{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 2.0, "eval_steps": 500, "global_step": 28930, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 6.91323885240235e-05, "grad_norm": 34.9440798526412, "learning_rate": 0.0, "loss": 0.9645, "step": 1 }, { "epoch": 0.000138264777048047, "grad_norm": 33.673685604530064, "learning_rate": 1.152073732718894e-08, "loss": 0.7393, "step": 2 }, { "epoch": 0.00020739716557207052, "grad_norm": 22.43385099453277, "learning_rate": 2.304147465437788e-08, "loss": 0.7936, "step": 3 }, { "epoch": 0.000276529554096094, "grad_norm": 28.42500854564411, "learning_rate": 3.4562211981566826e-08, "loss": 0.8649, "step": 4 }, { "epoch": 0.0003456619426201175, "grad_norm": 32.35221617602802, "learning_rate": 4.608294930875576e-08, "loss": 0.7936, "step": 5 }, { "epoch": 0.00041479433114414104, "grad_norm": 30.081850404027048, "learning_rate": 5.760368663594471e-08, "loss": 0.7451, "step": 6 }, { "epoch": 0.00048392671966816455, "grad_norm": 28.38948058174201, "learning_rate": 6.912442396313365e-08, "loss": 0.6852, "step": 7 }, { "epoch": 0.000553059108192188, "grad_norm": 31.206136075868837, "learning_rate": 8.064516129032259e-08, "loss": 0.9, "step": 8 }, { "epoch": 0.0006221914967162116, "grad_norm": 37.71844771383183, "learning_rate": 9.216589861751152e-08, "loss": 0.772, "step": 9 }, { "epoch": 0.000691323885240235, "grad_norm": 29.983780400427676, "learning_rate": 1.0368663594470047e-07, "loss": 0.7319, "step": 10 }, { "epoch": 0.0007604562737642585, "grad_norm": 36.19987067394532, "learning_rate": 1.1520737327188942e-07, "loss": 0.8056, "step": 11 }, { "epoch": 0.0008295886622882821, "grad_norm": 42.93606825513471, "learning_rate": 1.2672811059907837e-07, "loss": 0.859, "step": 12 }, { "epoch": 0.0008987210508123055, "grad_norm": 37.99604504993953, "learning_rate": 1.382488479262673e-07, "loss": 0.7705, "step": 13 }, { "epoch": 0.0009678534393363291, "grad_norm": 31.622258018943157, "learning_rate": 1.4976958525345624e-07, "loss": 0.6334, "step": 14 }, { "epoch": 0.0010369858278603526, "grad_norm": 34.33959870752537, "learning_rate": 1.6129032258064518e-07, "loss": 0.8504, "step": 15 }, { "epoch": 0.001106118216384376, "grad_norm": 37.13571973072957, "learning_rate": 1.7281105990783414e-07, "loss": 0.6594, "step": 16 }, { "epoch": 0.0011752506049083995, "grad_norm": 31.78585583926888, "learning_rate": 1.8433179723502305e-07, "loss": 0.6888, "step": 17 }, { "epoch": 0.0012443829934324232, "grad_norm": 25.094043754418248, "learning_rate": 1.9585253456221198e-07, "loss": 0.7976, "step": 18 }, { "epoch": 0.0013135153819564466, "grad_norm": 54.5488620461889, "learning_rate": 2.0737327188940094e-07, "loss": 0.9043, "step": 19 }, { "epoch": 0.00138264777048047, "grad_norm": 26.238293828837044, "learning_rate": 2.1889400921658988e-07, "loss": 0.6888, "step": 20 }, { "epoch": 0.0014517801590044936, "grad_norm": 17.066471390284395, "learning_rate": 2.3041474654377884e-07, "loss": 0.6545, "step": 21 }, { "epoch": 0.001520912547528517, "grad_norm": 21.117789247095214, "learning_rate": 2.4193548387096775e-07, "loss": 0.8644, "step": 22 }, { "epoch": 0.0015900449360525407, "grad_norm": 20.80264919839146, "learning_rate": 2.5345622119815674e-07, "loss": 0.6968, "step": 23 }, { "epoch": 0.0016591773245765642, "grad_norm": 14.811114509938184, "learning_rate": 2.649769585253456e-07, "loss": 0.6346, "step": 24 }, { "epoch": 0.0017283097131005876, "grad_norm": 13.111228442126537, "learning_rate": 2.764976958525346e-07, "loss": 0.7761, "step": 25 }, { "epoch": 0.001797442101624611, "grad_norm": 25.865684726112, "learning_rate": 2.8801843317972354e-07, "loss": 0.7814, "step": 26 }, { "epoch": 0.0018665744901486345, "grad_norm": 17.418568546561872, "learning_rate": 2.995391705069125e-07, "loss": 0.7032, "step": 27 }, { "epoch": 0.0019357068786726582, "grad_norm": 8.93723490102437, "learning_rate": 3.110599078341014e-07, "loss": 0.6165, "step": 28 }, { "epoch": 0.0020048392671966815, "grad_norm": 5.516023282237777, "learning_rate": 3.2258064516129035e-07, "loss": 0.8188, "step": 29 }, { "epoch": 0.002073971655720705, "grad_norm": 4.482361169685703, "learning_rate": 3.341013824884793e-07, "loss": 0.7102, "step": 30 }, { "epoch": 0.002143104044244729, "grad_norm": 4.676616632052571, "learning_rate": 3.456221198156683e-07, "loss": 0.6099, "step": 31 }, { "epoch": 0.002212236432768752, "grad_norm": 4.330483833858327, "learning_rate": 3.5714285714285716e-07, "loss": 0.6192, "step": 32 }, { "epoch": 0.0022813688212927757, "grad_norm": 4.1836388962362845, "learning_rate": 3.686635944700461e-07, "loss": 0.8131, "step": 33 }, { "epoch": 0.002350501209816799, "grad_norm": 3.494454663315565, "learning_rate": 3.801843317972351e-07, "loss": 0.5718, "step": 34 }, { "epoch": 0.0024196335983408227, "grad_norm": 3.579060751014094, "learning_rate": 3.9170506912442396e-07, "loss": 0.614, "step": 35 }, { "epoch": 0.0024887659868648463, "grad_norm": 4.156019643693866, "learning_rate": 4.032258064516129e-07, "loss": 0.7436, "step": 36 }, { "epoch": 0.0025578983753888696, "grad_norm": 4.328476048434388, "learning_rate": 