norm": 1.438963532447815, "learning_rate": 2.8133742582875677e-06, "loss": 0.09711125493049622, "step": 3902 }, { "epoch": 1.013239875389408, "grad_norm": 1.3789938688278198, "learning_rate": 2.8124286184780224e-06, "loss": 0.12545500695705414, "step": 3903 }, { "epoch": 1.0134994807892004, "grad_norm": 1.8135594129562378, "learning_rate": 2.8114829676325263e-06, "loss": 0.11866484582424164, "step": 3904 }, { "epoch": 1.0137590861889927, "grad_norm": 1.300871729850769, "learning_rate": 2.810537305918249e-06, "loss": 0.10130368173122406, "step": 3905 }, { "epoch": 1.014018691588785, "grad_norm": 1.5417916774749756, "learning_rate": 2.809591633502362e-06, "loss": 0.08368389308452606, "step": 3906 }, { "epoch": 1.0142782969885773, "grad_norm": 1.045426607131958, "learning_rate": 2.808645950552039e-06, "loss": 0.10910481214523315, "step": 3907 }, { "epoch": 1.0145379023883696, "grad_norm": 2.252429246902466, "learning_rate": 2.807700257234455e-06, "loss": 0.09629040211439133, "step": 3908 }, { "epoch": 1.014797507788162, "grad_norm": 1.2169761657714844, "learning_rate": 2.8067545537167874e-06, "loss": 0.09463657438755035, "step": 3909 }, { "epoch": 1.0150571131879542, "grad_norm": 1.400561809539795, "learning_rate": 2.805808840166212e-06, "loss": 0.12865933775901794, "step": 3910 }, { "epoch": 1.0153167185877465, "grad_norm": 1.6752711534500122, "learning_rate": 2.804863116749913e-06, "loss": 0.1142723560333252, "step": 3911 }, { "epoch": 1.0155763239875388, "grad_norm": 1.2134809494018555, "learning_rate": 2.8039173836350714e-06, "loss": 0.11512333899736404, "step": 3912 }, { "epoch": 1.0158359293873314, "grad_norm": 1.20999014377594, "learning_rate": 2.80297164098887e-06, "loss": 0.10620978474617004, "step": 3913 }, { "epoch": 1.0160955347871237, "grad_norm": 1.6887199878692627, "learning_rate": 2.802025888978496e-06, "loss": 0.13830289244651794, "step": 3914 }, { "epoch": 1.016355140186916, "grad_norm": 1.7436280250549316, "learning_rate": 2.801080127771136e-06, "loss": 0.08836863934993744, "step": 3915 }, { "epoch": 1.0166147455867083, "grad_norm": 1.3845853805541992, "learning_rate": 2.800134357533979e-06, "loss": 0.09865155071020126, "step": 3916 }, { "epoch": 1.0168743509865006, "grad_norm": 1.7707911729812622, "learning_rate": 2.7991885784342154e-06, "loss": 0.11437423527240753, "step": 3917 }, { "epoch": 1.017133956386293, "grad_norm": 1.432045340538025, "learning_rate": 2.7982427906390374e-06, "loss": 0.09974020719528198, "step": 3918 }, { "epoch": 1.0173935617860852, "grad_norm": 0.80413419008255, "learning_rate": 2.797296994315638e-06, "loss": 0.09959829598665237, "step": 3919 }, { "epoch": 1.0176531671858775, "grad_norm": 1.3016846179962158, "learning_rate": 2.796351189631214e-06, "loss": 0.11633264273405075, "step": 3920 }, { "epoch": 1.0179127725856698, "grad_norm": 1.201939344406128, "learning_rate": 2.795405376752961e-06, "loss": 0.13468678295612335, "step": 3921 }, { "epoch": 1.0181723779854621, "grad_norm": 0.6788123250007629, "learning_rate": 2.7944595558480765e-06, "loss": 0.09802497923374176, "step": 3922 }, { "epoch": 1.0184319833852544, "grad_norm": 1.2048505544662476, "learning_rate": 2.7935137270837605e-06, "loss": 0.1138576865196228, "step": 3923 }, { "epoch": 1.0186915887850467, "grad_norm": 0.8017233610153198, "learning_rate": 2.7925678906272143e-06, "loss": 0.10165975987911224, "step": 3924 }, { "epoch": 1.018951194184839, "grad_norm": 1.1235116720199585, "learning_rate": 2.7916220466456384e-06, "loss": 0.10487545281648636, "step": 3925 }, { "epoch": 1.0192107995846313, "grad_norm": 1.1464910507202148, "learning_rate": 2.7906761953062396e-06, "loss": 0.125993549823761, "step": 3926 }, { "epoch": 1.0194704049844237, "grad_norm": 1.2457261085510254, "learning_rate": 2.78973033677622e-06, "loss": 0.11488769948482513, "step": 3927 }, { "epoch": 1.019730010384216, "grad_norm": 2.420984983444214, "learning_rate": 2.788784471222786e-06, "loss": 0.10158448666334152, "step": 3928 }, { "epoch": 1.0199896157840083, "grad_norm": 0.7543880343437195, "learning_rate": 2.7878385988131455e-06, "loss": 0.1064959466457367, "step": 3929 }, { "epoch": 1.0202492211838006, "grad_norm": 0.5866715312004089, "learning_rate": 2.7868927197145086e-06, "loss": 0.10854463279247284, "step": 3930 }, { "epoch": 1.0205088265835929, "grad_norm": 1.803986668586731, "learning_rate": 2.785946834094082e-06, "loss": 0.10082395374774933, "step": 3931 }, { "epoch": 1.0207684319833852, "grad_norm": 0.7238885760307312, "learning_rate": 2.7850009421190787e-06, "loss": 0.11900736391544342, "step": 3932 }, { "epoch": 1.0210280373831775, "grad_norm": 1.7042080163955688, "learning_rate": 2.7840550439567104e-06, "loss": 0.10610990226268768, "step": 3933 }, { "epoch": 1.0212876427829698, "grad_norm": 1.443537950515747, "learning_rate": 2.7831091397741905e-06, "loss": 0.10829941928386688, "step": 3934 }, { "epoch": 1.021547248182762, "grad_norm": 1.593511700630188, "learning_rate": 2.782163229738732e-06, "loss": 0.0985422283411026, "step": 3935 }, { "epoch": 1.0218068535825544, "grad_norm": 0.5744121074676514, "learning_rate": 2.7812173140175506e-06, "loss": 0.11202459037303925, "step": 3936 }, { "epoch": 1.022066458982347, "grad_norm": 1.272524118423462, "learning_rate": 2.780271392777864e-06, "loss": 0.07372431457042694, "step": 3937 }, { "epoch": 1.0223260643821392, "grad_norm": 0.9576959013938904, "learning_rate": 2.7793254661868877e-06, "loss": 0.09829588234424591, "step": 3938 }, { "epoch": 1.0225856697819315, "grad_norm": 1.8664491176605225, "learning_rate": 2.778379534411841e-06, "loss": 0.12249462306499481, "step": 3939 }, { "epoch": 1.0228452751817239, "grad_norm": 0.6478845477104187, "learning_rate": 2.7774335976199422e-06, "loss": 0.11073589324951172, "step": 3940 }, { "epoch": 1.0231048805815162, "grad_norm": 1.618085503578186, "learning_rate": 2.7764876559784105e-06, "loss": 0.12067057192325592, "step": 3941 }, { "epoch": 1.0233644859813085, "grad_norm": 1.2684248685836792, "learning_rate": 2.775541709654469e-06, "loss": 0.12741868197917938, "step": 3942 }, { "epoch": 1.0236240913811008, "grad_norm": 1.78830087184906, "learning_rate": 2.7745957588153376e-06, "loss": 0.1389906108379364, "step": 3943 }, { "epoch": 1.023883696780893, "grad_norm": 1.9347078800201416, "learning_rate": 2.7736498036282388e-06, "loss": 0.1140076294541359, "step": 3944 }, { "epoch": 1.0241433021806854, "grad_norm": 0.9103916883468628, "learning_rate": 2.7727038442603967e-06, "loss": 0.08749522268772125, "step": 3945 }, { "epoch": 1.0244029075804777, "grad_norm": 0.8948709964752197, "learning_rate": 2.7717578808790342e-06, "loss": 0.11330470442771912, "step": 3946 }, { "epoch": 1.02466251298027, "grad_norm": 0.49263542890548706, "learning_rate": 2.7708119136513778e-06, "loss": 0.11274273693561554, "step": 3947 }, { "epoch": 1.0249221183800623, "grad_norm": 3.3229382038116455, "learning_rate": 2.7698659427446507e-06, "loss": 0.06656000763177872, "step": 3948 }, { "epoch": 1.0251817237798546, "grad_norm": 0.6914330720901489, "learning_rate": 2.768919968326079e-06, "loss": 0.10782860219478607, "step": 3949 }, { "epoch": 1.025441329179647, "grad_norm": 0.9257197380065918, "learning_rate": 2.767973990562891e-06, "loss": 0.0923338383436203, "step": 3950 }, { "epoch": 1.0257009345794392, "grad_norm": 1.2997838258743286, "learning_rate": 2.7670280096223136e-06, "loss": 0.08008883148431778, "step": 3951 }, { "epoch": 1.0259605399792315, "grad_norm": 1.9200317859649658, "learning_rate": 2.766082025671573e-06, "loss": 0.13676077127456665, "step": 3952 }, { "epoch": 1.0262201453790238, "grad_norm": 1.3760697841644287, "learning_rate": 2.7651360388778993e-06, "loss": 0.0958300530910492, "step": 3953 }, { "epoch": 1.0264797507788161, "grad_norm": 0.7651066184043884, "learning_rate": 2.7641900494085193e-06, "loss": 0.11647558957338333, "step": 3954 }, { "epoch": 1.0267393561786085, "grad_norm": 2.0738914012908936, "learning_rate": 2.763244057430665e-06, "loss": 0.08577944338321686, "step": 3955 }, { "epoch": 1.0269989615784008, "grad_norm": 1.8276768922805786, "learning_rate": 2.762298063111563e-06, "loss": 0.09185542166233063, "step": 3956 }, { "epoch": 1.027258566978193, "grad_norm": 3.3418450355529785, "learning_rate": 2.7613520666184466e-06, "loss": 0.11782185733318329, "step": 3957 }, { "epoch": 1.0275181723779854, "grad_norm": 1.6220811605453491, "learning_rate": 2.7604060681185442e-06, "loss": 0.13064482808113098, "step": 3958 }, { "epoch": 1.0277777777777777, "grad_norm": 1.3411171436309814, "learning_rate": 2.7594600677790866e-06, "loss": 0.11607711017131805, "step": 3959 }, { "epoch": 1.02803738317757, "grad_norm": 2.7491800785064697, "learning_rate": 2.758514065767306e-06, "loss": 0.10097096860408783, "step": 3960 }, { "epoch": 1.0282969885773625, "grad_norm": 1.1676503419876099, "learning_rate": 2.7575680622504342e-06, "loss": 0.10702136904001236, "step": 3961 }, { "epoch": 1.0285565939771548, "grad_norm": 1.3862093687057495, "learning_rate": 2.7566220573957015e-06, "loss": 0.10907164961099625, "step": 3962 }, { "epoch": 1.0288161993769471, "grad_norm": 1.571512222290039, "learning_rate": 2.755676051370341e-06, "loss": 0.12793296575546265, "step": 3963 }, { "epoch": 1.0290758047767394, "grad_norm": 0.6057969331741333, "learning_rate": 2.7547300443415836e-06, "loss": 0.10359916090965271, "step": 3964 }, { "epoch": 1.0293354101765317, "grad_norm": 0.5633028745651245, "learning_rate": 2.753784036476664e-06, "loss": 0.11862481385469437, "step": 3965 }, { "epoch": 1.029595015576324, "grad_norm": 1.4315931797027588, "learning_rate": 2.752838027942813e-06, "loss": 0.10846345126628876, "step": 3966 }, { "epoch": 1.0298546209761164, "grad_norm": 0.45681238174438477, "learning_rate": 2.751892018907263e-06, "loss": 0.11183209717273712, "step": 3967 }, { "epoch": 1.0301142263759087, "grad_norm": 0.5445169806480408, "learning_rate": 2.7509460095372477e-06, "loss": 0.12320645153522491, "step": 3968 }, { "epoch": 1.030373831775701, "grad_norm": 0.49648696184158325, "learning_rate": 2.7500000000000004e-06, "loss": 0.11495175957679749, "step": 3969 }, { "epoch": 1.0306334371754933, "grad_norm": 0.5946785807609558, "learning_rate": 2.749053990462752e-06, "loss": 0.10312432050704956, "step": 3970 }, { "epoch": 1.0308930425752856, "grad_norm": 0.8147770166397095, "learning_rate": 2.7481079810927374e-06, "loss": 0.11367550492286682, "step": 3971 }, { "epoch": 1.0311526479750779, "grad_norm": 0.5597261786460876, "learning_rate": 2.747161972057188e-06, "loss": 0.11017344146966934, "step": 3972 }, { "epoch": 1.0314122533748702, "grad_norm": 1.470530390739441, "learning_rate": 2.7462159635233372e-06, "loss": 0.07156550884246826, "step": 3973 }, { "epoch": 1.0316718587746625, "grad_norm": 1.4048322439193726, "learning_rate": 2.7452699556584167e-06, "loss": 0.1304040551185608, "step": 3974 }, { "epoch": 1.0319314641744548, "grad_norm": 1.121986985206604, "learning_rate": 2.7443239486296603e-06, "loss": 0.09555856883525848, "step": 3975 }, { "epoch": 1.032191069574247, "grad_norm": 2.490898609161377, "learning_rate": 2.7433779426042997e-06, "loss": 0.13808616995811462, "step": 3976 }, { "epoch": 1.0324506749740394, "grad_norm": 0.49772173166275024, "learning_rate": 2.742431937749567e-06, "loss": 0.10223080217838287, "step": 3977 }, { "epoch": 1.0327102803738317, "grad_norm": 1.2344396114349365, "learning_rate": 2.7414859342326943e-06, "loss": 0.11592510342597961, "step": 3978 }, { "epoch": 1.032969885773624, "grad_norm": 0.43611541390419006, "learning_rate": 2.740539932220915e-06, "loss": 0.1134045198559761, "step": 3979 }, { "epoch": 1.0332294911734163, "grad_norm": 1.2001941204071045, "learning_rate": 2.739593931881457e-06, "loss": 0.10182404518127441, "step": 3980 }, { "epoch": 1.0334890965732086, "grad_norm": 2.059147357940674, "learning_rate": 2.738647933381554e-06, "loss": 0.1152825653553009, "step": 3981 }, { "epoch": 1.033748701973001, "grad_norm": 1.2890201807022095, "learning_rate": 2.7377019368884367e-06, "loss": 0.10446758568286896, "step": 3982 }, { "epoch": 1.0340083073727933, "grad_norm": 0.647114098072052, "learning_rate": 2.7367559425693364e-06, "loss": 0.11758232116699219, "step": 3983 }, { "epoch": 1.0342679127725856, "grad_norm": 0.6945464015007019, "learning_rate": 2.7358099505914814e-06, "loss": 0.11733432859182358, "step": 3984 }, { "epoch": 1.034527518172378, "grad_norm": 1.3561468124389648, "learning_rate": 2.7348639611221014e-06, "loss": 0.0815856084227562, "step": 3985 }, { "epoch": 1.0347871235721704, "grad_norm": 1.9846501350402832, "learning_rate": 2.7339179743284273e-06, "loss": 0.11662788689136505, "step": 3986 }, { "epoch": 1.0350467289719627, "grad_norm": 2.0853874683380127, "learning_rate": 2.732971990377688e-06, "loss": 0.1080675721168518, "step": 3987 }, { "epoch": 1.035306334371755, "grad_norm": 1.2566685676574707, "learning_rate": 2.7320260094371097e-06, "loss": 0.11188755184412003, "step": 3988 }, { "epoch": 1.0355659397715473, "grad_norm": 1.2262588739395142, "learning_rate": 2.7310800316739215e-06, "loss": 0.10263531655073166, "step": 3989 }, { "epoch": 1.0358255451713396, "grad_norm": 1.5373601913452148, "learning_rate": 2.7301340572553505e-06, "loss": 0.11254139244556427, "step": 3990 }, { "epoch": 1.036085150571132, "grad_norm": 1.7191251516342163, "learning_rate": 2.729188086348624e-06, "loss": 0.11959468573331833, "step": 3991 }, { "epoch": 1.0363447559709242, "grad_norm": 1.7833349704742432, "learning_rate": 2.728242119120967e-06, "loss": 0.11597475409507751, "step": 3992 }, { "epoch": 1.0366043613707165, "grad_norm": 0.6970372796058655, "learning_rate": 2.7272961557396044e-06, "loss": 0.10294977575540543, "step": 3993 }, { "epoch": 1.0368639667705088, "grad_norm": 2.084064245223999, "learning_rate": 2.726350196371762e-06, "loss": 0.12963782250881195, "step": 3994 }, { "epoch": 1.0371235721703012, "grad_norm": 1.6065268516540527, "learning_rate": 2.7254042411846636e-06, "loss": 0.07687275856733322, "step": 3995 }, { "epoch": 1.0373831775700935, "grad_norm": 0.9766520261764526, "learning_rate": 2.724458290345532e-06, "loss": 0.10234059393405914, "step": 3996 }, { "epoch": 1.0376427829698858, "grad_norm": 1.296504259109497, "learning_rate": 2.7235123440215898e-06, "loss": 0.10087040066719055, "step": 3997 }, { "epoch": 1.037902388369678, "grad_norm": 1.6345179080963135, "learning_rate": 2.7225664023800597e-06, "loss": 0.11045363545417786, "step": 3998 }, { "epoch": 1.0381619937694704, "grad_norm": 2.0887718200683594, "learning_rate": 2.72162046558816e-06, "loss": 0.11045393347740173, "step": 3999 }, { "epoch": 1.0384215991692627, "grad_norm": 0.8221408128738403, "learning_rate": 2.720674533813113e-06, "loss": 0.11408201605081558, "step": 4000 }, { "epoch": 1.038681204569055, "grad_norm": 1.4526902437210083, "learning_rate": 2.719728607222136e-06, "loss": 0.1099502444267273, "step": 4001 }, { "epoch": 1.0389408099688473, "grad_norm": 0.7769321799278259, "learning_rate": 2.7187826859824497e-06, "loss": 0.10381652414798737, "step": 4002 }, { "epoch": 1.0392004153686396, "grad_norm": 1.6868833303451538, "learning_rate": 2.717836770261269e-06, "loss": 0.06977176666259766, "step": 4003 }, { "epoch": 1.039460020768432, "grad_norm": 1.205026388168335, "learning_rate": 2.7168908602258106e-06, "loss": 0.09031350165605545, "step": 4004 }, { "epoch": 1.0397196261682242, "grad_norm": 1.3016451597213745, "learning_rate": 2.7159449560432903e-06, "loss": 0.12063876539468765, "step": 4005 }, { "epoch": 1.0399792315680165, "grad_norm": 1.6297136545181274, "learning_rate": 2.714999057880922e-06, "loss": 0.09504091739654541, "step": 4006 }, { "epoch": 1.0402388369678088, "grad_norm": 0.7069912552833557, "learning_rate": 2.714053165905919e-06, "loss": 0.10341773927211761, "step": 4007 }, { "epoch": 1.0404984423676011, "grad_norm": 0.6498678922653198, "learning_rate": 2.7131072802854925e-06, "loss": 0.11703541874885559, "step": 4008 }, { "epoch": 1.0407580477673937, "grad_norm": 1.2803586721420288, "learning_rate": 2.7121614011868547e-06, "loss": 0.07824066281318665, "step": 4009 }, { "epoch": 1.041017653167186, "grad_norm": 1.1618196964263916, "learning_rate": 2.711215528777216e-06, "loss": 0.0866299495100975, "step": 4010 }, { "epoch": 1.0412772585669783, "grad_norm": 0.7585923671722412, "learning_rate": 2.7102696632237813e-06, "loss": 0.10300932824611664, "step": 4011 }, { "epoch": 1.0415368639667706, "grad_norm": 1.574832558631897, "learning_rate": 2.7093238046937615e-06, "loss": 0.09619969129562378, "step": 4012 }, { "epoch": 1.041796469366563, "grad_norm": 1.678396224975586, "learning_rate": 2.708377953354362e-06, "loss": 0.1031753420829773, "step": 4013 }, { "epoch": 1.0420560747663552, "grad_norm": 2.9860470294952393, "learning_rate": 2.7074321093727877e-06, "loss": 0.08488305658102036, "step": 4014 }, { "epoch": 1.0423156801661475, "grad_norm": 1.0343537330627441, "learning_rate": 2.706486272916241e-06, "loss": 0.11864928901195526, "step": 4015 }, { "epoch": 1.0425752855659398, "grad_norm": 1.2347207069396973, "learning_rate": 2.7055404441519246e-06, "loss": 0.11480699479579926, "step": 4016 }, { "epoch": 1.0428348909657321, "grad_norm": 2.9148638248443604, "learning_rate": 2.70459462324704e-06, "loss": 0.1510307788848877, "step": 4017 }, { "epoch": 1.0430944963655244, "grad_norm": 1.846718668937683, "learning_rate": 2.7036488103687868e-06, "loss": 0.11496168375015259, "step": 4018 }, { "epoch": 1.0433541017653167, "grad_norm": 0.8916647434234619, "learning_rate": 2.702703005684363e-06, "loss": 0.10214190185070038, "step": 4019 }, { "epoch": 1.043613707165109, "grad_norm": 1.4185750484466553, "learning_rate": 2.7017572093609633e-06, "loss": 0.11284556984901428, "step": 4020 }, { "epoch": 1.0438733125649013, "grad_norm": 1.119873285293579, "learning_rate": 2.700811421565785e-06, "loss": 0.11679325997829437, "step": 4021 }, { "epoch": 1.0441329179646937, "grad_norm": 1.3778148889541626, "learning_rate": 2.6998656424660226e-06, "loss": 0.10837540030479431, "step": 4022 }, { "epoch": 1.044392523364486, "grad_norm": 1.8874882459640503, "learning_rate": 2.698919872228865e-06, "loss": 0.11263018846511841, "step": 4023 }, { "epoch": 1.0446521287642783, "grad_norm": 1.119555950164795, "learning_rate": 2.6979741110215045e-06, "loss": 0.11908036470413208, "step": 4024 }, { "epoch": 1.0449117341640706, "grad_norm": 1.6246205568313599, "learning_rate": 2.6970283590111307e-06, "loss": 0.07856397330760956, "step": 4025 }, { "epoch": 1.0451713395638629, "grad_norm": 0.4791609048843384, "learning_rate": 2.6960826163649297e-06, "loss": 0.10844950377941132, "step": 4026 }, { "epoch": 1.0454309449636552, "grad_norm": 1.3635693788528442, "learning_rate": 2.695136883250088e-06, "loss": 0.11513840407133102, "step": 4027 }, { "epoch": 1.0456905503634475, "grad_norm": 1.559202790260315, "learning_rate": 2.6941911598337885e-06, "loss": 0.10338383913040161, "step": 4028 }, { "epoch": 1.0459501557632398, "grad_norm": 0.5029295086860657, "learning_rate": 2.6932454462832146e-06, "loss": 0.11188399791717529, "step": 4029 }, { "epoch": 1.046209761163032, "grad_norm": 1.323007345199585, "learning_rate": 2.6922997427655456e-06, "loss": 0.12185616046190262, "step": 4030 }, { "epoch": 1.0464693665628244, "grad_norm": 1.0110812187194824, "learning_rate": 2.6913540494479613e-06, "loss": 0.10902141034603119, "step": 4031 }, { "epoch": 1.0467289719626167, "grad_norm": 1.5434401035308838, "learning_rate": 2.6904083664976384e-06, "loss": 0.10781645029783249, "step": 4032 }, { "epoch": 1.0469885773624092, "grad_norm": 1.6450135707855225, "learning_rate": 2.689462694081752e-06, "loss": 0.11304759979248047, "step": 4033 }, { "epoch": 1.0472481827622016, "grad_norm": 0.9431352615356445, "learning_rate": 2.6885170323674748e-06, "loss": 0.11772231757640839, "step": 4034 }, { "epoch": 1.0475077881619939, "grad_norm": 1.3608789443969727, "learning_rate": 2.6875713815219783e-06, "loss": 0.11091412603855133, "step": 4035 }, { "epoch": 1.0477673935617862, "grad_norm": 0.6571729183197021, "learning_rate": 2.6866257417124326e-06, "loss": 0.11264145374298096, "step": 4036 }, { "epoch": 1.0480269989615785, "grad_norm": 0.7266103625297546, "learning_rate": 2.6856801131060046e-06, "loss": 0.10563478618860245, "step": 4037 }, { "epoch": 1.0482866043613708, "grad_norm": 1.8096513748168945, "learning_rate": 2.6847344958698594e-06, "loss": 0.09897451102733612, "step": 4038 }, { "epoch": 1.048546209761163, "grad_norm": 1.645833969116211, "learning_rate": 2.683788890171162e-06, "loss": 0.07647791504859924, "step": 4039 }, { "epoch": 1.0488058151609554, "grad_norm": 0.7773621678352356, "learning_rate": 2.6828432961770717e-06, "loss": 0.10936444997787476, "step": 4040 }, { "epoch": 1.0490654205607477, "grad_norm": 1.5274051427841187, "learning_rate": 2.68189771405475e-06, "loss": 0.09264339506626129, "step": 4041 }, { "epoch": 1.04932502596054, "grad_norm": 1.0537323951721191, "learning_rate": 2.680952143971352e-06, "loss": 0.1099226325750351, "step": 4042 }, { "epoch": 1.0495846313603323, "grad_norm": 1.0260236263275146, "learning_rate": 2.6800065860940343e-06, "loss": 0.10433943569660187, "step": 4043 }, { "epoch": 1.0498442367601246, "grad_norm": 1.7362252473831177, "learning_rate": 2.6790610405899496e-06, "loss": 0.0925493836402893, "step": 4044 }, { "epoch": 1.050103842159917, "grad_norm": 1.090184211730957, "learning_rate": 2.6781155076262494e-06, "loss": 0.09919504821300507, "step": 4045 }, { "epoch": 1.0503634475597092, "grad_norm": 1.1034215688705444, "learning_rate": 2.6771699873700806e-06, "loss": 0.11258963495492935, "step": 4046 }, { "epoch": 1.0506230529595015, "grad_norm": 1.1120389699935913, "learning_rate": 2.6762244799885906e-06, "loss": 0.10651334375143051, "step": 4047 }, { "epoch": 1.0508826583592938, "grad_norm": 1.0426025390625, "learning_rate": 2.675278985648923e-06, "loss": 0.10524391382932663, "step": 4048 }, { "epoch": 1.0511422637590861, "grad_norm": 1.9235159158706665, "learning_rate": 2.674333504518221e-06, "loss": 0.11514478176832199, "step": 4049 }, { "epoch": 1.0514018691588785, "grad_norm": 1.2406525611877441, "learning_rate": 2.673388036763622e-06, "loss": 0.10357450693845749, "step": 4050 }, { "epoch": 1.0516614745586708, "grad_norm": 1.1917682886123657, "learning_rate": 2.672442582552264e-06, "loss": 0.09465494751930237, "step": 4051 }, { "epoch": 1.051921079958463, "grad_norm": 1.421035647392273, "learning_rate": 2.671497142051282e-06, "loss": 0.09462177753448486, "step": 4052 }, { "epoch": 1.0521806853582554, "grad_norm": 1.1902976036071777, "learning_rate": 2.6705517154278073e-06, "loss": 0.0961611345410347, "step": 4053 }, { "epoch": 1.0524402907580477, "grad_norm": 2.644099235534668, "learning_rate": 2.66960630284897e-06, "loss": 0.1053415909409523, "step": 4054 }, { "epoch": 1.05269989615784, "grad_norm": 0.9853807687759399, "learning_rate": 2.668660904481898e-06, "loss": 0.09611696749925613, "step": 4055 }, { "epoch": 1.0529595015576323, "grad_norm": 1.0563054084777832, "learning_rate": 2.667715520493716e-06, "loss": 0.1054195687174797, "step": 4056 }, { "epoch": 1.0532191069574246, "grad_norm": 1.211535096168518, "learning_rate": 2.666770151051545e-06, "loss": 0.11944232881069183, "step": 4057 }, { "epoch": 1.0534787123572171, "grad_norm": 4.172733783721924, "learning_rate": 2.665824796322506e-06, "loss": 0.10448554158210754, "step": 4058 }, { "epoch": 1.0537383177570094, "grad_norm": 2.4177262783050537, "learning_rate": 2.6648794564737156e-06, "loss": 0.09320978820323944, "step": 4059 }, { "epoch": 1.0539979231568017, "grad_norm": 1.2431223392486572, "learning_rate": 2.663934131672289e-06, "loss": 0.11215037107467651, "step": 4060 }, { "epoch": 1.054257528556594, "grad_norm": 0.9310197830200195, "learning_rate": 2.6629888220853365e-06, "loss": 0.09896062314510345, "step": 4061 }, { "epoch": 1.0545171339563864, "grad_norm": 1.2711361646652222, "learning_rate": 2.662043527879968e-06, "loss": 0.09448230266571045, "step": 4062 }, { "epoch": 1.0547767393561787, "grad_norm": 1.6422704458236694, "learning_rate": 2.66109824922329e-06, "loss": 0.10541392862796783, "step": 4063 }, { "epoch": 1.055036344755971, "grad_norm": 0.9482632875442505, "learning_rate": 2.660152986282406e-06, "loss": 0.1142820492386818, "step": 4064 }, { "epoch": 1.0552959501557633, "grad_norm": 0.983386218547821, "learning_rate": 2.659207739224417e-06, "loss": 0.09710352122783661, "step": 4065 }, { "epoch": 1.0555555555555556, "grad_norm": 0.9347812533378601, "learning_rate": 2.65826250821642e-06, "loss": 0.11589615046977997, "step": 4066 }, { "epoch": 1.0558151609553479, "grad_norm": 2.1480941772460938, "learning_rate": 2.6573172934255116e-06, "loss": 0.1278148591518402, "step": 4067 }, { "epoch": 1.0560747663551402, "grad_norm": 1.073937177658081, "learning_rate": 2.656372095018784e-06, "loss": 0.12470511347055435, "step": 4068 }, { "epoch": 1.0563343717549325, "grad_norm": 1.525347352027893, "learning_rate": 2.6554269131633254e-06, "loss": 0.11511985212564468, "step": 4069 }, { "epoch": 1.0565939771547248, "grad_norm": 1.651882529258728, "learning_rate": 2.6544817480262237e-06, "loss": 0.1361863911151886, "step": 4070 }, { "epoch": 1.0568535825545171, "grad_norm": 0.8900969624519348, "learning_rate": 2.6535365997745616e-06, "loss": 0.10505513101816177, "step": 4071 }, { "epoch": 1.0571131879543094, "grad_norm": 0.8376300930976868, "learning_rate": 2.65259146857542e-06, "loss": 0.11128558218479156, "step": 4072 }, { "epoch": 1.0573727933541017, "grad_norm": 0.661980926990509, "learning_rate": 2.6516463545958766e-06, "loss": 0.09442558139562607, "step": 4073 }, { "epoch": 1.057632398753894, "grad_norm": 2.3984248638153076, "learning_rate": 2.650701258003005e-06, "loss": 0.10591982305049896, "step": 4074 }, { "epoch": 1.0578920041536863, "grad_norm": 0.5398861765861511, "learning_rate": 2.6497561789638783e-06, "loss": 0.10607457160949707, "step": 4075 }, { "epoch": 1.0581516095534786, "grad_norm": 0.599699079990387, "learning_rate": 2.648811117645564e-06, "loss": 0.12103217840194702, "step": 4076 }, { "epoch": 1.058411214953271, "grad_norm": 1.333528995513916, "learning_rate": 2.6478660742151257e-06, "loss": 0.10257856547832489, "step": 4077 }, { "epoch": 1.0586708203530633, "grad_norm": 1.4500318765640259, "learning_rate": 2.646921048839628e-06, "loss": 0.09305411577224731, "step": 4078 }, { "epoch": 1.0589304257528556, "grad_norm": 0.8250337839126587, "learning_rate": 2.6459760416861293e-06, "loss": 0.09387306869029999, "step": 4079 }, { "epoch": 1.0591900311526479, "grad_norm": 1.9660451412200928, "learning_rate": 2.6450310529216837e-06, "loss": 0.121984101831913, "step": 4080 }, { "epoch": 1.0594496365524404, "grad_norm": 0.6915801167488098, "learning_rate": 2.644086082713345e-06, "loss": 0.10682176053524017, "step": 4081 }, { "epoch": 1.0597092419522327, "grad_norm": 0.9067551493644714, "learning_rate": 2.64314113122816e-06, "loss": 0.09164093434810638, "step": 4082 }, { "epoch": 1.059968847352025, "grad_norm": 0.7572997808456421, "learning_rate": 2.6421961986331786e-06, "loss": 0.11429619044065475, "step": 4083 }, { "epoch": 1.0602284527518173, "grad_norm": 1.9900853633880615, "learning_rate": 2.6412512850954397e-06, "loss": 0.10207946598529816, "step": 4084 }, { "epoch": 1.0604880581516096, "grad_norm": 4.356836795806885, "learning_rate": 2.640306390781984e-06, "loss": 0.1345961093902588, "step": 4085 }, { "epoch": 1.060747663551402, "grad_norm": 1.2171200513839722, "learning_rate": 2.6393615158598464e-06, "loss": 0.0840890109539032, "step": 4086 }, { "epoch": 1.0610072689511942, "grad_norm": 1.976562261581421, "learning_rate": 2.6384166604960596e-06, "loss": 0.09226764738559723, "step": 4087 }, { "epoch": 1.0612668743509865, "grad_norm": 0.7766495943069458, "learning_rate": 2.6374718248576526e-06, "loss": 0.10927987098693848, "step": 4088 }, { "epoch": 1.0615264797507789, "grad_norm": 2.3239388465881348, "learning_rate": 2.6365270091116497e-06, "loss": 0.08612114191055298, "step": 4089 }, { "epoch": 1.0617860851505712, "grad_norm": 0.6052812337875366, "learning_rate": 2.635582213425074e-06, "loss": 0.1143002137541771, "step": 4090 }, { "epoch": 1.0620456905503635, "grad_norm": 0.8596175909042358, "learning_rate": 2.6346374379649432e-06, "loss": 0.10170318931341171, "step": 4091 }, { "epoch": 1.0623052959501558, "grad_norm": 0.6858527660369873, "learning_rate": 2.633692682898272e-06, "loss": 0.1129056066274643, "step": 4092 }, { "epoch": 1.062564901349948, "grad_norm": 2.087401866912842, "learning_rate": 2.632747948392071e-06, "loss": 0.10483085364103317, "step": 4093 }, { "epoch": 1.0628245067497404, "grad_norm": 2.1934115886688232, "learning_rate": 2.6318032346133476e-06, "loss": 0.08708155155181885, "step": 4094 }, { "epoch": 1.0630841121495327, "grad_norm": 0.4825480878353119, "learning_rate": 2.6308585417291073e-06, "loss": 0.11397454142570496, "step": 4095 }, { "epoch": 1.063343717549325, "grad_norm": 1.985511302947998, "learning_rate": 2.6299138699063475e-06, "loss": 0.08309510350227356, "step": 4096 }, { "epoch": 1.0636033229491173, "grad_norm": 0.9126203060150146, "learning_rate": 2.6289692193120664e-06, "loss": 0.12050671875476837, "step": 4097 }, { "epoch": 1.0638629283489096, "grad_norm": 0.7882723212242126, "learning_rate": 2.628024590113256e-06, "loss": 0.10264816880226135, "step": 4098 }, { "epoch": 1.064122533748702, "grad_norm": 2.224095582962036, "learning_rate": 2.6270799824769056e-06, "loss": 0.07717137783765793, "step": 4099 }, { "epoch": 1.0643821391484942, "grad_norm": 1.517728567123413, "learning_rate": 2.626135396569999e-06, "loss": 0.1105525940656662, "step": 4100 }, { "epoch": 1.0646417445482865, "grad_norm": 0.875839114189148, "learning_rate": 2.6251908325595187e-06, "loss": 0.11310988664627075, "step": 4101 }, { "epoch": 1.0649013499480788, "grad_norm": 1.621305227279663, "learning_rate": 2.6242462906124415e-06, "loss": 0.09683017432689667, "step": 4102 }, { "epoch": 1.0651609553478711, "grad_norm": 2.5618817806243896, "learning_rate": 2.623301770895741e-06, "loss": 0.13890521228313446, "step": 4103 }, { "epoch": 1.0654205607476634, "grad_norm": 1.4989774227142334, "learning_rate": 2.6223572735763864e-06, "loss": 0.11821639537811279, "step": 4104 }, { "epoch": 1.0656801661474558, "grad_norm": 2.518878221511841, "learning_rate": 2.621412798821343e-06, "loss": 0.12316161394119263, "step": 4105 }, { "epoch": 1.0659397715472483, "grad_norm": 2.1112942695617676, "learning_rate": 2.6204683467975716e-06, "loss": 0.10926535725593567, "step": 4106 }, { "epoch": 1.0661993769470406, "grad_norm": 1.1689201593399048, "learning_rate": 2.619523917672033e-06, "loss": 0.11077754199504852, "step": 4107 }, { "epoch": 1.066458982346833, "grad_norm": 2.0060503482818604, "learning_rate": 2.6185795116116756e-06, "loss": 0.11366386711597443, "step": 4108 }, { "epoch": 1.0667185877466252, "grad_norm": 1.2477540969848633, "learning_rate": 2.617635128783452e-06, "loss": 0.0817691832780838, "step": 4109 }, { "epoch": 1.0669781931464175, "grad_norm": 1.7522302865982056, "learning_rate": 2.6166907693543085e-06, "loss": 0.06805220246315002, "step": 4110 }, { "epoch": 1.0672377985462098, "grad_norm": 1.9138801097869873, "learning_rate": 2.615746433491183e-06, "loss": 0.12162304669618607, "step": 4111 }, { "epoch": 1.0674974039460021, "grad_norm": 1.1056476831436157, "learning_rate": 2.6148021213610147e-06, "loss": 0.10508900880813599, "step": 4112 }, { "epoch": 1.0677570093457944, "grad_norm": 0.8732638955116272, "learning_rate": 2.613857833130735e-06, "loss": 0.1270134150981903, "step": 4113 }, { "epoch": 1.0680166147455867, "grad_norm": 0.5443387627601624, "learning_rate": 2.6129135689672745e-06, "loss": 0.10750386863946915, "step": 4114 }, { "epoch": 1.068276220145379, "grad_norm": 0.8193955421447754, "learning_rate": 2.6119693290375555e-06, "loss": 0.10299680382013321, "step": 4115 }, { "epoch": 1.0685358255451713, "grad_norm": 0.6633593440055847, "learning_rate": 2.611025113508498e-06, "loss": 0.10665514320135117, "step": 4116 }, { "epoch": 1.0687954309449637, "grad_norm": 0.797003984451294, "learning_rate": 2.610080922547019e-06, "loss": 0.11351852864027023, "step": 4117 }, { "epoch": 1.069055036344756, "grad_norm": 1.4529677629470825, "learning_rate": 2.6091367563200294e-06, "loss": 0.10958683490753174, "step": 4118 }, { "epoch": 1.0693146417445483, "grad_norm": 0.7111524939537048, "learning_rate": 2.608192614994435e-06, "loss": 0.11669252812862396, "step": 4119 }, { "epoch": 1.0695742471443406, "grad_norm": 0.7558190226554871, "learning_rate": 2.6072484987371395e-06, "loss": 0.13324570655822754, "step": 4120 }, { "epoch": 1.0698338525441329, "grad_norm": 1.5151863098144531, "learning_rate": 2.6063044077150414e-06, "loss": 0.09106045961380005, "step": 4121 }, { "epoch": 1.0700934579439252, "grad_norm": 1.3701348304748535, "learning_rate": 2.6053603420950337e-06, "loss": 0.11678314208984375, "step": 4122 }, { "epoch": 1.0703530633437175, "grad_norm": 1.1219350099563599, "learning_rate": 2.6044163020440055e-06, "loss": 0.11609242856502533, "step": 4123 }, { "epoch": 1.0706126687435098, "grad_norm": 2.967900037765503, "learning_rate": 2.6034722877288415e-06, "loss": 0.10249350965023041, "step": 4124 }, { "epoch": 1.070872274143302, "grad_norm": 1.4325942993164062, "learning_rate": 2.6025282993164226e-06, "loss": 0.09866891801357269, "step": 4125 }, { "epoch": 1.0711318795430944, "grad_norm": 0.9265685677528381, "learning_rate": 2.6015843369736234e-06, "loss": 0.10095617175102234, "step": 4126 }, { "epoch": 1.0713914849428867, "grad_norm": 0.820939302444458, "learning_rate": 2.6006404008673153e-06, "loss": 0.10710900276899338, "step": 4127 }, { "epoch": 1.071651090342679, "grad_norm": 2.1390960216522217, "learning_rate": 2.599696491164364e-06, "loss": 0.12213864922523499, "step": 4128 }, { "epoch": 1.0719106957424716, "grad_norm": 1.356123447418213, "learning_rate": 2.5987526080316316e-06, "loss": 0.08624640852212906, "step": 4129 }, { "epoch": 1.0721703011422639, "grad_norm": 1.1710176467895508, "learning_rate": 2.5978087516359756e-06, "loss": 0.11659850180149078, "step": 4130 }, { "epoch": 1.0724299065420562, "grad_norm": 1.7501425743103027, "learning_rate": 2.5968649221442473e-06, "loss": 0.10283678025007248, "step": 4131 }, { "epoch": 1.0726895119418485, "grad_norm": 2.213583469390869, "learning_rate": 2.595921119723294e-06, "loss": 0.09831100702285767, "step": 4132 }, { "epoch": 1.0729491173416408, "grad_norm": 1.6601431369781494, "learning_rate": 2.5949773445399595e-06, "loss": 0.094840407371521, "step": 4133 }, { "epoch": 1.073208722741433, "grad_norm": 0.6969600915908813, "learning_rate": 2.5940335967610807e-06, "loss": 0.10845161974430084, "step": 4134 }, { "epoch": 1.0734683281412254, "grad_norm": 1.5083461999893188, "learning_rate": 2.5930898765534906e-06, "loss": 0.09629803895950317, "step": 4135 }, { "epoch": 1.0737279335410177, "grad_norm": 1.4460169076919556, "learning_rate": 2.592146184084018e-06, "loss": 0.11648160219192505, "step": 4136 }, { "epoch": 1.07398753894081, "grad_norm": 1.1582311391830444, "learning_rate": 2.5912025195194855e-06, "loss": 0.1282476782798767, "step": 4137 }, { "epoch": 1.0742471443406023, "grad_norm": 1.0386130809783936, "learning_rate": 2.5902588830267116e-06, "loss": 0.10814662277698517, "step": 4138 }, { "epoch": 1.0745067497403946, "grad_norm": 1.452853798866272, "learning_rate": 2.5893152747725094e-06, "loss": 0.10938650369644165, "step": 4139 }, { "epoch": 1.074766355140187, "grad_norm": 1.6555020809173584, "learning_rate": 2.588371694923687e-06, "loss": 0.12111406028270721, "step": 4140 }, { "epoch": 1.0750259605399792, "grad_norm": 1.9661180973052979, "learning_rate": 2.5874281436470494e-06, "loss": 0.10874421894550323, "step": 4141 }, { "epoch": 1.0752855659397715, "grad_norm": 0.7345647215843201, "learning_rate": 2.586484621109393e-06, "loss": 0.09980980306863785, "step": 4142 }, { "epoch": 1.0755451713395638, "grad_norm": 1.0747177600860596, "learning_rate": 2.585541127477511e-06, "loss": 0.10717210173606873, "step": 4143 }, { "epoch": 1.0758047767393562, "grad_norm": 0.9728832244873047, "learning_rate": 2.5845976629181917e-06, "loss": 0.12123366445302963, "step": 4144 }, { "epoch": 1.0760643821391485, "grad_norm": 2.4530255794525146, "learning_rate": 2.58365422759822e-06, "loss": 0.098723404109478, "step": 4145 }, { "epoch": 1.0763239875389408, "grad_norm": 0.823282778263092, "learning_rate": 2.582710821684371e-06, "loss": 0.10348431766033173, "step": 4146 }, { "epoch": 1.076583592938733, "grad_norm": 1.5496833324432373, "learning_rate": 2.5817674453434185e-06, "loss": 0.08360257744789124, "step": 4147 }, { "epoch": 1.0768431983385254, "grad_norm": 1.1046494245529175, "learning_rate": 2.5808240987421295e-06, "loss": 0.12575292587280273, "step": 4148 }, { "epoch": 1.0771028037383177, "grad_norm": 1.734848141670227, "learning_rate": 2.579880782047267e-06, "loss": 0.11507459729909897, "step": 4149 }, { "epoch": 1.07736240913811, "grad_norm": 1.9982742071151733, "learning_rate": 2.5789374954255864e-06, "loss": 0.1384306252002716, "step": 4150 }, { "epoch": 1.0776220145379023, "grad_norm": 0.9963706731796265, "learning_rate": 2.5779942390438396e-06, "loss": 0.10066027194261551, "step": 4151 }, { "epoch": 1.0778816199376946, "grad_norm": 2.336543083190918, "learning_rate": 2.5770510130687726e-06, "loss": 0.08926574885845184, "step": 4152 }, { "epoch": 1.078141225337487, "grad_norm": 1.9408934116363525, "learning_rate": 2.5761078176671275e-06, "loss": 0.09613482654094696, "step": 4153 }, { "epoch": 1.0784008307372794, "grad_norm": 1.418012261390686, "learning_rate": 2.5751646530056378e-06, "loss": 0.11145289987325668, "step": 4154 }, { "epoch": 1.0786604361370717, "grad_norm": 1.1056363582611084, "learning_rate": 2.574221519251035e-06, "loss": 0.08318756520748138, "step": 4155 }, { "epoch": 1.078920041536864, "grad_norm": 1.83256196975708, "learning_rate": 2.5732784165700418e-06, "loss": 0.09950388967990875, "step": 4156 }, { "epoch": 1.0791796469366564, "grad_norm": 1.340733528137207, "learning_rate": 2.5723353451293793e-06, "loss": 0.10450862348079681, "step": 4157 }, { "epoch": 1.0794392523364487, "grad_norm": 1.8450663089752197, "learning_rate": 2.571392305095759e-06, "loss": 0.10794264823198318, "step": 4158 }, { "epoch": 1.079698857736241, "grad_norm": 0.960726797580719, "learning_rate": 2.5704492966358897e-06, "loss": 0.11929121613502502, "step": 4159 }, { "epoch": 1.0799584631360333, "grad_norm": 1.3348394632339478, "learning_rate": 2.569506319916474e-06, "loss": 0.07922282069921494, "step": 4160 }, { "epoch": 1.0802180685358256, "grad_norm": 1.2756083011627197, "learning_rate": 2.5685633751042084e-06, "loss": 0.09314358234405518, "step": 4161 }, { "epoch": 1.0804776739356179, "grad_norm": 0.5301902890205383, "learning_rate": 2.5676204623657824e-06, "loss": 0.10468599200248718, "step": 4162 }, { "epoch": 1.0807372793354102, "grad_norm": 1.2583569288253784, "learning_rate": 2.566677581867884e-06, "loss": 0.12391610443592072, "step": 4163 }, { "epoch": 1.0809968847352025, "grad_norm": 1.6447484493255615, "learning_rate": 2.565734733777192e-06, "loss": 0.1137576475739479, "step": 4164 }, { "epoch": 1.0812564901349948, "grad_norm": 1.2877871990203857, "learning_rate": 2.564791918260378e-06, "loss": 0.12197602540254593, "step": 4165 }, { "epoch": 1.0815160955347871, "grad_norm": 1.1297355890274048, "learning_rate": 2.563849135484113e-06, "loss": 0.1079980731010437, "step": 4166 }, { "epoch": 1.0817757009345794, "grad_norm": 0.8218227624893188, "learning_rate": 2.562906385615059e-06, "loss": 0.10634259879589081, "step": 4167 }, { "epoch": 1.0820353063343717, "grad_norm": 1.160939335823059, "learning_rate": 2.561963668819873e-06, "loss": 0.09903126955032349, "step": 4168 }, { "epoch": 1.082294911734164, "grad_norm": 2.4371800422668457, "learning_rate": 2.5610209852652034e-06, "loss": 0.11075213551521301, "step": 4169 }, { "epoch": 1.0825545171339563, "grad_norm": 0.6440401673316956, "learning_rate": 2.5600783351176966e-06, "loss": 0.09601880609989166, "step": 4170 }, { "epoch": 1.0828141225337486, "grad_norm": 1.1619329452514648, "learning_rate": 2.559135718543992e-06, "loss": 0.10539465397596359, "step": 4171 }, { "epoch": 1.083073727933541, "grad_norm": 1.329147458076477, "learning_rate": 2.5581931357107226e-06, "loss": 0.07873204350471497, "step": 4172 }, { "epoch": 1.0833333333333333, "grad_norm": 0.9651356935501099, "learning_rate": 2.5572505867845145e-06, "loss": 0.10899728536605835, "step": 4173 }, { "epoch": 1.0835929387331256, "grad_norm": 1.134871244430542, "learning_rate": 2.5563080719319892e-06, "loss": 0.11177758872509003, "step": 4174 }, { "epoch": 1.0838525441329179, "grad_norm": 0.703917920589447, "learning_rate": 2.5553655913197612e-06, "loss": 0.10031837224960327, "step": 4175 }, { "epoch": 1.0841121495327102, "grad_norm": 0.748216986656189, "learning_rate": 2.554423145114442e-06, "loss": 0.10418634116649628, "step": 4176 }, { "epoch": 1.0843717549325027, "grad_norm": 1.339072585105896, "learning_rate": 2.553480733482631e-06, "loss": 0.11206929385662079, "step": 4177 }, { "epoch": 1.084631360332295, "grad_norm": 1.5450975894927979, "learning_rate": 2.5525383565909274e-06, "loss": 0.06305865943431854, "step": 4178 }, { "epoch": 1.0848909657320873, "grad_norm": 1.960640549659729, "learning_rate": 2.5515960146059204e-06, "loss": 0.0914054811000824, "step": 4179 }, { "epoch": 1.0851505711318796, "grad_norm": 1.4047281742095947, "learning_rate": 2.5506537076941955e-06, "loss": 0.09969660639762878, "step": 4180 }, { "epoch": 1.085410176531672, "grad_norm": 1.9494552612304688, "learning_rate": 2.5497114360223306e-06, "loss": 0.07780365645885468, "step": 4181 }, { "epoch": 1.0856697819314642, "grad_norm": 0.546002984046936, "learning_rate": 2.548769199756898e-06, "loss": 0.10517171025276184, "step": 4182 }, { "epoch": 1.0859293873312565, "grad_norm": 0.7272676825523376, "learning_rate": 2.547826999064463e-06, "loss": 0.10013356804847717, "step": 4183 }, { "epoch": 1.0861889927310489, "grad_norm": 1.2392303943634033, "learning_rate": 2.546884834111586e-06, "loss": 0.11830934882164001, "step": 4184 }, { "epoch": 1.0864485981308412, "grad_norm": 1.6948446035385132, "learning_rate": 2.545942705064819e-06, "loss": 0.11841622740030289, "step": 4185 }, { "epoch": 1.0867082035306335, "grad_norm": 1.6475998163223267, "learning_rate": 2.5450006120907095e-06, "loss": 0.10880494862794876, "step": 4186 }, { "epoch": 1.0869678089304258, "grad_norm": 2.3034725189208984, "learning_rate": 2.544058555355798e-06, "loss": 0.10566338896751404, "step": 4187 }, { "epoch": 1.087227414330218, "grad_norm": 1.7409005165100098, "learning_rate": 2.543116535026619e-06, "loss": 0.09773337095975876, "step": 4188 }, { "epoch": 1.0874870197300104, "grad_norm": 0.6165334582328796, "learning_rate": 2.5421745512696988e-06, "loss": 0.11362424492835999, "step": 4189 }, { "epoch": 1.0877466251298027, "grad_norm": 0.8938516974449158, "learning_rate": 2.54123260425156e-06, "loss": 0.11661914736032486, "step": 4190 }, { "epoch": 1.088006230529595, "grad_norm": 0.9386887550354004, "learning_rate": 2.540290694138717e-06, "loss": 0.09629863500595093, "step": 4191 }, { "epoch": 1.0882658359293873, "grad_norm": 1.047465205192566, "learning_rate": 2.5393488210976764e-06, "loss": 0.09756042063236237, "step": 4192 }, { "epoch": 1.0885254413291796, "grad_norm": 1.3783395290374756, "learning_rate": 2.5384069852949415e-06, "loss": 0.09948878735303879, "step": 4193 }, { "epoch": 1.088785046728972, "grad_norm": 1.178371787071228, "learning_rate": 2.537465186897006e-06, "loss": 0.10434645414352417, "step": 4194 }, { "epoch": 1.0890446521287642, "grad_norm": 0.7616375684738159, "learning_rate": 2.5365234260703607e-06, "loss": 0.10378049314022064, "step": 4195 }, { "epoch": 1.0893042575285565, "grad_norm": 1.2341277599334717, "learning_rate": 2.5355817029814844e-06, "loss": 0.097923144698143, "step": 4196 }, { "epoch": 1.0895638629283488, "grad_norm": 1.490099549293518, "learning_rate": 2.534640017796854e-06, "loss": 0.138931542634964, "step": 4197 }, { "epoch": 1.0898234683281411, "grad_norm": 1.7199040651321411, "learning_rate": 2.533698370682937e-06, "loss": 0.09666699916124344, "step": 4198 }, { "epoch": 1.0900830737279334, "grad_norm": 0.8599989414215088, "learning_rate": 2.5327567618061965e-06, "loss": 0.09987641870975494, "step": 4199 }, { "epoch": 1.0903426791277258, "grad_norm": 0.6447497010231018, "learning_rate": 2.5318151913330857e-06, "loss": 0.1150326281785965, "step": 4200 }, { "epoch": 1.090602284527518, "grad_norm": 1.141111969947815, "learning_rate": 2.5308736594300522e-06, "loss": 0.08671794831752777, "step": 4201 }, { "epoch": 1.0908618899273106, "grad_norm": 1.8143415451049805, "learning_rate": 2.529932166263539e-06, "loss": 0.12792879343032837, "step": 4202 }, { "epoch": 1.091121495327103, "grad_norm": 0.6936064958572388, "learning_rate": 2.5289907119999814e-06, "loss": 0.10116839408874512, "step": 4203 }, { "epoch": 1.0913811007268952, "grad_norm": 2.0204639434814453, "learning_rate": 2.528049296805804e-06, "loss": 0.11067177355289459, "step": 4204 }, { "epoch": 1.0916407061266875, "grad_norm": 0.9566385746002197, "learning_rate": 2.5271079208474288e-06, "loss": 0.09871124476194382, "step": 4205 }, { "epoch": 1.0919003115264798, "grad_norm": 1.3619251251220703, "learning_rate": 2.5261665842912696e-06, "loss": 0.0934697836637497, "step": 4206 }, { "epoch": 1.0921599169262721, "grad_norm": 2.2979300022125244, "learning_rate": 2.525225287303735e-06, "loss": 0.09129743278026581, "step": 4207 }, { "epoch": 1.0924195223260644, "grad_norm": 1.4586472511291504, "learning_rate": 2.5242840300512205e-06, "loss": 0.09261904656887054, "step": 4208 }, { "epoch": 1.0926791277258567, "grad_norm": 1.592861294746399, "learning_rate": 2.5233428127001226e-06, "loss": 0.12424929440021515, "step": 4209 }, { "epoch": 1.092938733125649, "grad_norm": 2.2805116176605225, "learning_rate": 2.522401635416825e-06, "loss": 0.12279827892780304, "step": 4210 }, { "epoch": 1.0931983385254413, "grad_norm": 1.1415367126464844, "learning_rate": 2.5214604983677084e-06, "loss": 0.09978878498077393, "step": 4211 }, { "epoch": 1.0934579439252337, "grad_norm": 1.571549892425537, "learning_rate": 2.5205194017191414e-06, "loss": 0.12911714613437653, "step": 4212 }, { "epoch": 1.093717549325026, "grad_norm": 3.6522767543792725, "learning_rate": 2.5195783456374902e-06, "loss": 0.08785964548587799, "step": 4213 }, { "epoch": 1.0939771547248183, "grad_norm": 1.567435383796692, "learning_rate": 2.518637330289111e-06, "loss": 0.10733364522457123, "step": 4214 }, { "epoch": 1.0942367601246106, "grad_norm": 1.8028690814971924, "learning_rate": 2.517696355840355e-06, "loss": 0.08227704465389252, "step": 4215 }, { "epoch": 1.0944963655244029, "grad_norm": 0.9726454615592957, "learning_rate": 2.5167554224575643e-06, "loss": 0.10702238976955414, "step": 4216 }, { "epoch": 1.0947559709241952, "grad_norm": 1.0579472780227661, "learning_rate": 2.5158145303070737e-06, "loss": 0.09946027398109436, "step": 4217 }, { "epoch": 1.0950155763239875, "grad_norm": 3.512981653213501, "learning_rate": 2.5148736795552125e-06, "loss": 0.07380004972219467, "step": 4218 }, { "epoch": 1.0952751817237798, "grad_norm": 1.7182592153549194, "learning_rate": 2.5139328703683007e-06, "loss": 0.11380620300769806, "step": 4219 }, { "epoch": 1.095534787123572, "grad_norm": 0.6548659801483154, "learning_rate": 2.5129921029126524e-06, "loss": 0.10944990813732147, "step": 4220 }, { "epoch": 1.0957943925233644, "grad_norm": 2.115196943283081, "learning_rate": 2.5120513773545737e-06, "loss": 0.09525709599256516, "step": 4221 }, { "epoch": 1.0960539979231567, "grad_norm": 0.7858138680458069, "learning_rate": 2.511110693860363e-06, "loss": 0.08430062234401703, "step": 4222 }, { "epoch": 1.096313603322949, "grad_norm": 1.8513163328170776, "learning_rate": 2.510170052596312e-06, "loss": 0.12624187767505646, "step": 4223 }, { "epoch": 1.0965732087227413, "grad_norm": 0.8238136172294617, "learning_rate": 2.5092294537287038e-06, "loss": 0.10887284576892853, "step": 4224 }, { "epoch": 1.0968328141225339, "grad_norm": 0.5874004364013672, "learning_rate": 2.5082888974238155e-06, "loss": 0.09126969426870346, "step": 4225 }, { "epoch": 1.0970924195223262, "grad_norm": 1.307328462600708, "learning_rate": 2.5073483838479164e-06, "loss": 0.10169100761413574, "step": 4226 }, { "epoch": 1.0973520249221185, "grad_norm": 1.846481442451477, "learning_rate": 2.5064079131672658e-06, "loss": 0.09840253740549088, "step": 4227 }, { "epoch": 1.0976116303219108, "grad_norm": 1.5782948732376099, "learning_rate": 2.5054674855481198e-06, "loss": 0.08216843754053116, "step": 4228 }, { "epoch": 1.097871235721703, "grad_norm": 2.159224510192871, "learning_rate": 2.5045271011567225e-06, "loss": 0.10171089321374893, "step": 4229 }, { "epoch": 1.0981308411214954, "grad_norm": 1.195103645324707, "learning_rate": 2.5035867601593145e-06, "loss": 0.12571123242378235, "step": 4230 }, { "epoch": 1.0983904465212877, "grad_norm": 1.4999433755874634, "learning_rate": 2.5026464627221233e-06, "loss": 0.0720609575510025, "step": 4231 }, { "epoch": 1.09865005192108, "grad_norm": 2.9598419666290283, "learning_rate": 2.501706209011374e-06, "loss": 0.11709066480398178, "step": 4232 }, { "epoch": 1.0989096573208723, "grad_norm": 1.8009798526763916, "learning_rate": 2.5007659991932826e-06, "loss": 0.08040972054004669, "step": 4233 }, { "epoch": 1.0991692627206646, "grad_norm": 1.6487783193588257, "learning_rate": 2.4998258334340563e-06, "loss": 0.0863153487443924, "step": 4234 }, { "epoch": 1.099428868120457, "grad_norm": 1.0245301723480225, "learning_rate": 2.498885711899894e-06, "loss": 0.09074528515338898, "step": 4235 }, { "epoch": 1.0996884735202492, "grad_norm": 1.005183458328247, "learning_rate": 2.497945634756987e-06, "loss": 0.117026187479496, "step": 4236 }, { "epoch": 1.0999480789200415, "grad_norm": 1.1736085414886475, "learning_rate": 2.497005602171522e-06, "loss": 0.11087416857481003, "step": 4237 }, { "epoch": 1.1002076843198338, "grad_norm": 1.6943937540054321, "learning_rate": 2.4960656143096732e-06, "loss": 0.12154296785593033, "step": 4238 }, { "epoch": 1.1004672897196262, "grad_norm": 2.094045639038086, "learning_rate": 2.4951256713376092e-06, "loss": 0.0892685204744339, "step": 4239 }, { "epoch": 1.1007268951194185, "grad_norm": 0.529097855091095, "learning_rate": 2.494185773421491e-06, "loss": 0.1142716258764267, "step": 4240 }, { "epoch": 1.1009865005192108, "grad_norm": 1.7511630058288574, "learning_rate": 2.4932459207274694e-06, "loss": 0.08068428933620453, "step": 4241 }, { "epoch": 1.101246105919003, "grad_norm": 0.7975839376449585, "learning_rate": 2.4923061134216927e-06, "loss": 0.10456281900405884, "step": 4242 }, { "epoch": 1.1015057113187954, "grad_norm": 2.1027238368988037, "learning_rate": 2.4913663516702927e-06, "loss": 0.11212195456027985, "step": 4243 }, { "epoch": 1.1017653167185877, "grad_norm": 1.0911930799484253, "learning_rate": 2.4904266356394003e-06, "loss": 0.09484539926052094, "step": 4244 }, { "epoch": 1.10202492211838, "grad_norm": 1.0713207721710205, "learning_rate": 2.4894869654951353e-06, "loss": 0.1108110323548317, "step": 4245 }, { "epoch": 1.1022845275181723, "grad_norm": 1.4329538345336914, "learning_rate": 2.48854734140361e-06, "loss": 0.11226882040500641, "step": 4246 }, { "epoch": 1.1025441329179646, "grad_norm": 0.9289718270301819, "learning_rate": 2.4876077635309283e-06, "loss": 0.10339900106191635, "step": 4247 }, { "epoch": 1.102803738317757, "grad_norm": 1.3483448028564453, "learning_rate": 2.4866682320431858e-06, "loss": 0.12832894921302795, "step": 4248 }, { "epoch": 1.1030633437175492, "grad_norm": 0.6341100931167603, "learning_rate": 2.4857287471064707e-06, "loss": 0.11473578959703445, "step": 4249 }, { "epoch": 1.1033229491173417, "grad_norm": 1.9279979467391968, "learning_rate": 2.484789308886862e-06, "loss": 0.07156671583652496, "step": 4250 }, { "epoch": 1.103582554517134, "grad_norm": 2.4480974674224854, "learning_rate": 2.4838499175504303e-06, "loss": 0.08032699674367905, "step": 4251 }, { "epoch": 1.1038421599169264, "grad_norm": 1.0610551834106445, "learning_rate": 2.4829105732632397e-06, "loss": 0.13661430776119232, "step": 4252 }, { "epoch": 1.1041017653167187, "grad_norm": 1.0920541286468506, "learning_rate": 2.4819712761913444e-06, "loss": 0.0870678722858429, "step": 4253 }, { "epoch": 1.104361370716511, "grad_norm": 1.4953206777572632, "learning_rate": 2.48103202650079e-06, "loss": 0.09465834498405457, "step": 4254 }, { "epoch": 1.1046209761163033, "grad_norm": 1.8006747961044312, "learning_rate": 2.480092824357615e-06, "loss": 0.10745728760957718, "step": 4255 }, { "epoch": 1.1048805815160956, "grad_norm": 2.605776786804199, "learning_rate": 2.4791536699278487e-06, "loss": 0.11686364561319351, "step": 4256 }, { "epoch": 1.105140186915888, "grad_norm": 0.9788538813591003, "learning_rate": 2.478214563377512e-06, "loss": 0.11058512330055237, "step": 4257 }, { "epoch": 1.1053997923156802, "grad_norm": 0.6690999269485474, "learning_rate": 2.4772755048726172e-06, "loss": 0.12032033503055573, "step": 4258 }, { "epoch": 1.1056593977154725, "grad_norm": 1.0514546632766724, "learning_rate": 2.4763364945791685e-06, "loss": 0.10390734672546387, "step": 4259 }, { "epoch": 1.1059190031152648, "grad_norm": 0.7881377935409546, "learning_rate": 2.475397532663161e-06, "loss": 0.1058397889137268, "step": 4260 }, { "epoch": 1.1061786085150571, "grad_norm": 1.2154139280319214, "learning_rate": 2.474458619290584e-06, "loss": 0.12085257470607758, "step": 4261 }, { "epoch": 1.1064382139148494, "grad_norm": 1.598276972770691, "learning_rate": 2.4735197546274125e-06, "loss": 0.07697385549545288, "step": 4262 }, { "epoch": 1.1066978193146417, "grad_norm": 1.179099202156067, "learning_rate": 2.4725809388396175e-06, "loss": 0.041386790573596954, "step": 4263 }, { "epoch": 1.106957424714434, "grad_norm": 1.6062880754470825, "learning_rate": 2.4716421720931616e-06, "loss": 0.08913623541593552, "step": 4264 }, { "epoch": 1.1072170301142263, "grad_norm": 1.3450043201446533, "learning_rate": 2.4707034545539963e-06, "loss": 0.10415361076593399, "step": 4265 }, { "epoch": 1.1074766355140186, "grad_norm": 0.761077880859375, "learning_rate": 2.4697647863880648e-06, "loss": 0.10116410255432129, "step": 4266 }, { "epoch": 1.107736240913811, "grad_norm": 0.8360349535942078, "learning_rate": 2.468826167761301e-06, "loss": 0.1105959415435791, "step": 4267 }, { "epoch": 1.1079958463136033, "grad_norm": 0.6058933734893799, "learning_rate": 2.4678875988396344e-06, "loss": 0.10368911921977997, "step": 4268 }, { "epoch": 1.1082554517133956, "grad_norm": 0.8574625253677368, "learning_rate": 2.466949079788981e-06, "loss": 0.10708189010620117, "step": 4269 }, { "epoch": 1.1085150571131879, "grad_norm": 1.822302222251892, "learning_rate": 2.4660106107752485e-06, "loss": 0.10175661742687225, "step": 4270 }, { "epoch": 1.1087746625129802, "grad_norm": 2.8706352710723877, "learning_rate": 2.4650721919643376e-06, "loss": 0.10900063812732697, "step": 4271 }, { "epoch": 1.1090342679127725, "grad_norm": 1.1738449335098267, "learning_rate": 2.46413382352214e-06, "loss": 0.0886528491973877, "step": 4272 }, { "epoch": 1.109293873312565, "grad_norm": 1.1670653820037842, "learning_rate": 2.463195505614536e-06, "loss": 0.09688359498977661, "step": 4273 }, { "epoch": 1.1095534787123573, "grad_norm": 1.6803380250930786, "learning_rate": 2.4622572384074e-06, "loss": 0.0788150280714035, "step": 4274 }, { "epoch": 1.1098130841121496, "grad_norm": 0.9431877136230469, "learning_rate": 2.4613190220665957e-06, "loss": 0.12419348210096359, "step": 4275 }, { "epoch": 1.110072689511942, "grad_norm": 1.5674984455108643, "learning_rate": 2.4603808567579787e-06, "loss": 0.1117384135723114, "step": 4276 }, { "epoch": 1.1103322949117342, "grad_norm": 1.8692229986190796, "learning_rate": 2.4594427426473944e-06, "loss": 0.08338241279125214, "step": 4277 }, { "epoch": 1.1105919003115265, "grad_norm": 1.1412792205810547, "learning_rate": 2.45850467990068e-06, "loss": 0.10995469242334366, "step": 4278 }, { "epoch": 1.1108515057113189, "grad_norm": 0.8590230345726013, "learning_rate": 2.457566668683664e-06, "loss": 0.11230258643627167, "step": 4279 }, { "epoch": 1.1111111111111112, "grad_norm": 2.240233898162842, "learning_rate": 2.456628709162166e-06, "loss": 0.09586222469806671, "step": 4280 }, { "epoch": 1.1113707165109035, "grad_norm": 1.5489022731781006, "learning_rate": 2.455690801501994e-06, "loss": 0.11895757913589478, "step": 4281 }, { "epoch": 1.1116303219106958, "grad_norm": 1.2584431171417236, "learning_rate": 2.4547529458689496e-06, "loss": 0.11793193221092224, "step": 4282 }, { "epoch": 1.111889927310488, "grad_norm": 2.100128650665283, "learning_rate": 2.4538151424288246e-06, "loss": 0.12146516144275665, "step": 4283 }, { "epoch": 1.1121495327102804, "grad_norm": 1.6718868017196655, "learning_rate": 2.452877391347401e-06, "loss": 0.06866662204265594, "step": 4284 }, { "epoch": 1.1124091381100727, "grad_norm": 1.4637898206710815, "learning_rate": 2.4519396927904505e-06, "loss": 0.10358348488807678, "step": 4285 }, { "epoch": 1.112668743509865, "grad_norm": 1.4698368310928345, "learning_rate": 2.4510020469237376e-06, "loss": 0.10384464263916016, "step": 4286 }, { "epoch": 1.1129283489096573, "grad_norm": 0.6325616836547852, "learning_rate": 2.450064453913017e-06, "loss": 0.10381516814231873, "step": 4287 }, { "epoch": 1.1131879543094496, "grad_norm": 0.8121567368507385, "learning_rate": 2.449126913924034e-06, "loss": 0.1010286882519722, "step": 4288 }, { "epoch": 1.113447559709242, "grad_norm": 1.3462440967559814, "learning_rate": 2.4481894271225225e-06, "loss": 0.12185665220022202, "step": 4289 }, { "epoch": 1.1137071651090342, "grad_norm": 1.848171353340149, "learning_rate": 2.44725199367421e-06, "loss": 0.10951857268810272, "step": 4290 }, { "epoch": 1.1139667705088265, "grad_norm": 1.1093195676803589, "learning_rate": 2.4463146137448127e-06, "loss": 0.10558392107486725, "step": 4291 }, { "epoch": 1.1142263759086188, "grad_norm": 1.9651559591293335, "learning_rate": 2.4453772875000398e-06, "loss": 0.11080946773290634, "step": 4292 }, { "epoch": 1.1144859813084111, "grad_norm": 1.4816288948059082, "learning_rate": 2.444440015105586e-06, "loss": 0.10158485174179077, "step": 4293 }, { "epoch": 1.1147455867082035, "grad_norm": 0.6975231766700745, "learning_rate": 2.4435027967271413e-06, "loss": 0.11589137464761734, "step": 4294 }, { "epoch": 1.1150051921079958, "grad_norm": 1.1668651103973389, "learning_rate": 2.442565632530384e-06, "loss": 0.1042175218462944, "step": 4295 }, { "epoch": 1.115264797507788, "grad_norm": 1.1044684648513794, "learning_rate": 2.441628522680985e-06, "loss": 0.06902109086513519, "step": 4296 }, { "epoch": 1.1155244029075804, "grad_norm": 0.7399016618728638, "learning_rate": 2.4406914673446012e-06, "loss": 0.0964137464761734, "step": 4297 }, { "epoch": 1.1157840083073727, "grad_norm": 0.5390303134918213, "learning_rate": 2.4397544666868837e-06, "loss": 0.10327216982841492, "step": 4298 }, { "epoch": 1.1160436137071652, "grad_norm": 1.2438400983810425, "learning_rate": 2.438817520873474e-06, "loss": 0.08313437551259995, "step": 4299 }, { "epoch": 1.1163032191069575, "grad_norm": 1.5189050436019897, "learning_rate": 2.4378806300700003e-06, "loss": 0.1117534190416336, "step": 4300 }, { "epoch": 1.1165628245067498, "grad_norm": 0.7906357049942017, "learning_rate": 2.436943794442085e-06, "loss": 0.09688113629817963, "step": 4301 }, { "epoch": 1.1168224299065421, "grad_norm": 0.8308025002479553, "learning_rate": 2.436007014155339e-06, "loss": 0.11722768843173981, "step": 4302 }, { "epoch": 1.1170820353063344, "grad_norm": 2.5163309574127197, "learning_rate": 2.435070289375363e-06, "loss": 0.09681698679924011, "step": 4303 }, { "epoch": 1.1173416407061267, "grad_norm": 0.8203453421592712, "learning_rate": 2.434133620267749e-06, "loss": 0.0993368849158287, "step": 4304 }, { "epoch": 1.117601246105919, "grad_norm": 0.7615283727645874, "learning_rate": 2.433197006998078e-06, "loss": 0.1175873801112175, "step": 4305 }, { "epoch": 1.1178608515057114, "grad_norm": 1.4317824840545654, "learning_rate": 2.432260449731923e-06, "loss": 0.09804056584835052, "step": 4306 }, { "epoch": 1.1181204569055037, "grad_norm": 1.1486117839813232, "learning_rate": 2.431323948634846e-06, "loss": 0.10223188996315002, "step": 4307 }, { "epoch": 1.118380062305296, "grad_norm": 1.8406909704208374, "learning_rate": 2.4303875038723975e-06, "loss": 0.09315359592437744, "step": 4308 }, { "epoch": 1.1186396677050883, "grad_norm": 2.229079246520996, "learning_rate": 2.429451115610119e-06, "loss": 0.11104267090559006, "step": 4309 }, { "epoch": 1.1188992731048806, "grad_norm": 0.7105711698532104, "learning_rate": 2.428514784013545e-06, "loss": 0.10629931092262268, "step": 4310 }, { "epoch": 1.1191588785046729, "grad_norm": 1.3428068161010742, "learning_rate": 2.4275785092481965e-06, "loss": 0.07951992750167847, "step": 4311 }, { "epoch": 1.1194184839044652, "grad_norm": 0.8047057390213013, "learning_rate": 2.426642291479584e-06, "loss": 0.09167380630970001, "step": 4312 }, { "epoch": 1.1196780893042575, "grad_norm": 3.155839204788208, "learning_rate": 2.4257061308732104e-06, "loss": 0.10675346851348877, "step": 4313 }, { "epoch": 1.1199376947040498, "grad_norm": 1.0162407159805298, "learning_rate": 2.424770027594568e-06, "loss": 0.09308852255344391, "step": 4314 }, { "epoch": 1.120197300103842, "grad_norm": 1.5873818397521973, "learning_rate": 2.4238339818091383e-06, "loss": 0.08678607642650604, "step": 4315 }, { "epoch": 1.1204569055036344, "grad_norm": 1.049943447113037, "learning_rate": 2.422897993682392e-06, "loss": 0.10962989926338196, "step": 4316 }, { "epoch": 1.1207165109034267, "grad_norm": 2.3240277767181396, "learning_rate": 2.421962063379791e-06, "loss": 0.1117195188999176, "step": 4317 }, { "epoch": 1.120976116303219, "grad_norm": 1.5794689655303955, "learning_rate": 2.421026191066785e-06, "loss": 0.1155800148844719, "step": 4318 }, { "epoch": 1.1212357217030113, "grad_norm": 1.440293312072754, "learning_rate": 2.4200903769088174e-06, "loss": 0.09775292128324509, "step": 4319 }, { "epoch": 1.1214953271028036, "grad_norm": 2.328019380569458, "learning_rate": 2.419154621071317e-06, "loss": 0.11464518308639526, "step": 4320 }, { "epoch": 1.1217549325025962, "grad_norm": 1.3875153064727783, "learning_rate": 2.4182189237197033e-06, "loss": 0.11210212111473083, "step": 4321 }, { "epoch": 1.1220145379023885, "grad_norm": 1.9318387508392334, "learning_rate": 2.4172832850193877e-06, "loss": 0.11194166541099548, "step": 4322 }, { "epoch": 1.1222741433021808, "grad_norm": 1.044551134109497, "learning_rate": 2.41634770513577e-06, "loss": 0.1027529388666153, "step": 4323 }, { "epoch": 1.122533748701973, "grad_norm": 0.9561600685119629, "learning_rate": 2.4154121842342363e-06, "loss": 0.1108727753162384, "step": 4324 }, { "epoch": 1.1227933541017654, "grad_norm": 1.2834080457687378, "learning_rate": 2.414476722480169e-06, "loss": 0.09994490444660187, "step": 4325 }, { "epoch": 1.1230529595015577, "grad_norm": 2.325542688369751, "learning_rate": 2.413541320038934e-06, "loss": 0.09056390821933746, "step": 4326 }, { "epoch": 1.12331256490135, "grad_norm": 0.9765632152557373, "learning_rate": 2.4126059770758907e-06, "loss": 0.11179609596729279, "step": 4327 }, { "epoch": 1.1235721703011423, "grad_norm": 0.8203895092010498, "learning_rate": 2.4116706937563848e-06, "loss": 0.12111398577690125, "step": 4328 }, { "epoch": 1.1238317757009346, "grad_norm": 0.9055829048156738, "learning_rate": 2.4107354702457524e-06, "loss": 0.0963592678308487, "step": 4329 }, { "epoch": 1.124091381100727, "grad_norm": 1.208454966545105, "learning_rate": 2.409800306709323e-06, "loss": 0.10082040727138519, "step": 4330 }, { "epoch": 1.1243509865005192, "grad_norm": 0.8353259563446045, "learning_rate": 2.4088652033124086e-06, "loss": 0.09325079619884491, "step": 4331 }, { "epoch": 1.1246105919003115, "grad_norm": 0.6426050662994385, "learning_rate": 2.407930160220315e-06, "loss": 0.11928203701972961, "step": 4332 }, { "epoch": 1.1248701973001038, "grad_norm": 2.299827814102173, "learning_rate": 2.406995177598337e-06, "loss": 0.07353129237890244, "step": 4333 }, { "epoch": 1.1251298026998962, "grad_norm": 0.7503188848495483, "learning_rate": 2.4060602556117578e-06, "loss": 0.10680104792118073, "step": 4334 }, { "epoch": 1.1253894080996885, "grad_norm": 1.7902098894119263, "learning_rate": 2.40512539442585e-06, "loss": 0.06182631477713585, "step": 4335 }, { "epoch": 1.1256490134994808, "grad_norm": 0.6934739351272583, "learning_rate": 2.404190594205876e-06, "loss": 0.10438719391822815, "step": 4336 }, { "epoch": 1.125908618899273, "grad_norm": 0.7267701029777527, "learning_rate": 2.4032558551170875e-06, "loss": 0.10481169819831848, "step": 4337 }, { "epoch": 1.1261682242990654, "grad_norm": 2.4624831676483154, "learning_rate": 2.402321177324724e-06, "loss": 0.08906758576631546, "step": 4338 }, { "epoch": 1.1264278296988577, "grad_norm": 1.6505070924758911, "learning_rate": 2.4013865609940156e-06, "loss": 0.12001322209835052, "step": 4339 }, { "epoch": 1.12668743509865, "grad_norm": 1.3725111484527588, "learning_rate": 2.400452006290181e-06, "loss": 0.09971727430820465, "step": 4340 }, { "epoch": 1.1269470404984423, "grad_norm": 1.9035242795944214, "learning_rate": 2.399517513378428e-06, "loss": 0.13138598203659058, "step": 4341 }, { "epoch": 1.1272066458982346, "grad_norm": 1.0673003196716309, "learning_rate": 2.398583082423954e-06, "loss": 0.1111227497458458, "step": 4342 }, { "epoch": 1.127466251298027, "grad_norm": 0.8644925355911255, "learning_rate": 2.3976487135919444e-06, "loss": 0.09399434924125671, "step": 4343 }, { "epoch": 1.1277258566978192, "grad_norm": 2.3697478771209717, "learning_rate": 2.3967144070475746e-06, "loss": 0.11235851794481277, "step": 4344 }, { "epoch": 1.1279854620976115, "grad_norm": 1.7145177125930786, "learning_rate": 2.395780162956008e-06, "loss": 0.10060441493988037, "step": 4345 }, { "epoch": 1.1282450674974038, "grad_norm": 2.0449140071868896, "learning_rate": 2.3948459814823986e-06, "loss": 0.10316136479377747, "step": 4346 }, { "epoch": 1.1285046728971961, "grad_norm": 1.7464412450790405, "learning_rate": 2.3939118627918874e-06, "loss": 0.1048264354467392, "step": 4347 }, { "epoch": 1.1287642782969887, "grad_norm": 1.4006708860397339, "learning_rate": 2.3929778070496055e-06, "loss": 0.11037217080593109, "step": 4348 }, { "epoch": 1.129023883696781, "grad_norm": 1.5851446390151978, "learning_rate": 2.3920438144206726e-06, "loss": 0.12118545174598694, "step": 4349 }, { "epoch": 1.1292834890965733, "grad_norm": 1.0622915029525757, "learning_rate": 2.3911098850701975e-06, "loss": 0.10449111461639404, "step": 4350 }, { "epoch": 1.1295430944963656, "grad_norm": 2.192049264907837, "learning_rate": 2.3901760191632774e-06, "loss": 0.08227626979351044, "step": 4351 }, { "epoch": 1.129802699896158, "grad_norm": 2.004770517349243, "learning_rate": 2.389242216864997e-06, "loss": 0.08085064589977264, "step": 4352 }, { "epoch": 1.1300623052959502, "grad_norm": 0.6535686254501343, "learning_rate": 2.3883084783404337e-06, "loss": 0.11239030957221985, "step": 4353 }, { "epoch": 1.1303219106957425, "grad_norm": 1.278141975402832, "learning_rate": 2.38737480375465e-06, "loss": 0.11821940541267395, "step": 4354 }, { "epoch": 1.1305815160955348, "grad_norm": 1.9028230905532837, "learning_rate": 2.3864411932726967e-06, "loss": 0.10374227166175842, "step": 4355 }, { "epoch": 1.1308411214953271, "grad_norm": 2.1595332622528076, "learning_rate": 2.3855076470596165e-06, "loss": 0.0824260264635086, "step": 4356 }, { "epoch": 1.1311007268951194, "grad_norm": 1.4295504093170166, "learning_rate": 2.3845741652804393e-06, "loss": 0.05293542146682739, "step": 4357 }, { "epoch": 1.1313603322949117, "grad_norm": 0.8609982132911682, "learning_rate": 2.3836407481001823e-06, "loss": 0.1034872829914093, "step": 4358 }, { "epoch": 1.131619937694704, "grad_norm": 1.854642629623413, "learning_rate": 2.3827073956838515e-06, "loss": 0.040594376623630524, "step": 4359 }, { "epoch": 1.1318795430944963, "grad_norm": 0.768905758857727, "learning_rate": 2.3817741081964433e-06, "loss": 0.10650031268596649, "step": 4360 }, { "epoch": 1.1321391484942886, "grad_norm": 1.134076714515686, "learning_rate": 2.3808408858029437e-06, "loss": 0.09900547564029694, "step": 4361 }, { "epoch": 1.132398753894081, "grad_norm": 1.7140746116638184, "learning_rate": 2.379907728668321e-06, "loss": 0.08451271057128906, "step": 4362 }, { "epoch": 1.1326583592938733, "grad_norm": 1.442422866821289, "learning_rate": 2.378974636957538e-06, "loss": 0.08498166501522064, "step": 4363 }, { "epoch": 1.1329179646936656, "grad_norm": 0.7176697254180908, "learning_rate": 2.3780416108355433e-06, "loss": 0.09068211913108826, "step": 4364 }, { "epoch": 1.1331775700934579, "grad_norm": 1.0424057245254517, "learning_rate": 2.3771086504672762e-06, "loss": 0.07956571877002716, "step": 4365 }, { "epoch": 1.1334371754932502, "grad_norm": 1.8056707382202148, "learning_rate": 2.3761757560176614e-06, "loss": 0.13409999012947083, "step": 4366 }, { "epoch": 1.1336967808930425, "grad_norm": 1.0518840551376343, "learning_rate": 2.3752429276516132e-06, "loss": 0.09285500645637512, "step": 4367 }, { "epoch": 1.1339563862928348, "grad_norm": 1.0021462440490723, "learning_rate": 2.3743101655340346e-06, "loss": 0.11130163073539734, "step": 4368 }, { "epoch": 1.1342159916926273, "grad_norm": 0.6502565145492554, "learning_rate": 2.3733774698298173e-06, "loss": 0.12276779115200043, "step": 4369 }, { "epoch": 1.1344755970924196, "grad_norm": 1.6288440227508545, "learning_rate": 2.3724448407038386e-06, "loss": 0.07915505766868591, "step": 4370 }, { "epoch": 1.134735202492212, "grad_norm": 1.6207919120788574, "learning_rate": 2.371512278320968e-06, "loss": 0.10893896967172623, "step": 4371 }, { "epoch": 1.1349948078920042, "grad_norm": 3.538769245147705, "learning_rate": 2.3705797828460596e-06, "loss": 0.127923846244812, "step": 4372 }, { "epoch": 1.1352544132917965, "grad_norm": 1.2556806802749634, "learning_rate": 2.369647354443959e-06, "loss": 0.11218670010566711, "step": 4373 }, { "epoch": 1.1355140186915889, "grad_norm": 0.9369907975196838, "learning_rate": 2.3687149932794967e-06, "loss": 0.09264862537384033, "step": 4374 }, { "epoch": 1.1357736240913812, "grad_norm": 1.001805305480957, "learning_rate": 2.3677826995174935e-06, "loss": 0.10658704489469528, "step": 4375 }, { "epoch": 1.1360332294911735, "grad_norm": 1.3638861179351807, "learning_rate": 2.366850473322757e-06, "loss": 0.0919300839304924, "step": 4376 }, { "epoch": 