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