4.147465437788019e-07, "loss": 0.8531, "step": 37 }, { "epoch": 0.0026270307639128933, "grad_norm": 3.892136166667997, "learning_rate": 4.262672811059908e-07, "loss": 0.6332, "step": 38 }, { "epoch": 0.0026961631524369165, "grad_norm": 3.650047004875848, "learning_rate": 4.3778801843317976e-07, "loss": 0.6688, "step": 39 }, { "epoch": 0.00276529554096094, "grad_norm": 3.903370727782822, "learning_rate": 4.493087557603687e-07, "loss": 0.8031, "step": 40 }, { "epoch": 0.002834427929484964, "grad_norm": 3.029749267331785, "learning_rate": 4.608294930875577e-07, "loss": 0.562, "step": 41 }, { "epoch": 0.002903560318008987, "grad_norm": 3.3724515708359744, "learning_rate": 4.7235023041474656e-07, "loss": 0.6285, "step": 42 }, { "epoch": 0.002972692706533011, "grad_norm": 3.120853255975001, "learning_rate": 4.838709677419355e-07, "loss": 0.6024, "step": 43 }, { "epoch": 0.003041825095057034, "grad_norm": 3.5007110596851354, "learning_rate": 4.953917050691245e-07, "loss": 0.6953, "step": 44 }, { "epoch": 0.0031109574835810577, "grad_norm": 2.903993254146862, "learning_rate": 5.069124423963135e-07, "loss": 0.6447, "step": 45 }, { "epoch": 0.0031800898721050814, "grad_norm": 2.9111249755713633, "learning_rate": 5.184331797235024e-07, "loss": 0.6124, "step": 46 }, { "epoch": 0.0032492222606291046, "grad_norm": 3.4763697212639713, "learning_rate": 5.299539170506912e-07, "loss": 0.77, "step": 47 }, { "epoch": 0.0033183546491531283, "grad_norm": 2.386832958474211, "learning_rate": 5.414746543778802e-07, "loss": 0.571, "step": 48 }, { "epoch": 0.0033874870376771516, "grad_norm": 3.030536050686801, "learning_rate": 5.529953917050692e-07, "loss": 0.5204, "step": 49 }, { "epoch": 0.0034566194262011752, "grad_norm": 3.1693393199753888, "learning_rate": 5.645161290322581e-07, "loss": 0.6501, "step": 50 }, { "epoch": 0.003525751814725199, "grad_norm": 3.9243951665329284, "learning_rate": 5.760368663594471e-07, "loss": 0.6689, "step": 51 }, { "epoch": 0.003594884203249222, "grad_norm": 3.077560835652264, "learning_rate": 5.87557603686636e-07, "loss": 0.5031, "step": 52 }, { "epoch": 0.003664016591773246, "grad_norm": 2.8163726562403153, "learning_rate": 5.99078341013825e-07, "loss": 0.5093, "step": 53 }, { "epoch": 0.003733148980297269, "grad_norm": 3.511571491792349, "learning_rate": 6.105990783410138e-07, "loss": 0.8205, "step": 54 }, { "epoch": 0.0038022813688212928, "grad_norm": 3.160864117354009, "learning_rate": 6.221198156682028e-07, "loss": 0.6377, "step": 55 }, { "epoch": 0.0038714137573453164, "grad_norm": 11.903969280756598, "learning_rate": 6.336405529953918e-07, "loss": 0.573, "step": 56 }, { "epoch": 0.00394054614586934, "grad_norm": 2.5897022628011963, "learning_rate": 6.451612903225807e-07, "loss": 0.5828, "step": 57 }, { "epoch": 0.004009678534393363, "grad_norm": 2.77798931956415, "learning_rate": 6.566820276497696e-07, "loss": 0.7159, "step": 58 }, { "epoch": 0.004078810922917387, "grad_norm": 2.1423009544505174, "learning_rate": 6.682027649769586e-07, "loss": 0.531, "step": 59 }, { "epoch": 0.00414794331144141, "grad_norm": 2.200586400758685, "learning_rate": 6.797235023041475e-07, "loss": 0.5159, "step": 60 }, { "epoch": 0.0042170756999654335, "grad_norm": 2.455628876534046, "learning_rate": 6.912442396313365e-07, "loss": 0.571, "step": 61 }, { "epoch": 0.004286208088489458, "grad_norm": 2.32773539468399, "learning_rate": 7.027649769585254e-07, "loss": 0.5781, "step": 62 }, { "epoch": 0.004355340477013481, "grad_norm": 1.972159443017681, "learning_rate": 7.142857142857143e-07, "loss": 0.5417, "step": 63 }, { "epoch": 0.004424472865537504, "grad_norm": 1.7613433264075753, "learning_rate": 7.258064516129033e-07, "loss": 0.6313, "step": 64 }, { "epoch": 0.004493605254061528, "grad_norm": 2.355455051144276, "learning_rate": 7.373271889400922e-07, "loss": 0.7466, "step": 65 }, { "epoch": 0.0045627376425855515, "grad_norm": 1.8155568424308584, "learning_rate": 7.488479262672811e-07, "loss": 0.5204, "step": 66 }, { "epoch": 0.004631870031109575, "grad_norm": 1.760410455690479, "learning_rate": 7.603686635944702e-07, "loss": 0.5106, "step": 67 }, { "epoch": 0.004701002419633598, "grad_norm": 2.294930339421628, "learning_rate": 7.71889400921659e-07, "loss": 0.6311, "step": 68 }, { "epoch": 0.004770134808157622, "grad_norm": 2.04434148978563, "learning_rate": 7.834101382488479e-07, "loss": 0.6379, "step": 69 }, { "epoch": 0.004839267196681645, "grad_norm": 1.8353461362292751, "learning_rate": 7.949308755760369e-07, "loss": 0.5099, "step": 70 }, { "epoch": 0.004908399585205669, "grad_norm": 1.4914258983127615, "learning_rate": 8.064516129032258e-07, "loss": 0.4675, "step": 71 }, { "epoch": 0.004977531973729693, "grad_norm": 2.095746279977775, "learning_rate": 8.179723502304149e-07, "loss": 0.662, "step": 72 }, { "epoch": 0.005046664362253716, "grad_norm": 1.5883903753892001, "learning_rate": 8.294930875576038e-07, "loss": 0.5489, "step": 73 }, { "epoch": 0.005115796750777739, "grad_norm": 1.623981030513798, "learning_rate": 8.410138248847927e-07, "loss": 0.5415, "step": 74 }, { "epoch": 0.005184929139301763, "grad_norm": 