1.1362928348909658, "grad_norm": 0.9094088077545166, "learning_rate": 2.3659183148600847e-06, "loss": 0.10608091205358505, "step": 4377 }, { "epoch": 1.136552440290758, "grad_norm": 0.6421760320663452, "learning_rate": 2.3649862242942595e-06, "loss": 0.11451096832752228, "step": 4378 }, { "epoch": 1.1368120456905504, "grad_norm": 0.7091236114501953, "learning_rate": 2.364054201790054e-06, "loss": 0.10442399978637695, "step": 4379 }, { "epoch": 1.1370716510903427, "grad_norm": 1.08834707736969, "learning_rate": 2.363122247512229e-06, "loss": 0.11544512212276459, "step": 4380 }, { "epoch": 1.137331256490135, "grad_norm": 1.7827129364013672, "learning_rate": 2.3621903616255322e-06, "loss": 0.1098882332444191, "step": 4381 }, { "epoch": 1.1375908618899273, "grad_norm": 2.1928653717041016, "learning_rate": 2.3612585442946994e-06, "loss": 0.11050808429718018, "step": 4382 }, { "epoch": 1.1378504672897196, "grad_norm": 1.114250898361206, "learning_rate": 2.3603267956844555e-06, "loss": 0.11805605888366699, "step": 4383 }, { "epoch": 1.138110072689512, "grad_norm": 0.9422768354415894, "learning_rate": 2.3593951159595125e-06, "loss": 0.10817033052444458, "step": 4384 }, { "epoch": 1.1383696780893042, "grad_norm": 0.6579753756523132, "learning_rate": 2.358463505284568e-06, "loss": 0.09776079654693604, "step": 4385 }, { "epoch": 1.1386292834890965, "grad_norm": 1.128190040588379, "learning_rate": 2.3575319638243103e-06, "loss": 0.11945868283510208, "step": 4386 }, { "epoch": 1.1388888888888888, "grad_norm": 3.2927284240722656, "learning_rate": 2.3566004917434165e-06, "loss": 0.1061900183558464, "step": 4387 }, { "epoch": 1.1391484942886811, "grad_norm": 1.4993547201156616, "learning_rate": 2.3556690892065487e-06, "loss": 0.075177863240242, "step": 4388 }, { "epoch": 1.1394080996884735, "grad_norm": 0.6495096683502197, "learning_rate": 2.354737756378355e-06, "loss": 0.10707490146160126, "step": 4389 }, { "epoch": 1.1396677050882658, "grad_norm": 0.7009621262550354, "learning_rate": 2.353806493423476e-06, "loss": 0.10380977392196655, "step": 4390 }, { "epoch": 1.139927310488058, "grad_norm": 1.6770870685577393, "learning_rate": 2.352875300506538e-06, "loss": 0.07788489758968353, "step": 4391 }, { "epoch": 1.1401869158878504, "grad_norm": 1.0194661617279053, "learning_rate": 2.351944177792155e-06, "loss": 0.10904770344495773, "step": 4392 }, { "epoch": 1.1404465212876427, "grad_norm": 0.6202057600021362, "learning_rate": 2.3510131254449253e-06, "loss": 0.11700282245874405, "step": 4393 }, { "epoch": 1.140706126687435, "grad_norm": 1.489181637763977, "learning_rate": 2.35008214362944e-06, "loss": 0.10890354961156845, "step": 4394 }, { "epoch": 1.1409657320872273, "grad_norm": 1.3954098224639893, "learning_rate": 2.349151232510274e-06, "loss": 0.09087663888931274, "step": 4395 }, { "epoch": 1.1412253374870198, "grad_norm": 2.2300591468811035, "learning_rate": 2.3482203922519937e-06, "loss": 0.13107794523239136, "step": 4396 }, { "epoch": 1.1414849428868121, "grad_norm": 1.0228581428527832, "learning_rate": 2.3472896230191473e-06, "loss": 0.09891146421432495, "step": 4397 }, { "epoch": 1.1417445482866044, "grad_norm": 0.9266546964645386, "learning_rate": 2.3463589249762747e-06, "loss": 0.061884865164756775, "step": 4398 }, { "epoch": 1.1420041536863967, "grad_norm": 0.6919386386871338, "learning_rate": 2.345428298287902e-06, "loss": 0.10801178216934204, "step": 4399 }, { "epoch": 1.142263759086189, "grad_norm": 1.068791151046753, "learning_rate": 2.3444977431185432e-06, "loss": 0.0914991945028305, "step": 4400 }, { "epoch": 1.1425233644859814, "grad_norm": 1.0556471347808838, "learning_rate": 2.3435672596326987e-06, "loss": 0.10949760675430298, "step": 4401 }, { "epoch": 1.1427829698857737, "grad_norm": 1.4118233919143677, "learning_rate": 2.3426368479948556e-06, "loss": 0.09260527789592743, "step": 4402 }, { "epoch": 1.143042575285566, "grad_norm": 1.475872278213501, "learning_rate": 2.3417065083694913e-06, "loss": 0.05625351518392563, "step": 4403 }, { "epoch": 1.1433021806853583, "grad_norm": 1.0298309326171875, "learning_rate": 2.3407762409210677e-06, "loss": 0.10350247472524643, "step": 4404 }, { "epoch": 1.1435617860851506, "grad_norm": 1.9240138530731201, "learning_rate": 2.339846045814034e-06, "loss": 0.1189717948436737, "step": 4405 }, { "epoch": 1.1438213914849429, "grad_norm": 1.2150708436965942, "learning_rate": 2.3389159232128285e-06, "loss": 0.1124582588672638, "step": 4406 }, { "epoch": 1.1440809968847352, "grad_norm": 1.0648010969161987, "learning_rate": 2.337985873281875e-06, "loss": 0.10791400074958801, "step": 4407 }, { "epoch": 1.1443406022845275, "grad_norm": 0.7943122386932373, "learning_rate": 2.337055896185586e-06, "loss": 0.1105341762304306, "step": 4408 }, { "epoch": 1.1446002076843198, "grad_norm": 1.720615029335022, "learning_rate": 2.336125992088359e-06, "loss": 0.058495961129665375, "step": 4409 }, { "epoch": 1.144859813084112, "grad_norm": 0.7004821300506592, "learning_rate": 2.3351961611545798e-06, "loss": 0.10837087780237198, "step": 4410 }, { "epoch": 1.1451194184839044, "grad_norm": 1.1647357940673828, "learning_rate": 2.3342664035486216e-06, "loss": 0.0948290228843689, "step": 4411 }, { "epoch": 1.1453790238836967, "grad_norm": 1.164368987083435, "learning_rate": 2.3333367194348443e-06, "loss": 0.10755547881126404, "step": 4412 }, { "epoch": 1.145638629283489, "grad_norm": 0.8534845113754272, "learning_rate": 2.3324071089775953e-06, "loss": 0.11656920611858368, "step": 4413 }, { "epoch": 1.1458982346832813, "grad_norm": 1.1953762769699097, "learning_rate": 2.3314775723412065e-06, "loss": 0.1067732647061348, "step": 4414 }, { "epoch": 1.1461578400830736, "grad_norm": 1.3485416173934937, "learning_rate": 2.330548109690002e-06, "loss": 0.09313680231571198, "step": 4415 }, { "epoch": 1.146417445482866, "grad_norm": 0.9472092390060425, "learning_rate": 2.3296187211882854e-06, "loss": 0.09323159605264664, "step": 4416 }, { "epoch": 1.1466770508826585, "grad_norm": 1.3689568042755127, "learning_rate": 2.328689407000354e-06, "loss": 0.11384659260511398, "step": 4417 }, { "epoch": 1.1469366562824508, "grad_norm": 1.8586674928665161, "learning_rate": 2.327760167290489e-06, "loss": 0.11340729892253876, "step": 4418 }, { "epoch": 1.147196261682243, "grad_norm": 1.8910760879516602, "learning_rate": 2.3268310022229597e-06, "loss": 0.09940376877784729, "step": 4419 }, { "epoch": 1.1474558670820354, "grad_norm": 1.85694420337677, "learning_rate": 2.3259019119620184e-06, "loss": 0.11307835578918457, "step": 4420 }, { "epoch": 1.1477154724818277, "grad_norm": 1.1081148386001587, "learning_rate": 2.324972896671908e-06, "loss": 0.10927946865558624, "step": 4421 }, { "epoch": 1.14797507788162, "grad_norm": 3.321838617324829, "learning_rate": 2.324043956516859e-06, "loss": 0.1376439481973648, "step": 4422 }, { "epoch": 1.1482346832814123, "grad_norm": 0.8872618675231934, "learning_rate": 2.323115091661085e-06, "loss": 0.09728182852268219, "step": 4423 }, { "epoch": 1.1484942886812046, "grad_norm": 0.7457396984100342, "learning_rate": 2.3221863022687877e-06, "loss": 0.10167980194091797, "step": 4424 }, { "epoch": 1.148753894080997, "grad_norm": 1.04591965675354, "learning_rate": 2.321257588504158e-06, "loss": 0.08465282618999481, "step": 4425 }, { "epoch": 1.1490134994807892, "grad_norm": 1.506716012954712, "learning_rate": 2.3203289505313677e-06, "loss": 0.07911841571331024, "step": 4426 }, { "epoch": 1.1492731048805815, "grad_norm": 1.4947651624679565, "learning_rate": 2.3194003885145826e-06, "loss": 0.09856444597244263, "step": 4427 }, { "epoch": 1.1495327102803738, "grad_norm": 0.8456995487213135, "learning_rate": 2.3184719026179485e-06, "loss": 0.09893780946731567, "step": 4428 }, { "epoch": 1.1497923156801662, "grad_norm": 1.3134331703186035, "learning_rate": 2.317543493005602e-06, "loss": 0.12735699117183685, "step": 4429 }, { "epoch": 1.1500519210799585, "grad_norm": 1.8565930128097534, "learning_rate": 2.316615159841663e-06, "loss": 0.11898043751716614, "step": 4430 }, { "epoch": 1.1503115264797508, "grad_norm": 1.9316636323928833, "learning_rate": 2.315686903290242e-06, "loss": 0.09750805050134659, "step": 4431 }, { "epoch": 1.150571131879543, "grad_norm": 0.9393948912620544, "learning_rate": 2.3147587235154307e-06, "loss": 0.09218500554561615, "step": 4432 }, { "epoch": 1.1508307372793354, "grad_norm": 1.7961041927337646, "learning_rate": 2.3138306206813117e-06, "loss": 0.07314403355121613, "step": 4433 }, { "epoch": 1.1510903426791277, "grad_norm": 2.000570774078369, "learning_rate": 2.3129025949519517e-06, "loss": 0.10143455862998962, "step": 4434 }, { "epoch": 1.15134994807892, "grad_norm": 0.8791841268539429, "learning_rate": 2.3119746464914055e-06, "loss": 0.10150055587291718, "step": 4435 }, { "epoch": 1.1516095534787123, "grad_norm": 1.793145775794983, "learning_rate": 2.3110467754637115e-06, "loss": 0.10272446274757385, "step": 4436 }, { "epoch": 1.1518691588785046, "grad_norm": 0.9657722115516663, "learning_rate": 2.3101189820328977e-06, "loss": 0.11488791555166245, "step": 4437 }, { "epoch": 1.152128764278297, "grad_norm": 1.8434735536575317, "learning_rate": 2.3091912663629758e-06, "loss": 0.09032037854194641, "step": 4438 }, { "epoch": 1.1523883696780892, "grad_norm": 2.9160826206207275, "learning_rate": 2.3082636286179446e-06, "loss": 0.10679259151220322, "step": 4439 }, { "epoch": 1.1526479750778815, "grad_norm": 1.7566896677017212, "learning_rate": 2.3073360689617893e-06, "loss": 0.09719331562519073, "step": 4440 }, { "epoch": 1.1529075804776738, "grad_norm": 1.3811990022659302, "learning_rate": 2.3064085875584814e-06, "loss": 0.06944697350263596, "step": 4441 }, { "epoch": 1.1531671858774661, "grad_norm": 2.5416715145111084, "learning_rate": 2.3054811845719786e-06, "loss": 0.08947455883026123, "step": 4442 }, { "epoch": 1.1534267912772584, "grad_norm": 1.9335685968399048, "learning_rate": 2.304553860166224e-06, "loss": 0.13234175741672516, "step": 4443 }, { "epoch": 1.153686396677051, "grad_norm": 1.4636236429214478, "learning_rate": 2.303626614505147e-06, "loss": 0.08485057204961777, "step": 4444 }, { "epoch": 1.1539460020768433, "grad_norm": 0.6148130893707275, "learning_rate": 2.302699447752665e-06, "loss": 0.10984544456005096, "step": 4445 }, { "epoch": 1.1542056074766356, "grad_norm": 0.8261554837226868, "learning_rate": 2.3017723600726795e-06, "loss": 0.10866548866033554, "step": 4446 }, { "epoch": 1.154465212876428, "grad_norm": 0.8562046885490417, "learning_rate": 2.3008453516290756e-06, "loss": 0.11516965925693512, "step": 4447 }, { "epoch": 1.1547248182762202, "grad_norm": 0.6787254810333252, "learning_rate": 2.29991842258573e-06, "loss": 0.09684920310974121, "step": 4448 }, { "epoch": 1.1549844236760125, "grad_norm": 1.2738057374954224, "learning_rate": 2.2989915731065023e-06, "loss": 0.08945205807685852, "step": 4449 }, { "epoch": 1.1552440290758048, "grad_norm": 0.7266321778297424, "learning_rate": 2.2980648033552384e-06, "loss": 0.10844144225120544, "step": 4450 }, { "epoch": 1.1555036344755971, "grad_norm": 2.0603013038635254, "learning_rate": 2.297138113495768e-06, "loss": 0.09008599817752838, "step": 4451 }, { "epoch": 1.1557632398753894, "grad_norm": 1.7090957164764404, "learning_rate": 2.2962115036919095e-06, "loss": 0.10776250064373016, "step": 4452 }, { "epoch": 1.1560228452751817, "grad_norm": 1.4187445640563965, "learning_rate": 2.295284974107468e-06, "loss": 0.10981941223144531, "step": 4453 }, { "epoch": 1.156282450674974, "grad_norm": 1.336417555809021, "learning_rate": 2.2943585249062316e-06, "loss": 0.09881135821342468, "step": 4454 }, { "epoch": 1.1565420560747663, "grad_norm": 0.739176869392395, "learning_rate": 2.2934321562519746e-06, "loss": 0.10936884582042694, "step": 4455 }, { "epoch": 1.1568016614745587, "grad_norm": 1.214894413948059, "learning_rate": 2.2925058683084582e-06, "loss": 0.12733125686645508, "step": 4456 }, { "epoch": 1.157061266874351, "grad_norm": 1.9969227313995361, "learning_rate": 2.291579661239429e-06, "loss": 0.09378641098737717, "step": 4457 }, { "epoch": 1.1573208722741433, "grad_norm": 1.896351933479309, "learning_rate": 2.2906535352086205e-06, "loss": 0.11540496349334717, "step": 4458 }, { "epoch": 1.1575804776739356, "grad_norm": 1.8922035694122314, "learning_rate": 2.2897274903797484e-06, "loss": 0.11905567348003387, "step": 4459 }, { "epoch": 1.1578400830737279, "grad_norm": 0.6693663597106934, "learning_rate": 2.2888015269165166e-06, "loss": 0.1080785021185875, "step": 4460 }, { "epoch": 1.1580996884735202, "grad_norm": 0.6831535696983337, "learning_rate": 2.2878756449826157e-06, "loss": 0.11392877995967865, "step": 4461 }, { "epoch": 1.1583592938733125, "grad_norm": 1.949723720550537, "learning_rate": 2.2869498447417193e-06, "loss": 0.09024390578269958, "step": 4462 }, { "epoch": 1.1586188992731048, "grad_norm": 1.3985058069229126, "learning_rate": 2.286024126357488e-06, "loss": 0.07520286738872528, "step": 4463 }, { "epoch": 1.158878504672897, "grad_norm": 1.6582635641098022, "learning_rate": 2.285098489993567e-06, "loss": 0.11482509970664978, "step": 4464 }, { "epoch": 1.1591381100726896, "grad_norm": 1.615842580795288, "learning_rate": 2.2841729358135885e-06, "loss": 0.11107946932315826, "step": 4465 }, { "epoch": 1.159397715472482, "grad_norm": 0.6240106821060181, "learning_rate": 2.2832474639811686e-06, "loss": 0.11452606320381165, "step": 4466 }, { "epoch": 1.1596573208722742, "grad_norm": 0.7649405598640442, "learning_rate": 2.28232207465991e-06, "loss": 0.08452245593070984, "step": 4467 }, { "epoch": 1.1599169262720666, "grad_norm": 2.7393271923065186, "learning_rate": 2.281396768013401e-06, "loss": 0.0799327865242958, "step": 4468 }, { "epoch": 1.1601765316718589, "grad_norm": 0.9167266488075256, "learning_rate": 2.280471544205213e-06, "loss": 0.10893204063177109, "step": 4469 }, { "epoch": 1.1604361370716512, "grad_norm": 0.9733408689498901, "learning_rate": 2.279546403398905e-06, "loss": 0.11285163462162018, "step": 4470 }, { "epoch": 1.1606957424714435, "grad_norm": 0.8560842871665955, "learning_rate": 2.2786213457580216e-06, "loss": 0.09485793113708496, "step": 4471 }, { "epoch": 1.1609553478712358, "grad_norm": 0.9873431921005249, "learning_rate": 2.27769637144609e-06, "loss": 0.12114407867193222, "step": 4472 }, { "epoch": 1.161214953271028, "grad_norm": 1.386628270149231, "learning_rate": 2.276771480626627e-06, "loss": 0.11368279159069061, "step": 4473 }, { "epoch": 1.1614745586708204, "grad_norm": 0.7996946573257446, "learning_rate": 2.27584667346313e-06, "loss": 0.11064739525318146, "step": 4474 }, { "epoch": 1.1617341640706127, "grad_norm": 1.2197047472000122, "learning_rate": 2.2749219501190846e-06, "loss": 0.0908200815320015, "step": 4475 }, { "epoch": 1.161993769470405, "grad_norm": 0.6570955514907837, "learning_rate": 2.2739973107579606e-06, "loss": 0.10490559041500092, "step": 4476 }, { "epoch": 1.1622533748701973, "grad_norm": 1.7936410903930664, "learning_rate": 2.2730727555432135e-06, "loss": 0.11911464482545853, "step": 4477 }, { "epoch": 1.1625129802699896, "grad_norm": 0.9180226922035217, "learning_rate": 2.272148284638282e-06, "loss": 0.11575636267662048, "step": 4478 }, { "epoch": 1.162772585669782, "grad_norm": 0.7123128175735474, "learning_rate": 2.271223898206593e-06, "loss": 0.0960497111082077, "step": 4479 }, { "epoch": 1.1630321910695742, "grad_norm": 0.586491584777832, "learning_rate": 2.270299596411556e-06, "loss": 0.11208892613649368, "step": 4480 }, { "epoch": 1.1632917964693665, "grad_norm": 1.538055181503296, "learning_rate": 2.269375379416568e-06, "loss": 0.09721849858760834, "step": 4481 }, { "epoch": 1.1635514018691588, "grad_norm": 0.7398699522018433, "learning_rate": 2.268451247385007e-06, "loss": 0.1067410409450531, "step": 4482 }, { "epoch": 1.1638110072689511, "grad_norm": 0.8009963035583496, "learning_rate": 2.2675272004802383e-06, "loss": 0.1080336943268776, "step": 4483 }, { "epoch": 1.1640706126687435, "grad_norm": 1.2811135053634644, "learning_rate": 2.2666032388656144e-06, "loss": 0.05376094952225685, "step": 4484 }, { "epoch": 1.1643302180685358, "grad_norm": 1.56028151512146, "learning_rate": 2.265679362704471e-06, "loss": 0.07298676669597626, "step": 4485 }, { "epoch": 1.164589823468328, "grad_norm": 1.817647933959961, "learning_rate": 2.264755572160125e-06, "loss": 0.09976334869861603, "step": 4486 }, { "epoch": 1.1648494288681204, "grad_norm": 2.1135239601135254, "learning_rate": 2.2638318673958836e-06, "loss": 0.09702073037624359, "step": 4487 }, { "epoch": 1.1651090342679127, "grad_norm": 1.7669038772583008, "learning_rate": 2.262908248575036e-06, "loss": 0.10990141332149506, "step": 4488 }, { "epoch": 1.165368639667705, "grad_norm": 1.7190380096435547, "learning_rate": 2.261984715860858e-06, "loss": 0.09071695804595947, "step": 4489 }, { "epoch": 1.1656282450674973, "grad_norm": 0.748201847076416, "learning_rate": 2.261061269416608e-06, "loss": 0.10783831775188446, "step": 4490 }, { "epoch": 1.1658878504672896, "grad_norm": 2.0629944801330566, "learning_rate": 2.2601379094055297e-06, "loss": 0.07397736608982086, "step": 4491 }, { "epoch": 1.1661474558670821, "grad_norm": 0.744611382484436, "learning_rate": 2.2592146359908534e-06, "loss": 0.09624417126178741, "step": 4492 }, { "epoch": 1.1664070612668744, "grad_norm": 1.2170385122299194, "learning_rate": 2.2582914493357912e-06, "loss": 0.1161261647939682, "step": 4493 }, { "epoch": 1.1666666666666667, "grad_norm": 1.673237681388855, "learning_rate": 2.2573683496035424e-06, "loss": 0.10106544196605682, "step": 4494 }, { "epoch": 1.166926272066459, "grad_norm": 0.8249514102935791, "learning_rate": 2.2564453369572898e-06, "loss": 0.11230581998825073, "step": 4495 }, { "epoch": 1.1671858774662514, "grad_norm": 0.7458687424659729, "learning_rate": 2.255522411560201e-06, "loss": 0.10154588520526886, "step": 4496 }, { "epoch": 1.1674454828660437, "grad_norm": 2.7465076446533203, "learning_rate": 2.254599573575427e-06, "loss": 0.06827225536108017, "step": 4497 }, { "epoch": 1.167705088265836, "grad_norm": 1.601072907447815, "learning_rate": 2.2536768231661053e-06, "loss": 0.09017512202262878, "step": 4498 }, { "epoch": 1.1679646936656283, "grad_norm": 1.946385145187378, "learning_rate": 2.252754160495357e-06, "loss": 0.07365120202302933, "step": 4499 }, { "epoch": 1.1682242990654206, "grad_norm": 2.251133441925049, "learning_rate": 2.251831585726289e-06, "loss": 0.11808853596448898, "step": 4500 }, { "epoch": 1.1684839044652129, "grad_norm": 0.8090120553970337, "learning_rate": 2.250909099021989e-06, "loss": 0.09503654390573502, "step": 4501 }, { "epoch": 1.1687435098650052, "grad_norm": 1.0404261350631714, "learning_rate": 2.249986700545533e-06, "loss": 0.10424233973026276, "step": 4502 }, { "epoch": 1.1690031152647975, "grad_norm": 1.6186777353286743, "learning_rate": 2.2490643904599787e-06, "loss": 0.11617693305015564, "step": 4503 }, { "epoch": 1.1692627206645898, "grad_norm": 0.8573946952819824, "learning_rate": 2.248142168928371e-06, "loss": 0.09889966994524002, "step": 4504 }, { "epoch": 1.1695223260643821, "grad_norm": 1.720043420791626, "learning_rate": 2.247220036113737e-06, "loss": 0.13214895129203796, "step": 4505 }, { "epoch": 1.1697819314641744, "grad_norm": 1.0888609886169434, "learning_rate": 2.2462979921790877e-06, "loss": 0.09651922434568405, "step": 4506 }, { "epoch": 1.1700415368639667, "grad_norm": 1.2964900732040405, "learning_rate": 2.2453760372874206e-06, "loss": 0.08334176987409592, "step": 4507 }, { "epoch": 1.170301142263759, "grad_norm": 1.5538713932037354, "learning_rate": 2.2444541716017167e-06, "loss": 0.09117549657821655, "step": 4508 }, { "epoch": 1.1705607476635513, "grad_norm": 0.7009642124176025, "learning_rate": 2.243532395284938e-06, "loss": 0.09216906130313873, "step": 4509 }, { "epoch": 1.1708203530633436, "grad_norm": 2.306175947189331, "learning_rate": 2.242610708500036e-06, "loss": 0.08139500021934509, "step": 4510 }, { "epoch": 1.171079958463136, "grad_norm": 1.6017030477523804, "learning_rate": 2.2416891114099433e-06, "loss": 0.1064191460609436, "step": 4511 }, { "epoch": 1.1713395638629283, "grad_norm": 1.615086555480957, "learning_rate": 2.240767604177577e-06, "loss": 0.11800646036863327, "step": 4512 }, { "epoch": 1.1715991692627208, "grad_norm": 1.1447829008102417, "learning_rate": 2.2398461869658384e-06, "loss": 0.10091876983642578, "step": 4513 }, { "epoch": 1.171858774662513, "grad_norm": 1.8664590120315552, "learning_rate": 2.2389248599376125e-06, "loss": 0.0952034741640091, "step": 4514 }, { "epoch": 1.1721183800623054, "grad_norm": 1.1480611562728882, "learning_rate": 2.238003623255769e-06, "loss": 0.10915690660476685, "step": 4515 }, { "epoch": 1.1723779854620977, "grad_norm": 1.883715033531189, "learning_rate": 2.2370824770831634e-06, "loss": 0.08083972334861755, "step": 4516 }, { "epoch": 1.17263759086189, "grad_norm": 1.2012273073196411, "learning_rate": 2.23616142158263e-06, "loss": 0.10354193300008774, "step": 4517 }, { "epoch": 1.1728971962616823, "grad_norm": 0.7049447298049927, "learning_rate": 2.2352404569169924e-06, "loss": 0.11784628033638, "step": 4518 }, { "epoch": 1.1731568016614746, "grad_norm": 0.8055465221405029, "learning_rate": 2.234319583249056e-06, "loss": 0.10209926962852478, "step": 4519 }, { "epoch": 1.173416407061267, "grad_norm": 1.707722544670105, "learning_rate": 2.233398800741609e-06, "loss": 0.08764398843050003, "step": 4520 }, { "epoch": 1.1736760124610592, "grad_norm": 1.8956999778747559, "learning_rate": 2.2324781095574255e-06, "loss": 0.1485287845134735, "step": 4521 }, { "epoch": 1.1739356178608515, "grad_norm": 1.2793806791305542, "learning_rate": 2.2315575098592627e-06, "loss": 0.09874873608350754, "step": 4522 }, { "epoch": 1.1741952232606438, "grad_norm": 1.7851461172103882, "learning_rate": 2.2306370018098623e-06, "loss": 0.13141164183616638, "step": 4523 }, { "epoch": 1.1744548286604362, "grad_norm": 1.6286113262176514, "learning_rate": 2.229716585571947e-06, "loss": 0.08224242925643921, "step": 4524 }, { "epoch": 1.1747144340602285, "grad_norm": 1.69109046459198, "learning_rate": 2.228796261308227e-06, "loss": 0.07860930263996124, "step": 4525 }, { "epoch": 1.1749740394600208, "grad_norm": 1.4658979177474976, "learning_rate": 2.2278760291813934e-06, "loss": 0.07354763150215149, "step": 4526 }, { "epoch": 1.175233644859813, "grad_norm": 1.6246483325958252, "learning_rate": 2.2269558893541237e-06, "loss": 0.08724162727594376, "step": 4527 }, { "epoch": 1.1754932502596054, "grad_norm": 2.2569379806518555, "learning_rate": 2.226035841989076e-06, "loss": 0.07161727547645569, "step": 4528 }, { "epoch": 1.1757528556593977, "grad_norm": 1.2339500188827515, "learning_rate": 2.225115887248894e-06, "loss": 0.11346692591905594, "step": 4529 }, { "epoch": 1.17601246105919, "grad_norm": 0.800662636756897, "learning_rate": 2.2241960252962054e-06, "loss": 0.10553543269634247, "step": 4530 }, { "epoch": 1.1762720664589823, "grad_norm": 1.6647140979766846, "learning_rate": 2.2232762562936204e-06, "loss": 0.0781160518527031, "step": 4531 }, { "epoch": 1.1765316718587746, "grad_norm": 0.6399361491203308, "learning_rate": 2.222356580403733e-06, "loss": 0.1105036586523056, "step": 4532 }, { "epoch": 1.176791277258567, "grad_norm": 0.6423635482788086, "learning_rate": 2.22143699778912e-06, "loss": 0.11442184448242188, "step": 4533 }, { "epoch": 1.1770508826583592, "grad_norm": 1.0192184448242188, "learning_rate": 2.2205175086123436e-06, "loss": 0.08724924921989441, "step": 4534 }, { "epoch": 1.1773104880581515, "grad_norm": 1.828208088874817, "learning_rate": 2.2195981130359482e-06, "loss": 0.11748029291629791, "step": 4535 }, { "epoch": 1.1775700934579438, "grad_norm": 1.8113712072372437, "learning_rate": 2.218678811222462e-06, "loss": 0.09370589256286621, "step": 4536 }, { "epoch": 1.1778296988577361, "grad_norm": 1.3887523412704468, "learning_rate": 2.217759603334396e-06, "loss": 0.08453921228647232, "step": 4537 }, { "epoch": 1.1780893042575284, "grad_norm": 1.0540262460708618, "learning_rate": 2.2168404895342456e-06, "loss": 0.09391874819993973, "step": 4538 }, { "epoch": 1.1783489096573208, "grad_norm": 2.5701489448547363, "learning_rate": 2.2159214699844895e-06, "loss": 0.09543560445308685, "step": 4539 }, { "epoch": 1.1786085150571133, "grad_norm": 2.2647128105163574, "learning_rate": 2.2150025448475876e-06, "loss": 0.08786055445671082, "step": 4540 }, { "epoch": 1.1788681204569056, "grad_norm": 1.1654201745986938, "learning_rate": 2.214083714285987e-06, "loss": 0.10884980857372284, "step": 4541 }, { "epoch": 1.179127725856698, "grad_norm": 1.5665236711502075, "learning_rate": 2.213164978462114e-06, "loss": 0.10418960452079773, "step": 4542 }, { "epoch": 1.1793873312564902, "grad_norm": 1.686216950416565, "learning_rate": 2.212246337538382e-06, "loss": 0.10485321283340454, "step": 4543 }, { "epoch": 1.1796469366562825, "grad_norm": 1.3227896690368652, "learning_rate": 2.211327791677183e-06, "loss": 0.09104688465595245, "step": 4544 }, { "epoch": 1.1799065420560748, "grad_norm": 1.7414140701293945, "learning_rate": 2.2104093410408972e-06, "loss": 0.11898189038038254, "step": 4545 }, { "epoch": 1.1801661474558671, "grad_norm": 0.933846116065979, "learning_rate": 2.209490985791886e-06, "loss": 0.09492182731628418, "step": 4546 }, { "epoch": 1.1804257528556594, "grad_norm": 1.1116673946380615, "learning_rate": 2.2085727260924916e-06, "loss": 0.12304575741291046, "step": 4547 }, { "epoch": 1.1806853582554517, "grad_norm": 1.3156713247299194, "learning_rate": 2.2076545621050416e-06, "loss": 0.09966187179088593, "step": 4548 }, { "epoch": 1.180944963655244, "grad_norm": 1.1341817378997803, "learning_rate": 2.2067364939918473e-06, "loss": 0.10846571624279022, "step": 4549 }, { "epoch": 1.1812045690550363, "grad_norm": 1.178478479385376, "learning_rate": 2.205818521915203e-06, "loss": 0.10577991604804993, "step": 4550 }, { "epoch": 1.1814641744548287, "grad_norm": 1.2523448467254639, "learning_rate": 2.2049006460373824e-06, "loss": 0.116240955889225, "step": 4551 }, { "epoch": 1.181723779854621, "grad_norm": 1.3691270351409912, "learning_rate": 2.203982866520647e-06, "loss": 0.10624746978282928, "step": 4552 }, { "epoch": 1.1819833852544133, "grad_norm": 0.7811548709869385, "learning_rate": 2.2030651835272383e-06, "loss": 0.098729208111763, "step": 4553 }, { "epoch": 1.1822429906542056, "grad_norm": 1.7577064037322998, "learning_rate": 2.202147597219383e-06, "loss": 0.09117984771728516, "step": 4554 }, { "epoch": 1.1825025960539979, "grad_norm": 1.4211246967315674, "learning_rate": 2.2012301077592874e-06, "loss": 0.08507753908634186, "step": 4555 }, { "epoch": 1.1827622014537902, "grad_norm": 1.4734938144683838, "learning_rate": 2.2003127153091437e-06, "loss": 0.12969642877578735, "step": 4556 }, { "epoch": 1.1830218068535825, "grad_norm": 1.9892948865890503, "learning_rate": 2.199395420031126e-06, "loss": 0.07992177456617355, "step": 4557 }, { "epoch": 1.1832814122533748, "grad_norm": 1.4018914699554443, "learning_rate": 2.198478222087391e-06, "loss": 0.10439097881317139, "step": 4558 }, { "epoch": 1.183541017653167, "grad_norm": 2.1544787883758545, "learning_rate": 2.1975611216400776e-06, "loss": 0.10294058173894882, "step": 4559 }, { "epoch": 1.1838006230529594, "grad_norm": 2.323665142059326, "learning_rate": 2.196644118851309e-06, "loss": 0.08285488933324814, "step": 4560 }, { "epoch": 1.184060228452752, "grad_norm": 0.9459753632545471, "learning_rate": 2.19572721388319e-06, "loss": 0.1228240355849266, "step": 4561 }, { "epoch": 1.1843198338525442, "grad_norm": 1.958260178565979, "learning_rate": 2.194810406897809e-06, "loss": 0.08624400198459625, "step": 4562 }, { "epoch": 1.1845794392523366, "grad_norm": 1.5099916458129883, "learning_rate": 2.193893698057235e-06, "loss": 0.11623266339302063, "step": 4563 }, { "epoch": 1.1848390446521289, "grad_norm": 2.046125888824463, "learning_rate": 2.1929770875235227e-06, "loss": 0.12504702806472778, "step": 4564 }, { "epoch": 1.1850986500519212, "grad_norm": 1.958598017692566, "learning_rate": 2.1920605754587073e-06, "loss": 0.06815354526042938, "step": 4565 }, { "epoch": 1.1853582554517135, "grad_norm": 1.8391612768173218, "learning_rate": 2.1911441620248066e-06, "loss": 0.084981270134449, "step": 4566 }, { "epoch": 1.1856178608515058, "grad_norm": 2.0087547302246094, "learning_rate": 2.1902278473838217e-06, "loss": 0.11607597023248672, "step": 4567 }, { "epoch": 1.185877466251298, "grad_norm": 2.7985475063323975, "learning_rate": 2.1893116316977365e-06, "loss": 0.06387986242771149, "step": 4568 }, { "epoch": 1.1861370716510904, "grad_norm": 0.8014139533042908, "learning_rate": 2.188395515128517e-06, "loss": 0.10433632135391235, "step": 4569 }, { "epoch": 1.1863966770508827, "grad_norm": 1.886704444885254, "learning_rate": 2.1874794978381114e-06, "loss": 0.06173597648739815, "step": 4570 }, { "epoch": 1.186656282450675, "grad_norm": 0.8855161666870117, "learning_rate": 2.1865635799884507e-06, "loss": 0.10149721801280975, "step": 4571 }, { "epoch": 1.1869158878504673, "grad_norm": 1.7222459316253662, "learning_rate": 2.185647761741447e-06, "loss": 0.0890849232673645, "step": 4572 }, { "epoch": 1.1871754932502596, "grad_norm": 1.641635775566101, "learning_rate": 2.184732043258998e-06, "loss": 0.0904659777879715, "step": 4573 }, { "epoch": 1.187435098650052, "grad_norm": 2.0457122325897217, "learning_rate": 2.183816424702982e-06, "loss": 0.10589152574539185, "step": 4574 }, { "epoch": 1.1876947040498442, "grad_norm": 1.9274945259094238, "learning_rate": 2.1829009062352558e-06, "loss": 0.08868300914764404, "step": 4575 }, { "epoch": 1.1879543094496365, "grad_norm": 3.208592176437378, "learning_rate": 2.181985488017666e-06, "loss": 0.08068109303712845, "step": 4576 }, { "epoch": 1.1882139148494288, "grad_norm": 1.5592719316482544, "learning_rate": 2.1810701702120367e-06, "loss": 0.10063393414020538, "step": 4577 }, { "epoch": 1.1884735202492211, "grad_norm": 1.883528470993042, "learning_rate": 2.180154952980173e-06, "loss": 0.09664037823677063, "step": 4578 }, { "epoch": 1.1887331256490135, "grad_norm": 1.8973339796066284, "learning_rate": 2.179239836483866e-06, "loss": 0.11084957420825958, "step": 4579 }, { "epoch": 1.1889927310488058, "grad_norm": 1.927881121635437, "learning_rate": 2.178324820884888e-06, "loss": 0.10123427212238312, "step": 4580 }, { "epoch": 1.189252336448598, "grad_norm": 3.21882963180542, "learning_rate": 2.177409906344992e-06, "loss": 0.09936277568340302, "step": 4581 }, { "epoch": 1.1895119418483904, "grad_norm": 0.6881288886070251, "learning_rate": 2.1764950930259143e-06, "loss": 0.103635273873806, "step": 4582 }, { "epoch": 1.1897715472481827, "grad_norm": 1.8106884956359863, "learning_rate": 2.1755803810893716e-06, "loss": 0.10801127552986145, "step": 4583 }, { "epoch": 1.190031152647975, "grad_norm": 1.3857372999191284, "learning_rate": 2.174665770697066e-06, "loss": 0.091135174036026, "step": 4584 }, { "epoch": 1.1902907580477673, "grad_norm": 2.7569005489349365, "learning_rate": 2.1737512620106784e-06, "loss": 0.13189171254634857, "step": 4585 }, { "epoch": 1.1905503634475596, "grad_norm": 1.1624959707260132, "learning_rate": 2.1728368551918726e-06, "loss": 0.10336282104253769, "step": 4586 }, { "epoch": 1.190809968847352, "grad_norm": 2.403596878051758, "learning_rate": 2.1719225504022968e-06, "loss": 0.09956235438585281, "step": 4587 }, { "epoch": 1.1910695742471444, "grad_norm": 1.844178318977356, "learning_rate": 2.1710083478035763e-06, "loss": 0.09318724274635315, "step": 4588 }, { "epoch": 1.1913291796469367, "grad_norm": 1.8054423332214355, "learning_rate": 2.1700942475573246e-06, "loss": 0.08002166450023651, "step": 4589 }, { "epoch": 1.191588785046729, "grad_norm": 1.6876448392868042, "learning_rate": 2.169180249825131e-06, "loss": 0.10878857970237732, "step": 4590 }, { "epoch": 1.1918483904465214, "grad_norm": 1.7541292905807495, "learning_rate": 2.1682663547685695e-06, "loss": 0.08645392954349518, "step": 4591 }, { "epoch": 1.1921079958463137, "grad_norm": 1.2176649570465088, "learning_rate": 2.167352562549197e-06, "loss": 0.11966592818498611, "step": 4592 }, { "epoch": 1.192367601246106, "grad_norm": 2.0416362285614014, "learning_rate": 2.1664388733285514e-06, "loss": 0.11322584748268127, "step": 4593 }, { "epoch": 1.1926272066458983, "grad_norm": 2.3450489044189453, "learning_rate": 2.1655252872681505e-06, "loss": 0.11405561864376068, "step": 4594 }, { "epoch": 1.1928868120456906, "grad_norm": 1.342619776725769, "learning_rate": 2.164611804529496e-06, "loss": 0.10744907706975937, "step": 4595 }, { "epoch": 1.1931464174454829, "grad_norm": 2.237339735031128, "learning_rate": 2.1636984252740705e-06, "loss": 0.07999907433986664, "step": 4596 }, { "epoch": 1.1934060228452752, "grad_norm": 1.4323488473892212, "learning_rate": 2.16278514966334e-06, "loss": 0.08223532140254974, "step": 4597 }, { "epoch": 1.1936656282450675, "grad_norm": 0.957847535610199, "learning_rate": 2.161871977858749e-06, "loss": 0.0950506255030632, "step": 4598 }, { "epoch": 1.1939252336448598, "grad_norm": 1.043179988861084, "learning_rate": 2.160958910021726e-06, "loss": 0.10989300161600113, "step": 4599 }, { "epoch": 1.1941848390446521, "grad_norm": 0.7816504240036011, "learning_rate": 2.16004594631368e-06, "loss": 0.10038349777460098, "step": 4600 }, { "epoch": 1.1944444444444444, "grad_norm": 1.285890817642212, "learning_rate": 2.1591330868960035e-06, "loss": 0.11116322875022888, "step": 4601 }, { "epoch": 1.1947040498442367, "grad_norm": 0.953270673751831, "learning_rate": 2.1582203319300677e-06, "loss": 0.10455237329006195, "step": 4602 }, { "epoch": 1.194963655244029, "grad_norm": 1.9557188749313354, "learning_rate": 2.157307681577227e-06, "loss": 0.09799829125404358, "step": 4603 }, { "epoch": 1.1952232606438213, "grad_norm": 1.3643156290054321, "learning_rate": 2.1563951359988183e-06, "loss": 0.10959364473819733, "step": 4604 }, { "epoch": 1.1954828660436136, "grad_norm": 1.3346564769744873, "learning_rate": 2.155482695356157e-06, "loss": 0.1046954095363617, "step": 4605 }, { "epoch": 1.195742471443406, "grad_norm": 1.7466199398040771, "learning_rate": 2.154570359810542e-06, "loss": 0.08167217671871185, "step": 4606 }, { "epoch": 1.1960020768431983, "grad_norm": 2.3000288009643555, "learning_rate": 2.1536581295232545e-06, "loss": 0.10098659247159958, "step": 4607 }, { "epoch": 1.1962616822429906, "grad_norm": 0.752533495426178, "learning_rate": 2.152746004655556e-06, "loss": 0.09821946918964386, "step": 4608 }, { "epoch": 1.196521287642783, "grad_norm": 2.0280566215515137, "learning_rate": 2.1518339853686874e-06, "loss": 0.10151466727256775, "step": 4609 }, { "epoch": 1.1967808930425754, "grad_norm": 1.4542518854141235, "learning_rate": 2.1509220718238744e-06, "loss": 0.10846341401338577, "step": 4610 }, { "epoch": 1.1970404984423677, "grad_norm": 2.0204944610595703, "learning_rate": 2.150010264182322e-06, "loss": 0.11946148425340652, "step": 4611 }, { "epoch": 1.19730010384216, "grad_norm": 1.2656090259552002, "learning_rate": 2.1490985626052173e-06, "loss": 0.10867127031087875, "step": 4612 }, { "epoch": 1.1975597092419523, "grad_norm": 1.7470619678497314, "learning_rate": 2.1481869672537277e-06, "loss": 0.07704995572566986, "step": 4613 }, { "epoch": 1.1978193146417446, "grad_norm": 1.6200016736984253, "learning_rate": 2.147275478289002e-06, "loss": 0.08369237929582596, "step": 4614 }, { "epoch": 1.198078920041537, "grad_norm": 1.496957778930664, "learning_rate": 2.1463640958721714e-06, "loss": 0.10507091879844666, "step": 4615 }, { "epoch": 1.1983385254413292, "grad_norm": 1.9776804447174072, "learning_rate": 2.145452820164348e-06, "loss": 0.08693476021289825, "step": 4616 }, { "epoch": 1.1985981308411215, "grad_norm": 1.1399797201156616, "learning_rate": 2.144541651326623e-06, "loss": 0.08932490646839142, "step": 4617 }, { "epoch": 1.1988577362409139, "grad_norm": 0.86251300573349, "learning_rate": 2.1436305895200715e-06, "loss": 0.09549067169427872, "step": 4618 }, { "epoch": 1.1991173416407062, "grad_norm": 2.315358877182007, "learning_rate": 2.142719634905747e-06, "loss": 0.0877697542309761, "step": 4619 }, { "epoch": 1.1993769470404985, "grad_norm": 0.9084955453872681, "learning_rate": 2.1418087876446867e-06, "loss": 0.11935747414827347, "step": 4620 }, { "epoch": 1.1996365524402908, "grad_norm": 1.9996752738952637, "learning_rate": 2.140898047897907e-06, "loss": 0.1126428097486496, "step": 4621 }, { "epoch": 1.199896157840083, "grad_norm": 0.6456223130226135, "learning_rate": 2.139987415826405e-06, "loss": 0.10350703448057175, "step": 4622 }, { "epoch": 1.2001557632398754, "grad_norm": 1.0546895265579224, "learning_rate": 2.1390768915911614e-06, "loss": 0.09036680310964584, "step": 4623 }, { "epoch": 1.2004153686396677, "grad_norm": 1.3810253143310547, "learning_rate": 2.138166475353135e-06, "loss": 0.09677884727716446, "step": 4624 }, { "epoch": 1.20067497403946, "grad_norm": 0.9361253380775452, "learning_rate": 2.137256167273266e-06, "loss": 0.10386759042739868, "step": 4625 }, { "epoch": 1.2009345794392523, "grad_norm": 1.116526484489441, "learning_rate": 2.136345967512477e-06, "loss": 0.11088588833808899, "step": 4626 }, { "epoch": 1.2011941848390446, "grad_norm": 0.9425929188728333, "learning_rate": 2.1354358762316703e-06, "loss": 0.09682779759168625, "step": 4627 }, { "epoch": 1.201453790238837, "grad_norm": 1.903933048248291, "learning_rate": 2.134525893591729e-06, "loss": 0.1222035139799118, "step": 4628 }, { "epoch": 1.2017133956386292, "grad_norm": 0.8552073836326599, "learning_rate": 2.133616019753517e-06, "loss": 0.10655969381332397, "step": 4629 }, { "epoch": 1.2019730010384215, "grad_norm": 0.7090772390365601, "learning_rate": 2.1327062548778797e-06, "loss": 0.095007985830307, "step": 4630 }, { "epoch": 1.2022326064382138, "grad_norm": 0.8879538178443909, "learning_rate": 2.131796599125643e-06, "loss": 0.08314165472984314, "step": 4631 }, { "epoch": 1.2024922118380061, "grad_norm": 1.1159650087356567, "learning_rate": 2.1308870526576112e-06, "loss": 0.09195791929960251, "step": 4632 }, { "epoch": 1.2027518172377984, "grad_norm": 0.9138805866241455, "learning_rate": 2.129977615634573e-06, "loss": 0.1154407486319542, "step": 4633 }, { "epoch": 1.2030114226375908, "grad_norm": 2.752368688583374, "learning_rate": 2.129068288217296e-06, "loss": 0.11749128997325897, "step": 4634 }, { "epoch": 1.203271028037383, "grad_norm": 2.35263991355896, "learning_rate": 2.128159070566529e-06, "loss": 0.10973730683326721, "step": 4635 }, { "epoch": 1.2035306334371756, "grad_norm": 1.5420446395874023, "learning_rate": 2.1272499628429993e-06, "loss": 0.1153005063533783, "step": 4636 }, { "epoch": 1.203790238836968, "grad_norm": 0.7174854874610901, "learning_rate": 2.1263409652074165e-06, "loss": 0.10569089651107788, "step": 4637 }, { "epoch": 1.2040498442367602, "grad_norm": 1.747840166091919, "learning_rate": 2.125432077820471e-06, "loss": 0.10533298552036285, "step": 4638 }, { "epoch": 1.2043094496365525, "grad_norm": 1.5858139991760254, "learning_rate": 2.1245233008428355e-06, "loss": 0.07963555306196213, "step": 4639 }, { "epoch": 1.2045690550363448, "grad_norm": 0.6683872938156128, "learning_rate": 2.123614634435158e-06, "loss": 0.10864847898483276, "step": 4640 }, { "epoch": 1.2048286604361371, "grad_norm": 0.863903820514679, "learning_rate": 2.1227060787580693e-06, "loss": 0.09673257917165756, "step": 4641 }, { "epoch": 1.2050882658359294, "grad_norm": 1.1786160469055176, "learning_rate": 2.1217976339721837e-06, "loss": 0.09195287525653839, "step": 4642 }, { "epoch": 1.2053478712357217, "grad_norm": 1.9643604755401611, "learning_rate": 2.120889300238093e-06, "loss": 0.08962178230285645, "step": 4643 }, { "epoch": 1.205607476635514, "grad_norm": 0.9738965630531311, "learning_rate": 2.1199810777163687e-06, "loss": 0.0905991941690445, "step": 4644 }, { "epoch": 1.2058670820353063, "grad_norm": 1.7965023517608643, "learning_rate": 2.1190729665675647e-06, "loss": 0.09035224467515945, "step": 4645 }, { "epoch": 1.2061266874350987, "grad_norm": 1.176874041557312, "learning_rate": 2.1181649669522138e-06, "loss": 0.10880941152572632, "step": 4646 }, { "epoch": 1.206386292834891, "grad_norm": 2.1584484577178955, "learning_rate": 2.1172570790308293e-06, "loss": 0.10655874013900757, "step": 4647 }, { "epoch": 1.2066458982346833, "grad_norm": 2.3585572242736816, "learning_rate": 2.1163493029639056e-06, "loss": 0.09437733888626099, "step": 4648 }, { "epoch": 1.2069055036344756, "grad_norm": 1.7597637176513672, "learning_rate": 2.1154416389119163e-06, "loss": 0.1082482635974884, "step": 4649 }, { "epoch": 1.2071651090342679, "grad_norm": 1.6744509935379028, "learning_rate": 2.114534087035316e-06, "loss": 0.13793742656707764, "step": 4650 }, { "epoch": 1.2074247144340602, "grad_norm": 0.8074899911880493, "learning_rate": 2.1136266474945387e-06, "loss": 0.10865800827741623, "step": 4651 }, { "epoch": 1.2076843198338525, "grad_norm": 1.581455945968628, "learning_rate": 2.112719320449999e-06, "loss": 0.08500590920448303, "step": 4652 }, { "epoch": 1.2079439252336448, "grad_norm": 1.2304021120071411, "learning_rate": 2.111812106062092e-06, "loss": 0.09842740744352341, "step": 4653 }, { "epoch": 1.208203530633437, "grad_norm": 1.6735141277313232, "learning_rate": 2.1109050044911922e-06, "loss": 0.09549102187156677, "step": 4654 }, { "epoch": 1.2084631360332294, "grad_norm": 1.8719223737716675, "learning_rate": 2.109998015897654e-06, "loss": 0.10116029530763626, "step": 4655 }, { "epoch": 1.2087227414330217, "grad_norm": 2.6185312271118164, "learning_rate": 2.1090911404418123e-06, "loss": 0.07591722160577774, "step": 4656 }, { "epoch": 1.208982346832814, "grad_norm": 2.6865334510803223, "learning_rate": 2.108184378283982e-06, "loss": 0.0932525247335434, "step": 4657 }, { "epoch": 1.2092419522326066, "grad_norm": 1.8123458623886108, "learning_rate": 2.107277729584459e-06, "loss": 0.09462239593267441, "step": 4658 }, { "epoch": 1.2095015576323989, "grad_norm": 0.9273370504379272, "learning_rate": 2.106371194503516e-06, "loss": 0.10822952538728714, "step": 4659 }, { "epoch": 1.2097611630321912, "grad_norm": 2.158878803253174, "learning_rate": 2.1054647732014088e-06, "loss": 0.1027936041355133, "step": 4660 }, { "epoch": 1.2100207684319835, "grad_norm": 1.403617024421692, "learning_rate": 2.1045584658383718e-06, "loss": 0.09301022440195084, "step": 4661 }, { "epoch": 1.2102803738317758, "grad_norm": 1.620278239250183, "learning_rate": 2.1036522725746207e-06, "loss": 0.07876865565776825, "step": 4662 }, { "epoch": 1.210539979231568, "grad_norm": 0.6039655208587646, "learning_rate": 2.102746193570346e-06, "loss": 0.10200108587741852, "step": 4663 }, { "epoch": 1.2107995846313604, "grad_norm": 2.0178418159484863, "learning_rate": 2.1018402289857254e-06, "loss": 0.08272634446620941, "step": 4664 }, { "epoch": 1.2110591900311527, "grad_norm": 1.558305025100708, "learning_rate": 2.1009343789809116e-06, "loss": 0.08992771804332733, "step": 4665 }, { "epoch": 1.211318795430945, "grad_norm": 1.561212182044983, "learning_rate": 2.1000286437160385e-06, "loss": 0.10950601100921631, "step": 4666 }, { "epoch": 1.2115784008307373, "grad_norm": 1.540473222732544, "learning_rate": 2.0991230233512177e-06, "loss": 0.09553079307079315, "step": 4667 }, { "epoch": 1.2118380062305296, "grad_norm": 1.4141393899917603, "learning_rate": 2.098217518046543e-06, "loss": 0.08240990340709686, "step": 4668 }, { "epoch": 1.212097611630322, "grad_norm": 0.9211397767066956, "learning_rate": 2.097312127962088e-06, "loss": 0.09342122822999954, "step": 4669 }, { "epoch": 1.2123572170301142, "grad_norm": 1.9294078350067139, "learning_rate": 2.0964068532579047e-06, "loss": 0.047900062054395676, "step": 4670 }, { "epoch": 1.2126168224299065, "grad_norm": 0.8332618474960327, "learning_rate": 2.095501694094024e-06, "loss": 0.11280281841754913, "step": 4671 }, { "epoch": 1.2128764278296988, "grad_norm": 1.7865148782730103, "learning_rate": 2.094596650630457e-06, "loss": 0.10666204243898392, "step": 4672 }, { "epoch": 1.2131360332294912, "grad_norm": 1.3066602945327759, "learning_rate": 2.093691723027196e-06, "loss": 0.10520625859498978, "step": 4673 }, { "epoch": 1.2133956386292835, "grad_norm": 1.6894140243530273, "learning_rate": 2.0927869114442118e-06, "loss": 0.09382510185241699, "step": 4674 }, { "epoch": 1.2136552440290758, "grad_norm": 1.9790555238723755, "learning_rate": 2.091882216041452e-06, "loss": 0.11567613482475281, "step": 4675 }, { "epoch": 1.213914849428868, "grad_norm": 0.75980144739151, "learning_rate": 2.090977636978848e-06, "loss": 0.10765508562326431, "step": 4676 }, { "epoch": 1.2141744548286604, "grad_norm": 1.3148119449615479, "learning_rate": 2.090073174416308e-06, "loss": 0.10215272009372711, "step": 4677 }, { "epoch": 1.2144340602284527, "grad_norm": 0.7467631101608276, "learning_rate": 2.089168828513722e-06, "loss": 0.09213908761739731, "step": 4678 }, { "epoch": 1.214693665628245, "grad_norm": 0.9681723713874817, "learning_rate": 2.0882645994309543e-06, "loss": 0.11211740970611572, "step": 4679 }, { "epoch": 1.2149532710280373, "grad_norm": 2.1656322479248047, "learning_rate": 2.0873604873278543e-06, "loss": 0.07154297828674316, "step": 4680 }, { "epoch": 1.2152128764278296, "grad_norm": 1.3516011238098145, "learning_rate": 2.086456492364247e-06, "loss": 0.10994870960712433, "step": 4681 }, { "epoch": 1.215472481827622, "grad_norm": 2.876793146133423, "learning_rate": 2.08555261469994e-06, "loss": 0.06926199048757553, "step": 4682 }, { "epoch": 1.2157320872274142, "grad_norm": 1.540960669517517, "learning_rate": 2.084648854494716e-06, "loss": 0.04313022643327713, "step": 4683 }, { "epoch": 1.2159916926272065, "grad_norm": 0.8928313851356506, "learning_rate": 2.08374521190834e-06, "loss": 0.1148916631937027, "step": 4684 }, { "epoch": 1.216251298026999, "grad_norm": 1.313925862312317, "learning_rate": 2.082841687100556e-06, "loss": 0.08824688196182251, "step": 4685 }, { "epoch": 1.2165109034267914, "grad_norm": 0.7481755018234253, "learning_rate": 2.0819382802310857e-06, "loss": 0.11139678955078125, "step": 4686 }, { "epoch": 1.2167705088265837, "grad_norm": 0.9774709343910217, "learning_rate": 2.08103499145963e-06, "loss": 0.10385673493146896, "step": 4687 }, { "epoch": 1.217030114226376, "grad_norm": 1.526374340057373, "learning_rate": 2.080131820945871e-06, "loss": 0.08085356652736664, "step": 4688 }, { "epoch": 1.2172897196261683, "grad_norm": 1.4202344417572021, "learning_rate": 2.0792287688494683e-06, "loss": 0.07181656360626221, "step": 4689 }, { "epoch": 1.2175493250259606, "grad_norm": 1.8349924087524414, "learning_rate": 2.0783258353300607e-06, "loss": 0.09350672364234924, "step": 4690 }, { "epoch": 1.217808930425753, "grad_norm": 0.6428411602973938, "learning_rate": 2.0774230205472657e-06, "loss": 0.09050796926021576, "step": 4691 }, { "epoch": 1.2180685358255452, "grad_norm": 1.8770372867584229, "learning_rate": 2.0765203246606806e-06, "loss": 0.04689274728298187, "step": 4692 }, { "epoch": 1.2183281412253375, "grad_norm": 1.4827790260314941, "learning_rate": 2.0756177478298824e-06, "loss": 0.1148410439491272, "step": 4693 }, { "epoch": 1.2185877466251298, "grad_norm": 1.4360389709472656, "learning_rate": 2.0747152902144243e-06, "loss": 0.07479144632816315, "step": 4694 }, { "epoch": 1.2188473520249221, "grad_norm": 1.0511682033538818, "learning_rate": 2.073812951973841e-06, "loss": 0.09781226515769958, "step": 4695 }, { "epoch": 1.2191069574247144, "grad_norm": 1.5068082809448242, "learning_rate": 2.072910733267645e-06, "loss": 0.10382319986820221, "step": 4696 }, { "epoch": 1.2193665628245067, "grad_norm": 1.3854299783706665, "learning_rate": 2.072008634255329e-06, "loss": 0.1001749336719513, "step": 4697 }, { "epoch": 1.219626168224299, "grad_norm": 1.6139576435089111, "learning_rate": 2.0711066550963615e-06, "loss": 0.09994262456893921, "step": 4698 }, { "epoch": 1.2198857736240913, "grad_norm": 1.6244022846221924, "learning_rate": 2.070204795950193e-06, "loss": 0.11738845705986023, "step": 4699 }, { "epoch": 1.2201453790238836, "grad_norm": 0.8226808905601501, "learning_rate": 2.069303056976251e-06, "loss": 0.11462894082069397, "step": 4700 }, { "epoch": 1.220404984423676, "grad_norm": 1.6951704025268555, "learning_rate": 2.0684014383339447e-06, "loss": 0.11291959881782532, "step": 4701 }, { "epoch": 1.2206645898234683, "grad_norm": 2.2280917167663574, "learning_rate": 2.0674999401826558e-06, "loss": 0.09589037299156189, "step": 4702 }, { "epoch": 1.2209241952232606, "grad_norm": 0.6539914011955261, "learning_rate": 2.0665985626817496e-06, "loss": 0.11172675341367722, "step": 4703 }, { "epoch": 1.2211838006230529, "grad_norm": 1.9385722875595093, "learning_rate": 2.06569730599057e-06, "loss": 0.08341071754693985, "step": 4704 }, { "epoch": 1.2214434060228452, "grad_norm": 0.5639216899871826, "learning_rate": 2.06479617026844e-06, "loss": 0.1072889044880867, "step": 4705 }, { "epoch": 1.2217030114226377, "grad_norm": 0.8622894287109375, "learning_rate": 2.063895155674657e-06, "loss": 0.08858449757099152, "step": 4706 }, { "epoch": 1.22196261682243, "grad_norm": 1.2791699171066284, "learning_rate": 2.0629942623685e-06, "loss": 0.07577704638242722, "step": 4707 }, { "epoch": 1.2222222222222223, "grad_norm": 2.9514150619506836, "learning_rate": 2.0620934905092273e-06, "loss": 0.08043460547924042, "step": 4708 }, { "epoch": 1.2224818276220146, "grad_norm": 1.760867714881897, "learning_rate": 2.0611928402560756e-06, "loss": 0.1135689914226532, "step": 4709 }, { "epoch": 1.222741433021807, "grad_norm": 1.083139419555664, "learning_rate": 2.0602923117682573e-06, "loss": 0.09614525735378265, "step": 4710 }, { "epoch": 1.2230010384215992, "grad_norm": 1.8714064359664917, "learning_rate": 2.059391905204966e-06, "loss": 0.1119903177022934, "step": 4711 }, { "epoch": 1.2232606438213915, "grad_norm": 0.7738422155380249, "learning_rate": 2.0584916207253736e-06, "loss": 0.11160793900489807, "step": 4712 }, { "epoch": 1.2235202492211839, "grad_norm": 1.2097479104995728, "learning_rate": 2.0575914584886285e-06, "loss": 0.10583512485027313, "step": 4713 }, { "epoch": 1.2237798546209762, "grad_norm": 1.2416579723358154, "learning_rate": 2.0566914186538596e-06, "loss": 0.09725228697061539, "step": 4714 }, { "epoch": 1.2240394600207685, "grad_norm": 1.914942979812622, "learning_rate": 2.0557915013801732e-06, "loss": 0.08582848310470581, "step": 4715 }, { "epoch": 1.2242990654205608, "grad_norm": 2.208847999572754, "learning_rate": 2.054891706826655e-06, "loss": 0.1131613627076149, "step": 4716 }, { "epoch": 1.224558670820353, "grad_norm": 0.8489353060722351, "learning_rate": 2.053992035152366e-06, "loss": 0.09513726830482483, "step": 4717 }, { "epoch": 1.2248182762201454, "grad_norm": 1.2893893718719482, "learning_rate": 2.053092486516349e-06, "loss": 0.10376983880996704, "step": 4718 }, { "epoch": 1.2250778816199377, "grad_norm": 1.16813325881958, "learning_rate": 2.0521930610776235e-06, "loss": 0.09997229278087616, "step": 4719 }, { "epoch": 1.22533748701973, "grad_norm": 1.525680661201477, "learning_rate": 2.0512937589951877e-06, "loss": 0.1099548488855362, "step": 4720 }, { "epoch": 1.2255970924195223, "grad_norm": 1.4571799039840698, "learning_rate": 2.050394580428016e-06, "loss": 0.10214337706565857, "step": 4721 }, { "epoch": 1.2258566978193146, "grad_norm": 2.0707192420959473, "learning_rate": 2.049495525535064e-06, "loss": 0.1218365877866745, "step": 4722 }, { "epoch": 1.226116303219107, "grad_norm": 1.3247753381729126, "learning_rate": 2.048596594475264e-06, "loss": 0.08812802284955978, "step": 4723 }, { "epoch": 1.2263759086188992, "grad_norm": 0.9207991361618042, "learning_rate": 2.047697787407526e-06, "loss": 0.10678558051586151, "step": 4724 }, { "epoch": 1.2266355140186915, "grad_norm": 1.659817099571228, "learning_rate": 2.046799104490738e-06, "loss": 0.0731409415602684, "step": 4725 }, { "epoch": 1.2268951194184838, "grad_norm": 2.525407075881958, "learning_rate": 2.045900545883768e-06, "loss": 0.08920980989933014, "step": 4726 }, { "epoch": 1.2271547248182761, "grad_norm": 1.8564342260360718, "learning_rate": 2.045002111745459e-06, "loss": 0.06975574791431427, "step": 4727 }, { "epoch": 1.2274143302180685, "grad_norm": 1.8517072200775146, "learning_rate": 2.0441038022346354e-06, "loss": 0.07924150675535202, "step": 4728 }, { "epoch": 1.2276739356178608, "grad_norm": 2.047266960144043, "learning_rate": 2.0432056175100957e-06, "loss": 0.11518135666847229, "step": 4729 }, { "epoch": 1.227933541017653, "grad_norm": 1.7580187320709229, "learning_rate": 2.0423075577306194e-06, "loss": 0.13211098313331604, "step": 4730 }, { "epoch": 1.2281931464174454, "grad_norm": 1.1342369318008423, "learning_rate": 2.041409623054963e-06, "loss": 0.10052161663770676, "step": 4731 }, { "epoch": 1.2284527518172377, "grad_norm": 1.4450818300247192, "learning_rate": 2.040511813641862e-06, "loss": 0.11939582973718643, "step": 4732 }, { "epoch": 1.2287123572170302, "grad_norm": 2.093379497528076, "learning_rate": 2.0396141296500256e-06, "loss": 0.0937478244304657, "step": 4733 }, { "epoch": 1.2289719626168225, "grad_norm": 2.30965518951416, "learning_rate": 2.038716571238145e-06, "loss": 0.061884235590696335, "step": 4734 }, { "epoch": 1.2292315680166148, "grad_norm": 0.6151968836784363, "learning_rate": 2.037819138564889e-06, "loss": 0.12790918350219727, "step": 4735 }, { "epoch": 1.2294911734164071, "grad_norm": 0.814793586730957, "learning_rate": 2.036921831788904e-06, "loss": 0.10945893824100494, "step": 4736 }, { "epoch": 1.2297507788161994, "grad_norm": 1.747654676437378, "learning_rate": 2.0360246510688102e-06, "loss": 0.10733655840158463, "step": 4737 }, { "epoch": 1.2300103842159917, "grad_norm": 1.625823736190796, "learning_rate": 2.0351275965632104e-06, "loss": 0.11750925332307816, "step": 4738 }, { "epoch": 1.230269989615784, "grad_norm": 2.379676342010498, "learning_rate": 2.0342306684306834e-06, "loss": 0.0817737877368927, "step": 4739 }, { "epoch": 1.2305295950155763, "grad_norm": 1.8309797048568726, "learning_rate": 2.033333866829785e-06, "loss": 0.1178707703948021, "step": 4740 }, { "epoch": 1.2307892004153687, "grad_norm": 1.349136233329773, "learning_rate": 2.03243719191905e-06, "loss": 0.0629582330584526, "step": 4741 }, { "epoch": 1.231048805815161, "grad_norm": 1.6710134744644165, "learning_rate": 2.031540643856989e-06, "loss": 0.08731186389923096, "step": 4742 }, { "epoch": 1.2313084112149533, "grad_norm": 1.7790601253509521, "learning_rate": 2.0306442228020917e-06, "loss": 0.10065925866365433, "step": 4743 }, { "epoch": 1.2315680166147456, "grad_norm": 0.8218916058540344, "learning_rate": 2.0297479289128246e-06, "loss": 0.12730003893375397, "step": 4744 }, { "epoch": 1.2318276220145379, "grad_norm": 1.7420454025268555, "learning_rate": 2.0288517623476324e-06, "loss": 0.10704504698514938, "step": 4745 }, { "epoch": 1.2320872274143302, "grad_norm": 0.833304762840271, "learning_rate": 2.027955723264936e-06, "loss": 0.1087205782532692, "step": 4746 }, { "epoch": 1.2323468328141225, "grad_norm": 1.786590576171875, "learning_rate": 2.027059811823136e-06, "loss": 0.07748767733573914, "step": 4747 }, { "epoch": 1.2326064382139148, "grad_norm": 0.7392938137054443, "learning_rate": 2.026164028180608e-06, "loss": 0.1130448579788208, "step": 4748 }, { "epoch": 1.232866043613707, "grad_norm": 2.077204465866089, "learning_rate": 2.0252683724957052e-06, "loss": 0.07010468095541, "step": 4749 }, { "epoch": 1.2331256490134994, "grad_norm": 1.0517722368240356, "learning_rate": 2.02437284492676e-06, "loss": 0.1039188951253891, "step": 4750 }, { "epoch": 1.2333852544132917, "grad_norm": 1.1268901824951172, "learning_rate": 2.0234774456320823e-06, "loss": 0.09905630350112915, "step": 4751 }, { "epoch": 1.233644859813084, "grad_norm": 0.967597246170044, "learning_rate": 2.0225821747699566e-06, "loss": 0.100420281291008, "step": 4752 }, { "epoch": 1.2339044652128763, "grad_norm": 1.866611361503601, "learning_rate": 2.0216870324986466e-06, "loss": 0.10509978234767914, "step": 4753 }, { "epoch": 1.2341640706126689, "grad_norm": 1.6913717985153198, "learning_rate": 2.020792018976393e-06, "loss": 0.11663911491632462, "step": 4754 }, { "epoch": 1.2344236760124612, "grad_norm": 1.0798447132110596, "learning_rate": 2.0198971343614136e-06, "loss": 0.10798263549804688, "step": 4755 }, { "epoch": 1.2346832814122535, "grad_norm": 0.9324933886528015, "learning_rate": 2.0190023788119044e-06, "loss": 0.10730204731225967, "step": 4756 }, { "epoch": 1.2349428868120458, "grad_norm": 1.5903671979904175, "learning_rate": 2.018107752486036e-06, "loss": 0.09100179374217987, "step": 4757 }, { "epoch": 1.235202492211838, "grad_norm": 2.3733885288238525, "learning_rate": 2.01721325554196e-06, "loss": 0.0819719135761261, "step": 4758 }, { "epoch": 1.2354620976116304, "grad_norm": 2.400402545928955, "learning_rate": 2.0163188881378017e-06, "loss": 0.0993385910987854, "step": 4759 }, { "epoch": 1.2357217030114227, "grad_norm": 1.2750372886657715, "learning_rate": 2.0154246504316637e-06, "loss": 0.08642677217721939, "step": 4760 }, { "epoch": 1.235981308411215, "grad_norm": 1.9648656845092773, "learning_rate": 2.0145305425816287e-06, "loss": 0.09115161001682281, "step": 4761 }, { "epoch": 1.2362409138110073, "grad_norm": 1.5801044702529907, "learning_rate": 2.013636564745754e-06, "loss": 0.09646575152873993, "step": 4762 }, { "epoch": 1.2365005192107996, "grad_norm": 0.6802464723587036, "learning_rate": 2.0127427170820746e-06, "loss": 0.10483389347791672, "step": 4763 }, { "epoch": 1.236760124610592, "grad_norm": 0.6802244186401367, "learning_rate": 2.0118489997486013e-06, "loss": 0.11123742908239365, "step": 4764 }, { "epoch": 1.2370197300103842, "grad_norm": 0.8311598896980286, "learning_rate": 2.0109554129033225e-06, "loss": 0.10342113673686981, "step": 4765 }, { "epoch": 1.2372793354101765, "grad_norm": 1.828099012374878, "learning_rate": 2.0100619567042068e-06, "loss": 0.059901200234889984, "step": 4766 }, { "epoch": 1.2375389408099688, "grad_norm": 1.757794737815857, "learning_rate": 2.0091686313091933e-06, "loss": 0.07298143208026886, "step": 4767 }, { "epoch": 1.2377985462097612, "grad_norm": 1.5962860584259033, "learning_rate": 2.0082754368762036e-06, "loss": 0.06800620257854462, "step": 4768 }, { "epoch": 1.2380581516095535, "grad_norm": 1.6343625783920288, "learning_rate": 2.0073823735631324e-06, "loss": 0.05683780834078789, "step": 4769 }, { "epoch": 1.2383177570093458, "grad_norm": 1.134319543838501, "learning_rate": 2.006489441527855e-06, "loss": 0.10802379250526428, "step": 4770 }, { "epoch": 1.238577362409138, "grad_norm": 0.9390475749969482, "learning_rate": 2.00559664092822e-06, "loss": 0.10162179172039032, "step": 4771 }, { "epoch": 1.2388369678089304, "grad_norm": 1.8514761924743652, "learning_rate": 2.004703971922054e-06, "loss": 0.07697127759456635, "step": 4772 }, { "epoch": 1.2390965732087227, "grad_norm": 2.296947717666626, "learning_rate": 2.0038114346671612e-06, "loss": 0.07274334877729416, "step": 4773 }, { "epoch": 1.239356178608515, "grad_norm": 1.468968152999878, "learning_rate": 2.0029190293213213e-06, "loss": 0.11922895908355713, "step": 4774 }, { "epoch": 1.2396157840083073, "grad_norm": 2.062366485595703, "learning_rate": 2.002026756042291e-06, "loss": 0.10315923392772675, "step": 4775 }, { "epoch": 1.2398753894080996, "grad_norm": 1.0541315078735352, "learning_rate": 2.0011346149878032e-06, "loss": 0.09372906386852264, "step": 4776 }, { "epoch": 1.240134994807892, "grad_norm": 1.0558140277862549, "learning_rate": 2.0002426063155694e-06, "loss": 0.11735202372074127, "step": 4777 }, { "epoch": 1.2403946002076842, "grad_norm": 1.8598949909210205, "learning_rate": 1.999350730183276e-06, "loss": 0.09243698418140411, "step": 4778 }, { "epoch": 1.2406542056074765, "grad_norm": 0.7786901593208313, "learning_rate": 1.998458986748585e-06, "loss": 0.11970491707324982, "step": 4779 }, { "epoch": 1.2409138110072688, "grad_norm": 2.947200298309326, "learning_rate": 1.997567376169137e-06, "loss": 0.08693112432956696, "step": 4780 }, { "epoch": 1.2411734164070614, "grad_norm": 1.4800245761871338, "learning_rate": 1.99667589860255e-06, "loss": 0.04562637209892273, "step": 4781 }, { "epoch": 1.2414330218068537, "grad_norm": 2.339153289794922, "learning_rate": 1.995784554206414e-06, "loss": 0.05361081659793854, "step": 4782 }, { "epoch": 1.241692627206646, "grad_norm": 1.8412498235702515, "learning_rate": 1.9948933431383e-06, "loss": 0.10701055824756622, "step": 4783 }, { "epoch": 1.2419522326064383, "grad_norm": 2.210320234298706, "learning_rate": 1.9940022655557527e-06, "loss": 0.0947682112455368, "step": 4784 }, { "epoch": 1.2422118380062306, "grad_norm": 1.5002073049545288, "learning_rate": 1.9931113216162956e-06, "loss": 0.08332467079162598, "step": 4785 }, { "epoch": 1.242471443406023, "grad_norm": 1.6309741735458374, "learning_rate": 1.992220511477426e-06, "loss": 0.0894312709569931, "step": 4786 }, { "epoch": 1.2427310488058152, "grad_norm": 1.5027821063995361, "learning_rate": 1.9913298352966193e-06, "loss": 0.10242348909378052, "step": 4787 }, { "epoch": 1.2429906542056075, "grad_norm": 0.8165651559829712, "learning_rate": 1.990439293231326e-06, "loss": 0.1005905270576477, "step": 4788 }, { "epoch": 1.2432502596053998, "grad_norm": 1.8168296813964844, "learning_rate": 1.9895488854389744e-06, "loss": 0.07950909435749054, "step": 4789 }, { "epoch": 1.2435098650051921, "grad_norm": 1.823508858680725, "learning_rate": 1.9886586120769685e-06, "loss": 0.07843995094299316, "step": 4790 }, { "epoch": 1.2437694704049844, "grad_norm": 1.8554226160049438, "learning_rate": 1.9877684733026865e-06, "loss": 0.15871194005012512, "step": 4791 }, { "epoch": 1.2440290758047767, "grad_norm": 2.1407840251922607, "learning_rate": 1.986878469273486e-06, "loss": 0.10796833038330078, "step": 4792 }, { "epoch": 1.244288681204569, "grad_norm": 0.6227608323097229, "learning_rate": 1.9859886001466988e-06, "loss": 0.11757767200469971, "step": 4793 }, { "epoch": 1.2445482866043613, "grad_norm": 3.0068447589874268, "learning_rate": 1.9850988660796344e-06, "loss": 0.08305655419826508, "step": 4794 }, { "epoch": 1.2448078920041536, "grad_norm": 1.2445285320281982, "learning_rate": 1.984209267229576e-06, "loss": 0.11423595249652863, "step": 4795 }, { "epoch": 1.245067497403946, "grad_norm": 1.6953917741775513, "learning_rate": 1.9833198037537844e-06, "loss": 0.11034323275089264, "step": 4796 }, { "epoch": 1.2453271028037383, "grad_norm": 1.1058330535888672, "learning_rate": 1.982430475809498e-06, "loss": 0.10172504931688309, "step": 4797 }, { "epoch": 1.2455867082035306, "grad_norm": 1.1315016746520996, "learning_rate": 1.9815412835539277e-06, "loss": 0.0956348404288292, "step": 4798 }, { "epoch": 1.2458463136033229, "grad_norm": 2.1238913536071777, "learning_rate": 1.9806522271442624e-06, "loss": 0.08525100350379944, "step": 4799 }, { "epoch": 1.2461059190031152, "grad_norm": 2.9480397701263428, "learning_rate": 1.979763306737668e-06, "loss": 0.08299963921308517, "step": 4800 }, { "epoch": 1.2463655244029075, "grad_norm": 0.975921094417572, "learning_rate": 1.978874522491286e-06, "loss": 0.09462817013263702, "step": 4801 }, { "epoch": 1.2466251298027, "grad_norm": 1.528088927268982, "learning_rate": 1.977985874562231e-06, "loss": 0.04033826291561127, "step": 4802 }, { "epoch": 1.2468847352024923, "grad_norm": 1.8933979272842407, "learning_rate": 1.977097363107597e-06, "loss": 0.08657872676849365, "step": 4803 }, { "epoch": 1.2471443406022846, "grad_norm": 1.4372785091400146, "learning_rate": 1.9762089882844517e-06, "loss": 0.09775520861148834, "step": 4804 }, { "epoch": 1.247403946002077, "grad_norm": 1.1602367162704468, "learning_rate": 1.97532075024984e-06, "loss": 0.11566731333732605, "step": 4805 }, { "epoch": 1.2476635514018692, "grad_norm": 2.271627187728882, "learning_rate": 1.974432649160782e-06, "loss": 0.08682756125926971, "step": 4806 }, { "epoch": 1.2479231568016615, "grad_norm": 1.8815243244171143, "learning_rate": 1.973544685174273e-06, "loss": 0.06523708254098892, "step": 4807 }, { "epoch": 1.2481827622014539, "grad_norm": 1.0506653785705566, "learning_rate": 1.972656858447285e-06, "loss": 0.09039673954248428, "step": 4808 }, { "epoch": 1.2484423676012462, "grad_norm": 1.576768159866333, "learning_rate": 1.971769169136767e-06, "loss": 0.057762064039707184, "step": 4809 }, { "epoch": 1.2487019730010385, "grad_norm": 0.669034481048584, "learning_rate": 1.9708816173996392e-06, "loss": 0.11969339847564697, "step": 4810 }, { "epoch": 1.2489615784008308, "grad_norm": 1.1027815341949463, "learning_rate": 1.969994203392803e-06, "loss": 0.10589846968650818, "step": 4811 }, { "epoch": 1.249221183800623, "grad_norm": 2.2276976108551025, "learning_rate": 1.969106927273132e-06, "loss": 0.09601536393165588, "step": 4812 }, { "epoch": 1.2494807892004154, "grad_norm": 1.0650813579559326, "learning_rate": 1.9682197891974753e-06, "loss": 0.08813127875328064, "step": 4813 }, { "epoch": 1.2497403946002077, "grad_norm": 2.627018928527832, "learning_rate": 1.9673327893226603e-06, "loss": 0.08876817673444748, "step": 4814 }, { "epoch": 1.25, "grad_norm": 1.4963213205337524, "learning_rate": 1.9664459278054866e-06, "loss": 0.10539543628692627, "step": 4815 }, { "epoch": 1.2502596053997923, "grad_norm": 1.6172162294387817, "learning_rate": 1.9655592048027324e-06, "loss": 0.0670037716627121, "step": 4816 }, { "epoch": 1.2505192107995846, "grad_norm": 2.082127571105957, "learning_rate": 1.964672620471149e-06, "loss": 0.09219756722450256, "step": 4817 }, { "epoch": 1.250778816199377, "grad_norm": 2.5313022136688232, "learning_rate": 1.963786174967465e-06, "loss": 0.06839495897293091, "step": 4818 }, { "epoch": 1.2510384215991692, "grad_norm": 1.9037845134735107, "learning_rate": 1.9628998684483835e-06, "loss": 0.10939992219209671, "step": 4819 }, { "epoch": 1.2512980269989615, "grad_norm": 2.407396078109741, "learning_rate": 1.962013701070583e-06, "loss": 0.06779249012470245, "step": 4820 }, { "epoch": 1.2515576323987538, "grad_norm": 1.6348568201065063, "learning_rate": 1.9611276729907187e-06, "loss": 0.06768002361059189, "step": 4821 }, { "epoch": 1.2518172377985461, "grad_norm": 1.8047782182693481, "learning_rate": 1.960241784365417e-06, "loss": 0.11074548214673996, "step": 4822 }, { "epoch": 1.2520768431983385, "grad_norm": 1.302378535270691, "learning_rate": 1.959356035351286e-06, "loss": 0.09621605277061462, "step": 4823 }, { "epoch": 1.2523364485981308, "grad_norm": 2.4484140872955322, "learning_rate": 1.9584704261049047e-06, "loss": 0.07918547093868256, "step": 4824 }, { "epoch": 1.252596053997923, "grad_norm": 1.7121210098266602, "learning_rate": 1.957584956782828e-06, "loss": 0.09343791007995605, "step": 4825 }, { "epoch": 1.2528556593977154, "grad_norm": 1.2207545042037964, "learning_rate": 1.956699627541587e-06, "loss": 0.1119050532579422, "step": 4826 }, { "epoch": 1.2531152647975077, "grad_norm": 2.263745069503784, "learning_rate": 1.9558144385376877e-06, "loss": 0.03890629857778549, "step": 4827 }, { "epoch": 1.2533748701973, "grad_norm": 3.082592248916626, "learning_rate": 1.9549293899276115e-06, "loss": 0.08749035745859146, "step": 4828 }, { "epoch": 1.2536344755970923, "grad_norm": 2.799783706665039, "learning_rate": 1.9540444818678146e-06, "loss": 0.10436701774597168, "step": 4829 }, { "epoch": 1.2538940809968846, "grad_norm": 0.8626585006713867, "learning_rate": 1.953159714514728e-06, "loss": 0.11443713307380676, "step": 4830 }, { "epoch": 1.2541536863966771, "grad_norm": 0.9839791059494019, "learning_rate": 1.952275088024758e-06, "loss": 0.09155015647411346, "step": 4831 }, { "epoch": 1.2544132917964694, "grad_norm": 2.614356756210327, "learning_rate": 1.9513906025542883e-06, "loss": 0.08671803027391434, "step": 4832 }, { "epoch": 1.2546728971962617, "grad_norm": 1.0109790563583374, "learning_rate": 1.9505062582596735e-06, "loss": 0.10283545404672623, "step": 4833 }, { "epoch": 1.254932502596054, "grad_norm": 1.8652557134628296, "learning_rate": 1.9496220552972467e-06, "loss": 0.08055712282657623, "step": 4834 }, { "epoch": 1.2551921079958464, "grad_norm": 1.6656099557876587, "learning_rate": 1.9487379938233136e-06, "loss": 0.07435799390077591, "step": 4835 }, { "epoch": 1.2554517133956387, "grad_norm": 0.8809136152267456, "learning_rate": 1.9478540739941577e-06, "loss": 0.1018739566206932, "step": 4836 }, { "epoch": 1.255711318795431, "grad_norm": 2.166677474975586, "learning_rate": 1.946970295966034e-06, "loss": 0.08598437905311584, "step": 4837 }, { "epoch": 1.2559709241952233, "grad_norm": 2.458033561706543, "learning_rate": 1.9460866598951748e-06, "loss": 0.104364313185215, "step": 4838 }, { "epoch": 1.2562305295950156, "grad_norm": 1.730401873588562, "learning_rate": 1.9452031659377874e-06, "loss": 0.08787904679775238, "step": 4839 }, { "epoch": 1.2564901349948079, "grad_norm": 1.725080132484436, "learning_rate": 1.9443198142500537e-06, "loss": 0.12021149694919586, "step": 4840 }, { "epoch": 1.2567497403946002, "grad_norm": 0.7746167182922363, "learning_rate": 1.943436604988128e-06, "loss": 0.11636011302471161, "step": 4841 }, { "epoch": 1.2570093457943925, "grad_norm": 2.288616180419922, "learning_rate": 1.9425535383081434e-06, "loss": 0.07034960389137268, "step": 4842 }, { "epoch": 1.2572689511941848, "grad_norm": 0.926501452922821, "learning_rate": 1.9416706143662044e-06, "loss": 0.10692664980888367, "step": 4843 }, { "epoch": 1.257528556593977, "grad_norm": 1.3213849067687988, "learning_rate": 1.9407878333183933e-06, "loss": 0.11409212648868561, "step": 4844 }, { "epoch": 1.2577881619937694, "grad_norm": 0.6838181018829346, "learning_rate": 1.9399051953207643e-06, "loss": 0.1097044050693512, "step": 4845 }, { "epoch": 1.2580477673935617, "grad_norm": 2.223217010498047, "learning_rate": 1.939022700529348e-06, "loss": 0.07327631860971451, "step": 4846 }, { "epoch": 1.258307372793354, "grad_norm": 1.3507041931152344, "learning_rate": 1.9381403491001492e-06, "loss": 0.10002008080482483, "step": 4847 }, { "epoch": 1.2585669781931463, "grad_norm": 1.1127694845199585, "learning_rate": 1.9372581411891485e-06, "loss": 0.10186998546123505, "step": 4848 }, { "epoch": 1.2588265835929389, "grad_norm": 1.1567420959472656, "learning_rate": 1.9363760769522982e-06, "loss": 0.1021980345249176, "step": 4849 }, { "epoch": 1.2590861889927312, "grad_norm": 0.8979769349098206, "learning_rate": 1.935494156545528e-06, "loss": 0.09103089570999146, "step": 4850 }, { "epoch": 1.2593457943925235, "grad_norm": 2.2396767139434814, "learning_rate": 1.9346123801247422e-06, "loss": 0.07384198158979416, "step": 4851 }, { "epoch": 1.2596053997923158, "grad_norm": 