1.9945653813608213, "learning_rate": 8.525345622119816e-07, "loss": 0.614, "step": 75 }, { "epoch": 0.0052540615278257865, "grad_norm": 1.6244603228338579, "learning_rate": 8.640552995391706e-07, "loss": 0.6063, "step": 76 }, { "epoch": 0.00532319391634981, "grad_norm": 1.5504256675866608, "learning_rate": 8.755760368663595e-07, "loss": 0.4822, "step": 77 }, { "epoch": 0.005392326304873833, "grad_norm": 1.6230935181446708, "learning_rate": 8.870967741935485e-07, "loss": 0.498, "step": 78 }, { "epoch": 0.005461458693397857, "grad_norm": 1.5878401175570196, "learning_rate": 8.986175115207374e-07, "loss": 0.5941, "step": 79 }, { "epoch": 0.00553059108192188, "grad_norm": 1.5282454826295082, "learning_rate": 9.101382488479263e-07, "loss": 0.6069, "step": 80 }, { "epoch": 0.005599723470445904, "grad_norm": 1.3342358503249527, "learning_rate": 9.216589861751154e-07, "loss": 0.4787, "step": 81 }, { "epoch": 0.005668855858969928, "grad_norm": 1.3519828544565198, "learning_rate": 9.331797235023042e-07, "loss": 0.5673, "step": 82 }, { "epoch": 0.005737988247493951, "grad_norm": 1.720173124274535, "learning_rate": 9.447004608294931e-07, "loss": 0.6532, "step": 83 }, { "epoch": 0.005807120636017974, "grad_norm": 1.4414404784620647, "learning_rate": 9.562211981566821e-07, "loss": 0.5463, "step": 84 }, { "epoch": 0.005876253024541998, "grad_norm": 1.3643538726191509, "learning_rate": 9.67741935483871e-07, "loss": 0.4178, "step": 85 }, { "epoch": 0.005945385413066022, "grad_norm": 2.257671934691231, "learning_rate": 9.7926267281106e-07, "loss": 0.5823, "step": 86 }, { "epoch": 0.006014517801590045, "grad_norm": 1.406624707555253, "learning_rate": 9.90783410138249e-07, "loss": 0.5507, "step": 87 }, { "epoch": 0.006083650190114068, "grad_norm": 1.492378979779291, "learning_rate": 1.0023041474654379e-06, "loss": 0.4768, "step": 88 }, { "epoch": 0.006152782578638092, "grad_norm": 1.3467179045676967, "learning_rate": 1.013824884792627e-06, "loss": 0.4563, "step": 89 }, { "epoch": 0.006221914967162115, "grad_norm": 1.463379966365056, "learning_rate": 1.0253456221198158e-06, "loss": 0.6569, "step": 90 }, { "epoch": 0.006291047355686139, "grad_norm": 1.3950042407141423, "learning_rate": 1.0368663594470047e-06, "loss": 0.4737, "step": 91 }, { "epoch": 0.006360179744210163, "grad_norm": 1.3628895079844756, "learning_rate": 1.0483870967741936e-06, "loss": 0.4894, "step": 92 }, { "epoch": 0.006429312132734186, "grad_norm": 2.513169965313829, "learning_rate": 1.0599078341013825e-06, "loss": 0.5406, "step": 93 }, { "epoch": 0.006498444521258209, "grad_norm": 1.2958164342851117, "learning_rate": 1.0714285714285714e-06, "loss": 0.6469, "step": 94 }, { "epoch": 0.006567576909782233, "grad_norm": 1.3514822355543998, "learning_rate": 1.0829493087557605e-06, "loss": 0.4691, "step": 95 }, { "epoch": 0.006636709298306257, "grad_norm": 1.2428207711646868, "learning_rate": 1.0944700460829493e-06, "loss": 0.4258, "step": 96 }, { "epoch": 0.00670584168683028, "grad_norm": 1.6951026437585202, "learning_rate": 1.1059907834101384e-06, "loss": 0.6153, "step": 97 }, { "epoch": 0.006774974075354303, "grad_norm": 1.2833770038526282, "learning_rate": 1.1175115207373273e-06, "loss": 0.5317, "step": 98 }, { "epoch": 0.006844106463878327, "grad_norm": 1.2179951376761788, "learning_rate": 1.1290322580645162e-06, "loss": 0.4052, "step": 99 }, { "epoch": 0.0069132388524023505, "grad_norm": 1.4028401719314814, "learning_rate": 1.1405529953917053e-06, "loss": 0.575, "step": 100 }, { "epoch": 0.006982371240926374, "grad_norm": 1.3500463589209022, "learning_rate": 1.1520737327188942e-06, "loss": 0.6326, "step": 101 }, { "epoch": 0.007051503629450398, "grad_norm": 1.2876210554183294, "learning_rate": 1.163594470046083e-06, "loss": 0.459, "step": 102 }, { "epoch": 0.007120636017974421, "grad_norm": 1.363895757402435, "learning_rate": 1.175115207373272e-06, "loss": 0.4788, "step": 103 }, { "epoch": 0.007189768406498444, "grad_norm": 1.45250295133778, "learning_rate": 1.1866359447004608e-06, "loss": 0.6651, "step": 104 }, { "epoch": 0.007258900795022468, "grad_norm": 1.4609093039386833, "learning_rate": 1.19815668202765e-06, "loss": 0.5358, "step": 105 }, { "epoch": 0.007328033183546492, "grad_norm": 1.2835484908672223, "learning_rate": 1.2096774193548388e-06, "loss": 0.4181, "step": 106 }, { "epoch": 0.007397165572070515, "grad_norm": 1.308011473903604, "learning_rate": 1.2211981566820277e-06, "loss": 0.4805, "step": 107 }, { "epoch": 0.007466297960594538, "grad_norm": 1.281782083998857, "learning_rate": 1.2327188940092166e-06, "loss": 0.6322, "step": 108 }, { "epoch": 0.007535430349118562, "grad_norm": 1.2610955703994753, "learning_rate": 1.2442396313364057e-06, "loss": 0.4923, "step": 109 }, { "epoch": 0.0076045627376425855, "grad_norm": 1.3249951641703495, "learning_rate": 1.2557603686635948e-06, "loss": 0.4221, "step": 110 }, { "epoch": 0.007673695126166609, "grad_norm": 1.1962700653721752, "learning_rate": 1.2672811059907836e-06, "loss": 0.5588, "step": 111 }, { "epoch": 0.007742827514690633, "grad_norm": 1.3775285139824998, "learning_rate": 1.2788018433179725e-06, "loss": 