2.141606092453003, "learning_rate": 1.9337307478458167e-06, "loss": 0.059893377125263214, "step": 4852 }, { "epoch": 1.259865005192108, "grad_norm": 1.9576925039291382, "learning_rate": 1.9328492598646047e-06, "loss": 0.0767287164926529, "step": 4853 }, { "epoch": 1.2601246105919004, "grad_norm": 1.8455737829208374, "learning_rate": 1.9319679163369336e-06, "loss": 0.10819415748119354, "step": 4854 }, { "epoch": 1.2603842159916927, "grad_norm": 1.3969671726226807, "learning_rate": 1.9310867174186054e-06, "loss": 0.10558774322271347, "step": 4855 }, { "epoch": 1.260643821391485, "grad_norm": 2.424755096435547, "learning_rate": 1.9302056632653937e-06, "loss": 0.09002278745174408, "step": 4856 }, { "epoch": 1.2609034267912773, "grad_norm": 2.137596845626831, "learning_rate": 1.92932475403305e-06, "loss": 0.1062317043542862, "step": 4857 }, { "epoch": 1.2611630321910696, "grad_norm": 0.7993842363357544, "learning_rate": 1.9284439898772994e-06, "loss": 0.10926675796508789, "step": 4858 }, { "epoch": 1.261422637590862, "grad_norm": 1.2013938426971436, "learning_rate": 1.9275633709538403e-06, "loss": 0.09833135455846786, "step": 4859 }, { "epoch": 1.2616822429906542, "grad_norm": 1.53896963596344, "learning_rate": 1.9266828974183456e-06, "loss": 0.06885141879320145, "step": 4860 }, { "epoch": 1.2619418483904465, "grad_norm": 1.6446304321289062, "learning_rate": 1.925802569426463e-06, "loss": 0.08336713165044785, "step": 4861 }, { "epoch": 1.2622014537902388, "grad_norm": 2.20824933052063, "learning_rate": 1.9249223871338142e-06, "loss": 0.07237112522125244, "step": 4862 }, { "epoch": 1.2624610591900312, "grad_norm": 1.1230889558792114, "learning_rate": 1.924042350695997e-06, "loss": 0.07294008135795593, "step": 4863 }, { "epoch": 1.2627206645898235, "grad_norm": 0.7335888743400574, "learning_rate": 1.9231624602685787e-06, "loss": 0.1020444929599762, "step": 4864 }, { "epoch": 1.2629802699896158, "grad_norm": 1.7276878356933594, "learning_rate": 1.922282716007106e-06, "loss": 0.08719903230667114, "step": 4865 }, { "epoch": 1.263239875389408, "grad_norm": 2.0514261722564697, "learning_rate": 1.9214031180670973e-06, "loss": 0.12657573819160461, "step": 4866 }, { "epoch": 1.2634994807892004, "grad_norm": 0.8210644721984863, "learning_rate": 1.9205236666040443e-06, "loss": 0.103711798787117, "step": 4867 }, { "epoch": 1.2637590861889927, "grad_norm": 1.662289023399353, "learning_rate": 1.919644361773415e-06, "loss": 0.07057381421327591, "step": 4868 }, { "epoch": 1.264018691588785, "grad_norm": 0.8655010461807251, "learning_rate": 1.9187652037306496e-06, "loss": 0.12194617092609406, "step": 4869 }, { "epoch": 1.2642782969885773, "grad_norm": 1.9243613481521606, "learning_rate": 1.917886192631163e-06, "loss": 0.08512820303440094, "step": 4870 }, { "epoch": 1.2645379023883696, "grad_norm": 0.9656262993812561, "learning_rate": 1.9170073286303458e-06, "loss": 0.10568580031394958, "step": 4871 }, { "epoch": 1.264797507788162, "grad_norm": 2.0861871242523193, "learning_rate": 1.9161286118835594e-06, "loss": 0.07847271114587784, "step": 4872 }, { "epoch": 1.2650571131879542, "grad_norm": 1.0090738534927368, "learning_rate": 1.9152500425461414e-06, "loss": 0.09038181602954865, "step": 4873 }, { "epoch": 1.2653167185877465, "grad_norm": 1.5519946813583374, "learning_rate": 1.914371620773402e-06, "loss": 0.09386815130710602, "step": 4874 }, { "epoch": 1.2655763239875388, "grad_norm": 0.8543466925621033, "learning_rate": 1.9134933467206274e-06, "loss": 0.11236654222011566, "step": 4875 }, { "epoch": 1.2658359293873311, "grad_norm": 1.1308915615081787, "learning_rate": 1.9126152205430754e-06, "loss": 0.09991523623466492, "step": 4876 }, { "epoch": 1.2660955347871234, "grad_norm": 0.7204552888870239, "learning_rate": 1.9117372423959786e-06, "loss": 0.11660924553871155, "step": 4877 }, { "epoch": 1.2663551401869158, "grad_norm": 0.792795717716217, "learning_rate": 1.9108594124345444e-06, "loss": 0.11300357431173325, "step": 4878 }, { "epoch": 1.2666147455867083, "grad_norm": 1.4239239692687988, "learning_rate": 1.9099817308139516e-06, "loss": 0.07534782588481903, "step": 4879 }, { "epoch": 1.2668743509865006, "grad_norm": 0.8764354586601257, "learning_rate": 1.9091041976893547e-06, "loss": 0.1105983778834343, "step": 4880 }, { "epoch": 1.267133956386293, "grad_norm": 1.7040098905563354, "learning_rate": 1.908226813215882e-06, "loss": 0.09347440302371979, "step": 4881 }, { "epoch": 1.2673935617860852, "grad_norm": 1.3780416250228882, "learning_rate": 1.9073495775486356e-06, "loss": 0.12107788026332855, "step": 4882 }, { "epoch": 1.2676531671858775, "grad_norm": 2.2725017070770264, "learning_rate": 1.9064724908426885e-06, "loss": 0.0917869582772255, "step": 4883 }, { "epoch": 1.2679127725856698, "grad_norm": 1.3371864557266235, "learning_rate": 1.9055955532530912e-06, "loss": 0.0995764434337616, "step": 4884 }, { "epoch": 1.2681723779854621, "grad_norm": 1.519915223121643, "learning_rate": 1.9047187649348652e-06, "loss": 0.11431561410427094, "step": 4885 }, { "epoch": 1.2684319833852544, "grad_norm": 1.5445683002471924, "learning_rate": 1.903842126043009e-06, "loss": 0.08921870589256287, "step": 4886 }, { "epoch": 1.2686915887850467, "grad_norm": 2.2133524417877197, "learning_rate": 1.902965636732489e-06, "loss": 0.06975196301937103, "step": 4887 }, { "epoch": 1.268951194184839, "grad_norm": 2.108081340789795, "learning_rate": 1.9020892971582495e-06, "loss": 0.06366802752017975, "step": 4888 }, { "epoch": 1.2692107995846313, "grad_norm": 1.8779354095458984, "learning_rate": 1.9012131074752083e-06, "loss": 0.08008570969104767, "step": 4889 }, { "epoch": 1.2694704049844237, "grad_norm": 2.412997245788574, "learning_rate": 1.900337067838256e-06, "loss": 0.08699291944503784, "step": 4890 }, { "epoch": 1.269730010384216, "grad_norm": 1.3614875078201294, "learning_rate": 1.8994611784022542e-06, "loss": 0.09187757223844528, "step": 4891 }, { "epoch": 1.2699896157840083, "grad_norm": 1.6848057508468628, "learning_rate": 1.8985854393220414e-06, "loss": 0.09278284758329391, "step": 4892 }, { "epoch": 1.2702492211838006, "grad_norm": 1.8636606931686401, "learning_rate": 1.8977098507524283e-06, "loss": 0.09849251806735992, "step": 4893 }, { "epoch": 1.2705088265835929, "grad_norm": 1.6477380990982056, "learning_rate": 1.8968344128481992e-06, "loss": 0.1025606095790863, "step": 4894 }, { "epoch": 1.2707684319833852, "grad_norm": 1.719349980354309, "learning_rate": 1.8959591257641107e-06, "loss": 0.10198469460010529, "step": 4895 }, { "epoch": 1.2710280373831775, "grad_norm": 0.9210487008094788, "learning_rate": 1.8950839896548934e-06, "loss": 0.09694548696279526, "step": 4896 }, { "epoch": 1.27128764278297, "grad_norm": 0.6898658275604248, "learning_rate": 1.894209004675252e-06, "loss": 0.0971701368689537, "step": 4897 }, { "epoch": 1.2715472481827623, "grad_norm": 1.49635648727417, "learning_rate": 1.8933341709798642e-06, "loss": 0.1022898256778717, "step": 4898 }, { "epoch": 1.2718068535825546, "grad_norm": 0.7310174703598022, "learning_rate": 1.8924594887233787e-06, "loss": 0.0988873541355133, "step": 4899 }, { "epoch": 1.272066458982347, "grad_norm": 0.7215217351913452, "learning_rate": 1.8915849580604213e-06, "loss": 0.10793793201446533, "step": 4900 }, { "epoch": 1.2723260643821392, "grad_norm": 0.9620080590248108, "learning_rate": 1.8907105791455877e-06, "loss": 0.10045459866523743, "step": 4901 }, { "epoch": 1.2725856697819315, "grad_norm": 1.8996326923370361, "learning_rate": 1.8898363521334494e-06, "loss": 0.10004107654094696, "step": 4902 }, { "epoch": 1.2728452751817239, "grad_norm": 1.8029547929763794, "learning_rate": 1.888962277178548e-06, "loss": 0.08190111815929413, "step": 4903 }, { "epoch": 1.2731048805815162, "grad_norm": 1.0387537479400635, "learning_rate": 1.8880883544354017e-06, "loss": 0.09108814597129822, "step": 4904 }, { "epoch": 1.2733644859813085, "grad_norm": 1.6038622856140137, "learning_rate": 1.8872145840584987e-06, "loss": 0.08920487761497498, "step": 4905 }, { "epoch": 1.2736240913811008, "grad_norm": 0.6382516622543335, "learning_rate": 1.8863409662023024e-06, "loss": 0.10765865445137024, "step": 4906 }, { "epoch": 1.273883696780893, "grad_norm": 0.628226101398468, "learning_rate": 1.8854675010212476e-06, "loss": 0.10392381995916367, "step": 4907 }, { "epoch": 1.2741433021806854, "grad_norm": 0.7733606696128845, "learning_rate": 1.8845941886697432e-06, "loss": 0.08718245476484299, "step": 4908 }, { "epoch": 1.2744029075804777, "grad_norm": 0.921006977558136, "learning_rate": 1.8837210293021724e-06, "loss": 0.08992837369441986, "step": 4909 }, { "epoch": 1.27466251298027, "grad_norm": 0.7481998205184937, "learning_rate": 1.8828480230728874e-06, "loss": 0.10957988351583481, "step": 4910 }, { "epoch": 1.2749221183800623, "grad_norm": 1.2908830642700195, "learning_rate": 1.8819751701362165e-06, "loss": 0.09267114102840424, "step": 4911 }, { "epoch": 1.2751817237798546, "grad_norm": 1.9467570781707764, "learning_rate": 1.8811024706464614e-06, "loss": 0.09427624940872192, "step": 4912 }, { "epoch": 1.275441329179647, "grad_norm": 1.1653364896774292, "learning_rate": 1.8802299247578943e-06, "loss": 0.10977944731712341, "step": 4913 }, { "epoch": 1.2757009345794392, "grad_norm": 1.8802770376205444, "learning_rate": 1.8793575326247603e-06, "loss": 0.09672713279724121, "step": 4914 }, { "epoch": 1.2759605399792315, "grad_norm": 1.5281627178192139, "learning_rate": 1.8784852944012799e-06, "loss": 0.09547396004199982, "step": 4915 }, { "epoch": 1.2762201453790238, "grad_norm": 1.031374216079712, "learning_rate": 1.8776132102416444e-06, "loss": 0.10312943160533905, "step": 4916 }, { "epoch": 1.2764797507788161, "grad_norm": 1.1504865884780884, "learning_rate": 1.8767412803000195e-06, "loss": 0.08774544298648834, "step": 4917 }, { "epoch": 1.2767393561786085, "grad_norm": 0.942642092704773, "learning_rate": 1.875869504730541e-06, "loss": 0.09522461891174316, "step": 4918 }, { "epoch": 1.2769989615784008, "grad_norm": 1.5560593605041504, "learning_rate": 1.8749978836873174e-06, "loss": 0.09732136130332947, "step": 4919 }, { "epoch": 1.277258566978193, "grad_norm": 2.561988353729248, "learning_rate": 1.874126417324434e-06, "loss": 0.056511085480451584, "step": 4920 }, { "epoch": 1.2775181723779854, "grad_norm": 0.7528212070465088, "learning_rate": 1.8732551057959459e-06, "loss": 0.10082804411649704, "step": 4921 }, { "epoch": 1.2777777777777777, "grad_norm": 2.482461452484131, "learning_rate": 1.8723839492558793e-06, "loss": 0.10791708528995514, "step": 4922 }, { "epoch": 1.27803738317757, "grad_norm": 1.11308753490448, "learning_rate": 1.8715129478582367e-06, "loss": 0.0489935576915741, "step": 4923 }, { "epoch": 1.2782969885773623, "grad_norm": 1.7264659404754639, "learning_rate": 1.8706421017569888e-06, "loss": 0.09418562054634094, "step": 4924 }, { "epoch": 1.2785565939771546, "grad_norm": 1.0075565576553345, "learning_rate": 1.8697714111060843e-06, "loss": 0.10197530686855316, "step": 4925 }, { "epoch": 1.278816199376947, "grad_norm": 0.889801025390625, "learning_rate": 1.868900876059439e-06, "loss": 0.10304834693670273, "step": 4926 }, { "epoch": 1.2790758047767394, "grad_norm": 1.1433039903640747, "learning_rate": 1.868030496770944e-06, "loss": 0.100599005818367, "step": 4927 }, { "epoch": 1.2793354101765317, "grad_norm": 1.8422927856445312, "learning_rate": 1.8671602733944633e-06, "loss": 0.11204510927200317, "step": 4928 }, { "epoch": 1.279595015576324, "grad_norm": 2.337397336959839, "learning_rate": 1.8662902060838318e-06, "loss": 0.1035575121641159, "step": 4929 }, { "epoch": 1.2798546209761164, "grad_norm": 1.2213844060897827, "learning_rate": 1.8654202949928574e-06, "loss": 0.12540313601493835, "step": 4930 }, { "epoch": 1.2801142263759087, "grad_norm": 1.8448903560638428, "learning_rate": 1.8645505402753205e-06, "loss": 0.07921917736530304, "step": 4931 }, { "epoch": 1.280373831775701, "grad_norm": 0.8535118699073792, "learning_rate": 1.8636809420849745e-06, "loss": 0.11006449162960052, "step": 4932 }, { "epoch": 1.2806334371754933, "grad_norm": 1.2361242771148682, "learning_rate": 1.8628115005755432e-06, "loss": 0.11095689982175827, "step": 4933 }, { "epoch": 1.2808930425752856, "grad_norm": 2.2325313091278076, "learning_rate": 1.861942215900725e-06, "loss": 0.11040027439594269, "step": 4934 }, { "epoch": 1.2811526479750779, "grad_norm": 1.3548877239227295, "learning_rate": 1.8610730882141893e-06, "loss": 0.09351082146167755, "step": 4935 }, { "epoch": 1.2814122533748702, "grad_norm": 0.7215409874916077, "learning_rate": 1.8602041176695779e-06, "loss": 0.10687364637851715, "step": 4936 }, { "epoch": 1.2816718587746625, "grad_norm": 0.5497516989707947, "learning_rate": 1.8593353044205047e-06, "loss": 0.11077573895454407, "step": 4937 }, { "epoch": 1.2819314641744548, "grad_norm": 0.8660454750061035, "learning_rate": 1.8584666486205567e-06, "loss": 0.10576347261667252, "step": 4938 }, { "epoch": 1.282191069574247, "grad_norm": 1.6374064683914185, "learning_rate": 1.857598150423291e-06, "loss": 0.09147538244724274, "step": 4939 }, { "epoch": 1.2824506749740394, "grad_norm": 2.518895387649536, "learning_rate": 1.8567298099822395e-06, "loss": 0.09011484682559967, "step": 4940 }, { "epoch": 1.2827102803738317, "grad_norm": 0.9145528078079224, "learning_rate": 1.8558616274509046e-06, "loss": 0.09894247353076935, "step": 4941 }, { "epoch": 1.282969885773624, "grad_norm": 1.0353277921676636, "learning_rate": 1.8549936029827603e-06, "loss": 0.09172634780406952, "step": 4942 }, { "epoch": 1.2832294911734163, "grad_norm": 0.7224407196044922, "learning_rate": 1.8541257367312543e-06, "loss": 0.10251383483409882, "step": 4943 }, { "epoch": 1.2834890965732086, "grad_norm": 0.9384313225746155, "learning_rate": 1.8532580288498065e-06, "loss": 0.08229048550128937, "step": 4944 }, { "epoch": 1.2837487019730012, "grad_norm": 1.188149333000183, "learning_rate": 1.8523904794918046e-06, "loss": 0.0969233363866806, "step": 4945 }, { "epoch": 1.2840083073727935, "grad_norm": 3.0188703536987305, "learning_rate": 1.851523088810614e-06, "loss": 0.08112110942602158, "step": 4946 }, { "epoch": 1.2842679127725858, "grad_norm": 1.89867103099823, "learning_rate": 1.850655856959569e-06, "loss": 0.09053179621696472, "step": 4947 }, { "epoch": 1.284527518172378, "grad_norm": 1.8294318914413452, "learning_rate": 1.849788784091977e-06, "loss": 0.09086914360523224, "step": 4948 }, { "epoch": 1.2847871235721704, "grad_norm": 1.1152877807617188, "learning_rate": 1.848921870361115e-06, "loss": 0.09711398929357529, "step": 4949 }, { "epoch": 1.2850467289719627, "grad_norm": 1.410899043083191, "learning_rate": 1.8480551159202342e-06, "loss": 0.12542398273944855, "step": 4950 }, { "epoch": 1.285306334371755, "grad_norm": 1.799659252166748, "learning_rate": 1.847188520922557e-06, "loss": 0.0813264399766922, "step": 4951 }, { "epoch": 1.2855659397715473, "grad_norm": 0.9907915592193604, "learning_rate": 1.8463220855212789e-06, "loss": 0.09415952861309052, "step": 4952 }, { "epoch": 1.2858255451713396, "grad_norm": 1.9687926769256592, "learning_rate": 1.845455809869564e-06, "loss": 0.04772327095270157, "step": 4953 }, { "epoch": 1.286085150571132, "grad_norm": 1.1444382667541504, "learning_rate": 1.8445896941205503e-06, "loss": 0.1014329046010971, "step": 4954 }, { "epoch": 1.2863447559709242, "grad_norm": 1.6956249475479126, "learning_rate": 1.8437237384273471e-06, "loss": 0.07203269004821777, "step": 4955 }, { "epoch": 1.2866043613707165, "grad_norm": 1.797334909439087, "learning_rate": 1.842857942943037e-06, "loss": 0.08821507543325424, "step": 4956 }, { "epoch": 1.2868639667705088, "grad_norm": 1.3134334087371826, "learning_rate": 1.8419923078206714e-06, "loss": 0.08233381807804108, "step": 4957 }, { "epoch": 1.2871235721703012, "grad_norm": 1.8644988536834717, "learning_rate": 1.8411268332132747e-06, "loss": 0.13013869524002075, "step": 4958 }, { "epoch": 1.2873831775700935, "grad_norm": 0.8916760087013245, "learning_rate": 1.8402615192738443e-06, "loss": 0.11173571646213531, "step": 4959 }, { "epoch": 1.2876427829698858, "grad_norm": 1.6910005807876587, "learning_rate": 1.8393963661553466e-06, "loss": 0.047008365392684937, "step": 4960 }, { "epoch": 1.287902388369678, "grad_norm": 1.4002207517623901, "learning_rate": 1.8385313740107208e-06, "loss": 0.06255980581045151, "step": 4961 }, { "epoch": 1.2881619937694704, "grad_norm": 0.876453161239624, "learning_rate": 1.8376665429928786e-06, "loss": 0.11130937933921814, "step": 4962 }, { "epoch": 1.2884215991692627, "grad_norm": 2.1573822498321533, "learning_rate": 1.8368018732547026e-06, "loss": 0.10278597474098206, "step": 4963 }, { "epoch": 1.288681204569055, "grad_norm": 1.9567006826400757, "learning_rate": 1.8359373649490452e-06, "loss": 0.08385571092367172, "step": 4964 }, { "epoch": 1.2889408099688473, "grad_norm": 1.0189908742904663, "learning_rate": 1.8350730182287323e-06, "loss": 0.09077081084251404, "step": 4965 }, { "epoch": 1.2892004153686396, "grad_norm": 1.2332594394683838, "learning_rate": 1.8342088332465612e-06, "loss": 0.06855422258377075, "step": 4966 }, { "epoch": 1.289460020768432, "grad_norm": 2.2448294162750244, "learning_rate": 1.8333448101552993e-06, "loss": 0.05798068270087242, "step": 4967 }, { "epoch": 1.2897196261682242, "grad_norm": 1.3928409814834595, "learning_rate": 1.8324809491076861e-06, "loss": 0.08001552522182465, "step": 4968 }, { "epoch": 1.2899792315680165, "grad_norm": 1.484084963798523, "learning_rate": 1.8316172502564334e-06, "loss": 0.1069699078798294, "step": 4969 }, { "epoch": 1.2902388369678088, "grad_norm": 1.130466341972351, "learning_rate": 1.8307537137542219e-06, "loss": 0.10533522814512253, "step": 4970 }, { "epoch": 1.2904984423676011, "grad_norm": 2.773263692855835, "learning_rate": 1.829890339753706e-06, "loss": 0.11379852890968323, "step": 4971 }, { "epoch": 1.2907580477673934, "grad_norm": 1.6600093841552734, "learning_rate": 1.8290271284075106e-06, "loss": 0.11764736473560333, "step": 4972 }, { "epoch": 1.2910176531671858, "grad_norm": 2.081948757171631, "learning_rate": 1.8281640798682316e-06, "loss": 0.05538302659988403, "step": 4973 }, { "epoch": 1.291277258566978, "grad_norm": 0.9145585298538208, "learning_rate": 1.8273011942884356e-06, "loss": 0.11368018388748169, "step": 4974 }, { "epoch": 1.2915368639667706, "grad_norm": 1.1089812517166138, "learning_rate": 1.8264384718206624e-06, "loss": 0.09307290613651276, "step": 4975 }, { "epoch": 1.291796469366563, "grad_norm": 2.187260627746582, "learning_rate": 1.8255759126174188e-06, "loss": 0.09824703633785248, "step": 4976 }, { "epoch": 1.2920560747663552, "grad_norm": 2.2623562812805176, "learning_rate": 1.8247135168311885e-06, "loss": 0.08741535246372223, "step": 4977 }, { "epoch": 1.2923156801661475, "grad_norm": 1.82276451587677, "learning_rate": 1.823851284614422e-06, "loss": 0.10073759406805038, "step": 4978 }, { "epoch": 1.2925752855659398, "grad_norm": 1.7067333459854126, "learning_rate": 1.8229892161195435e-06, "loss": 0.08577220886945724, "step": 4979 }, { "epoch": 1.2928348909657321, "grad_norm": 0.7387046813964844, "learning_rate": 1.8221273114989447e-06, "loss": 0.11478707194328308, "step": 4980 }, { "epoch": 1.2930944963655244, "grad_norm": 1.7754110097885132, "learning_rate": 1.8212655709049908e-06, "loss": 0.09618349373340607, "step": 4981 }, { "epoch": 1.2933541017653167, "grad_norm": 0.9069662094116211, "learning_rate": 1.8204039944900198e-06, "loss": 0.10963216423988342, "step": 4982 }, { "epoch": 1.293613707165109, "grad_norm": 0.7376523613929749, "learning_rate": 1.8195425824063384e-06, "loss": 0.09929913282394409, "step": 4983 }, { "epoch": 1.2938733125649013, "grad_norm": 0.7268335223197937, "learning_rate": 1.8186813348062232e-06, "loss": 0.12064263224601746, "step": 4984 }, { "epoch": 1.2941329179646937, "grad_norm": 1.586085557937622, "learning_rate": 1.817820251841923e-06, "loss": 0.08486179262399673, "step": 4985 }, { "epoch": 1.294392523364486, "grad_norm": 1.1095062494277954, "learning_rate": 1.8169593336656594e-06, "loss": 0.11501765251159668, "step": 4986 }, { "epoch": 1.2946521287642783, "grad_norm": 1.2990349531173706, "learning_rate": 1.816098580429621e-06, "loss": 0.10334787517786026, "step": 4987 }, { "epoch": 1.2949117341640706, "grad_norm": 1.9226102828979492, "learning_rate": 1.8152379922859697e-06, "loss": 0.06338106095790863, "step": 4988 }, { "epoch": 1.2951713395638629, "grad_norm": 2.0504887104034424, "learning_rate": 1.8143775693868387e-06, "loss": 0.08202405273914337, "step": 4989 }, { "epoch": 1.2954309449636552, "grad_norm": 1.803162932395935, "learning_rate": 1.8135173118843313e-06, "loss": 0.0995875746011734, "step": 4990 }, { "epoch": 1.2956905503634475, "grad_norm": 1.7430073022842407, "learning_rate": 1.8126572199305192e-06, "loss": 0.06880881637334824, "step": 4991 }, { "epoch": 1.2959501557632398, "grad_norm": 0.9211831092834473, "learning_rate": 1.811797293677448e-06, "loss": 0.0893508568406105, "step": 4992 }, { "epoch": 1.2962097611630323, "grad_norm": 1.839077115058899, "learning_rate": 1.8109375332771346e-06, "loss": 0.07209859043359756, "step": 4993 }, { "epoch": 1.2964693665628246, "grad_norm": 0.9719936847686768, "learning_rate": 1.8100779388815631e-06, "loss": 0.10943108052015305, "step": 4994 }, { "epoch": 1.296728971962617, "grad_norm": 0.9359202980995178, "learning_rate": 1.8092185106426913e-06, "loss": 0.10295574367046356, "step": 4995 }, { "epoch": 1.2969885773624092, "grad_norm": 1.0542527437210083, "learning_rate": 1.808359248712444e-06, "loss": 0.10154297947883606, "step": 4996 }, { "epoch": 1.2972481827622016, "grad_norm": 0.9940427541732788, "learning_rate": 1.8075001532427216e-06, "loss": 0.09263833612203598, "step": 4997 }, { "epoch": 1.2975077881619939, "grad_norm": 0.7570685148239136, "learning_rate": 1.806641224385393e-06, "loss": 0.11148155480623245, "step": 4998 }, { "epoch": 1.2977673935617862, "grad_norm": 1.6295709609985352, "learning_rate": 1.8057824622922948e-06, "loss": 0.08656396716833115, "step": 4999 }, { "epoch": 1.2980269989615785, "grad_norm": 2.1062815189361572, "learning_rate": 1.8049238671152381e-06, "loss": 0.08703595399856567, "step": 5000 }, { "epoch": 1.2982866043613708, "grad_norm": 1.8208476305007935, "learning_rate": 1.804065439006001e-06, "loss": 0.08524604141712189, "step": 5001 }, { "epoch": 1.298546209761163, "grad_norm": 1.7853790521621704, "learning_rate": 1.8032071781163363e-06, "loss": 0.0901721864938736, "step": 5002 }, { "epoch": 1.2988058151609554, "grad_norm": 2.0108771324157715, "learning_rate": 1.802349084597964e-06, "loss": 0.0685400664806366, "step": 5003 }, { "epoch": 1.2990654205607477, "grad_norm": 1.8850218057632446, "learning_rate": 1.8014911586025747e-06, "loss": 0.08939540386199951, "step": 5004 }, { "epoch": 1.29932502596054, "grad_norm": 1.4095593690872192, "learning_rate": 1.800633400281831e-06, "loss": 0.09722776710987091, "step": 5005 }, { "epoch": 1.2995846313603323, "grad_norm": 2.039494037628174, "learning_rate": 1.7997758097873647e-06, "loss": 0.0673384964466095, "step": 5006 }, { "epoch": 1.2998442367601246, "grad_norm": 1.57309889793396, "learning_rate": 1.7989183872707767e-06, "loss": 0.08827182650566101, "step": 5007 }, { "epoch": 1.300103842159917, "grad_norm": 2.427859306335449, "learning_rate": 1.7980611328836422e-06, "loss": 0.08643168210983276, "step": 5008 }, { "epoch": 1.3003634475597092, "grad_norm": 1.96788489818573, "learning_rate": 1.7972040467775021e-06, "loss": 0.10094838589429855, "step": 5009 }, { "epoch": 1.3006230529595015, "grad_norm": 1.7753342390060425, "learning_rate": 1.7963471291038708e-06, "loss": 0.10202015936374664, "step": 5010 }, { "epoch": 1.3008826583592938, "grad_norm": 1.155860185623169, "learning_rate": 1.7954903800142321e-06, "loss": 0.09620825946331024, "step": 5011 }, { "epoch": 1.3011422637590861, "grad_norm": 1.696862816810608, "learning_rate": 1.7946337996600371e-06, "loss": 0.11175764352083206, "step": 5012 }, { "epoch": 1.3014018691588785, "grad_norm": 1.278415560722351, "learning_rate": 1.793777388192713e-06, "loss": 0.10897757858037949, "step": 5013 }, { "epoch": 1.3016614745586708, "grad_norm": 1.159961462020874, "learning_rate": 1.7929211457636514e-06, "loss": 0.094966821372509, "step": 5014 }, { "epoch": 1.301921079958463, "grad_norm": 0.7689101696014404, "learning_rate": 1.7920650725242167e-06, "loss": 0.10413309931755066, "step": 5015 }, { "epoch": 1.3021806853582554, "grad_norm": 1.3697742223739624, "learning_rate": 1.7912091686257438e-06, "loss": 0.10010640323162079, "step": 5016 }, { "epoch": 1.3024402907580477, "grad_norm": 1.7575366497039795, "learning_rate": 1.7903534342195366e-06, "loss": 0.08016490936279297, "step": 5017 }, { "epoch": 1.30269989615784, "grad_norm": 2.5068342685699463, "learning_rate": 1.7894978694568688e-06, "loss": 0.09367348253726959, "step": 5018 }, { "epoch": 1.3029595015576323, "grad_norm": 1.5408666133880615, "learning_rate": 1.7886424744889852e-06, "loss": 0.0764082670211792, "step": 5019 }, { "epoch": 1.3032191069574246, "grad_norm": 1.9413719177246094, "learning_rate": 1.7877872494670996e-06, "loss": 0.07913719862699509, "step": 5020 }, { "epoch": 1.303478712357217, "grad_norm": 1.834448218345642, "learning_rate": 1.7869321945423985e-06, "loss": 0.07938352227210999, "step": 5021 }, { "epoch": 1.3037383177570092, "grad_norm": 1.5289329290390015, "learning_rate": 1.7860773098660316e-06, "loss": 0.0746985673904419, "step": 5022 }, { "epoch": 1.3039979231568017, "grad_norm": 1.0915687084197998, "learning_rate": 1.7852225955891256e-06, "loss": 0.09609170258045197, "step": 5023 }, { "epoch": 1.304257528556594, "grad_norm": 2.2860326766967773, "learning_rate": 1.7843680518627753e-06, "loss": 0.10280642658472061, "step": 5024 }, { "epoch": 1.3045171339563864, "grad_norm": 0.9972696900367737, "learning_rate": 1.7835136788380436e-06, "loss": 0.1005650982260704, "step": 5025 }, { "epoch": 1.3047767393561787, "grad_norm": 1.40913724899292, "learning_rate": 1.7826594766659638e-06, "loss": 0.091154545545578, "step": 5026 }, { "epoch": 1.305036344755971, "grad_norm": 1.7213668823242188, "learning_rate": 1.7818054454975387e-06, "loss": 0.07956188172101974, "step": 5027 }, { "epoch": 1.3052959501557633, "grad_norm": 1.9573733806610107, "learning_rate": 1.7809515854837418e-06, "loss": 0.08848810195922852, "step": 5028 }, { "epoch": 1.3055555555555556, "grad_norm": 1.3880937099456787, "learning_rate": 1.7800978967755179e-06, "loss": 0.0952645018696785, "step": 5029 }, { "epoch": 1.3058151609553479, "grad_norm": 1.0850284099578857, "learning_rate": 1.7792443795237764e-06, "loss": 0.11106666922569275, "step": 5030 }, { "epoch": 1.3060747663551402, "grad_norm": 1.6155424118041992, "learning_rate": 1.7783910338794019e-06, "loss": 0.10667698830366135, "step": 5031 }, { "epoch": 1.3063343717549325, "grad_norm": 1.3053226470947266, "learning_rate": 1.7775378599932452e-06, "loss": 0.09330153465270996, "step": 5032 }, { "epoch": 1.3065939771547248, "grad_norm": 1.9279711246490479, "learning_rate": 1.776684858016129e-06, "loss": 0.06521078944206238, "step": 5033 }, { "epoch": 1.3068535825545171, "grad_norm": 1.1408963203430176, "learning_rate": 1.7758320280988444e-06, "loss": 0.09530483186244965, "step": 5034 }, { "epoch": 1.3071131879543094, "grad_norm": 1.5248456001281738, "learning_rate": 1.774979370392151e-06, "loss": 0.06606833636760712, "step": 5035 }, { "epoch": 1.3073727933541017, "grad_norm": 0.7527453899383545, "learning_rate": 1.7741268850467796e-06, "loss": 0.11295610666275024, "step": 5036 }, { "epoch": 1.307632398753894, "grad_norm": 1.70902681350708, "learning_rate": 1.7732745722134315e-06, "loss": 0.06763696670532227, "step": 5037 }, { "epoch": 1.3078920041536863, "grad_norm": 1.0016359090805054, "learning_rate": 1.7724224320427741e-06, "loss": 0.10196969658136368, "step": 5038 }, { "epoch": 1.3081516095534786, "grad_norm": 2.2750649452209473, "learning_rate": 1.7715704646854476e-06, "loss": 0.0957225039601326, "step": 5039 }, { "epoch": 1.308411214953271, "grad_norm": 2.058605432510376, "learning_rate": 1.7707186702920587e-06, "loss": 0.046889528632164, "step": 5040 }, { "epoch": 1.3086708203530635, "grad_norm": 1.4133124351501465, "learning_rate": 1.7698670490131875e-06, "loss": 0.08865290135145187, "step": 5041 }, { "epoch": 1.3089304257528558, "grad_norm": 2.2128074169158936, "learning_rate": 1.7690156009993803e-06, "loss": 0.09472204744815826, "step": 5042 }, { "epoch": 1.309190031152648, "grad_norm": 2.818209648132324, "learning_rate": 1.7681643264011523e-06, "loss": 0.13521632552146912, "step": 5043 }, { "epoch": 1.3094496365524404, "grad_norm": 1.5568156242370605, "learning_rate": 1.7673132253689913e-06, "loss": 0.05798737704753876, "step": 5044 }, { "epoch": 1.3097092419522327, "grad_norm": 1.1975964307785034, "learning_rate": 1.7664622980533508e-06, "loss": 0.10398496687412262, "step": 5045 }, { "epoch": 1.309968847352025, "grad_norm": 1.7278258800506592, "learning_rate": 1.7656115446046558e-06, "loss": 0.06002102047204971, "step": 5046 }, { "epoch": 1.3102284527518173, "grad_norm": 1.3096463680267334, "learning_rate": 1.764760965173301e-06, "loss": 0.08647437393665314, "step": 5047 }, { "epoch": 1.3104880581516096, "grad_norm": 1.4835972785949707, "learning_rate": 1.7639105599096484e-06, "loss": 0.08392171561717987, "step": 5048 }, { "epoch": 1.310747663551402, "grad_norm": 1.8294953107833862, "learning_rate": 1.7630603289640298e-06, "loss": 0.0675918310880661, "step": 5049 }, { "epoch": 1.3110072689511942, "grad_norm": 2.413435459136963, "learning_rate": 1.7622102724867473e-06, "loss": 0.07026897370815277, "step": 5050 }, { "epoch": 1.3112668743509865, "grad_norm": 1.1099427938461304, "learning_rate": 1.7613603906280708e-06, "loss": 0.11422748863697052, "step": 5051 }, { "epoch": 1.3115264797507789, "grad_norm": 1.6390002965927124, "learning_rate": 1.7605106835382424e-06, "loss": 0.06735639274120331, "step": 5052 }, { "epoch": 1.3117860851505712, "grad_norm": 2.400676727294922, "learning_rate": 1.7596611513674667e-06, "loss": 0.07116980105638504, "step": 5053 }, { "epoch": 1.3120456905503635, "grad_norm": 0.9815429449081421, "learning_rate": 1.7588117942659239e-06, "loss": 0.1025952696800232, "step": 5054 }, { "epoch": 1.3123052959501558, "grad_norm": 0.9681453704833984, "learning_rate": 1.7579626123837612e-06, "loss": 0.10219116508960724, "step": 5055 }, { "epoch": 1.312564901349948, "grad_norm": 0.6869997382164001, "learning_rate": 1.7571136058710942e-06, "loss": 0.0964328795671463, "step": 5056 }, { "epoch": 1.3128245067497404, "grad_norm": 0.5667349100112915, "learning_rate": 1.7562647748780076e-06, "loss": 0.10807107388973236, "step": 5057 }, { "epoch": 1.3130841121495327, "grad_norm": 0.9744312763214111, "learning_rate": 1.7554161195545545e-06, "loss": 0.09176282584667206, "step": 5058 }, { "epoch": 1.313343717549325, "grad_norm": 1.6821876764297485, "learning_rate": 1.754567640050759e-06, "loss": 0.07752963900566101, "step": 5059 }, { "epoch": 1.3136033229491173, "grad_norm": 0.838627278804779, "learning_rate": 1.7537193365166133e-06, "loss": 0.11617110669612885, "step": 5060 }, { "epoch": 1.3138629283489096, "grad_norm": 1.718875765800476, "learning_rate": 1.7528712091020756e-06, "loss": 0.10258333384990692, "step": 5061 }, { "epoch": 1.314122533748702, "grad_norm": 1.33382248878479, "learning_rate": 1.7520232579570768e-06, "loss": 0.09690790623426437, "step": 5062 }, { "epoch": 1.3143821391484942, "grad_norm": 1.075677752494812, "learning_rate": 1.7511754832315165e-06, "loss": 0.08598051965236664, "step": 5063 }, { "epoch": 1.3146417445482865, "grad_norm": 2.0800366401672363, "learning_rate": 1.7503278850752604e-06, "loss": 0.09391899406909943, "step": 5064 }, { "epoch": 1.3149013499480788, "grad_norm": 0.6730133891105652, "learning_rate": 1.7494804636381446e-06, "loss": 0.10247322171926498, "step": 5065 }, { "epoch": 1.3151609553478711, "grad_norm": 1.751095175743103, "learning_rate": 1.748633219069974e-06, "loss": 0.0636238157749176, "step": 5066 }, { "epoch": 1.3154205607476634, "grad_norm": 1.0070768594741821, "learning_rate": 1.7477861515205213e-06, "loss": 0.09841476380825043, "step": 5067 }, { "epoch": 1.3156801661474558, "grad_norm": 1.953928828239441, "learning_rate": 1.7469392611395309e-06, "loss": 0.05901001766324043, "step": 5068 }, { "epoch": 1.315939771547248, "grad_norm": 2.0720980167388916, "learning_rate": 1.7460925480767107e-06, "loss": 0.0845760852098465, "step": 5069 }, { "epoch": 1.3161993769470404, "grad_norm": 1.936699390411377, "learning_rate": 1.7452460124817424e-06, "loss": 0.10090354084968567, "step": 5070 }, { "epoch": 1.316458982346833, "grad_norm": 1.0343163013458252, "learning_rate": 1.7443996545042734e-06, "loss": 0.09261175990104675, "step": 5071 }, { "epoch": 1.3167185877466252, "grad_norm": 1.6702309846878052, "learning_rate": 1.7435534742939193e-06, "loss": 0.07663043588399887, "step": 5072 }, { "epoch": 1.3169781931464175, "grad_norm": 0.8193424940109253, "learning_rate": 1.7427074720002673e-06, "loss": 0.09174564480781555, "step": 5073 }, { "epoch": 1.3172377985462098, "grad_norm": 1.480992317199707, "learning_rate": 1.7418616477728694e-06, "loss": 0.11188167333602905, "step": 5074 }, { "epoch": 1.3174974039460021, "grad_norm": 1.182507038116455, "learning_rate": 1.74101600176125e-06, "loss": 0.0843099057674408, "step": 5075 }, { "epoch": 1.3177570093457944, "grad_norm": 1.5928915739059448, "learning_rate": 1.740170534114898e-06, "loss": 0.09513647109270096, "step": 5076 }, { "epoch": 1.3180166147455867, "grad_norm": 1.9685451984405518, "learning_rate": 1.7393252449832734e-06, "loss": 0.07661495357751846, "step": 5077 }, { "epoch": 1.318276220145379, "grad_norm": 1.2498301267623901, "learning_rate": 1.7384801345158048e-06, "loss": 0.09732312709093094, "step": 5078 }, { "epoch": 1.3185358255451713, "grad_norm": 2.0732064247131348, "learning_rate": 1.7376352028618877e-06, "loss": 0.09336875379085541, "step": 5079 }, { "epoch": 1.3187954309449637, "grad_norm": 0.9396673440933228, "learning_rate": 1.7367904501708856e-06, "loss": 0.0953366681933403, "step": 5080 }, { "epoch": 1.319055036344756, "grad_norm": 2.0732975006103516, "learning_rate": 1.7359458765921334e-06, "loss": 0.10631662607192993, "step": 5081 }, { "epoch": 1.3193146417445483, "grad_norm": 1.3397228717803955, "learning_rate": 1.7351014822749308e-06, "loss": 0.0919666513800621, "step": 5082 }, { "epoch": 1.3195742471443406, "grad_norm": 2.2015349864959717, "learning_rate": 1.734257267368549e-06, "loss": 0.07008545100688934, "step": 5083 }, { "epoch": 1.3198338525441329, "grad_norm": 2.2104082107543945, "learning_rate": 1.7334132320222244e-06, "loss": 0.10147248208522797, "step": 5084 }, { "epoch": 1.3200934579439252, "grad_norm": 1.1297916173934937, "learning_rate": 1.732569376385163e-06, "loss": 0.10065193474292755, "step": 5085 }, { "epoch": 1.3203530633437175, "grad_norm": 1.120904564857483, "learning_rate": 1.7317257006065407e-06, "loss": 0.10215792059898376, "step": 5086 }, { "epoch": 1.3206126687435098, "grad_norm": 2.1136205196380615, "learning_rate": 1.7308822048354996e-06, "loss": 0.055029094219207764, "step": 5087 }, { "epoch": 1.320872274143302, "grad_norm": 0.626913845539093, "learning_rate": 1.7300388892211483e-06, "loss": 0.10997819155454636, "step": 5088 }, { "epoch": 1.3211318795430946, "grad_norm": 1.303159475326538, "learning_rate": 1.7291957539125686e-06, "loss": 0.10661458969116211, "step": 5089 }, { "epoch": 1.321391484942887, "grad_norm": 1.2194136381149292, "learning_rate": 1.7283527990588055e-06, "loss": 0.08893401175737381, "step": 5090 }, { "epoch": 1.3216510903426792, "grad_norm": 1.8157081604003906, "learning_rate": 1.7275100248088766e-06, "loss": 0.07279831171035767, "step": 5091 }, { "epoch": 1.3219106957424716, "grad_norm": 3.077052593231201, "learning_rate": 1.7266674313117615e-06, "loss": 0.049088750034570694, "step": 5092 }, { "epoch": 1.3221703011422639, "grad_norm": 1.338759183883667, "learning_rate": 1.7258250187164139e-06, "loss": 0.08184511959552765, "step": 5093 }, { "epoch": 1.3224299065420562, "grad_norm": 2.6809616088867188, "learning_rate": 1.7249827871717526e-06, "loss": 0.05343027412891388, "step": 5094 }, { "epoch": 1.3226895119418485, "grad_norm": 1.8992756605148315, "learning_rate": 1.7241407368266649e-06, "loss": 0.07935518026351929, "step": 5095 }, { "epoch": 1.3229491173416408, "grad_norm": 0.6035615801811218, "learning_rate": 1.7232988678300062e-06, "loss": 0.0966990739107132, "step": 5096 }, { "epoch": 1.323208722741433, "grad_norm": 0.8608377575874329, "learning_rate": 1.7224571803305984e-06, "loss": 0.09320635348558426, "step": 5097 }, { "epoch": 1.3234683281412254, "grad_norm": 0.8370407819747925, "learning_rate": 1.721615674477234e-06, "loss": 0.09775302559137344, "step": 5098 }, { "epoch": 1.3237279335410177, "grad_norm": 0.7960683703422546, "learning_rate": 1.720774350418672e-06, "loss": 0.09557199478149414, "step": 5099 }, { "epoch": 1.32398753894081, "grad_norm": 1.0107325315475464, "learning_rate": 1.7199332083036374e-06, "loss": 0.09941849112510681, "step": 5100 }, { "epoch": 1.3242471443406023, "grad_norm": 2.2036619186401367, "learning_rate": 1.719092248280827e-06, "loss": 0.055956434458494186, "step": 5101 }, { "epoch": 1.3245067497403946, "grad_norm": 0.8729590773582458, "learning_rate": 1.7182514704989029e-06, "loss": 0.10018940269947052, "step": 5102 }, { "epoch": 1.324766355140187, "grad_norm": 1.7258538007736206, "learning_rate": 1.7174108751064939e-06, "loss": 0.0869404673576355, "step": 5103 }, { "epoch": 1.3250259605399792, "grad_norm": 0.6424934267997742, "learning_rate": 1.7165704622521994e-06, "loss": 0.09913147985935211, "step": 5104 }, { "epoch": 1.3252855659397715, "grad_norm": 0.8366549611091614, "learning_rate": 1.7157302320845848e-06, "loss": 0.11641500890254974, "step": 5105 }, { "epoch": 1.3255451713395638, "grad_norm": 1.6887822151184082, "learning_rate": 1.7148901847521837e-06, "loss": 0.09409157931804657, "step": 5106 }, { "epoch": 1.3258047767393562, "grad_norm": 0.8267247080802917, "learning_rate": 1.7140503204034973e-06, "loss": 0.1104573905467987, "step": 5107 }, { "epoch": 1.3260643821391485, "grad_norm": 0.6157257556915283, "learning_rate": 1.713210639186993e-06, "loss": 0.10908951610326767, "step": 5108 }, { "epoch": 1.3263239875389408, "grad_norm": 1.765519142150879, "learning_rate": 1.7123711412511096e-06, "loss": 0.10564014315605164, "step": 5109 }, { "epoch": 1.326583592938733, "grad_norm": 0.8858361840248108, "learning_rate": 1.7115318267442496e-06, "loss": 0.10270510613918304, "step": 5110 }, { "epoch": 1.3268431983385254, "grad_norm": 1.8503081798553467, "learning_rate": 1.7106926958147841e-06, "loss": 0.0798049196600914, "step": 5111 }, { "epoch": 1.3271028037383177, "grad_norm": 1.1533552408218384, "learning_rate": 1.7098537486110539e-06, "loss": 0.10608266294002533, "step": 5112 }, { "epoch": 1.32736240913811, "grad_norm": 1.2287570238113403, "learning_rate": 1.709014985281364e-06, "loss": 0.08852748572826385, "step": 5113 }, { "epoch": 1.3276220145379023, "grad_norm": 0.5731212496757507, "learning_rate": 1.70817640597399e-06, "loss": 0.1130814403295517, "step": 5114 }, { "epoch": 1.3278816199376946, "grad_norm": 0.8058143258094788, "learning_rate": 1.7073380108371728e-06, "loss": 0.10701246559619904, "step": 5115 }, { "epoch": 1.328141225337487, "grad_norm": 1.7213112115859985, "learning_rate": 1.7064998000191207e-06, "loss": 0.10510361194610596, "step": 5116 }, { "epoch": 1.3284008307372792, "grad_norm": 0.7133524417877197, "learning_rate": 1.705661773668012e-06, "loss": 0.10905898362398148, "step": 5117 }, { "epoch": 1.3286604361370715, "grad_norm": 1.3097184896469116, "learning_rate": 1.7048239319319894e-06, "loss": 0.11076328903436661, "step": 5118 }, { "epoch": 1.328920041536864, "grad_norm": 1.5647989511489868, "learning_rate": 1.7039862749591635e-06, "loss": 0.0806100070476532, "step": 5119 }, { "epoch": 1.3291796469366564, "grad_norm": 0.951311469078064, "learning_rate": 1.703148802897614e-06, "loss": 0.0998443216085434, "step": 5120 }, { "epoch": 1.3294392523364487, "grad_norm": 0.9836848378181458, "learning_rate": 1.7023115158953862e-06, "loss": 0.09030307829380035, "step": 5121 }, { "epoch": 1.329698857736241, "grad_norm": 2.324127674102783, "learning_rate": 1.7014744141004948e-06, "loss": 0.10553618520498276, "step": 5122 }, { "epoch": 1.3299584631360333, "grad_norm": 1.5289442539215088, "learning_rate": 1.7006374976609174e-06, "loss": 0.06232062727212906, "step": 5123 }, { "epoch": 1.3302180685358256, "grad_norm": 0.6871544718742371, "learning_rate": 1.6998007667246037e-06, "loss": 0.10278233885765076, "step": 5124 }, { "epoch": 1.3304776739356179, "grad_norm": 1.125856637954712, "learning_rate": 1.698964221439469e-06, "loss": 0.10376627743244171, "step": 5125 }, { "epoch": 1.3307372793354102, "grad_norm": 1.6055545806884766, "learning_rate": 1.698127861953393e-06, "loss": 0.11781982332468033, "step": 5126 }, { "epoch": 1.3309968847352025, "grad_norm": 0.9325611591339111, "learning_rate": 1.697291688414227e-06, "loss": 0.08928489685058594, "step": 5127 }, { "epoch": 1.3312564901349948, "grad_norm": 0.9858977198600769, "learning_rate": 1.6964557009697858e-06, "loss": 0.0770261138677597, "step": 5128 }, { "epoch": 1.3315160955347871, "grad_norm": 1.052505612373352, "learning_rate": 1.6956198997678546e-06, "loss": 0.09800462424755096, "step": 5129 }, { "epoch": 1.3317757009345794, "grad_norm": 2.4963691234588623, "learning_rate": 1.6947842849561835e-06, "loss": 0.12416506558656693, "step": 5130 }, { "epoch": 1.3320353063343717, "grad_norm": 1.3947975635528564, "learning_rate": 1.6939488566824885e-06, "loss": 0.10845345258712769, "step": 5131 }, { "epoch": 1.332294911734164, "grad_norm": 2.5681588649749756, "learning_rate": 1.6931136150944561e-06, "loss": 0.06390316039323807, "step": 5132 }, { "epoch": 1.3325545171339563, "grad_norm": 2.36020827293396, "learning_rate": 1.692278560339737e-06, "loss": 0.09923751652240753, "step": 5133 }, { "epoch": 1.3328141225337486, "grad_norm": 2.144134759902954, "learning_rate": 1.6914436925659488e-06, "loss": 0.05589212477207184, "step": 5134 }, { "epoch": 1.333073727933541, "grad_norm": 0.8377406001091003, "learning_rate": 1.6906090119206792e-06, "loss": 0.11083347350358963, "step": 5135 }, { "epoch": 1.3333333333333333, "grad_norm": 1.2027149200439453, "learning_rate": 1.689774518551478e-06, "loss": 0.09256263077259064, "step": 5136 }, { "epoch": 1.3335929387331256, "grad_norm": 0.8404366374015808, "learning_rate": 1.6889402126058674e-06, "loss": 0.09611284732818604, "step": 5137 }, { "epoch": 1.333852544132918, "grad_norm": 1.4954057931900024, "learning_rate": 1.6881060942313322e-06, "loss": 0.061994120478630066, "step": 5138 }, { "epoch": 1.3341121495327104, "grad_norm": 1.4585034847259521, "learning_rate": 1.687272163575324e-06, "loss": 0.11385656893253326, "step": 5139 }, { "epoch": 1.3343717549325027, "grad_norm": 1.7379014492034912, "learning_rate": 1.6864384207852655e-06, "loss": 0.08197532594203949, "step": 5140 }, { "epoch": 1.334631360332295, "grad_norm": 1.3879505395889282, "learning_rate": 1.6856048660085418e-06, "loss": 0.09891419857740402, "step": 5141 }, { "epoch": 1.3348909657320873, "grad_norm": 1.3283171653747559, "learning_rate": 1.6847714993925051e-06, "loss": 0.09594491124153137, "step": 5142 }, { "epoch": 1.3351505711318796, "grad_norm": 1.095567226409912, "learning_rate": 1.6839383210844779e-06, "loss": 0.09103378653526306, "step": 5143 }, { "epoch": 1.335410176531672, "grad_norm": 0.9426709413528442, "learning_rate": 1.683105331231745e-06, "loss": 0.08727887272834778, "step": 5144 }, { "epoch": 1.3356697819314642, "grad_norm": 1.0929019451141357, "learning_rate": 1.6822725299815618e-06, "loss": 0.10534603148698807, "step": 5145 }, { "epoch": 1.3359293873312565, "grad_norm": 3.0080771446228027, "learning_rate": 1.6814399174811475e-06, "loss": 0.09509234130382538, "step": 5146 }, { "epoch": 1.3361889927310489, "grad_norm": 0.8167038559913635, "learning_rate": 1.6806074938776884e-06, "loss": 0.10175721347332001, "step": 5147 }, { "epoch": 1.3364485981308412, "grad_norm": 1.5393602848052979, "learning_rate": 1.6797752593183393e-06, "loss": 0.08106888085603714, "step": 5148 }, { "epoch": 1.3367082035306335, "grad_norm": 2.1178181171417236, "learning_rate": 1.6789432139502192e-06, "loss": 0.08816681057214737, "step": 5149 }, { "epoch": 1.3369678089304258, "grad_norm": 1.2341914176940918, "learning_rate": 1.6781113579204145e-06, "loss": 0.12042249739170074, "step": 5150 }, { "epoch": 1.337227414330218, "grad_norm": 2.0603134632110596, "learning_rate": 1.6772796913759792e-06, "loss": 0.12234324961900711, "step": 5151 }, { "epoch": 1.3374870197300104, "grad_norm": 2.552460193634033, "learning_rate": 1.6764482144639326e-06, "loss": 0.11166740953922272, "step": 5152 }, { "epoch": 1.3377466251298027, "grad_norm": 2.344496488571167, "learning_rate": 1.6756169273312606e-06, "loss": 0.09415430575609207, "step": 5153 }, { "epoch": 1.338006230529595, "grad_norm": 1.882314682006836, "learning_rate": 1.6747858301249148e-06, "loss": 0.0966164618730545, "step": 5154 }, { "epoch": 1.3382658359293873, "grad_norm": 0.7743442058563232, "learning_rate": 1.6739549229918154e-06, "loss": 0.11379647254943848, "step": 5155 }, { "epoch": 1.3385254413291796, "grad_norm": 1.7939151525497437, "learning_rate": 1.6731242060788489e-06, "loss": 0.08646373450756073, "step": 5156 }, { "epoch": 1.338785046728972, "grad_norm": 1.6105319261550903, "learning_rate": 1.6722936795328637e-06, "loss": 0.08467410504817963, "step": 5157 }, { "epoch": 1.3390446521287642, "grad_norm": 2.105311393737793, "learning_rate": 1.671463343500681e-06, "loss": 0.09082774817943573, "step": 5158 }, { "epoch": 1.3393042575285565, "grad_norm": 0.7733020782470703, "learning_rate": 1.670633198129083e-06, "loss": 0.11480225622653961, "step": 5159 }, { "epoch": 1.3395638629283488, "grad_norm": 1.6456936597824097, "learning_rate": 1.669803243564822e-06, "loss": 0.10482099652290344, "step": 5160 }, { "epoch": 1.3398234683281411, "grad_norm": 1.6534948348999023, "learning_rate": 1.6689734799546145e-06, "loss": 0.07752469927072525, "step": 5161 }, { "epoch": 1.3400830737279334, "grad_norm": 2.5140278339385986, "learning_rate": 1.6681439074451424e-06, "loss": 0.10942210257053375, "step": 5162 }, { "epoch": 1.3403426791277258, "grad_norm": 1.11195707321167, "learning_rate": 1.6673145261830577e-06, "loss": 0.08857649564743042, "step": 5163 }, { "epoch": 1.340602284527518, "grad_norm": 1.9925673007965088, "learning_rate": 1.6664853363149739e-06, "loss": 0.06471937149763107, "step": 5164 }, { "epoch": 1.3408618899273104, "grad_norm": 4.28483247756958, "learning_rate": 1.6656563379874724e-06, "loss": 0.07420331239700317, "step": 5165 }, { "epoch": 1.3411214953271027, "grad_norm": 2.6515581607818604, "learning_rate": 1.6648275313471039e-06, "loss": 0.08545859158039093, "step": 5166 }, { "epoch": 1.341381100726895, "grad_norm": 0.7688214778900146, "learning_rate": 1.6639989165403794e-06, "loss": 0.09523023664951324, "step": 5167 }, { "epoch": 1.3416407061266875, "grad_norm": 1.450846552848816, "learning_rate": 1.6631704937137811e-06, "loss": 0.11149273067712784, "step": 5168 }, { "epoch": 1.3419003115264798, "grad_norm": 3.131690263748169, "learning_rate": 1.6623422630137542e-06, "loss": 0.08775751292705536, "step": 5169 }, { "epoch": 1.3421599169262721, "grad_norm": 0.9110947847366333, "learning_rate": 1.6615142245867107e-06, "loss": 0.09957236051559448, "step": 5170 }, { "epoch": 1.3424195223260644, "grad_norm": 1.3665359020233154, "learning_rate": 1.66068637857903e-06, "loss": 0.10950644314289093, "step": 5171 }, { "epoch": 1.3426791277258567, "grad_norm": 1.3102394342422485, "learning_rate": 1.6598587251370557e-06, "loss": 0.08366797864437103, "step": 5172 }, { "epoch": 1.342938733125649, "grad_norm": 1.9560073614120483, "learning_rate": 1.659031264407097e-06, "loss": 0.035155780613422394, "step": 5173 }, { "epoch": 1.3431983385254413, "grad_norm": 2.3364336490631104, "learning_rate": 1.6582039965354313e-06, "loss": 0.1177777573466301, "step": 5174 }, { "epoch": 1.3434579439252337, "grad_norm": 1.9577982425689697, "learning_rate": 1.6573769216683e-06, "loss": 0.09151066839694977, "step": 5175 }, { "epoch": 1.343717549325026, "grad_norm": 1.9738633632659912, "learning_rate": 1.6565500399519116e-06, "loss": 0.06992147117853165, "step": 5176 }, { "epoch": 1.3439771547248183, "grad_norm": 1.635305404663086, "learning_rate": 1.6557233515324398e-06, "loss": 0.08610910177230835, "step": 5177 }, { "epoch": 1.3442367601246106, "grad_norm": 0.8515183329582214, "learning_rate": 1.6548968565560226e-06, "loss": 0.09433291852474213, "step": 5178 }, { "epoch": 1.3444963655244029, "grad_norm": 1.0659033060073853, "learning_rate": 1.6540705551687686e-06, "loss": 0.0942760705947876, "step": 5179 }, { "epoch": 1.3447559709241952, "grad_norm": 1.1450929641723633, "learning_rate": 1.653244447516745e-06, "loss": 0.09696999192237854, "step": 5180 }, { "epoch": 1.3450155763239875, "grad_norm": 2.2936065196990967, "learning_rate": 1.6524185337459909e-06, "loss": 0.046207427978515625, "step": 5181 }, { "epoch": 1.3452751817237798, "grad_norm": 0.9144432544708252, "learning_rate": 1.6515928140025095e-06, "loss": 0.10533938556909561, "step": 5182 }, { "epoch": 1.345534787123572, "grad_norm": 1.101541519165039, "learning_rate": 1.6507672884322685e-06, "loss": 0.10665228962898254, "step": 5183 }, { "epoch": 1.3457943925233644, "grad_norm": 0.8859853148460388, "learning_rate": 1.649941957181202e-06, "loss": 0.09361016750335693, "step": 5184 }, { "epoch": 1.3460539979231567, "grad_norm": 1.307438611984253, "learning_rate": 1.6491168203952085e-06, "loss": 0.08412928879261017, "step": 5185 }, { "epoch": 1.3463136033229492, "grad_norm": 0.9372792840003967, "learning_rate": 1.6482918782201537e-06, "loss": 0.0952037051320076, "step": 5186 }, { "epoch": 1.3465732087227416, "grad_norm": 0.7670215368270874, "learning_rate": 1.6474671308018714e-06, "loss": 0.09970393776893616, "step": 5187 }, { "epoch": 1.3468328141225339, "grad_norm": 2.022919178009033, "learning_rate": 1.646642578286154e-06, "loss": 0.07744358479976654, "step": 5188 }, { "epoch": 1.3470924195223262, "grad_norm": 1.8393374681472778, "learning_rate": 1.645818220818766e-06, "loss": 0.061123259365558624, "step": 5189 }, { "epoch": 1.3473520249221185, "grad_norm": 0.8797036409378052, "learning_rate": 1.644994058545433e-06, "loss": 0.10369568318128586, "step": 5190 }, { "epoch": 1.3476116303219108, "grad_norm": 1.6434391736984253, "learning_rate": 1.6441700916118505e-06, "loss": 0.09690868854522705, "step": 5191 }, { "epoch": 1.347871235721703, "grad_norm": 2.0661776065826416, "learning_rate": 1.6433463201636755e-06, "loss": 0.032894160598516464, "step": 5192 }, { "epoch": 1.3481308411214954, "grad_norm": 1.7711721658706665, "learning_rate": 1.642522744346531e-06, "loss": 0.08364634215831757, "step": 5193 }, { "epoch": 1.3483904465212877, "grad_norm": 2.1514391899108887, "learning_rate": 1.6416993643060081e-06, "loss": 0.08405034244060516, "step": 5194 }, { "epoch": 1.34865005192108, "grad_norm": 1.9052571058273315, "learning_rate": 1.6408761801876612e-06, "loss": 0.09769954532384872, "step": 5195 }, { "epoch": 1.3489096573208723, "grad_norm": 2.0501556396484375, "learning_rate": 1.6400531921370095e-06, "loss": 0.08526705205440521, "step": 5196 }, { "epoch": 1.3491692627206646, "grad_norm": 2.257371425628662, "learning_rate": 1.6392304002995397e-06, "loss": 0.10113078355789185, "step": 5197 }, { "epoch": 1.349428868120457, "grad_norm": 0.7516101002693176, "learning_rate": 1.6384078048207008e-06, "loss": 0.10249299556016922, "step": 5198 }, { "epoch": 1.3496884735202492, "grad_norm": 1.5096698999404907, "learning_rate": 1.6375854058459107e-06, "loss": 0.09561985731124878, "step": 5199 }, { "epoch": 1.3499480789200415, "grad_norm": 1.9372674226760864, "learning_rate": 1.63676320352055e-06, "loss": 0.09782135486602783, "step": 5200 }, { "epoch": 1.3502076843198338, "grad_norm": 1.157460331916809, "learning_rate": 1.6359411979899648e-06, "loss": 0.09556093066930771, "step": 5201 }, { "epoch": 1.3504672897196262, "grad_norm": 1.4310044050216675, "learning_rate": 1.6351193893994675e-06, "loss": 0.10462254285812378, "step": 5202 }, { "epoch": 1.3507268951194185, "grad_norm": 1.2048358917236328, "learning_rate": 1.6342977778943353e-06, "loss": 0.08448949456214905, "step": 5203 }, { "epoch": 1.3509865005192108, "grad_norm": 1.2924342155456543, "learning_rate": 1.6334763636198083e-06, "loss": 0.07384341955184937, "step": 5204 }, { "epoch": 1.351246105919003, "grad_norm": 1.3329856395721436, "learning_rate": 1.6326551467210965e-06, "loss": 0.09145724773406982, "step": 5205 }, { "epoch": 1.3515057113187954, "grad_norm": 2.029691457748413, "learning_rate": 1.6318341273433704e-06, "loss": 0.09485410153865814, "step": 5206 }, { "epoch": 1.3517653167185877, "grad_norm": 0.9505450129508972, "learning_rate": 1.6310133056317678e-06, "loss": 0.08858853578567505, "step": 5207 }, { "epoch": 1.35202492211838, "grad_norm": 1.6214860677719116, "learning_rate": 1.6301926817313923e-06, "loss": 0.06260433048009872, "step": 5208 }, { "epoch": 1.3522845275181723, "grad_norm": 1.4838706254959106, "learning_rate": 1.629372255787309e-06, "loss": 0.08851490914821625, "step": 5209 }, { "epoch": 1.3525441329179646, "grad_norm": 1.023572325706482, "learning_rate": 1.6285520279445539e-06, "loss": 0.09436215460300446, "step": 5210 }, { "epoch": 1.352803738317757, "grad_norm": 1.8122189044952393, "learning_rate": 1.627731998348121e-06, "loss": 0.09527707099914551, "step": 5211 }, { "epoch": 1.3530633437175492, "grad_norm": 2.259655475616455, "learning_rate": 1.626912167142974e-06, "loss": 0.07758316397666931, "step": 5212 }, { "epoch": 1.3533229491173415, "grad_norm": 1.9019447565078735, "learning_rate": 1.6260925344740414e-06, "loss": 0.08106313645839691, "step": 5213 }, { "epoch": 1.3535825545171338, "grad_norm": 1.3832517862319946, "learning_rate": 1.6252731004862149e-06, "loss": 0.07999657839536667, "step": 5214 }, { "epoch": 1.3538421599169261, "grad_norm": 1.6756677627563477, "learning_rate": 1.6244538653243508e-06, "loss": 0.0786684900522232, "step": 5215 }, { "epoch": 1.3541017653167187, "grad_norm": 1.8212919235229492, "learning_rate": 1.6236348291332716e-06, "loss": 0.09053952991962433, "step": 5216 }, { "epoch": 1.354361370716511, "grad_norm": 0.9490439891815186, "learning_rate": 1.6228159920577638e-06, "loss": 0.09845992922782898, "step": 5217 }, { "epoch": 1.3546209761163033, "grad_norm": 0.818274736404419, "learning_rate": 1.6219973542425816e-06, "loss": 0.09179829806089401, "step": 5218 }, { "epoch": 1.3548805815160956, "grad_norm": 1.9367988109588623, "learning_rate": 1.621178915832437e-06, "loss": 0.08927646279335022, "step": 5219 }, { "epoch": 1.355140186915888, "grad_norm": 0.7724813222885132, "learning_rate": 1.6203606769720148e-06, "loss": 0.10675030946731567, "step": 5220 }, { "epoch": 1.3553997923156802, "grad_norm": 1.1198976039886475, "learning_rate": 1.6195426378059587e-06, "loss": 0.11119500547647476, "step": 5221 }, { "epoch": 1.3556593977154725, "grad_norm": 1.0130336284637451, "learning_rate": 1.6187247984788811e-06, "loss": 0.10804849117994308, "step": 5222 }, { "epoch": 1.3559190031152648, "grad_norm": 2.0664308071136475, "learning_rate": 1.6179071591353563e-06, "loss": 0.0973174124956131, "step": 5223 }, { "epoch": 1.3561786085150571, "grad_norm": 1.9545090198516846, "learning_rate": 1.6170897199199231e-06, "loss": 0.06337320059537888, "step": 5224 }, { "epoch": 1.3564382139148494, "grad_norm": 0.8373878002166748, "learning_rate": 1.6162724809770885e-06, "loss": 0.08497676253318787, "step": 5225 }, { "epoch": 1.3566978193146417, "grad_norm": 1.7478277683258057, "learning_rate": 1.6154554424513203e-06, "loss": 0.09011715650558472, "step": 5226 }, { "epoch": 1.356957424714434, "grad_norm": 1.7414262294769287, "learning_rate": 1.6146386044870514e-06, "loss": 0.061612553894519806, "step": 5227 }, { "epoch": 1.3572170301142263, "grad_norm": 1.0941442251205444, "learning_rate": 1.6138219672286814e-06, "loss": 0.08595485240221024, "step": 5228 }, { "epoch": 1.3574766355140186, "grad_norm": 2.4693243503570557, "learning_rate": 1.6130055308205723e-06, "loss": 0.08636277914047241, "step": 5229 }, { "epoch": 1.357736240913811, "grad_norm": 0.8190666437149048, "learning_rate": 1.612189295407052e-06, "loss": 0.09378372132778168, "step": 5230 }, { "epoch": 1.3579958463136033, "grad_norm": 0.9679145216941833, "learning_rate": 1.6113732611324118e-06, "loss": 0.09781187772750854, "step": 5231 }, { "epoch": 1.3582554517133956, "grad_norm": 1.9354196786880493, "learning_rate": 1.6105574281409073e-06, "loss": 0.053648486733436584, "step": 5232 }, { "epoch": 1.3585150571131879, "grad_norm": 1.1642709970474243, "learning_rate": 1.6097417965767608e-06, "loss": 0.08495347201824188, "step": 5233 }, { "epoch": 1.3587746625129804, "grad_norm": 0.6860016584396362, "learning_rate": 1.608926366584156e-06, "loss": 0.10500258207321167, "step": 5234 }, { "epoch": 1.3590342679127727, "grad_norm": 2.436781883239746, "learning_rate": 1.6081111383072414e-06, "loss": 0.06871400773525238, "step": 5235 }, { "epoch": 1.359293873312565, "grad_norm": 0.939520537853241, "learning_rate": 1.6072961118901328e-06, "loss": 0.09508030116558075, "step": 5236 }, { "epoch": 1.3595534787123573, "grad_norm": 0.7377943992614746, "learning_rate": 1.6064812874769073e-06, "loss": 0.09945444762706757, "step": 5237 }, { "epoch": 1.3598130841121496, "grad_norm": 1.1906712055206299, "learning_rate": 1.605666665211606e-06, "loss": 0.0784301906824112, "step": 5238 }, { "epoch": 1.360072689511942, "grad_norm": 0.894995391368866, "learning_rate": 1.6048522452382375e-06, "loss": 0.08597405254840851, "step": 5239 }, { "epoch": 1.3603322949117342, "grad_norm": 1.3395514488220215, "learning_rate": 1.604038027700771e-06, "loss": 0.12227277457714081, "step": 5240 }, { "epoch": 1.3605919003115265, "grad_norm": 0.9950277805328369, "learning_rate": 1.6032240127431433e-06, "loss": 0.10547386109828949, "step": 5241 }, { "epoch": 1.3608515057113189, "grad_norm": 1.499782681465149, "learning_rate": 1.6024102005092513e-06, "loss": 0.10403885692358017, "step": 5242 }, { "epoch": 1.3611111111111112, "grad_norm": 3.0387635231018066, "learning_rate": 1.6015965911429593e-06, "loss": 0.0830051600933075, "step": 5243 }, { "epoch": 1.3613707165109035, "grad_norm": 0.9191167950630188, "learning_rate": 1.600783184788096e-06, "loss": 0.11374545097351074, "step": 5244 }, { "epoch": 1.3616303219106958, "grad_norm": 0.9000369310379028, "learning_rate": 1.5999699815884518e-06, "loss": 0.10265921801328659, "step": 5245 }, { "epoch": 1.361889927310488, "grad_norm": 1.4850963354110718, "learning_rate": 1.5991569816877827e-06, "loss": 0.08774083852767944, "step": 5246 }, { "epoch": 1.3621495327102804, "grad_norm": 0.9092705249786377, "learning_rate": 1.5983441852298075e-06, "loss": 0.0923246294260025, "step": 5247 }, { "epoch": 1.3624091381100727, "grad_norm": 1.0964990854263306, "learning_rate": 1.5975315923582113e-06, "loss": 0.11152797937393188, "step": 5248 }, { "epoch": 1.362668743509865, "grad_norm": 1.0379050970077515, "learning_rate": 1.5967192032166429e-06, "loss": 0.09389062970876694, "step": 5249 }, { "epoch": 1.3629283489096573, "grad_norm": 1.5395863056182861, "learning_rate": 1.5959070179487112e-06, "loss": 0.056361790746450424, "step": 5250 }, { "epoch": 1.3631879543094496, "grad_norm": 1.9314711093902588, "learning_rate": 1.5950950366979943e-06, "loss": 0.08462906628847122, "step": 5251 }, { "epoch": 1.363447559709242, "grad_norm": 1.6968573331832886, "learning_rate": 1.5942832596080304e-06, "loss": 0.07912393659353256, "step": 5252 }, { "epoch": 1.3637071651090342, "grad_norm": 1.8183726072311401, "learning_rate": 1.5934716868223244e-06, "loss": 0.10236421227455139, "step": 5253 }, { "epoch": 1.3639667705088265, "grad_norm": 1.6404143571853638, "learning_rate": 1.592660318484343e-06, "loss": 0.07448167353868484, "step": 5254 }, { "epoch": 1.3642263759086188, "grad_norm": 1.312504768371582, "learning_rate": 1.5918491547375175e-06, "loss": 0.0959487035870552, "step": 5255 }, { "epoch": 1.3644859813084111, "grad_norm": 2.935727596282959, "learning_rate": 1.591038195725244e-06, "loss": 0.10250979661941528, "step": 5256 }, { "epoch": 1.3647455867082035, "grad_norm": 0.7919148802757263, "learning_rate": 1.5902274415908802e-06, "loss": 0.10060763359069824, "step": 5257 }, { "epoch": 1.3650051921079958, "grad_norm": 0.9971367716789246, "learning_rate": 1.5894168924777493e-06, "loss": 0.10984253138303757, "step": 5258 }, { "epoch": 1.365264797507788, "grad_norm": 1.1945204734802246, "learning_rate": 1.5886065485291385e-06, "loss": 0.10748639702796936, "step": 5259 }, { "epoch": 1.3655244029075804, "grad_norm": 1.4561413526535034, "learning_rate": 1.587796409888297e-06, "loss": 0.10533461719751358, "step": 5260 }, { "epoch": 1.3657840083073727, "grad_norm": 1.4144396781921387, "learning_rate": 1.5869864766984397e-06, "loss": 0.09166991710662842, "step": 5261 }, { "epoch": 1.366043613707165, "grad_norm": 2.0766525268554688, "learning_rate": 1.586176749102744e-06, "loss": 0.09751869738101959, "step": 5262 }, { "epoch": 1.3663032191069573, "grad_norm": 1.068116545677185, "learning_rate": 1.5853672272443499e-06, "loss": 0.12196813523769379, "step": 5263 }, { "epoch": 1.3665628245067498, "grad_norm": 1.053043007850647, "learning_rate": 1.5845579112663651e-06, "loss": 0.10514704883098602, "step": 5264 }, { "epoch": 1.3668224299065421, "grad_norm": 2.0521156787872314, "learning_rate": 1.5837488013118558e-06, "loss": 0.04807126522064209, "step": 5265 }, { "epoch": 1.3670820353063344, "grad_norm": 1.0332788228988647, "learning_rate": 1.5829398975238542e-06, "loss": 0.08831143379211426, "step": 5266 }, { "epoch": 1.3673416407061267, "grad_norm": 2.030405282974243, "learning_rate": 1.5821312000453575e-06, "loss": 0.08792868256568909, "step": 5267 }, { "epoch": 1.367601246105919, "grad_norm": 1.917402744293213, "learning_rate": 1.581322709019324e-06, "loss": 0.10460501909255981, "step": 5268 }, { "epoch": 1.3678608515057114, "grad_norm": 2.2389912605285645, "learning_rate": 1.580514424588675e-06, "loss": 0.06114555895328522, "step": 5269 }, { "epoch": 1.3681204569055037, "grad_norm": 2.009392023086548, "learning_rate": 1.5797063468962998e-06, "loss": 0.0858621597290039, "step": 5270 }, { "epoch": 1.368380062305296, "grad_norm": 2.2687714099884033, "learning_rate": 1.5788984760850445e-06, "loss": 0.06650868058204651, "step": 5271 }, { "epoch": 1.3686396677050883, "grad_norm": 2.0304744243621826, "learning_rate": 1.5780908122977265e-06, "loss": 0.11351649463176727, "step": 5272 }, { "epoch": 1.3688992731048806, "grad_norm": 2.2466769218444824, "learning_rate": 1.5772833556771177e-06, "loss": 0.06695536524057388, "step": 5273 }, { "epoch": 1.3691588785046729, "grad_norm": 0.6564857363700867, "learning_rate": 1.57647610636596e-06, "loss": 0.09580932557582855, "step": 5274 }, { "epoch": 1.3694184839044652, "grad_norm": 1.8325858116149902, "learning_rate": 1.5756690645069573e-06, "loss": 0.06989321112632751, "step": 5275 }, { "epoch": 1.3696780893042575, "grad_norm": 1.5046415328979492, "learning_rate": 1.5748622302427755e-06, "loss": 0.11971612274646759, "step": 5276 }, { "epoch": 1.3699376947040498, "grad_norm": 2.2593352794647217, "learning_rate": 1.5740556037160443e-06, "loss": 0.1489749401807785, "step": 5277 }, { "epoch": 1.370197300103842, "grad_norm": 1.3308252096176147, "learning_rate": 1.5732491850693557e-06, "loss": 0.09188304096460342, "step": 5278 }, { "epoch": 1.3704569055036344, "grad_norm": 1.2880947589874268, "learning_rate": 1.5724429744452674e-06, "loss": 0.08629830926656723, "step": 5279 }, { "epoch": 1.3707165109034267, "grad_norm": 1.215866208076477, "learning_rate": 1.5716369719863003e-06, "loss": 0.11703748255968094, "step": 5280 }, { "epoch": 1.370976116303219, "grad_norm": 1.9275444746017456, "learning_rate": 1.5708311778349341e-06, "loss": 0.08642884343862534, "step": 5281 }, { "epoch": 1.3712357217030116, "grad_norm": 1.9557236433029175, "learning_rate": 1.5700255921336168e-06, "loss": 0.10253143310546875, "step": 5282 }, { "epoch": 1.3714953271028039, "grad_norm": 1.9282760620117188, "learning_rate": 1.5692202150247566e-06, "loss": 0.09547386318445206, "step": 5283 }, { "epoch": 1.3717549325025962, "grad_norm": 0.9743326306343079, "learning_rate": 1.5684150466507263e-06, "loss": 0.11497433483600616, "step": 5284 }, { "epoch": 1.3720145379023885, "grad_norm": 0.8976438641548157, "learning_rate": 1.567610087153862e-06, "loss": 0.10132530331611633, "step": 5285 }, { "epoch": 1.3722741433021808, "grad_norm": 0.9340794682502747, "learning_rate": 1.5668053366764596e-06, "loss": 0.09603682160377502, "step": 5286 }, { "epoch": 1.372533748701973, "grad_norm": 1.8227587938308716, "learning_rate": 1.566000795360783e-06, "loss": 0.07815256714820862, "step": 5287 }, { "epoch": 1.3727933541017654, "grad_norm": 0.8755418658256531, "learning_rate": 1.565196463349056e-06, "loss": 0.09156768769025803, "step": 5288 }, { "epoch": 1.3730529595015577, "grad_norm": 1.7193381786346436, "learning_rate": 1.564392340783465e-06, "loss": 0.07587450742721558, "step": 5289 }, { "epoch": 1.37331256490135, "grad_norm": 1.5868970155715942, "learning_rate": 1.563588427806162e-06, "loss": 0.07615289092063904, "step": 5290 }, { "epoch": 1.3735721703011423, "grad_norm": 1.322640299797058, "learning_rate": 1.5627847245592598e-06, "loss": 0.1151583269238472, "step": 5291 }, { "epoch": 1.3738317757009346, "grad_norm": 1.0377113819122314, "learning_rate": 1.5619812311848336e-06, "loss": 0.10647653043270111, "step": 5292 }, { "epoch": 1.374091381100727, "grad_norm": 1.1660685539245605, "learning_rate": 1.561177947824925e-06, "loss": 0.10547361522912979, "step": 5293 }, { "epoch": 1.3743509865005192, "grad_norm": 2.063826084136963, "learning_rate": 1.5603748746215338e-06, "loss": 0.08137871325016022, "step": 5294 }, { "epoch": 1.3746105919003115, "grad_norm": 0.6970618963241577, "learning_rate": 1.5595720117166268e-06, "loss": 0.1011979803442955, "step": 5295 }, { "epoch": 1.3748701973001038, "grad_norm": 1.9414372444152832, "learning_rate": 1.5587693592521307e-06, "loss": 0.0806158259510994, "step": 5296 }, { "epoch": 1.3751298026998962, "grad_norm": 1.9362478256225586, "learning_rate": 1.557966917369936e-06, "loss": 0.08045521378517151, "step": 5297 }, { "epoch": 1.3753894080996885, "grad_norm": 2.6395974159240723, "learning_rate": 1.5571646862118968e-06, "loss": 0.05990555137395859, "step": 5298 }, { "epoch": 1.3756490134994808, "grad_norm": 2.001452922821045, "learning_rate": 1.5563626659198289e-06, "loss": 0.08175022155046463, "step": 5299 }, { "epoch": 1.375908618899273, "grad_norm": 0.9516861438751221, "learning_rate": 1.55556085663551e-06, "loss": 0.08928598463535309, "step": 5300 }, { "epoch": 1.3761682242990654, "grad_norm": 3.2923216819763184, "learning_rate": 1.5547592585006839e-06, "loss": 0.09715556353330612, "step": 5301 }, { "epoch": 1.3764278296988577, "grad_norm": 1.1962944269180298, "learning_rate": 1.5539578716570524e-06, "loss": 0.09151912480592728, "step": 5302 }, { "epoch": 1.37668743509865, "grad_norm": 0.7691511511802673, "learning_rate": 1.5531566962462853e-06, "loss": 0.10224343836307526, "step": 5303 }, { "epoch": 1.3769470404984423, "grad_norm": 2.176278829574585, "learning_rate": 1.552355732410008e-06, "loss": 0.09286494553089142, "step": 5304 }, { "epoch": 1.3772066458982346, "grad_norm": 2.0870230197906494, "learning_rate": 1.551554980289815e-06, "loss": 0.08479846268892288, "step": 5305 }, { "epoch": 1.377466251298027, "grad_norm": 1.146116018295288, "learning_rate": 1.550754440027262e-06, "loss": 0.10096123814582825, "step": 5306 }, { "epoch": 1.3777258566978192, "grad_norm": 2.647033214569092, "learning_rate": 1.5499541117638645e-06, "loss": 0.08897694945335388, "step": 5307 }, { "epoch": 1.3779854620976115, "grad_norm": 0.8056673407554626, "learning_rate": 1.5491539956411027e-06, "loss": 0.10310237109661102, "step": 5308 }, { "epoch": 1.3782450674974038, "grad_norm": 1.8601597547531128, "learning_rate": 1.548354091800418e-06, "loss": 0.07255491614341736, "step": 5309 }, { "epoch": 1.3785046728971961, "grad_norm": 0.9988092184066772, "learning_rate": 1.5475544003832161e-06, "loss": 0.09706442058086395, "step": 5310 }, { "epoch": 1.3787642782969884, "grad_norm": 0.9287011623382568, "learning_rate": 1.546754921530865e-06, "loss": 0.08855310082435608, "step": 5311 }, { "epoch": 1.379023883696781, "grad_norm": 1.6118525266647339, "learning_rate": 1.5459556553846915e-06, "loss": 0.0764891505241394, "step": 5312 }, { "epoch": 1.3792834890965733, "grad_norm": 1.6333945989608765, "learning_rate": 1.5451566020859903e-06, "loss": 0.10151085257530212, "step": 5313 }, { "epoch": 1.3795430944963656, "grad_norm": 1.7912601232528687, "learning_rate": 1.5443577617760137e-06, "loss": 0.08309335261583328, "step": 5314 }, { "epoch": 1.379802699896158, "grad_norm": 0.7820817232131958, "learning_rate": 1.54355913459598e-06, "loss": 0.1008792519569397, "step": 5315 }, { "epoch": 1.3800623052959502, "grad_norm": 2.130380153656006, "learning_rate": 1.5427607206870677e-06, "loss": 0.08539221435785294, "step": 5316 }, { "epoch": 1.3803219106957425, "grad_norm": 1.5997415781021118, "learning_rate": 1.5419625201904164e-06, "loss": 0.0900307297706604, "step": 5317 }, { "epoch": 1.3805815160955348, "grad_norm": 1.0550719499588013, "learning_rate": 1.5411645332471326e-06, "loss": 0.09674093127250671, "step": 5318 }, { "epoch": 1.3808411214953271, "grad_norm": 1.5718226432800293, "learning_rate": 1.5403667599982808e-06, "loss": 0.09042906761169434, "step": 5319 }, { "epoch": 1.3811007268951194, "grad_norm": 3.699394941329956, "learning_rate": 1.5395692005848876e-06, "loss": 0.043552882969379425, "step": 5320 }, { "epoch": 1.3813603322949117, "grad_norm": 1.256629467010498, "learning_rate": 1.5387718551479452e-06, "loss": 0.09348945319652557, "step": 5321 }, { "epoch": 1.381619937694704, "grad_norm": 2.6781766414642334, "learning_rate": 1.5379747238284059e-06, "loss": 0.04824291542172432, "step": 5322 }, { "epoch": 1.3818795430944963, "grad_norm": 2.00785231590271, "learning_rate": 1.5371778067671826e-06, "loss": 0.06002098321914673, "step": 5323 }, { "epoch": 1.3821391484942886, "grad_norm": 1.7371656894683838, "learning_rate": 1.5363811041051541e-06, "loss": 0.07477635145187378, "step": 5324 }, { "epoch": 1.382398753894081, "grad_norm": 0.9483927488327026, "learning_rate": 1.5355846159831572e-06, "loss": 0.09743642807006836, "step": 5325 }, { "epoch": 1.3826583592938733, "grad_norm": 