0.5623, "step": 112 }, { "epoch": 0.007811959903214656, "grad_norm": 1.337987055738114, "learning_rate": 1.2903225806451614e-06, "loss": 0.4688, "step": 113 }, { "epoch": 0.00788109229173868, "grad_norm": 1.3532327559108164, "learning_rate": 1.3018433179723503e-06, "loss": 0.4213, "step": 114 }, { "epoch": 0.007950224680262703, "grad_norm": 1.3122018246380716, "learning_rate": 1.3133640552995392e-06, "loss": 0.5818, "step": 115 }, { "epoch": 0.008019357068786726, "grad_norm": 1.3509173463944713, "learning_rate": 1.3248847926267283e-06, "loss": 0.5203, "step": 116 }, { "epoch": 0.00808848945731075, "grad_norm": 1.186681241697288, "learning_rate": 1.3364055299539171e-06, "loss": 0.3754, "step": 117 }, { "epoch": 0.008157621845834774, "grad_norm": 1.4453813834623022, "learning_rate": 1.347926267281106e-06, "loss": 0.5607, "step": 118 }, { "epoch": 0.008226754234358797, "grad_norm": 1.270171415461237, "learning_rate": 1.359447004608295e-06, "loss": 0.552, "step": 119 }, { "epoch": 0.00829588662288282, "grad_norm": 1.3994578502293613, "learning_rate": 1.3709677419354838e-06, "loss": 0.4408, "step": 120 }, { "epoch": 0.008365019011406844, "grad_norm": 1.188064646041792, "learning_rate": 1.382488479262673e-06, "loss": 0.531, "step": 121 }, { "epoch": 0.008434151399930867, "grad_norm": 1.4921759577255946, "learning_rate": 1.394009216589862e-06, "loss": 0.5897, "step": 122 }, { "epoch": 0.00850328378845489, "grad_norm": 2.16371156305486, "learning_rate": 1.4055299539170509e-06, "loss": 0.4203, "step": 123 }, { "epoch": 0.008572416176978915, "grad_norm": 1.12994585361747, "learning_rate": 1.4170506912442397e-06, "loss": 0.5092, "step": 124 }, { "epoch": 0.008641548565502939, "grad_norm": 1.3372577776785866, "learning_rate": 1.4285714285714286e-06, "loss": 0.5319, "step": 125 }, { "epoch": 0.008710680954026962, "grad_norm": 1.1494616786627583, "learning_rate": 1.4400921658986175e-06, "loss": 0.5768, "step": 126 }, { "epoch": 0.008779813342550985, "grad_norm": 1.189213499698125, "learning_rate": 1.4516129032258066e-06, "loss": 0.4304, "step": 127 }, { "epoch": 0.008848945731075008, "grad_norm": 1.532624062546067, "learning_rate": 1.4631336405529955e-06, "loss": 0.417, "step": 128 }, { "epoch": 0.008918078119599031, "grad_norm": 1.151192470533236, "learning_rate": 1.4746543778801844e-06, "loss": 0.5397, "step": 129 }, { "epoch": 0.008987210508123056, "grad_norm": 1.4538247804184805, "learning_rate": 1.4861751152073733e-06, "loss": 0.5097, "step": 130 }, { "epoch": 0.00905634289664708, "grad_norm": 1.0850544595872205, "learning_rate": 1.4976958525345621e-06, "loss": 0.3907, "step": 131 }, { "epoch": 0.009125475285171103, "grad_norm": 1.185332492083601, "learning_rate": 1.5092165898617514e-06, "loss": 0.4739, "step": 132 }, { "epoch": 0.009194607673695126, "grad_norm": 1.2759436112853506, "learning_rate": 1.5207373271889403e-06, "loss": 0.5734, "step": 133 }, { "epoch": 0.00926374006221915, "grad_norm": 1.2371382662733574, "learning_rate": 1.5322580645161292e-06, "loss": 0.4263, "step": 134 }, { "epoch": 0.009332872450743173, "grad_norm": 1.1060648594095186, "learning_rate": 1.543778801843318e-06, "loss": 0.377, "step": 135 }, { "epoch": 0.009402004839267196, "grad_norm": 1.2211885180066708, "learning_rate": 1.555299539170507e-06, "loss": 0.604, "step": 136 }, { "epoch": 0.009471137227791221, "grad_norm": 1.3341426217177204, "learning_rate": 1.5668202764976959e-06, "loss": 0.5747, "step": 137 }, { "epoch": 0.009540269616315244, "grad_norm": 1.0817353003709607, "learning_rate": 1.578341013824885e-06, "loss": 0.4176, "step": 138 }, { "epoch": 0.009609402004839267, "grad_norm": 1.2508473442714214, "learning_rate": 1.5898617511520738e-06, "loss": 0.4687, "step": 139 }, { "epoch": 0.00967853439336329, "grad_norm": 1.1284326356737397, "learning_rate": 1.6013824884792627e-06, "loss": 0.5708, "step": 140 }, { "epoch": 0.009747666781887314, "grad_norm": 1.233162064846663, "learning_rate": 1.6129032258064516e-06, "loss": 0.4688, "step": 141 }, { "epoch": 0.009816799170411337, "grad_norm": 1.2889017511916188, "learning_rate": 1.6244239631336407e-06, "loss": 0.403, "step": 142 }, { "epoch": 0.00988593155893536, "grad_norm": 1.4093614082269197, "learning_rate": 1.6359447004608298e-06, "loss": 0.5309, "step": 143 }, { "epoch": 0.009955063947459385, "grad_norm": 1.2250658214240213, "learning_rate": 1.6474654377880187e-06, "loss": 0.5017, "step": 144 }, { "epoch": 0.010024196335983409, "grad_norm": 1.2074848825987257, "learning_rate": 1.6589861751152075e-06, "loss": 0.4504, "step": 145 }, { "epoch": 0.010093328724507432, "grad_norm": 1.1027790128134336, "learning_rate": 1.6705069124423964e-06, "loss": 0.4444, "step": 146 }, { "epoch": 0.010162461113031455, "grad_norm": 1.1442550749085312, "learning_rate": 1.6820276497695853e-06, "loss": 0.5837, "step": 147 }, { "epoch": 0.010231593501555478, "grad_norm": 1.634041975115378, "learning_rate": 1.6935483870967742e-06, "loss": 0.4638, "step": 148 }, { "epoch": 0.010300725890079502, "grad_norm": 1.094249606033789, "learning_rate": 1.7050691244239633e-06, "loss": 0.4046, "step": 149 }, { "epoch": 0.010369858278603527, "grad_norm": 1.1536126720458706, "learning_rate": 1.7165898617511522e-06, "loss": 0.4635, "step": 150 }, { "epoch": 0.01043899066712755, "grad_norm": 1.2448910998421052, "learning_rate": 1.7281105990783413e-06, "loss": 0.5963, "step": 151 }, { "epoch": 0.010508123055651573, "grad_norm": 1.1310113620927191, "learning_rate": 1.7396313364055301e-06, "loss": 0.4019, "step": 152 }, { "epoch": 0.010577255444175596, "grad_norm": 1.5507805009326183, "learning_rate": 1.751152073732719e-06, "loss": 0.4587, "step": 153 }, { "epoch": 0.01064638783269962, "grad_norm": 1.155500845113847, "learning_rate": 1.762672811059908e-06, "loss": 0.5539, "step": 154 }, { "epoch": 0.010715520221223643, "grad_norm": 1.2384797989322383, "learning_rate": 1.774193548387097e-06, "loss": 0.508, "step": 155 }, { "epoch": 0.010784652609747666, "grad_norm": 1.2115712786096806, "learning_rate": 1.7857142857142859e-06, "loss": 0.4287, "step": 156 }, { "epoch": 0.010853784998271691, "grad_norm": 1.1885268019092419, "learning_rate": 1.7972350230414748e-06, "loss": 0.4242, "step": 157 }, { "epoch": 0.010922917386795714, "grad_norm": 1.0980986308901606, "learning_rate": 1.8087557603686637e-06, "loss": 0.5273, "step": 158 }, { "epoch": 0.010992049775319737, "grad_norm": 1.2534332993585773, "learning_rate": 1.8202764976958525e-06, "loss": 0.3849, "step": 159 }, { "epoch": 0.01106118216384376, "grad_norm": 1.1321438660816612, "learning_rate": 1.8317972350230418e-06, "loss": 0.422, "step": 160 }, { "epoch": 0.011130314552367784, "grad_norm": 1.121671582474042, "learning_rate": 1.8433179723502307e-06, "loss": 0.5088, "step": 161 }, { "epoch": 0.011199446940891807, "grad_norm": 1.1103806895451467, "learning_rate": 1.8548387096774196e-06, "loss": 0.5011, "step": 162 }, { "epoch": 0.01126857932941583, "grad_norm": 1.1752212423567987, "learning_rate": 1.8663594470046085e-06, "loss": 0.362, "step": 163 }, { "epoch": 0.011337711717939855, "grad_norm": 1.4572113737756471, "learning_rate": 1.8778801843317974e-06, "loss": 0.4121, "step": 164 }, { "epoch": 0.011406844106463879, "grad_norm": 1.3449725880675136, "learning_rate": 1.8894009216589863e-06, "loss": 0.5903, "step": 165 }, { "epoch": 0.011475976494987902, "grad_norm": 1.1634171781961238, "learning_rate": 1.9009216589861754e-06, "loss": 0.5223, "step": 166 }, { "epoch": 0.011545108883511925, "grad_norm": 1.1151411074538389, "learning_rate": 1.9124423963133642e-06, "loss": 0.3819, "step": 167 }, { "epoch": 0.011614241272035948, "grad_norm": 1.1725566595412353, "learning_rate": 1.923963133640553e-06, "loss": 0.5071, "step": 168 }, { "epoch": 0.011683373660559972, "grad_norm": 1.1308906410218031, "learning_rate": 1.935483870967742e-06, "loss": 0.5372, "step": 169 }, { "epoch": 0.011752506049083997, "grad_norm": 1.2557807970717432, "learning_rate": 1.947004608294931e-06, "loss": 0.4238, "step": 170 }, { "epoch": 0.01182163843760802, "grad_norm": 1.1332992880098687, "learning_rate": 1.95852534562212e-06, "loss": 0.3869, "step": 171 }, { "epoch": 0.011890770826132043, "grad_norm": 1.2914049602435962, "learning_rate": 1.970046082949309e-06, "loss": 0.6199, "step": 172 }, { "epoch": 0.011959903214656066, "grad_norm": 1.114855518018034, "learning_rate": 1.981566820276498e-06, "loss": 0.4641, "step": 173 }, { "epoch": 0.01202903560318009, "grad_norm": 1.0703710600210101, "learning_rate": 1.993087557603687e-06, "loss": 0.4512, "step": 174 }, { "epoch": 0.012098167991704113, "grad_norm": 1.1299266783906492, "learning_rate": 2.0046082949308757e-06, "loss": 0.4568, "step": 175 }, { "epoch": 0.012167300380228136, "grad_norm": 1.2468737121320466, "learning_rate": 2.0161290322580646e-06, "loss": 0.5509, "step": 176 }, { "epoch": 0.012236432768752161, "grad_norm": 1.4200039327672183, "learning_rate": 2.027649769585254e-06, "loss": 0.4353, "step": 177 }, { "epoch": 0.012305565157276184, "grad_norm": 1.1459944662780834, "learning_rate": 2.0391705069124428e-06, "loss": 0.3853, "step": 178 }, { "epoch": 0.012374697545800208, "grad_norm": 1.361765695066618, "learning_rate": 2.0506912442396317e-06, "loss": 0.5217, "step": 179 }, { "epoch": 0.01244382993432423, "grad_norm": 1.2168060961425464, "learning_rate": 2.0622119815668206e-06, "loss": 0.5047, "step": 180 }, { "epoch": 0.012512962322848254, "grad_norm": 1.6357908883825478, "learning_rate": 2.0737327188940094e-06, "loss": 0.4246, "step": 181 }, { "epoch": 0.012582094711372277, "grad_norm": 1.251392144669402, "learning_rate": 2.0852534562211983e-06, "loss": 0.4141, "step": 182 }, { "epoch": 0.0126512270998963, "grad_norm": 1.1423379775438067, "learning_rate": 2.096774193548387e-06, "loss": 0.5201, "step": 183 }, { "epoch": 0.012720359488420326, "grad_norm": 1.2526793398262468, "learning_rate": 2.108294930875576e-06, "loss": 0.4889, "step": 184 }, { "epoch": 0.012789491876944349, "grad_norm": 1.0851759019126994, "learning_rate": 2.119815668202765e-06, "loss": 0.3825, "step": 185 }, { "epoch": 0.012858624265468372, "grad_norm": 1.1196840204426646, "learning_rate": 2.131336405529954e-06, "loss": 0.5323, "step": 186 }, { "epoch": 0.012927756653992395, "grad_norm": 