2.924588441848755, "learning_rate": 1.5347883425419945e-06, "loss": 0.1021377295255661, "step": 5326 }, { "epoch": 1.3829179646936656, "grad_norm": 1.9400163888931274, "learning_rate": 1.5339922839224285e-06, "loss": 0.07206069678068161, "step": 5327 }, { "epoch": 1.3831775700934579, "grad_norm": 1.275150179862976, "learning_rate": 1.5331964402651827e-06, "loss": 0.09521915018558502, "step": 5328 }, { "epoch": 1.3834371754932502, "grad_norm": 1.7754157781600952, "learning_rate": 1.5324008117109463e-06, "loss": 0.0778515562415123, "step": 5329 }, { "epoch": 1.3836967808930427, "grad_norm": 1.436007022857666, "learning_rate": 1.531605398400367e-06, "loss": 0.10987070947885513, "step": 5330 }, { "epoch": 1.383956386292835, "grad_norm": 2.551018714904785, "learning_rate": 1.5308102004740549e-06, "loss": 0.09128344058990479, "step": 5331 }, { "epoch": 1.3842159916926273, "grad_norm": 0.8184898495674133, "learning_rate": 1.530015218072584e-06, "loss": 0.1087726280093193, "step": 5332 }, { "epoch": 1.3844755970924196, "grad_norm": 1.4671196937561035, "learning_rate": 1.5292204513364884e-06, "loss": 0.07999056577682495, "step": 5333 }, { "epoch": 1.384735202492212, "grad_norm": 0.8564237952232361, "learning_rate": 1.528425900406266e-06, "loss": 0.10029932856559753, "step": 5334 }, { "epoch": 1.3849948078920042, "grad_norm": 1.234570860862732, "learning_rate": 1.5276315654223728e-06, "loss": 0.10210496187210083, "step": 5335 }, { "epoch": 1.3852544132917965, "grad_norm": 0.8265160918235779, "learning_rate": 1.5268374465252301e-06, "loss": 0.10931813716888428, "step": 5336 }, { "epoch": 1.3855140186915889, "grad_norm": 1.1566625833511353, "learning_rate": 1.5260435438552208e-06, "loss": 0.09879037737846375, "step": 5337 }, { "epoch": 1.3857736240913812, "grad_norm": 0.9371954202651978, "learning_rate": 1.5252498575526887e-06, "loss": 0.08737345784902573, "step": 5338 }, { "epoch": 1.3860332294911735, "grad_norm": 0.8400828242301941, "learning_rate": 1.524456387757938e-06, "loss": 0.09089764952659607, "step": 5339 }, { "epoch": 1.3862928348909658, "grad_norm": 1.9536831378936768, "learning_rate": 1.5236631346112364e-06, "loss": 0.06338811665773392, "step": 5340 }, { "epoch": 1.386552440290758, "grad_norm": 1.9611624479293823, "learning_rate": 1.5228700982528127e-06, "loss": 0.07977970689535141, "step": 5341 }, { "epoch": 1.3868120456905504, "grad_norm": 1.5089588165283203, "learning_rate": 1.5220772788228599e-06, "loss": 0.09724568575620651, "step": 5342 }, { "epoch": 1.3870716510903427, "grad_norm": 1.4557099342346191, "learning_rate": 1.5212846764615271e-06, "loss": 0.09766631573438644, "step": 5343 }, { "epoch": 1.387331256490135, "grad_norm": 1.6862422227859497, "learning_rate": 1.5204922913089305e-06, "loss": 0.08499829471111298, "step": 5344 }, { "epoch": 1.3875908618899273, "grad_norm": 1.9048212766647339, "learning_rate": 1.5197001235051446e-06, "loss": 0.07416325807571411, "step": 5345 }, { "epoch": 1.3878504672897196, "grad_norm": 0.9625958800315857, "learning_rate": 1.5189081731902062e-06, "loss": 0.10114744305610657, "step": 5346 }, { "epoch": 1.388110072689512, "grad_norm": 1.2334425449371338, "learning_rate": 1.5181164405041149e-06, "loss": 0.09187227487564087, "step": 5347 }, { "epoch": 1.3883696780893042, "grad_norm": 0.9384467601776123, "learning_rate": 1.5173249255868299e-06, "loss": 0.0901578813791275, "step": 5348 }, { "epoch": 1.3886292834890965, "grad_norm": 0.9071186184883118, "learning_rate": 1.5165336285782739e-06, "loss": 0.0837501585483551, "step": 5349 }, { "epoch": 1.3888888888888888, "grad_norm": 3.065178394317627, "learning_rate": 1.5157425496183303e-06, "loss": 0.08811558783054352, "step": 5350 }, { "epoch": 1.3891484942886811, "grad_norm": 2.2226476669311523, "learning_rate": 1.5149516888468423e-06, "loss": 0.06266838312149048, "step": 5351 }, { "epoch": 1.3894080996884735, "grad_norm": 1.6323785781860352, "learning_rate": 1.5141610464036177e-06, "loss": 0.06715177744626999, "step": 5352 }, { "epoch": 1.3896677050882658, "grad_norm": 2.0664985179901123, "learning_rate": 1.513370622428423e-06, "loss": 0.10079899430274963, "step": 5353 }, { "epoch": 1.389927310488058, "grad_norm": 1.6231249570846558, "learning_rate": 1.5125804170609865e-06, "loss": 0.08284871280193329, "step": 5354 }, { "epoch": 1.3901869158878504, "grad_norm": 1.465438723564148, "learning_rate": 1.5117904304409997e-06, "loss": 0.09125905483961105, "step": 5355 }, { "epoch": 1.3904465212876427, "grad_norm": 1.257238745689392, "learning_rate": 1.511000662708113e-06, "loss": 0.11321589350700378, "step": 5356 }, { "epoch": 1.390706126687435, "grad_norm": 1.7694942951202393, "learning_rate": 1.5102111140019404e-06, "loss": 0.07653352618217468, "step": 5357 }, { "epoch": 1.3909657320872273, "grad_norm": 2.0747101306915283, "learning_rate": 1.5094217844620558e-06, "loss": 0.09925009310245514, "step": 5358 }, { "epoch": 1.3912253374870196, "grad_norm": 1.825908899307251, "learning_rate": 1.508632674227993e-06, "loss": 0.10456857085227966, "step": 5359 }, { "epoch": 1.3914849428868121, "grad_norm": 1.517425298690796, "learning_rate": 1.5078437834392508e-06, "loss": 0.10401608794927597, "step": 5360 }, { "epoch": 1.3917445482866044, "grad_norm": 2.4257869720458984, "learning_rate": 1.507055112235286e-06, "loss": 0.10294854640960693, "step": 5361 }, { "epoch": 1.3920041536863967, "grad_norm": 1.0925437211990356, "learning_rate": 1.5062666607555175e-06, "loss": 0.09646797180175781, "step": 5362 }, { "epoch": 1.392263759086189, "grad_norm": 0.8560322523117065, "learning_rate": 1.505478429139326e-06, "loss": 0.11430742591619492, "step": 5363 }, { "epoch": 1.3925233644859814, "grad_norm": 1.9971673488616943, "learning_rate": 1.5046904175260513e-06, "loss": 0.08067814260721207, "step": 5364 }, { "epoch": 1.3927829698857737, "grad_norm": 2.29315185546875, "learning_rate": 1.5039026260549988e-06, "loss": 0.08255042135715485, "step": 5365 }, { "epoch": 1.393042575285566, "grad_norm": 0.6962767243385315, "learning_rate": 1.5031150548654288e-06, "loss": 0.10116444528102875, "step": 5366 }, { "epoch": 1.3933021806853583, "grad_norm": 1.1614059209823608, "learning_rate": 1.502327704096567e-06, "loss": 0.1075403094291687, "step": 5367 }, { "epoch": 1.3935617860851506, "grad_norm": 1.5055485963821411, "learning_rate": 1.5015405738876002e-06, "loss": 0.06693601608276367, "step": 5368 }, { "epoch": 1.3938213914849429, "grad_norm": 1.6055458784103394, "learning_rate": 1.5007536643776737e-06, "loss": 0.1038282960653305, "step": 5369 }, { "epoch": 1.3940809968847352, "grad_norm": 1.4672183990478516, "learning_rate": 1.4999669757058956e-06, "loss": 0.04995633661746979, "step": 5370 }, { "epoch": 1.3943406022845275, "grad_norm": 0.9908756613731384, "learning_rate": 1.4991805080113337e-06, "loss": 0.09213865548372269, "step": 5371 }, { "epoch": 1.3946002076843198, "grad_norm": 1.9435248374938965, "learning_rate": 1.4983942614330182e-06, "loss": 0.08002284169197083, "step": 5372 }, { "epoch": 1.394859813084112, "grad_norm": 0.9205184578895569, "learning_rate": 1.4976082361099397e-06, "loss": 0.10067189484834671, "step": 5373 }, { "epoch": 1.3951194184839044, "grad_norm": 2.362337589263916, "learning_rate": 1.496822432181048e-06, "loss": 0.08225482702255249, "step": 5374 }, { "epoch": 1.3953790238836967, "grad_norm": 0.789635956287384, "learning_rate": 1.4960368497852568e-06, "loss": 0.10535048693418503, "step": 5375 }, { "epoch": 1.395638629283489, "grad_norm": 1.163055181503296, "learning_rate": 1.49525148906144e-06, "loss": 0.08684976398944855, "step": 5376 }, { "epoch": 1.3958982346832813, "grad_norm": 0.8803633451461792, "learning_rate": 1.4944663501484285e-06, "loss": 0.10775931179523468, "step": 5377 }, { "epoch": 1.3961578400830739, "grad_norm": 1.314159631729126, "learning_rate": 1.4936814331850188e-06, "loss": 0.08210548013448715, "step": 5378 }, { "epoch": 1.3964174454828662, "grad_norm": 0.9833315014839172, "learning_rate": 1.4928967383099655e-06, "loss": 0.09883569180965424, "step": 5379 }, { "epoch": 1.3966770508826585, "grad_norm": 1.122422456741333, "learning_rate": 1.4921122656619855e-06, "loss": 0.08415412902832031, "step": 5380 }, { "epoch": 1.3969366562824508, "grad_norm": 1.1602107286453247, "learning_rate": 1.4913280153797553e-06, "loss": 0.09511256217956543, "step": 5381 }, { "epoch": 1.397196261682243, "grad_norm": 0.8578269481658936, "learning_rate": 1.4905439876019113e-06, "loss": 0.10349045693874359, "step": 5382 }, { "epoch": 1.3974558670820354, "grad_norm": 1.8347517251968384, "learning_rate": 1.4897601824670534e-06, "loss": 0.08622072637081146, "step": 5383 }, { "epoch": 1.3977154724818277, "grad_norm": 1.8625491857528687, "learning_rate": 1.4889766001137395e-06, "loss": 0.08015888184309006, "step": 5384 }, { "epoch": 1.39797507788162, "grad_norm": 0.9165411591529846, "learning_rate": 1.4881932406804883e-06, "loss": 0.09579126536846161, "step": 5385 }, { "epoch": 1.3982346832814123, "grad_norm": 1.3995438814163208, "learning_rate": 1.4874101043057814e-06, "loss": 0.09475546330213547, "step": 5386 }, { "epoch": 1.3984942886812046, "grad_norm": 3.2025222778320312, "learning_rate": 1.4866271911280578e-06, "loss": 0.10142166912555695, "step": 5387 }, { "epoch": 1.398753894080997, "grad_norm": 2.7151834964752197, "learning_rate": 1.4858445012857203e-06, "loss": 0.0963984876871109, "step": 5388 }, { "epoch": 1.3990134994807892, "grad_norm": 1.2822718620300293, "learning_rate": 1.4850620349171297e-06, "loss": 0.09304049611091614, "step": 5389 }, { "epoch": 1.3992731048805815, "grad_norm": 0.9310765862464905, "learning_rate": 1.4842797921606073e-06, "loss": 0.10171209275722504, "step": 5390 }, { "epoch": 1.3995327102803738, "grad_norm": 2.417120933532715, "learning_rate": 1.483497773154437e-06, "loss": 0.10138674825429916, "step": 5391 }, { "epoch": 1.3997923156801662, "grad_norm": 2.0394558906555176, "learning_rate": 1.4827159780368616e-06, "loss": 0.08293132483959198, "step": 5392 }, { "epoch": 1.4000519210799585, "grad_norm": 2.427415132522583, "learning_rate": 1.4819344069460833e-06, "loss": 0.08913370966911316, "step": 5393 }, { "epoch": 1.4003115264797508, "grad_norm": 2.1451005935668945, "learning_rate": 1.4811530600202683e-06, "loss": 0.072015181183815, "step": 5394 }, { "epoch": 1.400571131879543, "grad_norm": 2.2782297134399414, "learning_rate": 1.4803719373975382e-06, "loss": 0.068380206823349, "step": 5395 }, { "epoch": 1.4008307372793354, "grad_norm": 2.7514567375183105, "learning_rate": 1.4795910392159798e-06, "loss": 0.08821230381727219, "step": 5396 }, { "epoch": 1.4010903426791277, "grad_norm": 1.0741053819656372, "learning_rate": 1.4788103656136372e-06, "loss": 0.07992976903915405, "step": 5397 }, { "epoch": 1.40134994807892, "grad_norm": 1.1644033193588257, "learning_rate": 1.478029916728515e-06, "loss": 0.09106336534023285, "step": 5398 }, { "epoch": 1.4016095534787123, "grad_norm": 1.6760213375091553, "learning_rate": 1.4772496926985808e-06, "loss": 0.09079146385192871, "step": 5399 }, { "epoch": 1.4018691588785046, "grad_norm": 2.2812323570251465, "learning_rate": 1.476469693661757e-06, "loss": 0.062438733875751495, "step": 5400 }, { "epoch": 1.402128764278297, "grad_norm": 1.5594110488891602, "learning_rate": 1.4756899197559317e-06, "loss": 0.06793583184480667, "step": 5401 }, { "epoch": 1.4023883696780892, "grad_norm": 1.3858494758605957, "learning_rate": 1.4749103711189517e-06, "loss": 0.0864597037434578, "step": 5402 }, { "epoch": 1.4026479750778815, "grad_norm": 1.2154954671859741, "learning_rate": 1.4741310478886217e-06, "loss": 0.0921611487865448, "step": 5403 }, { "epoch": 1.4029075804776738, "grad_norm": 0.9705194234848022, "learning_rate": 1.4733519502027093e-06, "loss": 0.11115129292011261, "step": 5404 }, { "epoch": 1.4031671858774661, "grad_norm": 2.1158480644226074, "learning_rate": 1.47257307819894e-06, "loss": 0.10928012430667877, "step": 5405 }, { "epoch": 1.4034267912772584, "grad_norm": 1.830750584602356, "learning_rate": 1.4717944320150013e-06, "loss": 0.07060038298368454, "step": 5406 }, { "epoch": 1.4036863966770508, "grad_norm": 1.3286828994750977, "learning_rate": 1.4710160117885414e-06, "loss": 0.10977330803871155, "step": 5407 }, { "epoch": 1.4039460020768433, "grad_norm": 2.717801332473755, "learning_rate": 1.4702378176571641e-06, "loss": 0.10010191053152084, "step": 5408 }, { "epoch": 1.4042056074766356, "grad_norm": 1.6886156797409058, "learning_rate": 1.4694598497584392e-06, "loss": 0.09391845017671585, "step": 5409 }, { "epoch": 1.404465212876428, "grad_norm": 0.8195159435272217, "learning_rate": 1.4686821082298914e-06, "loss": 0.09559056162834167, "step": 5410 }, { "epoch": 1.4047248182762202, "grad_norm": 1.246151328086853, "learning_rate": 1.4679045932090092e-06, "loss": 0.1006552055478096, "step": 5411 }, { "epoch": 1.4049844236760125, "grad_norm": 2.4629063606262207, "learning_rate": 1.4671273048332389e-06, "loss": 0.10883386433124542, "step": 5412 }, { "epoch": 1.4052440290758048, "grad_norm": 1.1941287517547607, "learning_rate": 1.4663502432399861e-06, "loss": 0.1094353199005127, "step": 5413 }, { "epoch": 1.4055036344755971, "grad_norm": 1.2663629055023193, "learning_rate": 1.4655734085666198e-06, "loss": 0.10707588493824005, "step": 5414 }, { "epoch": 1.4057632398753894, "grad_norm": 1.5051548480987549, "learning_rate": 1.4647968009504654e-06, "loss": 0.05738130956888199, "step": 5415 }, { "epoch": 1.4060228452751817, "grad_norm": 1.332828402519226, "learning_rate": 1.4640204205288084e-06, "loss": 0.0987115427851677, "step": 5416 }, { "epoch": 1.406282450674974, "grad_norm": 1.457950472831726, "learning_rate": 1.4632442674388969e-06, "loss": 0.10402954369783401, "step": 5417 }, { "epoch": 1.4065420560747663, "grad_norm": 2.080117702484131, "learning_rate": 1.4624683418179353e-06, "loss": 0.05754304677248001, "step": 5418 }, { "epoch": 1.4068016614745587, "grad_norm": 1.6812317371368408, "learning_rate": 1.461692643803091e-06, "loss": 0.08115822076797485, "step": 5419 }, { "epoch": 1.407061266874351, "grad_norm": 1.218604326248169, "learning_rate": 1.4609171735314892e-06, "loss": 0.10153673589229584, "step": 5420 }, { "epoch": 1.4073208722741433, "grad_norm": 1.2534794807434082, "learning_rate": 1.4601419311402142e-06, "loss": 0.09137829393148422, "step": 5421 }, { "epoch": 1.4075804776739356, "grad_norm": 1.6284736394882202, "learning_rate": 1.4593669167663126e-06, "loss": 0.05733746290206909, "step": 5422 }, { "epoch": 1.4078400830737279, "grad_norm": 1.379406213760376, "learning_rate": 1.4585921305467887e-06, "loss": 0.08902528882026672, "step": 5423 }, { "epoch": 1.4080996884735202, "grad_norm": 2.6632654666900635, "learning_rate": 1.4578175726186056e-06, "loss": 0.09998463094234467, "step": 5424 }, { "epoch": 1.4083592938733125, "grad_norm": 0.8416876792907715, "learning_rate": 1.45704324311869e-06, "loss": 0.11650075763463974, "step": 5425 }, { "epoch": 1.408618899273105, "grad_norm": 1.5556591749191284, "learning_rate": 1.4562691421839237e-06, "loss": 0.08315210789442062, "step": 5426 }, { "epoch": 1.4088785046728973, "grad_norm": 1.4472458362579346, "learning_rate": 1.4554952699511498e-06, "loss": 0.09982912242412567, "step": 5427 }, { "epoch": 1.4091381100726896, "grad_norm": 0.772532045841217, "learning_rate": 1.454721626557173e-06, "loss": 0.09369201958179474, "step": 5428 }, { "epoch": 1.409397715472482, "grad_norm": 1.9743458032608032, "learning_rate": 1.4539482121387539e-06, "loss": 0.10685201734304428, "step": 5429 }, { "epoch": 1.4096573208722742, "grad_norm": 1.518354058265686, "learning_rate": 1.4531750268326166e-06, "loss": 0.08029346168041229, "step": 5430 }, { "epoch": 1.4099169262720666, "grad_norm": 2.001919984817505, "learning_rate": 1.45240207077544e-06, "loss": 0.06438659876585007, "step": 5431 }, { "epoch": 1.4101765316718589, "grad_norm": 1.5292444229125977, "learning_rate": 1.4516293441038659e-06, "loss": 0.042877331376075745, "step": 5432 }, { "epoch": 1.4104361370716512, "grad_norm": 3.7648122310638428, "learning_rate": 1.450856846954496e-06, "loss": 0.09919741004705429, "step": 5433 }, { "epoch": 1.4106957424714435, "grad_norm": 1.3873673677444458, "learning_rate": 1.4500845794638886e-06, "loss": 0.10974586755037308, "step": 5434 }, { "epoch": 1.4109553478712358, "grad_norm": 2.6128973960876465, "learning_rate": 1.449312541768564e-06, "loss": 0.07970000803470612, "step": 5435 }, { "epoch": 1.411214953271028, "grad_norm": 2.1465775966644287, "learning_rate": 1.4485407340049991e-06, "loss": 0.11976763606071472, "step": 5436 }, { "epoch": 1.4114745586708204, "grad_norm": 0.943084716796875, "learning_rate": 1.447769156309633e-06, "loss": 0.11235867440700531, "step": 5437 }, { "epoch": 1.4117341640706127, "grad_norm": 2.592036247253418, "learning_rate": 1.4469978088188646e-06, "loss": 0.08685477077960968, "step": 5438 }, { "epoch": 1.411993769470405, "grad_norm": 1.494664192199707, "learning_rate": 1.4462266916690474e-06, "loss": 0.04823945090174675, "step": 5439 }, { "epoch": 1.4122533748701973, "grad_norm": 1.7521076202392578, "learning_rate": 1.4454558049964993e-06, "loss": 0.08024448156356812, "step": 5440 }, { "epoch": 1.4125129802699896, "grad_norm": 3.782893180847168, "learning_rate": 1.4446851489374942e-06, "loss": 0.08000379800796509, "step": 5441 }, { "epoch": 1.412772585669782, "grad_norm": 1.3335040807724, "learning_rate": 1.4439147236282674e-06, "loss": 0.09401368349790573, "step": 5442 }, { "epoch": 1.4130321910695742, "grad_norm": 0.9526726603507996, "learning_rate": 1.4431445292050123e-06, "loss": 0.10142326354980469, "step": 5443 }, { "epoch": 1.4132917964693665, "grad_norm": 0.952166736125946, "learning_rate": 1.4423745658038808e-06, "loss": 0.0996813103556633, "step": 5444 }, { "epoch": 1.4135514018691588, "grad_norm": 2.3943965435028076, "learning_rate": 1.4416048335609856e-06, "loss": 0.067239910364151, "step": 5445 }, { "epoch": 1.4138110072689511, "grad_norm": 1.6630666255950928, "learning_rate": 1.4408353326123981e-06, "loss": 0.07234061509370804, "step": 5446 }, { "epoch": 1.4140706126687435, "grad_norm": 1.4186261892318726, "learning_rate": 1.4400660630941469e-06, "loss": 0.07123464345932007, "step": 5447 }, { "epoch": 1.4143302180685358, "grad_norm": 1.1273201704025269, "learning_rate": 1.439297025142223e-06, "loss": 0.07738295197486877, "step": 5448 }, { "epoch": 1.414589823468328, "grad_norm": 2.2349231243133545, "learning_rate": 1.4385282188925732e-06, "loss": 0.1379469335079193, "step": 5449 }, { "epoch": 1.4148494288681204, "grad_norm": 1.5902460813522339, "learning_rate": 1.4377596444811064e-06, "loss": 0.08577509224414825, "step": 5450 }, { "epoch": 1.4151090342679127, "grad_norm": 1.3633686304092407, "learning_rate": 1.4369913020436882e-06, "loss": 0.08046507835388184, "step": 5451 }, { "epoch": 1.415368639667705, "grad_norm": 1.690893530845642, "learning_rate": 1.436223191716143e-06, "loss": 0.08357946574687958, "step": 5452 }, { "epoch": 1.4156282450674973, "grad_norm": 2.3734805583953857, "learning_rate": 1.4354553136342569e-06, "loss": 0.05940568447113037, "step": 5453 }, { "epoch": 1.4158878504672896, "grad_norm": 1.527306079864502, "learning_rate": 1.4346876679337721e-06, "loss": 0.12969738245010376, "step": 5454 }, { "epoch": 1.416147455867082, "grad_norm": 0.9771069288253784, "learning_rate": 1.43392025475039e-06, "loss": 0.10726368427276611, "step": 5455 }, { "epoch": 1.4164070612668744, "grad_norm": 0.8648546934127808, "learning_rate": 1.4331530742197738e-06, "loss": 0.09551429748535156, "step": 5456 }, { "epoch": 1.4166666666666667, "grad_norm": 1.1207524538040161, "learning_rate": 1.432386126477542e-06, "loss": 0.09488169103860855, "step": 5457 }, { "epoch": 1.416926272066459, "grad_norm": 2.4823808670043945, "learning_rate": 1.431619411659273e-06, "loss": 0.11621935665607452, "step": 5458 }, { "epoch": 1.4171858774662514, "grad_norm": 1.3537369966506958, "learning_rate": 1.4308529299005059e-06, "loss": 0.06357432901859283, "step": 5459 }, { "epoch": 1.4174454828660437, "grad_norm": 1.4884754419326782, "learning_rate": 1.4300866813367356e-06, "loss": 0.07232372462749481, "step": 5460 }, { "epoch": 1.417705088265836, "grad_norm": 2.9187076091766357, "learning_rate": 1.4293206661034198e-06, "loss": 0.06017729267477989, "step": 5461 }, { "epoch": 1.4179646936656283, "grad_norm": 2.282845973968506, "learning_rate": 1.4285548843359686e-06, "loss": 0.09502940624952316, "step": 5462 }, { "epoch": 1.4182242990654206, "grad_norm": 2.1787595748901367, "learning_rate": 1.427789336169757e-06, "loss": 0.053249724209308624, "step": 5463 }, { "epoch": 1.4184839044652129, "grad_norm": 2.0856783390045166, "learning_rate": 1.4270240217401171e-06, "loss": 0.07401984930038452, "step": 5464 }, { "epoch": 1.4187435098650052, "grad_norm": 1.9891303777694702, "learning_rate": 1.426258941182338e-06, "loss": 0.06851667165756226, "step": 5465 }, { "epoch": 1.4190031152647975, "grad_norm": 1.6114139556884766, "learning_rate": 1.4254940946316684e-06, "loss": 0.1004115492105484, "step": 5466 }, { "epoch": 1.4192627206645898, "grad_norm": 0.9431700706481934, "learning_rate": 1.4247294822233146e-06, "loss": 0.08931749314069748, "step": 5467 }, { "epoch": 1.4195223260643821, "grad_norm": 2.2483277320861816, "learning_rate": 1.423965104092444e-06, "loss": 0.07825204730033875, "step": 5468 }, { "epoch": 1.4197819314641744, "grad_norm": 1.1149860620498657, "learning_rate": 1.4232009603741822e-06, "loss": 0.08566334843635559, "step": 5469 }, { "epoch": 1.4200415368639667, "grad_norm": 2.0328009128570557, "learning_rate": 1.4224370512036095e-06, "loss": 0.04537423327565193, "step": 5470 }, { "epoch": 1.420301142263759, "grad_norm": 2.3093512058258057, "learning_rate": 1.4216733767157698e-06, "loss": 0.06312225013971329, "step": 5471 }, { "epoch": 1.4205607476635513, "grad_norm": 1.1934142112731934, "learning_rate": 1.420909937045662e-06, "loss": 0.07196170836687088, "step": 5472 }, { "epoch": 1.4208203530633436, "grad_norm": 0.9175509214401245, "learning_rate": 1.4201467323282453e-06, "loss": 0.1053071841597557, "step": 5473 }, { "epoch": 1.421079958463136, "grad_norm": 2.442641496658325, "learning_rate": 1.419383762698437e-06, "loss": 0.10152499377727509, "step": 5474 }, { "epoch": 1.4213395638629285, "grad_norm": 1.3156585693359375, "learning_rate": 1.4186210282911118e-06, "loss": 0.08405358344316483, "step": 5475 }, { "epoch": 1.4215991692627208, "grad_norm": 1.4146878719329834, "learning_rate": 1.4178585292411046e-06, "loss": 0.09415294229984283, "step": 5476 }, { "epoch": 1.421858774662513, "grad_norm": 2.7321815490722656, "learning_rate": 1.4170962656832077e-06, "loss": 0.14615163207054138, "step": 5477 }, { "epoch": 1.4221183800623054, "grad_norm": 1.0785027742385864, "learning_rate": 1.416334237752171e-06, "loss": 0.0982193797826767, "step": 5478 }, { "epoch": 1.4223779854620977, "grad_norm": 1.1843550205230713, "learning_rate": 1.4155724455827046e-06, "loss": 0.10420718789100647, "step": 5479 }, { "epoch": 1.42263759086189, "grad_norm": 1.9708480834960938, "learning_rate": 1.4148108893094756e-06, "loss": 0.10516341030597687, "step": 5480 }, { "epoch": 1.4228971962616823, "grad_norm": 1.9327665567398071, "learning_rate": 1.414049569067109e-06, "loss": 0.07358808815479279, "step": 5481 }, { "epoch": 1.4231568016614746, "grad_norm": 0.7805708050727844, "learning_rate": 1.41328848499019e-06, "loss": 0.1057073324918747, "step": 5482 }, { "epoch": 1.423416407061267, "grad_norm": 0.8804479241371155, "learning_rate": 1.4125276372132594e-06, "loss": 0.0927874743938446, "step": 5483 }, { "epoch": 1.4236760124610592, "grad_norm": 1.8902266025543213, "learning_rate": 1.4117670258708194e-06, "loss": 0.08841625601053238, "step": 5484 }, { "epoch": 1.4239356178608515, "grad_norm": 0.9024019241333008, "learning_rate": 1.4110066510973275e-06, "loss": 0.08994550257921219, "step": 5485 }, { "epoch": 1.4241952232606438, "grad_norm": 1.5638184547424316, "learning_rate": 1.4102465130272003e-06, "loss": 0.09012355655431747, "step": 5486 }, { "epoch": 1.4244548286604362, "grad_norm": 1.4889856576919556, "learning_rate": 1.4094866117948136e-06, "loss": 0.07910090684890747, "step": 5487 }, { "epoch": 1.4247144340602285, "grad_norm": 3.224169969558716, "learning_rate": 1.4087269475345009e-06, "loss": 0.13517816364765167, "step": 5488 }, { "epoch": 1.4249740394600208, "grad_norm": 2.6774260997772217, "learning_rate": 1.4079675203805516e-06, "loss": 0.042717382311820984, "step": 5489 }, { "epoch": 1.425233644859813, "grad_norm": 1.5963724851608276, "learning_rate": 1.4072083304672174e-06, "loss": 0.06732894480228424, "step": 5490 }, { "epoch": 1.4254932502596054, "grad_norm": 1.999118447303772, "learning_rate": 1.4064493779287033e-06, "loss": 0.07981859147548676, "step": 5491 }, { "epoch": 1.4257528556593977, "grad_norm": 2.819875955581665, "learning_rate": 1.4056906628991775e-06, "loss": 0.12804171442985535, "step": 5492 }, { "epoch": 1.42601246105919, "grad_norm": 1.7817034721374512, "learning_rate": 1.4049321855127608e-06, "loss": 0.07214696705341339, "step": 5493 }, { "epoch": 1.4262720664589823, "grad_norm": 1.8700088262557983, "learning_rate": 1.4041739459035352e-06, "loss": 0.08030404150485992, "step": 5494 }, { "epoch": 1.4265316718587746, "grad_norm": 1.82258939743042, "learning_rate": 1.4034159442055417e-06, "loss": 0.05720890313386917, "step": 5495 }, { "epoch": 1.426791277258567, "grad_norm": 1.141995906829834, "learning_rate": 1.4026581805527767e-06, "loss": 0.09768866747617722, "step": 5496 }, { "epoch": 1.4270508826583592, "grad_norm": 0.927881121635437, "learning_rate": 1.4019006550791953e-06, "loss": 0.0981748104095459, "step": 5497 }, { "epoch": 1.4273104880581515, "grad_norm": 1.3742460012435913, "learning_rate": 1.4011433679187098e-06, "loss": 0.08388325572013855, "step": 5498 }, { "epoch": 1.4275700934579438, "grad_norm": 1.570218563079834, "learning_rate": 1.400386319205192e-06, "loss": 0.02962559275329113, "step": 5499 }, { "epoch": 1.4278296988577361, "grad_norm": 0.8412952423095703, "learning_rate": 1.3996295090724723e-06, "loss": 0.10581006109714508, "step": 5500 }, { "epoch": 1.4280893042575284, "grad_norm": 1.0722569227218628, "learning_rate": 1.3988729376543347e-06, "loss": 0.09351101517677307, "step": 5501 }, { "epoch": 1.4283489096573208, "grad_norm": 0.8974300026893616, "learning_rate": 1.3981166050845258e-06, "loss": 0.10856063663959503, "step": 5502 }, { "epoch": 1.428608515057113, "grad_norm": 2.0958852767944336, "learning_rate": 1.3973605114967465e-06, "loss": 0.08051689714193344, "step": 5503 }, { "epoch": 1.4288681204569054, "grad_norm": 1.4941067695617676, "learning_rate": 1.396604657024658e-06, "loss": 0.10017213225364685, "step": 5504 }, { "epoch": 1.429127725856698, "grad_norm": 0.8697652220726013, "learning_rate": 1.3958490418018778e-06, "loss": 0.10180847346782684, "step": 5505 }, { "epoch": 1.4293873312564902, "grad_norm": 1.8342598676681519, "learning_rate": 1.3950936659619795e-06, "loss": 0.05897211283445358, "step": 5506 }, { "epoch": 1.4296469366562825, "grad_norm": 2.208386182785034, "learning_rate": 1.3943385296384996e-06, "loss": 0.07817265391349792, "step": 5507 }, { "epoch": 1.4299065420560748, "grad_norm": 2.9474380016326904, "learning_rate": 1.3935836329649266e-06, "loss": 0.1097116768360138, "step": 5508 }, { "epoch": 1.4301661474558671, "grad_norm": 1.8299205303192139, "learning_rate": 1.3928289760747088e-06, "loss": 0.09495235234498978, "step": 5509 }, { "epoch": 1.4304257528556594, "grad_norm": 0.9327763319015503, "learning_rate": 1.3920745591012536e-06, "loss": 0.11650241166353226, "step": 5510 }, { "epoch": 1.4306853582554517, "grad_norm": 1.2162184715270996, "learning_rate": 1.3913203821779239e-06, "loss": 0.10660241544246674, "step": 5511 }, { "epoch": 1.430944963655244, "grad_norm": 1.37444269657135, "learning_rate": 1.3905664454380401e-06, "loss": 0.09223763644695282, "step": 5512 }, { "epoch": 1.4312045690550363, "grad_norm": 2.1717355251312256, "learning_rate": 1.3898127490148832e-06, "loss": 0.05914686620235443, "step": 5513 }, { "epoch": 1.4314641744548287, "grad_norm": 1.6870089769363403, "learning_rate": 1.389059293041687e-06, "loss": 0.08747787773609161, "step": 5514 }, { "epoch": 1.431723779854621, "grad_norm": 2.233616828918457, "learning_rate": 1.3883060776516473e-06, "loss": 0.12361036241054535, "step": 5515 }, { "epoch": 1.4319833852544133, "grad_norm": 1.1297000646591187, "learning_rate": 1.3875531029779137e-06, "loss": 0.11900731176137924, "step": 5516 }, { "epoch": 1.4322429906542056, "grad_norm": 2.206170082092285, "learning_rate": 1.3868003691535958e-06, "loss": 0.060464873909950256, "step": 5517 }, { "epoch": 1.4325025960539979, "grad_norm": 0.9545180201530457, "learning_rate": 1.3860478763117596e-06, "loss": 0.09845087677240372, "step": 5518 }, { "epoch": 1.4327622014537902, "grad_norm": 2.4292373657226562, "learning_rate": 1.3852956245854288e-06, "loss": 0.09900152683258057, "step": 5519 }, { "epoch": 1.4330218068535825, "grad_norm": 2.0883872509002686, "learning_rate": 1.3845436141075832e-06, "loss": 0.08490142226219177, "step": 5520 }, { "epoch": 1.4332814122533748, "grad_norm": 1.7746319770812988, "learning_rate": 1.3837918450111626e-06, "loss": 0.05946850776672363, "step": 5521 }, { "epoch": 1.433541017653167, "grad_norm": 2.585435390472412, "learning_rate": 1.3830403174290611e-06, "loss": 0.1416945606470108, "step": 5522 }, { "epoch": 1.4338006230529596, "grad_norm": 1.6560174226760864, "learning_rate": 1.382289031494134e-06, "loss": 0.061953604221343994, "step": 5523 }, { "epoch": 1.434060228452752, "grad_norm": 1.3296743631362915, "learning_rate": 1.3815379873391876e-06, "loss": 0.09410557150840759, "step": 5524 }, { "epoch": 1.4343198338525442, "grad_norm": 2.5498554706573486, "learning_rate": 1.3807871850969918e-06, "loss": 0.05725138634443283, "step": 5525 }, { "epoch": 1.4345794392523366, "grad_norm": 0.8985838890075684, "learning_rate": 1.3800366249002715e-06, "loss": 0.08607283234596252, "step": 5526 }, { "epoch": 1.4348390446521289, "grad_norm": 1.9511009454727173, "learning_rate": 1.3792863068817079e-06, "loss": 0.09212648868560791, "step": 5527 }, { "epoch": 1.4350986500519212, "grad_norm": 2.1922006607055664, "learning_rate": 1.3785362311739397e-06, "loss": 0.11805547773838043, "step": 5528 }, { "epoch": 1.4353582554517135, "grad_norm": 2.0487120151519775, "learning_rate": 1.377786397909563e-06, "loss": 0.08569052815437317, "step": 5529 }, { "epoch": 1.4356178608515058, "grad_norm": 1.7793958187103271, "learning_rate": 1.377036807221132e-06, "loss": 0.09410528838634491, "step": 5530 }, { "epoch": 1.435877466251298, "grad_norm": 1.1122866868972778, "learning_rate": 1.376287459241158e-06, "loss": 0.09098494797945023, "step": 5531 }, { "epoch": 1.4361370716510904, "grad_norm": 1.4880738258361816, "learning_rate": 1.3755383541021056e-06, "loss": 0.0922633484005928, "step": 5532 }, { "epoch": 1.4363966770508827, "grad_norm": 1.4750995635986328, "learning_rate": 1.3747894919364024e-06, "loss": 0.07154165953397751, "step": 5533 }, { "epoch": 1.436656282450675, "grad_norm": 1.0845284461975098, "learning_rate": 1.3740408728764277e-06, "loss": 0.09270547330379486, "step": 5534 }, { "epoch": 1.4369158878504673, "grad_norm": 1.276590347290039, "learning_rate": 1.3732924970545225e-06, "loss": 0.101833775639534, "step": 5535 }, { "epoch": 1.4371754932502596, "grad_norm": 1.3607633113861084, "learning_rate": 1.3725443646029811e-06, "loss": 0.10307000577449799, "step": 5536 }, { "epoch": 1.437435098650052, "grad_norm": 1.4647700786590576, "learning_rate": 1.3717964756540557e-06, "loss": 0.0872625857591629, "step": 5537 }, { "epoch": 1.4376947040498442, "grad_norm": 2.6306684017181396, "learning_rate": 1.3710488303399577e-06, "loss": 0.07057608664035797, "step": 5538 }, { "epoch": 1.4379543094496365, "grad_norm": 0.8394936323165894, "learning_rate": 1.3703014287928528e-06, "loss": 0.09259139001369476, "step": 5539 }, { "epoch": 1.4382139148494288, "grad_norm": 2.069093942642212, "learning_rate": 1.3695542711448636e-06, "loss": 0.07821031659841537, "step": 5540 }, { "epoch": 1.4384735202492211, "grad_norm": 1.6900527477264404, "learning_rate": 1.368807357528072e-06, "loss": 0.06373687833547592, "step": 5541 }, { "epoch": 1.4387331256490135, "grad_norm": 2.7954232692718506, "learning_rate": 1.368060688074514e-06, "loss": 0.06687815487384796, "step": 5542 }, { "epoch": 1.4389927310488058, "grad_norm": 1.537164568901062, "learning_rate": 1.367314262916184e-06, "loss": 0.09516583383083344, "step": 5543 }, { "epoch": 1.439252336448598, "grad_norm": 1.782835602760315, "learning_rate": 1.3665680821850335e-06, "loss": 0.110873743891716, "step": 5544 }, { "epoch": 1.4395119418483904, "grad_norm": 1.4813916683197021, "learning_rate": 1.365822146012969e-06, "loss": 0.09120619297027588, "step": 5545 }, { "epoch": 1.4397715472481827, "grad_norm": 1.3634207248687744, "learning_rate": 1.3650764545318562e-06, "loss": 0.09584929049015045, "step": 5546 }, { "epoch": 1.440031152647975, "grad_norm": 1.5996681451797485, "learning_rate": 1.3643310078735156e-06, "loss": 0.08409257233142853, "step": 5547 }, { "epoch": 1.4402907580477673, "grad_norm": 2.3052854537963867, "learning_rate": 1.3635858061697246e-06, "loss": 0.06456848978996277, "step": 5548 }, { "epoch": 1.4405503634475596, "grad_norm": 1.3975627422332764, "learning_rate": 1.3628408495522195e-06, "loss": 0.07618723809719086, "step": 5549 }, { "epoch": 1.440809968847352, "grad_norm": 1.4784283638000488, "learning_rate": 1.36209613815269e-06, "loss": 0.07763178646564484, "step": 5550 }, { "epoch": 1.4410695742471442, "grad_norm": 1.4012764692306519, "learning_rate": 1.3613516721027837e-06, "loss": 0.0940573513507843, "step": 5551 }, { "epoch": 1.4413291796469365, "grad_norm": 1.0728421211242676, "learning_rate": 1.360607451534107e-06, "loss": 0.09998201578855515, "step": 5552 }, { "epoch": 1.441588785046729, "grad_norm": 1.5085986852645874, "learning_rate": 1.3598634765782192e-06, "loss": 0.09060589969158173, "step": 5553 }, { "epoch": 1.4418483904465214, "grad_norm": 1.972074270248413, "learning_rate": 1.3591197473666401e-06, "loss": 0.06097966805100441, "step": 5554 }, { "epoch": 1.4421079958463137, "grad_norm": 2.0789878368377686, "learning_rate": 1.3583762640308412e-06, "loss": 0.07799579948186874, "step": 5555 }, { "epoch": 1.442367601246106, "grad_norm": 1.221176266670227, "learning_rate": 1.3576330267022552e-06, "loss": 0.09567118436098099, "step": 5556 }, { "epoch": 1.4426272066458983, "grad_norm": 1.78324556350708, "learning_rate": 1.3568900355122697e-06, "loss": 0.06681147962808609, "step": 5557 }, { "epoch": 1.4428868120456906, "grad_norm": 1.4442529678344727, "learning_rate": 1.356147290592228e-06, "loss": 0.07873836904764175, "step": 5558 }, { "epoch": 1.4431464174454829, "grad_norm": 2.524501085281372, "learning_rate": 1.3554047920734301e-06, "loss": 0.08582118153572083, "step": 5559 }, { "epoch": 1.4434060228452752, "grad_norm": 1.2483302354812622, "learning_rate": 1.3546625400871321e-06, "loss": 0.0825248658657074, "step": 5560 }, { "epoch": 1.4436656282450675, "grad_norm": 1.3459399938583374, "learning_rate": 1.3539205347645485e-06, "loss": 0.09096066653728485, "step": 5561 }, { "epoch": 1.4439252336448598, "grad_norm": 1.6701596975326538, "learning_rate": 1.3531787762368498e-06, "loss": 0.09073203802108765, "step": 5562 }, { "epoch": 1.4441848390446521, "grad_norm": 1.207314372062683, "learning_rate": 1.3524372646351585e-06, "loss": 0.09813626110553741, "step": 5563 }, { "epoch": 1.4444444444444444, "grad_norm": 1.6617817878723145, "learning_rate": 1.3516960000905601e-06, "loss": 0.08142507076263428, "step": 5564 }, { "epoch": 1.4447040498442367, "grad_norm": 3.7855424880981445, "learning_rate": 1.3509549827340917e-06, "loss": 0.04263627156615257, "step": 5565 }, { "epoch": 1.444963655244029, "grad_norm": 3.103343963623047, "learning_rate": 1.350214212696748e-06, "loss": 0.13971146941184998, "step": 5566 }, { "epoch": 1.4452232606438213, "grad_norm": 1.7529047727584839, "learning_rate": 1.3494736901094814e-06, "loss": 0.03215964511036873, "step": 5567 }, { "epoch": 1.4454828660436136, "grad_norm": 1.8195693492889404, "learning_rate": 1.3487334151031975e-06, "loss": 0.053660180419683456, "step": 5568 }, { "epoch": 1.445742471443406, "grad_norm": 2.2969419956207275, "learning_rate": 1.347993387808762e-06, "loss": 0.10516650974750519, "step": 5569 }, { "epoch": 1.4460020768431983, "grad_norm": 1.6264100074768066, "learning_rate": 1.3472536083569937e-06, "loss": 0.071523517370224, "step": 5570 }, { "epoch": 1.4462616822429908, "grad_norm": 2.1850128173828125, "learning_rate": 1.3465140768786683e-06, "loss": 0.07154932618141174, "step": 5571 }, { "epoch": 1.446521287642783, "grad_norm": 2.0144498348236084, "learning_rate": 1.3457747935045193e-06, "loss": 0.05826356261968613, "step": 5572 }, { "epoch": 1.4467808930425754, "grad_norm": 0.887910783290863, "learning_rate": 1.345035758365234e-06, "loss": 0.09912936389446259, "step": 5573 }, { "epoch": 1.4470404984423677, "grad_norm": 0.8975045084953308, "learning_rate": 1.3442969715914568e-06, "loss": 0.09403949975967407, "step": 5574 }, { "epoch": 1.44730010384216, "grad_norm": 3.0898661613464355, "learning_rate": 1.343558433313789e-06, "loss": 0.057015981525182724, "step": 5575 }, { "epoch": 1.4475597092419523, "grad_norm": 1.1362682580947876, "learning_rate": 1.3428201436627863e-06, "loss": 0.08643198013305664, "step": 5576 }, { "epoch": 1.4478193146417446, "grad_norm": 1.1064552068710327, "learning_rate": 1.342082102768963e-06, "loss": 0.0911404937505722, "step": 5577 }, { "epoch": 1.448078920041537, "grad_norm": 1.8565151691436768, "learning_rate": 1.3413443107627872e-06, "loss": 0.06267070770263672, "step": 5578 }, { "epoch": 1.4483385254413292, "grad_norm": 1.1398260593414307, "learning_rate": 1.340606767774682e-06, "loss": 0.07519452273845673, "step": 5579 }, { "epoch": 1.4485981308411215, "grad_norm": 1.0612819194793701, "learning_rate": 1.3398694739350303e-06, "loss": 0.08628260344266891, "step": 5580 }, { "epoch": 1.4488577362409139, "grad_norm": 0.7188932299613953, "learning_rate": 1.3391324293741681e-06, "loss": 0.09046466648578644, "step": 5581 }, { "epoch": 1.4491173416407062, "grad_norm": 2.482975482940674, "learning_rate": 1.3383956342223872e-06, "loss": 0.07540398091077805, "step": 5582 }, { "epoch": 1.4493769470404985, "grad_norm": 1.3741512298583984, "learning_rate": 1.3376590886099372e-06, "loss": 0.09227129817008972, "step": 5583 }, { "epoch": 1.4496365524402908, "grad_norm": 3.1874947547912598, "learning_rate": 1.3369227926670215e-06, "loss": 0.07607052475214005, "step": 5584 }, { "epoch": 1.449896157840083, "grad_norm": 2.181605100631714, "learning_rate": 1.336186746523802e-06, "loss": 0.06770507991313934, "step": 5585 }, { "epoch": 1.4501557632398754, "grad_norm": 1.000555396080017, "learning_rate": 1.3354509503103922e-06, "loss": 0.092898428440094, "step": 5586 }, { "epoch": 1.4504153686396677, "grad_norm": 1.2781637907028198, "learning_rate": 1.3347154041568655e-06, "loss": 0.09508414566516876, "step": 5587 }, { "epoch": 1.45067497403946, "grad_norm": 1.0157794952392578, "learning_rate": 1.3339801081932507e-06, "loss": 0.09446574002504349, "step": 5588 }, { "epoch": 1.4509345794392523, "grad_norm": 1.7199351787567139, "learning_rate": 1.3332450625495298e-06, "loss": 0.07364746928215027, "step": 5589 }, { "epoch": 1.4511941848390446, "grad_norm": 1.6800693273544312, "learning_rate": 1.332510267355643e-06, "loss": 0.07699365168809891, "step": 5590 }, { "epoch": 1.451453790238837, "grad_norm": 0.8258773684501648, "learning_rate": 1.3317757227414835e-06, "loss": 0.0936555415391922, "step": 5591 }, { "epoch": 1.4517133956386292, "grad_norm": 2.3397152423858643, "learning_rate": 1.331041428836904e-06, "loss": 0.09485261887311935, "step": 5592 }, { "epoch": 1.4519730010384215, "grad_norm": 1.7442909479141235, "learning_rate": 1.3303073857717103e-06, "loss": 0.06751951575279236, "step": 5593 }, { "epoch": 1.4522326064382138, "grad_norm": 1.4535250663757324, "learning_rate": 1.3295735936756631e-06, "loss": 0.07565809041261673, "step": 5594 }, { "epoch": 1.4524922118380061, "grad_norm": 1.054301381111145, "learning_rate": 1.328840052678482e-06, "loss": 0.08510340750217438, "step": 5595 }, { "epoch": 1.4527518172377984, "grad_norm": 1.4101475477218628, "learning_rate": 1.3281067629098394e-06, "loss": 0.05367811769247055, "step": 5596 }, { "epoch": 1.4530114226375908, "grad_norm": 1.006090521812439, "learning_rate": 1.3273737244993632e-06, "loss": 0.11501339077949524, "step": 5597 }, { "epoch": 1.453271028037383, "grad_norm": 1.580352783203125, "learning_rate": 1.3266409375766392e-06, "loss": 0.03135039657354355, "step": 5598 }, { "epoch": 1.4535306334371754, "grad_norm": 2.1214213371276855, "learning_rate": 1.3259084022712065e-06, "loss": 0.09573540091514587, "step": 5599 }, { "epoch": 1.4537902388369677, "grad_norm": 2.3939077854156494, "learning_rate": 1.3251761187125614e-06, "loss": 0.1223086267709732, "step": 5600 }, { "epoch": 1.4540498442367602, "grad_norm": 1.1254281997680664, "learning_rate": 1.3244440870301545e-06, "loss": 0.07420440018177032, "step": 5601 }, { "epoch": 1.4543094496365525, "grad_norm": 2.431887626647949, "learning_rate": 1.3237123073533915e-06, "loss": 0.08980583399534225, "step": 5602 }, { "epoch": 1.4545690550363448, "grad_norm": 1.1096527576446533, "learning_rate": 1.3229807798116351e-06, "loss": 0.08528539538383484, "step": 5603 }, { "epoch": 1.4548286604361371, "grad_norm": 1.8893113136291504, "learning_rate": 1.3222495045342032e-06, "loss": 0.08436676859855652, "step": 5604 }, { "epoch": 1.4550882658359294, "grad_norm": 1.0305273532867432, "learning_rate": 1.3215184816503663e-06, "loss": 0.092296302318573, "step": 5605 }, { "epoch": 1.4553478712357217, "grad_norm": 2.053166627883911, "learning_rate": 1.3207877112893553e-06, "loss": 0.06548468768596649, "step": 5606 }, { "epoch": 1.455607476635514, "grad_norm": 2.1175215244293213, "learning_rate": 1.3200571935803513e-06, "loss": 0.10202641785144806, "step": 5607 }, { "epoch": 1.4558670820353063, "grad_norm": 2.19356369972229, "learning_rate": 1.3193269286524945e-06, "loss": 0.10152791440486908, "step": 5608 }, { "epoch": 1.4561266874350987, "grad_norm": 1.1641613245010376, "learning_rate": 1.318596916634879e-06, "loss": 0.09886765480041504, "step": 5609 }, { "epoch": 1.456386292834891, "grad_norm": 2.9760358333587646, "learning_rate": 1.3178671576565533e-06, "loss": 0.04681891202926636, "step": 5610 }, { "epoch": 1.4566458982346833, "grad_norm": 4.132040500640869, "learning_rate": 1.3171376518465228e-06, "loss": 0.06802040338516235, "step": 5611 }, { "epoch": 1.4569055036344756, "grad_norm": 2.3393874168395996, "learning_rate": 1.3164083993337478e-06, "loss": 0.06933934986591339, "step": 5612 }, { "epoch": 1.4571651090342679, "grad_norm": 0.8118723630905151, "learning_rate": 1.3156794002471417e-06, "loss": 0.10912582278251648, "step": 5613 }, { "epoch": 1.4574247144340602, "grad_norm": 0.7967187166213989, "learning_rate": 1.314950654715577e-06, "loss": 0.10067909210920334, "step": 5614 }, { "epoch": 1.4576843198338525, "grad_norm": 1.7333143949508667, "learning_rate": 1.3142221628678778e-06, "loss": 0.07894779741764069, "step": 5615 }, { "epoch": 1.4579439252336448, "grad_norm": 0.9741872549057007, "learning_rate": 1.3134939248328265e-06, "loss": 0.10628602653741837, "step": 5616 }, { "epoch": 1.458203530633437, "grad_norm": 1.7329058647155762, "learning_rate": 1.3127659407391564e-06, "loss": 0.09128459542989731, "step": 5617 }, { "epoch": 1.4584631360332294, "grad_norm": 0.854745626449585, "learning_rate": 1.3120382107155595e-06, "loss": 0.08601236343383789, "step": 5618 }, { "epoch": 1.458722741433022, "grad_norm": 1.1549644470214844, "learning_rate": 1.3113107348906833e-06, "loss": 0.07588547468185425, "step": 5619 }, { "epoch": 1.4589823468328142, "grad_norm": 2.623004674911499, "learning_rate": 1.3105835133931264e-06, "loss": 0.06534261256456375, "step": 5620 }, { "epoch": 1.4592419522326066, "grad_norm": 0.9963952302932739, "learning_rate": 1.3098565463514467e-06, "loss": 0.07129192352294922, "step": 5621 }, { "epoch": 1.4595015576323989, "grad_norm": 2.0665645599365234, "learning_rate": 1.3091298338941538e-06, "loss": 0.10658321529626846, "step": 5622 }, { "epoch": 1.4597611630321912, "grad_norm": 1.7797561883926392, "learning_rate": 1.3084033761497156e-06, "loss": 0.07978932559490204, "step": 5623 }, { "epoch": 1.4600207684319835, "grad_norm": 1.9452227354049683, "learning_rate": 1.3076771732465521e-06, "loss": 0.07553223520517349, "step": 5624 }, { "epoch": 1.4602803738317758, "grad_norm": 0.810589075088501, "learning_rate": 1.306951225313039e-06, "loss": 0.1041233241558075, "step": 5625 }, { "epoch": 1.460539979231568, "grad_norm": 1.954848051071167, "learning_rate": 1.3062255324775086e-06, "loss": 0.053359877318143845, "step": 5626 }, { "epoch": 1.4607995846313604, "grad_norm": 1.8831576108932495, "learning_rate": 1.3055000948682458e-06, "loss": 0.0723959431052208, "step": 5627 }, { "epoch": 1.4610591900311527, "grad_norm": 1.8446556329727173, "learning_rate": 1.3047749126134912e-06, "loss": 0.09833551943302155, "step": 5628 }, { "epoch": 1.461318795430945, "grad_norm": 2.555307388305664, "learning_rate": 1.304049985841441e-06, "loss": 0.06051081418991089, "step": 5629 }, { "epoch": 1.4615784008307373, "grad_norm": 1.5011131763458252, "learning_rate": 1.303325314680245e-06, "loss": 0.04425887018442154, "step": 5630 }, { "epoch": 1.4618380062305296, "grad_norm": 1.474104881286621, "learning_rate": 1.3026008992580096e-06, "loss": 0.10594537854194641, "step": 5631 }, { "epoch": 1.462097611630322, "grad_norm": 2.160998821258545, "learning_rate": 1.3018767397027942e-06, "loss": 0.06516115367412567, "step": 5632 }, { "epoch": 1.4623572170301142, "grad_norm": 1.7767980098724365, "learning_rate": 1.301152836142613e-06, "loss": 0.10120688378810883, "step": 5633 }, { "epoch": 1.4626168224299065, "grad_norm": 1.7510905265808105, "learning_rate": 1.3004291887054366e-06, "loss": 0.10097165405750275, "step": 5634 }, { "epoch": 1.4628764278296988, "grad_norm": 1.591636061668396, "learning_rate": 1.299705797519189e-06, "loss": 0.10833658277988434, "step": 5635 }, { "epoch": 1.4631360332294912, "grad_norm": 1.1513221263885498, "learning_rate": 1.2989826627117482e-06, "loss": 0.08912007510662079, "step": 5636 }, { "epoch": 1.4633956386292835, "grad_norm": 1.5545201301574707, "learning_rate": 1.2982597844109498e-06, "loss": 0.06683840602636337, "step": 5637 }, { "epoch": 1.4636552440290758, "grad_norm": 2.3255956172943115, "learning_rate": 1.29753716274458e-06, "loss": 0.037200283259153366, "step": 5638 }, { "epoch": 1.463914849428868, "grad_norm": 1.9822745323181152, "learning_rate": 1.2968147978403835e-06, "loss": 0.0775606632232666, "step": 5639 }, { "epoch": 1.4641744548286604, "grad_norm": 0.9315198659896851, "learning_rate": 1.2960926898260575e-06, "loss": 0.11026741564273834, "step": 5640 }, { "epoch": 1.4644340602284527, "grad_norm": 2.390897035598755, "learning_rate": 1.2953708388292533e-06, "loss": 0.0615990087389946, "step": 5641 }, { "epoch": 1.464693665628245, "grad_norm": 2.306804656982422, "learning_rate": 1.294649244977579e-06, "loss": 0.10476495325565338, "step": 5642 }, { "epoch": 1.4649532710280373, "grad_norm": 2.9180996417999268, "learning_rate": 1.2939279083985945e-06, "loss": 0.10579992830753326, "step": 5643 }, { "epoch": 1.4652128764278296, "grad_norm": 1.4095759391784668, "learning_rate": 1.2932068292198157e-06, "loss": 0.1191992536187172, "step": 5644 }, { "epoch": 1.465472481827622, "grad_norm": 1.6373028755187988, "learning_rate": 1.292486007568714e-06, "loss": 0.0843011885881424, "step": 5645 }, { "epoch": 1.4657320872274142, "grad_norm": 1.3894901275634766, "learning_rate": 1.2917654435727137e-06, "loss": 0.07696254551410675, "step": 5646 }, { "epoch": 1.4659916926272065, "grad_norm": 1.0449519157409668, "learning_rate": 1.2910451373591936e-06, "loss": 0.10056976974010468, "step": 5647 }, { "epoch": 1.4662512980269988, "grad_norm": 1.499512791633606, "learning_rate": 1.290325089055487e-06, "loss": 0.1034451425075531, "step": 5648 }, { "epoch": 1.4665109034267914, "grad_norm": 1.720740795135498, "learning_rate": 1.2896052987888825e-06, "loss": 0.10246852785348892, "step": 5649 }, { "epoch": 1.4667705088265837, "grad_norm": 3.1493091583251953, "learning_rate": 1.2888857666866243e-06, "loss": 0.050814516842365265, "step": 5650 }, { "epoch": 1.467030114226376, "grad_norm": 1.2680795192718506, "learning_rate": 1.2881664928759051e-06, "loss": 0.09447404742240906, "step": 5651 }, { "epoch": 1.4672897196261683, "grad_norm": 0.7362880706787109, "learning_rate": 1.2874474774838802e-06, "loss": 0.08619478344917297, "step": 5652 }, { "epoch": 1.4675493250259606, "grad_norm": 1.0968103408813477, "learning_rate": 1.2867287206376522e-06, "loss": 0.08586306124925613, "step": 5653 }, { "epoch": 1.467808930425753, "grad_norm": 0.689751148223877, "learning_rate": 1.2860102224642823e-06, "loss": 0.09268057346343994, "step": 5654 }, { "epoch": 1.4680685358255452, "grad_norm": 1.3819077014923096, "learning_rate": 1.2852919830907846e-06, "loss": 0.08247940987348557, "step": 5655 }, { "epoch": 1.4683281412253375, "grad_norm": 1.713128924369812, "learning_rate": 1.284574002644126e-06, "loss": 0.05314340442419052, "step": 5656 }, { "epoch": 1.4685877466251298, "grad_norm": 1.7334986925125122, "learning_rate": 1.2838562812512306e-06, "loss": 0.08096186816692352, "step": 5657 }, { "epoch": 1.4688473520249221, "grad_norm": 2.0749847888946533, "learning_rate": 1.2831388190389747e-06, "loss": 0.09092340618371964, "step": 5658 }, { "epoch": 1.4691069574247144, "grad_norm": 0.9961044192314148, "learning_rate": 1.2824216161341879e-06, "loss": 0.10566022247076035, "step": 5659 }, { "epoch": 1.4693665628245067, "grad_norm": 2.839719772338867, "learning_rate": 1.2817046726636573e-06, "loss": 0.09973575174808502, "step": 5660 }, { "epoch": 1.469626168224299, "grad_norm": 1.0837925672531128, "learning_rate": 1.28098798875412e-06, "loss": 0.0840739905834198, "step": 5661 }, { "epoch": 1.4698857736240913, "grad_norm": 2.0118982791900635, "learning_rate": 1.280271564532271e-06, "loss": 0.11009722948074341, "step": 5662 }, { "epoch": 1.4701453790238836, "grad_norm": 1.8924154043197632, "learning_rate": 1.2795554001247574e-06, "loss": 0.08563186228275299, "step": 5663 }, { "epoch": 1.470404984423676, "grad_norm": 0.916720986366272, "learning_rate": 1.2788394956581795e-06, "loss": 0.11981067061424255, "step": 5664 }, { "epoch": 1.4706645898234683, "grad_norm": 0.8267751932144165, "learning_rate": 1.2781238512590944e-06, "loss": 0.09231294691562653, "step": 5665 }, { "epoch": 1.4709241952232606, "grad_norm": 2.127951145172119, "learning_rate": 1.277408467054011e-06, "loss": 0.06189677119255066, "step": 5666 }, { "epoch": 1.471183800623053, "grad_norm": 2.0656847953796387, "learning_rate": 1.2766933431693919e-06, "loss": 0.07308537513017654, "step": 5667 }, { "epoch": 1.4714434060228454, "grad_norm": 2.0184311866760254, "learning_rate": 1.275978479731656e-06, "loss": 0.04012870043516159, "step": 5668 }, { "epoch": 1.4717030114226377, "grad_norm": 1.360625982284546, "learning_rate": 1.2752638768671738e-06, "loss": 0.10270427912473679, "step": 5669 }, { "epoch": 1.47196261682243, "grad_norm": 2.399157762527466, "learning_rate": 1.2745495347022722e-06, "loss": 0.0880652517080307, "step": 5670 }, { "epoch": 1.4722222222222223, "grad_norm": 2.463886260986328, "learning_rate": 1.273835453363229e-06, "loss": 0.08116603642702103, "step": 5671 }, { "epoch": 1.4724818276220146, "grad_norm": 1.529787540435791, "learning_rate": 1.2731216329762775e-06, "loss": 0.07879548519849777, "step": 5672 }, { "epoch": 1.472741433021807, "grad_norm": 1.4216951131820679, "learning_rate": 1.2724080736676064e-06, "loss": 0.07961148023605347, "step": 5673 }, { "epoch": 1.4730010384215992, "grad_norm": 2.298336982727051, "learning_rate": 1.2716947755633547e-06, "loss": 0.06347643584012985, "step": 5674 }, { "epoch": 1.4732606438213915, "grad_norm": 2.434246063232422, "learning_rate": 1.2709817387896173e-06, "loss": 0.053038790822029114, "step": 5675 }, { "epoch": 1.4735202492211839, "grad_norm": 1.415435552597046, "learning_rate": 1.2702689634724446e-06, "loss": 0.09925098717212677, "step": 5676 }, { "epoch": 1.4737798546209762, "grad_norm": 2.3400275707244873, "learning_rate": 1.2695564497378377e-06, "loss": 0.07463937997817993, "step": 5677 }, { "epoch": 1.4740394600207685, "grad_norm": 1.268211007118225, "learning_rate": 1.2688441977117526e-06, "loss": 0.09807127714157104, "step": 5678 }, { "epoch": 1.4742990654205608, "grad_norm": 0.91861492395401, "learning_rate": 1.2681322075200988e-06, "loss": 0.11884298920631409, "step": 5679 }, { "epoch": 1.474558670820353, "grad_norm": 0.8520916104316711, "learning_rate": 1.2674204792887406e-06, "loss": 0.10867027938365936, "step": 5680 }, { "epoch": 1.4748182762201454, "grad_norm": 2.902252197265625, "learning_rate": 1.2667090131434961e-06, "loss": 0.07032962143421173, "step": 5681 }, { "epoch": 1.4750778816199377, "grad_norm": 3.1837639808654785, "learning_rate": 1.2659978092101343e-06, "loss": 0.06527499109506607, "step": 5682 }, { "epoch": 1.47533748701973, "grad_norm": 2.3142902851104736, "learning_rate": 1.2652868676143806e-06, "loss": 0.03867443650960922, "step": 5683 }, { "epoch": 1.4755970924195223, "grad_norm": 2.495178461074829, "learning_rate": 1.2645761884819136e-06, "loss": 0.05621065944433212, "step": 5684 }, { "epoch": 1.4758566978193146, "grad_norm": 1.1836827993392944, "learning_rate": 1.263865771938365e-06, "loss": 0.108674556016922, "step": 5685 }, { "epoch": 1.476116303219107, "grad_norm": 1.3085609674453735, "learning_rate": 1.2631556181093202e-06, "loss": 0.09581151604652405, "step": 5686 }, { "epoch": 1.4763759086188992, "grad_norm": 1.8924717903137207, "learning_rate": 1.2624457271203172e-06, "loss": 0.08370251208543777, "step": 5687 }, { "epoch": 1.4766355140186915, "grad_norm": 1.9715509414672852, "learning_rate": 1.2617360990968486e-06, "loss": 0.08328446000814438, "step": 5688 }, { "epoch": 1.4768951194184838, "grad_norm": 1.4983943700790405, "learning_rate": 1.2610267341643628e-06, "loss": 0.09264165163040161, "step": 5689 }, { "epoch": 1.4771547248182761, "grad_norm": 0.9341095089912415, "learning_rate": 1.260317632448256e-06, "loss": 0.10644209384918213, "step": 5690 }, { "epoch": 1.4774143302180685, "grad_norm": 1.0531935691833496, "learning_rate": 1.2596087940738832e-06, "loss": 0.0981166809797287, "step": 5691 }, { "epoch": 1.4776739356178608, "grad_norm": 0.8248613476753235, "learning_rate": 1.2589002191665495e-06, "loss": 0.09297877550125122, "step": 5692 }, { "epoch": 1.477933541017653, "grad_norm": 1.7246915102005005, "learning_rate": 1.2581919078515161e-06, "loss": 0.07419446110725403, "step": 5693 }, { "epoch": 1.4781931464174454, "grad_norm": 0.9986926317214966, "learning_rate": 1.2574838602539955e-06, "loss": 0.11331997811794281, "step": 5694 }, { "epoch": 1.4784527518172377, "grad_norm": 1.0584954023361206, "learning_rate": 1.2567760764991536e-06, "loss": 0.10745544731616974, "step": 5695 }, { "epoch": 1.47871235721703, "grad_norm": 2.3085227012634277, "learning_rate": 1.2560685567121116e-06, "loss": 0.0835479274392128, "step": 5696 }, { "epoch": 1.4789719626168225, "grad_norm": 1.9986516237258911, "learning_rate": 1.2553613010179427e-06, "loss": 0.034180015325546265, "step": 5697 }, { "epoch": 1.4792315680166148, "grad_norm": 1.7995917797088623, "learning_rate": 1.2546543095416718e-06, "loss": 0.0868450254201889, "step": 5698 }, { "epoch": 1.4794911734164071, "grad_norm": 1.9591569900512695, "learning_rate": 1.2539475824082813e-06, "loss": 0.06041466444730759, "step": 5699 }, { "epoch": 1.4797507788161994, "grad_norm": 1.7182599306106567, "learning_rate": 1.2532411197427034e-06, "loss": 0.06769169867038727, "step": 5700 }, { "epoch": 1.4800103842159917, "grad_norm": 1.2773650884628296, "learning_rate": 1.2525349216698229e-06, "loss": 0.09465811401605606, "step": 5701 }, { "epoch": 1.480269989615784, "grad_norm": 0.9308030009269714, "learning_rate": 1.2518289883144819e-06, "loss": 0.09842954576015472, "step": 5702 }, { "epoch": 1.4805295950155763, "grad_norm": 2.780235767364502, "learning_rate": 1.2511233198014716e-06, "loss": 0.05155376344919205, "step": 5703 }, { "epoch": 1.4807892004153687, "grad_norm": 1.4054419994354248, "learning_rate": 1.2504179162555392e-06, "loss": 0.08762387931346893, "step": 5704 }, { "epoch": 1.481048805815161, "grad_norm": 1.6226224899291992, "learning_rate": 1.2497127778013833e-06, "loss": 0.094053715467453, "step": 5705 }, { "epoch": 1.4813084112149533, "grad_norm": 2.948937177658081, "learning_rate": 1.2490079045636554e-06, "loss": 0.07260679453611374, "step": 5706 }, { "epoch": 1.4815680166147456, "grad_norm": 2.9181272983551025, "learning_rate": 1.248303296666963e-06, "loss": 0.046356067061424255, "step": 5707 }, { "epoch": 1.4818276220145379, "grad_norm": 1.3606159687042236, "learning_rate": 1.247598954235863e-06, "loss": 0.07554192841053009, "step": 5708 }, { "epoch": 1.4820872274143302, "grad_norm": 2.9878454208374023, "learning_rate": 1.2468948773948669e-06, "loss": 0.08727576583623886, "step": 5709 }, { "epoch": 1.4823468328141225, "grad_norm": 2.3973701000213623, "learning_rate": 1.2461910662684404e-06, "loss": 0.046026188880205154, "step": 5710 }, { "epoch": 1.4826064382139148, "grad_norm": 1.0969562530517578, "learning_rate": 1.2454875209810005e-06, "loss": 0.11267851293087006, "step": 5711 }, { "epoch": 1.482866043613707, "grad_norm": 2.6374216079711914, "learning_rate": 1.2447842416569192e-06, "loss": 0.11075352132320404, "step": 5712 }, { "epoch": 1.4831256490134994, "grad_norm": 2.0368244647979736, "learning_rate": 1.244081228420518e-06, "loss": 0.07452729344367981, "step": 5713 }, { "epoch": 1.4833852544132917, "grad_norm": 2.0523061752319336, "learning_rate": 1.2433784813960743e-06, "loss": 0.05204875022172928, "step": 5714 }, { "epoch": 1.4836448598130842, "grad_norm": 1.6806813478469849, "learning_rate": 1.2426760007078188e-06, "loss": 0.0568968765437603, "step": 5715 }, { "epoch": 1.4839044652128766, "grad_norm": 1.203555703163147, "learning_rate": 1.2419737864799334e-06, "loss": 0.09440246224403381, "step": 5716 }, { "epoch": 1.4841640706126689, "grad_norm": 2.361259937286377, "learning_rate": 1.2412718388365532e-06, "loss": 0.0639498233795166, "step": 5717 }, { "epoch": 1.4844236760124612, "grad_norm": 3.184847593307495, "learning_rate": 1.2405701579017658e-06, "loss": 0.09396247565746307, "step": 5718 }, { "epoch": 1.4846832814122535, "grad_norm": 2.435622453689575, "learning_rate": 1.2398687437996133e-06, "loss": 0.07179130613803864, "step": 5719 }, { "epoch": 1.4849428868120458, "grad_norm": 1.4138110876083374, "learning_rate": 1.239167596654091e-06, "loss": 0.08902369439601898, "step": 5720 }, { "epoch": 1.485202492211838, "grad_norm": 1.0773680210113525, "learning_rate": 1.2384667165891424e-06, "loss": 0.0865655168890953, "step": 5721 }, { "epoch": 1.4854620976116304, "grad_norm": 2.8593995571136475, "learning_rate": 1.2377661037286698e-06, "loss": 0.042107872664928436, "step": 5722 }, { "epoch": 1.4857217030114227, "grad_norm": 1.6515899896621704, "learning_rate": 1.2370657581965236e-06, "loss": 0.11191229522228241, "step": 5723 }, { "epoch": 1.485981308411215, "grad_norm": 3.658540725708008, "learning_rate": 1.2363656801165104e-06, "loss": 0.07566960901021957, "step": 5724 }, { "epoch": 1.4862409138110073, "grad_norm": 1.7108403444290161, "learning_rate": 1.2356658696123873e-06, "loss": 0.08095668256282806, "step": 5725 }, { "epoch": 1.4865005192107996, "grad_norm": 2.6335995197296143, "learning_rate": 1.234966326807864e-06, "loss": 0.0764516294002533, "step": 5726 }, { "epoch": 1.486760124610592, "grad_norm": 1.8402124643325806, "learning_rate": 1.2342670518266057e-06, "loss": 0.05888313800096512, "step": 5727 }, { "epoch": 1.4870197300103842, "grad_norm": 2.2578351497650146, "learning_rate": 1.2335680447922266e-06, "loss": 0.12267117202281952, "step": 5728 }, { "epoch": 1.4872793354101765, "grad_norm": 1.3422932624816895, "learning_rate": 1.2328693058282945e-06, "loss": 0.09164048731327057, "step": 5729 }, { "epoch": 1.4875389408099688, "grad_norm": 1.882962942123413, "learning_rate": 1.2321708350583324e-06, "loss": 0.07343505322933197, "step": 5730 }, { "epoch": 1.4877985462097612, "grad_norm": 1.5193208456039429, "learning_rate": 1.231472632605813e-06, "loss": 0.09521572291851044, "step": 5731 }, { "epoch": 1.4880581516095535, "grad_norm": 1.8988357782363892, "learning_rate": 1.230774698594162e-06, "loss": 0.10152153670787811, "step": 5732 }, { "epoch": 1.4883177570093458, "grad_norm": 1.6441453695297241, "learning_rate": 1.230077033146759e-06, "loss": 0.09823909401893616, "step": 5733 }, { "epoch": 1.488577362409138, "grad_norm": 0.8628275394439697, "learning_rate": 1.2293796363869345e-06, "loss": 0.10111837089061737, "step": 5734 }, { "epoch": 1.4888369678089304, "grad_norm": 1.327332854270935, "learning_rate": 1.2286825084379733e-06, "loss": 0.09744225442409515, "step": 5735 }, { "epoch": 1.4890965732087227, "grad_norm": 1.034943699836731, "learning_rate": 1.227985649423111e-06, "loss": 0.08435948938131332, "step": 5736 }, { "epoch": 1.489356178608515, "grad_norm": 1.7696892023086548, "learning_rate": 1.2272890594655356e-06, "loss": 0.06239713355898857, "step": 5737 }, { "epoch": 1.4896157840083073, "grad_norm": 2.4956374168395996, "learning_rate": 1.2265927386883898e-06, "loss": 0.08168935030698776, "step": 5738 }, { "epoch": 1.4898753894080996, "grad_norm": 1.6976205110549927, "learning_rate": 1.2258966872147664e-06, "loss": 0.0812419056892395, "step": 5739 }, { "epoch": 1.490134994807892, "grad_norm": 1.0400067567825317, "learning_rate": 1.2252009051677106e-06, "loss": 0.10496683418750763, "step": 5740 }, { "epoch": 1.4903946002076842, "grad_norm": 1.751832127571106, "learning_rate": 1.2245053926702224e-06, "loss": 0.03455689549446106, "step": 5741 }, { "epoch": 1.4906542056074765, "grad_norm": 0.9957082867622375, "learning_rate": 1.2238101498452507e-06, "loss": 0.1005820631980896, "step": 5742 }, { "epoch": 1.4909138110072688, "grad_norm": 1.172450065612793, "learning_rate": 1.2231151768157008e-06, "loss": 0.09898596256971359, "step": 5743 }, { "epoch": 1.4911734164070611, "grad_norm": 2.1748461723327637, "learning_rate": 1.2224204737044252e-06, "loss": 0.05263746902346611, "step": 5744 }, { "epoch": 1.4914330218068537, "grad_norm": 2.595632314682007, "learning_rate": 1.221726040634233e-06, "loss": 0.06835458427667618, "step": 5745 }, { "epoch": 1.491692627206646, "grad_norm": 2.4144632816314697, "learning_rate": 1.2210318777278844e-06, "loss": 0.0977252647280693, "step": 5746 }, { "epoch": 1.4919522326064383, "grad_norm": 1.1003177165985107, "learning_rate": 1.2203379851080914e-06, "loss": 0.09769140183925629, "step": 5747 }, { "epoch": 1.4922118380062306, "grad_norm": 2.0031490325927734, "learning_rate": 1.219644362897518e-06, "loss": 0.06074557080864906, "step": 5748 }, { "epoch": 1.492471443406023, "grad_norm": 1.3136252164840698, "learning_rate": 1.21895101121878e-06, "loss": 0.08865530043840408, "step": 5749 }, { "epoch": 1.4927310488058152, "grad_norm": 2.325861692428589, "learning_rate": 1.218257930194447e-06, "loss": 0.0951375663280487, "step": 5750 }, { "epoch": 1.4929906542056075, "grad_norm": 2.4470999240875244, "learning_rate": 1.2175651199470408e-06, "loss": 0.045542433857917786, "step": 5751 }, { "epoch": 1.4932502596053998, "grad_norm": 1.6070786714553833, "learning_rate": 1.2168725805990319e-06, "loss": 0.04633967950940132, "step": 5752 }, { "epoch": 1.4935098650051921, "grad_norm": 2.310304641723633, "learning_rate": 1.2161803122728476e-06, "loss": 0.06171228736639023, "step": 5753 }, { "epoch": 1.4937694704049844, "grad_norm": 1.7906163930892944, "learning_rate": 1.2154883150908635e-06, "loss": 0.08289233595132828, "step": 5754 }, { "epoch": 1.4940290758047767, "grad_norm": 2.1916778087615967, "learning_rate": 1.2147965891754104e-06, "loss": 0.05682789534330368, "step": 5755 }, { "epoch": 1.494288681204569, "grad_norm": 1.8379619121551514, "learning_rate": 1.2141051346487692e-06, "loss": 0.0762040987610817, "step": 5756 }, { "epoch": 1.4945482866043613, "grad_norm": 1.0547549724578857, "learning_rate": 1.2134139516331715e-06, "loss": 0.09755740314722061, "step": 5757 }, { "epoch": 1.4948078920041536, "grad_norm": 1.3927891254425049, "learning_rate": 1.212723040250805e-06, "loss": 0.09816627204418182, "step": 5758 }, { "epoch": 1.495067497403946, "grad_norm": 2.5994112491607666, "learning_rate": 1.212032400623806e-06, "loss": 0.07766634225845337, "step": 5759 }, { "epoch": 1.4953271028037383, "grad_norm": 1.8464524745941162, "learning_rate": 1.2113420328742629e-06, "loss": 0.0880374014377594, "step": 5760 }, { "epoch": 1.4955867082035306, "grad_norm": 0.2612735629081726, "learning_rate": 1.2106519371242188e-06, "loss": 0.05049097537994385, "step": 5761 }, { "epoch": 1.4958463136033229, "grad_norm": 1.6355018615722656, "learning_rate": 1.2099621134956656e-06, "loss": 0.06480468809604645, "step": 5762 }, { "epoch": 1.4961059190031154, "grad_norm": 2.265650749206543, "learning_rate": 1.209272562110548e-06, "loss": 0.0703086256980896, "step": 5763 }, { "epoch": 1.4963655244029077, "grad_norm": 1.815948724746704, "learning_rate": 1.2085832830907642e-06, "loss": 0.06967715919017792, "step": 5764 }, { "epoch": 1.4966251298027, "grad_norm": 1.9809038639068604, "learning_rate": 1.2078942765581611e-06, "loss": 0.06524708867073059, "step": 5765 }, { "epoch": 1.4968847352024923, "grad_norm": 2.447256088256836, "learning_rate": 1.2072055426345414e-06, "loss": 0.0424753800034523, "step": 5766 }, { "epoch": 1.4971443406022846, "grad_norm": 1.2722976207733154, "learning_rate": 1.2065170814416564e-06, "loss": 0.08206459134817123, "step": 5767 }, { "epoch": 1.497403946002077, "grad_norm": 1.9310564994812012, "learning_rate": 1.2058288931012095e-06, "loss": 0.0979795902967453, "step": 5768 }, { "epoch": 1.4976635514018692, "grad_norm": 0.9769346117973328, "learning_rate": 1.205140977734858e-06, "loss": 0.11122790724039078, "step": 5769 }, { "epoch": 1.4979231568016615, "grad_norm": 1.9449039697647095, "learning_rate": 1.2044533354642095e-06, "loss": 0.08771148324012756, "step": 5770 }, { "epoch": 1.4981827622014539, "grad_norm": 1.2140088081359863, "learning_rate": 1.2037659664108222e-06, "loss": 0.10005956143140793, "step": 5771 }, { "epoch": 1.4984423676012462, "grad_norm": 2.6532633304595947, "learning_rate": 1.2030788706962085e-06, "loss": 0.0717071145772934, "step": 5772 }, { "epoch": 1.4987019730010385, "grad_norm": 0.9837157726287842, "learning_rate": 1.20239204844183e-06, "loss": 0.10126663744449615, "step": 5773 }, { "epoch": 1.4989615784008308, "grad_norm": 1.1744115352630615, "learning_rate": 1.2017054997691032e-06, "loss": 0.060050949454307556, "step": 5774 }, { "epoch": 1.499221183800623, "grad_norm": 1.4299930334091187, "learning_rate": 1.2010192247993915e-06, "loss": 0.08206599950790405, "step": 5775 }, { "epoch": 1.4994807892004154, "grad_norm": 0.8468270301818848, "learning_rate": 1.200333223654014e-06, "loss": 0.09252676367759705, "step": 5776 }, { "epoch": 1.4997403946002077, "grad_norm": 2.3176424503326416, "learning_rate": 1.19964749645424e-06, "loss": 0.057071998715400696, "step": 5777 }, { "epoch": 1.5, "grad_norm": 1.8632142543792725, "learning_rate": 1.1989620433212908e-06, "loss": 0.07106736302375793, "step": 5778 }, { "epoch": 1.5002596053997923, "grad_norm": 1.2760462760925293, "learning_rate": 1.1982768643763382e-06, "loss": 0.08406925201416016, "step": 5779 }, { "epoch": 1.5005192107995846, "grad_norm": 0.9271950721740723, "learning_rate": 1.1975919597405054e-06, "loss": 0.08835961669683456, "step": 5780 }, { "epoch": 1.500778816199377, "grad_norm": 1.5447776317596436, "learning_rate": 1.1969073295348686e-06, "loss": 0.09516650438308716, "step": 5781 }, { "epoch": 1.5010384215991692, "grad_norm": 2.149942636489868, "learning_rate": 1.1962229738804565e-06, "loss": 0.08867911994457245, "step": 5782 }, { "epoch": 1.5012980269989615, "grad_norm": 1.8775110244750977, "learning_rate": 1.1955388928982444e-06, "loss": 0.07917845249176025, "step": 5783 }, { "epoch": 1.5015576323987538, "grad_norm": 2.2977819442749023, "learning_rate": 1.1948550867091643e-06, "loss": 0.07869499921798706, "step": 5784 }, { "epoch": 1.5018172377985461, "grad_norm": 1.3614263534545898, "learning_rate": 1.1941715554340965e-06, "loss": 0.08016819506883621, "step": 5785 }, { "epoch": 1.5020768431983385, "grad_norm": 1.4871180057525635, "learning_rate": 1.1934882991938737e-06, "loss": 0.08839903771877289, "step": 5786 }, { "epoch": 1.5023364485981308, "grad_norm": 1.3625625371932983, "learning_rate": 1.1928053181092804e-06, "loss": 0.10540559142827988, "step": 5787 }, { "epoch": 1.502596053997923, "grad_norm": 1.9542540311813354, "learning_rate": 1.1921226123010515e-06, "loss": 0.07575495541095734, "step": 5788 }, { "epoch": 1.5028556593977154, "grad_norm": 1.8201959133148193, "learning_rate": 1.1914401818898745e-06, "loss": 0.09374161809682846, "step": 5789 }, { "epoch": 1.5031152647975077, "grad_norm": 2.0099387168884277, "learning_rate": 1.1907580269963874e-06, "loss": 0.03918146342039108, "step": 5790 }, { "epoch": 1.5033748701973, "grad_norm": 1.2159491777420044, "learning_rate": 1.1900761477411782e-06, "loss": 0.08705278486013412, "step": 5791 }, { "epoch": 1.5036344755970923, "grad_norm": 1.7154775857925415, "learning_rate": 1.1893945442447892e-06, "loss": 0.10810242593288422, "step": 5792 }, { "epoch": 1.5038940809968846, "grad_norm": 0.9757118821144104, "learning_rate": 1.188713216627712e-06, "loss": 0.0917116105556488, "step": 5793 }, { "epoch": 1.504153686396677, "grad_norm": 1.1762350797653198, "learning_rate": 1.1880321650103884e-06, "loss": 0.10762188583612442, "step": 5794 }, { "epoch": 1.5044132917964692, "grad_norm": 1.0582555532455444, "learning_rate": 1.1873513895132145e-06, "loss": 0.09704665839672089, "step": 5795 }, { "epoch": 1.5046728971962615, "grad_norm": 1.1265954971313477, "learning_rate": 1.1866708902565346e-06, "loss": 0.08284904062747955, "step": 5796 }, { "epoch": 1.504932502596054, "grad_norm": 1.171717882156372, "learning_rate": 1.1859906673606462e-06, "loss": 0.07881435006856918, "step": 5797 }, { "epoch": 1.5051921079958464, "grad_norm": 1.1078985929489136, "learning_rate": 1.1853107209457966e-06, "loss": 0.1015382707118988, "step": 5798 }, { "epoch": 1.5054517133956387, "grad_norm": 1.0388264656066895, "learning_rate": 1.1846310511321845e-06, "loss": 0.11410538852214813, "step": 5799 }, { "epoch": 1.505711318795431, "grad_norm": 2.6792261600494385, "learning_rate": 1.1839516580399607e-06, "loss": 0.07751795649528503, "step": 5800 }, { "epoch": 1.5059709241952233, "grad_norm": 2.867060661315918, "learning_rate": 1.1832725417892263e-06, "loss": 0.10994866490364075, "step": 5801 }, { "epoch": 1.5062305295950156, "grad_norm": 1.4404475688934326, "learning_rate": 1.1825937025000325e-06, "loss": 0.09765414893627167, "step": 5802 }, { "epoch": 1.5064901349948079, "grad_norm": 1.8358343839645386, "learning_rate": 1.181915140292384e-06, "loss": 0.10955282300710678, "step": 5803 }, { "epoch": 1.5067497403946002, "grad_norm": 1.546090006828308, "learning_rate": 1.1812368552862336e-06, "loss": 0.08055579662322998, "step": 5804 }, { "epoch": 1.5070093457943925, "grad_norm": 2.0532968044281006, "learning_rate": 1.1805588476014886e-06, "loss": 0.07610231637954712, "step": 5805 }, { "epoch": 1.5072689511941848, "grad_norm": 1.0706595182418823, "learning_rate": 1.1798811173580024e-06, "loss": 0.09671146422624588, "step": 5806 }, { "epoch": 1.507528556593977, "grad_norm": 1.9327586889266968, "learning_rate": 1.1792036646755838e-06, "loss": 0.06804097443819046, "step": 5807 }, { "epoch": 1.5077881619937694, "grad_norm": 1.2185311317443848, "learning_rate": 1.1785264896739914e-06, "loss": 0.08770112693309784, "step": 5808 }, { "epoch": 1.508047767393562, "grad_norm": 2.6607582569122314, "learning_rate": 1.177849592472934e-06, "loss": 0.05288448929786682, "step": 5809 }, { "epoch": 1.5083073727933543, "grad_norm": 0.8761628270149231, "learning_rate": 1.1771729731920707e-06, "loss": 0.09159867465496063, "step": 5810 }, { "epoch": 1.5085669781931466, "grad_norm": 0.9034680128097534, "learning_rate": 1.1764966319510124e-06, "loss": 0.10463310778141022, "step": 5811 }, { "epoch": 1.5088265835929389, "grad_norm": 0.9802106022834778, "learning_rate": 1.1758205688693219e-06, "loss": 0.10163730382919312, "step": 5812 }, { "epoch": 1.5090861889927312, "grad_norm": 2.310793161392212, "learning_rate": 1.1751447840665105e-06, "loss": 0.06777505576610565, "step": 5813 }, { "epoch": 1.5093457943925235, "grad_norm": 2.1753487586975098, "learning_rate": 1.1744692776620414e-06, "loss": 0.09601671993732452, "step": 5814 }, { "epoch": 1.5096053997923158, "grad_norm": 2.443192720413208, "learning_rate": 1.17379404977533e-06, "loss": 0.09404326975345612, "step": 5815 }, { "epoch": 1.509865005192108, "grad_norm": 1.7750352621078491, "learning_rate": 1.1731191005257399e-06, "loss": 0.06695036590099335, "step": 5816 }, { "epoch": 1.5101246105919004, "grad_norm": 2.5221972465515137, "learning_rate": 1.1724444300325864e-06, "loss": 0.04635847732424736, "step": 5817 }, { "epoch": 1.5103842159916927, "grad_norm": 1.4641765356063843, "learning_rate": 1.1717700384151375e-06, "loss": 0.11309404671192169, "step": 5818 }, { "epoch": 1.510643821391485, "grad_norm": 0.7971563935279846, "learning_rate": 1.1710959257926084e-06, "loss": 0.1154252290725708, "step": 5819 }, { "epoch": 1.5109034267912773, "grad_norm": 2.1520743370056152, "learning_rate": 1.1704220922841678e-06, "loss": 0.04799816012382507, "step": 5820 }, { "epoch": 1.5111630321910696, "grad_norm": 2.3805530071258545, "learning_rate": 1.169748538008934e-06, "loss": 0.07850069552659988, "step": 5821 }, { "epoch": 1.511422637590862, "grad_norm": 1.588114857673645, "learning_rate": 1.1690752630859748e-06, "loss": 0.07140526920557022, "step": 5822 }, { "epoch": 1.5116822429906542, "grad_norm": 1.4065258502960205, "learning_rate": 1.1684022676343118e-06, "loss": 0.10363844037055969, "step": 5823 }, { "epoch": 1.5119418483904465, "grad_norm": 1.3956371545791626, "learning_rate": 1.1677295517729138e-06, "loss": 0.08315222710371017, "step": 5824 }, { "epoch": 1.5122014537902388, "grad_norm": 1.8374898433685303, "learning_rate": 1.1670571156207013e-06, "loss": 0.08022747933864594, "step": 5825 }, { "epoch": 1.5124610591900312, "grad_norm": 1.9068835973739624, "learning_rate": 1.1663849592965466e-06, "loss": 0.08344580978155136, "step": 5826 }, { "epoch": 1.5127206645898235, "grad_norm": 1.0507676601409912, "learning_rate": 1.1657130829192708e-06, "loss": 0.08987107127904892, "step": 5827 }, { "epoch": 1.5129802699896158, "grad_norm": 1.2006851434707642, "learning_rate": 1.1650414866076472e-06, "loss": 0.08297563344240189, "step": 5828 }, { "epoch": 1.513239875389408, "grad_norm": 1.4668716192245483, "learning_rate": 1.1643701704803977e-06, "loss": 0.10421878099441528, "step": 5829 }, { "epoch": 1.5134994807892004, "grad_norm": 1.5741909742355347, "learning_rate": 1.1636991346561956e-06, "loss": 0.08259332180023193, "step": 5830 }, { "epoch": 1.5137590861889927, "grad_norm": 3.428436756134033, "learning_rate": 1.1630283792536657e-06, "loss": 0.09747858345508575, "step": 5831 }, { "epoch": 1.514018691588785, "grad_norm": 1.8216413259506226, "learning_rate": 1.1623579043913814e-06, "loss": 0.07828821241855621, "step": 5832 }, { "epoch": 1.5142782969885773, "grad_norm": 1.648822546005249, "learning_rate": 1.1616877101878665e-06, "loss": 0.08433644473552704, "step": 5833 }, { "epoch": 1.5145379023883696, "grad_norm": 0.9031714797019958, "learning_rate": 1.161017796761598e-06, "loss": 0.10085232555866241, "step": 5834 }, { "epoch": 1.514797507788162, "grad_norm": 2.4060072898864746, "learning_rate": 1.1603481642309992e-06, "loss": 0.07324910163879395, "step": 5835 }, { "epoch": 1.5150571131879542, "grad_norm": 1.9873666763305664, "learning_rate": 1.1596788127144485e-06, "loss": 0.11132475733757019, "step": 5836 }, { "epoch": 1.5153167185877465, "grad_norm": 3.581393241882324, "learning_rate": 1.1590097423302683e-06, "loss": 0.09182844310998917, "step": 5837 }, { "epoch": 1.5155763239875388, "grad_norm": 1.0517174005508423, "learning_rate": 1.1583409531967372e-06, "loss": 0.09724234789609909, "step": 5838 }, { "epoch": 1.5158359293873311, "grad_norm": 2.3837313652038574, "learning_rate": 1.1576724454320828e-06, "loss": 0.08827230334281921, "step": 5839 }, { "epoch": 1.5160955347871234, "grad_norm": 0.8421794772148132, "learning_rate": 1.1570042191544787e-06, "loss": 0.10070481896400452, "step": 5840 }, { "epoch": 1.5163551401869158, "grad_norm": 2.8190770149230957, "learning_rate": 1.156336274482055e-06, "loss": 0.0886351466178894, "step": 5841 }, { "epoch": 1.516614745586708, "grad_norm": 1.875177264213562, "learning_rate": 1.1556686115328872e-06, "loss": 0.08483252674341202, "step": 5842 }, { "epoch": 1.5168743509865004, "grad_norm": 2.8100593090057373, "learning_rate": 1.1550012304250043e-06, "loss": 0.07415728271007538, "step": 5843 }, { "epoch": 1.5171339563862927, "grad_norm": 2.703791379928589, "learning_rate": 1.154334131276383e-06, "loss": 0.063306525349617, "step": 5844 }, { "epoch": 1.5173935617860852, "grad_norm": 2.2567360401153564, "learning_rate": 1.1536673142049507e-06, "loss": 0.07167874276638031, "step": 5845 }, { "epoch": 1.5176531671858775, "grad_norm": 1.3372658491134644, "learning_rate": 1.153000779328587e-06, "loss": 0.07448028028011322, "step": 5846 }, { "epoch": 1.5179127725856698, "grad_norm": 2.3266685009002686, "learning_rate": 1.1523345267651188e-06, "loss": 0.07329260557889938, "step": 5847 }, { "epoch": 1.5181723779854621, "grad_norm": 1.381507396697998, "learning_rate": 1.1516685566323243e-06, "loss": 0.09969665110111237, "step": 5848 }, { "epoch": 1.5184319833852544, "grad_norm": 1.5878626108169556, "learning_rate": 1.1510028690479323e-06, "loss": 0.09962528944015503, "step": 5849 }, { "epoch": 1.5186915887850467, "grad_norm": 1.803161859512329, "learning_rate": 1.1503374641296204e-06, "loss": 0.0837327316403389, "step": 5850 }, { "epoch": 1.518951194184839, "grad_norm": 1.249117374420166, "learning_rate": 1.1496723419950178e-06, "loss": 0.10206258296966553, "step": 5851 }, { "epoch": 1.5192107995846313, "grad_norm": 1.356480598449707, "learning_rate": 1.149007502761703e-06, "loss": 0.1091877818107605, "step": 5852 }, { "epoch": 1.5194704049844237, "grad_norm": 1.1940269470214844, "learning_rate": 1.1483429465472026e-06, "loss": 0.10270051658153534, "step": 5853 }, { "epoch": 1.519730010384216, "grad_norm": 2.057420253753662, "learning_rate": 1.1476786734689973e-06, "loss": 0.08206522464752197, "step": 5854 }, { "epoch": 1.5199896157840083, "grad_norm": 1.2337254285812378, "learning_rate": 1.1470146836445141e-06, "loss": 0.10427269339561462, "step": 5855 }, { "epoch": 1.5202492211838006, "grad_norm": 0.9999926686286926, "learning_rate": 1.1463509771911305e-06, "loss": 0.1170228123664856, "step": 5856 }, { "epoch": 1.520508826583593, "grad_norm": 2.706402063369751, "learning_rate": 1.1456875542261765e-06, "loss": 0.10557707399129868, "step": 5857 }, { "epoch": 1.5207684319833854, "grad_norm": 1.668836236000061, "learning_rate": 1.145024414866928e-06, "loss": 0.07919663190841675, "step": 5858 }, { "epoch": 1.5210280373831777, "grad_norm": 1.8590058088302612, "learning_rate": 1.1443615592306147e-06, "loss": 0.0779566690325737, "step": 5859 }, { "epoch": 1.52128764278297, "grad_norm": 2.9656214714050293, "learning_rate": 1.1436989874344135e-06, "loss": 0.05410052090883255, "step": 5860 }, { "epoch": 1.5215472481827623, "grad_norm": 1.9943851232528687, "learning_rate": 1.143036699595451e-06, "loss": 0.05495370179414749, "step": 5861 }, { "epoch": 1.5218068535825546, "grad_norm": 1.0608376264572144, "learning_rate": 1.1423746958308066e-06, "loss": 0.09123346954584122, "step": 5862 }, { "epoch": 1.522066458982347, "grad_norm": 1.119866132736206, "learning_rate": 1.1417129762575057e-06, "loss": 0.07703015208244324, "step": 5863 }, { "epoch": 1.5223260643821392, "grad_norm": 1.0342732667922974, "learning_rate": 1.1410515409925252e-06, "loss": 0.09457436203956604, "step": 5864 }, { "epoch": 1.5225856697819315, "grad_norm": 1.3068370819091797, "learning_rate": 1.1403903901527929e-06, "loss": 0.08127337694168091, "step": 5865 }, { "epoch": 1.5228452751817239, "grad_norm": 0.9507443904876709, "learning_rate": 1.1397295238551847e-06, "loss": 0.10530754923820496, "step": 5866 }, { "epoch": 1.5231048805815162, "grad_norm": 1.3955154418945312, "learning_rate": 1.1390689422165262e-06, "loss": 0.06124638766050339, "step": 5867 }, { "epoch": 1.5233644859813085, "grad_norm": 0.897025465965271, "learning_rate": 1.1384086453535925e-06, "loss": 0.08828877657651901, "step": 5868 }, { "epoch": 1.5236240913811008, "grad_norm": 2.0262744426727295, "learning_rate": 1.1377486333831099e-06, "loss": 0.09674763679504395, "step": 5869 }, { "epoch": 1.523883696780893, "grad_norm": 2.0246031284332275, "learning_rate": 1.1370889064217545e-06, "loss": 0.08498959243297577, "step": 5870 }, { "epoch": 1.5241433021806854, "grad_norm": 1.7955573797225952, "learning_rate": 1.1364294645861483e-06, "loss": 0.08530449867248535, "step": 5871 }, { "epoch": 1.5244029075804777, "grad_norm": 1.8091404438018799, "learning_rate": 1.1357703079928676e-06, "loss": 0.06464444100856781, "step": 5872 }, { "epoch": 1.52466251298027, "grad_norm": 0.9872140288352966, "learning_rate": 1.1351114367584347e-06, "loss": 0.09971894323825836, "step": 5873 }, { "epoch": 1.5249221183800623, "grad_norm": 1.6760300397872925, "learning_rate": 1.1344528509993246e-06, "loss": 0.07571645081043243, "step": 5874 }, { "epoch": 1.5251817237798546, "grad_norm": 2.00217604637146, "learning_rate": 1.1337945508319593e-06, "loss": 0.09325873851776123, "step": 5875 }, { "epoch": 1.525441329179647, "grad_norm": 2.64308762550354, "learning_rate": 1.1331365363727106e-06, "loss": 0.08904168754816055, "step": 5876 }, { "epoch": 1.5257009345794392, "grad_norm": 2.1677675247192383, "learning_rate": 1.1324788077379015e-06, "loss": 0.033731624484062195, "step": 5877 }, { "epoch": 1.5259605399792315, "grad_norm": 1.1646326780319214, "learning_rate": 1.1318213650438028e-06, "loss": 0.08474485576152802, "step": 5878 }, { "epoch": 1.5262201453790238, "grad_norm": 2.092038154602051, "learning_rate": 1.131164208406635e-06, "loss": 0.08385191857814789, "step": 5879 }, { "epoch": 1.5264797507788161, "grad_norm": 1.9521901607513428, "learning_rate": 1.130507337942569e-06, "loss": 0.0719255656003952, "step": 5880 }, { "epoch": 1.5267393561786085, "grad_norm": 1.3952186107635498, "learning_rate": 1.129850753767724e-06, "loss": 0.0758938118815422, "step": 5881 }, { "epoch": 1.5269989615784008, "grad_norm": 2.2499945163726807, "learning_rate": 1.1291944559981697e-06, "loss": 0.07185406982898712, "step": 5882 }, { "epoch": 1.527258566978193, "grad_norm": 2.137253761291504, "learning_rate": 1.128538444749924e-06, "loss": 0.05678912252187729, "step": 5883 }, { "epoch": 1.5275181723779854, "grad_norm": 1.814542531967163, "learning_rate": 1.1278827201389539e-06, "loss": 0.07464494556188583, "step": 5884 }, { "epoch": 1.5277777777777777, "grad_norm": 2.1878576278686523, "learning_rate": 1.1272272822811784e-06, "loss": 0.04065092280507088, "step": 5885 }, { "epoch": 1.52803738317757, "grad_norm": 2.8025293350219727, "learning_rate": 1.1265721312924624e-06, "loss": 0.07984146475791931, "step": 5886 }, { "epoch": 1.5282969885773623, "grad_norm": 1.1053236722946167, "learning_rate": 1.1259172672886216e-06, "loss": 0.08024618029594421, "step": 5887 }, { "epoch": 1.5285565939771546, "grad_norm": 2.068321466445923, "learning_rate": 1.1252626903854218e-06, "loss": 0.060270145535469055, "step": 5888 }, { "epoch": 1.528816199376947, "grad_norm": 1.0689607858657837, "learning_rate": 1.1246084006985764e-06, "loss": 0.11017032712697983, "step": 5889 }, { "epoch": 1.5290758047767392, "grad_norm": 2.7909185886383057, "learning_rate": 1.1239543983437498e-06, "loss": 0.09223061054944992, "step": 5890 }, { "epoch": 1.5293354101765315, "grad_norm": 1.3023602962493896, "learning_rate": 1.1233006834365542e-06, "loss": 0.08692699670791626, "step": 5891 }, { "epoch": 1.5295950155763238, "grad_norm": 1.0879043340682983, "learning_rate": 1.1226472560925502e-06, "loss": 0.10699359327554703, "step": 5892 }, { "epoch": 1.5298546209761164, "grad_norm": 1.9316412210464478, "learning_rate": 1.1219941164272515e-06, "loss": 0.059960126876831055, "step": 5893 }, { "epoch": 1.5301142263759087, "grad_norm": 1.9012889862060547, "learning_rate": 1.1213412645561154e-06, "loss": 0.0637778490781784, "step": 5894 }, { "epoch": 1.530373831775701, "grad_norm": 2.382544994354248, "learning_rate": 1.1206887005945521e-06, "loss": 0.07231378555297852, "step": 5895 }, { "epoch": 1.5306334371754933, "grad_norm": 1.728304386138916, "learning_rate": 1.1200364246579213e-06, "loss": 0.07849890738725662, "step": 5896 }, { "epoch": 1.5308930425752856, "grad_norm": 2.2154061794281006, "learning_rate": 1.119384436861529e-06, "loss": 0.08376012742519379, "step": 5897 }, { "epoch": 1.5311526479750779, "grad_norm": 2.1187996864318848, "learning_rate": 1.1187327373206323e-06, "loss": 0.04348108172416687, "step": 5898 }, { "epoch": 1.5314122533748702, "grad_norm": 2.1931252479553223, "learning_rate": 1.118081326150436e-06, "loss": 0.07941049337387085, "step": 5899 }, { "epoch": 1.5316718587746625, "grad_norm": 2.016223907470703, "learning_rate": 1.117430203466095e-06, "loss": 0.08044205605983734, "step": 5900 }, { "epoch": 1.5319314641744548, "grad_norm": 2.4664618968963623, "learning_rate": 1.1167793693827145e-06, "loss": 0.10029575228691101, "step": 5901 }, { "epoch": 1.532191069574247, "grad_norm": 1.8014073371887207, "learning_rate": 1.1161288240153446e-06, "loss": 0.053211234509944916, "step": 5902 }, { "epoch": 1.5324506749740394, "grad_norm": 1.0412076711654663, "learning_rate": 1.1154785674789883e-06, "loss": 0.08953572064638138, "step": 5903 }, { "epoch": 1.5327102803738317, "grad_norm": 2.473123073577881, "learning_rate": 1.1148285998885956e-06, "loss": 0.06042682006955147, "step": 5904 }, { "epoch": 1.5329698857736243, "grad_norm": 2.074002742767334, "learning_rate": 1.114178921359066e-06, "loss": 0.06860284507274628, "step": 5905 }, { "epoch": 1.5332294911734166, "grad_norm": 1.0775072574615479, "learning_rate": 1.113529532005248e-06, "loss": 0.09581921994686127, "step": 5906 }, { "epoch": 1.5334890965732089, "grad_norm": 1.161952257156372, "learning_rate": 1.112880431941938e-06, "loss": 0.09173540771007538, "step": 5907 }, { "epoch": 1.5337487019730012, "grad_norm": 1.0712785720825195, "learning_rate": 1.112231621283883e-06, "loss": 0.08780352771282196, "step": 5908 }, { "epoch": 1.5340083073727935, "grad_norm": 1.0159274339675903, "learning_rate": 1.1115831001457775e-06, "loss": 0.08758530020713806, "step": 5909 }, { "epoch": 1.5342679127725858, "grad_norm": 1.875688076019287, "learning_rate": 1.1109348686422643e-06, "loss": 0.09176848828792572, "step": 5910 }, { "epoch": 1.534527518172378, "grad_norm": 2.258369207382202, "learning_rate": 1.1102869268879379e-06, "loss": 0.09075839817523956, "step": 5911 }, { "epoch": 1.5347871235721704, "grad_norm": 0.8374190926551819, "learning_rate": 1.1096392749973373e-06, "loss": 0.08896776288747787, "step": 5912 }, { "epoch": 1.5350467289719627, "grad_norm": 2.7251393795013428, "learning_rate": 1.1089919130849543e-06, "loss": 0.07161704450845718, "step": 5913 }, { "epoch": 1.535306334371755, "grad_norm": 2.15234375, "learning_rate": 1.108344841265227e-06, "loss": 0.06100110709667206, "step": 5914 }, { "epoch": 1.5355659397715473, "grad_norm": 3.068291664123535, "learning_rate": 1.1076980596525422e-06, "loss": 0.09514187276363373, "step": 5915 }, { "epoch": 1.5358255451713396, "grad_norm": 2.4938855171203613, "learning_rate": 1.1070515683612373e-06, "loss": 0.06469926238059998, "step": 5916 }, { "epoch": 1.536085150571132, "grad_norm": 2.255168914794922, "learning_rate": 1.1064053675055967e-06, "loss": 0.09565214812755585, "step": 5917 }, { "epoch": 1.5363447559709242, "grad_norm": 2.463813066482544, "learning_rate": 1.1057594571998528e-06, "loss": 0.05613786727190018, "step": 5918 }, { "epoch": 1.5366043613707165, "grad_norm": 1.5252999067306519, "learning_rate": 1.10511383755819e-06, "loss": 0.056713469326496124, "step": 5919 }, { "epoch": 1.5368639667705088, "grad_norm": 0.8105835914611816, "learning_rate": 1.1044685086947376e-06, "loss": 0.08680049329996109, "step": 5920 }, { "epoch": 1.5371235721703012, "grad_norm": 2.319265127182007, "learning_rate": 1.1038234707235742e-06, "loss": 0.08861161768436432, "step": 5921 }, { "epoch": 1.5373831775700935, "grad_norm": 1.3692545890808105, "learning_rate": 1.1031787237587296e-06, "loss": 0.08376135677099228, "step": 5922 }, { "epoch": 1.5376427829698858, "grad_norm": 1.8669941425323486, "learning_rate": 1.102534267914179e-06, "loss": 0.08315816521644592, "step": 5923 }, { "epoch": 1.537902388369678, "grad_norm": 1.8570722341537476, "learning_rate": 1.101890103303849e-06, "loss": 0.059590063989162445, "step": 5924 }, { "epoch": 1.5381619937694704, "grad_norm": 1.6171973943710327, "learning_rate": 1.1012462300416104e-06, "loss": 0.04596174880862236, "step": 5925 }, { "epoch": 1.5384215991692627, "grad_norm": 2.3687362670898438, "learning_rate": 1.100602648241287e-06, "loss": 0.10099482536315918, "step": 5926 }, { "epoch": 1.538681204569055, "grad_norm": 1.0463370084762573, "learning_rate": 1.0999593580166501e-06, "loss": 0.09870936721563339, "step": 5927 }, { "epoch": 1.5389408099688473, "grad_norm": 1.570513367652893, "learning_rate": 1.099316359481418e-06, "loss": 0.05798112601041794, "step": 5928 }, { "epoch": 1.5392004153686396, "grad_norm": 0.9751373529434204, "learning_rate": 1.0986736527492573e-06, "loss": 0.10834896564483643, "step": 5929 }, { "epoch": 1.539460020768432, "grad_norm": 2.4260916709899902, "learning_rate": 1.0980312379337844e-06, "loss": 0.11138235032558441, "step": 5930 }, { "epoch": 1.5397196261682242, "grad_norm": 1.202669382095337, "learning_rate": 1.097389115148563e-06, "loss": 0.08391304314136505, "step": 5931 }, { "epoch": 1.5399792315680165, "grad_norm": 1.5917757749557495, "learning_rate": 1.0967472845071078e-06, "loss": 0.028681006282567978, "step": 5932 }, { "epoch": 1.5402388369678088, "grad_norm": 2.0228652954101562, "learning_rate": 1.0961057461228768e-06, "loss": 0.08474241197109222, "step": 5933 }, { "epoch": 1.5404984423676011, "grad_norm": 2.0304510593414307, "learning_rate": 1.0954645001092818e-06, "loss": 0.044685617089271545, "step": 5934 }, { "epoch": 1.5407580477673934, "grad_norm": 1.885511040687561, "learning_rate": 1.0948235465796784e-06, "loss": 0.07300519943237305, "step": 5935 }, { "epoch": 1.5410176531671858, "grad_norm": 2.1458792686462402, "learning_rate": 1.0941828856473738e-06, "loss": 0.10996004939079285, "step": 5936 }, { "epoch": 1.541277258566978, "grad_norm": 2.3857064247131348, "learning_rate": 1.0935425174256224e-06, "loss": 0.07510793209075928, "step": 5937 }, { "epoch": 1.5415368639667704, "grad_norm": 2.067589521408081, "learning_rate": 1.092902442027625e-06, "loss": 0.05173785239458084, "step": 5938 }, { "epoch": 1.5417964693665627, "grad_norm": 1.705296277999878, "learning_rate": 1.0922626595665343e-06, "loss": 0.07079563289880753, "step": 5939 }, { "epoch": 1.542056074766355, "grad_norm": 1.1254098415374756, "learning_rate": 1.091623170155448e-06, "loss": 0.08963233232498169, "step": 5940 }, { "epoch": 1.5423156801661475, "grad_norm": 2.466247797012329, "learning_rate": 1.0909839739074126e-06, "loss": 0.0595567487180233, "step": 5941 }, { "epoch": 1.5425752855659398, "grad_norm": 1.996450662612915, "learning_rate": 1.0903450709354253e-06, "loss": 0.054763488471508026, "step": 5942 }, { "epoch": 1.5428348909657321, "grad_norm": 2.0409975051879883, "learning_rate": 1.0897064613524273e-06, "loss": 0.09125161170959473, "step": 5943 }, { "epoch": 1.5430944963655244, "grad_norm": 2.0096259117126465, "learning_rate": 1.0890681452713123e-06, "loss": 0.07967916131019592, "step": 5944 }, { "epoch": 1.5433541017653167, "grad_norm": 1.18385910987854, "learning_rate": 1.0884301228049187e-06, "loss": 0.06624393165111542, "step": 5945 }, { "epoch": 1.543613707165109, "grad_norm": 1.6968759298324585, "learning_rate": 1.0877923940660338e-06, "loss": 0.07664497196674347, "step": 5946 }, { "epoch": 1.5438733125649013, "grad_norm": 2.223393440246582, "learning_rate": 1.087154959167395e-06, "loss": 0.07341691851615906, "step": 5947 }, { "epoch": 1.5441329179646937, "grad_norm": 2.0515177249908447, "learning_rate": 1.0865178182216852e-06, "loss": 0.052643559873104095, "step": 5948 }, { "epoch": 1.544392523364486, "grad_norm": 1.7670931816101074, "learning_rate": 1.0858809713415355e-06, "loss": 0.10088475048542023, "step": 5949 }, { "epoch": 1.5446521287642783, "grad_norm": 1.7452009916305542, "learning_rate": 1.0852444186395279e-06, "loss": 0.05758816748857498, "step": 5950 }, { "epoch": 1.5449117341640706, "grad_norm": 1.663089632987976, "learning_rate": 1.0846081602281889e-06, "loss": 0.09032410383224487, "step": 5951 }, { "epoch": 1.5451713395638629, "grad_norm": 2.4358677864074707, "learning_rate": 1.0839721962199942e-06, "loss": 0.07618533074855804, "step": 5952 }, { "epoch": 1.5454309449636554, "grad_norm": 1.4477732181549072, "learning_rate": 1.083336526727369e-06, "loss": 0.09758014976978302, "step": 5953 }, { "epoch": 1.5456905503634477, "grad_norm": 1.3861182928085327, "learning_rate": 1.0827011518626836e-06, "loss": 0.09721972048282623, "step": 5954 }, { "epoch": 1.54595015576324, "grad_norm": 1.8003337383270264, "learning_rate": 1.0820660717382598e-06, "loss": 0.06780886650085449, "step": 5955 }, { "epoch": 1.5462097611630323, "grad_norm": 1.5142900943756104, "learning_rate": 1.0814312864663625e-06, "loss": 0.08933378756046295, "step": 5956 }, { "epoch": 1.5464693665628246, "grad_norm": 2.464782238006592, "learning_rate": 1.0807967961592086e-06, "loss": 0.08892744779586792, "step": 5957 }, { "epoch": 1.546728971962617, "grad_norm": 0.9430475831031799, "learning_rate": 1.0801626009289621e-06, "loss": 0.11945274472236633, "step": 5958 }, { "epoch": 1.5469885773624092, "grad_norm": 3.6356890201568604, "learning_rate": 1.0795287008877334e-06, "loss": 0.09159818291664124, "step": 5959 }, { "epoch": 1.5472481827622016, "grad_norm": 1.1948353052139282, "learning_rate": 1.0788950961475819e-06, "loss": 0.09441187977790833, "step": 5960 }, { "epoch": 1.5475077881619939, "grad_norm": 3.2875218391418457, "learning_rate": 1.0782617868205138e-06, "loss": 0.06938505172729492, "step": 5961 }, { "epoch": 1.5477673935617862, "grad_norm": 0.9128623008728027, "learning_rate": 1.0776287730184837e-06, "loss": 0.10684406757354736, "step": 5962 }, { "epoch": 1.5480269989615785, "grad_norm": 1.6035733222961426, "learning_rate": 1.0769960548533955e-06, "loss": 0.08461470901966095, "step": 5963 }, { "epoch": 1.5482866043613708, "grad_norm": 1.1189491748809814, "learning_rate": 1.076363632437097e-06, "loss": 0.0888393446803093, "step": 5964 }, { "epoch": 1.548546209761163, "grad_norm": 1.9569964408874512, "learning_rate": 1.0757315058813877e-06, "loss": 0.05991591885685921, "step": 5965 }, { "epoch": 1.5488058151609554, "grad_norm": 2.626361131668091, "learning_rate": 1.0750996752980117e-06, "loss": 0.12007100880146027, "step": 5966 }, { "epoch": 1.5490654205607477, "grad_norm": 0.7756895422935486, "learning_rate": 1.0744681407986636e-06, "loss": 0.10402464866638184, "step": 5967 }, { "epoch": 1.54932502596054, "grad_norm": 1.1473429203033447, "learning_rate": 1.073836902494984e-06, "loss": 0.10138577222824097, "step": 5968 }, { "epoch": 1.5495846313603323, "grad_norm": 0.8722885847091675, "learning_rate": 1.0732059604985602e-06, "loss": 0.10378310829401016, "step": 5969 }, { "epoch": 1.5498442367601246, "grad_norm": 2.161135196685791, "learning_rate": 1.0725753149209295e-06, "loss": 0.09570495784282684, "step": 5970 }, { "epoch": 1.550103842159917, "grad_norm": 1.3648200035095215, "learning_rate": 1.0719449658735754e-06, "loss": 0.05612984299659729, "step": 5971 }, { "epoch": 1.5503634475597092, "grad_norm": 3.748950242996216, "learning_rate": 1.0713149134679281e-06, "loss": 0.07052658498287201, "step": 5972 }, { "epoch": 1.5506230529595015, "grad_norm": 1.7094436883926392, "learning_rate": 1.0706851578153681e-06, "loss": 0.043802037835121155, "step": 5973 }, { "epoch": 1.5508826583592938, "grad_norm": 0.8666313886642456, "learning_rate": 1.0700556990272203e-06, "loss": 0.10900184512138367, "step": 5974 }, { "epoch": 1.5511422637590861, "grad_norm": 1.4844446182250977, "learning_rate": 1.06942653721476e-06, "loss": 0.0820506140589714, "step": 5975 }, { "epoch": 1.5514018691588785, "grad_norm": 1.3920783996582031, "learning_rate": 1.0687976724892078e-06, "loss": 0.0972975492477417, "step": 5976 }, { "epoch": 1.5516614745586708, "grad_norm": 2.258906364440918, "learning_rate": 1.0681691049617322e-06, "loss": 0.07622990012168884, "step": 5977 }, { "epoch": 1.551921079958463, "grad_norm": 1.1793261766433716, "learning_rate": 1.0675408347434504e-06, "loss": 0.09030319005250931, "step": 5978 }, { "epoch": 1.5521806853582554, "grad_norm": 1.9590553045272827, "learning_rate": 1.066912861945426e-06, "loss": 0.05454747751355171, "step": 5979 }, { "epoch": 1.5524402907580477, "grad_norm": 1.569707989692688, "learning_rate": 1.0662851866786694e-06, "loss": 0.09068014472723007, "step": 5980 }, { "epoch": 1.55269989615784, "grad_norm": 1.3129007816314697, "learning_rate": 1.0656578090541402e-06, "loss": 0.08584757149219513, "step": 5981 }, { "epoch": 1.5529595015576323, "grad_norm": 1.5905967950820923, "learning_rate": 1.0650307291827443e-06, "loss": 0.09893740713596344, "step": 5982 }, { "epoch": 1.5532191069574246, "grad_norm": 1.3335070610046387, "learning_rate": 1.0644039471753343e-06, "loss": 0.09877856075763702, "step": 5983 }, { "epoch": 1.553478712357217, "grad_norm": 2.397242546081543, "learning_rate": 1.0637774631427117e-06, "loss": 0.051177941262722015, "step": 5984 }, { "epoch": 1.5537383177570092, "grad_norm": 2.921319007873535, "learning_rate": 1.0631512771956238e-06, "loss": 0.07376348972320557, "step": 5985 }, { "epoch": 1.5539979231568015, "grad_norm": 1.1866257190704346, "learning_rate": 1.0625253894447676e-06, "loss": 0.09968172013759613, "step": 5986 }, { "epoch": 1.5542575285565938, "grad_norm": 2.5980262756347656, "learning_rate": 1.0618998000007832e-06, "loss": 0.07805516570806503, "step": 5987 }, { "epoch": 1.5545171339563861, "grad_norm": 1.1087676286697388, "learning_rate": 1.061274508974262e-06, "loss": 0.031959645450115204, "step": 5988 }, { "epoch": 1.5547767393561787, "grad_norm": 1.171015739440918, "learning_rate": 1.060649516475741e-06, "loss": 0.08943591266870499, "step": 5989 }, { "epoch": 1.555036344755971, "grad_norm": 1.6398059129714966, "learning_rate": 1.0600248226157051e-06, "loss": 0.07190883159637451, "step": 5990 }, { "epoch": 1.5552959501557633, "grad_norm": 2.135258197784424, "learning_rate": 1.0594004275045849e-06, "loss": 0.048442691564559937, "step": 5991 }, { "epoch": 1.5555555555555556, "grad_norm": 1.7062385082244873, "learning_rate": 1.058776331252759e-06, "loss": 0.07222253084182739, "step": 5992 }, { "epoch": 1.5558151609553479, "grad_norm": 1.4753800630569458, "learning_rate": 1.058152533970554e-06, "loss": 0.09594085812568665, "step": 5993 }, { "epoch": 1.5560747663551402, "grad_norm": 1.361330509185791, "learning_rate": 1.0575290357682439e-06, "loss": 0.08110220730304718, "step": 5994 }, { "epoch": 1.5563343717549325, "grad_norm": 2.1211981773376465, "learning_rate": 1.0569058367560468e-06, "loss": 0.06400386244058609, "step": 5995 }, { "epoch": 1.5565939771547248, "grad_norm": 1.592323899269104, "learning_rate": 1.0562829370441304e-06, "loss": 0.08627843111753464, "step": 5996 }, { "epoch": 1.5568535825545171, "grad_norm": 2.2356388568878174, "learning_rate": 1.055660336742611e-06, "loss": 0.039577022194862366, "step": 5997 }, { "epoch": 1.5571131879543094, "grad_norm": 0.7992229461669922, "learning_rate": 1.0550380359615487e-06, "loss": 0.088558129966259, "step": 5998 }, { "epoch": 1.5573727933541017, "grad_norm": 1.8543915748596191, "learning_rate": 1.054416034810952e-06, "loss": 0.05616449564695358, "step": 5999 }, { "epoch": 1.557632398753894, "grad_norm": 2.23537015914917, "learning_rate": 1.0537943334007762e-06, "loss": 0.08849579095840454, "step": 6000 }, { "epoch": 1.5578920041536866, "grad_norm": 2.3783740997314453, "learning_rate": 1.0531729318409242e-06, "loss": 0.07064476609230042, "step": 6001 }, { "epoch": 1.5581516095534789, "grad_norm": 1.1646288633346558, "learning_rate": 1.052551830241247e-06, "loss": 0.08904772996902466, "step": 6002 }, { "epoch": 1.5584112149532712, "grad_norm": 1.934389352798462, "learning_rate": 1.0519310287115383e-06, "loss": 0.05755787342786789, "step": 6003 }, { "epoch": 1.5586708203530635, "grad_norm": 2.3524763584136963, "learning_rate": 1.051310527361544e-06, "loss": 0.051617808640003204, "step": 6004 }, { "epoch": 1.5589304257528558, "grad_norm": 1.5526752471923828, "learning_rate": 1.0506903263009536e-06, "loss": 0.07671065628528595, "step": 6005 }, { "epoch": 1.559190031152648, "grad_norm": 1.4573789834976196, "learning_rate": 1.0500704256394037e-06, "loss": 0.0942082554101944, "step": 6006 }, { "epoch": 1.5594496365524404, "grad_norm": 2.5055289268493652, "learning_rate": 1.04945082548648e-06, "loss": 0.07920192182064056, "step": 6007 }, { "epoch": 1.5597092419522327, "grad_norm": 1.0316025018692017, "learning_rate": 1.0488315259517124e-06, "loss": 0.09964659065008163, "step": 6008 }, { "epoch": 1.559968847352025, "grad_norm": 2.662199020385742, "learning_rate": 1.0482125271445798e-06, "loss": 0.05017231032252312, "step": 6009 }, { "epoch": 1.5602284527518173, "grad_norm": 1.648695707321167, "learning_rate": 1.0475938291745069e-06, "loss": 0.06395022571086884, "step": 6010 }, { "epoch": 1.5604880581516096, "grad_norm": 1.0733568668365479, "learning_rate": 1.0469754321508641e-06, "loss": 0.08882426470518112, "step": 6011 }, { "epoch": 1.560747663551402, "grad_norm": 0.8474011421203613, "learning_rate": 1.046357336182972e-06, "loss": 0.122789166867733, "step": 6012 }, { "epoch": 1.5610072689511942, "grad_norm": 1.3454252481460571, "learning_rate": 1.045739541380094e-06, "loss": 0.09444201737642288, "step": 6013 }, { "epoch": 1.5612668743509865, "grad_norm": 1.3023254871368408, "learning_rate": 1.0451220478514424e-06, "loss": 0.06471452862024307, "step": 6014 }, { "epoch": 1.5615264797507789, "grad_norm": 1.4778265953063965, "learning_rate": 1.044504855706177e-06, "loss": 0.07651277631521225, "step": 6015 }, { "epoch": 1.5617860851505712, "grad_norm": 3.4426002502441406, "learning_rate": 1.0438879650534016e-06, "loss": 0.06733302772045135, "step": 6016 }, { "epoch": 1.5620456905503635, "grad_norm": 2.3659517765045166, "learning_rate": 1.0432713760021696e-06, "loss": 0.07566571980714798, "step": 6017 }, { "epoch": 1.5623052959501558, "grad_norm": 1.6272128820419312, "learning_rate": 1.0426550886614797e-06, "loss": 0.07934263348579407, "step": 6018 }, { "epoch": 1.562564901349948, "grad_norm": 1.785159945487976, "learning_rate": 1.0420391031402763e-06, "loss": 0.05892294645309448, "step": 6019 }, { "epoch": 1.5628245067497404, "grad_norm": 1.0810701847076416, "learning_rate": 1.041423419547453e-06, "loss": 0.08474431186914444, "step": 6020 }, { "epoch": 1.5630841121495327, "grad_norm": 1.4543123245239258, "learning_rate": 1.0408080379918481e-06, "loss": 0.08031681180000305, "step": 6021 }, { "epoch": 1.563343717549325, "grad_norm": 2.429208755493164, "learning_rate": 1.0401929585822459e-06, "loss": 0.10000836849212646, "step": 6022 }, { "epoch": 1.5636033229491173, "grad_norm": 1.8598854541778564, "learning_rate": 1.0395781814273801e-06, "loss": 0.08853572607040405, "step": 6023 }, { "epoch": 1.5638629283489096, "grad_norm": 1.5444996356964111, "learning_rate": 1.0389637066359278e-06, "loss": 0.07918647676706314, "step": 6024 }, { "epoch": 1.564122533748702, "grad_norm": 2.3224399089813232, "learning_rate": 1.0383495343165154e-06, "loss": 0.08406790345907211, "step": 6025 }, { "epoch": 1.5643821391484942, "grad_norm": 1.7902377843856812, "learning_rate": 1.037735664577713e-06, "loss": 0.07694902271032333, "step": 6026 }, { "epoch": 1.5646417445482865, "grad_norm": 1.2982217073440552, "learning_rate": 1.0371220975280393e-06, "loss": 0.07837854325771332, "step": 6027 }, { "epoch": 1.5649013499480788, "grad_norm": 1.7747238874435425, "learning_rate": 1.0365088332759596e-06, "loss": 0.0601622648537159, "step": 6028 }, { "epoch": 1.5651609553478711, "grad_norm": 2.8230671882629395, "learning_rate": 1.0358958719298845e-06, "loss": 0.059058889746665955, "step": 6029 }, { "epoch": 1.5654205607476634, "grad_norm": 3.676715135574341, "learning_rate": 1.0352832135981711e-06, "loss": 0.028110139071941376, "step": 6030 }, { "epoch": 1.5656801661474558, "grad_norm": 1.1602373123168945, "learning_rate": 1.0346708583891237e-06, "loss": 0.0946439728140831, "step": 6031 }, { "epoch": 1.565939771547248, "grad_norm": 1.4692538976669312, "learning_rate": 1.034058806410993e-06, "loss": 0.08345486223697662, "step": 6032 }, { "epoch": 1.5661993769470404, "grad_norm": 1.6239076852798462, "learning_rate": 1.0334470577719755e-06, "loss": 0.0789816826581955, "step": 6033 }, { "epoch": 1.5664589823468327, "grad_norm": 1.0591564178466797, "learning_rate": 1.0328356125802136e-06, "loss": 0.11294466257095337, "step": 6034 }, { "epoch": 1.566718587746625, "grad_norm": 1.2330132722854614, "learning_rate": 1.0322244709437983e-06, "loss": 0.08318421244621277, "step": 6035 }, { "epoch": 1.5669781931464173, "grad_norm": 2.1802587509155273, "learning_rate": 1.0316136329707649e-06, "loss": 0.09324286133050919, "step": 6036 }, { "epoch": 1.5672377985462098, "grad_norm": 2.183981418609619, "learning_rate": 1.0310030987690945e-06, "loss": 0.07759514451026917, "step": 6037 }, { "epoch": 1.5674974039460021, "grad_norm": 1.6893408298492432, "learning_rate": 1.030392868446717e-06, "loss": 0.07664933800697327, "step": 6038 }, { "epoch": 1.5677570093457944, "grad_norm": 1.9691989421844482, "learning_rate": 1.029782942111506e-06, "loss": 0.08318433165550232, "step": 6039 }, { "epoch": 1.5680166147455867, "grad_norm": 2.140294075012207, "learning_rate": 1.0291733198712838e-06, "loss": 0.0758683979511261, "step": 6040 }, { "epoch": 1.568276220145379, "grad_norm": 1.4325587749481201, "learning_rate": 1.0285640018338166e-06, "loss": 0.07307692617177963, "step": 6041 }, { "epoch": 1.5685358255451713, "grad_norm": 2.804560422897339, "learning_rate": 1.027954988106818e-06, "loss": 0.09665507078170776, "step": 6042 }, { "epoch": 1.5687954309449637, "grad_norm": 1.3367815017700195, "learning_rate": 1.0273462787979485e-06, "loss": 0.06793618947267532, "step": 6043 }, { "epoch": 1.569055036344756, "grad_norm": 1.1213245391845703, "learning_rate": 1.0267378740148131e-06, "loss": 0.10802378505468369, "step": 6044 }, { "epoch": 1.5693146417445483, "grad_norm": 2.2777302265167236, "learning_rate": 1.0261297738649637e-06, "loss": 0.05957474187016487, "step": 6045 }, { "epoch": 1.5695742471443406, "grad_norm": 2.328894853591919, "learning_rate": 1.0255219784558992e-06, "loss": 0.08342821896076202, "step": 6046 }, { "epoch": 1.5698338525441329, "grad_norm": 2.591844081878662, "learning_rate": 1.0249144878950634e-06, "loss": 0.06517252326011658, "step": 6047 }, { "epoch": 1.5700934579439252, "grad_norm": 3.2984490394592285, "learning_rate": 1.0243073022898472e-06, "loss": 0.08367905020713806, "step": 6048 }, { "epoch": 1.5703530633437177, "grad_norm": 3.296880006790161, "learning_rate": 1.023700421747587e-06, "loss": 0.03693362697958946, "step": 6049 }, { "epoch": 1.57061266874351, "grad_norm": 1.9752914905548096, "learning_rate": 1.0230938463755648e-06, "loss": 0.0771501287817955, "step": 6050 }, { "epoch": 1.5708722741433023, "grad_norm": 0.9184299111366272, "learning_rate": 1.0224875762810102e-06, "loss": 0.10943619161844254, "step": 6051 }, { "epoch": 1.5711318795430946, "grad_norm": 2.521049976348877, "learning_rate": 1.021881611571097e-06, "loss": 0.04783886671066284, "step": 6052 }, { "epoch": 1.571391484942887, "grad_norm": 1.1316463947296143, "learning_rate": 1.0212759523529456e-06, "loss": 0.08376044034957886, "step": 6053 }, { "epoch": 1.5716510903426792, "grad_norm": 1.1367353200912476, "learning_rate": 1.020670598733624e-06, "loss": 0.09541178494691849, "step": 6054 }, { "epoch": 1.5719106957424716, "grad_norm": 2.8396732807159424, "learning_rate": 1.0200655508201438e-06, "loss": 0.06486783921718597, "step": 6055 }, { "epoch": 1.5721703011422639, "grad_norm": 1.1447618007659912, "learning_rate": 1.019460808719465e-06, "loss": 0.10555700957775116, "step": 6056 }, { "epoch": 1.5724299065420562, "grad_norm": 1.0826016664505005, "learning_rate": 1.0188563725384898e-06, "loss": 0.09422542154788971, "step": 6057 }, { "epoch": 1.5726895119418485, "grad_norm": 2.08689546585083, "learning_rate": 1.01825224238407e-06, "loss": 0.0869084894657135, "step": 6058 }, { "epoch": 1.5729491173416408, "grad_norm": 1.2910898923873901, "learning_rate": 1.0176484183630035e-06, "loss": 0.07216355204582214, "step": 6059 }, { "epoch": 1.573208722741433, "grad_norm": 0.8986571431159973, "learning_rate": 1.0170449005820296e-06, "loss": 0.09801703691482544, "step": 6060 }, { "epoch": 1.5734683281412254, "grad_norm": 2.6856298446655273, "learning_rate": 1.0164416891478382e-06, "loss": 0.043433841317892075, "step": 6061 }, { "epoch": 1.5737279335410177, "grad_norm": 1.7277212142944336, "learning_rate": 1.0158387841670626e-06, "loss": 0.07513932138681412, "step": 6062 }, { "epoch": 1.57398753894081, "grad_norm": 2.3058204650878906, "learning_rate": 1.0152361857462836e-06, "loss": 0.07032407075166702, "step": 6063 }, { "epoch": 1.5742471443406023, "grad_norm": 1.3450485467910767, "learning_rate": 1.0146338939920262e-06, "loss": 0.07219764590263367, "step": 6064 }, { "epoch": 1.5745067497403946, "grad_norm": 2.233898639678955, "learning_rate": 1.014031909010761e-06, "loss": 0.0858054906129837, "step": 6065 }, { "epoch": 1.574766355140187, "grad_norm": 1.2294669151306152, "learning_rate": 1.0134302309089065e-06, "loss": 0.09202608466148376, "step": 6066 }, { "epoch": 1.5750259605399792, "grad_norm": 1.149354100227356, "learning_rate": 1.0128288597928252e-06, "loss": 0.07439355552196503, "step": 6067 }, { "epoch": 1.5752855659397715, "grad_norm": 2.1694748401641846, "learning_rate": 1.0122277957688244e-06, "loss": 0.08416637778282166, "step": 6068 }, { "epoch": 1.5755451713395638, "grad_norm": 1.5713344812393188, "learning_rate": 1.0116270389431606e-06, "loss": 0.059271190315485, "step": 6069 }, { "epoch": 1.5758047767393562, "grad_norm": 2.8026952743530273, "learning_rate": 1.0110265894220322e-06, "loss": 0.08533775806427002, "step": 6070 }, { "epoch": 1.5760643821391485, "grad_norm": 1.490256905555725, "learning_rate": 1.010426447311586e-06, "loss": 0.07082581520080566, "step": 6071 }, { "epoch": 1.5763239875389408, "grad_norm": 1.3162482976913452, "learning_rate": 1.0098266127179132e-06, "loss": 0.09017488360404968, "step": 6072 }, { "epoch": 1.576583592938733, "grad_norm": 2.8712151050567627, "learning_rate": 1.00922708574705e-06, "loss": 0.04308103024959564, "step": 6073 }, { "epoch": 1.5768431983385254, "grad_norm": 1.5883300304412842, "learning_rate": 1.0086278665049801e-06, "loss": 0.0939520001411438, "step": 6074 }, { "epoch": 1.5771028037383177, "grad_norm": 3.161482095718384, "learning_rate": 1.0080289550976314e-06, "loss": 0.0581890344619751, "step": 6075 }, { "epoch": 1.57736240913811, "grad_norm": 1.5903140306472778, "learning_rate": 1.0074303516308767e-06, "loss": 0.10218745470046997, "step": 6076 }, { "epoch": 1.5776220145379023, "grad_norm": 1.7155605554580688, "learning_rate": 1.0068320562105369e-06, "loss": 0.11738792061805725, "step": 6077 }, { "epoch": 1.5778816199376946, "grad_norm": 1.9470221996307373, "learning_rate": 1.0062340689423762e-06, "loss": 0.04183310642838478, "step": 6078 }, { "epoch": 1.578141225337487, "grad_norm": 1.2876535654067993, "learning_rate": 1.0056363899321053e-06, "loss": 0.07597589492797852, "step": 6079 }, { "epoch": 1.5784008307372792, "grad_norm": 2.5674660205841064, "learning_rate": 1.0050390192853801e-06, "loss": 0.08106155693531036, "step": 6080 }, { "epoch": 1.5786604361370715, "grad_norm": 1.9637043476104736, "learning_rate": 1.0044419571078016e-06, "loss": 0.07716383039951324, "step": 6081 }, { "epoch": 1.5789200415368638, "grad_norm": 1.1976619958877563, "learning_rate": 1.0038452035049174e-06, "loss": 0.10545198619365692, "step": 6082 }, { "epoch": 1.5791796469366561, "grad_norm": 1.873313307762146, "learning_rate": 1.0032487585822196e-06, "loss": 0.10379631817340851, "step": 6083 }, { "epoch": 1.5794392523364484, "grad_norm": 1.8095868825912476, "learning_rate": 1.0026526224451455e-06, "loss": 0.06067759543657303, "step": 6084 }, { "epoch": 1.579698857736241, "grad_norm": 1.0608868598937988, "learning_rate": 1.0020567951990794e-06, "loss": 0.07668692618608475, "step": 6085 }, { "epoch": 1.5799584631360333, "grad_norm": 1.24435555934906, "learning_rate": 1.0014612769493494e-06, "loss": 0.08463361859321594, "step": 6086 }, { "epoch": 1.5802180685358256, "grad_norm": 1.2398103475570679, "learning_rate": 1.000866067801229e-06, "loss": 0.09549593925476074, "step": 6087 }, { "epoch": 1.5804776739356179, "grad_norm": 2.409507989883423, "learning_rate": 1.0002711678599378e-06, "loss": 0.05551242455840111, "step": 6088 }, { "epoch": 1.5807372793354102, "grad_norm": 1.347996473312378, "learning_rate": 9.996765772306405e-07, "loss": 0.08371710777282715, "step": 6089 }, { "epoch": 1.5809968847352025, "grad_norm": 1.8880901336669922, "learning_rate": 9.990822960184488e-07, "loss": 0.08186201751232147, "step": 6090 }, { "epoch": 1.5812564901349948, "grad_norm": 2.094942092895508, "learning_rate": 9.984883243284146e-07, "loss": 0.07011152803897858, "step": 6091 }, { "epoch": 1.5815160955347871, "grad_norm": 2.1061389446258545, "learning_rate": 9.978946622655414e-07, "loss": 0.08058494329452515, "step": 6092 }, { "epoch": 1.5817757009345794, "grad_norm": 3.401148796081543, "learning_rate": 9.973013099347734e-07, "loss": 0.08639975637197495, "step": 6093 }, { "epoch": 1.5820353063343717, "grad_norm": 2.0214951038360596, "learning_rate": 9.967082674410028e-07, "loss": 0.07546569406986237, "step": 6094 }, { "epoch": 1.582294911734164, "grad_norm": 1.5400265455245972, "learning_rate": 9.961155348890654e-07, "loss": 0.09870176017284393, "step": 6095 }, { "epoch": 1.5825545171339563, "grad_norm": 1.6822298765182495, "learning_rate": 9.955231123837422e-07, "loss": 0.09558405727148056, "step": 6096 }, { "epoch": 1.5828141225337489, "grad_norm": 1.1249655485153198, "learning_rate": 9.94931000029761e-07, "loss": 0.08420605957508087, "step": 6097 }, { "epoch": 1.5830737279335412, "grad_norm": 1.2712353467941284, "learning_rate": 9.943391979317938e-07, "loss": 0.0844879224896431, "step": 6098 }, { "epoch": 1.5833333333333335, "grad_norm": 1.9617359638214111, "learning_rate": 9.93747706194456e-07, "loss": 0.03531111776828766, "step": 6099 }, { "epoch": 1.5835929387331258, "grad_norm": 2.779381513595581, "learning_rate": 9.931565249223116e-07, "loss": 0.08972206711769104, "step": 6100 }, { "epoch": 1.583852544132918, "grad_norm": 1.7724953889846802, "learning_rate": 9.925656542198666e-07, "loss": 0.08500641584396362, "step": 6101 }, { "epoch": 1.5841121495327104, "grad_norm": 1.2339333295822144, "learning_rate": 9.919750941915748e-07, "loss": 0.09647046029567719, "step": 6102 }, { "epoch": 1.5843717549325027, "grad_norm": 1.9689987897872925, "learning_rate": 9.91384844941833e-07, "loss": 0.09169982373714447, "step": 6103 }, { "epoch": 1.584631360332295, "grad_norm": 1.8940378427505493, "learning_rate": 9.907949065749827e-07, "loss": 0.04640091955661774, "step": 6104 }, { "epoch": 1.5848909657320873, "grad_norm": 2.6122889518737793, "learning_rate": 9.902052791953131e-07, "loss": 0.0628657415509224, "step": 6105 }, { "epoch": 1.5851505711318796, "grad_norm": 1.285253643989563, "learning_rate": 9.896159629070564e-07, "loss": 0.08354566991329193, "step": 6106 }, { "epoch": 1.585410176531672, "grad_norm": 1.0120586156845093, "learning_rate": 9.890269578143894e-07, "loss": 0.08819891512393951, "step": 6107 }, { "epoch": 1.5856697819314642, "grad_norm": 1.1507296562194824, "learning_rate": 9.884382640214357e-07, "loss": 0.08883897960186005, "step": 6108 }, { "epoch": 1.5859293873312565, "grad_norm": 2.924661159515381, "learning_rate": 9.878498816322621e-07, "loss": 0.09316301345825195, "step": 6109 }, { "epoch": 1.5861889927310489, "grad_norm": 1.4970853328704834, "learning_rate": 9.87261810750882e-07, "loss": 0.07438605278730392, "step": 6110 }, { "epoch": 1.5864485981308412, "grad_norm": 1.002127766609192, "learning_rate": 9.866740514812524e-07, "loss": 0.0943244993686676, "step": 6111 }, { "epoch": 1.5867082035306335, "grad_norm": 1.751143455505371, "learning_rate": 9.86086603927275e-07, "loss": 0.04865844547748566, "step": 6112 }, { "epoch": 1.5869678089304258, "grad_norm": 1.7690833806991577, "learning_rate": 9.854994681927992e-07, "loss": 0.06378083676099777, "step": 6113 }, { "epoch": 1.587227414330218, "grad_norm": 2.455989360809326, "learning_rate": 9.849126443816146e-07, "loss": 0.031909700483083725, "step": 6114 }, { "epoch": 1.5874870197300104, "grad_norm": 1.162693738937378, "learning_rate": 9.843261325974591e-07, "loss": 0.032582834362983704, "step": 6115 }, { "epoch": 1.5877466251298027, "grad_norm": 1.1120893955230713, "learning_rate": 9.837399329440155e-07, "loss": 0.11378929018974304, "step": 6116 }, { "epoch": 1.588006230529595, "grad_norm": 1.070883870124817, "learning_rate": 9.831540455249097e-07, "loss": 0.09994000196456909, "step": 6117 }, { "epoch": 1.5882658359293873, "grad_norm": 2.8815407752990723, "learning_rate": 9.825684704437132e-07, "loss": 0.07097972184419632, "step": 6118 }, { "epoch": 1.5885254413291796, "grad_norm": 1.542039394378662, "learning_rate": 9.819832078039419e-07, "loss": 0.10529028624296188, "step": 6119 }, { "epoch": 1.588785046728972, "grad_norm": 1.7995392084121704, "learning_rate": 9.813982577090572e-07, "loss": 0.0962611734867096, "step": 6120 }, { "epoch": 1.5890446521287642, "grad_norm": 2.0107719898223877, "learning_rate": 9.808136202624657e-07, "loss": 0.11408552527427673, "step": 6121 }, { "epoch": 1.5893042575285565, "grad_norm": 2.3409430980682373, "learning_rate": 9.802292955675163e-07, "loss": 0.0745001882314682, "step": 6122 }, { "epoch": 1.5895638629283488, "grad_norm": 2.204756736755371, "learning_rate": 9.796452837275055e-07, "loss": 0.0345219150185585, "step": 6123 }, { "epoch": 1.5898234683281411, "grad_norm": 1.1766369342803955, "learning_rate": 9.790615848456725e-07, "loss": 0.09558561444282532, "step": 6124 }, { "epoch": 1.5900830737279334, "grad_norm": 1.521924376487732, "learning_rate": 9.784781990252027e-07, "loss": 0.1152442991733551, "step": 6125 }, { "epoch": 1.5903426791277258, "grad_norm": 0.9648968577384949, "learning_rate": 9.778951263692247e-07, "loss": 0.10260136425495148, "step": 6126 }, { "epoch": 1.590602284527518, "grad_norm": 1.4788743257522583, "learning_rate": 9.773123669808118e-07, "loss": 0.08317574858665466, "step": 6127 }, { "epoch": 1.5908618899273104, "grad_norm": 1.2478387355804443, "learning_rate": 9.767299209629841e-07, "loss": 0.10234574973583221, "step": 6128 }, { "epoch": 1.5911214953271027, "grad_norm": 3.3503925800323486, "learning_rate": 9.761477884187038e-07, "loss": 0.047283247113227844, "step": 6129 }, { "epoch": 1.591381100726895, "grad_norm": 1.6113765239715576, "learning_rate": 9.755659694508781e-07, "loss": 0.07288974523544312, "step": 6130 }, { "epoch": 1.5916407061266873, "grad_norm": 2.1356170177459717, "learning_rate": 9.749844641623605e-07, "loss": 0.07674858719110489, "step": 6131 }, { "epoch": 1.5919003115264796, "grad_norm": 0.9727609753608704, "learning_rate": 9.744032726559465e-07, "loss": 0.09146337956190109, "step": 6132 }, { "epoch": 1.592159916926272, "grad_norm": 1.6354570388793945, "learning_rate": 9.738223950343789e-07, "loss": 0.09498828649520874, "step": 6133 }, { "epoch": 1.5924195223260644, "grad_norm": 1.7421940565109253, "learning_rate": 9.732418314003422e-07, "loss": 0.08544714748859406, "step": 6134 }, { "epoch": 1.5926791277258567, "grad_norm": 2.0846450328826904, "learning_rate": 9.726615818564675e-07, "loss": 0.09836870431900024, "step": 6135 }, { "epoch": 1.592938733125649, "grad_norm": 2.2379415035247803, "learning_rate": 9.720816465053296e-07, "loss": 0.040433384478092194, "step": 6136 }, { "epoch": 1.5931983385254413, "grad_norm": 1.5306917428970337, "learning_rate": 9.715020254494475e-07, "loss": 0.059246741235256195, "step": 6137 }, { "epoch": 1.5934579439252337, "grad_norm": 1.4385390281677246, "learning_rate": 9.709227187912852e-07, "loss": 0.10741458833217621, "step": 6138 }, { "epoch": 1.593717549325026, "grad_norm": 1.4851657152175903, "learning_rate": 9.703437266332506e-07, "loss": 0.09109412133693695, "step": 6139 }, { "epoch": 1.5939771547248183, "grad_norm": 1.1681221723556519, "learning_rate": 9.697650490776968e-07, "loss": 0.09482919424772263, "step": 6140 }, { "epoch": 1.5942367601246106, "grad_norm": 0.9456115961074829, "learning_rate": 9.691866862269197e-07, "loss": 0.08220840990543365, "step": 6141 }, { "epoch": 1.5944963655244029, "grad_norm": 1.5994988679885864, "learning_rate": 9.686086381831617e-07, "loss": 0.037913113832473755, "step": 6142 }, { "epoch": 1.5947559709241952, "grad_norm": 2.202545166015625, "learning_rate": 9.680309050486075e-07, "loss": 0.07347086071968079, "step": 6143 }, { "epoch": 1.5950155763239875, "grad_norm": 2.0897631645202637, "learning_rate": 9.674534869253887e-07, "loss": 0.03712991625070572, "step": 6144 }, { "epoch": 1.5952751817237798, "grad_norm": 2.5197739601135254, "learning_rate": 9.668763839155776e-07, "loss": 0.052152086049318314, "step": 6145 }, { "epoch": 1.5955347871235723, "grad_norm": 1.5605113506317139, "learning_rate": 9.662995961211935e-07, "loss": 0.08922526240348816, "step": 6146 }, { "epoch": 1.5957943925233646, "grad_norm": 2.497342348098755, "learning_rate": 9.657231236442002e-07, "loss": 0.03218647092580795, "step": 6147 }, { "epoch": 1.596053997923157, "grad_norm": 1.6547503471374512, "learning_rate": 9.65146966586504e-07, "loss": 0.09476131200790405, "step": 6148 }, { "epoch": 1.5963136033229492, "grad_norm": 1.8695465326309204, "learning_rate": 9.645711250499565e-07, "loss": 0.09134788066148758, "step": 6149 }, { "epoch": 1.5965732087227416, "grad_norm": 1.3578015565872192, "learning_rate": 9.639955991363526e-07, "loss": 0.1024589091539383, "step": 6150 }, { "epoch": 1.5968328141225339, "grad_norm": 1.1559131145477295, "learning_rate": 9.634203889474324e-07, "loss": 0.08944524824619293, "step": 6151 }, { "epoch": 1.5970924195223262, "grad_norm": 0.9861889481544495, "learning_rate": 9.628454945848815e-07, "loss": 0.10267975926399231, "step": 6152 }, { "epoch": 1.5973520249221185, "grad_norm": 2.0185015201568604, "learning_rate": 9.622709161503257e-07, "loss": 0.03393243998289108, "step": 6153 }, { "epoch": 1.5976116303219108, "grad_norm": 2.6210618019104004, "learning_rate": 9.616966537453389e-07, "loss": 0.040963008999824524, "step": 6154 }, { "epoch": 1.597871235721703, "grad_norm": 1.722512125968933, "learning_rate": 9.611227074714364e-07, "loss": 0.09902866184711456, "step": 6155 }, { "epoch": 1.5981308411214954, "grad_norm": 1.1540443897247314, "learning_rate": 9.6054907743008e-07, "loss": 0.09427473694086075, "step": 6156 }, { "epoch": 1.5983904465212877, "grad_norm": 1.440485954284668, "learning_rate": 9.599757637226739e-07, "loss": 0.04939635097980499, "step": 6157 }, { "epoch": 1.59865005192108, "grad_norm": 1.3791521787643433, "learning_rate": 9.594027664505656e-07, "loss": 0.09998196363449097, "step": 6158 }, { "epoch": 1.5989096573208723, "grad_norm": 1.5336077213287354, "learning_rate": 9.588300857150498e-07, "loss": 0.03678065910935402, "step": 6159 }, { "epoch": 1.5991692627206646, "grad_norm": 1.9710125923156738, "learning_rate": 9.582577216173626e-07, "loss": 0.07958707213401794, "step": 6160 }, { "epoch": 1.599428868120457, "grad_norm": 2.2896652221679688, "learning_rate": 9.576856742586837e-07, "loss": 0.08143539726734161, "step": 6161 }, { "epoch": 1.5996884735202492, "grad_norm": 2.4498918056488037, "learning_rate": 9.5711394374014e-07, "loss": 0.050532661378383636, "step": 6162 }, { "epoch": 1.5999480789200415, "grad_norm": 1.4272384643554688, "learning_rate": 9.565425301627987e-07, "loss": 0.09276328235864639, "step": 6163 }, { "epoch": 1.6002076843198338, "grad_norm": 0.9791120290756226, "learning_rate": 9.55971433627674e-07, "loss": 0.10542580485343933, "step": 6164 }, { "epoch": 1.6004672897196262, "grad_norm": 1.335355281829834, "learning_rate": 9.554006542357219e-07, "loss": 0.07980373501777649, "step": 6165 }, { "epoch": 1.6007268951194185, "grad_norm": 2.269930839538574, "learning_rate": 9.548301920878428e-07, "loss": 0.06491931527853012, "step": 6166 }, { "epoch": 1.6009865005192108, "grad_norm": 2.107928514480591, "learning_rate": 9.542600472848825e-07, "loss": 0.02710852026939392, "step": 6167 }, { "epoch": 1.601246105919003, "grad_norm": 0.9283004999160767, "learning_rate": 9.536902199276284e-07, "loss": 0.1079859584569931, "step": 6168 }, { "epoch": 1.6015057113187954, "grad_norm": 1.9947947263717651, "learning_rate": 9.531207101168132e-07, "loss": 0.08576375246047974, "step": 6169 }, { "epoch": 1.6017653167185877, "grad_norm": 1.1300941705703735, "learning_rate": 9.525515179531139e-07, "loss": 0.08865601569414139, "step": 6170 }, { "epoch": 1.60202492211838, "grad_norm": 1.707851767539978, "learning_rate": 9.519826435371502e-07, "loss": 0.09617302566766739, "step": 6171 }, { "epoch": 1.6022845275181723, "grad_norm": 1.6912431716918945, "learning_rate": 9.514140869694852e-07, "loss": 0.08626917004585266, "step": 6172 }, { "epoch": 1.6025441329179646, "grad_norm": 2.6937367916107178, "learning_rate": 9.508458483506281e-07, "loss": 0.08702857047319412, "step": 6173 }, { "epoch": 1.602803738317757, "grad_norm": 1.4566317796707153, "learning_rate": 9.502779277810292e-07, "loss": 0.07552346587181091, "step": 6174 }, { "epoch": 1.6030633437175492, "grad_norm": 2.1049418449401855, "learning_rate": 9.497103253610857e-07, "loss": 0.02314796671271324, "step": 6175 }, { "epoch": 1.6033229491173415, "grad_norm": 3.2509877681732178, "learning_rate": 9.491430411911343e-07, "loss": 0.08062507212162018, "step": 6176 }, { "epoch": 1.6035825545171338, "grad_norm": 1.4938987493515015, "learning_rate": 9.485760753714592e-07, "loss": 0.07362337410449982, "step": 6177 }, { "epoch": 1.6038421599169261, "grad_norm": 1.022855520248413, "learning_rate": 9.480094280022873e-07, "loss": 0.09277419745922089, "step": 6178 }, { "epoch": 1.6041017653167184, "grad_norm": 2.129868507385254, "learning_rate": 9.474430991837886e-07, "loss": 0.07223029434680939, "step": 6179 }, { "epoch": 1.6043613707165107, "grad_norm": 1.7997244596481323, "learning_rate": 9.468770890160767e-07, "loss": 0.03698170557618141, "step": 6180 }, { "epoch": 1.604620976116303, "grad_norm": 1.7108246088027954, "learning_rate": 9.463113975992086e-07, "loss": 0.09431932866573334, "step": 6181 }, { "epoch": 1.6048805815160956, "grad_norm": 2.3159632682800293, "learning_rate": 9.457460250331868e-07, "loss": 0.06641227751970291, "step": 6182 }, { "epoch": 1.605140186915888, "grad_norm": 1.239151954650879, "learning_rate": 9.451809714179569e-07, "loss": 0.09501616656780243, "step": 6183 }, { "epoch": 1.6053997923156802, "grad_norm": 1.6093443632125854, "learning_rate": 9.446162368534051e-07, "loss": 0.08935916423797607, "step": 6184 }, { "epoch": 1.6056593977154725, "grad_norm": 1.1986711025238037, "learning_rate": 9.440518214393657e-07, "loss": 0.09575404971837997, "step": 6185 }, { "epoch": 1.6059190031152648, "grad_norm": 1.8479880094528198, "learning_rate": 9.434877252756125e-07, "loss": 0.06010580062866211, "step": 6186 }, { "epoch": 1.6061786085150571, "grad_norm": 1.891932487487793, "learning_rate": 9.429239484618669e-07, "loss": 0.06200970709323883, "step": 6187 }, { "epoch": 1.6064382139148494, "grad_norm": 1.7085798978805542, "learning_rate": 9.423604910977904e-07, "loss": 0.08367212116718292, "step": 6188 }, { "epoch": 1.6066978193146417, "grad_norm": 2.621931314468384, "learning_rate": 9.41797353282989e-07, "loss": 0.08068011701107025, "step": 6189 }, { "epoch": 1.606957424714434, "grad_norm": 2.4050872325897217, "learning_rate": 9.412345351170138e-07, "loss": 0.10644640773534775, "step": 6190 }, { "epoch": 1.6072170301142263, "grad_norm": 2.2514307498931885, "learning_rate": 9.406720366993575e-07, "loss": 0.023459143936634064, "step": 6191 }, { "epoch": 1.6074766355140186, "grad_norm": 1.623583436012268, "learning_rate": 9.401098581294563e-07, "loss": 0.091646708548069, "step": 6192 }, { "epoch": 1.607736240913811, "grad_norm": 1.7606457471847534, "learning_rate": 9.395479995066919e-07, "loss": 0.09417660534381866, "step": 6193 }, { "epoch": 1.6079958463136035, "grad_norm": 0.7192879319190979, "learning_rate": 9.38986460930387e-07, "loss": 0.025028076022863388, "step": 6194 }, { "epoch": 1.6082554517133958, "grad_norm": 0.9866868257522583, "learning_rate": 9.384252424998094e-07, "loss": 0.10373125970363617, "step": 6195 }, { "epoch": 1.608515057113188, "grad_norm": 1.0723165273666382, "learning_rate": 9.378643443141694e-07, "loss": 0.09848420321941376, "step": 6196 }, { "epoch": 1.6087746625129804, "grad_norm": 1.9726558923721313, "learning_rate": 9.373037664726204e-07, "loss": 0.091504767537117, "step": 6197 }, { "epoch": 1.6090342679127727, "grad_norm": 2.6369478702545166, "learning_rate": 9.367435090742608e-07, "loss": 0.07515844702720642, "step": 6198 }, { "epoch": 1.609293873312565, "grad_norm": 3.1358885765075684, "learning_rate": 9.361835722181306e-07, "loss": 0.06812085956335068, "step": 6199 }, { "epoch": 1.6095534787123573, "grad_norm": 2.1620583534240723, "learning_rate": 9.356239560032135e-07, "loss": 0.08783017098903656, "step": 6200 }, { "epoch": 1.6098130841121496, "grad_norm": 2.6502530574798584, "learning_rate": 9.350646605284378e-07, "loss": 0.05481591075658798, "step": 6201 }, { "epoch": 1.610072689511942, "grad_norm": 1.0682423114776611, "learning_rate": 9.345056858926735e-07, "loss": 0.10182024538516998, "step": 6202 }, { "epoch": 1.6103322949117342, "grad_norm": 0.9656057357788086, "learning_rate": 9.339470321947341e-07, "loss": 0.027233321219682693, "step": 6203 }, { "epoch": 1.6105919003115265, "grad_norm": 2.657465696334839, "learning_rate": 9.333886995333782e-07, "loss": 0.046343348920345306, "step": 6204 }, { "epoch": 1.6108515057113189, "grad_norm": 2.6102144718170166, "learning_rate": 9.328306880073046e-07, "loss": 0.0897085964679718, "step": 6205 }, { "epoch": 1.6111111111111112, "grad_norm": 1.8016705513000488, "learning_rate": 9.322729977151587e-07, "loss": 0.07311704754829407, "step": 6206 }, { "epoch": 1.6113707165109035, "grad_norm": 1.0922622680664062, "learning_rate": 9.317156287555253e-07, "loss": 0.08386850357055664, "step": 6207 }, { "epoch": 1.6116303219106958, "grad_norm": 1.9199827909469604, "learning_rate": 9.311585812269356e-07, "loss": 0.047205034643411636, "step": 6208 }, { "epoch": 1.611889927310488, "grad_norm": 1.8865864276885986, "learning_rate": 9.306018552278631e-07, "loss": 0.06728263199329376, "step": 6209 }, { "epoch": 1.6121495327102804, "grad_norm": 1.1475927829742432, "learning_rate": 9.300454508567242e-07, "loss": 0.10193848609924316, "step": 6210 }, { "epoch": 1.6124091381100727, "grad_norm": 2.243162155151367, "learning_rate": 9.294893682118783e-07, "loss": 0.058992430567741394, "step": 6211 }, { "epoch": 1.612668743509865, "grad_norm": 2.4835257530212402, "learning_rate": 9.28933607391627e-07, "loss": 0.0591256320476532, "step": 6212 }, { "epoch": 1.6129283489096573, "grad_norm": 3.6000521183013916, "learning_rate": 9.28378168494217e-07, "loss": 0.05449417978525162, "step": 6213 }, { "epoch": 1.6131879543094496, "grad_norm": 2.550304889678955, "learning_rate": 9.278230516178385e-07, "loss": 0.06814974546432495, "step": 6214 }, { "epoch": 1.613447559709242, "grad_norm": 1.7343870401382446, "learning_rate": 9.272682568606209e-07, "loss": 0.08835625648498535, "step": 6215 }, { "epoch": 1.6137071651090342, "grad_norm": 2.1523725986480713, "learning_rate": 9.267137843206411e-07, "loss": 0.02848856709897518, "step": 6216 }, { "epoch": 1.6139667705088265, "grad_norm": 1.204899787902832, "learning_rate": 9.261596340959161e-07, "loss": 0.10145269334316254, "step": 6217 }, { "epoch": 1.6142263759086188, "grad_norm": 3.308232307434082, "learning_rate": 9.256058062844078e-07, "loss": 0.0975317731499672, "step": 6218 }, { "epoch": 1.6144859813084111, "grad_norm": 1.231540560722351, "learning_rate": 9.250523009840201e-07, "loss": 0.08596450090408325, "step": 6219 }, { "epoch": 1.6147455867082035, "grad_norm": 2.831268072128296, "learning_rate": 9.244991182925987e-07, "loss": 0.04560595750808716, "step": 6220 }, { "epoch": 1.6150051921079958, "grad_norm": 1.447739601135254, "learning_rate": 9.239462583079359e-07, "loss": 0.07613097131252289, "step": 6221 }, { "epoch": 1.615264797507788, "grad_norm": 1.6906819343566895, "learning_rate": 9.233937211277633e-07, "loss": 0.08476895838975906, "step": 6222 }, { "epoch": 1.6155244029075804, "grad_norm": 1.0199923515319824, "learning_rate": 9.228415068497565e-07, "loss": 0.11342701315879822, "step": 6223 }, { "epoch": 1.6157840083073727, "grad_norm": 1.1601204872131348, "learning_rate": 9.222896155715352e-07, "loss": 0.08691871166229248, "step": 6224 }, { "epoch": 1.616043613707165, "grad_norm": 2.1980738639831543, "learning_rate": 9.217380473906612e-07, "loss": 0.07012910395860672, "step": 6225 }, { "epoch": 1.6163032191069573, "grad_norm": 3.088587760925293, "learning_rate": 9.211868024046381e-07, "loss": 0.035192638635635376, "step": 6226 }, { "epoch": 1.6165628245067496, "grad_norm": 1.501689076423645, "learning_rate": 9.206358807109147e-07, "loss": 0.09642428159713745, "step": 6227 }, { "epoch": 1.616822429906542, "grad_norm": 2.570058822631836, "learning_rate": 9.200852824068801e-07, "loss": 0.05589129775762558, "step": 6228 }, { "epoch": 1.6170820353063342, "grad_norm": 1.1282228231430054, "learning_rate": 9.195350075898684e-07, "loss": 0.09546087682247162, "step": 6229 }, { "epoch": 1.6173416407061267, "grad_norm": 2.2996973991394043, "learning_rate": 9.189850563571556e-07, "loss": 0.06520772725343704, "step": 6230 }, { "epoch": 1.617601246105919, "grad_norm": 3.278193712234497, "learning_rate": 9.184354288059596e-07, "loss": 0.08071711659431458, "step": 6231 }, { "epoch": 1.6178608515057114, "grad_norm": 2.1731276512145996, "learning_rate": 9.178861250334429e-07, "loss": 0.06112588196992874, "step": 6232 }, { "epoch": 1.6181204569055037, "grad_norm": 1.029514193534851, "learning_rate": 9.173371451367097e-07, "loss": 0.10042339563369751, "step": 6233 }, { "epoch": 1.618380062305296, "grad_norm": 1.4827449321746826, "learning_rate": 9.167884892128059e-07, "loss": 0.09089043736457825, "step": 6234 }, { "epoch": 1.6186396677050883, "grad_norm": 1.6580712795257568, "learning_rate": 9.162401573587229e-07, "loss": 0.09851022064685822, "step": 6235 }, { "epoch": 1.6188992731048806, "grad_norm": 2.3572118282318115, "learning_rate": 9.156921496713922e-07, "loss": 0.08482196927070618, "step": 6236 }, { "epoch": 1.6191588785046729, "grad_norm": 1.475961685180664, "learning_rate": 9.151444662476902e-07, "loss": 0.07247503101825714, "step": 6237 }, { "epoch": 1.6194184839044652, "grad_norm": 1.7859104871749878, "learning_rate": 9.145971071844332e-07, "loss": 0.04497482627630234, "step": 6238 }, { "epoch": 1.6196780893042575, "grad_norm": 2.696775197982788, "learning_rate": 9.140500725783818e-07, "loss": 0.07755742222070694, "step": 6239 }, { "epoch": 1.6199376947040498, "grad_norm": 2.576812505722046, "learning_rate": 9.135033625262412e-07, "loss": 0.07655825465917587, "step": 6240 }, { "epoch": 1.620197300103842, "grad_norm": 3.1878929138183594, "learning_rate": 9.129569771246555e-07, "loss": 0.06077192723751068, "step": 6241 }, { "epoch": 1.6204569055036346, "grad_norm": 0.8095784783363342, "learning_rate": 9.124109164702135e-07, "loss": 0.10641824454069138, "step": 6242 }, { "epoch": 1.620716510903427, "grad_norm": 0.9430158138275146, "learning_rate": 9.118651806594459e-07, "loss": 0.10409274697303772, "step": 6243 }, { "epoch": 1.6209761163032192, "grad_norm": 1.185451626777649, "learning_rate": 9.113197697888265e-07, "loss": 0.09753158688545227, "step": 6244 }, { "epoch": 1.6212357217030116, "grad_norm": 1.1384131908416748, "learning_rate": 9.107746839547729e-07, "loss": 0.09745414555072784, "step": 6245 }, { "epoch": 1.6214953271028039, "grad_norm": 1.6757336854934692, "learning_rate": 9.102299232536413e-07, "loss": 0.08329597860574722, "step": 6246 }, { "epoch": 1.6217549325025962, "grad_norm": 1.353380560874939, "learning_rate": 9.096854877817346e-07, "loss": 0.08761949837207794, "step": 6247 }, { "epoch": 1.6220145379023885, "grad_norm": 1.5324443578720093, "learning_rate": 9.091413776352954e-07, "loss": 0.07421671599149704, "step": 6248 }, { "epoch": 1.6222741433021808, "grad_norm": 2.11991286277771, "learning_rate": 9.085975929105113e-07, "loss": 0.043829526752233505, "step": 6249 }, { "epoch": 1.622533748701973, "grad_norm": 1.6498823165893555, "learning_rate": 9.080541337035102e-07, "loss": 0.0628131553530693, "step": 6250 }, { "epoch": 1.6227933541017654, "grad_norm": 1.551942229270935, "learning_rate": 9.075110001103627e-07, "loss": 0.08027373254299164, "step": 6251 }, { "epoch": 1.6230529595015577, "grad_norm": 3.004621982574463, "learning_rate": 9.069681922270837e-07, "loss": 0.035060618072748184, "step": 6252 }, { "epoch": 1.62331256490135, "grad_norm": 2.246523141860962, "learning_rate": 9.064257101496285e-07, "loss": 0.06887707859277725, "step": 6253 }, { "epoch": 1.6235721703011423, "grad_norm": 1.5695509910583496, "learning_rate": 9.058835539738947e-07, "loss": 0.07901803404092789, "step": 6254 }, { "epoch": 1.6238317757009346, "grad_norm": 1.9931635856628418, "learning_rate": 9.053417237957243e-07, "loss": 0.07455745339393616, "step": 6255 }, { "epoch": 1.624091381100727, "grad_norm": 2.5035088062286377, "learning_rate": 9.048002197109004e-07, "loss": 0.030133482068777084, "step": 6256 }, { "epoch": 1.6243509865005192, "grad_norm": 1.4799833297729492, "learning_rate": 9.042590418151474e-07, "loss": 0.09065840393304825, "step": 6257 }, { "epoch": 1.6246105919003115, "grad_norm": 1.4527637958526611, "learning_rate": 9.037181902041348e-07, "loss": 0.053313545882701874, "step": 6258 }, { "epoch": 1.6248701973001038, "grad_norm": 2.1630771160125732, "learning_rate": 9.031776649734708e-07, "loss": 0.04040665552020073, "step": 6259 }, { "epoch": 1.6251298026998962, "grad_norm": 3.101062297821045, "learning_rate": 9.026374662187102e-07, "loss": 0.03237282112240791, "step": 6260 }, { "epoch": 1.6253894080996885, "grad_norm": 1.6297576427459717, "learning_rate": 9.020975940353463e-07, "loss": 0.06495204567909241, "step": 6261 }, { "epoch": 1.6256490134994808, "grad_norm": 2.6005847454071045, "learning_rate": 9.015580485188158e-07, "loss": 0.10863398760557175, "step": 6262 }, { "epoch": 1.625908618899273, "grad_norm": 1.9756828546524048, "learning_rate": 9.010188297644994e-07, "loss": 0.05312586575746536, "step": 6263 }, { "epoch": 1.6261682242990654, "grad_norm": 2.083435535430908, "learning_rate": 9.004799378677176e-07, "loss": 0.0604742094874382, "step": 6264 }, { "epoch": 1.6264278296988577, "grad_norm": 1.7363635301589966, "learning_rate": 8.999413729237339e-07, "loss": 0.076970174908638, "step": 6265 }, { "epoch": 1.62668743509865, "grad_norm": 1.685844898223877, "learning_rate": 8.994031350277552e-07, "loss": 0.10055333375930786, "step": 6266 }, { "epoch": 1.6269470404984423, "grad_norm": 1.3439046144485474, "learning_rate": 8.988652242749288e-07, "loss": 0.10198850929737091, "step": 6267 }, { "epoch": 1.6272066458982346, "grad_norm": 1.8562952280044556, "learning_rate": 8.983276407603464e-07, "loss": 0.043179966509342194, "step": 6268 }, { "epoch": 1.627466251298027, "grad_norm": 1.1559672355651855, "learning_rate": 8.977903845790384e-07, "loss": 0.09000782668590546, "step": 6269 }, { "epoch": 1.6277258566978192, "grad_norm": 2.030702590942383, "learning_rate": 8.972534558259803e-07, "loss": 0.07223950326442719, "step": 6270 }, { "epoch": 1.6279854620976115, "grad_norm": 1.4603698253631592, "learning_rate": 8.967168545960897e-07, "loss": 0.07066722214221954, "step": 6271 }, { "epoch": 1.6282450674974038, "grad_norm": 2.3633594512939453, "learning_rate": 8.961805809842244e-07, "loss": 0.06615129858255386, "step": 6272 }, { "epoch": 1.6285046728971961, "grad_norm": 2.5636799335479736, "learning_rate": 8.956446350851855e-07, "loss": 0.05425149202346802, "step": 6273 }, { "epoch": 1.6287642782969884, "grad_norm": 1.7185187339782715, "learning_rate": 8.951090169937158e-07, "loss": 0.0681116059422493, "step": 6274 }, { "epoch": 1.6290238836967808, "grad_norm": 