1.339356870616822, "learning_rate": 2.1428571428571427e-06, "loss": 0.4467, "step": 187 }, { "epoch": 0.012996889042516419, "grad_norm": 2.0694130424228177, "learning_rate": 2.154377880184332e-06, "loss": 0.3817, "step": 188 }, { "epoch": 0.013066021431040442, "grad_norm": 1.2742122428424147, "learning_rate": 2.165898617511521e-06, "loss": 0.4752, "step": 189 }, { "epoch": 0.013135153819564467, "grad_norm": 2.183700159269381, "learning_rate": 2.17741935483871e-06, "loss": 0.5271, "step": 190 }, { "epoch": 0.01320428620808849, "grad_norm": 1.0351783961380059, "learning_rate": 2.1889400921658987e-06, "loss": 0.4445, "step": 191 }, { "epoch": 0.013273418596612513, "grad_norm": 1.1153044966893517, "learning_rate": 2.2004608294930876e-06, "loss": 0.4444, "step": 192 }, { "epoch": 0.013342550985136536, "grad_norm": 1.136013382205602, "learning_rate": 2.211981566820277e-06, "loss": 0.4901, "step": 193 }, { "epoch": 0.01341168337366056, "grad_norm": 1.2056713921306719, "learning_rate": 2.2235023041474658e-06, "loss": 0.5801, "step": 194 }, { "epoch": 0.013480815762184583, "grad_norm": 1.1815860439649806, "learning_rate": 2.2350230414746546e-06, "loss": 0.4335, "step": 195 }, { "epoch": 0.013549948150708606, "grad_norm": 1.0947669244843998, "learning_rate": 2.2465437788018435e-06, "loss": 0.4637, "step": 196 }, { "epoch": 0.013619080539232631, "grad_norm": 1.0593317756997989, "learning_rate": 2.2580645161290324e-06, "loss": 0.5375, "step": 197 }, { "epoch": 0.013688212927756654, "grad_norm": 1.1009394656970166, "learning_rate": 2.2695852534562213e-06, "loss": 0.4978, "step": 198 }, { "epoch": 0.013757345316280678, "grad_norm": 1.2362604627517317, "learning_rate": 2.2811059907834106e-06, "loss": 0.4323, "step": 199 }, { "epoch": 0.013826477704804701, "grad_norm": 1.0944020993186645, "learning_rate": 2.2926267281105995e-06, "loss": 0.5209, "step": 200 }, { "epoch": 0.013895610093328724, "grad_norm": 1.069710740790938, "learning_rate": 2.3041474654377884e-06, "loss": 0.5142, "step": 201 }, { "epoch": 0.013964742481852747, "grad_norm": 1.2453487278419197, "learning_rate": 2.3156682027649772e-06, "loss": 0.409, "step": 202 }, { "epoch": 0.01403387487037677, "grad_norm": 1.0503571128962466, "learning_rate": 2.327188940092166e-06, "loss": 0.4256, "step": 203 }, { "epoch": 0.014103007258900796, "grad_norm": 1.0701677716380908, "learning_rate": 2.338709677419355e-06, "loss": 0.5383, "step": 204 }, { "epoch": 0.014172139647424819, "grad_norm": 1.2213453250802513, "learning_rate": 2.350230414746544e-06, "loss": 0.4557, "step": 205 }, { "epoch": 0.014241272035948842, "grad_norm": 1.1516101788611701, "learning_rate": 2.3617511520737328e-06, "loss": 0.3673, "step": 206 }, { "epoch": 0.014310404424472865, "grad_norm": 1.1693088429524088, "learning_rate": 2.3732718894009217e-06, "loss": 0.4198, "step": 207 }, { "epoch": 0.014379536812996889, "grad_norm": 1.1533922686891838, "learning_rate": 2.3847926267281105e-06, "loss": 0.554, "step": 208 }, { "epoch": 0.014448669201520912, "grad_norm": 1.1439359352294427, "learning_rate": 2.3963133640553e-06, "loss": 0.384, "step": 209 }, { "epoch": 0.014517801590044937, "grad_norm": 1.2285367419297846, "learning_rate": 2.4078341013824887e-06, "loss": 0.4027, "step": 210 }, { "epoch": 0.01458693397856896, "grad_norm": 1.1133312272769853, "learning_rate": 2.4193548387096776e-06, "loss": 0.4961, "step": 211 }, { "epoch": 0.014656066367092983, "grad_norm": 1.1630366736493916, "learning_rate": 2.4308755760368665e-06, "loss": 0.4598, "step": 212 }, { "epoch": 0.014725198755617007, "grad_norm": 1.0552097175091026, "learning_rate": 2.4423963133640554e-06, "loss": 0.4215, "step": 213 }, { "epoch": 0.01479433114414103, "grad_norm": 1.1165062481403663, "learning_rate": 2.4539170506912443e-06, "loss": 0.4483, "step": 214 }, { "epoch": 0.014863463532665053, "grad_norm": 1.0963898536637655, "learning_rate": 2.465437788018433e-06, "loss": 0.6367, "step": 215 }, { "epoch": 0.014932595921189076, "grad_norm": 1.3322507709640006, "learning_rate": 2.4769585253456224e-06, "loss": 0.4216, "step": 216 }, { "epoch": 0.015001728309713101, "grad_norm": 1.049235825802239, "learning_rate": 2.4884792626728113e-06, "loss": 0.378, "step": 217 }, { "epoch": 0.015070860698237125, "grad_norm": 1.130836915727969, "learning_rate": 2.5e-06, "loss": 0.5373, "step": 218 }, { "epoch": 0.015139993086761148, "grad_norm": 1.062829638240166, "learning_rate": 2.5115207373271895e-06, "loss": 0.4903, "step": 219 }, { "epoch": 0.015209125475285171, "grad_norm": 1.0338784884458483, "learning_rate": 2.523041474654378e-06, "loss": 0.4346, "step": 220 }, { "epoch": 0.015278257863809194, "grad_norm": 1.0684674648792372, "learning_rate": 2.5345622119815673e-06, "loss": 0.4056, "step": 221 }, { "epoch": 0.015347390252333218, "grad_norm": 1.218106431112793, "learning_rate": 2.5460829493087557e-06, "loss": 0.5386, "step": 222 }, { "epoch": 0.015416522640857242, "grad_norm": 2.673942537017698, "learning_rate": 2.557603686635945e-06, "loss": 0.4451, "step": 223 }, { "epoch": 0.015485655029381266, "grad_norm": 1.1590525711008295, "learning_rate": 2.569124423963134e-06, "loss": 