2.316102981567383, "learning_rate": 8.945737268045004e-07, "loss": 0.039450280368328094, "step": 6275 }, { "epoch": 1.629283489096573, "grad_norm": 2.095618486404419, "learning_rate": 8.940387646121673e-07, "loss": 0.05522090569138527, "step": 6276 }, { "epoch": 1.6295430944963654, "grad_norm": 1.3464866876602173, "learning_rate": 8.935041305112838e-07, "loss": 0.0945620983839035, "step": 6277 }, { "epoch": 1.629802699896158, "grad_norm": 1.5699437856674194, "learning_rate": 8.929698245963625e-07, "loss": 0.08847863972187042, "step": 6278 }, { "epoch": 1.6300623052959502, "grad_norm": 2.051771402359009, "learning_rate": 8.92435846961856e-07, "loss": 0.05965118110179901, "step": 6279 }, { "epoch": 1.6303219106957425, "grad_norm": 2.1009607315063477, "learning_rate": 8.919021977021584e-07, "loss": 0.08747975528240204, "step": 6280 }, { "epoch": 1.6305815160955348, "grad_norm": 1.2324861288070679, "learning_rate": 8.91368876911608e-07, "loss": 0.09589973092079163, "step": 6281 }, { "epoch": 1.6308411214953271, "grad_norm": 2.7869832515716553, "learning_rate": 8.908358846844828e-07, "loss": 0.04545474052429199, "step": 6282 }, { "epoch": 1.6311007268951194, "grad_norm": 1.606809377670288, "learning_rate": 8.903032211150042e-07, "loss": 0.07630696892738342, "step": 6283 }, { "epoch": 1.6313603322949117, "grad_norm": 1.2121533155441284, "learning_rate": 8.89770886297335e-07, "loss": 0.03071986883878708, "step": 6284 }, { "epoch": 1.631619937694704, "grad_norm": 2.4116129875183105, "learning_rate": 8.892388803255788e-07, "loss": 0.0361812487244606, "step": 6285 }, { "epoch": 1.6318795430944963, "grad_norm": 1.7403823137283325, "learning_rate": 8.887072032937833e-07, "loss": 0.08316375315189362, "step": 6286 }, { "epoch": 1.6321391484942886, "grad_norm": 1.5812841653823853, "learning_rate": 8.881758552959366e-07, "loss": 0.08881276845932007, "step": 6287 }, { "epoch": 1.632398753894081, "grad_norm": 1.992758870124817, "learning_rate": 8.876448364259677e-07, "loss": 0.03440956398844719, "step": 6288 }, { "epoch": 1.6326583592938733, "grad_norm": 1.2539353370666504, "learning_rate": 8.871141467777501e-07, "loss": 0.07440802454948425, "step": 6289 }, { "epoch": 1.6329179646936658, "grad_norm": 1.265480399131775, "learning_rate": 8.865837864450962e-07, "loss": 0.08865395188331604, "step": 6290 }, { "epoch": 1.633177570093458, "grad_norm": 1.1828174591064453, "learning_rate": 8.86053755521763e-07, "loss": 0.09833326935768127, "step": 6291 }, { "epoch": 1.6334371754932504, "grad_norm": 3.0429728031158447, "learning_rate": 8.855240541014471e-07, "loss": 0.050162408500909805, "step": 6292 }, { "epoch": 1.6336967808930427, "grad_norm": 2.3702785968780518, "learning_rate": 8.849946822777871e-07, "loss": 0.09758677333593369, "step": 6293 }, { "epoch": 1.633956386292835, "grad_norm": 2.77288818359375, "learning_rate": 8.844656401443646e-07, "loss": 0.04942808300256729, "step": 6294 }, { "epoch": 1.6342159916926273, "grad_norm": 2.9595258235931396, "learning_rate": 8.839369277947021e-07, "loss": 0.05698695406317711, "step": 6295 }, { "epoch": 1.6344755970924196, "grad_norm": 1.4203801155090332, "learning_rate": 8.83408545322263e-07, "loss": 0.06444445252418518, "step": 6296 }, { "epoch": 1.634735202492212, "grad_norm": 2.549541473388672, "learning_rate": 8.828804928204543e-07, "loss": 0.07179844379425049, "step": 6297 }, { "epoch": 1.6349948078920042, "grad_norm": 1.4770290851593018, "learning_rate": 8.823527703826227e-07, "loss": 0.07806380093097687, "step": 6298 }, { "epoch": 1.6352544132917965, "grad_norm": 2.3512961864471436, "learning_rate": 8.818253781020586e-07, "loss": 0.06400511413812637, "step": 6299 }, { "epoch": 1.6355140186915889, "grad_norm": 1.0489492416381836, "learning_rate": 8.812983160719912e-07, "loss": 0.09269094467163086, "step": 6300 }, { "epoch": 1.6357736240913812, "grad_norm": 2.1737372875213623, "learning_rate": 8.807715843855941e-07, "loss": 0.049977242946624756, "step": 6301 }, { "epoch": 1.6360332294911735, "grad_norm": 1.364508867263794, "learning_rate": 8.802451831359815e-07, "loss": 0.09535686671733856, "step": 6302 }, { "epoch": 1.6362928348909658, "grad_norm": 1.275258183479309, "learning_rate": 8.797191124162091e-07, "loss": 0.08067290484905243, "step": 6303 }, { "epoch": 1.636552440290758, "grad_norm": 1.9472476243972778, "learning_rate": 8.791933723192731e-07, "loss": 0.08750363439321518, "step": 6304 }, { "epoch": 1.6368120456905504, "grad_norm": 3.472194194793701, "learning_rate": 8.786679629381136e-07, "loss": 0.08926378190517426, "step": 6305 }, { "epoch": 1.6370716510903427, "grad_norm": 1.5445444583892822, "learning_rate": 8.781428843656106e-07, "loss": 0.09301196038722992, "step": 6306 }, { "epoch": 1.637331256490135, "grad_norm": 3.761751413345337, "learning_rate": 8.776181366945858e-07, "loss": 0.0738721638917923, "step": 6307 }, { "epoch": 1.6375908618899273, "grad_norm": 1.9794292449951172, "learning_rate": 8.77093720017802e-07, "loss": 0.06833053380250931, "step": 6308 }, { "epoch": 1.6378504672897196, "grad_norm": 1.9078712463378906, "learning_rate": 8.765696344279645e-07, "loss": 0.03704241290688515, "step": 6309 }, { "epoch": 1.638110072689512, "grad_norm": 1.1082496643066406, "learning_rate": 8.76045880017721e-07, "loss": 0.09502490609884262, "step": 6310 }, { "epoch": 1.6383696780893042, "grad_norm": 2.1378400325775146, "learning_rate": 8.755224568796569e-07, "loss": 0.09288675338029861, "step": 6311 }, { "epoch": 1.6386292834890965, "grad_norm": 1.9254324436187744, "learning_rate": 8.749993651063033e-07, "loss": 0.05408333241939545, "step": 6312 }, { "epoch": 1.6388888888888888, "grad_norm": 2.2559492588043213, "learning_rate": 8.744766047901291e-07, "loss": 0.07009482383728027, "step": 6313 }, { "epoch": 1.6391484942886811, "grad_norm": 2.6943106651306152, "learning_rate": 8.739541760235484e-07, "loss": 0.0666966438293457, "step": 6314 }, { "epoch": 1.6394080996884735, "grad_norm": 1.1270484924316406, "learning_rate": 8.734320788989133e-07, "loss": 0.08144529908895493, "step": 6315 }, { "epoch": 1.6396677050882658, "grad_norm": 1.7099756002426147, "learning_rate": 8.729103135085184e-07, "loss": 0.060740966349840164, "step": 6316 }, { "epoch": 1.639927310488058, "grad_norm": 3.174616575241089, "learning_rate": 8.723888799446012e-07, "loss": 0.04596267640590668, "step": 6317 }, { "epoch": 1.6401869158878504, "grad_norm": 1.9286597967147827, "learning_rate": 8.718677782993383e-07, "loss": 0.08550409227609634, "step": 6318 }, { "epoch": 1.6404465212876427, "grad_norm": 1.0966986417770386, "learning_rate": 8.713470086648482e-07, "loss": 0.10720959305763245, "step": 6319 }, { "epoch": 1.640706126687435, "grad_norm": 3.538055181503296, "learning_rate": 8.70826571133192e-07, "loss": 0.04058544337749481, "step": 6320 }, { "epoch": 1.6409657320872273, "grad_norm": 2.7400882244110107, "learning_rate": 8.703064657963703e-07, "loss": 0.06811760365962982, "step": 6321 }, { "epoch": 1.6412253374870196, "grad_norm": 1.3368967771530151, "learning_rate": 8.697866927463264e-07, "loss": 0.028489677235484123, "step": 6322 }, { "epoch": 1.641484942886812, "grad_norm": 1.7954356670379639, "learning_rate": 8.692672520749443e-07, "loss": 0.06109418720006943, "step": 6323 }, { "epoch": 1.6417445482866042, "grad_norm": 3.00793194770813, "learning_rate": 8.687481438740481e-07, "loss": 0.036792054772377014, "step": 6324 }, { "epoch": 1.6420041536863965, "grad_norm": 1.586061954498291, "learning_rate": 8.682293682354059e-07, "loss": 0.03816767409443855, "step": 6325 }, { "epoch": 1.642263759086189, "grad_norm": 2.9340980052948, "learning_rate": 8.677109252507246e-07, "loss": 0.05980562046170235, "step": 6326 }, { "epoch": 1.6425233644859814, "grad_norm": 1.1787118911743164, "learning_rate": 8.671928150116523e-07, "loss": 0.09629453718662262, "step": 6327 }, { "epoch": 1.6427829698857737, "grad_norm": 1.4815285205841064, "learning_rate": 8.666750376097805e-07, "loss": 0.11215105652809143, "step": 6328 }, { "epoch": 1.643042575285566, "grad_norm": 1.3358441591262817, "learning_rate": 8.661575931366387e-07, "loss": 0.0875689834356308, "step": 6329 }, { "epoch": 1.6433021806853583, "grad_norm": 2.8204028606414795, "learning_rate": 8.656404816837011e-07, "loss": 0.10395972430706024, "step": 6330 }, { "epoch": 1.6435617860851506, "grad_norm": 3.0337417125701904, "learning_rate": 8.651237033423801e-07, "loss": 0.07584653794765472, "step": 6331 }, { "epoch": 1.6438213914849429, "grad_norm": 1.1554536819458008, "learning_rate": 8.646072582040296e-07, "loss": 0.09536339342594147, "step": 6332 }, { "epoch": 1.6440809968847352, "grad_norm": 2.805889844894409, "learning_rate": 8.640911463599475e-07, "loss": 0.08742748200893402, "step": 6333 }, { "epoch": 1.6443406022845275, "grad_norm": 1.5252563953399658, "learning_rate": 8.635753679013681e-07, "loss": 0.08302798867225647, "step": 6334 }, { "epoch": 1.6446002076843198, "grad_norm": 3.209838628768921, "learning_rate": 8.6305992291947e-07, "loss": 0.07461308687925339, "step": 6335 }, { "epoch": 1.644859813084112, "grad_norm": 3.3564770221710205, "learning_rate": 8.62544811505373e-07, "loss": 0.07664334028959274, "step": 6336 }, { "epoch": 1.6451194184839044, "grad_norm": 1.1346714496612549, "learning_rate": 8.620300337501361e-07, "loss": 0.09398753941059113, "step": 6337 }, { "epoch": 1.645379023883697, "grad_norm": 1.502850890159607, "learning_rate": 8.615155897447606e-07, "loss": 0.058411695063114166, "step": 6338 }, { "epoch": 1.6456386292834893, "grad_norm": 3.352208137512207, "learning_rate": 8.610014795801879e-07, "loss": 0.06841129064559937, "step": 6339 }, { "epoch": 1.6458982346832816, "grad_norm": 3.494746685028076, "learning_rate": 8.604877033473011e-07, "loss": 0.051090456545352936, "step": 6340 }, { "epoch": 1.6461578400830739, "grad_norm": 1.2651066780090332, "learning_rate": 8.599742611369255e-07, "loss": 0.07530159503221512, "step": 6341 }, { "epoch": 1.6464174454828662, "grad_norm": 2.6406631469726562, "learning_rate": 8.594611530398235e-07, "loss": 0.05468865483999252, "step": 6342 }, { "epoch": 1.6466770508826585, "grad_norm": 1.9402426481246948, "learning_rate": 8.589483791467024e-07, "loss": 0.07385271787643433, "step": 6343 }, { "epoch": 1.6469366562824508, "grad_norm": 1.0502851009368896, "learning_rate": 8.584359395482081e-07, "loss": 0.10586459189653397, "step": 6344 }, { "epoch": 1.647196261682243, "grad_norm": 2.01999568939209, "learning_rate": 8.579238343349295e-07, "loss": 0.06101779639720917, "step": 6345 }, { "epoch": 1.6474558670820354, "grad_norm": 1.1058485507965088, "learning_rate": 8.574120635973937e-07, "loss": 0.09808668494224548, "step": 6346 }, { "epoch": 1.6477154724818277, "grad_norm": 1.8303128480911255, "learning_rate": 8.569006274260702e-07, "loss": 0.08039042353630066, "step": 6347 }, { "epoch": 1.64797507788162, "grad_norm": 1.2248384952545166, "learning_rate": 8.563895259113703e-07, "loss": 0.09465692937374115, "step": 6348 }, { "epoch": 1.6482346832814123, "grad_norm": 2.5535027980804443, "learning_rate": 8.558787591436437e-07, "loss": 0.10673338174819946, "step": 6349 }, { "epoch": 1.6484942886812046, "grad_norm": 1.1417303085327148, "learning_rate": 8.553683272131827e-07, "loss": 0.029671479016542435, "step": 6350 }, { "epoch": 1.648753894080997, "grad_norm": 1.4945361614227295, "learning_rate": 8.548582302102207e-07, "loss": 0.09303434193134308, "step": 6351 }, { "epoch": 1.6490134994807892, "grad_norm": 1.4462026357650757, "learning_rate": 8.543484682249301e-07, "loss": 0.07625877112150192, "step": 6352 }, { "epoch": 1.6492731048805815, "grad_norm": 1.4343844652175903, "learning_rate": 8.53839041347426e-07, "loss": 0.09154519438743591, "step": 6353 }, { "epoch": 1.6495327102803738, "grad_norm": 3.1043853759765625, "learning_rate": 8.53329949667763e-07, "loss": 0.08775098621845245, "step": 6354 }, { "epoch": 1.6497923156801662, "grad_norm": 1.7129087448120117, "learning_rate": 8.528211932759367e-07, "loss": 0.07594691216945648, "step": 6355 }, { "epoch": 1.6500519210799585, "grad_norm": 1.5262432098388672, "learning_rate": 8.52312772261884e-07, "loss": 0.07567735761404037, "step": 6356 }, { "epoch": 1.6503115264797508, "grad_norm": 1.69338059425354, "learning_rate": 8.518046867154823e-07, "loss": 0.053664542734622955, "step": 6357 }, { "epoch": 1.650571131879543, "grad_norm": 1.4259474277496338, "learning_rate": 8.512969367265483e-07, "loss": 0.10318340361118317, "step": 6358 }, { "epoch": 1.6508307372793354, "grad_norm": 1.6148463487625122, "learning_rate": 8.507895223848417e-07, "loss": 0.073415607213974, "step": 6359 }, { "epoch": 1.6510903426791277, "grad_norm": 1.752038598060608, "learning_rate": 8.502824437800615e-07, "loss": 0.10563290864229202, "step": 6360 }, { "epoch": 1.65134994807892, "grad_norm": 2.7622737884521484, "learning_rate": 8.49775701001847e-07, "loss": 0.06920492649078369, "step": 6361 }, { "epoch": 1.6516095534787123, "grad_norm": 1.456576943397522, "learning_rate": 8.492692941397798e-07, "loss": 0.09308740496635437, "step": 6362 }, { "epoch": 1.6518691588785046, "grad_norm": 1.27218496799469, "learning_rate": 8.487632232833798e-07, "loss": 0.10165951400995255, "step": 6363 }, { "epoch": 1.652128764278297, "grad_norm": 1.2515913248062134, "learning_rate": 8.482574885221106e-07, "loss": 0.09205329418182373, "step": 6364 }, { "epoch": 1.6523883696780892, "grad_norm": 1.4410823583602905, "learning_rate": 8.477520899453719e-07, "loss": 0.0708448737859726, "step": 6365 }, { "epoch": 1.6526479750778815, "grad_norm": 2.9729433059692383, "learning_rate": 8.472470276425082e-07, "loss": 0.03838695213198662, "step": 6366 }, { "epoch": 1.6529075804776738, "grad_norm": 1.4622578620910645, "learning_rate": 8.467423017028035e-07, "loss": 0.07177866250276566, "step": 6367 }, { "epoch": 1.6531671858774661, "grad_norm": 1.193825602531433, "learning_rate": 8.462379122154807e-07, "loss": 0.08920475840568542, "step": 6368 }, { "epoch": 1.6534267912772584, "grad_norm": 3.0043554306030273, "learning_rate": 8.457338592697046e-07, "loss": 0.06238364800810814, "step": 6369 }, { "epoch": 1.6536863966770508, "grad_norm": 6.20644998550415, "learning_rate": 8.452301429545798e-07, "loss": 0.06797962635755539, "step": 6370 }, { "epoch": 1.653946002076843, "grad_norm": 2.700916051864624, "learning_rate": 8.447267633591525e-07, "loss": 0.0811849907040596, "step": 6371 }, { "epoch": 1.6542056074766354, "grad_norm": 1.7249960899353027, "learning_rate": 8.442237205724094e-07, "loss": 0.022762708365917206, "step": 6372 }, { "epoch": 1.6544652128764277, "grad_norm": 2.0865397453308105, "learning_rate": 8.437210146832749e-07, "loss": 0.09035594761371613, "step": 6373 }, { "epoch": 1.6547248182762202, "grad_norm": 1.826719045639038, "learning_rate": 8.432186457806175e-07, "loss": 0.07390870153903961, "step": 6374 }, { "epoch": 1.6549844236760125, "grad_norm": 1.5272135734558105, "learning_rate": 8.427166139532436e-07, "loss": 0.07567788660526276, "step": 6375 }, { "epoch": 1.6552440290758048, "grad_norm": 2.761233329772949, "learning_rate": 8.422149192899018e-07, "loss": 0.07027827203273773, "step": 6376 }, { "epoch": 1.6555036344755971, "grad_norm": 2.269932746887207, "learning_rate": 8.417135618792799e-07, "loss": 0.05132441222667694, "step": 6377 }, { "epoch": 1.6557632398753894, "grad_norm": 1.6077008247375488, "learning_rate": 8.412125418100055e-07, "loss": 0.06967903673648834, "step": 6378 }, { "epoch": 1.6560228452751817, "grad_norm": 1.9305940866470337, "learning_rate": 8.407118591706493e-07, "loss": 0.08818428218364716, "step": 6379 }, { "epoch": 1.656282450674974, "grad_norm": 1.6508095264434814, "learning_rate": 8.402115140497189e-07, "loss": 0.08691748231649399, "step": 6380 }, { "epoch": 1.6565420560747663, "grad_norm": 1.0917210578918457, "learning_rate": 8.397115065356641e-07, "loss": 0.10024824738502502, "step": 6381 }, { "epoch": 1.6568016614745587, "grad_norm": 1.567116618156433, "learning_rate": 8.392118367168757e-07, "loss": 0.08532217890024185, "step": 6382 }, { "epoch": 1.657061266874351, "grad_norm": 1.3928910493850708, "learning_rate": 8.38712504681683e-07, "loss": 0.08857567608356476, "step": 6383 }, { "epoch": 1.6573208722741433, "grad_norm": 1.284807562828064, "learning_rate": 8.38213510518357e-07, "loss": 0.09090577811002731, "step": 6384 }, { "epoch": 1.6575804776739356, "grad_norm": 1.8246285915374756, "learning_rate": 8.377148543151084e-07, "loss": 0.07452039420604706, "step": 6385 }, { "epoch": 1.657840083073728, "grad_norm": 1.7136338949203491, "learning_rate": 8.372165361600876e-07, "loss": 0.07516755163669586, "step": 6386 }, { "epoch": 1.6580996884735204, "grad_norm": 2.9743096828460693, "learning_rate": 8.367185561413866e-07, "loss": 0.09491956233978271, "step": 6387 }, { "epoch": 1.6583592938733127, "grad_norm": 2.0259876251220703, "learning_rate": 8.362209143470368e-07, "loss": 0.04054476320743561, "step": 6388 }, { "epoch": 1.658618899273105, "grad_norm": 1.9377498626708984, "learning_rate": 8.357236108650091e-07, "loss": 0.04561326652765274, "step": 6389 }, { "epoch": 1.6588785046728973, "grad_norm": 1.4281237125396729, "learning_rate": 8.352266457832158e-07, "loss": 0.07452711462974548, "step": 6390 }, { "epoch": 1.6591381100726896, "grad_norm": 1.8333243131637573, "learning_rate": 8.347300191895097e-07, "loss": 0.07893837243318558, "step": 6391 }, { "epoch": 1.659397715472482, "grad_norm": 2.289268732070923, "learning_rate": 8.342337311716817e-07, "loss": 0.09135361015796661, "step": 6392 }, { "epoch": 1.6596573208722742, "grad_norm": 1.0633518695831299, "learning_rate": 8.337377818174651e-07, "loss": 0.09402650594711304, "step": 6393 }, { "epoch": 1.6599169262720666, "grad_norm": 1.7859476804733276, "learning_rate": 8.332421712145316e-07, "loss": 0.05186077207326889, "step": 6394 }, { "epoch": 1.6601765316718589, "grad_norm": 1.6300774812698364, "learning_rate": 8.327468994504953e-07, "loss": 0.08197177946567535, "step": 6395 }, { "epoch": 1.6604361370716512, "grad_norm": 2.1780521869659424, "learning_rate": 8.32251966612907e-07, "loss": 0.06771625578403473, "step": 6396 }, { "epoch": 1.6606957424714435, "grad_norm": 1.0970007181167603, "learning_rate": 8.317573727892603e-07, "loss": 0.10763457417488098, "step": 6397 }, { "epoch": 1.6609553478712358, "grad_norm": 1.0741956233978271, "learning_rate": 8.312631180669889e-07, "loss": 0.08441248536109924, "step": 6398 }, { "epoch": 1.661214953271028, "grad_norm": 1.949671983718872, "learning_rate": 8.307692025334648e-07, "loss": 0.07488936185836792, "step": 6399 }, { "epoch": 1.6614745586708204, "grad_norm": 1.0739126205444336, "learning_rate": 8.302756262760009e-07, "loss": 0.1161700040102005, "step": 6400 }, { "epoch": 1.6617341640706127, "grad_norm": 6.5178704261779785, "learning_rate": 8.297823893818502e-07, "loss": 0.048856426030397415, "step": 6401 }, { "epoch": 1.661993769470405, "grad_norm": 1.4959386587142944, "learning_rate": 8.292894919382058e-07, "loss": 0.06460646539926529, "step": 6402 }, { "epoch": 1.6622533748701973, "grad_norm": 1.2255001068115234, "learning_rate": 8.287969340322017e-07, "loss": 0.0968666672706604, "step": 6403 }, { "epoch": 1.6625129802699896, "grad_norm": 1.570447564125061, "learning_rate": 8.283047157509091e-07, "loss": 0.05121217668056488, "step": 6404 }, { "epoch": 1.662772585669782, "grad_norm": 1.4408552646636963, "learning_rate": 8.278128371813423e-07, "loss": 0.07540065050125122, "step": 6405 }, { "epoch": 1.6630321910695742, "grad_norm": 1.1591966152191162, "learning_rate": 8.273212984104531e-07, "loss": 0.10607452690601349, "step": 6406 }, { "epoch": 1.6632917964693665, "grad_norm": 2.0110645294189453, "learning_rate": 8.268300995251353e-07, "loss": 0.07665278762578964, "step": 6407 }, { "epoch": 1.6635514018691588, "grad_norm": 1.5637075901031494, "learning_rate": 8.263392406122212e-07, "loss": 0.07222025096416473, "step": 6408 }, { "epoch": 1.6638110072689511, "grad_norm": 1.6512972116470337, "learning_rate": 8.258487217584827e-07, "loss": 0.08259165287017822, "step": 6409 }, { "epoch": 1.6640706126687435, "grad_norm": 2.9049935340881348, "learning_rate": 8.253585430506333e-07, "loss": 0.04081197828054428, "step": 6410 }, { "epoch": 1.6643302180685358, "grad_norm": 2.10499906539917, "learning_rate": 8.248687045753256e-07, "loss": 0.045775726437568665, "step": 6411 }, { "epoch": 1.664589823468328, "grad_norm": 1.0749475955963135, "learning_rate": 8.243792064191503e-07, "loss": 0.10348130762577057, "step": 6412 }, { "epoch": 1.6648494288681204, "grad_norm": 1.5653272867202759, "learning_rate": 8.238900486686413e-07, "loss": 0.08705338835716248, "step": 6413 }, { "epoch": 1.6651090342679127, "grad_norm": 1.2072583436965942, "learning_rate": 8.234012314102695e-07, "loss": 0.11480334401130676, "step": 6414 }, { "epoch": 1.665368639667705, "grad_norm": 1.341468095779419, "learning_rate": 8.22912754730446e-07, "loss": 0.08605347573757172, "step": 6415 }, { "epoch": 1.6656282450674973, "grad_norm": 1.9859296083450317, "learning_rate": 8.224246187155238e-07, "loss": 0.05605266988277435, "step": 6416 }, { "epoch": 1.6658878504672896, "grad_norm": 1.7032657861709595, "learning_rate": 8.219368234517927e-07, "loss": 0.07370461523532867, "step": 6417 }, { "epoch": 1.666147455867082, "grad_norm": 1.884658932685852, "learning_rate": 8.21449369025485e-07, "loss": 0.08596813678741455, "step": 6418 }, { "epoch": 1.6664070612668742, "grad_norm": 2.7537033557891846, "learning_rate": 8.209622555227709e-07, "loss": 0.06950847059488297, "step": 6419 }, { "epoch": 1.6666666666666665, "grad_norm": 2.05605411529541, "learning_rate": 8.204754830297602e-07, "loss": 0.08922471106052399, "step": 6420 }, { "epoch": 1.6669262720664588, "grad_norm": 2.0802371501922607, "learning_rate": 8.199890516325046e-07, "loss": 0.04039076715707779, "step": 6421 }, { "epoch": 1.6671858774662511, "grad_norm": 1.9502339363098145, "learning_rate": 8.195029614169934e-07, "loss": 0.08615627139806747, "step": 6422 }, { "epoch": 1.6674454828660437, "grad_norm": 1.097766637802124, "learning_rate": 8.190172124691553e-07, "loss": 0.09381146728992462, "step": 6423 }, { "epoch": 1.667705088265836, "grad_norm": 2.5508880615234375, "learning_rate": 8.185318048748611e-07, "loss": 0.0769227147102356, "step": 6424 }, { "epoch": 1.6679646936656283, "grad_norm": 1.2293193340301514, "learning_rate": 8.180467387199184e-07, "loss": 0.11119867861270905, "step": 6425 }, { "epoch": 1.6682242990654206, "grad_norm": 1.6265336275100708, "learning_rate": 8.175620140900774e-07, "loss": 0.07911065965890884, "step": 6426 }, { "epoch": 1.6684839044652129, "grad_norm": 2.791973829269409, "learning_rate": 8.170776310710246e-07, "loss": 0.044023990631103516, "step": 6427 }, { "epoch": 1.6687435098650052, "grad_norm": 2.982069253921509, "learning_rate": 8.165935897483881e-07, "loss": 0.07792897522449493, "step": 6428 }, { "epoch": 1.6690031152647975, "grad_norm": 1.3822418451309204, "learning_rate": 8.161098902077366e-07, "loss": 0.10226854681968689, "step": 6429 }, { "epoch": 1.6692627206645898, "grad_norm": 2.3894946575164795, "learning_rate": 8.156265325345764e-07, "loss": 0.08371947705745697, "step": 6430 }, { "epoch": 1.6695223260643821, "grad_norm": 1.2691015005111694, "learning_rate": 8.151435168143537e-07, "loss": 0.09606769680976868, "step": 6431 }, { "epoch": 1.6697819314641744, "grad_norm": 2.346733808517456, "learning_rate": 8.146608431324548e-07, "loss": 0.043691739439964294, "step": 6432 }, { "epoch": 1.6700415368639667, "grad_norm": 1.6647380590438843, "learning_rate": 8.141785115742049e-07, "loss": 0.09659554064273834, "step": 6433 }, { "epoch": 1.670301142263759, "grad_norm": 1.953577995300293, "learning_rate": 8.136965222248713e-07, "loss": 0.05297382175922394, "step": 6434 }, { "epoch": 1.6705607476635516, "grad_norm": 1.2001700401306152, "learning_rate": 8.132148751696556e-07, "loss": 0.10293132066726685, "step": 6435 }, { "epoch": 1.6708203530633439, "grad_norm": 1.191983938217163, "learning_rate": 8.127335704937042e-07, "loss": 0.09209059178829193, "step": 6436 }, { "epoch": 1.6710799584631362, "grad_norm": 1.0104362964630127, "learning_rate": 8.122526082820994e-07, "loss": 0.10228706151247025, "step": 6437 }, { "epoch": 1.6713395638629285, "grad_norm": 1.3938416242599487, "learning_rate": 8.117719886198655e-07, "loss": 0.10400702804327011, "step": 6438 }, { "epoch": 1.6715991692627208, "grad_norm": 3.007416248321533, "learning_rate": 8.112917115919644e-07, "loss": 0.10056883096694946, "step": 6439 }, { "epoch": 1.671858774662513, "grad_norm": 2.7198402881622314, "learning_rate": 8.108117772832977e-07, "loss": 0.0535820834338665, "step": 6440 }, { "epoch": 1.6721183800623054, "grad_norm": 1.3525784015655518, "learning_rate": 8.103321857787077e-07, "loss": 0.10318601131439209, "step": 6441 }, { "epoch": 1.6723779854620977, "grad_norm": 2.5632529258728027, "learning_rate": 8.09852937162975e-07, "loss": 0.10208840668201447, "step": 6442 }, { "epoch": 1.67263759086189, "grad_norm": 1.515732765197754, "learning_rate": 8.093740315208188e-07, "loss": 0.08900725841522217, "step": 6443 }, { "epoch": 1.6728971962616823, "grad_norm": 1.7205299139022827, "learning_rate": 8.088954689368999e-07, "loss": 0.07471688091754913, "step": 6444 }, { "epoch": 1.6731568016614746, "grad_norm": 2.2358148097991943, "learning_rate": 8.084172494958168e-07, "loss": 0.07203377783298492, "step": 6445 }, { "epoch": 1.673416407061267, "grad_norm": 1.4725360870361328, "learning_rate": 8.07939373282107e-07, "loss": 0.08382351696491241, "step": 6446 }, { "epoch": 1.6736760124610592, "grad_norm": 1.8976283073425293, "learning_rate": 8.074618403802494e-07, "loss": 0.05738735944032669, "step": 6447 }, { "epoch": 1.6739356178608515, "grad_norm": 2.096250295639038, "learning_rate": 8.069846508746594e-07, "loss": 0.07278002798557281, "step": 6448 }, { "epoch": 1.6741952232606438, "grad_norm": 1.6522059440612793, "learning_rate": 8.065078048496946e-07, "loss": 0.07345163077116013, "step": 6449 }, { "epoch": 1.6744548286604362, "grad_norm": 1.290840744972229, "learning_rate": 8.060313023896499e-07, "loss": 0.10128960013389587, "step": 6450 }, { "epoch": 1.6747144340602285, "grad_norm": 1.6204620599746704, "learning_rate": 8.055551435787595e-07, "loss": 0.067634716629982, "step": 6451 }, { "epoch": 1.6749740394600208, "grad_norm": 2.1218550205230713, "learning_rate": 8.050793285011983e-07, "loss": 0.06359545886516571, "step": 6452 }, { "epoch": 1.675233644859813, "grad_norm": 2.261432409286499, "learning_rate": 8.04603857241079e-07, "loss": 0.07115539163351059, "step": 6453 }, { "epoch": 1.6754932502596054, "grad_norm": 2.158346176147461, "learning_rate": 8.041287298824535e-07, "loss": 0.06876610219478607, "step": 6454 }, { "epoch": 1.6757528556593977, "grad_norm": 2.4445767402648926, "learning_rate": 8.036539465093146e-07, "loss": 0.07179165631532669, "step": 6455 }, { "epoch": 1.67601246105919, "grad_norm": 2.1166138648986816, "learning_rate": 8.031795072055921e-07, "loss": 0.06974770128726959, "step": 6456 }, { "epoch": 1.6762720664589823, "grad_norm": 3.466080904006958, "learning_rate": 8.027054120551572e-07, "loss": 0.05988018587231636, "step": 6457 }, { "epoch": 1.6765316718587746, "grad_norm": 1.32033371925354, "learning_rate": 8.022316611418176e-07, "loss": 0.08392873406410217, "step": 6458 }, { "epoch": 1.676791277258567, "grad_norm": 2.3455686569213867, "learning_rate": 8.017582545493221e-07, "loss": 0.06152409315109253, "step": 6459 }, { "epoch": 1.6770508826583592, "grad_norm": 1.4250460863113403, "learning_rate": 8.01285192361359e-07, "loss": 0.08897790312767029, "step": 6460 }, { "epoch": 1.6773104880581515, "grad_norm": 2.101304054260254, "learning_rate": 8.008124746615542e-07, "loss": 0.05367983877658844, "step": 6461 }, { "epoch": 1.6775700934579438, "grad_norm": 1.8463231325149536, "learning_rate": 8.003401015334734e-07, "loss": 0.0848567932844162, "step": 6462 }, { "epoch": 1.6778296988577361, "grad_norm": 1.7880767583847046, "learning_rate": 7.99868073060621e-07, "loss": 0.0989476889371872, "step": 6463 }, { "epoch": 1.6780893042575284, "grad_norm": 1.4973489046096802, "learning_rate": 7.99396389326441e-07, "loss": 0.07696128636598587, "step": 6464 }, { "epoch": 1.6783489096573208, "grad_norm": 1.207213282585144, "learning_rate": 7.989250504143174e-07, "loss": 0.10308215767145157, "step": 6465 }, { "epoch": 1.678608515057113, "grad_norm": 1.816024899482727, "learning_rate": 7.984540564075702e-07, "loss": 0.01910896971821785, "step": 6466 }, { "epoch": 1.6788681204569054, "grad_norm": 2.539217710494995, "learning_rate": 7.979834073894621e-07, "loss": 0.09319277107715607, "step": 6467 }, { "epoch": 1.6791277258566977, "grad_norm": 2.2213282585144043, "learning_rate": 7.975131034431917e-07, "loss": 0.06195490062236786, "step": 6468 }, { "epoch": 1.67938733125649, "grad_norm": 1.6422065496444702, "learning_rate": 7.970431446518991e-07, "loss": 0.08690430223941803, "step": 6469 }, { "epoch": 1.6796469366562823, "grad_norm": 4.008612155914307, "learning_rate": 7.965735310986615e-07, "loss": 0.07793448865413666, "step": 6470 }, { "epoch": 1.6799065420560748, "grad_norm": 1.8974239826202393, "learning_rate": 7.961042628664958e-07, "loss": 0.10397543013095856, "step": 6471 }, { "epoch": 1.6801661474558671, "grad_norm": 1.8724342584609985, "learning_rate": 7.956353400383586e-07, "loss": 0.09806199371814728, "step": 6472 }, { "epoch": 1.6804257528556594, "grad_norm": 1.6734036207199097, "learning_rate": 7.951667626971443e-07, "loss": 0.07063285261392593, "step": 6473 }, { "epoch": 1.6806853582554517, "grad_norm": 1.5241410732269287, "learning_rate": 7.94698530925686e-07, "loss": 0.09278473258018494, "step": 6474 }, { "epoch": 1.680944963655244, "grad_norm": 1.3520056009292603, "learning_rate": 7.942306448067574e-07, "loss": 0.0880584642291069, "step": 6475 }, { "epoch": 1.6812045690550363, "grad_norm": 2.0141282081604004, "learning_rate": 7.937631044230699e-07, "loss": 0.057640790939331055, "step": 6476 }, { "epoch": 1.6814641744548287, "grad_norm": 2.003495693206787, "learning_rate": 7.932959098572731e-07, "loss": 0.08314405381679535, "step": 6477 }, { "epoch": 1.681723779854621, "grad_norm": 1.9200068712234497, "learning_rate": 7.928290611919573e-07, "loss": 0.06241118907928467, "step": 6478 }, { "epoch": 1.6819833852544133, "grad_norm": 1.3850253820419312, "learning_rate": 7.923625585096496e-07, "loss": 0.08484987169504166, "step": 6479 }, { "epoch": 1.6822429906542056, "grad_norm": 2.3150596618652344, "learning_rate": 7.91896401892818e-07, "loss": 0.06239655613899231, "step": 6480 }, { "epoch": 1.6825025960539979, "grad_norm": 2.0201003551483154, "learning_rate": 7.914305914238682e-07, "loss": 0.0784674882888794, "step": 6481 }, { "epoch": 1.6827622014537902, "grad_norm": 2.2138314247131348, "learning_rate": 7.909651271851438e-07, "loss": 0.05254215747117996, "step": 6482 }, { "epoch": 1.6830218068535827, "grad_norm": 2.2992160320281982, "learning_rate": 7.905000092589296e-07, "loss": 0.0639997348189354, "step": 6483 }, { "epoch": 1.683281412253375, "grad_norm": 2.994629144668579, "learning_rate": 7.900352377274471e-07, "loss": 0.06122151017189026, "step": 6484 }, { "epoch": 1.6835410176531673, "grad_norm": 2.861893892288208, "learning_rate": 7.895708126728569e-07, "loss": 0.024752024561166763, "step": 6485 }, { "epoch": 1.6838006230529596, "grad_norm": 1.179407000541687, "learning_rate": 7.891067341772596e-07, "loss": 0.08596646785736084, "step": 6486 }, { "epoch": 1.684060228452752, "grad_norm": 2.254335403442383, "learning_rate": 7.886430023226927e-07, "loss": 0.041039638221263885, "step": 6487 }, { "epoch": 1.6843198338525442, "grad_norm": 2.406231641769409, "learning_rate": 7.881796171911349e-07, "loss": 0.08229127526283264, "step": 6488 }, { "epoch": 1.6845794392523366, "grad_norm": 2.1684181690216064, "learning_rate": 7.877165788644998e-07, "loss": 0.0650898739695549, "step": 6489 }, { "epoch": 1.6848390446521289, "grad_norm": 1.9882370233535767, "learning_rate": 7.872538874246436e-07, "loss": 0.08783434331417084, "step": 6490 }, { "epoch": 1.6850986500519212, "grad_norm": 2.16463565826416, "learning_rate": 7.867915429533596e-07, "loss": 0.10116872191429138, "step": 6491 }, { "epoch": 1.6853582554517135, "grad_norm": 1.1019666194915771, "learning_rate": 7.863295455323791e-07, "loss": 0.10954839736223221, "step": 6492 }, { "epoch": 1.6856178608515058, "grad_norm": 1.5283515453338623, "learning_rate": 7.85867895243373e-07, "loss": 0.08235436677932739, "step": 6493 }, { "epoch": 1.685877466251298, "grad_norm": 1.8114030361175537, "learning_rate": 7.854065921679499e-07, "loss": 0.03311125189065933, "step": 6494 }, { "epoch": 1.6861370716510904, "grad_norm": 1.0458347797393799, "learning_rate": 7.849456363876579e-07, "loss": 0.08191180229187012, "step": 6495 }, { "epoch": 1.6863966770508827, "grad_norm": 0.93790203332901, "learning_rate": 7.844850279839847e-07, "loss": 0.09679493308067322, "step": 6496 }, { "epoch": 1.686656282450675, "grad_norm": 1.675896406173706, "learning_rate": 7.84024767038353e-07, "loss": 0.07560575008392334, "step": 6497 }, { "epoch": 1.6869158878504673, "grad_norm": 2.169269323348999, "learning_rate": 7.835648536321285e-07, "loss": 0.08962717652320862, "step": 6498 }, { "epoch": 1.6871754932502596, "grad_norm": 1.6257996559143066, "learning_rate": 7.831052878466124e-07, "loss": 0.09382608532905579, "step": 6499 }, { "epoch": 1.687435098650052, "grad_norm": 2.496943473815918, "learning_rate": 7.826460697630449e-07, "loss": 0.04880600422620773, "step": 6500 }, { "epoch": 1.6876947040498442, "grad_norm": 2.2454121112823486, "learning_rate": 7.821871994626062e-07, "loss": 0.08537280559539795, "step": 6501 }, { "epoch": 1.6879543094496365, "grad_norm": 2.74599552154541, "learning_rate": 7.817286770264134e-07, "loss": 0.0722728818655014, "step": 6502 }, { "epoch": 1.6882139148494288, "grad_norm": 1.0327144861221313, "learning_rate": 7.812705025355234e-07, "loss": 0.099643275141716, "step": 6503 }, { "epoch": 1.6884735202492211, "grad_norm": 3.427330255508423, "learning_rate": 7.808126760709309e-07, "loss": 0.0586513876914978, "step": 6504 }, { "epoch": 1.6887331256490135, "grad_norm": 1.6931445598602295, "learning_rate": 7.803551977135681e-07, "loss": 0.039730194956064224, "step": 6505 }, { "epoch": 1.6889927310488058, "grad_norm": 1.5435994863510132, "learning_rate": 7.798980675443079e-07, "loss": 0.08441195636987686, "step": 6506 }, { "epoch": 1.689252336448598, "grad_norm": 2.567746877670288, "learning_rate": 7.794412856439602e-07, "loss": 0.047220610082149506, "step": 6507 }, { "epoch": 1.6895119418483904, "grad_norm": 2.069852590560913, "learning_rate": 7.789848520932729e-07, "loss": 0.05799632519483566, "step": 6508 }, { "epoch": 1.6897715472481827, "grad_norm": 2.077388048171997, "learning_rate": 7.785287669729338e-07, "loss": 0.08302589505910873, "step": 6509 }, { "epoch": 1.690031152647975, "grad_norm": 1.5663665533065796, "learning_rate": 7.780730303635675e-07, "loss": 0.08247379958629608, "step": 6510 }, { "epoch": 1.6902907580477673, "grad_norm": 2.2488796710968018, "learning_rate": 7.776176423457391e-07, "loss": 0.023639028891921043, "step": 6511 }, { "epoch": 1.6905503634475596, "grad_norm": 1.5271780490875244, "learning_rate": 7.771626029999496e-07, "loss": 0.0818367749452591, "step": 6512 }, { "epoch": 1.690809968847352, "grad_norm": 2.993926525115967, "learning_rate": 7.767079124066397e-07, "loss": 0.06825617700815201, "step": 6513 }, { "epoch": 1.6910695742471442, "grad_norm": 1.2350437641143799, "learning_rate": 7.762535706461887e-07, "loss": 0.10546356439590454, "step": 6514 }, { "epoch": 1.6913291796469365, "grad_norm": 2.276855945587158, "learning_rate": 7.757995777989139e-07, "loss": 0.08207212388515472, "step": 6515 }, { "epoch": 1.6915887850467288, "grad_norm": 1.2682808637619019, "learning_rate": 7.753459339450699e-07, "loss": 0.08655976504087448, "step": 6516 }, { "epoch": 1.6918483904465211, "grad_norm": 1.8644741773605347, "learning_rate": 7.748926391648516e-07, "loss": 0.07960353791713715, "step": 6517 }, { "epoch": 1.6921079958463134, "grad_norm": 2.1488490104675293, "learning_rate": 7.744396935383904e-07, "loss": 0.03883454203605652, "step": 6518 }, { "epoch": 1.692367601246106, "grad_norm": 1.234410285949707, "learning_rate": 7.739870971457581e-07, "loss": 0.03512030467391014, "step": 6519 }, { "epoch": 1.6926272066458983, "grad_norm": 3.2094032764434814, "learning_rate": 7.735348500669611e-07, "loss": 0.043461743742227554, "step": 6520 }, { "epoch": 1.6928868120456906, "grad_norm": 2.049318552017212, "learning_rate": 7.730829523819477e-07, "loss": 0.09322839230298996, "step": 6521 }, { "epoch": 1.6931464174454829, "grad_norm": 1.7112902402877808, "learning_rate": 7.726314041706031e-07, "loss": 0.09541598707437515, "step": 6522 }, { "epoch": 1.6934060228452752, "grad_norm": 2.8715898990631104, "learning_rate": 7.721802055127509e-07, "loss": 0.041238926351070404, "step": 6523 }, { "epoch": 1.6936656282450675, "grad_norm": 2.492448329925537, "learning_rate": 7.717293564881521e-07, "loss": 0.05143604055047035, "step": 6524 }, { "epoch": 1.6939252336448598, "grad_norm": 1.1554731130599976, "learning_rate": 7.712788571765065e-07, "loss": 0.09017448127269745, "step": 6525 }, { "epoch": 1.6941848390446521, "grad_norm": 1.8600960969924927, "learning_rate": 7.708287076574523e-07, "loss": 0.08325241506099701, "step": 6526 }, { "epoch": 1.6944444444444444, "grad_norm": 2.2647953033447266, "learning_rate": 7.703789080105657e-07, "loss": 0.06577633321285248, "step": 6527 }, { "epoch": 1.6947040498442367, "grad_norm": 1.4484788179397583, "learning_rate": 7.699294583153605e-07, "loss": 0.0851621925830841, "step": 6528 }, { "epoch": 1.694963655244029, "grad_norm": 1.8545869588851929, "learning_rate": 7.694803586512899e-07, "loss": 0.0726260244846344, "step": 6529 }, { "epoch": 1.6952232606438213, "grad_norm": 2.359443426132202, "learning_rate": 7.690316090977437e-07, "loss": 0.08181767165660858, "step": 6530 }, { "epoch": 1.6954828660436139, "grad_norm": 1.8615903854370117, "learning_rate": 7.685832097340508e-07, "loss": 0.08144894987344742, "step": 6531 }, { "epoch": 1.6957424714434062, "grad_norm": 1.4974712133407593, "learning_rate": 7.681351606394784e-07, "loss": 0.08170641213655472, "step": 6532 }, { "epoch": 1.6960020768431985, "grad_norm": 2.2285847663879395, "learning_rate": 7.676874618932302e-07, "loss": 0.081515833735466, "step": 6533 }, { "epoch": 1.6962616822429908, "grad_norm": 1.323750376701355, "learning_rate": 7.672401135744503e-07, "loss": 0.07932787388563156, "step": 6534 }, { "epoch": 1.696521287642783, "grad_norm": 2.2289516925811768, "learning_rate": 7.667931157622192e-07, "loss": 0.06484930962324142, "step": 6535 }, { "epoch": 1.6967808930425754, "grad_norm": 1.3101439476013184, "learning_rate": 7.663464685355555e-07, "loss": 0.07705655694007874, "step": 6536 }, { "epoch": 1.6970404984423677, "grad_norm": 1.9081729650497437, "learning_rate": 7.65900171973417e-07, "loss": 0.08336016535758972, "step": 6537 }, { "epoch": 1.69730010384216, "grad_norm": 1.466562271118164, "learning_rate": 7.65454226154698e-07, "loss": 0.07249851524829865, "step": 6538 }, { "epoch": 1.6975597092419523, "grad_norm": 1.9896225929260254, "learning_rate": 7.650086311582314e-07, "loss": 0.03256934881210327, "step": 6539 }, { "epoch": 1.6978193146417446, "grad_norm": 1.1896528005599976, "learning_rate": 7.645633870627889e-07, "loss": 0.08783447742462158, "step": 6540 }, { "epoch": 1.698078920041537, "grad_norm": 1.3562264442443848, "learning_rate": 7.641184939470788e-07, "loss": 0.05375359579920769, "step": 6541 }, { "epoch": 1.6983385254413292, "grad_norm": 2.2511847019195557, "learning_rate": 7.636739518897488e-07, "loss": 0.05069494619965553, "step": 6542 }, { "epoch": 1.6985981308411215, "grad_norm": 1.6111931800842285, "learning_rate": 7.632297609693828e-07, "loss": 0.08510806411504745, "step": 6543 }, { "epoch": 1.6988577362409139, "grad_norm": 1.6950150728225708, "learning_rate": 7.627859212645041e-07, "loss": 0.04882027208805084, "step": 6544 }, { "epoch": 1.6991173416407062, "grad_norm": 1.4669711589813232, "learning_rate": 7.623424328535738e-07, "loss": 0.038320787250995636, "step": 6545 }, { "epoch": 1.6993769470404985, "grad_norm": 2.9096806049346924, "learning_rate": 7.618992958149897e-07, "loss": 0.030843954533338547, "step": 6546 }, { "epoch": 1.6996365524402908, "grad_norm": 1.480969786643982, "learning_rate": 7.614565102270885e-07, "loss": 0.08304178714752197, "step": 6547 }, { "epoch": 1.699896157840083, "grad_norm": 2.9340550899505615, "learning_rate": 7.610140761681449e-07, "loss": 0.04275555536150932, "step": 6548 }, { "epoch": 1.7001557632398754, "grad_norm": 1.8587918281555176, "learning_rate": 7.605719937163705e-07, "loss": 0.06476134806871414, "step": 6549 }, { "epoch": 1.7004153686396677, "grad_norm": 1.8355411291122437, "learning_rate": 7.601302629499168e-07, "loss": 0.08328711986541748, "step": 6550 }, { "epoch": 1.70067497403946, "grad_norm": 2.6313841342926025, "learning_rate": 7.596888839468696e-07, "loss": 0.05445322394371033, "step": 6551 }, { "epoch": 1.7009345794392523, "grad_norm": 2.152325391769409, "learning_rate": 7.592478567852557e-07, "loss": 0.07356654852628708, "step": 6552 }, { "epoch": 1.7011941848390446, "grad_norm": 2.923067331314087, "learning_rate": 7.588071815430393e-07, "loss": 0.07410446554422379, "step": 6553 }, { "epoch": 1.701453790238837, "grad_norm": 1.1306278705596924, "learning_rate": 7.583668582981202e-07, "loss": 0.09285618364810944, "step": 6554 }, { "epoch": 1.7017133956386292, "grad_norm": 1.1921101808547974, "learning_rate": 7.579268871283385e-07, "loss": 0.09463024139404297, "step": 6555 }, { "epoch": 1.7019730010384215, "grad_norm": 1.5822888612747192, "learning_rate": 7.574872681114705e-07, "loss": 0.06433798372745514, "step": 6556 }, { "epoch": 1.7022326064382138, "grad_norm": 2.429471969604492, "learning_rate": 7.570480013252313e-07, "loss": 0.0769626796245575, "step": 6557 }, { "epoch": 1.7024922118380061, "grad_norm": 2.002030372619629, "learning_rate": 7.566090868472728e-07, "loss": 0.08577494323253632, "step": 6558 }, { "epoch": 1.7027518172377984, "grad_norm": 1.8257428407669067, "learning_rate": 7.56170524755185e-07, "loss": 0.02975662238895893, "step": 6559 }, { "epoch": 1.7030114226375908, "grad_norm": 1.5287071466445923, "learning_rate": 7.557323151264958e-07, "loss": 0.09089391678571701, "step": 6560 }, { "epoch": 1.703271028037383, "grad_norm": 2.0357680320739746, "learning_rate": 7.552944580386709e-07, "loss": 0.05317620933055878, "step": 6561 }, { "epoch": 1.7035306334371754, "grad_norm": 2.001725673675537, "learning_rate": 7.548569535691127e-07, "loss": 0.054988689720630646, "step": 6562 }, { "epoch": 1.7037902388369677, "grad_norm": 3.106879949569702, "learning_rate": 7.544198017951626e-07, "loss": 0.05154892057180405, "step": 6563 }, { "epoch": 1.70404984423676, "grad_norm": 2.034677505493164, "learning_rate": 7.539830027940986e-07, "loss": 0.03895267844200134, "step": 6564 }, { "epoch": 1.7043094496365523, "grad_norm": 2.9195737838745117, "learning_rate": 7.535465566431374e-07, "loss": 0.047226909548044205, "step": 6565 }, { "epoch": 1.7045690550363446, "grad_norm": 1.18998384475708, "learning_rate": 7.531104634194323e-07, "loss": 0.10099007189273834, "step": 6566 }, { "epoch": 1.7048286604361371, "grad_norm": 1.1626052856445312, "learning_rate": 7.526747232000737e-07, "loss": 0.10078109055757523, "step": 6567 }, { "epoch": 1.7050882658359294, "grad_norm": 1.3182363510131836, "learning_rate": 7.522393360620922e-07, "loss": 0.09052585065364838, "step": 6568 }, { "epoch": 1.7053478712357217, "grad_norm": 1.3844417333602905, "learning_rate": 7.518043020824535e-07, "loss": 0.07760043442249298, "step": 6569 }, { "epoch": 1.705607476635514, "grad_norm": 2.177809953689575, "learning_rate": 7.513696213380608e-07, "loss": 0.05764910206198692, "step": 6570 }, { "epoch": 1.7058670820353063, "grad_norm": 1.4141857624053955, "learning_rate": 7.509352939057573e-07, "loss": 0.0842357724905014, "step": 6571 }, { "epoch": 1.7061266874350987, "grad_norm": 1.444898009300232, "learning_rate": 7.505013198623207e-07, "loss": 0.11788642406463623, "step": 6572 }, { "epoch": 1.706386292834891, "grad_norm": 2.1647095680236816, "learning_rate": 7.500676992844686e-07, "loss": 0.09671364724636078, "step": 6573 }, { "epoch": 1.7066458982346833, "grad_norm": 3.010054588317871, "learning_rate": 7.49634432248855e-07, "loss": 0.04424816742539406, "step": 6574 }, { "epoch": 1.7069055036344756, "grad_norm": 2.1595537662506104, "learning_rate": 7.492015188320711e-07, "loss": 0.11447601020336151, "step": 6575 }, { "epoch": 1.7071651090342679, "grad_norm": 2.2732245922088623, "learning_rate": 7.487689591106467e-07, "loss": 0.08929155766963959, "step": 6576 }, { "epoch": 1.7074247144340602, "grad_norm": 1.4430207014083862, "learning_rate": 7.483367531610483e-07, "loss": 0.06779851764440536, "step": 6577 }, { "epoch": 1.7076843198338525, "grad_norm": 1.3338934183120728, "learning_rate": 7.479049010596795e-07, "loss": 0.09125933796167374, "step": 6578 }, { "epoch": 1.707943925233645, "grad_norm": 1.2604018449783325, "learning_rate": 7.474734028828828e-07, "loss": 0.10802806913852692, "step": 6579 }, { "epoch": 1.7082035306334373, "grad_norm": 1.3882639408111572, "learning_rate": 7.470422587069368e-07, "loss": 0.09761957824230194, "step": 6580 }, { "epoch": 1.7084631360332296, "grad_norm": 3.12497615814209, "learning_rate": 7.466114686080576e-07, "loss": 0.054441966116428375, "step": 6581 }, { "epoch": 1.708722741433022, "grad_norm": 1.94308340549469, "learning_rate": 7.461810326623988e-07, "loss": 0.09880030155181885, "step": 6582 }, { "epoch": 1.7089823468328142, "grad_norm": 2.37250018119812, "learning_rate": 7.457509509460522e-07, "loss": 0.06435172259807587, "step": 6583 }, { "epoch": 1.7092419522326066, "grad_norm": 2.676663398742676, "learning_rate": 7.453212235350471e-07, "loss": 0.03963472694158554, "step": 6584 }, { "epoch": 1.7095015576323989, "grad_norm": 2.1237502098083496, "learning_rate": 7.448918505053476e-07, "loss": 0.05987415090203285, "step": 6585 }, { "epoch": 1.7097611630321912, "grad_norm": 2.016618490219116, "learning_rate": 7.444628319328586e-07, "loss": 0.08095227181911469, "step": 6586 }, { "epoch": 1.7100207684319835, "grad_norm": 3.059905767440796, "learning_rate": 7.440341678934197e-07, "loss": 0.05849333852529526, "step": 6587 }, { "epoch": 1.7102803738317758, "grad_norm": 2.125175952911377, "learning_rate": 7.436058584628098e-07, "loss": 0.0913119688630104, "step": 6588 }, { "epoch": 1.710539979231568, "grad_norm": 3.1105360984802246, "learning_rate": 7.431779037167437e-07, "loss": 0.08960910141468048, "step": 6589 }, { "epoch": 1.7107995846313604, "grad_norm": 2.1753673553466797, "learning_rate": 7.42750303730874e-07, "loss": 0.042567696422338486, "step": 6590 }, { "epoch": 1.7110591900311527, "grad_norm": 2.263216733932495, "learning_rate": 7.423230585807907e-07, "loss": 0.07997673749923706, "step": 6591 }, { "epoch": 1.711318795430945, "grad_norm": 1.6470228433609009, "learning_rate": 7.418961683420211e-07, "loss": 0.06090661883354187, "step": 6592 }, { "epoch": 1.7115784008307373, "grad_norm": 2.148578405380249, "learning_rate": 7.414696330900291e-07, "loss": 0.07779084146022797, "step": 6593 }, { "epoch": 1.7118380062305296, "grad_norm": 1.1753867864608765, "learning_rate": 7.410434529002172e-07, "loss": 0.10399352759122849, "step": 6594 }, { "epoch": 1.712097611630322, "grad_norm": 1.995680332183838, "learning_rate": 7.406176278479232e-07, "loss": 0.08029837161302567, "step": 6595 }, { "epoch": 1.7123572170301142, "grad_norm": 2.7957632541656494, "learning_rate": 7.401921580084244e-07, "loss": 0.07140211760997772, "step": 6596 }, { "epoch": 1.7126168224299065, "grad_norm": 3.188992738723755, "learning_rate": 7.397670434569335e-07, "loss": 0.062142618000507355, "step": 6597 }, { "epoch": 1.7128764278296988, "grad_norm": 2.33357572555542, "learning_rate": 7.393422842686008e-07, "loss": 0.07926957309246063, "step": 6598 }, { "epoch": 1.7131360332294912, "grad_norm": 2.046165704727173, "learning_rate": 7.389178805185145e-07, "loss": 0.08474661409854889, "step": 6599 }, { "epoch": 1.7133956386292835, "grad_norm": 0.9959151148796082, "learning_rate": 7.384938322816995e-07, "loss": 0.09870778024196625, "step": 6600 }, { "epoch": 1.7136552440290758, "grad_norm": 0.9580398797988892, "learning_rate": 7.380701396331168e-07, "loss": 0.11947083473205566, "step": 6601 }, { "epoch": 1.713914849428868, "grad_norm": 2.9951655864715576, "learning_rate": 7.376468026476669e-07, "loss": 0.07142682373523712, "step": 6602 }, { "epoch": 1.7141744548286604, "grad_norm": 2.372795820236206, "learning_rate": 7.372238214001852e-07, "loss": 0.029156440868973732, "step": 6603 }, { "epoch": 1.7144340602284527, "grad_norm": 2.1310296058654785, "learning_rate": 7.368011959654456e-07, "loss": 0.06103108823299408, "step": 6604 }, { "epoch": 1.714693665628245, "grad_norm": 2.04410982131958, "learning_rate": 7.363789264181586e-07, "loss": 0.03284183889627457, "step": 6605 }, { "epoch": 1.7149532710280373, "grad_norm": 2.226121187210083, "learning_rate": 7.359570128329711e-07, "loss": 0.06470994651317596, "step": 6606 }, { "epoch": 1.7152128764278296, "grad_norm": 2.003514051437378, "learning_rate": 7.355354552844693e-07, "loss": 0.07837729156017303, "step": 6607 }, { "epoch": 1.715472481827622, "grad_norm": 1.288948893547058, "learning_rate": 7.351142538471728e-07, "loss": 0.08834633231163025, "step": 6608 }, { "epoch": 1.7157320872274142, "grad_norm": 1.6431137323379517, "learning_rate": 7.346934085955416e-07, "loss": 0.05205089598894119, "step": 6609 }, { "epoch": 1.7159916926272065, "grad_norm": 1.465664029121399, "learning_rate": 7.342729196039716e-07, "loss": 0.08328511565923691, "step": 6610 }, { "epoch": 1.7162512980269988, "grad_norm": 1.3969156742095947, "learning_rate": 7.338527869467954e-07, "loss": 0.09128876030445099, "step": 6611 }, { "epoch": 1.7165109034267911, "grad_norm": 1.197123646736145, "learning_rate": 7.33433010698283e-07, "loss": 0.08775870501995087, "step": 6612 }, { "epoch": 1.7167705088265834, "grad_norm": 1.2985303401947021, "learning_rate": 7.330135909326405e-07, "loss": 0.08904077112674713, "step": 6613 }, { "epoch": 1.7170301142263757, "grad_norm": 1.3927998542785645, "learning_rate": 7.325945277240124e-07, "loss": 0.08158086240291595, "step": 6614 }, { "epoch": 1.7172897196261683, "grad_norm": 1.4426997900009155, "learning_rate": 7.3217582114648e-07, "loss": 0.08171328157186508, "step": 6615 }, { "epoch": 1.7175493250259606, "grad_norm": 1.2017768621444702, "learning_rate": 7.317574712740595e-07, "loss": 0.08592475205659866, "step": 6616 }, { "epoch": 1.717808930425753, "grad_norm": 1.299126148223877, "learning_rate": 7.313394781807066e-07, "loss": 0.10001876950263977, "step": 6617 }, { "epoch": 1.7180685358255452, "grad_norm": 3.2293341159820557, "learning_rate": 7.309218419403132e-07, "loss": 0.06167423725128174, "step": 6618 }, { "epoch": 1.7183281412253375, "grad_norm": 1.3710651397705078, "learning_rate": 7.305045626267074e-07, "loss": 0.06528948247432709, "step": 6619 }, { "epoch": 1.7185877466251298, "grad_norm": 2.148672342300415, "learning_rate": 7.300876403136544e-07, "loss": 0.02974085882306099, "step": 6620 }, { "epoch": 1.7188473520249221, "grad_norm": 1.0456980466842651, "learning_rate": 7.296710750748563e-07, "loss": 0.10935233533382416, "step": 6621 }, { "epoch": 1.7191069574247144, "grad_norm": 1.9632270336151123, "learning_rate": 7.292548669839531e-07, "loss": 0.08894359320402145, "step": 6622 }, { "epoch": 1.7193665628245067, "grad_norm": 2.9443705081939697, "learning_rate": 7.288390161145209e-07, "loss": 0.07158803939819336, "step": 6623 }, { "epoch": 1.719626168224299, "grad_norm": 2.0157742500305176, "learning_rate": 7.284235225400716e-07, "loss": 0.06490842252969742, "step": 6624 }, { "epoch": 1.7198857736240913, "grad_norm": 2.4821712970733643, "learning_rate": 7.280083863340559e-07, "loss": 0.0518374964594841, "step": 6625 }, { "epoch": 1.7201453790238836, "grad_norm": 2.2051076889038086, "learning_rate": 7.275936075698597e-07, "loss": 0.09571842104196548, "step": 6626 }, { "epoch": 1.7204049844236762, "grad_norm": 2.1074371337890625, "learning_rate": 7.271791863208067e-07, "loss": 0.05809875577688217, "step": 6627 }, { "epoch": 1.7206645898234685, "grad_norm": 1.2272320985794067, "learning_rate": 7.267651226601577e-07, "loss": 0.10089132189750671, "step": 6628 }, { "epoch": 1.7209241952232608, "grad_norm": 2.5028011798858643, "learning_rate": 7.263514166611083e-07, "loss": 0.040413159877061844, "step": 6629 }, { "epoch": 1.721183800623053, "grad_norm": 1.0776456594467163, "learning_rate": 7.259380683967936e-07, "loss": 0.10344883799552917, "step": 6630 }, { "epoch": 1.7214434060228454, "grad_norm": 1.9528888463974, "learning_rate": 7.255250779402835e-07, "loss": 0.058811258524656296, "step": 6631 }, { "epoch": 1.7217030114226377, "grad_norm": 2.006368398666382, "learning_rate": 7.251124453645849e-07, "loss": 0.05267078056931496, "step": 6632 }, { "epoch": 1.72196261682243, "grad_norm": 1.21024489402771, "learning_rate": 7.247001707426424e-07, "loss": 0.08394210040569305, "step": 6633 }, { "epoch": 1.7222222222222223, "grad_norm": 2.425215721130371, "learning_rate": 7.242882541473367e-07, "loss": 0.08226372301578522, "step": 6634 }, { "epoch": 1.7224818276220146, "grad_norm": 0.9033045768737793, "learning_rate": 7.238766956514844e-07, "loss": 0.030611690133810043, "step": 6635 }, { "epoch": 1.722741433021807, "grad_norm": 2.010739326477051, "learning_rate": 7.23465495327841e-07, "loss": 0.04378077760338783, "step": 6636 }, { "epoch": 1.7230010384215992, "grad_norm": 1.5161865949630737, "learning_rate": 7.230546532490958e-07, "loss": 0.08385762572288513, "step": 6637 }, { "epoch": 1.7232606438213915, "grad_norm": 1.1090604066848755, "learning_rate": 7.226441694878774e-07, "loss": 0.08969444036483765, "step": 6638 }, { "epoch": 1.7235202492211839, "grad_norm": 1.910886526107788, "learning_rate": 7.222340441167498e-07, "loss": 0.08083392679691315, "step": 6639 }, { "epoch": 1.7237798546209762, "grad_norm": 2.2658989429473877, "learning_rate": 7.218242772082131e-07, "loss": 0.0616963692009449, "step": 6640 }, { "epoch": 1.7240394600207685, "grad_norm": 1.6958723068237305, "learning_rate": 7.214148688347055e-07, "loss": 0.031177662312984467, "step": 6641 }, { "epoch": 1.7242990654205608, "grad_norm": 1.618383526802063, "learning_rate": 7.210058190686003e-07, "loss": 0.08957258611917496, "step": 6642 }, { "epoch": 1.724558670820353, "grad_norm": 1.893794298171997, "learning_rate": 7.205971279822084e-07, "loss": 0.09630763530731201, "step": 6643 }, { "epoch": 1.7248182762201454, "grad_norm": 1.5793176889419556, "learning_rate": 7.201887956477774e-07, "loss": 0.0712747871875763, "step": 6644 }, { "epoch": 1.7250778816199377, "grad_norm": 4.1135430335998535, "learning_rate": 7.197808221374904e-07, "loss": 0.06734354048967361, "step": 6645 }, { "epoch": 1.72533748701973, "grad_norm": 2.7595300674438477, "learning_rate": 7.193732075234693e-07, "loss": 0.03913411870598793, "step": 6646 }, { "epoch": 1.7255970924195223, "grad_norm": 2.768069267272949, "learning_rate": 7.189659518777687e-07, "loss": 0.034094568341970444, "step": 6647 }, { "epoch": 1.7258566978193146, "grad_norm": 1.713606595993042, "learning_rate": 7.185590552723834e-07, "loss": 0.026959307491779327, "step": 6648 }, { "epoch": 1.726116303219107, "grad_norm": 2.0713608264923096, "learning_rate": 7.181525177792436e-07, "loss": 0.03321655094623566, "step": 6649 }, { "epoch": 1.7263759086188992, "grad_norm": 2.714477300643921, "learning_rate": 7.177463394702157e-07, "loss": 0.09514698386192322, "step": 6650 }, { "epoch": 1.7266355140186915, "grad_norm": 1.727353811264038, "learning_rate": 7.173405204171028e-07, "loss": 0.06921111047267914, "step": 6651 }, { "epoch": 1.7268951194184838, "grad_norm": 1.4285537004470825, "learning_rate": 7.169350606916434e-07, "loss": 0.041571952402591705, "step": 6652 }, { "epoch": 1.7271547248182761, "grad_norm": 2.4600670337677, "learning_rate": 7.165299603655141e-07, "loss": 0.0688866525888443, "step": 6653 }, { "epoch": 1.7274143302180685, "grad_norm": 2.053046703338623, "learning_rate": 7.161252195103286e-07, "loss": 0.05796576663851738, "step": 6654 }, { "epoch": 1.7276739356178608, "grad_norm": 2.43062162399292, "learning_rate": 7.157208381976338e-07, "loss": 0.047514695674180984, "step": 6655 }, { "epoch": 1.727933541017653, "grad_norm": 1.040937066078186, "learning_rate": 7.153168164989166e-07, "loss": 0.11826777458190918, "step": 6656 }, { "epoch": 1.7281931464174454, "grad_norm": 1.8660262823104858, "learning_rate": 7.149131544855976e-07, "loss": 0.05684617906808853, "step": 6657 }, { "epoch": 1.7284527518172377, "grad_norm": 3.2153186798095703, "learning_rate": 7.145098522290362e-07, "loss": 0.052866481244564056, "step": 6658 }, { "epoch": 1.72871235721703, "grad_norm": 1.275315761566162, "learning_rate": 7.141069098005259e-07, "loss": 0.06390824168920517, "step": 6659 }, { "epoch": 1.7289719626168223, "grad_norm": 2.165371894836426, "learning_rate": 7.137043272712981e-07, "loss": 0.06943869590759277, "step": 6660 }, { "epoch": 1.7292315680166146, "grad_norm": 3.2625184059143066, "learning_rate": 7.133021047125206e-07, "loss": 0.075263112783432, "step": 6661 }, { "epoch": 1.729491173416407, "grad_norm": 1.9399257898330688, "learning_rate": 7.129002421952968e-07, "loss": 0.039072856307029724, "step": 6662 }, { "epoch": 1.7297507788161994, "grad_norm": 2.7445969581604004, "learning_rate": 7.124987397906661e-07, "loss": 0.05150236934423447, "step": 6663 }, { "epoch": 1.7300103842159917, "grad_norm": 1.9271217584609985, "learning_rate": 7.120975975696064e-07, "loss": 0.09165555238723755, "step": 6664 }, { "epoch": 1.730269989615784, "grad_norm": 1.5815668106079102, "learning_rate": 7.116968156030297e-07, "loss": 0.09934081137180328, "step": 6665 }, { "epoch": 1.7305295950155763, "grad_norm": 3.0584285259246826, "learning_rate": 7.112963939617845e-07, "loss": 0.05814596638083458, "step": 6666 }, { "epoch": 1.7307892004153687, "grad_norm": 2.666186571121216, "learning_rate": 7.108963327166571e-07, "loss": 0.08165848255157471, "step": 6667 }, { "epoch": 1.731048805815161, "grad_norm": 1.6078550815582275, "learning_rate": 7.104966319383684e-07, "loss": 0.028118547052145004, "step": 6668 }, { "epoch": 1.7313084112149533, "grad_norm": 3.3696773052215576, "learning_rate": 7.100972916975771e-07, "loss": 0.09941165149211884, "step": 6669 }, { "epoch": 1.7315680166147456, "grad_norm": 3.612748146057129, "learning_rate": 7.096983120648771e-07, "loss": 0.0699671059846878, "step": 6670 }, { "epoch": 1.7318276220145379, "grad_norm": 1.517237901687622, "learning_rate": 7.092996931107983e-07, "loss": 0.08698543906211853, "step": 6671 }, { "epoch": 1.7320872274143302, "grad_norm": 2.2630317211151123, "learning_rate": 7.089014349058085e-07, "loss": 0.07271747291088104, "step": 6672 }, { "epoch": 1.7323468328141225, "grad_norm": 2.9787497520446777, "learning_rate": 7.085035375203098e-07, "loss": 0.06771843880414963, "step": 6673 }, { "epoch": 1.7326064382139148, "grad_norm": 1.1924302577972412, "learning_rate": 7.081060010246415e-07, "loss": 0.11592510342597961, "step": 6674 }, { "epoch": 1.7328660436137073, "grad_norm": 1.364316701889038, "learning_rate": 7.077088254890792e-07, "loss": 0.0959082692861557, "step": 6675 }, { "epoch": 1.7331256490134996, "grad_norm": 1.9317365884780884, "learning_rate": 7.07312010983834e-07, "loss": 0.08017387241125107, "step": 6676 }, { "epoch": 1.733385254413292, "grad_norm": 2.760648250579834, "learning_rate": 7.069155575790547e-07, "loss": 0.07567417621612549, "step": 6677 }, { "epoch": 1.7336448598130842, "grad_norm": 2.0417556762695312, "learning_rate": 7.065194653448234e-07, "loss": 0.06543722003698349, "step": 6678 }, { "epoch": 1.7339044652128766, "grad_norm": 2.690678119659424, "learning_rate": 7.061237343511613e-07, "loss": 0.05457354709506035, "step": 6679 }, { "epoch": 1.7341640706126689, "grad_norm": 1.2774412631988525, "learning_rate": 7.057283646680249e-07, "loss": 0.11392857134342194, "step": 6680 }, { "epoch": 1.7344236760124612, "grad_norm": 1.4796439409255981, "learning_rate": 7.05333356365306e-07, "loss": 0.07053180038928986, "step": 6681 }, { "epoch": 1.7346832814122535, "grad_norm": 1.1399248838424683, "learning_rate": 7.04938709512833e-07, "loss": 0.10862824320793152, "step": 6682 }, { "epoch": 1.7349428868120458, "grad_norm": 1.2051246166229248, "learning_rate": 7.045444241803699e-07, "loss": 0.09413324296474457, "step": 6683 }, { "epoch": 1.735202492211838, "grad_norm": 1.3412725925445557, "learning_rate": 7.04150500437618e-07, "loss": 0.10733427107334137, "step": 6684 }, { "epoch": 1.7354620976116304, "grad_norm": 1.5527960062026978, "learning_rate": 7.037569383542147e-07, "loss": 0.09115961194038391, "step": 6685 }, { "epoch": 1.7357217030114227, "grad_norm": 1.5648986101150513, "learning_rate": 7.033637379997308e-07, "loss": 0.08282878994941711, "step": 6686 }, { "epoch": 1.735981308411215, "grad_norm": 1.2960044145584106, "learning_rate": 7.029708994436771e-07, "loss": 0.081082284450531, "step": 6687 }, { "epoch": 1.7362409138110073, "grad_norm": 2.4365146160125732, "learning_rate": 7.02578422755497e-07, "loss": 0.03711868077516556, "step": 6688 }, { "epoch": 1.7365005192107996, "grad_norm": 2.2063443660736084, "learning_rate": 7.021863080045726e-07, "loss": 0.033936165273189545, "step": 6689 }, { "epoch": 1.736760124610592, "grad_norm": 2.0485169887542725, "learning_rate": 7.017945552602201e-07, "loss": 0.044793300330638885, "step": 6690 }, { "epoch": 1.7370197300103842, "grad_norm": 1.2697584629058838, "learning_rate": 7.014031645916923e-07, "loss": 0.10068787634372711, "step": 6691 }, { "epoch": 1.7372793354101765, "grad_norm": 1.9809134006500244, "learning_rate": 7.010121360681787e-07, "loss": 0.018233973532915115, "step": 6692 }, { "epoch": 1.7375389408099688, "grad_norm": 2.30564022064209, "learning_rate": 7.006214697588036e-07, "loss": 0.04947352036833763, "step": 6693 }, { "epoch": 1.7377985462097612, "grad_norm": 1.8965952396392822, "learning_rate": 7.002311657326279e-07, "loss": 0.043812647461891174, "step": 6694 }, { "epoch": 1.7380581516095535, "grad_norm": 1.2632514238357544, "learning_rate": 6.99841224058649e-07, "loss": 0.09453131258487701, "step": 6695 }, { "epoch": 1.7383177570093458, "grad_norm": 1.9725151062011719, "learning_rate": 6.994516448057993e-07, "loss": 0.06160775572061539, "step": 6696 }, { "epoch": 1.738577362409138, "grad_norm": 1.5573371648788452, "learning_rate": 6.99062428042947e-07, "loss": 0.08295533061027527, "step": 6697 }, { "epoch": 1.7388369678089304, "grad_norm": 1.7311969995498657, "learning_rate": 6.986735738388976e-07, "loss": 0.09991371631622314, "step": 6698 }, { "epoch": 1.7390965732087227, "grad_norm": 2.673246145248413, "learning_rate": 6.982850822623907e-07, "loss": 0.07667402923107147, "step": 6699 }, { "epoch": 1.739356178608515, "grad_norm": 2.6980819702148438, "learning_rate": 6.978969533821036e-07, "loss": 0.03541853278875351, "step": 6700 }, { "epoch": 1.7396157840083073, "grad_norm": 2.128143072128296, "learning_rate": 6.975091872666484e-07, "loss": 0.06340349465608597, "step": 6701 }, { "epoch": 1.7398753894080996, "grad_norm": 1.2122608423233032, "learning_rate": 6.971217839845726e-07, "loss": 0.11294415593147278, "step": 6702 }, { "epoch": 1.740134994807892, "grad_norm": 2.4355857372283936, "learning_rate": 6.96734743604361e-07, "loss": 0.09141701459884644, "step": 6703 }, { "epoch": 1.7403946002076842, "grad_norm": 3.7179877758026123, "learning_rate": 6.963480661944335e-07, "loss": 0.042205460369586945, "step": 6704 }, { "epoch": 1.7406542056074765, "grad_norm": 1.8214377164840698, "learning_rate": 6.959617518231448e-07, "loss": 0.05834716558456421, "step": 6705 }, { "epoch": 1.7409138110072688, "grad_norm": 1.3265619277954102, "learning_rate": 6.955758005587876e-07, "loss": 0.09788170456886292, "step": 6706 }, { "epoch": 1.7411734164070611, "grad_norm": 2.6079909801483154, "learning_rate": 6.951902124695885e-07, "loss": 0.04799748957157135, "step": 6707 }, { "epoch": 1.7414330218068534, "grad_norm": 1.445577621459961, "learning_rate": 6.948049876237116e-07, "loss": 0.09397874027490616, "step": 6708 }, { "epoch": 1.7416926272066457, "grad_norm": 2.5251810550689697, "learning_rate": 6.944201260892544e-07, "loss": 0.08459730446338654, "step": 6709 }, { "epoch": 1.741952232606438, "grad_norm": 1.484876036643982, "learning_rate": 6.940356279342524e-07, "loss": 0.09599336981773376, "step": 6710 }, { "epoch": 1.7422118380062306, "grad_norm": 1.6490875482559204, "learning_rate": 6.936514932266763e-07, "loss": 0.10765679180622101, "step": 6711 }, { "epoch": 1.742471443406023, "grad_norm": 2.1627230644226074, "learning_rate": 6.932677220344321e-07, "loss": 0.06480522453784943, "step": 6712 }, { "epoch": 1.7427310488058152, "grad_norm": 2.4225730895996094, "learning_rate": 6.928843144253619e-07, "loss": 0.04335001856088638, "step": 6713 }, { "epoch": 1.7429906542056075, "grad_norm": 1.7856563329696655, "learning_rate": 6.925012704672426e-07, "loss": 0.03954897075891495, "step": 6714 }, { "epoch": 1.7432502596053998, "grad_norm": 2.4557113647460938, "learning_rate": 6.921185902277882e-07, "loss": 0.06270758807659149, "step": 6715 }, { "epoch": 1.7435098650051921, "grad_norm": 1.7899584770202637, "learning_rate": 6.917362737746487e-07, "loss": 0.04174809157848358, "step": 6716 }, { "epoch": 1.7437694704049844, "grad_norm": 1.982311725616455, "learning_rate": 6.913543211754075e-07, "loss": 0.05619465187191963, "step": 6717 }, { "epoch": 1.7440290758047767, "grad_norm": 1.8666962385177612, "learning_rate": 6.909727324975854e-07, "loss": 0.07082945108413696, "step": 6718 }, { "epoch": 1.744288681204569, "grad_norm": 1.393868327140808, "learning_rate": 6.905915078086392e-07, "loss": 0.1044861450791359, "step": 6719 }, { "epoch": 1.7445482866043613, "grad_norm": 1.750421166419983, "learning_rate": 6.902106471759595e-07, "loss": 0.07092368602752686, "step": 6720 }, { "epoch": 1.7448078920041536, "grad_norm": 1.8999699354171753, "learning_rate": 6.898301506668749e-07, "loss": 0.0868811309337616, "step": 6721 }, { "epoch": 1.745067497403946, "grad_norm": 2.1933400630950928, "learning_rate": 6.894500183486478e-07, "loss": 0.03433830291032791, "step": 6722 }, { "epoch": 1.7453271028037385, "grad_norm": 2.6021833419799805, "learning_rate": 6.890702502884775e-07, "loss": 0.06651656329631805, "step": 6723 }, { "epoch": 1.7455867082035308, "grad_norm": 2.9075927734375, "learning_rate": 6.886908465534975e-07, "loss": 0.067429319024086, "step": 6724 }, { "epoch": 1.745846313603323, "grad_norm": 3.154264211654663, "learning_rate": 6.883118072107783e-07, "loss": 0.04692567512392998, "step": 6725 }, { "epoch": 1.7461059190031154, "grad_norm": 2.7630677223205566, "learning_rate": 6.879331323273253e-07, "loss": 0.02591950073838234, "step": 6726 }, { "epoch": 1.7463655244029077, "grad_norm": 3.021310567855835, "learning_rate": 6.875548219700792e-07, "loss": 0.0348600372672081, "step": 6727 }, { "epoch": 1.7466251298027, "grad_norm": 2.5153534412384033, "learning_rate": 6.871768762059167e-07, "loss": 0.06164387986063957, "step": 6728 }, { "epoch": 1.7468847352024923, "grad_norm": 2.7634835243225098, "learning_rate": 6.867992951016506e-07, "loss": 0.0920405238866806, "step": 6729 }, { "epoch": 1.7471443406022846, "grad_norm": 1.7562124729156494, "learning_rate": 6.864220787240275e-07, "loss": 0.06941434741020203, "step": 6730 }, { "epoch": 1.747403946002077, "grad_norm": 1.091983437538147, "learning_rate": 6.860452271397317e-07, "loss": 0.10862623155117035, "step": 6731 }, { "epoch": 1.7476635514018692, "grad_norm": 1.608365774154663, "learning_rate": 6.856687404153814e-07, "loss": 0.0841442197561264, "step": 6732 }, { "epoch": 1.7479231568016615, "grad_norm": 1.3580505847930908, "learning_rate": 6.852926186175305e-07, "loss": 0.10501915216445923, "step": 6733 }, { "epoch": 1.7481827622014539, "grad_norm": 2.150991201400757, "learning_rate": 6.849168618126695e-07, "loss": 0.07089859247207642, "step": 6734 }, { "epoch": 1.7484423676012462, "grad_norm": 2.3649158477783203, "learning_rate": 6.845414700672231e-07, "loss": 0.08535492420196533, "step": 6735 }, { "epoch": 1.7487019730010385, "grad_norm": 1.9011478424072266, "learning_rate": 6.841664434475518e-07, "loss": 0.0280052050948143, "step": 6736 }, { "epoch": 1.7489615784008308, "grad_norm": 1.962438941001892, "learning_rate": 6.837917820199525e-07, "loss": 0.05690060928463936, "step": 6737 }, { "epoch": 1.749221183800623, "grad_norm": 3.5733349323272705, "learning_rate": 6.834174858506555e-07, "loss": 0.06985000520944595, "step": 6738 }, { "epoch": 1.7494807892004154, "grad_norm": 1.0087890625, "learning_rate": 6.830435550058297e-07, "loss": 0.10416610538959503, "step": 6739 }, { "epoch": 1.7497403946002077, "grad_norm": 2.248413562774658, "learning_rate": 6.826699895515751e-07, "loss": 0.06272809207439423, "step": 6740 }, { "epoch": 1.75, "grad_norm": 1.4085458517074585, "learning_rate": 6.822967895539311e-07, "loss": 0.08582157641649246, "step": 6741 }, { "epoch": 1.7502596053997923, "grad_norm": 1.447259783744812, "learning_rate": 6.81923955078871e-07, "loss": 0.04279697686433792, "step": 6742 }, { "epoch": 1.7505192107995846, "grad_norm": 2.4621787071228027, "learning_rate": 6.815514861923028e-07, "loss": 0.057595863938331604, "step": 6743 }, { "epoch": 1.750778816199377, "grad_norm": 2.075063943862915, "learning_rate": 6.811793829600707e-07, "loss": 0.08025684952735901, "step": 6744 }, { "epoch": 1.7510384215991692, "grad_norm": 1.3365131616592407, "learning_rate": 6.808076454479534e-07, "loss": 0.0846622884273529, "step": 6745 }, { "epoch": 1.7512980269989615, "grad_norm": 1.0469220876693726, "learning_rate": 6.80436273721667e-07, "loss": 0.08588080108165741, "step": 6746 }, { "epoch": 1.7515576323987538, "grad_norm": 1.22266685962677, "learning_rate": 6.800652678468604e-07, "loss": 0.07526945322751999, "step": 6747 }, { "epoch": 1.7518172377985461, "grad_norm": 1.956801176071167, "learning_rate": 6.79694627889119e-07, "loss": 0.05045241862535477, "step": 6748 }, { "epoch": 1.7520768431983385, "grad_norm": 1.1451102495193481, "learning_rate": 6.793243539139636e-07, "loss": 0.10870543122291565, "step": 6749 }, { "epoch": 1.7523364485981308, "grad_norm": 3.554368495941162, "learning_rate": 6.789544459868508e-07, "loss": 0.06160247325897217, "step": 6750 }, { "epoch": 1.752596053997923, "grad_norm": 3.9537508487701416, "learning_rate": 6.785849041731705e-07, "loss": 0.08631541579961777, "step": 6751 }, { "epoch": 1.7528556593977154, "grad_norm": 