0.3855, "step": 224 }, { "epoch": 0.015554787417905289, "grad_norm": 1.7137194934810192, "learning_rate": 2.580645161290323e-06, "loss": 0.468, "step": 225 }, { "epoch": 0.015623919806429312, "grad_norm": 1.1401310565526825, "learning_rate": 2.5921658986175117e-06, "loss": 0.5145, "step": 226 }, { "epoch": 0.015693052194953337, "grad_norm": 1.4624995999164183, "learning_rate": 2.6036866359447006e-06, "loss": 0.3865, "step": 227 }, { "epoch": 0.01576218458347736, "grad_norm": 1.1735394534334942, "learning_rate": 2.61520737327189e-06, "loss": 0.4085, "step": 228 }, { "epoch": 0.015831316972001384, "grad_norm": 1.1014890147680734, "learning_rate": 2.6267281105990783e-06, "loss": 0.5265, "step": 229 }, { "epoch": 0.015900449360525405, "grad_norm": 1.2361924976340946, "learning_rate": 2.6382488479262676e-06, "loss": 0.5106, "step": 230 }, { "epoch": 0.01596958174904943, "grad_norm": 1.1208728155130347, "learning_rate": 2.6497695852534565e-06, "loss": 0.373, "step": 231 }, { "epoch": 0.01603871413757345, "grad_norm": 1.2745732387149677, "learning_rate": 2.6612903225806454e-06, "loss": 0.4705, "step": 232 }, { "epoch": 0.016107846526097477, "grad_norm": 1.0625268788593725, "learning_rate": 2.6728110599078343e-06, "loss": 0.532, "step": 233 }, { "epoch": 0.0161769789146215, "grad_norm": 1.4338547367742491, "learning_rate": 2.684331797235023e-06, "loss": 0.4658, "step": 234 }, { "epoch": 0.016246111303145523, "grad_norm": 1.1661311559624397, "learning_rate": 2.695852534562212e-06, "loss": 0.4033, "step": 235 }, { "epoch": 0.016315243691669548, "grad_norm": 1.2616208244422273, "learning_rate": 2.7073732718894014e-06, "loss": 0.4546, "step": 236 }, { "epoch": 0.01638437608019357, "grad_norm": 1.5496764448462512, "learning_rate": 2.71889400921659e-06, "loss": 0.4981, "step": 237 }, { "epoch": 0.016453508468717595, "grad_norm": 1.2930140547937898, "learning_rate": 2.730414746543779e-06, "loss": 0.3965, "step": 238 }, { "epoch": 0.016522640857241616, "grad_norm": 1.1342886569128705, "learning_rate": 2.7419354838709676e-06, "loss": 0.3715, "step": 239 }, { "epoch": 0.01659177324576564, "grad_norm": 1.1560044281843784, "learning_rate": 2.753456221198157e-06, "loss": 0.5873, "step": 240 }, { "epoch": 0.016660905634289666, "grad_norm": 1.1411192617320278, "learning_rate": 2.764976958525346e-06, "loss": 0.3985, "step": 241 }, { "epoch": 0.016730038022813688, "grad_norm": 0.9274617477090726, "learning_rate": 2.7764976958525347e-06, "loss": 0.3979, "step": 242 }, { "epoch": 0.016799170411337713, "grad_norm": 1.5236039286442629, "learning_rate": 2.788018433179724e-06, "loss": 0.5025, "step": 243 }, { "epoch": 0.016868302799861734, "grad_norm": 1.6675911675437223, "learning_rate": 2.7995391705069124e-06, "loss": 0.4885, "step": 244 }, { "epoch": 0.01693743518838576, "grad_norm": 1.1220871508545842, "learning_rate": 2.8110599078341017e-06, "loss": 0.4554, "step": 245 }, { "epoch": 0.01700656757690978, "grad_norm": 1.116830291116706, "learning_rate": 2.822580645161291e-06, "loss": 0.4259, "step": 246 }, { "epoch": 0.017075699965433806, "grad_norm": 1.1282636923259928, "learning_rate": 2.8341013824884795e-06, "loss": 0.5628, "step": 247 }, { "epoch": 0.01714483235395783, "grad_norm": 1.121721884920774, "learning_rate": 2.845622119815669e-06, "loss": 0.4185, "step": 248 }, { "epoch": 0.017213964742481852, "grad_norm": 1.0450951793980237, "learning_rate": 2.8571428571428573e-06, "loss": 0.3641, "step": 249 }, { "epoch": 0.017283097131005877, "grad_norm": 1.090447718455109, "learning_rate": 2.8686635944700466e-06, "loss": 0.51, "step": 250 }, { "epoch": 0.0173522295195299, "grad_norm": 1.254133880621293, "learning_rate": 2.880184331797235e-06, "loss": 0.5228, "step": 251 }, { "epoch": 0.017421361908053924, "grad_norm": 1.1945463830377683, "learning_rate": 2.8917050691244243e-06, "loss": 0.4138, "step": 252 }, { "epoch": 0.01749049429657795, "grad_norm": 1.04787076368285, "learning_rate": 2.903225806451613e-06, "loss": 0.3533, "step": 253 }, { "epoch": 0.01755962668510197, "grad_norm": 1.2315243304726673, "learning_rate": 2.914746543778802e-06, "loss": 0.4792, "step": 254 }, { "epoch": 0.017628759073625995, "grad_norm": 1.172910099297306, "learning_rate": 2.926267281105991e-06, "loss": 0.4744, "step": 255 }, { "epoch": 0.017697891462150016, "grad_norm": 1.1761016660489836, "learning_rate": 2.93778801843318e-06, "loss": 0.3531, "step": 256 }, { "epoch": 0.01776702385067404, "grad_norm": 1.3748129458891465, "learning_rate": 2.9493087557603687e-06, "loss": 0.4579, "step": 257 }, { "epoch": 0.017836156239198063, "grad_norm": 1.133810302824861, "learning_rate": 2.960829493087558e-06, "loss": 0.555, "step": 258 }, { "epoch": 0.017905288627722088, "grad_norm": 1.2084757159542465, "learning_rate": 2.9723502304147465e-06, "loss": 0.3977, "step": 259 }, { "epoch": 0.017974421016246113, "grad_norm": 1.2588922062836412, "learning_rate": 2.983870967741936e-06, "loss": 0.3749, "step": 260 }, { "epoch": 0.018043553404770134, "grad_norm": 1.0853723976115512, "learning_rate": 2.9953917050691243e-06, "loss": 0.4993, "step": 261 }, { "epoch": 