2.119558811187744, "learning_rate": 6.782157285382508e-07, "loss": 0.03655324876308441, "step": 6752 }, { "epoch": 1.7531152647975077, "grad_norm": 2.6231789588928223, "learning_rate": 6.778469191473519e-07, "loss": 0.08742457628250122, "step": 6753 }, { "epoch": 1.7533748701973, "grad_norm": 1.140571117401123, "learning_rate": 6.774784760656725e-07, "loss": 0.10981402546167374, "step": 6754 }, { "epoch": 1.7536344755970923, "grad_norm": 3.1142630577087402, "learning_rate": 6.771103993583437e-07, "loss": 0.034836724400520325, "step": 6755 }, { "epoch": 1.7538940809968846, "grad_norm": 2.7945640087127686, "learning_rate": 6.767426890904332e-07, "loss": 0.048904214054346085, "step": 6756 }, { "epoch": 1.754153686396677, "grad_norm": 1.703900694847107, "learning_rate": 6.763753453269445e-07, "loss": 0.08841133117675781, "step": 6757 }, { "epoch": 1.7544132917964692, "grad_norm": 1.2633640766143799, "learning_rate": 6.760083681328147e-07, "loss": 0.09704241156578064, "step": 6758 }, { "epoch": 1.7546728971962615, "grad_norm": 1.5385980606079102, "learning_rate": 6.756417575729171e-07, "loss": 0.08315491676330566, "step": 6759 }, { "epoch": 1.754932502596054, "grad_norm": 2.276386260986328, "learning_rate": 6.752755137120606e-07, "loss": 0.08129709959030151, "step": 6760 }, { "epoch": 1.7551921079958464, "grad_norm": 2.575270652770996, "learning_rate": 6.749096366149881e-07, "loss": 0.08837351948022842, "step": 6761 }, { "epoch": 1.7554517133956387, "grad_norm": 2.7022275924682617, "learning_rate": 6.745441263463791e-07, "loss": 0.04336126893758774, "step": 6762 }, { "epoch": 1.755711318795431, "grad_norm": 2.689450979232788, "learning_rate": 6.741789829708467e-07, "loss": 0.05588030070066452, "step": 6763 }, { "epoch": 1.7559709241952233, "grad_norm": 3.1504266262054443, "learning_rate": 6.738142065529396e-07, "loss": 0.0449778288602829, "step": 6764 }, { "epoch": 1.7562305295950156, "grad_norm": 2.1181678771972656, "learning_rate": 6.73449797157143e-07, "loss": 0.07213133573532104, "step": 6765 }, { "epoch": 1.7564901349948079, "grad_norm": 1.4374903440475464, "learning_rate": 6.730857548478753e-07, "loss": 0.05937886983156204, "step": 6766 }, { "epoch": 1.7567497403946002, "grad_norm": 2.707420587539673, "learning_rate": 6.727220796894913e-07, "loss": 0.09523209184408188, "step": 6767 }, { "epoch": 1.7570093457943925, "grad_norm": 2.2065787315368652, "learning_rate": 6.723587717462801e-07, "loss": 0.08120740950107574, "step": 6768 }, { "epoch": 1.7572689511941848, "grad_norm": 1.9294394254684448, "learning_rate": 6.719958310824662e-07, "loss": 0.06232085078954697, "step": 6769 }, { "epoch": 1.757528556593977, "grad_norm": 2.4783849716186523, "learning_rate": 6.7163325776221e-07, "loss": 0.08272407948970795, "step": 6770 }, { "epoch": 1.7577881619937694, "grad_norm": 2.529876232147217, "learning_rate": 6.712710518496049e-07, "loss": 0.0815982073545456, "step": 6771 }, { "epoch": 1.758047767393562, "grad_norm": 1.808303952217102, "learning_rate": 6.709092134086813e-07, "loss": 0.05509010702371597, "step": 6772 }, { "epoch": 1.7583073727933543, "grad_norm": 1.3213131427764893, "learning_rate": 6.705477425034046e-07, "loss": 0.1107335165143013, "step": 6773 }, { "epoch": 1.7585669781931466, "grad_norm": 1.117804765701294, "learning_rate": 6.701866391976733e-07, "loss": 0.08955998718738556, "step": 6774 }, { "epoch": 1.7588265835929389, "grad_norm": 3.094834327697754, "learning_rate": 6.698259035553228e-07, "loss": 0.08713790774345398, "step": 6775 }, { "epoch": 1.7590861889927312, "grad_norm": 3.1140079498291016, "learning_rate": 6.694655356401226e-07, "loss": 0.09874395281076431, "step": 6776 }, { "epoch": 1.7593457943925235, "grad_norm": 1.315736174583435, "learning_rate": 6.691055355157783e-07, "loss": 0.10204029083251953, "step": 6777 }, { "epoch": 1.7596053997923158, "grad_norm": 2.2207508087158203, "learning_rate": 6.687459032459289e-07, "loss": 0.08084223419427872, "step": 6778 }, { "epoch": 1.759865005192108, "grad_norm": 2.8606021404266357, "learning_rate": 6.683866388941495e-07, "loss": 0.07560206949710846, "step": 6779 }, { "epoch": 1.7601246105919004, "grad_norm": 2.8981637954711914, "learning_rate": 6.680277425239496e-07, "loss": 0.06922958046197891, "step": 6780 }, { "epoch": 1.7603842159916927, "grad_norm": 1.7232426404953003, "learning_rate": 6.676692141987744e-07, "loss": 0.051877737045288086, "step": 6781 }, { "epoch": 1.760643821391485, "grad_norm": 1.503714919090271, "learning_rate": 6.673110539820023e-07, "loss": 0.08590681850910187, "step": 6782 }, { "epoch": 1.7609034267912773, "grad_norm": 2.5759425163269043, "learning_rate": 6.669532619369493e-07, "loss": 0.06246725097298622, "step": 6783 }, { "epoch": 1.7611630321910696, "grad_norm": 1.3105722665786743, "learning_rate": 6.665958381268634e-07, "loss": 0.10756213963031769, "step": 6784 }, { "epoch": 1.761422637590862, "grad_norm": 2.640503168106079, "learning_rate": 6.662387826149303e-07, "loss": 0.07439540326595306, "step": 6785 }, { "epoch": 1.7616822429906542, "grad_norm": 1.6399656534194946, "learning_rate": 6.658820954642687e-07, "loss": 0.0770924761891365, "step": 6786 }, { "epoch": 1.7619418483904465, "grad_norm": 1.0856131315231323, "learning_rate": 6.655257767379321e-07, "loss": 0.09788313508033752, "step": 6787 }, { "epoch": 1.7622014537902388, "grad_norm": 2.5860018730163574, "learning_rate": 6.651698264989105e-07, "loss": 0.0708560049533844, "step": 6788 }, { "epoch": 1.7624610591900312, "grad_norm": 1.2545156478881836, "learning_rate": 6.648142448101274e-07, "loss": 0.09572702646255493, "step": 6789 }, { "epoch": 1.7627206645898235, "grad_norm": 2.7240371704101562, "learning_rate": 6.644590317344409e-07, "loss": 0.03711908683180809, "step": 6790 }, { "epoch": 1.7629802699896158, "grad_norm": 1.7552515268325806, "learning_rate": 6.641041873346453e-07, "loss": 0.08511319756507874, "step": 6791 }, { "epoch": 1.763239875389408, "grad_norm": 1.92020845413208, "learning_rate": 6.637497116734686e-07, "loss": 0.04667174816131592, "step": 6792 }, { "epoch": 1.7634994807892004, "grad_norm": 1.220494270324707, "learning_rate": 6.633956048135741e-07, "loss": 0.09389710426330566, "step": 6793 }, { "epoch": 1.7637590861889927, "grad_norm": 1.4062057733535767, "learning_rate": 6.630418668175598e-07, "loss": 0.08083036541938782, "step": 6794 }, { "epoch": 1.764018691588785, "grad_norm": 1.7017079591751099, "learning_rate": 6.626884977479582e-07, "loss": 0.029512468725442886, "step": 6795 }, { "epoch": 1.7642782969885773, "grad_norm": 2.216965913772583, "learning_rate": 6.623354976672376e-07, "loss": 0.04895754158496857, "step": 6796 }, { "epoch": 1.7645379023883696, "grad_norm": 1.9454326629638672, "learning_rate": 6.619828666377998e-07, "loss": 0.06927341222763062, "step": 6797 }, { "epoch": 1.764797507788162, "grad_norm": 1.2842838764190674, "learning_rate": 6.616306047219813e-07, "loss": 0.10648301243782043, "step": 6798 }, { "epoch": 1.7650571131879542, "grad_norm": 2.0504074096679688, "learning_rate": 6.612787119820552e-07, "loss": 0.08142244815826416, "step": 6799 }, { "epoch": 1.7653167185877465, "grad_norm": 2.5941734313964844, "learning_rate": 6.609271884802272e-07, "loss": 0.0973961353302002, "step": 6800 }, { "epoch": 1.7655763239875388, "grad_norm": 2.707540273666382, "learning_rate": 6.605760342786391e-07, "loss": 0.05751977115869522, "step": 6801 }, { "epoch": 1.7658359293873311, "grad_norm": 2.077547550201416, "learning_rate": 6.602252494393663e-07, "loss": 0.0660313293337822, "step": 6802 }, { "epoch": 1.7660955347871234, "grad_norm": 2.9104461669921875, "learning_rate": 6.598748340244198e-07, "loss": 0.05541219562292099, "step": 6803 }, { "epoch": 1.7663551401869158, "grad_norm": 3.9226856231689453, "learning_rate": 6.595247880957459e-07, "loss": 0.058212026953697205, "step": 6804 }, { "epoch": 1.766614745586708, "grad_norm": 1.2883461713790894, "learning_rate": 6.59175111715223e-07, "loss": 0.10602496564388275, "step": 6805 }, { "epoch": 1.7668743509865004, "grad_norm": 2.1173205375671387, "learning_rate": 6.588258049446673e-07, "loss": 0.06992203742265701, "step": 6806 }, { "epoch": 1.7671339563862927, "grad_norm": 3.283149480819702, "learning_rate": 6.584768678458274e-07, "loss": 0.06824126839637756, "step": 6807 }, { "epoch": 1.7673935617860852, "grad_norm": 1.533017873764038, "learning_rate": 6.58128300480388e-07, "loss": 0.09063491970300674, "step": 6808 }, { "epoch": 1.7676531671858775, "grad_norm": 1.493021011352539, "learning_rate": 6.577801029099676e-07, "loss": 0.08813071250915527, "step": 6809 }, { "epoch": 1.7679127725856698, "grad_norm": 2.446324348449707, "learning_rate": 6.574322751961193e-07, "loss": 0.07024619728326797, "step": 6810 }, { "epoch": 1.7681723779854621, "grad_norm": 7.9827446937561035, "learning_rate": 6.570848174003316e-07, "loss": 0.06007272005081177, "step": 6811 }, { "epoch": 1.7684319833852544, "grad_norm": 1.5375152826309204, "learning_rate": 6.567377295840267e-07, "loss": 0.08290845900774002, "step": 6812 }, { "epoch": 1.7686915887850467, "grad_norm": 1.7771474123001099, "learning_rate": 6.563910118085616e-07, "loss": 0.053741663694381714, "step": 6813 }, { "epoch": 1.768951194184839, "grad_norm": 2.0918772220611572, "learning_rate": 6.560446641352288e-07, "loss": 0.07074860483407974, "step": 6814 }, { "epoch": 1.7692107995846313, "grad_norm": 1.5403096675872803, "learning_rate": 6.556986866252535e-07, "loss": 0.09438635408878326, "step": 6815 }, { "epoch": 1.7694704049844237, "grad_norm": 1.4712085723876953, "learning_rate": 6.553530793397975e-07, "loss": 0.0968974232673645, "step": 6816 }, { "epoch": 1.769730010384216, "grad_norm": 2.198401689529419, "learning_rate": 6.550078423399565e-07, "loss": 0.09023450314998627, "step": 6817 }, { "epoch": 1.7699896157840083, "grad_norm": 1.2363752126693726, "learning_rate": 6.546629756867594e-07, "loss": 0.10325441509485245, "step": 6818 }, { "epoch": 1.7702492211838006, "grad_norm": 2.108210802078247, "learning_rate": 6.543184794411715e-07, "loss": 0.05362284183502197, "step": 6819 }, { "epoch": 1.770508826583593, "grad_norm": 1.6635135412216187, "learning_rate": 6.539743536640921e-07, "loss": 0.0759018063545227, "step": 6820 }, { "epoch": 1.7707684319833854, "grad_norm": 3.2203094959259033, "learning_rate": 6.536305984163538e-07, "loss": 0.07429471611976624, "step": 6821 }, { "epoch": 1.7710280373831777, "grad_norm": 1.4844856262207031, "learning_rate": 6.532872137587254e-07, "loss": 0.07961653172969818, "step": 6822 }, { "epoch": 1.77128764278297, "grad_norm": 2.1787161827087402, "learning_rate": 6.529441997519091e-07, "loss": 0.07789655774831772, "step": 6823 }, { "epoch": 1.7715472481827623, "grad_norm": 2.0793933868408203, "learning_rate": 6.526015564565423e-07, "loss": 0.08765949308872223, "step": 6824 }, { "epoch": 1.7718068535825546, "grad_norm": 1.0421682596206665, "learning_rate": 6.522592839331962e-07, "loss": 0.11369045823812485, "step": 6825 }, { "epoch": 1.772066458982347, "grad_norm": 1.6085760593414307, "learning_rate": 6.519173822423768e-07, "loss": 0.07594103366136551, "step": 6826 }, { "epoch": 1.7723260643821392, "grad_norm": 1.3187516927719116, "learning_rate": 6.51575851444525e-07, "loss": 0.07833360135555267, "step": 6827 }, { "epoch": 1.7725856697819315, "grad_norm": 2.0802481174468994, "learning_rate": 6.512346916000144e-07, "loss": 0.06445291638374329, "step": 6828 }, { "epoch": 1.7728452751817239, "grad_norm": 2.1101536750793457, "learning_rate": 6.508939027691548e-07, "loss": 0.07484032213687897, "step": 6829 }, { "epoch": 1.7731048805815162, "grad_norm": 2.8635549545288086, "learning_rate": 6.505534850121904e-07, "loss": 0.03246684372425079, "step": 6830 }, { "epoch": 1.7733644859813085, "grad_norm": 1.086414098739624, "learning_rate": 6.502134383892993e-07, "loss": 0.1111370399594307, "step": 6831 }, { "epoch": 1.7736240913811008, "grad_norm": 2.1828739643096924, "learning_rate": 6.498737629605933e-07, "loss": 0.06089155375957489, "step": 6832 }, { "epoch": 1.773883696780893, "grad_norm": 1.993422031402588, "learning_rate": 6.49534458786119e-07, "loss": 0.07883479446172714, "step": 6833 }, { "epoch": 1.7741433021806854, "grad_norm": 2.6747260093688965, "learning_rate": 6.491955259258584e-07, "loss": 0.05427651107311249, "step": 6834 }, { "epoch": 1.7744029075804777, "grad_norm": 1.8718401193618774, "learning_rate": 6.488569644397272e-07, "loss": 0.06123126298189163, "step": 6835 }, { "epoch": 1.77466251298027, "grad_norm": 1.9310057163238525, "learning_rate": 6.485187743875743e-07, "loss": 0.07049312442541122, "step": 6836 }, { "epoch": 1.7749221183800623, "grad_norm": 2.273831367492676, "learning_rate": 6.481809558291848e-07, "loss": 0.05533549189567566, "step": 6837 }, { "epoch": 1.7751817237798546, "grad_norm": 1.0465409755706787, "learning_rate": 6.478435088242768e-07, "loss": 0.09709178656339645, "step": 6838 }, { "epoch": 1.775441329179647, "grad_norm": 1.2181705236434937, "learning_rate": 6.475064334325039e-07, "loss": 0.10625246912240982, "step": 6839 }, { "epoch": 1.7757009345794392, "grad_norm": 0.8825112581253052, "learning_rate": 6.471697297134526e-07, "loss": 0.02645733579993248, "step": 6840 }, { "epoch": 1.7759605399792315, "grad_norm": 1.4040952920913696, "learning_rate": 6.468333977266446e-07, "loss": 0.09774988144636154, "step": 6841 }, { "epoch": 1.7762201453790238, "grad_norm": 1.1606167554855347, "learning_rate": 6.464974375315357e-07, "loss": 0.0811973437666893, "step": 6842 }, { "epoch": 1.7764797507788161, "grad_norm": 2.27722430229187, "learning_rate": 6.461618491875163e-07, "loss": 0.08332516998052597, "step": 6843 }, { "epoch": 1.7767393561786085, "grad_norm": 3.461724281311035, "learning_rate": 6.4582663275391e-07, "loss": 0.051186613738536835, "step": 6844 }, { "epoch": 1.7769989615784008, "grad_norm": 1.9881644248962402, "learning_rate": 6.454917882899764e-07, "loss": 0.08368700742721558, "step": 6845 }, { "epoch": 1.777258566978193, "grad_norm": 2.4511475563049316, "learning_rate": 6.451573158549071e-07, "loss": 0.05606713518500328, "step": 6846 }, { "epoch": 1.7775181723779854, "grad_norm": 4.165424823760986, "learning_rate": 6.448232155078303e-07, "loss": 0.0566912442445755, "step": 6847 }, { "epoch": 1.7777777777777777, "grad_norm": 2.355116128921509, "learning_rate": 6.44489487307807e-07, "loss": 0.05754171311855316, "step": 6848 }, { "epoch": 1.77803738317757, "grad_norm": 1.807234764099121, "learning_rate": 6.441561313138321e-07, "loss": 0.09559624642133713, "step": 6849 }, { "epoch": 1.7782969885773623, "grad_norm": 2.1762733459472656, "learning_rate": 6.438231475848361e-07, "loss": 0.0288666021078825, "step": 6850 }, { "epoch": 1.7785565939771546, "grad_norm": 1.4719374179840088, "learning_rate": 6.434905361796825e-07, "loss": 0.08596889674663544, "step": 6851 }, { "epoch": 1.778816199376947, "grad_norm": 1.841166377067566, "learning_rate": 6.431582971571693e-07, "loss": 0.08927446603775024, "step": 6852 }, { "epoch": 1.7790758047767392, "grad_norm": 2.5959460735321045, "learning_rate": 6.428264305760288e-07, "loss": 0.08031413704156876, "step": 6853 }, { "epoch": 1.7793354101765315, "grad_norm": 4.18360710144043, "learning_rate": 6.424949364949279e-07, "loss": 0.06164907291531563, "step": 6854 }, { "epoch": 1.7795950155763238, "grad_norm": 2.5247957706451416, "learning_rate": 6.421638149724661e-07, "loss": 0.048007141798734665, "step": 6855 }, { "epoch": 1.7798546209761164, "grad_norm": 1.0152771472930908, "learning_rate": 6.418330660671794e-07, "loss": 0.0952807068824768, "step": 6856 }, { "epoch": 1.7801142263759087, "grad_norm": 2.3261048793792725, "learning_rate": 6.415026898375355e-07, "loss": 0.058951448649168015, "step": 6857 }, { "epoch": 1.780373831775701, "grad_norm": 2.637784242630005, "learning_rate": 6.411726863419386e-07, "loss": 0.07088947296142578, "step": 6858 }, { "epoch": 1.7806334371754933, "grad_norm": 2.615739107131958, "learning_rate": 6.408430556387243e-07, "loss": 0.06122802942991257, "step": 6859 }, { "epoch": 1.7808930425752856, "grad_norm": 2.08597731590271, "learning_rate": 6.405137977861643e-07, "loss": 0.08618205785751343, "step": 6860 }, { "epoch": 1.7811526479750779, "grad_norm": 2.8719990253448486, "learning_rate": 6.401849128424645e-07, "loss": 0.0550522580742836, "step": 6861 }, { "epoch": 1.7814122533748702, "grad_norm": 1.1546821594238281, "learning_rate": 6.398564008657638e-07, "loss": 0.03712402656674385, "step": 6862 }, { "epoch": 1.7816718587746625, "grad_norm": 1.2787768840789795, "learning_rate": 6.395282619141354e-07, "loss": 0.08819323778152466, "step": 6863 }, { "epoch": 1.7819314641744548, "grad_norm": 1.1288049221038818, "learning_rate": 6.392004960455867e-07, "loss": 0.10242517292499542, "step": 6864 }, { "epoch": 1.782191069574247, "grad_norm": 2.5185282230377197, "learning_rate": 6.388731033180594e-07, "loss": 0.03743383288383484, "step": 6865 }, { "epoch": 1.7824506749740394, "grad_norm": 3.28497314453125, "learning_rate": 6.385460837894296e-07, "loss": 0.056621845811605453, "step": 6866 }, { "epoch": 1.7827102803738317, "grad_norm": 1.5934131145477295, "learning_rate": 6.382194375175058e-07, "loss": 0.06666503846645355, "step": 6867 }, { "epoch": 1.7829698857736243, "grad_norm": 1.2231892347335815, "learning_rate": 6.378931645600321e-07, "loss": 0.09877817332744598, "step": 6868 }, { "epoch": 1.7832294911734166, "grad_norm": 2.8571183681488037, "learning_rate": 6.375672649746861e-07, "loss": 0.042253680527210236, "step": 6869 }, { "epoch": 1.7834890965732089, "grad_norm": 2.018446922302246, "learning_rate": 6.372417388190796e-07, "loss": 0.0644407793879509, "step": 6870 }, { "epoch": 1.7837487019730012, "grad_norm": 1.3613559007644653, "learning_rate": 6.369165861507577e-07, "loss": 0.09083347022533417, "step": 6871 }, { "epoch": 1.7840083073727935, "grad_norm": 2.701143741607666, "learning_rate": 6.365918070271999e-07, "loss": 0.07135777175426483, "step": 6872 }, { "epoch": 1.7842679127725858, "grad_norm": 1.913231372833252, "learning_rate": 6.362674015058207e-07, "loss": 0.058384090662002563, "step": 6873 }, { "epoch": 1.784527518172378, "grad_norm": 1.78445565700531, "learning_rate": 6.359433696439665e-07, "loss": 0.029821768403053284, "step": 6874 }, { "epoch": 1.7847871235721704, "grad_norm": 1.8699487447738647, "learning_rate": 6.356197114989187e-07, "loss": 0.06305938214063644, "step": 6875 }, { "epoch": 1.7850467289719627, "grad_norm": 1.592635989189148, "learning_rate": 6.352964271278934e-07, "loss": 0.0804213434457779, "step": 6876 }, { "epoch": 1.785306334371755, "grad_norm": 1.570473313331604, "learning_rate": 6.349735165880393e-07, "loss": 0.07277136296033859, "step": 6877 }, { "epoch": 1.7855659397715473, "grad_norm": 1.429251790046692, "learning_rate": 6.346509799364402e-07, "loss": 0.09320482611656189, "step": 6878 }, { "epoch": 1.7858255451713396, "grad_norm": 1.9676730632781982, "learning_rate": 6.343288172301125e-07, "loss": 0.0885305255651474, "step": 6879 }, { "epoch": 1.786085150571132, "grad_norm": 2.3198890686035156, "learning_rate": 6.340070285260071e-07, "loss": 0.06537249684333801, "step": 6880 }, { "epoch": 1.7863447559709242, "grad_norm": 2.8361122608184814, "learning_rate": 6.3368561388101e-07, "loss": 0.03505956754088402, "step": 6881 }, { "epoch": 1.7866043613707165, "grad_norm": 4.318500518798828, "learning_rate": 6.333645733519387e-07, "loss": 0.0808529481291771, "step": 6882 }, { "epoch": 1.7868639667705088, "grad_norm": 3.369521379470825, "learning_rate": 6.330439069955461e-07, "loss": 0.06298209726810455, "step": 6883 }, { "epoch": 1.7871235721703012, "grad_norm": 3.2942028045654297, "learning_rate": 6.327236148685192e-07, "loss": 0.032507188618183136, "step": 6884 }, { "epoch": 1.7873831775700935, "grad_norm": 2.589446544647217, "learning_rate": 6.32403697027478e-07, "loss": 0.05055747926235199, "step": 6885 }, { "epoch": 1.7876427829698858, "grad_norm": 2.5861380100250244, "learning_rate": 6.320841535289765e-07, "loss": 0.07730144262313843, "step": 6886 }, { "epoch": 1.787902388369678, "grad_norm": 2.6414783000946045, "learning_rate": 6.317649844295029e-07, "loss": 0.04432031884789467, "step": 6887 }, { "epoch": 1.7881619937694704, "grad_norm": 2.8786962032318115, "learning_rate": 6.314461897854787e-07, "loss": 0.04946950823068619, "step": 6888 }, { "epoch": 1.7884215991692627, "grad_norm": 1.7696456909179688, "learning_rate": 6.311277696532603e-07, "loss": 0.08246845752000809, "step": 6889 }, { "epoch": 1.788681204569055, "grad_norm": 2.5050222873687744, "learning_rate": 6.308097240891359e-07, "loss": 0.05060822516679764, "step": 6890 }, { "epoch": 1.7889408099688473, "grad_norm": 2.991877555847168, "learning_rate": 6.304920531493292e-07, "loss": 0.09808081388473511, "step": 6891 }, { "epoch": 1.7892004153686396, "grad_norm": 1.5138369798660278, "learning_rate": 6.301747568899975e-07, "loss": 0.08051414787769318, "step": 6892 }, { "epoch": 1.789460020768432, "grad_norm": 2.284614324569702, "learning_rate": 6.298578353672312e-07, "loss": 0.05434989929199219, "step": 6893 }, { "epoch": 1.7897196261682242, "grad_norm": 3.365846633911133, "learning_rate": 6.29541288637055e-07, "loss": 0.07072971761226654, "step": 6894 }, { "epoch": 1.7899792315680165, "grad_norm": 1.7985990047454834, "learning_rate": 6.292251167554264e-07, "loss": 0.06632521003484726, "step": 6895 }, { "epoch": 1.7902388369678088, "grad_norm": 1.3040106296539307, "learning_rate": 6.289093197782378e-07, "loss": 0.09725455939769745, "step": 6896 }, { "epoch": 1.7904984423676011, "grad_norm": 1.842320203781128, "learning_rate": 6.285938977613156e-07, "loss": 0.09028337150812149, "step": 6897 }, { "epoch": 1.7907580477673934, "grad_norm": 2.4907524585723877, "learning_rate": 6.282788507604179e-07, "loss": 0.039500992745161057, "step": 6898 }, { "epoch": 1.7910176531671858, "grad_norm": 3.2450623512268066, "learning_rate": 6.279641788312388e-07, "loss": 0.03788290172815323, "step": 6899 }, { "epoch": 1.791277258566978, "grad_norm": 3.738107442855835, "learning_rate": 6.276498820294043e-07, "loss": 0.08189387619495392, "step": 6900 }, { "epoch": 1.7915368639667704, "grad_norm": 0.8257058262825012, "learning_rate": 6.273359604104756e-07, "loss": 0.051484014838933945, "step": 6901 }, { "epoch": 1.7917964693665627, "grad_norm": 2.192553758621216, "learning_rate": 6.270224140299466e-07, "loss": 0.06053538620471954, "step": 6902 }, { "epoch": 1.792056074766355, "grad_norm": 2.9119160175323486, "learning_rate": 6.267092429432446e-07, "loss": 0.011499208398163319, "step": 6903 }, { "epoch": 1.7923156801661475, "grad_norm": 1.728197693824768, "learning_rate": 6.26396447205732e-07, "loss": 0.06568308919668198, "step": 6904 }, { "epoch": 1.7925752855659398, "grad_norm": 3.458439826965332, "learning_rate": 6.260840268727032e-07, "loss": 0.03945091739296913, "step": 6905 }, { "epoch": 1.7928348909657321, "grad_norm": 2.8227500915527344, "learning_rate": 6.257719819993873e-07, "loss": 0.06031573563814163, "step": 6906 }, { "epoch": 1.7930944963655244, "grad_norm": 2.3386104106903076, "learning_rate": 6.254603126409466e-07, "loss": 0.06913809478282928, "step": 6907 }, { "epoch": 1.7933541017653167, "grad_norm": 0.9679714441299438, "learning_rate": 6.251490188524768e-07, "loss": 0.1059197336435318, "step": 6908 }, { "epoch": 1.793613707165109, "grad_norm": 3.536191701889038, "learning_rate": 6.248381006890083e-07, "loss": 0.030856752768158913, "step": 6909 }, { "epoch": 1.7938733125649013, "grad_norm": 2.3129849433898926, "learning_rate": 6.24527558205504e-07, "loss": 0.0478590726852417, "step": 6910 }, { "epoch": 1.7941329179646937, "grad_norm": 3.377791404724121, "learning_rate": 6.2421739145686e-07, "loss": 0.020922336727380753, "step": 6911 }, { "epoch": 1.794392523364486, "grad_norm": 2.533806085586548, "learning_rate": 6.239076004979075e-07, "loss": 0.0623868852853775, "step": 6912 }, { "epoch": 1.7946521287642783, "grad_norm": 2.677060127258301, "learning_rate": 6.235981853834102e-07, "loss": 0.07103578746318817, "step": 6913 }, { "epoch": 1.7949117341640706, "grad_norm": 3.237905979156494, "learning_rate": 6.232891461680653e-07, "loss": 0.0586816743016243, "step": 6914 }, { "epoch": 1.7951713395638629, "grad_norm": 3.2498273849487305, "learning_rate": 6.229804829065044e-07, "loss": 0.049101755023002625, "step": 6915 }, { "epoch": 1.7954309449636554, "grad_norm": 2.088059425354004, "learning_rate": 6.226721956532921e-07, "loss": 0.09510905295610428, "step": 6916 }, { "epoch": 1.7956905503634477, "grad_norm": 2.0323030948638916, "learning_rate": 6.223642844629255e-07, "loss": 0.04291682690382004, "step": 6917 }, { "epoch": 1.79595015576324, "grad_norm": 3.320683240890503, "learning_rate": 6.220567493898376e-07, "loss": 0.06664245575666428, "step": 6918 }, { "epoch": 1.7962097611630323, "grad_norm": 2.0444960594177246, "learning_rate": 6.217495904883928e-07, "loss": 0.0679267942905426, "step": 6919 }, { "epoch": 1.7964693665628246, "grad_norm": 2.6996188163757324, "learning_rate": 6.214428078128902e-07, "loss": 0.07359542697668076, "step": 6920 }, { "epoch": 1.796728971962617, "grad_norm": 1.488203763961792, "learning_rate": 6.211364014175613e-07, "loss": 0.08512649685144424, "step": 6921 }, { "epoch": 1.7969885773624092, "grad_norm": 3.4644339084625244, "learning_rate": 6.208303713565721e-07, "loss": 0.04537327587604523, "step": 6922 }, { "epoch": 1.7972481827622016, "grad_norm": 2.144676446914673, "learning_rate": 6.205247176840217e-07, "loss": 0.05741564929485321, "step": 6923 }, { "epoch": 1.7975077881619939, "grad_norm": 3.1405129432678223, "learning_rate": 6.202194404539431e-07, "loss": 0.014526870101690292, "step": 6924 }, { "epoch": 1.7977673935617862, "grad_norm": 1.6842595338821411, "learning_rate": 6.199145397203017e-07, "loss": 0.08444158732891083, "step": 6925 }, { "epoch": 1.7980269989615785, "grad_norm": 3.01084566116333, "learning_rate": 6.19610015536997e-07, "loss": 0.0401887521147728, "step": 6926 }, { "epoch": 1.7982866043613708, "grad_norm": 2.4522385597229004, "learning_rate": 6.193058679578619e-07, "loss": 0.04941173642873764, "step": 6927 }, { "epoch": 1.798546209761163, "grad_norm": 0.9774278998374939, "learning_rate": 6.190020970366636e-07, "loss": 0.1023554727435112, "step": 6928 }, { "epoch": 1.7988058151609554, "grad_norm": 1.5824648141860962, "learning_rate": 6.186987028271007e-07, "loss": 0.08573982119560242, "step": 6929 }, { "epoch": 1.7990654205607477, "grad_norm": 1.4583929777145386, "learning_rate": 6.183956853828069e-07, "loss": 0.07874573767185211, "step": 6930 }, { "epoch": 1.79932502596054, "grad_norm": 3.140401840209961, "learning_rate": 6.180930447573486e-07, "loss": 0.03707218915224075, "step": 6931 }, { "epoch": 1.7995846313603323, "grad_norm": 2.4960527420043945, "learning_rate": 6.177907810042259e-07, "loss": 0.04827365279197693, "step": 6932 }, { "epoch": 1.7998442367601246, "grad_norm": 2.572941303253174, "learning_rate": 6.174888941768721e-07, "loss": 0.05048742890357971, "step": 6933 }, { "epoch": 1.800103842159917, "grad_norm": 1.2048749923706055, "learning_rate": 6.171873843286534e-07, "loss": 0.09621763229370117, "step": 6934 }, { "epoch": 1.8003634475597092, "grad_norm": 1.5925087928771973, "learning_rate": 6.168862515128703e-07, "loss": 0.08396652340888977, "step": 6935 }, { "epoch": 1.8006230529595015, "grad_norm": 2.076932907104492, "learning_rate": 6.165854957827565e-07, "loss": 0.06110372394323349, "step": 6936 }, { "epoch": 1.8008826583592938, "grad_norm": 1.8052771091461182, "learning_rate": 6.16285117191478e-07, "loss": 0.03820008039474487, "step": 6937 }, { "epoch": 1.8011422637590861, "grad_norm": 2.8369204998016357, "learning_rate": 6.159851157921354e-07, "loss": 0.06109822914004326, "step": 6938 }, { "epoch": 1.8014018691588785, "grad_norm": 2.2876362800598145, "learning_rate": 6.156854916377618e-07, "loss": 0.10527486354112625, "step": 6939 }, { "epoch": 1.8016614745586708, "grad_norm": 1.0555171966552734, "learning_rate": 6.153862447813236e-07, "loss": 0.10668835043907166, "step": 6940 }, { "epoch": 1.801921079958463, "grad_norm": 3.5334112644195557, "learning_rate": 6.150873752757217e-07, "loss": 0.05662599951028824, "step": 6941 }, { "epoch": 1.8021806853582554, "grad_norm": 1.867287278175354, "learning_rate": 6.147888831737882e-07, "loss": 0.0647432878613472, "step": 6942 }, { "epoch": 1.8024402907580477, "grad_norm": 2.150942802429199, "learning_rate": 6.144907685282909e-07, "loss": 0.03055000677704811, "step": 6943 }, { "epoch": 1.80269989615784, "grad_norm": 2.6921942234039307, "learning_rate": 6.14193031391929e-07, "loss": 0.04932200536131859, "step": 6944 }, { "epoch": 1.8029595015576323, "grad_norm": 1.4177911281585693, "learning_rate": 6.13895671817335e-07, "loss": 0.09388893842697144, "step": 6945 }, { "epoch": 1.8032191069574246, "grad_norm": 2.110156297683716, "learning_rate": 6.135986898570765e-07, "loss": 0.07390041649341583, "step": 6946 }, { "epoch": 1.803478712357217, "grad_norm": 1.7753127813339233, "learning_rate": 6.133020855636526e-07, "loss": 0.09564930200576782, "step": 6947 }, { "epoch": 1.8037383177570092, "grad_norm": 1.8175028562545776, "learning_rate": 6.130058589894958e-07, "loss": 0.08041682839393616, "step": 6948 }, { "epoch": 1.8039979231568015, "grad_norm": 1.9325580596923828, "learning_rate": 6.127100101869725e-07, "loss": 0.07884475588798523, "step": 6949 }, { "epoch": 1.8042575285565938, "grad_norm": 2.53507399559021, "learning_rate": 6.124145392083817e-07, "loss": 0.050437334924936295, "step": 6950 }, { "epoch": 1.8045171339563861, "grad_norm": 2.8015832901000977, "learning_rate": 6.121194461059566e-07, "loss": 0.09206147491931915, "step": 6951 }, { "epoch": 1.8047767393561787, "grad_norm": 1.2467859983444214, "learning_rate": 6.118247309318625e-07, "loss": 0.09228014945983887, "step": 6952 }, { "epoch": 1.805036344755971, "grad_norm": 1.3858119249343872, "learning_rate": 6.115303937381976e-07, "loss": 0.0858096331357956, "step": 6953 }, { "epoch": 1.8052959501557633, "grad_norm": 3.1359660625457764, "learning_rate": 6.112364345769953e-07, "loss": 0.06523091346025467, "step": 6954 }, { "epoch": 1.8055555555555556, "grad_norm": 1.4268978834152222, "learning_rate": 6.1094285350022e-07, "loss": 0.08904305100440979, "step": 6955 }, { "epoch": 1.8058151609553479, "grad_norm": 2.7092604637145996, "learning_rate": 6.1064965055977e-07, "loss": 0.08647584915161133, "step": 6956 }, { "epoch": 1.8060747663551402, "grad_norm": 2.1219170093536377, "learning_rate": 6.103568258074773e-07, "loss": 0.056774917989969254, "step": 6957 }, { "epoch": 1.8063343717549325, "grad_norm": 2.081064224243164, "learning_rate": 6.100643792951065e-07, "loss": 0.05084068328142166, "step": 6958 }, { "epoch": 1.8065939771547248, "grad_norm": 1.6588605642318726, "learning_rate": 6.097723110743556e-07, "loss": 0.05656041204929352, "step": 6959 }, { "epoch": 1.8068535825545171, "grad_norm": 1.9519902467727661, "learning_rate": 6.094806211968551e-07, "loss": 0.03783261030912399, "step": 6960 }, { "epoch": 1.8071131879543094, "grad_norm": 2.339320182800293, "learning_rate": 6.091893097141695e-07, "loss": 0.0619039312005043, "step": 6961 }, { "epoch": 1.8073727933541017, "grad_norm": 3.622236490249634, "learning_rate": 6.088983766777959e-07, "loss": 0.04710721969604492, "step": 6962 }, { "epoch": 1.807632398753894, "grad_norm": 1.5853406190872192, "learning_rate": 6.086078221391644e-07, "loss": 0.07846362888813019, "step": 6963 }, { "epoch": 1.8078920041536866, "grad_norm": 2.0128796100616455, "learning_rate": 6.083176461496389e-07, "loss": 0.05602439492940903, "step": 6964 }, { "epoch": 1.8081516095534789, "grad_norm": 1.5677101612091064, "learning_rate": 6.08027848760515e-07, "loss": 0.07196571677923203, "step": 6965 }, { "epoch": 1.8084112149532712, "grad_norm": 1.14310622215271, "learning_rate": 6.077384300230233e-07, "loss": 0.1006033793091774, "step": 6966 }, { "epoch": 1.8086708203530635, "grad_norm": 2.7811360359191895, "learning_rate": 6.074493899883259e-07, "loss": 0.07547435164451599, "step": 6967 }, { "epoch": 1.8089304257528558, "grad_norm": 1.249145269393921, "learning_rate": 6.07160728707518e-07, "loss": 0.0873355120420456, "step": 6968 }, { "epoch": 1.809190031152648, "grad_norm": 1.8149702548980713, "learning_rate": 6.068724462316291e-07, "loss": 0.05717622861266136, "step": 6969 }, { "epoch": 1.8094496365524404, "grad_norm": 2.4261109828948975, "learning_rate": 6.065845426116207e-07, "loss": 0.05846180021762848, "step": 6970 }, { "epoch": 1.8097092419522327, "grad_norm": 2.363914966583252, "learning_rate": 6.062970178983869e-07, "loss": 0.04223708063364029, "step": 6971 }, { "epoch": 1.809968847352025, "grad_norm": 1.0952526330947876, "learning_rate": 6.060098721427566e-07, "loss": 0.035042084753513336, "step": 6972 }, { "epoch": 1.8102284527518173, "grad_norm": 1.4480708837509155, "learning_rate": 6.057231053954895e-07, "loss": 0.03808460012078285, "step": 6973 }, { "epoch": 1.8104880581516096, "grad_norm": 1.2139595746994019, "learning_rate": 6.054367177072803e-07, "loss": 0.09437981992959976, "step": 6974 }, { "epoch": 1.810747663551402, "grad_norm": 1.8625774383544922, "learning_rate": 6.051507091287554e-07, "loss": 0.05646289512515068, "step": 6975 }, { "epoch": 1.8110072689511942, "grad_norm": 1.2697563171386719, "learning_rate": 6.048650797104746e-07, "loss": 0.09592390060424805, "step": 6976 }, { "epoch": 1.8112668743509865, "grad_norm": 2.0719950199127197, "learning_rate": 6.045798295029305e-07, "loss": 0.046029526740312576, "step": 6977 }, { "epoch": 1.8115264797507789, "grad_norm": 1.3653497695922852, "learning_rate": 6.042949585565491e-07, "loss": 0.005002706311643124, "step": 6978 }, { "epoch": 1.8117860851505712, "grad_norm": 1.5233677625656128, "learning_rate": 6.040104669216889e-07, "loss": 0.020754750818014145, "step": 6979 }, { "epoch": 1.8120456905503635, "grad_norm": 2.222273826599121, "learning_rate": 6.037263546486413e-07, "loss": 0.0698624774813652, "step": 6980 }, { "epoch": 1.8123052959501558, "grad_norm": 2.8099496364593506, "learning_rate": 6.03442621787631e-07, "loss": 0.0636235773563385, "step": 6981 }, { "epoch": 1.812564901349948, "grad_norm": 2.333449363708496, "learning_rate": 6.031592683888158e-07, "loss": 0.06240859255194664, "step": 6982 }, { "epoch": 1.8128245067497404, "grad_norm": 1.880344271659851, "learning_rate": 6.028762945022857e-07, "loss": 0.053342826664447784, "step": 6983 }, { "epoch": 1.8130841121495327, "grad_norm": 2.0584683418273926, "learning_rate": 6.02593700178064e-07, "loss": 0.060662612318992615, "step": 6984 }, { "epoch": 1.813343717549325, "grad_norm": 1.3513511419296265, "learning_rate": 6.023114854661073e-07, "loss": 0.09508147835731506, "step": 6985 }, { "epoch": 1.8136033229491173, "grad_norm": 1.2310487031936646, "learning_rate": 6.020296504163043e-07, "loss": 0.10133812576532364, "step": 6986 }, { "epoch": 1.8138629283489096, "grad_norm": 2.6249139308929443, "learning_rate": 6.017481950784767e-07, "loss": 0.05544254183769226, "step": 6987 }, { "epoch": 1.814122533748702, "grad_norm": 3.196259021759033, "learning_rate": 6.014671195023805e-07, "loss": 0.054264068603515625, "step": 6988 }, { "epoch": 1.8143821391484942, "grad_norm": 1.3613897562026978, "learning_rate": 6.01186423737702e-07, "loss": 0.07684537768363953, "step": 6989 }, { "epoch": 1.8146417445482865, "grad_norm": 1.5581252574920654, "learning_rate": 6.009061078340631e-07, "loss": 0.07931587100028992, "step": 6990 }, { "epoch": 1.8149013499480788, "grad_norm": 1.5305687189102173, "learning_rate": 6.00626171841016e-07, "loss": 0.06613953411579132, "step": 6991 }, { "epoch": 1.8151609553478711, "grad_norm": 2.7690577507019043, "learning_rate": 6.003466158080478e-07, "loss": 0.0740012526512146, "step": 6992 }, { "epoch": 1.8154205607476634, "grad_norm": 3.3579368591308594, "learning_rate": 6.000674397845777e-07, "loss": 0.03311030566692352, "step": 6993 }, { "epoch": 1.8156801661474558, "grad_norm": 2.275456428527832, "learning_rate": 5.997886438199571e-07, "loss": 0.034566182643175125, "step": 6994 }, { "epoch": 1.815939771547248, "grad_norm": 1.6019835472106934, "learning_rate": 5.995102279634708e-07, "loss": 0.10169975459575653, "step": 6995 }, { "epoch": 1.8161993769470404, "grad_norm": 2.777693033218384, "learning_rate": 5.992321922643364e-07, "loss": 0.050400860607624054, "step": 6996 }, { "epoch": 1.8164589823468327, "grad_norm": 2.8988864421844482, "learning_rate": 5.98954536771705e-07, "loss": 0.04386585205793381, "step": 6997 }, { "epoch": 1.816718587746625, "grad_norm": 3.335421323776245, "learning_rate": 5.986772615346586e-07, "loss": 0.08231634646654129, "step": 6998 }, { "epoch": 1.8169781931464173, "grad_norm": 3.2876017093658447, "learning_rate": 5.984003666022135e-07, "loss": 0.09511074423789978, "step": 6999 }, { "epoch": 1.8172377985462098, "grad_norm": 1.9785751104354858, "learning_rate": 5.981238520233188e-07, "loss": 0.07771390676498413, "step": 7000 }, { "epoch": 1.8174974039460021, "grad_norm": 3.214491844177246, "learning_rate": 5.978477178468556e-07, "loss": 0.07166115939617157, "step": 7001 }, { "epoch": 1.8177570093457944, "grad_norm": 2.1488630771636963, "learning_rate": 5.975719641216379e-07, "loss": 0.040799736976623535, "step": 7002 }, { "epoch": 1.8180166147455867, "grad_norm": 1.412427544593811, "learning_rate": 5.972965908964129e-07, "loss": 0.10019712150096893, "step": 7003 }, { "epoch": 1.818276220145379, "grad_norm": 1.4131320714950562, "learning_rate": 5.970215982198602e-07, "loss": 0.08111979812383652, "step": 7004 }, { "epoch": 1.8185358255451713, "grad_norm": 1.3612172603607178, "learning_rate": 5.967469861405927e-07, "loss": 0.08579495549201965, "step": 7005 }, { "epoch": 1.8187954309449637, "grad_norm": 1.2490730285644531, "learning_rate": 5.964727547071548e-07, "loss": 0.08201315999031067, "step": 7006 }, { "epoch": 1.819055036344756, "grad_norm": 4.034292697906494, "learning_rate": 5.961989039680243e-07, "loss": 0.07997126877307892, "step": 7007 }, { "epoch": 1.8193146417445483, "grad_norm": 1.9792617559432983, "learning_rate": 5.959254339716127e-07, "loss": 0.039113469421863556, "step": 7008 }, { "epoch": 1.8195742471443406, "grad_norm": 1.6290733814239502, "learning_rate": 5.956523447662624e-07, "loss": 0.03599679842591286, "step": 7009 }, { "epoch": 1.8198338525441329, "grad_norm": 1.951759696006775, "learning_rate": 5.953796364002491e-07, "loss": 0.02999589778482914, "step": 7010 }, { "epoch": 1.8200934579439252, "grad_norm": 2.211460828781128, "learning_rate": 5.951073089217822e-07, "loss": 0.05785481631755829, "step": 7011 }, { "epoch": 1.8203530633437177, "grad_norm": 1.642525553703308, "learning_rate": 5.948353623790024e-07, "loss": 0.09639954566955566, "step": 7012 }, { "epoch": 1.82061266874351, "grad_norm": 1.662732720375061, "learning_rate": 5.94563796819984e-07, "loss": 0.0767025500535965, "step": 7013 }, { "epoch": 1.8208722741433023, "grad_norm": 3.4920904636383057, "learning_rate": 5.942926122927334e-07, "loss": 0.05801020562648773, "step": 7014 }, { "epoch": 1.8211318795430946, "grad_norm": 2.0800206661224365, "learning_rate": 5.940218088451894e-07, "loss": 0.027736226096749306, "step": 7015 }, { "epoch": 1.821391484942887, "grad_norm": 1.8642719984054565, "learning_rate": 5.937513865252247e-07, "loss": 0.07634906470775604, "step": 7016 }, { "epoch": 1.8216510903426792, "grad_norm": 4.260894298553467, "learning_rate": 5.934813453806434e-07, "loss": 0.02113248035311699, "step": 7017 }, { "epoch": 1.8219106957424716, "grad_norm": 1.1915100812911987, "learning_rate": 5.932116854591819e-07, "loss": 0.08907988667488098, "step": 7018 }, { "epoch": 1.8221703011422639, "grad_norm": 1.337662696838379, "learning_rate": 5.929424068085109e-07, "loss": 0.09325329959392548, "step": 7019 }, { "epoch": 1.8224299065420562, "grad_norm": 3.8806097507476807, "learning_rate": 5.926735094762325e-07, "loss": 0.06800641119480133, "step": 7020 }, { "epoch": 1.8226895119418485, "grad_norm": 1.660308599472046, "learning_rate": 5.924049935098812e-07, "loss": 0.08628081530332565, "step": 7021 }, { "epoch": 1.8229491173416408, "grad_norm": 2.0051703453063965, "learning_rate": 5.921368589569246e-07, "loss": 0.07632789760828018, "step": 7022 }, { "epoch": 1.823208722741433, "grad_norm": 1.9568829536437988, "learning_rate": 5.918691058647626e-07, "loss": 0.04762737825512886, "step": 7023 }, { "epoch": 1.8234683281412254, "grad_norm": 2.196716070175171, "learning_rate": 5.916017342807287e-07, "loss": 0.04752755165100098, "step": 7024 }, { "epoch": 1.8237279335410177, "grad_norm": 1.5156999826431274, "learning_rate": 5.913347442520865e-07, "loss": 0.05488187074661255, "step": 7025 }, { "epoch": 1.82398753894081, "grad_norm": 1.5135884284973145, "learning_rate": 5.910681358260355e-07, "loss": 0.058098871260881424, "step": 7026 }, { "epoch": 1.8242471443406023, "grad_norm": 1.4347279071807861, "learning_rate": 5.908019090497045e-07, "loss": 0.10241179168224335, "step": 7027 }, { "epoch": 1.8245067497403946, "grad_norm": 1.7462012767791748, "learning_rate": 5.905360639701571e-07, "loss": 0.0745425894856453, "step": 7028 }, { "epoch": 1.824766355140187, "grad_norm": 1.2782135009765625, "learning_rate": 5.902706006343887e-07, "loss": 0.09678591787815094, "step": 7029 }, { "epoch": 1.8250259605399792, "grad_norm": 1.663150668144226, "learning_rate": 5.900055190893262e-07, "loss": 0.05998988077044487, "step": 7030 }, { "epoch": 1.8252855659397715, "grad_norm": 1.7685041427612305, "learning_rate": 5.897408193818311e-07, "loss": 0.0251249261200428, "step": 7031 }, { "epoch": 1.8255451713395638, "grad_norm": 2.934274196624756, "learning_rate": 5.894765015586959e-07, "loss": 0.06681688129901886, "step": 7032 }, { "epoch": 1.8258047767393562, "grad_norm": 3.9654765129089355, "learning_rate": 5.892125656666454e-07, "loss": 0.03385275602340698, "step": 7033 }, { "epoch": 1.8260643821391485, "grad_norm": 1.1897658109664917, "learning_rate": 5.889490117523381e-07, "loss": 0.09820882976055145, "step": 7034 }, { "epoch": 1.8263239875389408, "grad_norm": 1.4234312772750854, "learning_rate": 5.886858398623635e-07, "loss": 0.10062803328037262, "step": 7035 }, { "epoch": 1.826583592938733, "grad_norm": 1.6561388969421387, "learning_rate": 5.884230500432455e-07, "loss": 0.06978664547204971, "step": 7036 }, { "epoch": 1.8268431983385254, "grad_norm": 1.5003769397735596, "learning_rate": 5.881606423414383e-07, "loss": 0.08112161606550217, "step": 7037 }, { "epoch": 1.8271028037383177, "grad_norm": 2.0311288833618164, "learning_rate": 5.878986168033301e-07, "loss": 0.04601301997900009, "step": 7038 }, { "epoch": 1.82736240913811, "grad_norm": 3.0405235290527344, "learning_rate": 5.876369734752411e-07, "loss": 0.031982067972421646, "step": 7039 }, { "epoch": 1.8276220145379023, "grad_norm": 2.9689114093780518, "learning_rate": 5.873757124034234e-07, "loss": 0.06303805112838745, "step": 7040 }, { "epoch": 1.8278816199376946, "grad_norm": 1.736883521080017, "learning_rate": 5.871148336340623e-07, "loss": 0.07974912226200104, "step": 7041 }, { "epoch": 1.828141225337487, "grad_norm": 1.5291460752487183, "learning_rate": 5.868543372132751e-07, "loss": 0.09144198894500732, "step": 7042 }, { "epoch": 1.8284008307372792, "grad_norm": 1.3781428337097168, "learning_rate": 5.865942231871113e-07, "loss": 0.09899541735649109, "step": 7043 }, { "epoch": 1.8286604361370715, "grad_norm": 2.408766031265259, "learning_rate": 5.863344916015538e-07, "loss": 0.04371140897274017, "step": 7044 }, { "epoch": 1.8289200415368638, "grad_norm": 2.2876596450805664, "learning_rate": 5.86075142502517e-07, "loss": 0.07089430838823318, "step": 7045 }, { "epoch": 1.8291796469366561, "grad_norm": 1.1097508668899536, "learning_rate": 5.858161759358472e-07, "loss": 0.08874369412660599, "step": 7046 }, { "epoch": 1.8294392523364484, "grad_norm": 1.7540377378463745, "learning_rate": 5.855575919473248e-07, "loss": 0.06171216070652008, "step": 7047 }, { "epoch": 1.829698857736241, "grad_norm": 1.4184932708740234, "learning_rate": 5.852993905826606e-07, "loss": 0.08555351197719574, "step": 7048 }, { "epoch": 1.8299584631360333, "grad_norm": 2.552164316177368, "learning_rate": 5.850415718874988e-07, "loss": 0.04337412491440773, "step": 7049 }, { "epoch": 1.8302180685358256, "grad_norm": 1.3194503784179688, "learning_rate": 5.847841359074165e-07, "loss": 0.0813361182808876, "step": 7050 }, { "epoch": 1.8304776739356179, "grad_norm": 1.7732791900634766, "learning_rate": 5.845270826879221e-07, "loss": 0.10185234993696213, "step": 7051 }, { "epoch": 1.8307372793354102, "grad_norm": 1.9007766246795654, "learning_rate": 5.842704122744569e-07, "loss": 0.07896015048027039, "step": 7052 }, { "epoch": 1.8309968847352025, "grad_norm": 1.1215380430221558, "learning_rate": 5.840141247123938e-07, "loss": 0.09215778112411499, "step": 7053 }, { "epoch": 1.8312564901349948, "grad_norm": 1.3441162109375, "learning_rate": 5.837582200470388e-07, "loss": 0.08570453524589539, "step": 7054 }, { "epoch": 1.8315160955347871, "grad_norm": 1.7632420063018799, "learning_rate": 5.835026983236308e-07, "loss": 0.058166973292827606, "step": 7055 }, { "epoch": 1.8317757009345794, "grad_norm": 1.1112332344055176, "learning_rate": 5.832475595873389e-07, "loss": 0.11262109875679016, "step": 7056 }, { "epoch": 1.8320353063343717, "grad_norm": 1.2784178256988525, "learning_rate": 5.829928038832668e-07, "loss": 0.0993492379784584, "step": 7057 }, { "epoch": 1.832294911734164, "grad_norm": 2.386428117752075, "learning_rate": 5.827384312564488e-07, "loss": 0.053949933499097824, "step": 7058 }, { "epoch": 1.8325545171339563, "grad_norm": 1.466478943824768, "learning_rate": 5.824844417518527e-07, "loss": 0.08458688855171204, "step": 7059 }, { "epoch": 1.8328141225337489, "grad_norm": 1.5836235284805298, "learning_rate": 5.822308354143776e-07, "loss": 0.05155277997255325, "step": 7060 }, { "epoch": 1.8330737279335412, "grad_norm": 1.6136573553085327, "learning_rate": 5.819776122888553e-07, "loss": 0.07458634674549103, "step": 7061 }, { "epoch": 1.8333333333333335, "grad_norm": 1.5571075677871704, "learning_rate": 5.817247724200503e-07, "loss": 0.08362628519535065, "step": 7062 }, { "epoch": 1.8335929387331258, "grad_norm": 2.239379405975342, "learning_rate": 5.814723158526586e-07, "loss": 0.09512616693973541, "step": 7063 }, { "epoch": 1.833852544132918, "grad_norm": 1.4477635622024536, "learning_rate": 5.812202426313087e-07, "loss": 0.09593158215284348, "step": 7064 }, { "epoch": 1.8341121495327104, "grad_norm": 3.472691059112549, "learning_rate": 5.809685528005614e-07, "loss": 0.06587904691696167, "step": 7065 }, { "epoch": 1.8343717549325027, "grad_norm": 2.2679028511047363, "learning_rate": 5.807172464049096e-07, "loss": 0.0662396103143692, "step": 7066 }, { "epoch": 1.834631360332295, "grad_norm": 1.1670401096343994, "learning_rate": 5.80466323488779e-07, "loss": 0.09312635660171509, "step": 7067 }, { "epoch": 1.8348909657320873, "grad_norm": 1.5688661336898804, "learning_rate": 5.802157840965264e-07, "loss": 0.048709943890571594, "step": 7068 }, { "epoch": 1.8351505711318796, "grad_norm": 1.7620502710342407, "learning_rate": 5.799656282724417e-07, "loss": 0.07427439838647842, "step": 7069 }, { "epoch": 1.835410176531672, "grad_norm": 2.681354522705078, "learning_rate": 5.797158560607467e-07, "loss": 0.056229155510663986, "step": 7070 }, { "epoch": 1.8356697819314642, "grad_norm": 3.199444055557251, "learning_rate": 5.794664675055956e-07, "loss": 0.056655559688806534, "step": 7071 }, { "epoch": 1.8359293873312565, "grad_norm": 2.570410966873169, "learning_rate": 5.792174626510742e-07, "loss": 0.036950282752513885, "step": 7072 }, { "epoch": 1.8361889927310489, "grad_norm": 2.1183841228485107, "learning_rate": 5.789688415412015e-07, "loss": 0.04155125468969345, "step": 7073 }, { "epoch": 1.8364485981308412, "grad_norm": 1.343191146850586, "learning_rate": 5.787206042199273e-07, "loss": 0.0793817788362503, "step": 7074 }, { "epoch": 1.8367082035306335, "grad_norm": 1.7084650993347168, "learning_rate": 5.784727507311344e-07, "loss": 0.039919413626194, "step": 7075 }, { "epoch": 1.8369678089304258, "grad_norm": 3.756014108657837, "learning_rate": 5.78225281118638e-07, "loss": 0.03448256477713585, "step": 7076 }, { "epoch": 1.837227414330218, "grad_norm": 2.095703363418579, "learning_rate": 5.779781954261845e-07, "loss": 0.051239483058452606, "step": 7077 }, { "epoch": 1.8374870197300104, "grad_norm": 1.2606490850448608, "learning_rate": 5.777314936974541e-07, "loss": 0.07652656733989716, "step": 7078 }, { "epoch": 1.8377466251298027, "grad_norm": 1.611302137374878, "learning_rate": 5.774851759760566e-07, "loss": 0.046185202896595, "step": 7079 }, { "epoch": 1.838006230529595, "grad_norm": 1.7786353826522827, "learning_rate": 5.77239242305536e-07, "loss": 0.07938110083341599, "step": 7080 }, { "epoch": 1.8382658359293873, "grad_norm": 2.042827844619751, "learning_rate": 5.769936927293679e-07, "loss": 0.04041393846273422, "step": 7081 }, { "epoch": 1.8385254413291796, "grad_norm": 2.651853084564209, "learning_rate": 5.767485272909598e-07, "loss": 0.07734154164791107, "step": 7082 }, { "epoch": 1.838785046728972, "grad_norm": 2.9740712642669678, "learning_rate": 5.765037460336512e-07, "loss": 0.07747037708759308, "step": 7083 }, { "epoch": 1.8390446521287642, "grad_norm": 2.1026179790496826, "learning_rate": 5.762593490007135e-07, "loss": 0.04078897088766098, "step": 7084 }, { "epoch": 1.8393042575285565, "grad_norm": 1.1183712482452393, "learning_rate": 5.76015336235351e-07, "loss": 0.09275007992982864, "step": 7085 }, { "epoch": 1.8395638629283488, "grad_norm": 1.2383148670196533, "learning_rate": 5.757717077807e-07, "loss": 0.08707208931446075, "step": 7086 }, { "epoch": 1.8398234683281411, "grad_norm": 1.5544302463531494, "learning_rate": 5.75528463679827e-07, "loss": 0.07812345027923584, "step": 7087 }, { "epoch": 1.8400830737279334, "grad_norm": 1.5234843492507935, "learning_rate": 5.752856039757334e-07, "loss": 0.08600124716758728, "step": 7088 }, { "epoch": 1.8403426791277258, "grad_norm": 1.0909693241119385, "learning_rate": 5.750431287113501e-07, "loss": 0.031749509274959564, "step": 7089 }, { "epoch": 1.840602284527518, "grad_norm": 1.7179806232452393, "learning_rate": 5.74801037929542e-07, "loss": 0.062182337045669556, "step": 7090 }, { "epoch": 1.8408618899273104, "grad_norm": 2.6752021312713623, "learning_rate": 5.745593316731051e-07, "loss": 0.07572424411773682, "step": 7091 }, { "epoch": 1.8411214953271027, "grad_norm": 1.5333988666534424, "learning_rate": 5.74318009984767e-07, "loss": 0.07981676608324051, "step": 7092 }, { "epoch": 1.841381100726895, "grad_norm": 1.7389171123504639, "learning_rate": 5.740770729071887e-07, "loss": 0.06854695826768875, "step": 7093 }, { "epoch": 1.8416407061266873, "grad_norm": 1.3917428255081177, "learning_rate": 5.738365204829617e-07, "loss": 0.09740681946277618, "step": 7094 }, { "epoch": 1.8419003115264796, "grad_norm": 0.9536347389221191, "learning_rate": 5.735963527546102e-07, "loss": 0.03218274936079979, "step": 7095 }, { "epoch": 1.842159916926272, "grad_norm": 2.9624996185302734, "learning_rate": 5.733565697645906e-07, "loss": 0.06660883128643036, "step": 7096 }, { "epoch": 1.8424195223260644, "grad_norm": 2.371778964996338, "learning_rate": 5.731171715552908e-07, "loss": 0.061240456998348236, "step": 7097 }, { "epoch": 1.8426791277258567, "grad_norm": 2.875049352645874, "learning_rate": 5.728781581690312e-07, "loss": 0.04385076463222504, "step": 7098 }, { "epoch": 1.842938733125649, "grad_norm": 3.611854314804077, "learning_rate": 5.726395296480637e-07, "loss": 0.06684312224388123, "step": 7099 }, { "epoch": 1.8431983385254413, "grad_norm": 2.14725923538208, "learning_rate": 5.724012860345723e-07, "loss": 0.06440086662769318, "step": 7100 }, { "epoch": 1.8434579439252337, "grad_norm": 2.555089235305786, "learning_rate": 5.721634273706735e-07, "loss": 0.07570743560791016, "step": 7101 }, { "epoch": 1.843717549325026, "grad_norm": 3.3096671104431152, "learning_rate": 5.719259536984147e-07, "loss": 0.02889937162399292, "step": 7102 }, { "epoch": 1.8439771547248183, "grad_norm": 1.5273785591125488, "learning_rate": 5.716888650597758e-07, "loss": 0.08111004531383514, "step": 7103 }, { "epoch": 1.8442367601246106, "grad_norm": 3.228947401046753, "learning_rate": 5.714521614966689e-07, "loss": 0.027918275445699692, "step": 7104 }, { "epoch": 1.8444963655244029, "grad_norm": 3.3861968517303467, "learning_rate": 5.71215843050938e-07, "loss": 0.053528960794210434, "step": 7105 }, { "epoch": 1.8447559709241952, "grad_norm": 2.3823180198669434, "learning_rate": 5.709799097643583e-07, "loss": 0.022771280258893967, "step": 7106 }, { "epoch": 1.8450155763239875, "grad_norm": 1.3662010431289673, "learning_rate": 5.707443616786376e-07, "loss": 0.09458792209625244, "step": 7107 }, { "epoch": 1.8452751817237798, "grad_norm": 1.6666123867034912, "learning_rate": 5.705091988354152e-07, "loss": 0.08599069714546204, "step": 7108 }, { "epoch": 1.8455347871235723, "grad_norm": 2.052866220474243, "learning_rate": 5.702744212762634e-07, "loss": 0.06066840887069702, "step": 7109 }, { "epoch": 1.8457943925233646, "grad_norm": 1.7547920942306519, "learning_rate": 5.700400290426841e-07, "loss": 0.052975527942180634, "step": 7110 }, { "epoch": 1.846053997923157, "grad_norm": 2.7716732025146484, "learning_rate": 5.698060221761133e-07, "loss": 0.06657262146472931, "step": 7111 }, { "epoch": 1.8463136033229492, "grad_norm": 2.8348281383514404, "learning_rate": 5.695724007179182e-07, "loss": 0.06196019798517227, "step": 7112 }, { "epoch": 1.8465732087227416, "grad_norm": 2.227423906326294, "learning_rate": 5.693391647093975e-07, "loss": 0.04801774024963379, "step": 7113 }, { "epoch": 1.8468328141225339, "grad_norm": 1.8913335800170898, "learning_rate": 5.69106314191782e-07, "loss": 0.06527165323495865, "step": 7114 }, { "epoch": 1.8470924195223262, "grad_norm": 2.3194751739501953, "learning_rate": 5.688738492062342e-07, "loss": 0.08032073080539703, "step": 7115 }, { "epoch": 1.8473520249221185, "grad_norm": 1.9783313274383545, "learning_rate": 5.686417697938483e-07, "loss": 0.057693000882864, "step": 7116 }, { "epoch": 1.8476116303219108, "grad_norm": 1.8343628644943237, "learning_rate": 5.684100759956516e-07, "loss": 0.04938192293047905, "step": 7117 }, { "epoch": 1.847871235721703, "grad_norm": 3.0450098514556885, "learning_rate": 5.681787678526015e-07, "loss": 0.06834442913532257, "step": 7118 }, { "epoch": 1.8481308411214954, "grad_norm": 1.4410899877548218, "learning_rate": 5.679478454055882e-07, "loss": 0.08671992272138596, "step": 7119 }, { "epoch": 1.8483904465212877, "grad_norm": 1.9634960889816284, "learning_rate": 5.677173086954331e-07, "loss": 0.06057975813746452, "step": 7120 }, { "epoch": 1.84865005192108, "grad_norm": 2.9801084995269775, "learning_rate": 5.674871577628907e-07, "loss": 0.024871688336133957, "step": 7121 }, { "epoch": 1.8489096573208723, "grad_norm": 1.1189202070236206, "learning_rate": 5.672573926486457e-07, "loss": 0.09923960268497467, "step": 7122 }, { "epoch": 1.8491692627206646, "grad_norm": 2.4192025661468506, "learning_rate": 5.670280133933152e-07, "loss": 0.10007795691490173, "step": 7123 }, { "epoch": 1.849428868120457, "grad_norm": 2.7425801753997803, "learning_rate": 5.66799020037449e-07, "loss": 0.05116640776395798, "step": 7124 }, { "epoch": 1.8496884735202492, "grad_norm": 2.7801594734191895, "learning_rate": 5.66570412621527e-07, "loss": 0.022816075012087822, "step": 7125 }, { "epoch": 1.8499480789200415, "grad_norm": 1.722102165222168, "learning_rate": 5.663421911859622e-07, "loss": 0.06401734054088593, "step": 7126 }, { "epoch": 1.8502076843198338, "grad_norm": 2.0958781242370605, "learning_rate": 5.661143557710988e-07, "loss": 0.08182752132415771, "step": 7127 }, { "epoch": 1.8504672897196262, "grad_norm": 1.1252976655960083, "learning_rate": 5.658869064172132e-07, "loss": 0.09694287925958633, "step": 7128 }, { "epoch": 1.8507268951194185, "grad_norm": 2.9791154861450195, "learning_rate": 5.656598431645127e-07, "loss": 0.06568285822868347, "step": 7129 }, { "epoch": 1.8509865005192108, "grad_norm": 2.2512590885162354, "learning_rate": 5.654331660531374e-07, "loss": 0.07425650954246521, "step": 7130 }, { "epoch": 1.851246105919003, "grad_norm": 1.6906564235687256, "learning_rate": 5.652068751231583e-07, "loss": 0.07151858508586884, "step": 7131 }, { "epoch": 1.8515057113187954, "grad_norm": 2.3572733402252197, "learning_rate": 5.649809704145786e-07, "loss": 0.046365633606910706, "step": 7132 }, { "epoch": 1.8517653167185877, "grad_norm": 1.5364265441894531, "learning_rate": 5.64755451967333e-07, "loss": 0.04132421314716339, "step": 7133 }, { "epoch": 1.85202492211838, "grad_norm": 2.062997817993164, "learning_rate": 5.645303198212878e-07, "loss": 0.03597059100866318, "step": 7134 }, { "epoch": 1.8522845275181723, "grad_norm": 4.137160778045654, "learning_rate": 5.643055740162418e-07, "loss": 0.03389241546392441, "step": 7135 }, { "epoch": 1.8525441329179646, "grad_norm": 2.052739143371582, "learning_rate": 5.640812145919245e-07, "loss": 0.06966477632522583, "step": 7136 }, { "epoch": 1.852803738317757, "grad_norm": 2.0269320011138916, "learning_rate": 5.638572415879972e-07, "loss": 0.07722099125385284, "step": 7137 }, { "epoch": 1.8530633437175492, "grad_norm": 2.7728724479675293, "learning_rate": 5.63633655044054e-07, "loss": 0.05548319220542908, "step": 7138 }, { "epoch": 1.8533229491173415, "grad_norm": 2.156992197036743, "learning_rate": 5.63410454999619e-07, "loss": 0.0521087720990181, "step": 7139 }, { "epoch": 1.8535825545171338, "grad_norm": 1.2483187913894653, "learning_rate": 5.6318764149415e-07, "loss": 0.0889304131269455, "step": 7140 }, { "epoch": 1.8538421599169261, "grad_norm": 2.5509591102600098, "learning_rate": 5.629652145670338e-07, "loss": 0.046675242483615875, "step": 7141 }, { "epoch": 1.8541017653167184, "grad_norm": 1.8796852827072144, "learning_rate": 5.627431742575914e-07, "loss": 0.07822427153587341, "step": 7142 }, { "epoch": 1.8543613707165107, "grad_norm": 1.9820215702056885, "learning_rate": 5.625215206050745e-07, "loss": 0.06584002077579498, "step": 7143 }, { "epoch": 1.854620976116303, "grad_norm": 2.001718521118164, "learning_rate": 5.623002536486661e-07, "loss": 0.07336613535881042, "step": 7144 }, { "epoch": 1.8548805815160956, "grad_norm": 1.6674765348434448, "learning_rate": 5.620793734274812e-07, "loss": 0.06323391199111938, "step": 7145 }, { "epoch": 1.855140186915888, "grad_norm": 1.3103089332580566, "learning_rate": 5.618588799805662e-07, "loss": 0.09284624457359314, "step": 7146 }, { "epoch": 1.8553997923156802, "grad_norm": 1.321955680847168, "learning_rate": 5.616387733468992e-07, "loss": 0.10196252912282944, "step": 7147 }, { "epoch": 1.8556593977154725, "grad_norm": 2.0213606357574463, "learning_rate": 5.614190535653909e-07, "loss": 0.05257360637187958, "step": 7148 }, { "epoch": 1.8559190031152648, "grad_norm": 2.111698627471924, "learning_rate": 5.611997206748815e-07, "loss": 0.04384759068489075, "step": 7149 }, { "epoch": 1.8561786085150571, "grad_norm": 2.505484104156494, "learning_rate": 5.60980774714145e-07, "loss": 0.07031410187482834, "step": 7150 }, { "epoch": 1.8564382139148494, "grad_norm": 1.9934964179992676, "learning_rate": 5.607622157218851e-07, "loss": 0.028956813737750053, "step": 7151 }, { "epoch": 1.8566978193146417, "grad_norm": 0.475263386964798, "learning_rate": 5.60544043736739e-07, "loss": 0.040220052003860474, "step": 7152 }, { "epoch": 1.856957424714434, "grad_norm": 3.118082284927368, "learning_rate": 5.60326258797274e-07, "loss": 0.03933624178171158, "step": 7153 }, { "epoch": 1.8572170301142263, "grad_norm": 2.738222360610962, "learning_rate": 5.601088609419893e-07, "loss": 0.029872599989175797, "step": 7154 }, { "epoch": 1.8574766355140186, "grad_norm": 1.991898536682129, "learning_rate": 5.598918502093165e-07, "loss": 0.039287060499191284, "step": 7155 }, { "epoch": 1.857736240913811, "grad_norm": 2.552272081375122, "learning_rate": 5.596752266376175e-07, "loss": 0.039059460163116455, "step": 7156 }, { "epoch": 1.8579958463136035, "grad_norm": 1.2502588033676147, "learning_rate": 5.594589902651866e-07, "loss": 0.11493326723575592, "step": 7157 }, { "epoch": 1.8582554517133958, "grad_norm": 2.0200600624084473, "learning_rate": 5.592431411302498e-07, "loss": 0.06934788823127747, "step": 7158 }, { "epoch": 1.858515057113188, "grad_norm": 2.9341723918914795, "learning_rate": 5.590276792709639e-07, "loss": 0.06536795198917389, "step": 7159 }, { "epoch": 1.8587746625129804, "grad_norm": 1.447871208190918, "learning_rate": 5.588126047254173e-07, "loss": 0.09085738658905029, "step": 7160 }, { "epoch": 1.8590342679127727, "grad_norm": 1.8653486967086792, "learning_rate": 5.58597917531631e-07, "loss": 0.03188139945268631, "step": 7161 }, { "epoch": 1.859293873312565, "grad_norm": 1.4727171659469604, "learning_rate": 5.583836177275561e-07, "loss": 0.06309160590171814, "step": 7162 }, { "epoch": 1.8595534787123573, "grad_norm": 3.1142168045043945, "learning_rate": 5.581697053510768e-07, "loss": 0.060690607875585556, "step": 7163 }, { "epoch": 1.8598130841121496, "grad_norm": 3.0053720474243164, "learning_rate": 5.579561804400068e-07, "loss": 0.05731711909174919, "step": 7164 }, { "epoch": 1.860072689511942, "grad_norm": 2.0649266242980957, "learning_rate": 5.577430430320932e-07, "loss": 0.05152491107583046, "step": 7165 }, { "epoch": 1.8603322949117342, "grad_norm": 1.6516352891921997, "learning_rate": 5.575302931650135e-07, "loss": 0.06014912948012352, "step": 7166 }, { "epoch": 1.8605919003115265, "grad_norm": 1.2344369888305664, "learning_rate": 5.573179308763772e-07, "loss": 0.10155508667230606, "step": 7167 }, { "epoch": 1.8608515057113189, "grad_norm": 2.827587127685547, "learning_rate": 5.571059562037244e-07, "loss": 0.05979536846280098, "step": 7168 }, { "epoch": 1.8611111111111112, "grad_norm": 1.907150149345398, "learning_rate": 5.568943691845285e-07, "loss": 0.0655461773276329, "step": 7169 }, { "epoch": 1.8613707165109035, "grad_norm": 3.095856189727783, "learning_rate": 5.566831698561925e-07, "loss": 0.063877634704113, "step": 7170 }, { "epoch": 1.8616303219106958, "grad_norm": 1.36872398853302, "learning_rate": 5.56472358256052e-07, "loss": 0.10893528908491135, "step": 7171 }, { "epoch": 1.861889927310488, "grad_norm": 1.5645896196365356, "learning_rate": 5.56261934421373e-07, "loss": 0.07485348731279373, "step": 7172 }, { "epoch": 1.8621495327102804, "grad_norm": 1.625811219215393, "learning_rate": 5.560518983893542e-07, "loss": 0.050916340202093124, "step": 7173 }, { "epoch": 1.8624091381100727, "grad_norm": 1.222854495048523, "learning_rate": 5.558422501971251e-07, "loss": 0.10284251719713211, "step": 7174 }, { "epoch": 1.862668743509865, "grad_norm": 4.011045932769775, "learning_rate": 5.556329898817467e-07, "loss": 0.052676159888505936, "step": 7175 }, { "epoch": 1.8629283489096573, "grad_norm": 2.7600033283233643, "learning_rate": 5.554241174802112e-07, "loss": 0.04667083919048309, "step": 7176 }, { "epoch": 1.8631879543094496, "grad_norm": 3.6929235458374023, "learning_rate": 5.552156330294427e-07, "loss": 0.04407970979809761, "step": 7177 }, { "epoch": 1.863447559709242, "grad_norm": 2.870166540145874, "learning_rate": 5.55007536566296e-07, "loss": 0.06476444005966187, "step": 7178 }, { "epoch": 1.8637071651090342, "grad_norm": 2.5365219116210938, "learning_rate": 5.547998281275589e-07, "loss": 0.03278786689043045, "step": 7179 }, { "epoch": 1.8639667705088265, "grad_norm": 1.995769739151001, "learning_rate": 5.545925077499481e-07, "loss": 0.043899718672037125, "step": 7180 }, { "epoch": 1.8642263759086188, "grad_norm": 2.7859485149383545, "learning_rate": 5.543855754701142e-07, "loss": 0.04427701607346535, "step": 7181 }, { "epoch": 1.8644859813084111, "grad_norm": 3.153670310974121, "learning_rate": 5.541790313246371e-07, "loss": 0.03375053405761719, "step": 7182 }, { "epoch": 1.8647455867082035, "grad_norm": 2.0544443130493164, "learning_rate": 5.539728753500302e-07, "loss": 0.06840628385543823, "step": 7183 }, { "epoch": 1.8650051921079958, "grad_norm": 1.6247950792312622, "learning_rate": 5.537671075827364e-07, "loss": 0.05296798795461655, "step": 7184 }, { "epoch": 1.865264797507788, "grad_norm": 2.160846710205078, "learning_rate": 5.535617280591308e-07, "loss": 0.029627583920955658, "step": 7185 }, { "epoch": 1.8655244029075804, "grad_norm": 3.969337224960327, "learning_rate": 5.5335673681552e-07, "loss": 0.038410529494285583, "step": 7186 }, { "epoch": 1.8657840083073727, "grad_norm": 1.2558056116104126, "learning_rate": 5.531521338881417e-07, "loss": 0.0953197106719017, "step": 7187 }, { "epoch": 1.866043613707165, "grad_norm": 2.1702704429626465, "learning_rate": 5.529479193131649e-07, "loss": 0.09213104844093323, "step": 7188 }, { "epoch": 1.8663032191069573, "grad_norm": 3.412900447845459, "learning_rate": 5.527440931266902e-07, "loss": 0.03390642628073692, "step": 7189 }, { "epoch": 1.8665628245067496, "grad_norm": 1.8839139938354492, "learning_rate": 5.525406553647495e-07, "loss": 0.06344301998615265, "step": 7190 }, { "epoch": 1.866822429906542, "grad_norm": 1.852896809577942, "learning_rate": 5.523376060633054e-07, "loss": 0.04628432169556618, "step": 7191 }, { "epoch": 1.8670820353063342, "grad_norm": 1.9753142595291138, "learning_rate": 5.52134945258253e-07, "loss": 0.031078889966011047, "step": 7192 }, { "epoch": 1.8673416407061267, "grad_norm": 2.9121336936950684, "learning_rate": 5.519326729854176e-07, "loss": 0.06996820867061615, "step": 7193 }, { "epoch": 1.867601246105919, "grad_norm": 2.6899876594543457, "learning_rate": 5.517307892805569e-07, "loss": 0.09022223204374313, "step": 7194 }, { "epoch": 1.8678608515057114, "grad_norm": 2.3476972579956055, "learning_rate": 5.515292941793587e-07, "loss": 0.03685808926820755, "step": 7195 }, { "epoch": 1.8681204569055037, "grad_norm": 3.5275652408599854, "learning_rate": 5.513281877174431e-07, "loss": 0.07008738815784454, "step": 7196 }, { "epoch": 1.868380062305296, "grad_norm": 2.6970391273498535, "learning_rate": 5.51127469930361e-07, "loss": 0.04838114231824875, "step": 7197 }, { "epoch": 1.8686396677050883, "grad_norm": 1.994193196296692, "learning_rate": 5.509271408535949e-07, "loss": 0.09732715785503387, "step": 7198 }, { "epoch": 1.8688992731048806, "grad_norm": 2.882844924926758, "learning_rate": 5.507272005225578e-07, "loss": 0.06445015221834183, "step": 7199 }, { "epoch": 1.8691588785046729, "grad_norm": 2.3364784717559814, "learning_rate": 5.505276489725951e-07, "loss": 0.053324922919273376, "step": 7200 }, { "epoch": 1.8694184839044652, "grad_norm": 2.824132204055786, "learning_rate": 5.503284862389828e-07, "loss": 0.049420714378356934, "step": 7201 }, { "epoch": 1.8696780893042575, "grad_norm": 3.9442458152770996, "learning_rate": 5.501297123569287e-07, "loss": 0.08071783930063248, "step": 7202 }, { "epoch": 1.8699376947040498, "grad_norm": 1.9043912887573242, "learning_rate": 5.499313273615706e-07, "loss": 0.0772644579410553, "step": 7203 }, { "epoch": 1.870197300103842, "grad_norm": 2.100532054901123, "learning_rate": 5.49733331287979e-07, "loss": 0.031517673283815384, "step": 7204 }, { "epoch": 1.8704569055036346, "grad_norm": 2.011747121810913, "learning_rate": 5.495357241711554e-07, "loss": 0.055769212543964386, "step": 7205 }, { "epoch": 1.870716510903427, "grad_norm": 1.5445022583007812, "learning_rate": 5.493385060460316e-07, "loss": 0.09042543172836304, "step": 7206 }, { "epoch": 1.8709761163032192, "grad_norm": 1.7549000978469849, "learning_rate": 5.491416769474714e-07, "loss": 0.0701138898730278, "step": 7207 }, { "epoch": 1.8712357217030116, "grad_norm": 1.6204867362976074, "learning_rate": 5.489452369102696e-07, "loss": 0.07606621086597443, "step": 7208 }, { "epoch": 1.8714953271028039, "grad_norm": 1.2219535112380981, "learning_rate": 5.487491859691525e-07, "loss": 0.09827135503292084, "step": 7209 }, { "epoch": 1.8717549325025962, "grad_norm": 2.1428003311157227, "learning_rate": 5.485535241587777e-07, "loss": 0.08852627873420715, "step": 7210 }, { "epoch": 1.8720145379023885, "grad_norm": 2.216468334197998, "learning_rate": 5.483582515137331e-07, "loss": 0.0820828378200531, "step": 7211 }, { "epoch": 1.8722741433021808, "grad_norm": 2.2403950691223145, "learning_rate": 5.481633680685388e-07, "loss": 0.07790782302618027, "step": 7212 }, { "epoch": 1.872533748701973, "grad_norm": 1.887462854385376, "learning_rate": 5.479688738576457e-07, "loss": 0.07671099901199341, "step": 7213 }, { "epoch": 1.8727933541017654, "grad_norm": 2.0997633934020996, "learning_rate": 5.477747689154355e-07, "loss": 0.07080894708633423, "step": 7214 }, { "epoch": 1.8730529595015577, "grad_norm": 4.004630088806152, "learning_rate": 5.475810532762226e-07, "loss": 0.032463766634464264, "step": 7215 }, { "epoch": 1.87331256490135, "grad_norm": 1.9574999809265137, "learning_rate": 5.473877269742502e-07, "loss": 0.06643374264240265, "step": 7216 }, { "epoch": 1.8735721703011423, "grad_norm": 2.6187469959259033, "learning_rate": 5.471947900436947e-07, "loss": 0.017606280744075775, "step": 7217 }, { "epoch": 1.8738317757009346, "grad_norm": 3.0662567615509033, "learning_rate": 5.470022425186627e-07, "loss": 0.058349430561065674, "step": 7218 }, { "epoch": 1.874091381100727, "grad_norm": 1.6935930252075195, "learning_rate": 5.468100844331922e-07, "loss": 0.07234371453523636, "step": 7219 }, { "epoch": 1.8743509865005192, "grad_norm": 2.311542510986328, "learning_rate": 5.466183158212525e-07, "loss": 0.07209601998329163, "step": 7220 }, { "epoch": 1.8746105919003115, "grad_norm": 1.6826496124267578, "learning_rate": 5.464269367167439e-07, "loss": 0.062497179955244064, "step": 7221 }, { "epoch": 1.8748701973001038, "grad_norm": 1.2406011819839478, "learning_rate": 5.462359471534976e-07, "loss": 0.0980529859662056, "step": 7222 }, { "epoch": 1.8751298026998962, "grad_norm": 2.7152397632598877, "learning_rate": 5.460453471652765e-07, "loss": 0.0780695229768753, "step": 7223 }, { "epoch": 1.8753894080996885, "grad_norm": 2.0355639457702637, "learning_rate": 5.45855136785774e-07, "loss": 0.09746035933494568, "step": 7224 }, { "epoch": 1.8756490134994808, "grad_norm": 1.8916116952896118, "learning_rate": 5.45665316048615e-07, "loss": 0.08477144688367844, "step": 7225 }, { "epoch": 1.875908618899273, "grad_norm": 2.05802845954895, "learning_rate": 5.454758849873559e-07, "loss": 0.08989831805229187, "step": 7226 }, { "epoch": 1.8761682242990654, "grad_norm": 1.2651567459106445, "learning_rate": 5.45286843635483e-07, "loss": 0.0799824595451355, "step": 7227 }, { "epoch": 1.8764278296988577, "grad_norm": 1.3240751028060913, "learning_rate": 5.450981920264152e-07, "loss": 0.08450756222009659, "step": 7228 }, { "epoch": 1.87668743509865, "grad_norm": 2.7550413608551025, "learning_rate": 5.449099301935015e-07, "loss": 0.06116699427366257, "step": 7229 }, { "epoch": 1.8769470404984423, "grad_norm": 0.5885164141654968, "learning_rate": 5.447220581700222e-07, "loss": 0.035047926008701324, "step": 7230 }, { "epoch": 1.8772066458982346, "grad_norm": 4.070500373840332, "learning_rate": 5.445345759891889e-07, "loss": 0.07664401084184647, "step": 7231 }, { "epoch": 1.877466251298027, "grad_norm": 1.990390658378601, "learning_rate": 5.44347483684144e-07, "loss": 0.08037592470645905, "step": 7232 }, { "epoch": 1.8777258566978192, "grad_norm": 5.03997802734375, "learning_rate": 5.441607812879614e-07, "loss": 0.06851410120725632, "step": 7233 }, { "epoch": 1.8779854620976115, "grad_norm": 1.6392254829406738, "learning_rate": 5.439744688336453e-07, "loss": 0.062297724187374115, "step": 7234 }, { "epoch": 1.8782450674974038, "grad_norm": 1.4650583267211914, "learning_rate": 5.43788546354132e-07, "loss": 0.08492116630077362, "step": 7235 }, { "epoch": 1.8785046728971961, "grad_norm": 1.5084290504455566, "learning_rate": 5.436030138822881e-07, "loss": 0.06343288719654083, "step": 7236 }, { "epoch": 1.8787642782969884, "grad_norm": 2.3556694984436035, "learning_rate": 5.434178714509117e-07, "loss": 0.04239549860358238, "step": 7237 }, { "epoch": 1.8790238836967808, "grad_norm": 3.060015916824341, "learning_rate": 5.432331190927316e-07, "loss": 0.04333975538611412, "step": 7238 }, { "epoch": 1.879283489096573, "grad_norm": 1.9497798681259155, "learning_rate": 5.430487568404078e-07, "loss": 0.07804543524980545, "step": 7239 }, { "epoch": 1.8795430944963654, "grad_norm": 3.378786087036133, "learning_rate": 5.428647847265313e-07, "loss": 0.08169816434383392, "step": 7240 }, { "epoch": 1.879802699896158, "grad_norm": 1.604292631149292, "learning_rate": 5.426812027836241e-07, "loss": 0.04644296318292618, "step": 7241 }, { "epoch": 1.8800623052959502, "grad_norm": 2.6089792251586914, "learning_rate": 5.424980110441392e-07, "loss": 0.041414208710193634, "step": 7242 }, { "epoch": 1.8803219106957425, "grad_norm": 1.2384477853775024, "learning_rate": 5.423152095404608e-07, "loss": 0.10937109589576721, "step": 7243 }, { "epoch": 1.8805815160955348, "grad_norm": 1.862714409828186, "learning_rate": 5.421327983049046e-07, "loss": 0.0878058597445488, "step": 7244 }, { "epoch": 1.8808411214953271, "grad_norm": 1.696884036064148, "learning_rate": 5.419507773697157e-07, "loss": 0.06101612374186516, "step": 7245 }, { "epoch": 1.8811007268951194, "grad_norm": 1.7919634580612183, "learning_rate": 5.417691467670719e-07, "loss": 0.0686187818646431, "step": 7246 }, { "epoch": 1.8813603322949117, "grad_norm": 1.6381665468215942, "learning_rate": 5.415879065290811e-07, "loss": 0.07795805484056473, "step": 7247 }, { "epoch": 1.881619937694704, "grad_norm": 1.3706485033035278, "learning_rate": 5.414070566877826e-07, "loss": 0.08527199923992157, "step": 7248 }, { "epoch": 1.8818795430944963, "grad_norm": 1.3415945768356323, "learning_rate": 5.412265972751465e-07, "loss": 0.0990915596485138, "step": 7249 }, { "epoch": 1.8821391484942886, "grad_norm": 3.7007052898406982, "learning_rate": 5.410465283230735e-07, "loss": 0.03232192248106003, "step": 7250 }, { "epoch": 1.882398753894081, "grad_norm": 1.5344291925430298, "learning_rate": 5.408668498633959e-07, "loss": 0.08689523488283157, "step": 7251 }, { "epoch": 1.8826583592938733, "grad_norm": 2.0912888050079346, "learning_rate": 5.406875619278772e-07, "loss": 0.05869665741920471, "step": 7252 }, { "epoch": 1.8829179646936658, "grad_norm": 2.537407159805298, "learning_rate": 5.405086645482108e-07, "loss": 0.04833827167749405, "step": 7253 }, { "epoch": 1.883177570093458, "grad_norm": 1.5086331367492676, "learning_rate": 5.40330157756022e-07, "loss": 0.08868201076984406, "step": 7254 }, { "epoch": 1.8834371754932504, "grad_norm": 3.7553162574768066, "learning_rate": 5.401520415828664e-07, "loss": 0.06205041706562042, "step": 7255 }, { "epoch": 1.8836967808930427, "grad_norm": 1.6138535737991333, "learning_rate": 5.399743160602311e-07, "loss": 0.09319469332695007, "step": 7256 }, { "epoch": 1.883956386292835, "grad_norm": 2.8776538372039795, "learning_rate": 5.397969812195336e-07, "loss": 0.06907933950424194, "step": 7257 }, { "epoch": 1.8842159916926273, "grad_norm": 1.2806040048599243, "learning_rate": 5.396200370921232e-07, "loss": 0.10509011894464493, "step": 7258 }, { "epoch": 1.8844755970924196, "grad_norm": 2.010413885116577, "learning_rate": 5.394434837092792e-07, "loss": 0.052223291248083115, "step": 7259 }, { "epoch": 1.884735202492212, "grad_norm": 2.162227153778076, "learning_rate": 5.392673211022122e-07, "loss": 0.047316621989011765, "step": 7260 }, { "epoch": 1.8849948078920042, "grad_norm": 2.706059217453003, "learning_rate": 5.390915493020638e-07, "loss": 0.058458112180233, "step": 7261 }, { "epoch": 1.8852544132917965, "grad_norm": 2.4329934120178223, "learning_rate": 5.389161683399063e-07, "loss": 0.03431377932429314, "step": 7262 }, { "epoch": 1.8855140186915889, "grad_norm": 2.080951690673828, "learning_rate": 5.387411782467429e-07, "loss": 0.06434285640716553, "step": 7263 }, { "epoch": 1.8857736240913812, "grad_norm": 1.7840458154678345, "learning_rate": 5.385665790535087e-07, "loss": 0.06618332862854004, "step": 7264 }, { "epoch": 1.8860332294911735, "grad_norm": 1.5007435083389282, "learning_rate": 5.383923707910677e-07, "loss": 0.09115926921367645, "step": 7265 }, { "epoch": 1.8862928348909658, "grad_norm": 2.7507214546203613, "learning_rate": 5.382185534902165e-07, "loss": 0.10803970694541931, "step": 7266 }, { "epoch": 1.886552440290758, "grad_norm": 1.3037465810775757, "learning_rate": 5.380451271816824e-07, "loss": 0.09702640026807785, "step": 7267 }, { "epoch": 1.8868120456905504, "grad_norm": 1.1270434856414795, "learning_rate": 5.378720918961223e-07, "loss": 0.08255596458911896, "step": 7268 }, { "epoch": 1.8870716510903427, "grad_norm": 1.9674606323242188, "learning_rate": 5.376994476641252e-07, "loss": 0.05734347552061081, "step": 7269 }, { "epoch": 1.887331256490135, "grad_norm": 1.3048129081726074, "learning_rate": 5.375271945162114e-07, "loss": 0.028094466775655746, "step": 7270 }, { "epoch": 1.8875908618899273, "grad_norm": 2.935762643814087, "learning_rate": 5.373553324828303e-07, "loss": 0.058932021260261536, "step": 7271 }, { "epoch": 1.8878504672897196, "grad_norm": 1.8799816370010376, "learning_rate": 5.371838615943637e-07, "loss": 0.031783126294612885, "step": 7272 }, { "epoch": 1.888110072689512, "grad_norm": 2.6773788928985596, "learning_rate": 5.370127818811234e-07, "loss": 0.06814919412136078, "step": 7273 }, { "epoch": 1.8883696780893042, "grad_norm": 4.232272148132324, "learning_rate": 5.368420933733525e-07, "loss": 0.07052066177129745, "step": 7274 }, { "epoch": 1.8886292834890965, "grad_norm": 3.1274027824401855, "learning_rate": 5.366717961012253e-07, "loss": 0.06221286207437515, "step": 7275 }, { "epoch": 1.8888888888888888, "grad_norm": 2.858930826187134, "learning_rate": 5.365018900948458e-07, "loss": 0.05445896089076996, "step": 7276 }, { "epoch": 1.8891484942886811, "grad_norm": 1.2804113626480103, "learning_rate": 5.363323753842498e-07, "loss": 0.08071129769086838, "step": 7277 }, { "epoch": 1.8894080996884735, "grad_norm": 3.7878949642181396, "learning_rate": 5.361632519994034e-07, "loss": 0.03146248310804367, "step": 7278 }, { "epoch": 1.8896677050882658, "grad_norm": 2.2433485984802246, "learning_rate": 5.359945199702042e-07, "loss": 0.03910640627145767, "step": 7279 }, { "epoch": 1.889927310488058, "grad_norm": 1.1445984840393066, "learning_rate": 5.358261793264796e-07, "loss": 0.11015906929969788, "step": 7280 }, { "epoch": 1.8901869158878504, "grad_norm": 1.695696473121643, "learning_rate": 5.356582300979887e-07, "loss": 0.05327192693948746, "step": 7281 }, { "epoch": 1.8904465212876427, "grad_norm": 1.2794114351272583, "learning_rate": 5.354906723144211e-07, "loss": 0.08869466185569763, "step": 7282 }, { "epoch": 1.890706126687435, "grad_norm": 2.752908229827881, "learning_rate": 5.353235060053971e-07, "loss": 0.03429798781871796, "step": 7283 }, { "epoch": 1.8909657320872273, "grad_norm": 1.6759836673736572, "learning_rate": 5.351567312004675e-07, "loss": 0.08978734910488129, "step": 7284 }, { "epoch": 1.8912253374870196, "grad_norm": 2.1669774055480957, "learning_rate": 5.349903479291153e-07, "loss": 0.05788474529981613, "step": 7285 }, { "epoch": 1.891484942886812, "grad_norm": 2.2342467308044434, "learning_rate": 5.348243562207519e-07, "loss": 0.04311051964759827, "step": 7286 }, { "epoch": 1.8917445482866042, "grad_norm": 1.771019458770752, "learning_rate": 5.34658756104722e-07, "loss": 0.0776401236653328, "step": 7287 }, { "epoch": 1.8920041536863965, "grad_norm": 1.8092021942138672, "learning_rate": 5.344935476102992e-07, "loss": 0.07985533028841019, "step": 7288 }, { "epoch": 1.892263759086189, "grad_norm": 2.5777220726013184, "learning_rate": 5.34328730766689e-07, "loss": 0.059049077332019806, "step": 7289 }, { "epoch": 1.8925233644859814, "grad_norm": 1.2160885334014893, "learning_rate": 5.341643056030269e-07, "loss": 0.05241958796977997, "step": 7290 }, { "epoch": 1.8927829698857737, "grad_norm": 2.6448705196380615, "learning_rate": 5.340002721483798e-07, "loss": 0.03847411274909973, "step": 7291 }, { "epoch": 1.893042575285566, "grad_norm": 2.5853655338287354, "learning_rate": 5.338366304317445e-07, "loss": 0.041642360389232635, "step": 7292 }, { "epoch": 1.8933021806853583, "grad_norm": 2.550676107406616, "learning_rate": 5.336733804820501e-07, "loss": 0.03503049910068512, "step": 7293 }, { "epoch": 1.8935617860851506, "grad_norm": 2.68114972114563, "learning_rate": 5.335105223281544e-07, "loss": 0.05423580855131149, "step": 7294 }, { "epoch": 1.8938213914849429, "grad_norm": 1.5732444524765015, "learning_rate": 5.333480559988477e-07, "loss": 0.07866833359003067, "step": 7295 }, { "epoch": 1.8940809968847352, "grad_norm": 2.1210055351257324, "learning_rate": 5.331859815228501e-07, "loss": 0.05840904265642166, "step": 7296 }, { "epoch": 1.8943406022845275, "grad_norm": 1.486438274383545, "learning_rate": 5.330242989288127e-07, "loss": 0.07167083024978638, "step": 7297 }, { "epoch": 1.8946002076843198, "grad_norm": 1.896733045578003, "learning_rate": 5.328630082453173e-07, "loss": 0.06502191722393036, "step": 7298 }, { "epoch": 1.894859813084112, "grad_norm": 2.864384889602661, "learning_rate": 5.327021095008762e-07, "loss": 0.06827370077371597, "step": 7299 }, { "epoch": 1.8951194184839044, "grad_norm": 2.7896299362182617, "learning_rate": 5.325416027239327e-07, "loss": 0.06766174733638763, "step": 7300 }, { "epoch": 1.895379023883697, "grad_norm": 3.0193936824798584, "learning_rate": 5.323814879428608e-07, "loss": 0.053964391350746155, "step": 7301 }, { "epoch": 1.8956386292834893, "grad_norm": 1.9105150699615479, "learning_rate": 5.322217651859652e-07, "loss": 0.05500335246324539, "step": 7302 }, { "epoch": 1.8958982346832816, "grad_norm": 2.2361626625061035, "learning_rate": 5.320624344814812e-07, "loss": 0.062233902513980865, "step": 7303 }, { "epoch": 1.8961578400830739, "grad_norm": 1.5325415134429932, "learning_rate": 5.319034958575744e-07, "loss": 0.09186608344316483, "step": 7304 }, { "epoch": 1.8964174454828662, "grad_norm": 3.0689573287963867, "learning_rate": 5.317449493423419e-07, "loss": 0.0629660040140152, "step": 7305 }, { "epoch": 1.8966770508826585, "grad_norm": 2.781794309616089, "learning_rate": 5.315867949638114e-07, "loss": 0.05969161540269852, "step": 7306 }, { "epoch": 1.8969366562824508, "grad_norm": 1.877953290939331, "learning_rate": 5.314290327499404e-07, "loss": 0.07464419305324554, "step": 7307 }, { "epoch": 1.897196261682243, "grad_norm": 1.6039637327194214, "learning_rate": 5.312716627286179e-07, "loss": 0.10120820254087448, "step": 7308 }, { "epoch": 1.8974558670820354, "grad_norm": 1.4612491130828857, "learning_rate": 5.311146849276632e-07, "loss": 0.08870091289281845, "step": 7309 }, { "epoch": 1.8977154724818277, "grad_norm": 2.14309024810791, "learning_rate": 5.309580993748266e-07, "loss": 0.03820960968732834, "step": 7310 }, { "epoch": 1.89797507788162, "grad_norm": 1.2788382768630981, "learning_rate": 5.308019060977887e-07, "loss": 0.0973401889204979, "step": 7311 }, { "epoch": 1.8982346832814123, "grad_norm": 3.2110650539398193, "learning_rate": 5.306461051241607e-07, "loss": 0.03697246313095093, "step": 7312 }, { "epoch": 1.8984942886812046, "grad_norm": 1.1001335382461548, "learning_rate": 5.304906964814849e-07, "loss": 0.09799317270517349, "step": 7313 }, { "epoch": 1.898753894080997, "grad_norm": 1.2539924383163452, "learning_rate": 5.303356801972342e-07, "loss": 0.09916560351848602, "step": 7314 }, { "epoch": 1.8990134994807892, "grad_norm": 3.070831775665283, "learning_rate": 5.301810562988113e-07, "loss": 0.054396167397499084, "step": 7315 }, { "epoch": 1.8992731048805815, "grad_norm": 1.4021042585372925, "learning_rate": 5.300268248135509e-07, "loss": 0.09142239391803741, "step": 7316 }, { "epoch": 1.8995327102803738, "grad_norm": 2.2203733921051025, "learning_rate": 5.29872985768717e-07, "loss": 0.024209577590227127, "step": 7317 }, { "epoch": 1.8997923156801662, "grad_norm": 1.9415515661239624, "learning_rate": 5.297195391915051e-07, "loss": 0.026305673643946648, "step": 7318 }, { "epoch": 1.9000519210799585, "grad_norm": 3.5319623947143555, "learning_rate": 5.295664851090409e-07, "loss": 0.04753035306930542, "step": 7319 }, { "epoch": 1.9003115264797508, "grad_norm": 1.4749433994293213, "learning_rate": 5.29413823548381e-07, "loss": 0.07093380391597748, "step": 7320 }, { "epoch": 1.900571131879543, "grad_norm": 1.5096511840820312, "learning_rate": 5.292615545365124e-07, "loss": 0.06770461797714233, "step": 7321 }, { "epoch": 1.9008307372793354, "grad_norm": 1.7042044401168823, "learning_rate": 5.29109678100353e-07, "loss": 0.06726548075675964, "step": 7322 }, { "epoch": 1.9010903426791277, "grad_norm": 1.3512861728668213, "learning_rate": 5.289581942667505e-07, "loss": 0.08210764825344086, "step": 7323 }, { "epoch": 1.90134994807892, "grad_norm": 1.7200963497161865, "learning_rate": 5.288071030624846e-07, "loss": 0.0614427849650383, "step": 7324 }, { "epoch": 1.9016095534787123, "grad_norm": 1.1789770126342773, "learning_rate": 5.286564045142638e-07, "loss": 0.10214214026927948, "step": 7325 }, { "epoch": 1.9018691588785046, "grad_norm": 2.0870847702026367, "learning_rate": 5.28506098648729e-07, "loss": 0.06834931671619415, "step": 7326 }, { "epoch": 1.902128764278297, "grad_norm": 1.2893192768096924, "learning_rate": 5.283561854924504e-07, "loss": 0.11092295497655869, "step": 7327 }, { "epoch": 1.9023883696780892, "grad_norm": 2.7408673763275146, "learning_rate": 5.282066650719291e-07, "loss": 0.04016325995326042, "step": 7328 }, { "epoch": 1.9026479750778815, "grad_norm": 2.84108829498291, "learning_rate": 5.280575374135975e-07, "loss": 0.06079496070742607, "step": 7329 }, { "epoch": 1.9029075804776738, "grad_norm": 1.4154393672943115, "learning_rate": 5.279088025438173e-07, "loss": 0.08229338377714157, "step": 7330 }, { "epoch": 1.9031671858774661, "grad_norm": 2.1758434772491455, "learning_rate": 5.277604604888816e-07, "loss": 0.06233108043670654, "step": 7331 }, { "epoch": 1.9034267912772584, "grad_norm": 2.936866521835327, "learning_rate": 5.27612511275014e-07, "loss": 0.07831043004989624, "step": 7332 }, { "epoch": 1.9036863966770508, "grad_norm": 2.2642669677734375, "learning_rate": 5.274649549283685e-07, "loss": 0.09932490438222885, "step": 7333 }, { "epoch": 1.903946002076843, "grad_norm": 2.4284067153930664, "learning_rate": 5.273177914750298e-07, "loss": 0.07891280949115753, "step": 7334 }, { "epoch": 1.9042056074766354, "grad_norm": 2.3808603286743164, "learning_rate": 5.271710209410127e-07, "loss": 0.0393657311797142, "step": 7335 }, { "epoch": 1.9044652128764277, "grad_norm": 1.2457200288772583, "learning_rate": 5.270246433522629e-07, "loss": 0.10133376717567444, "step": 7336 }, { "epoch": 1.9047248182762202, "grad_norm": 1.512174129486084, "learning_rate": 5.268786587346573e-07, "loss": 0.07500813901424408, "step": 7337 }, { "epoch": 1.9049844236760125, "grad_norm": 2.781677484512329, "learning_rate": 5.267330671140016e-07, "loss": 0.07185329496860504, "step": 7338 }, { "epoch": 1.9052440290758048, "grad_norm": 2.280827283859253, "learning_rate": 5.265878685160339e-07, "loss": 0.05104751139879227, "step": 7339 }, { "epoch": 1.9055036344755971, "grad_norm": 2.3221986293792725, "learning_rate": 5.264430629664214e-07, "loss": 0.04679974168539047, "step": 7340 }, { "epoch": 1.9057632398753894, "grad_norm": 3.369584560394287, "learning_rate": 5.262986504907629e-07, "loss": 0.05632369592785835, "step": 7341 }, { "epoch": 1.9060228452751817, "grad_norm": 1.2186721563339233, "learning_rate": 5.261546311145868e-07, "loss": 0.11288312822580338, "step": 7342 }, { "epoch": 1.906282450674974, "grad_norm": 2.888965129852295, "learning_rate": 5.260110048633527e-07, "loss": 0.059428825974464417, "step": 7343 }, { "epoch": 1.9065420560747663, "grad_norm": 2.4870874881744385, "learning_rate": 5.258677717624502e-07, "loss": 0.05176474153995514, "step": 7344 }, { "epoch": 1.9068016614745587, "grad_norm": 2.4360010623931885, "learning_rate": 5.257249318372001e-07, "loss": 0.07006342709064484, "step": 7345 }, { "epoch": 1.907061266874351, "grad_norm": 2.0135912895202637, "learning_rate": 5.255824851128525e-07, "loss": 0.03261631727218628, "step": 7346 }, { "epoch": 1.9073208722741433, "grad_norm": 1.508127212524414, "learning_rate": 5.254404316145894e-07, "loss": 0.08954451978206635, "step": 7347 }, { "epoch": 1.9075804776739356, "grad_norm": 1.4061510562896729, "learning_rate": 5.252987713675222e-07, "loss": 0.08403737843036652, "step": 7348 }, { "epoch": 1.907840083073728, "grad_norm": 1.4047681093215942, "learning_rate": 5.251575043966932e-07, "loss": 0.1101854220032692, "step": 7349 }, { "epoch": 1.9080996884735204, "grad_norm": 1.5904147624969482, "learning_rate": 5.250166307270754e-07, "loss": 0.07885314524173737, "step": 7350 }, { "epoch": 1.9083592938733127, "grad_norm": 1.0841927528381348, "learning_rate": 5.24876150383572e-07, "loss": 0.10811731219291687, "step": 7351 }, { "epoch": 1.908618899273105, "grad_norm": 1.1947945356369019, "learning_rate": 5.247360633910168e-07, "loss": 0.08496154844760895, "step": 7352 }, { "epoch": 1.9088785046728973, "grad_norm": 2.0489258766174316, "learning_rate": 5.245963697741734e-07, "loss": 0.06690298765897751, "step": 7353 }, { "epoch": 1.9091381100726896, "grad_norm": 1.9883970022201538, "learning_rate": 5.244570695577371e-07, "loss": 0.062272727489471436, "step": 7354 }, { "epoch": 1.909397715472482, "grad_norm": 2.3504750728607178, "learning_rate": 5.243181627663325e-07, "loss": 0.05689443647861481, "step": 7355 }, { "epoch": 1.9096573208722742, "grad_norm": 2.681380271911621, "learning_rate": 5.241796494245157e-07, "loss": 0.05956236645579338, "step": 7356 }, { "epoch": 1.9099169262720666, "grad_norm": 3.0589776039123535, "learning_rate": 5.240415295567719e-07, "loss": 0.08989696949720383, "step": 7357 }, { "epoch": 1.9101765316718589, "grad_norm": 1.125362515449524, "learning_rate": 5.239038031875184e-07, "loss": 0.08893334120512009, "step": 7358 }, { "epoch": 1.9104361370716512, "grad_norm": 2.2526440620422363, "learning_rate": 5.237664703411014e-07, "loss": 0.05944515019655228, "step": 7359 }, { "epoch": 1.9106957424714435, "grad_norm": 3.824575901031494, "learning_rate": 5.236295310417988e-07, "loss": 0.06257858127355576, "step": 7360 }, { "epoch": 1.9109553478712358, "grad_norm": 2.7960352897644043, "learning_rate": 5.234929853138177e-07, "loss": 0.0687563344836235, "step": 7361 }, { "epoch": 1.911214953271028, "grad_norm": 2.1835412979125977, "learning_rate": 5.233568331812964e-07, "loss": 0.04515792429447174, "step": 7362 }, { "epoch": 1.9114745586708204, "grad_norm": 1.4524468183517456, "learning_rate": 5.232210746683039e-07, "loss": 0.069346584379673, "step": 7363 }, { "epoch": 1.9117341640706127, "grad_norm": 1.8972746133804321, "learning_rate": 5.230857097988388e-07, "loss": 0.026280798017978668, "step": 7364 }, { "epoch": 1.911993769470405, "grad_norm": 2.557520627975464, "learning_rate": 5.229507385968308e-07, "loss": 0.06953943520784378, "step": 7365 }, { "epoch": 1.9122533748701973, "grad_norm": 2.009683609008789, "learning_rate": 5.228161610861392e-07, "loss": 0.06257285177707672, "step": 7366 }, { "epoch": 1.9125129802699896, "grad_norm": 1.2628939151763916, "learning_rate": 5.226819772905546e-07, "loss": 0.08497118949890137, "step": 7367 }, { "epoch": 1.912772585669782, "grad_norm": 2.968550443649292, "learning_rate": 5.22548187233798e-07, "loss": 0.08603470027446747, "step": 7368 }, { "epoch": 1.9130321910695742, "grad_norm": 1.5163896083831787, "learning_rate": 5.224147909395197e-07, "loss": 0.03910179063677788, "step": 7369 }, { "epoch": 1.9132917964693665, "grad_norm": 2.1174867153167725, "learning_rate": 5.222817884313015e-07, "loss": 0.02769630402326584, "step": 7370 }, { "epoch": 1.9135514018691588, "grad_norm": 2.6205430030822754, "learning_rate": 5.22149179732655e-07, "loss": 0.08240695297718048, "step": 7371 }, { "epoch": 1.9138110072689511, "grad_norm": 1.359692096710205, "learning_rate": 5.220169648670227e-07, "loss": 0.09552737325429916, "step": 7372 }, { "epoch": 1.9140706126687435, "grad_norm": 3.4876861572265625, "learning_rate": 5.21885143857777e-07, "loss": 0.063310906291008, "step": 7373 }, { "epoch": 1.9143302180685358, "grad_norm": 1.2614545822143555, "learning_rate": 5.217537167282206e-07, "loss": 0.0868564248085022, "step": 7374 }, { "epoch": 1.914589823468328, "grad_norm": 3.41119647026062, "learning_rate": 5.21622683501587e-07, "loss": 0.07800083607435226, "step": 7375 }, { "epoch": 1.9148494288681204, "grad_norm": 1.9771524667739868, "learning_rate": 5.214920442010399e-07, "loss": 0.048428088426589966, "step": 7376 }, { "epoch": 1.9151090342679127, "grad_norm": 1.422064185142517, "learning_rate": 5.213617988496733e-07, "loss": 0.08883599936962128, "step": 7377 }, { "epoch": 1.915368639667705, "grad_norm": 1.781522512435913, "learning_rate": 5.212319474705119e-07, "loss": 0.07605449110269547, "step": 7378 }, { "epoch": 1.9156282450674973, "grad_norm": 2.017775297164917, "learning_rate": 5.211024900865099e-07, "loss": 0.06885462254285812, "step": 7379 }, { "epoch": 1.9158878504672896, "grad_norm": 2.528745174407959, "learning_rate": 5.209734267205526e-07, "loss": 0.04029364511370659, "step": 7380 }, { "epoch": 1.916147455867082, "grad_norm": 3.642747402191162, "learning_rate": 5.208447573954557e-07, "loss": 0.09927932918071747, "step": 7381 }, { "epoch": 1.9164070612668742, "grad_norm": 2.661914110183716, "learning_rate": 5.207164821339647e-07, "loss": 0.05807174742221832, "step": 7382 }, { "epoch": 1.9166666666666665, "grad_norm": 1.1881822347640991, "learning_rate": 5.205886009587559e-07, "loss": 0.10186739265918732, "step": 7383 }, { "epoch": 1.9169262720664588, "grad_norm": 2.011632204055786, "learning_rate": 5.204611138924358e-07, "loss": 0.026757966727018356, "step": 7384 }, { "epoch": 1.9171858774662511, "grad_norm": 2.3981263637542725, "learning_rate": 5.203340209575409e-07, "loss": 0.09261655807495117, "step": 7385 }, { "epoch": 1.9174454828660437, "grad_norm": 1.3305714130401611, "learning_rate": 5.202073221765384e-07, "loss": 0.09240797162055969, "step": 7386 }, { "epoch": 1.917705088265836, "grad_norm": 1.7605080604553223, "learning_rate": 5.20081017571826e-07, "loss": 0.07526083290576935, "step": 7387 }, { "epoch": 1.9179646936656283, "grad_norm": 2.2722580432891846, "learning_rate": 5.199551071657315e-07, "loss": 0.08422622084617615, "step": 7388 }, { "epoch": 1.9182242990654206, "grad_norm": 1.9785338640213013, "learning_rate": 5.198295909805126e-07, "loss": 0.07183951139450073, "step": 7389 }, { "epoch": 1.9184839044652129, "grad_norm": 2.084357500076294, "learning_rate": 5.197044690383577e-07, "loss": 0.05571846663951874, "step": 7390 }, { "epoch": 1.9187435098650052, "grad_norm": 2.277475357055664, "learning_rate": 5.195797413613861e-07, "loss": 0.055154718458652496, "step": 7391 }, { "epoch": 1.9190031152647975, "grad_norm": 4.108753681182861, "learning_rate": 5.194554079716461e-07, "loss": 0.05233587324619293, "step": 7392 }, { "epoch": 1.9192627206645898, "grad_norm": 2.2492237091064453, "learning_rate": 5.19331468891117e-07, "loss": 0.06534083932638168, "step": 7393 }, { "epoch": 1.9195223260643821, "grad_norm": 1.270282506942749, "learning_rate": 5.192079241417092e-07, "loss": 0.08770772814750671, "step": 7394 }, { "epoch": 1.9197819314641744, "grad_norm": 1.1165341138839722, "learning_rate": 5.190847737452617e-07, "loss": 0.0990588441491127, "step": 7395 }, { "epoch": 1.9200415368639667, "grad_norm": 1.210025668144226, "learning_rate": 5.189620177235454e-07, "loss": 0.09501416236162186, "step": 7396 }, { "epoch": 1.920301142263759, "grad_norm": 1.8121867179870605, "learning_rate": 5.1883965609826e-07, "loss": 0.08437502384185791, "step": 7397 }, { "epoch": 1.9205607476635516, "grad_norm": 1.5773847103118896, "learning_rate": 5.187176888910366e-07, "loss": 0.05088964104652405, "step": 7398 }, { "epoch": 1.9208203530633439, "grad_norm": 4.583192348480225, "learning_rate": 5.185961161234364e-07, "loss": 0.04454783350229263, "step": 7399 }, { "epoch": 1.9210799584631362, "grad_norm": 1.62906813621521, "learning_rate": 5.184749378169505e-07, "loss": 0.07819102704524994, "step": 7400 }, { "epoch": 1.9213395638629285, "grad_norm": 2.5082995891571045, "learning_rate": 5.183541539930002e-07, "loss": 0.03783271089196205, "step": 7401 }, { "epoch": 1.9215991692627208, "grad_norm": 2.697598695755005, "learning_rate": 5.182337646729378e-07, "loss": 0.020590487867593765, "step": 7402 }, { "epoch": 1.921858774662513, "grad_norm": 2.7785496711730957, "learning_rate": 5.18113769878045e-07, "loss": 0.03712283819913864, "step": 7403 }, { "epoch": 1.9221183800623054, "grad_norm": 1.9113543033599854, "learning_rate": 5.179941696295345e-07, "loss": 0.07444076985120773, "step": 7404 }, { "epoch": 1.9223779854620977, "grad_norm": 1.4848120212554932, "learning_rate": 5.178749639485482e-07, "loss": 0.08123272657394409, "step": 7405 }, { "epoch": 1.92263759086189, "grad_norm": 2.4166269302368164, "learning_rate": 5.177561528561597e-07, "loss": 0.07889614254236221, "step": 7406 }, { "epoch": 1.9228971962616823, "grad_norm": 1.6422795057296753, "learning_rate": 5.176377363733718e-07, "loss": 0.07773641496896744, "step": 7407 }, { "epoch": 1.9231568016614746, "grad_norm": 2.2995762825012207, "learning_rate": 5.175197145211176e-07, "loss": 0.05912020057439804, "step": 7408 }, { "epoch": 1.923416407061267, "grad_norm": 3.303053379058838, "learning_rate": 5.174020873202611e-07, "loss": 0.040612898766994476, "step": 7409 }, { "epoch": 1.9236760124610592, "grad_norm": 1.916748285293579, "learning_rate": 5.172848547915956e-07, "loss": 0.04769746586680412, "step": 7410 }, { "epoch": 1.9239356178608515, "grad_norm": 2.2321503162384033, "learning_rate": 5.171680169558456e-07, "loss": 0.057256974279880524, "step": 7411 }, { "epoch": 1.9241952232606438, "grad_norm": 3.0355920791625977, "learning_rate": 5.170515738336647e-07, "loss": 0.07254397124052048, "step": 7412 }, { "epoch": 1.9244548286604362, "grad_norm": 3.5463759899139404, "learning_rate": 5.169355254456381e-07, "loss": 0.041633859276771545, "step": 7413 }, { "epoch": 1.9247144340602285, "grad_norm": 2.845258951187134, "learning_rate": 5.168198718122804e-07, "loss": 0.05051680654287338, "step": 7414 }, { "epoch": 1.9249740394600208, "grad_norm": 1.2929636240005493, "learning_rate": 5.167046129540361e-07, "loss": 0.09107039868831635, "step": 7415 }, { "epoch": 1.925233644859813, "grad_norm": 1.775193452835083, "learning_rate": 5.165897488912805e-07, "loss": 0.07279209792613983, "step": 7416 }, { "epoch": 1.9254932502596054, "grad_norm": 1.356252908706665, "learning_rate": 5.164752796443191e-07, "loss": 0.04780589044094086, "step": 7417 }, { "epoch": 1.9257528556593977, "grad_norm": 2.1206045150756836, "learning_rate": 5.163612052333871e-07, "loss": 0.06201550364494324, "step": 7418 }, { "epoch": 1.92601246105919, "grad_norm": 3.3220245838165283, "learning_rate": 5.162475256786508e-07, "loss": 0.03131671994924545, "step": 7419 }, { "epoch": 1.9262720664589823, "grad_norm": 1.514266848564148, "learning_rate": 5.161342410002055e-07, "loss": 0.08724634349346161, "step": 7420 }, { "epoch": 1.9265316718587746, "grad_norm": 3.3586061000823975, "learning_rate": 5.160213512180778e-07, "loss": 0.01501481980085373, "step": 7421 }, { "epoch": 1.926791277258567, "grad_norm": 1.8846315145492554, "learning_rate": 5.159088563522239e-07, "loss": 0.08963146060705185, "step": 7422 }, { "epoch": 1.9270508826583592, "grad_norm": 2.3328566551208496, "learning_rate": 5.157967564225303e-07, "loss": 0.04375319182872772, "step": 7423 }, { "epoch": 1.9273104880581515, "grad_norm": 2.3030717372894287, "learning_rate": 5.156850514488136e-07, "loss": 0.05326252430677414, "step": 7424 }, { "epoch": 1.9275700934579438, "grad_norm": 1.751187801361084, "learning_rate": 5.155737414508208e-07, "loss": 0.07261815667152405, "step": 7425 }, { "epoch": 1.9278296988577361, "grad_norm": 1.9122910499572754, "learning_rate": 5.154628264482289e-07, "loss": 0.05953161418437958, "step": 7426 }, { "epoch": 1.9280893042575284, "grad_norm": 1.2129971981048584, "learning_rate": 5.153523064606452e-07, "loss": 0.10255911946296692, "step": 7427 }, { "epoch": 1.9283489096573208, "grad_norm": 1.5634913444519043, "learning_rate": 5.15242181507607e-07, "loss": 0.09889692813158035, "step": 7428 }, { "epoch": 1.928608515057113, "grad_norm": 1.7445446252822876, "learning_rate": 5.151324516085819e-07, "loss": 0.09820430725812912, "step": 7429 }, { "epoch": 1.9288681204569054, "grad_norm": 1.2181981801986694, "learning_rate": 5.15023116782968e-07, "loss": 0.10337254405021667, "step": 7430 }, { "epoch": 1.9291277258566977, "grad_norm": 2.4693965911865234, "learning_rate": 5.149141770500926e-07, "loss": 0.07985658943653107, "step": 7431 }, { "epoch": 1.92938733125649, "grad_norm": 1.1890647411346436, "learning_rate": 5.148056324292142e-07, "loss": 0.041432637721300125, "step": 7432 }, { "epoch": 1.9296469366562823, "grad_norm": 2.224872350692749, "learning_rate": 5.146974829395207e-07, "loss": 0.033287905156612396, "step": 7433 }, { "epoch": 1.9299065420560748, "grad_norm": 2.1099202632904053, "learning_rate": 5.145897286001304e-07, "loss": 0.08617336302995682, "step": 7434 }, { "epoch": 1.9301661474558671, "grad_norm": 1.30787992477417, "learning_rate": 5.144823694300921e-07, "loss": 0.0943697988986969, "step": 7435 }, { "epoch": 1.9304257528556594, "grad_norm": 2.6678545475006104, "learning_rate": 5.143754054483844e-07, "loss": 0.03302524983882904, "step": 7436 }, { "epoch": 1.9306853582554517, "grad_norm": 2.943755865097046, "learning_rate": 5.14268836673916e-07, "loss": 0.08059057593345642, "step": 7437 }, { "epoch": 1.930944963655244, "grad_norm": 2.477290391921997, "learning_rate": 5.141626631255257e-07, "loss": 0.0654904693365097, "step": 7438 }, { "epoch": 1.9312045690550363, "grad_norm": 2.708751916885376, "learning_rate": 5.140568848219826e-07, "loss": 0.06151711195707321, "step": 7439 }, { "epoch": 1.9314641744548287, "grad_norm": 2.734832525253296, "learning_rate": 5.139515017819862e-07, "loss": 0.06402048468589783, "step": 7440 }, { "epoch": 1.931723779854621, "grad_norm": 2.076617479324341, "learning_rate": 5.138465140241657e-07, "loss": 0.06776285916566849, "step": 7441 }, { "epoch": 1.9319833852544133, "grad_norm": 3.8182554244995117, "learning_rate": 5.137419215670798e-07, "loss": 0.08036233484745026, "step": 7442 }, { "epoch": 1.9322429906542056, "grad_norm": 1.2741726636886597, "learning_rate": 5.136377244292192e-07, "loss": 0.09558694064617157, "step": 7443 }, { "epoch": 1.9325025960539979, "grad_norm": 2.108386516571045, "learning_rate": 5.135339226290027e-07, "loss": 0.036678213626146317, "step": 7444 }, { "epoch": 1.9327622014537902, "grad_norm": 2.2269363403320312, "learning_rate": 5.134305161847805e-07, "loss": 0.0893649160861969, "step": 7445 }, { "epoch": 1.9330218068535827, "grad_norm": 3.0102484226226807, "learning_rate": 5.133275051148322e-07, "loss": 0.04447717219591141, "step": 7446 }, { "epoch": 1.933281412253375, "grad_norm": 2.0427098274230957, "learning_rate": 5.132248894373682e-07, "loss": 0.06890727579593658, "step": 7447 }, { "epoch": 1.9335410176531673, "grad_norm": 1.5284013748168945, "learning_rate": 5.131226691705285e-07, "loss": 0.09061598032712936, "step": 7448 }, { "epoch": 1.9338006230529596, "grad_norm": 1.3402003049850464, "learning_rate": 5.130208443323829e-07, "loss": 0.10498829185962677, "step": 7449 }, { "epoch": 1.934060228452752, "grad_norm": 2.1373283863067627, "learning_rate": 5.129194149409321e-07, "loss": 0.05061047896742821, "step": 7450 }, { "epoch": 1.9343198338525442, "grad_norm": 1.2085217237472534, "learning_rate": 5.128183810141066e-07, "loss": 0.09565811604261398, "step": 7451 }, { "epoch": 1.9345794392523366, "grad_norm": 1.7748862504959106, "learning_rate": 5.127177425697662e-07, "loss": 0.05065179616212845, "step": 7452 }, { "epoch": 1.9348390446521289, "grad_norm": 1.1731985807418823, "learning_rate": 5.126174996257024e-07, "loss": 0.0862247422337532, "step": 7453 }, { "epoch": 1.9350986500519212, "grad_norm": 1.8475030660629272, "learning_rate": 5.125176521996353e-07, "loss": 0.07912188023328781, "step": 7454 }, { "epoch": 1.9353582554517135, "grad_norm": 1.8100453615188599, "learning_rate": 5.124182003092153e-07, "loss": 0.09942914545536041, "step": 7455 }, { "epoch": 1.9356178608515058, "grad_norm": 2.025348663330078, "learning_rate": 5.123191439720241e-07, "loss": 0.05948314815759659, "step": 7456 }, { "epoch": 1.935877466251298, "grad_norm": 1.8599539995193481, "learning_rate": 5.122204832055723e-07, "loss": 0.01715906523168087, "step": 7457 }, { "epoch": 1.9361370716510904, "grad_norm": 1.462825894355774, "learning_rate": 5.121222180273004e-07, "loss": 0.1081211268901825, "step": 7458 }, { "epoch": 1.9363966770508827, "grad_norm": 2.742733955383301, "learning_rate": 5.120243484545798e-07, "loss": 0.06105063110589981, "step": 7459 }, { "epoch": 1.936656282450675, "grad_norm": 2.097278356552124, "learning_rate": 5.119268745047115e-07, "loss": 0.07616496086120605, "step": 7460 }, { "epoch": 1.9369158878504673, "grad_norm": 1.6929700374603271, "learning_rate": 5.118297961949268e-07, "loss": 0.05828595906496048, "step": 7461 }, { "epoch": 1.9371754932502596, "grad_norm": 1.5422446727752686, "learning_rate": 5.117331135423866e-07, "loss": 0.06523942947387695, "step": 7462 }, { "epoch": 1.937435098650052, "grad_norm": 1.5287466049194336, "learning_rate": 5.116368265641828e-07, "loss": 0.0853840559720993, "step": 7463 }, { "epoch": 1.9376947040498442, "grad_norm": 1.6494321823120117, "learning_rate": 5.115409352773361e-07, "loss": 0.08693838119506836, "step": 7464 }, { "epoch": 1.9379543094496365, "grad_norm": 3.8782572746276855, "learning_rate": 5.11445439698798e-07, "loss": 0.022143134847283363, "step": 7465 }, { "epoch": 1.9382139148494288, "grad_norm": 1.6686431169509888, "learning_rate": 5.113503398454503e-07, "loss": 0.0902959406375885, "step": 7466 }, { "epoch": 1.9384735202492211, "grad_norm": 1.5577163696289062, "learning_rate": 5.11255635734104e-07, "loss": 0.08171379566192627, "step": 7467 }, { "epoch": 1.9387331256490135, "grad_norm": 2.979421615600586, "learning_rate": 5.11161327381501e-07, "loss": 0.03382732346653938, "step": 7468 }, { "epoch": 1.9389927310488058, "grad_norm": 1.476989984512329, "learning_rate": 5.110674148043129e-07, "loss": 0.08709333091974258, "step": 7469 }, { "epoch": 1.939252336448598, "grad_norm": 3.280722141265869, "learning_rate": 5.109738980191408e-07, "loss": 0.06503184139728546, "step": 7470 }, { "epoch": 1.9395119418483904, "grad_norm": 3.15346360206604, "learning_rate": 5.108807770425167e-07, "loss": 0.06227519363164902, "step": 7471 }, { "epoch": 1.9397715472481827, "grad_norm": 2.262608766555786, "learning_rate": 5.107880518909023e-07, "loss": 0.06694810837507248, "step": 7472 }, { "epoch": 1.940031152647975, "grad_norm": 1.4982975721359253, "learning_rate": 5.106957225806889e-07, "loss": 0.09782378375530243, "step": 7473 }, { "epoch": 1.9402907580477673, "grad_norm": 3.339843511581421, "learning_rate": 5.106037891281988e-07, "loss": 0.03974422439932823, "step": 7474 }, { "epoch": 1.9405503634475596, "grad_norm": 2.978393077850342, "learning_rate": 5.105122515496832e-07, "loss": 0.016780471429228783, "step": 7475 }, { "epoch": 1.940809968847352, "grad_norm": 3.4753191471099854, "learning_rate": 5.10421109861324e-07, "loss": 0.04750813543796539, "step": 7476 }, { "epoch": 1.9410695742471442, "grad_norm": 2.097590208053589, "learning_rate": 5.103303640792333e-07, "loss": 0.06712307035923004, "step": 7477 }, { "epoch": 1.9413291796469365, "grad_norm": 1.9140722751617432, "learning_rate": 5.102400142194524e-07, "loss": 0.07431228458881378, "step": 7478 }, { "epoch": 1.9415887850467288, "grad_norm": 2.560793161392212, "learning_rate": 5.101500602979533e-07, "loss": 0.028833594173192978, "step": 7479 }, { "epoch": 1.9418483904465211, "grad_norm": 3.210181474685669, "learning_rate": 5.100605023306377e-07, "loss": 0.08556883037090302, "step": 7480 }, { "epoch": 1.9421079958463134, "grad_norm": 3.605156421661377, "learning_rate": 5.099713403333375e-07, "loss": 0.036633338779211044, "step": 7481 }, { "epoch": 1.942367601246106, "grad_norm": 1.9508568048477173, "learning_rate": 5.098825743218146e-07, "loss": 0.0895804613828659, "step": 7482 }, { "epoch": 1.9426272066458983, "grad_norm": 1.7651793956756592, "learning_rate": 5.097942043117604e-07, "loss": 0.07618942856788635, "step": 7483 }, { "epoch": 1.9428868120456906, "grad_norm": 2.255542755126953, "learning_rate": 5.097062303187973e-07, "loss": 0.060614585876464844, "step": 7484 }, { "epoch": 1.9431464174454829, "grad_norm": 1.9553011655807495, "learning_rate": 5.096186523584765e-07, "loss": 0.09202969074249268, "step": 7485 }, { "epoch": 1.9434060228452752, "grad_norm": 3.3904924392700195, "learning_rate": 5.095314704462803e-07, "loss": 0.06726758927106857, "step": 7486 }, { "epoch": 1.9436656282450675, "grad_norm": 1.1866285800933838, "learning_rate": 5.0944468459762e-07, "loss": 0.0946442186832428, "step": 7487 }, { "epoch": 1.9439252336448598, "grad_norm": 1.762980341911316, "learning_rate": 5.093582948278375e-07, "loss": 0.07031647861003876, "step": 7488 }, { "epoch": 1.9441848390446521, "grad_norm": 1.5390465259552002, "learning_rate": 5.092723011522047e-07, "loss": 0.10103723406791687, "step": 7489 }, { "epoch": 1.9444444444444444, "grad_norm": 1.9893646240234375, "learning_rate": 5.091867035859232e-07, "loss": 0.06646261364221573, "step": 7490 }, { "epoch": 1.9447040498442367, "grad_norm": 2.2734646797180176, "learning_rate": 5.091015021441246e-07, "loss": 0.06217927858233452, "step": 7491 }, { "epoch": 1.944963655244029, "grad_norm": 2.4813387393951416, "learning_rate": 5.090166968418707e-07, "loss": 0.0365162268280983, "step": 7492 }, { "epoch": 1.9452232606438213, "grad_norm": 2.2427237033843994, "learning_rate": 5.08932287694153e-07, "loss": 0.017949027940630913, "step": 7493 }, { "epoch": 1.9454828660436139, "grad_norm": 2.4431867599487305, "learning_rate": 5.088482747158933e-07, "loss": 0.046446558088064194, "step": 7494 }, { "epoch": 1.9457424714434062, "grad_norm": 2.065382957458496, "learning_rate": 5.087646579219432e-07, "loss": 0.08727556467056274, "step": 7495 }, { "epoch": 1.9460020768431985, "grad_norm": 2.3444435596466064, "learning_rate": 5.086814373270836e-07, "loss": 0.026540406048297882, "step": 7496 }, { "epoch": 1.9462616822429908, "grad_norm": 1.445884346961975, "learning_rate": 5.085986129460268e-07, "loss": 0.033106882125139236, "step": 7497 }, { "epoch": 1.946521287642783, "grad_norm": 2.7622883319854736, "learning_rate": 5.085161847934141e-07, "loss": 0.03924419730901718, "step": 7498 }, { "epoch": 1.9467808930425754, "grad_norm": 1.5799787044525146, "learning_rate": 5.084341528838165e-07, "loss": 0.08269000053405762, "step": 7499 }, { "epoch": 1.9470404984423677, "grad_norm": 2.296159505844116, "learning_rate": 5.083525172317356e-07, "loss": 0.04104591906070709, "step": 7500 }, { "epoch": 1.94730010384216, "grad_norm": 1.9673917293548584, "learning_rate": 5.082712778516029e-07, "loss": 0.035898059606552124, "step": 7501 }, { "epoch": 1.9475597092419523, "grad_norm": 1.7033684253692627, "learning_rate": 5.081904347577796e-07, "loss": 0.09063739329576492, "step": 7502 }, { "epoch": 1.9478193146417446, "grad_norm": 3.9672365188598633, "learning_rate": 5.081099879645564e-07, "loss": 0.03772086277604103, "step": 7503 }, { "epoch": 1.948078920041537, "grad_norm": 3.003079652786255, "learning_rate": 5.080299374861551e-07, "loss": 0.045853447169065475, "step": 7504 }, { "epoch": 1.9483385254413292, "grad_norm": 1.757948637008667, "learning_rate": 5.079502833367265e-07, "loss": 0.08225766569375992, "step": 7505 }, { "epoch": 1.9485981308411215, "grad_norm": 1.269392728805542, "learning_rate": 5.078710255303515e-07, "loss": 0.08725916594266891, "step": 7506 }, { "epoch": 1.9488577362409139, "grad_norm": 1.509224534034729, "learning_rate": 5.077921640810416e-07, "loss": 0.09289193898439407, "step": 7507 }, { "epoch": 1.9491173416407062, "grad_norm": 2.171570301055908, "learning_rate": 5.077136990027372e-07, "loss": 0.07104922831058502, "step": 7508 }, { "epoch": 1.9493769470404985, "grad_norm": 2.0105786323547363, "learning_rate": 5.076356303093091e-07, "loss": 0.024783752858638763, "step": 7509 }, { "epoch": 1.9496365524402908, "grad_norm": 1.668602466583252, "learning_rate": 5.075579580145585e-07, "loss": 0.026493441313505173, "step": 7510 }, { "epoch": 1.949896157840083, "grad_norm": 1.8643462657928467, "learning_rate": 5.074806821322155e-07, "loss": 0.0673501044511795, "step": 7511 }, { "epoch": 1.9501557632398754, "grad_norm": 1.1696885824203491, "learning_rate": 5.074038026759412e-07, "loss": 0.10836585611104965, "step": 7512 }, { "epoch": 1.9504153686396677, "grad_norm": 2.2405447959899902, "learning_rate": 5.07327319659326e-07, "loss": 0.051977984607219696, "step": 7513 }, { "epoch": 1.95067497403946, "grad_norm": 2.540788412094116, "learning_rate": 5.072512330958902e-07, "loss": 0.039536021649837494, "step": 7514 }, { "epoch": 1.9509345794392523, "grad_norm": 1.2329416275024414, "learning_rate": 5.071755429990844e-07, "loss": 0.10099165886640549, "step": 7515 }, { "epoch": 1.9511941848390446, "grad_norm": 2.139153242111206, "learning_rate": 5.071002493822885e-07, "loss": 0.05752935633063316, "step": 7516 }, { "epoch": 1.951453790238837, "grad_norm": 1.9524208307266235, "learning_rate": 5.070253522588128e-07, "loss": 0.02793625183403492, "step": 7517 }, { "epoch": 1.9517133956386292, "grad_norm": 3.0072760581970215, "learning_rate": 5.06950851641898e-07, "loss": 0.03399357944726944, "step": 7518 }, { "epoch": 1.9519730010384215, "grad_norm": 3.0265250205993652, "learning_rate": 5.068767475447131e-07, "loss": 0.0445338636636734, "step": 7519 }, { "epoch": 1.9522326064382138, "grad_norm": 1.7633473873138428, "learning_rate": 5.068030399803587e-07, "loss": 0.09469695389270782, "step": 7520 }, { "epoch": 1.9524922118380061, "grad_norm": 1.679100751876831, "learning_rate": 5.067297289618643e-07, "loss": 0.06835145503282547, "step": 7521 }, { "epoch": 1.9527518172377984, "grad_norm": 1.2087763547897339, "learning_rate": 5.0665681450219e-07, "loss": 0.11058487743139267, "step": 7522 }, { "epoch": 1.9530114226375908, "grad_norm": 1.4427231550216675, "learning_rate": 5.065842966142251e-07, "loss": 0.08915063738822937, "step": 7523 }, { "epoch": 1.953271028037383, "grad_norm": 3.378633499145508, "learning_rate": 5.065121753107889e-07, "loss": 0.050291966646909714, "step": 7524 }, { "epoch": 1.9535306334371754, "grad_norm": 1.4112430810928345, "learning_rate": 5.064404506046314e-07, "loss": 0.09696569293737411, "step": 7525 }, { "epoch": 1.9537902388369677, "grad_norm": 2.253882884979248, "learning_rate": 5.063691225084313e-07, "loss": 0.04487059637904167, "step": 7526 }, { "epoch": 1.95404984423676, "grad_norm": 2.4661238193511963, "learning_rate": 5.062981910347979e-07, "loss": 0.03651702031493187, "step": 7527 }, { "epoch": 1.9543094496365523, "grad_norm": 2.4150257110595703, "learning_rate": 5.062276561962705e-07, "loss": 0.0599394366145134, "step": 7528 }, { "epoch": 1.9545690550363446, "grad_norm": 1.5239005088806152, "learning_rate": 5.06157518005318e-07, "loss": 0.09399125725030899, "step": 7529 }, { "epoch": 1.9548286604361371, "grad_norm": 2.9189600944519043, "learning_rate": 5.060877764743391e-07, "loss": 0.03539860248565674, "step": 7530 }, { "epoch": 1.9550882658359294, "grad_norm": 3.595968723297119, "learning_rate": 5.060184316156627e-07, "loss": 0.07753909379243851, "step": 7531 }, { "epoch": 1.9553478712357217, "grad_norm": 2.3016469478607178, "learning_rate": 5.059494834415468e-07, "loss": 0.016769424080848694, "step": 7532 }, { "epoch": 1.955607476635514, "grad_norm": 2.0601389408111572, "learning_rate": 5.058809319641804e-07, "loss": 0.03694761544466019, "step": 7533 }, { "epoch": 1.9558670820353063, "grad_norm": 3.9383366107940674, "learning_rate": 5.05812777195682e-07, "loss": 0.060100022703409195, "step": 7534 }, { "epoch": 1.9561266874350987, "grad_norm": 2.03096866607666, "learning_rate": 5.057450191480994e-07, "loss": 0.03948241472244263, "step": 7535 }, { "epoch": 1.956386292834891, "grad_norm": 2.6057703495025635, "learning_rate": 5.056776578334107e-07, "loss": 0.046893730759620667, "step": 7536 }, { "epoch": 1.9566458982346833, "grad_norm": 2.487269163131714, "learning_rate": 5.056106932635236e-07, "loss": 0.05304668843746185, "step": 7537 }, { "epoch": 1.9569055036344756, "grad_norm": 1.4954187870025635, "learning_rate": 5.055441254502768e-07, "loss": 0.09550328552722931, "step": 7538 }, { "epoch": 1.9571651090342679, "grad_norm": 2.7982921600341797, "learning_rate": 5.054779544054372e-07, "loss": 0.057979121804237366, "step": 7539 }, { "epoch": 1.9574247144340602, "grad_norm": 1.8020809888839722, "learning_rate": 5.054121801407024e-07, "loss": 0.02779519185423851, "step": 7540 }, { "epoch": 1.9576843198338525, "grad_norm": 1.6915472745895386, "learning_rate": 5.053468026677003e-07, "loss": 0.08900544047355652, "step": 7541 }, { "epoch": 1.957943925233645, "grad_norm": 5.13455057144165, "learning_rate": 5.052818219979871e-07, "loss": 0.07745469361543655, "step": 7542 }, { "epoch": 1.9582035306334373, "grad_norm": 1.5240693092346191, "learning_rate": 5.05217238143051e-07, "loss": 0.10416094958782196, "step": 7543 }, { "epoch": 1.9584631360332296, "grad_norm": 2.966235399246216, "learning_rate": 5.051530511143083e-07, "loss": 0.042550019919872284, "step": 7544 }, { "epoch": 1.958722741433022, "grad_norm": 1.4786360263824463, "learning_rate": 5.050892609231061e-07, "loss": 0.10275860130786896, "step": 7545 }, { "epoch": 1.9589823468328142, "grad_norm": 2.630980968475342, "learning_rate": 5.050258675807208e-07, "loss": 0.052731581032276154, "step": 7546 }, { "epoch": 1.9592419522326066, "grad_norm": 2.6576735973358154, "learning_rate": 5.049628710983592e-07, "loss": 0.03757006302475929, "step": 7547 }, { "epoch": 1.9595015576323989, "grad_norm": 2.603917121887207, "learning_rate": 5.049002714871572e-07, "loss": 0.0687759593129158, "step": 7548 }, { "epoch": 1.9597611630321912, "grad_norm": 3.020347833633423, "learning_rate": 5.048380687581814e-07, "loss": 0.0635070651769638, "step": 7549 }, { "epoch": 1.9600207684319835, "grad_norm": 1.3563027381896973, "learning_rate": 5.047762629224276e-07, "loss": 0.09171169251203537, "step": 7550 }, { "epoch": 1.9602803738317758, "grad_norm": 2.068349838256836, "learning_rate": 5.047148539908217e-07, "loss": 0.07570590078830719, "step": 7551 }, { "epoch": 1.960539979231568, "grad_norm": 2.875526189804077, "learning_rate": 5.046538419742194e-07, "loss": 0.04397118464112282, "step": 7552 }, { "epoch": 1.9607995846313604, "grad_norm": 2.7065370082855225, "learning_rate": 5.045932268834061e-07, "loss": 0.056687526404857635, "step": 7553 }, { "epoch": 1.9610591900311527, "grad_norm": 1.2517879009246826, "learning_rate": 5.045330087290976e-07, "loss": 0.08345586061477661, "step": 7554 }, { "epoch": 1.961318795430945, "grad_norm": 1.6050739288330078, "learning_rate": 5.044731875219385e-07, "loss": 0.0800209790468216, "step": 7555 }, { "epoch": 1.9615784008307373, "grad_norm": 4.487667083740234, "learning_rate": 5.044137632725041e-07, "loss": 0.06676207482814789, "step": 7556 }, { "epoch": 1.9618380062305296, "grad_norm": 2.733078718185425, "learning_rate": 5.043547359912995e-07, "loss": 0.046076685190200806, "step": 7557 }, { "epoch": 1.962097611630322, "grad_norm": 2.79866361618042, "learning_rate": 5.042961056887591e-07, "loss": 0.06848785281181335, "step": 7558 }, { "epoch": 1.9623572170301142, "grad_norm": 1.4952201843261719, "learning_rate": 5.042378723752473e-07, "loss": 0.10881800949573517, "step": 7559 }, { "epoch": 1.9626168224299065, "grad_norm": 1.87258780002594, "learning_rate": 5.041800360610583e-07, "loss": 0.05019083991646767, "step": 7560 }, { "epoch": 1.9628764278296988, "grad_norm": 2.3393609523773193, "learning_rate": 5.041225967564167e-07, "loss": 0.0566113144159317, "step": 7561 }, { "epoch": 1.9631360332294912, "grad_norm": 3.2326438426971436, "learning_rate": 5.040655544714762e-07, "loss": 0.018337026238441467, "step": 7562 }, { "epoch": 1.9633956386292835, "grad_norm": 1.7086124420166016, "learning_rate": 5.040089092163204e-07, "loss": 0.0665978491306305, "step": 7563 }, { "epoch": 1.9636552440290758, "grad_norm": 1.3855236768722534, "learning_rate": 5.039526610009634e-07, "loss": 0.08218804001808167, "step": 7564 }, { "epoch": 1.963914849428868, "grad_norm": 1.531093955039978, "learning_rate": 5.038968098353482e-07, "loss": 0.09255607426166534, "step": 7565 }, { "epoch": 1.9641744548286604, "grad_norm": 1.8990371227264404, "learning_rate": 5.038413557293478e-07, "loss": 0.09900495409965515, "step": 7566 }, { "epoch": 1.9644340602284527, "grad_norm": 1.921247124671936, "learning_rate": 5.037862986927657e-07, "loss": 0.07436146587133408, "step": 7567 }, { "epoch": 1.964693665628245, "grad_norm": 1.7629841566085815, "learning_rate": 5.037316387353347e-07, "loss": 0.0978461354970932, "step": 7568 }, { "epoch": 1.9649532710280373, "grad_norm": 2.7069833278656006, "learning_rate": 5.036773758667168e-07, "loss": 0.03954937309026718, "step": 7569 }, { "epoch": 1.9652128764278296, "grad_norm": 1.8266985416412354, "learning_rate": 5.036235100965053e-07, "loss": 0.09570728242397308, "step": 7570 }, { "epoch": 1.965472481827622, "grad_norm": 1.593684434890747, "learning_rate": 5.035700414342218e-07, "loss": 0.08994805067777634, "step": 7571 }, { "epoch": 1.9657320872274142, "grad_norm": 1.7650514841079712, "learning_rate": 5.035169698893185e-07, "loss": 0.0661553144454956, "step": 7572 }, { "epoch": 1.9659916926272065, "grad_norm": 2.2013416290283203, "learning_rate": 5.034642954711773e-07, "loss": 0.03207538276910782, "step": 7573 }, { "epoch": 1.9662512980269988, "grad_norm": 1.2607561349868774, "learning_rate": 5.034120181891097e-07, "loss": 0.0718233585357666, "step": 7574 }, { "epoch": 1.9665109034267911, "grad_norm": 1.184170126914978, "learning_rate": 5.033601380523576e-07, "loss": 0.0974072739481926, "step": 7575 }, { "epoch": 1.9667705088265834, "grad_norm": 2.2946536540985107, "learning_rate": 5.033086550700915e-07, "loss": 0.022976232692599297, "step": 7576 }, { "epoch": 1.9670301142263757, "grad_norm": 1.383217215538025, "learning_rate": 5.03257569251413e-07, "loss": 0.09991490840911865, "step": 7577 }, { "epoch": 1.9672897196261683, "grad_norm": 1.833756685256958, "learning_rate": 5.032068806053528e-07, "loss": 0.026093047112226486, "step": 7578 }, { "epoch": 1.9675493250259606, "grad_norm": 1.6224445104599, "learning_rate": 5.031565891408709e-07, "loss": 0.08316491544246674, "step": 7579 }, { "epoch": 1.967808930425753, "grad_norm": 2.4221601486206055, "learning_rate": 5.031066948668585e-07, "loss": 0.04535667225718498, "step": 7580 }, { "epoch": 1.9680685358255452, "grad_norm": 2.7136318683624268, "learning_rate": 5.030571977921354e-07, "loss": 0.04252788424491882, "step": 7581 }, { "epoch": 1.9683281412253375, "grad_norm": 2.617201328277588, "learning_rate": 5.030080979254515e-07, "loss": 0.06463600695133209, "step": 7582 }, { "epoch": 1.9685877466251298, "grad_norm": 2.8724143505096436, "learning_rate": 5.029593952754865e-07, "loss": 0.061590246856212616, "step": 7583 }, { "epoch": 1.9688473520249221, "grad_norm": 2.095482587814331, "learning_rate": 5.029110898508502e-07, "loss": 0.09164990484714508, "step": 7584 }, { "epoch": 1.9691069574247144, "grad_norm": 1.572028398513794, "learning_rate": 5.028631816600815e-07, "loss": 0.0790092945098877, "step": 7585 }, { "epoch": 1.9693665628245067, "grad_norm": 3.2916452884674072, "learning_rate": 5.028156707116497e-07, "loss": 0.06480664014816284, "step": 7586 }, { "epoch": 1.969626168224299, "grad_norm": 3.6320202350616455, "learning_rate": 5.027685570139535e-07, "loss": 0.03954426571726799, "step": 7587 }, { "epoch": 1.9698857736240913, "grad_norm": 2.049487590789795, "learning_rate": 5.02721840575322e-07, "loss": 0.07202333211898804, "step": 7588 }, { "epoch": 1.9701453790238836, "grad_norm": 3.186851978302002, "learning_rate": 5.02675521404013e-07, "loss": 0.055566735565662384, "step": 7589 }, { "epoch": 1.9704049844236762, "grad_norm": 1.823207974433899, "learning_rate": 5.026295995082151e-07, "loss": 0.08600667119026184, "step": 7590 }, { "epoch": 1.9706645898234685, "grad_norm": 1.948906421661377, "learning_rate": 5.025840748960458e-07, "loss": 0.06813022494316101, "step": 7591 }, { "epoch": 1.9709241952232608, "grad_norm": 1.6774576902389526, "learning_rate": 5.025389475755532e-07, "loss": 0.055470213294029236, "step": 7592 }, { "epoch": 1.971183800623053, "grad_norm": 1.4019930362701416, "learning_rate": 5.024942175547146e-07, "loss": 0.09219971299171448, "step": 7593 }, { "epoch": 1.9714434060228454, "grad_norm": 1.6995102167129517, "learning_rate": 5.024498848414372e-07, "loss": 0.03968588262796402, "step": 7594 }, { "epoch": 1.9717030114226377, "grad_norm": 1.372101902961731, "learning_rate": 5.024059494435582e-07, "loss": 0.09318259358406067, "step": 7595 }, { "epoch": 1.97196261682243, "grad_norm": 1.4065935611724854, "learning_rate": 5.023624113688442e-07, "loss": 0.0939580649137497, "step": 7596 }, { "epoch": 1.9722222222222223, "grad_norm": 1.6873221397399902, "learning_rate": 5.023192706249917e-07, "loss": 0.09438347816467285, "step": 7597 }, { "epoch": 1.9724818276220146, "grad_norm": 2.013601303100586, "learning_rate": 5.022765272196273e-07, "loss": 0.042233120650053024, "step": 7598 }, { "epoch": 1.972741433021807, "grad_norm": 1.5470956563949585, "learning_rate": 5.022341811603066e-07, "loss": 0.07529165595769882, "step": 7599 }, { "epoch": 1.9730010384215992, "grad_norm": 3.4799349308013916, "learning_rate": 5.021922324545157e-07, "loss": 0.04035193473100662, "step": 7600 }, { "epoch": 1.9732606438213915, "grad_norm": 2.0042989253997803, "learning_rate": 5.021506811096703e-07, "loss": 0.07059355825185776, "step": 7601 }, { "epoch": 1.9735202492211839, "grad_norm": 1.9689987897872925, "learning_rate": 5.021095271331153e-07, "loss": 0.07929807156324387, "step": 7602 }, { "epoch": 1.9737798546209762, "grad_norm": 2.0664989948272705, "learning_rate": 5.020687705321261e-07, "loss": 0.04366575926542282, "step": 7603 }, { "epoch": 1.9740394600207685, "grad_norm": 1.5220314264297485, "learning_rate": 5.020284113139077e-07, "loss": 0.085391104221344, "step": 7604 }, { "epoch": 1.9742990654205608, "grad_norm": 1.6859320402145386, "learning_rate": 5.019884494855942e-07, "loss": 0.08042481541633606, "step": 7605 }, { "epoch": 1.974558670820353, "grad_norm": 2.8777220249176025, "learning_rate": 5.019488850542504e-07, "loss": 0.06564566493034363, "step": 7606 }, { "epoch": 1.9748182762201454, "grad_norm": 1.9051649570465088, "learning_rate": 5.0190971802687e-07, "loss": 0.07754673808813095, "step": 7607 }, { "epoch": 1.9750778816199377, "grad_norm": 2.198599100112915, "learning_rate": 5.018709484103771e-07, "loss": 0.08840900659561157, "step": 7608 }, { "epoch": 1.97533748701973, "grad_norm": 1.7090321779251099, "learning_rate": 5.018325762116251e-07, "loss": 0.0756164938211441, "step": 7609 }, { "epoch": 1.9755970924195223, "grad_norm": 3.2151317596435547, "learning_rate": 5.017946014373977e-07, "loss": 0.015406077727675438, "step": 7610 }, { "epoch": 1.9758566978193146, "grad_norm": 2.3218913078308105, "learning_rate": 5.017570240944074e-07, "loss": 0.028969723731279373, "step": 7611 }, { "epoch": 1.976116303219107, "grad_norm": 2.643324851989746, "learning_rate": 5.017198441892972e-07, "loss": 0.07934356480836868, "step": 7612 }, { "epoch": 1.9763759086188992, "grad_norm": 2.6325345039367676, "learning_rate": 5.016830617286399e-07, "loss": 0.07263703644275665, "step": 7613 }, { "epoch": 1.9766355140186915, "grad_norm": 4.927050590515137, "learning_rate": 5.016466767189378e-07, "loss": 0.06042082980275154, "step": 7614 }, { "epoch": 1.9768951194184838, "grad_norm": 2.416717529296875, "learning_rate": 5.016106891666227e-07, "loss": 0.04615069180727005, "step": 7615 }, { "epoch": 1.9771547248182761, "grad_norm": 1.2711825370788574, "learning_rate": 5.015750990780564e-07, "loss": 0.01182909682393074, "step": 7616 }, { "epoch": 1.9774143302180685, "grad_norm": 2.162055730819702, "learning_rate": 5.015399064595305e-07, "loss": 0.0622231662273407, "step": 7617 }, { "epoch": 1.9776739356178608, "grad_norm": 1.7689777612686157, "learning_rate": 5.01505111317266e-07, "loss": 0.10324521362781525, "step": 7618 }, { "epoch": 1.977933541017653, "grad_norm": 2.5320465564727783, "learning_rate": 5.014707136574146e-07, "loss": 0.06623880565166473, "step": 7619 }, { "epoch": 1.9781931464174454, "grad_norm": 1.72808837890625, "learning_rate": 5.014367134860563e-07, "loss": 0.08451399207115173, "step": 7620 }, { "epoch": 1.9784527518172377, "grad_norm": 2.028581142425537, "learning_rate": 5.014031108092017e-07, "loss": 0.05404966324567795, "step": 7621 }, { "epoch": 1.97871235721703, "grad_norm": 2.832362174987793, "learning_rate": 5.01369905632791e-07, "loss": 0.05658766254782677, "step": 7622 }, { "epoch": 1.9789719626168223, "grad_norm": 1.8250025510787964, "learning_rate": 5.013370979626944e-07, "loss": 0.09817276895046234, "step": 7623 }, { "epoch": 1.9792315680166146, "grad_norm": 0.8822161555290222, "learning_rate": 5.013046878047113e-07, "loss": 0.013166259042918682, "step": 7624 }, { "epoch": 1.979491173416407, "grad_norm": 2.93461537361145, "learning_rate": 5.01272675164571e-07, "loss": 0.0529542900621891, "step": 7625 }, { "epoch": 1.9797507788161994, "grad_norm": 1.75931715965271, "learning_rate": 5.012410600479326e-07, "loss": 0.07238578796386719, "step": 7626 }, { "epoch": 1.9800103842159917, "grad_norm": 1.4441697597503662, "learning_rate": 5.012098424603852e-07, "loss": 0.09748569130897522, "step": 7627 }, { "epoch": 1.980269989615784, "grad_norm": 2.3690285682678223, "learning_rate": 5.011790224074471e-07, "loss": 0.04768328368663788, "step": 7628 }, { "epoch": 1.9805295950155763, "grad_norm": 3.326603651046753, "learning_rate": 5.011485998945668e-07, "loss": 0.09210164844989777, "step": 7629 }, { "epoch": 1.9807892004153687, "grad_norm": 3.710452079772949, "learning_rate": 5.01118574927122e-07, "loss": 0.04181351512670517, "step": 7630 }, { "epoch": 1.981048805815161, "grad_norm": 1.8321658372879028, "learning_rate": 5.010889475104205e-07, "loss": 0.08593320101499557, "step": 7631 }, { "epoch": 1.9813084112149533, "grad_norm": 2.0335750579833984, "learning_rate": 5.010597176497002e-07, "loss": 0.07735608518123627, "step": 7632 }, { "epoch": 1.9815680166147456, "grad_norm": 2.700472116470337, "learning_rate": 5.010308853501275e-07, "loss": 0.06621278077363968, "step": 7633 }, { "epoch": 1.9818276220145379, "grad_norm": 1.2705007791519165, "learning_rate": 5.010024506168001e-07, "loss": 0.09884325414896011, "step": 7634 }, { "epoch": 1.9820872274143302, "grad_norm": 3.1115574836730957, "learning_rate": 5.009744134547438e-07, "loss": 0.08807194977998734, "step": 7635 }, { "epoch": 1.9823468328141225, "grad_norm": 1.3965504169464111, "learning_rate": 5.009467738689157e-07, "loss": 0.09492120146751404, "step": 7636 }, { "epoch": 1.9826064382139148, "grad_norm": 1.1677532196044922, "learning_rate": 5.009195318642012e-07, "loss": 0.10174675285816193, "step": 7637 }, { "epoch": 1.9828660436137073, "grad_norm": 1.3990025520324707, "learning_rate": 5.008926874454166e-07, "loss": 0.09306704998016357, "step": 7638 }, { "epoch": 1.9831256490134996, "grad_norm": 1.6945353746414185, "learning_rate": 5.008662406173071e-07, "loss": 0.08657180517911911, "step": 7639 }, { "epoch": 1.983385254413292, "grad_norm": 1.6021965742111206, "learning_rate": 5.008401913845478e-07, "loss": 0.04959368333220482, "step": 7640 }, { "epoch": 1.9836448598130842, "grad_norm": 2.574524402618408, "learning_rate": 5.008145397517439e-07, "loss": 0.0431305468082428, "step": 7641 }, { "epoch": 1.9839044652128766, "grad_norm": 2.605161190032959, "learning_rate": 5.007892857234296e-07, "loss": 0.05555075407028198, "step": 7642 }, { "epoch": 1.9841640706126689, "grad_norm": 1.6888700723648071, "learning_rate": 5.007644293040695e-07, "loss": 0.08208144456148148, "step": 7643 }, { "epoch": 1.9844236760124612, "grad_norm": 3.789928913116455, "learning_rate": 5.007399704980578e-07, "loss": 0.04744364321231842, "step": 7644 }, { "epoch": 1.9846832814122535, "grad_norm": 2.4384467601776123, "learning_rate": 5.007159093097181e-07, "loss": 0.0514814518392086, "step": 7645 }, { "epoch": 1.9849428868120458, "grad_norm": 1.4345169067382812, "learning_rate": 5.006922457433039e-07, "loss": 0.051253996789455414, "step": 7646 }, { "epoch": 1.985202492211838, "grad_norm": 4.146392822265625, "learning_rate": 5.006689798029981e-07, "loss": 0.05522489547729492, "step": 7647 }, { "epoch": 1.9854620976116304, "grad_norm": 2.1691064834594727, "learning_rate": 5.00646111492914e-07, "loss": 0.0339425690472126, "step": 7648 }, { "epoch": 1.9857217030114227, "grad_norm": 1.910693645477295, "learning_rate": 5.006236408170939e-07, "loss": 0.06377508491277695, "step": 7649 }, { "epoch": 1.985981308411215, "grad_norm": 2.5149669647216797, "learning_rate": 5.006015677795102e-07, "loss": 0.03739326074719429, "step": 7650 }, { "epoch": 1.9862409138110073, "grad_norm": 3.145343780517578, "learning_rate": 5.005798923840652e-07, "loss": 0.04374285042285919, "step": 7651 }, { "epoch": 1.9865005192107996, "grad_norm": 1.4832382202148438, "learning_rate": 5.005586146345901e-07, "loss": 0.07887154817581177, "step": 7652 }, { "epoch": 1.986760124610592, "grad_norm": 2.480626344680786, "learning_rate": 5.005377345348463e-07, "loss": 0.02602575160562992, "step": 7653 }, { "epoch": 1.9870197300103842, "grad_norm": 1.3214482069015503, "learning_rate": 5.005172520885257e-07, "loss": 0.10800068080425262, "step": 7654 }, { "epoch": 1.9872793354101765, "grad_norm": 1.3355056047439575, "learning_rate": 5.004971672992484e-07, "loss": 0.07155978679656982, "step": 7655 }, { "epoch": 1.9875389408099688, "grad_norm": 2.175248861312866, "learning_rate": 5.004774801705649e-07, "loss": 0.07562453299760818, "step": 7656 }, { "epoch": 1.9877985462097612, "grad_norm": 4.16956090927124, "learning_rate": 5.00458190705956e-07, "loss": 0.06684092432260513, "step": 7657 }, { "epoch": 1.9880581516095535, "grad_norm": 2.9625625610351562, "learning_rate": 5.00439298908831e-07, "loss": 0.02597794309258461, "step": 7658 }, { "epoch": 1.9883177570093458, "grad_norm": 2.649956464767456, "learning_rate": 5.004208047825299e-07, "loss": 0.05677703395485878, "step": 7659 }, { "epoch": 1.988577362409138, "grad_norm": 1.8722232580184937, "learning_rate": 5.004027083303222e-07, "loss": 0.05201847851276398, "step": 7660 }, { "epoch": 1.9888369678089304, "grad_norm": 2.39107608795166, "learning_rate": 5.003850095554065e-07, "loss": 0.059901148080825806, "step": 7661 }, { "epoch": 1.9890965732087227, "grad_norm": 3.977266788482666, "learning_rate": 5.00367708460912e-07, "loss": 0.05649450421333313, "step": 7662 }, { "epoch": 1.989356178608515, "grad_norm": 1.4340389966964722, "learning_rate": 5.003508050498967e-07, "loss": 0.09782363474369049, "step": 7663 }, { "epoch": 1.9896157840083073, "grad_norm": 1.6004629135131836, "learning_rate": 5.00334299325349e-07, "loss": 0.04189876466989517, "step": 7664 }, { "epoch": 1.9898753894080996, "grad_norm": 2.3338260650634766, "learning_rate": 5.003181912901865e-07, "loss": 0.06678564846515656, "step": 7665 }, { "epoch": 1.990134994807892, "grad_norm": 2.449531316757202, "learning_rate": 5.003024809472572e-07, "loss": 0.06830041110515594, "step": 7666 }, { "epoch": 1.9903946002076842, "grad_norm": 2.1409101486206055, "learning_rate": 5.002871682993378e-07, "loss": 0.05936598777770996, "step": 7667 }, { "epoch": 1.9906542056074765, "grad_norm": 2.92907977104187, "learning_rate": 5.002722533491354e-07, "loss": 0.06522024422883987, "step": 7668 }, { "epoch": 1.9909138110072688, "grad_norm": 2.219045639038086, "learning_rate": 5.002577360992869e-07, "loss": 0.05373181030154228, "step": 7669 }, { "epoch": 1.9911734164070611, "grad_norm": 1.7775267362594604, "learning_rate": 5.002436165523583e-07, "loss": 0.04156951978802681, "step": 7670 }, { "epoch": 1.9914330218068534, "grad_norm": 2.4047963619232178, "learning_rate": 5.002298947108454e-07, "loss": 0.038577742874622345, "step": 7671 }, { "epoch": 1.9916926272066457, "grad_norm": 1.56131112575531, "learning_rate": 5.002165705771746e-07, "loss": 0.085055872797966, "step": 7672 }, { "epoch": 1.991952232606438, "grad_norm": 1.3594938516616821, "learning_rate": 5.002036441537007e-07, "loss": 0.09053932130336761, "step": 7673 }, { "epoch": 1.9922118380062306, "grad_norm": 2.414135694503784, "learning_rate": 5.00191115442709e-07, "loss": 0.06986020505428314, "step": 7674 }, { "epoch": 1.992471443406023, "grad_norm": 1.9892802238464355, "learning_rate": 5.001789844464145e-07, "loss": 0.07014727592468262, "step": 7675 }, { "epoch": 1.9927310488058152, "grad_norm": 1.9072118997573853, "learning_rate": 5.001672511669613e-07, "loss": 0.07203157246112823, "step": 7676 }, { "epoch": 1.9929906542056075, "grad_norm": 2.0830376148223877, "learning_rate": 5.001559156064238e-07, "loss": 0.07158809900283813, "step": 7677 }, { "epoch": 1.9932502596053998, "grad_norm": 1.7707465887069702, "learning_rate": 5.00144977766806e-07, "loss": 0.042444802820682526, "step": 7678 }, { "epoch": 1.9935098650051921, "grad_norm": 2.8923542499542236, "learning_rate": 5.001344376500408e-07, "loss": 0.03503238782286644, "step": 7679 }, { "epoch": 1.9937694704049844, "grad_norm": 2.3337881565093994, "learning_rate": 5.001242952579924e-07, "loss": 0.05890018492937088, "step": 7680 }, { "epoch": 1.9940290758047767, "grad_norm": 2.0874979496002197, "learning_rate": 5.001145505924532e-07, "loss": 0.07469760626554489, "step": 7681 }, { "epoch": 1.994288681204569, "grad_norm": 1.693061113357544, "learning_rate": 5.001052036551458e-07, "loss": 0.09677666425704956, "step": 7682 }, { "epoch": 1.9945482866043613, "grad_norm": 2.354555130004883, "learning_rate": 5.000962544477227e-07, "loss": 0.03406219184398651, "step": 7683 }, { "epoch": 1.9948078920041536, "grad_norm": 2.2990882396698, "learning_rate": 5.000877029717655e-07, "loss": 0.07407793402671814, "step": 7684 }, { "epoch": 1.995067497403946, "grad_norm": 4.055701732635498, "learning_rate": 5.000795492287867e-07, "loss": 0.055747564882040024, "step": 7685 }, { "epoch": 1.9953271028037385, "grad_norm": 2.3519155979156494, "learning_rate": 5.000717932202271e-07, "loss": 0.04997289180755615, "step": 7686 }, { "epoch": 1.9955867082035308, "grad_norm": 2.0138792991638184, "learning_rate": 5.000644349474577e-07, "loss": 0.08424818515777588, "step": 7687 }, { "epoch": 1.995846313603323, "grad_norm": 2.241953134536743, "learning_rate": 5.000574744117799e-07, "loss": 0.06445389240980148, "step": 7688 }, { "epoch": 1.9961059190031154, "grad_norm": 2.929614305496216, "learning_rate": 5.000509116144236e-07, "loss": 0.06162720173597336, "step": 7689 }, { "epoch": 1.9963655244029077, "grad_norm": 1.9447435140609741, "learning_rate": 5.000447465565491e-07, "loss": 0.08256976306438446, "step": 7690 }, { "epoch": 1.9966251298027, "grad_norm": 1.7833008766174316, "learning_rate": 5.000389792392462e-07, "loss": 0.0900188535451889, "step": 7691 }, { "epoch": 1.9968847352024923, "grad_norm": 1.8954367637634277, "learning_rate": 5.000336096635346e-07, "loss": 0.06498772650957108, "step": 7692 }, { "epoch": 1.9971443406022846, "grad_norm": 2.800274133682251, "learning_rate": 5.000286378303634e-07, "loss": 0.06215269863605499, "step": 7693 }, { "epoch": 1.997403946002077, "grad_norm": 1.713019847869873, "learning_rate": 5.000240637406114e-07, "loss": 0.10645126551389694, "step": 7694 }, { "epoch": 1.9976635514018692, "grad_norm": 1.6680662631988525, "learning_rate": 5.000198873950875e-07, "loss": 0.05673588067293167, "step": 7695 }, { "epoch": 1.9979231568016615, "grad_norm": 2.2716898918151855, "learning_rate": 5.000161087945296e-07, "loss": 0.046487316489219666, "step": 7696 }, { "epoch": 1.9981827622014539, "grad_norm": 1.7474796772003174, "learning_rate": 5.000127279396061e-07, "loss": 0.10008780658245087, "step": 7697 }, { "epoch": 1.9984423676012462, "grad_norm": 3.551811695098877, "learning_rate": 5.000097448309143e-07, "loss": 0.03176377713680267, "step": 7698 }, { "epoch": 1.9987019730010385, "grad_norm": 2.150904417037964, "learning_rate": 5.000071594689815e-07, "loss": 0.025767400860786438, "step": 7699 }, { "epoch": 1.9989615784008308, "grad_norm": 1.6857056617736816, "learning_rate": 5.000049718542651e-07, "loss": 0.09427676349878311, "step": 7700 }, { "epoch": 1.999221183800623, "grad_norm": 1.8868073225021362, "learning_rate": 5.000031819871517e-07, "loss": 0.07848390191793442, "step": 7701 }, { "epoch": 1.9994807892004154, "grad_norm": 2.3158187866210938, "learning_rate": 5.000017898679574e-07, "loss": 0.07643073797225952, "step": 7702 }, { "epoch": 1.9997403946002077, "grad_norm": 1.9772340059280396, "learning_rate": 5.000007954969287e-07, "loss": 0.040937237441539764, "step": 7703 }, { "epoch": 2.0, "grad_norm": 3.555635690689087, "learning_rate": 5.000001988742409e-07, "loss": 0.04859667271375656, "step": 7704 } ], "logging_steps": 1, "max_steps": 7704, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 1, "trial_name": null, "trial_params": null }