0.01811268579329416, "grad_norm": 1.483299933341881, "learning_rate": 3.0069124423963136e-06, "loss": 0.4829, "step": 262 }, { "epoch": 0.01818181818181818, "grad_norm": 1.0634035465072342, "learning_rate": 3.018433179723503e-06, "loss": 0.4013, "step": 263 }, { "epoch": 0.018250950570342206, "grad_norm": 2.3840286597697204, "learning_rate": 3.0299539170506913e-06, "loss": 0.3481, "step": 264 }, { "epoch": 0.018320082958866227, "grad_norm": 1.0838058896415377, "learning_rate": 3.0414746543778806e-06, "loss": 0.6022, "step": 265 }, { "epoch": 0.018389215347390252, "grad_norm": 2.20502277464021, "learning_rate": 3.052995391705069e-06, "loss": 0.5195, "step": 266 }, { "epoch": 0.018458347735914277, "grad_norm": 1.037586155899078, "learning_rate": 3.0645161290322584e-06, "loss": 0.3991, "step": 267 }, { "epoch": 0.0185274801244383, "grad_norm": 1.1511597811582115, "learning_rate": 3.076036866359447e-06, "loss": 0.4388, "step": 268 }, { "epoch": 0.018596612512962324, "grad_norm": 1.7535102407089453, "learning_rate": 3.087557603686636e-06, "loss": 0.471, "step": 269 }, { "epoch": 0.018665744901486345, "grad_norm": 1.085684364581938, "learning_rate": 3.0990783410138255e-06, "loss": 0.4105, "step": 270 }, { "epoch": 0.01873487729001037, "grad_norm": 1.069737621786555, "learning_rate": 3.110599078341014e-06, "loss": 0.3559, "step": 271 }, { "epoch": 0.018804009678534392, "grad_norm": 1.1167118385365813, "learning_rate": 3.1221198156682032e-06, "loss": 0.5655, "step": 272 }, { "epoch": 0.018873142067058417, "grad_norm": 1.0377647026571575, "learning_rate": 3.1336405529953917e-06, "loss": 0.4397, "step": 273 }, { "epoch": 0.018942274455582442, "grad_norm": 1.0418184227030611, "learning_rate": 3.145161290322581e-06, "loss": 0.3577, "step": 274 }, { "epoch": 0.019011406844106463, "grad_norm": 1.1311748827186927, "learning_rate": 3.15668202764977e-06, "loss": 0.4861, "step": 275 }, { "epoch": 0.01908053923263049, "grad_norm": 1.0328284198398903, "learning_rate": 3.1682027649769588e-06, "loss": 0.4893, "step": 276 }, { "epoch": 0.01914967162115451, "grad_norm": 1.0988838175982854, "learning_rate": 3.1797235023041477e-06, "loss": 0.4582, "step": 277 }, { "epoch": 0.019218804009678535, "grad_norm": 1.1031587537339402, "learning_rate": 3.1912442396313365e-06, "loss": 0.3449, "step": 278 }, { "epoch": 0.019287936398202556, "grad_norm": 1.1634672821954648, "learning_rate": 3.2027649769585254e-06, "loss": 0.5016, "step": 279 }, { "epoch": 0.01935706878672658, "grad_norm": 1.2378819346998877, "learning_rate": 3.2142857142857147e-06, "loss": 0.4821, "step": 280 }, { "epoch": 0.019426201175250606, "grad_norm": 1.212192841530856, "learning_rate": 3.225806451612903e-06, "loss": 0.3775, "step": 281 }, { "epoch": 0.019495333563774628, "grad_norm": 1.273417924695868, "learning_rate": 3.2373271889400925e-06, "loss": 0.4289, "step": 282 }, { "epoch": 0.019564465952298653, "grad_norm": 1.0605368982541856, "learning_rate": 3.2488479262672814e-06, "loss": 0.488, "step": 283 }, { "epoch": 0.019633598340822674, "grad_norm": 1.1174184076291074, "learning_rate": 3.2603686635944703e-06, "loss": 0.4324, "step": 284 }, { "epoch": 0.0197027307293467, "grad_norm": 1.1766882189237, "learning_rate": 3.2718894009216596e-06, "loss": 0.3844, "step": 285 }, { "epoch": 0.01977186311787072, "grad_norm": 1.2218001800427445, "learning_rate": 3.283410138248848e-06, "loss": 0.5236, "step": 286 }, { "epoch": 0.019840995506394746, "grad_norm": 1.0757220936008491, "learning_rate": 3.2949308755760373e-06, "loss": 0.5496, "step": 287 }, { "epoch": 0.01991012789491877, "grad_norm": 2.961308215826865, "learning_rate": 3.306451612903226e-06, "loss": 0.4121, "step": 288 }, { "epoch": 0.019979260283442792, "grad_norm": 1.1522510073103214, "learning_rate": 3.317972350230415e-06, "loss": 0.4065, "step": 289 }, { "epoch": 0.020048392671966817, "grad_norm": 1.2292315221616472, "learning_rate": 3.3294930875576036e-06, "loss": 0.5414, "step": 290 }, { "epoch": 0.02011752506049084, "grad_norm": 1.1504007882216973, "learning_rate": 3.341013824884793e-06, "loss": 0.3495, "step": 291 }, { "epoch": 0.020186657449014864, "grad_norm": 1.0788134203908983, "learning_rate": 3.352534562211982e-06, "loss": 0.3317, "step": 292 }, { "epoch": 0.02025578983753889, "grad_norm": 1.1592979078775822, "learning_rate": 3.3640552995391706e-06, "loss": 0.5025, "step": 293 }, { "epoch": 0.02032492222606291, "grad_norm": 1.1260990262001613, "learning_rate": 3.37557603686636e-06, "loss": 0.5162, "step": 294 }, { "epoch": 0.020394054614586935, "grad_norm": 1.153151293188433, "learning_rate": 3.3870967741935484e-06, "loss": 0.3915, "step": 295 }, { "epoch": 0.020463187003110957, "grad_norm": 1.0794556224463014, "learning_rate": 3.3986175115207377e-06, "loss": 0.4193, "step": 296 }, { "epoch": 0.02053231939163498, "grad_norm": 1.1920226514132668, "learning_rate": 3.4101382488479266e-06, "loss": 0.5291, "step": 297 }, { "epoch": 0.020601451780159003, "grad_norm": 1.2024345408794637, "learning_rate": 3.4216589861751155e-06, "loss": 0.5031, "step": 298 }, { "epoch": 0.020670584168683028, "grad_norm": 1.3406577783304463, "learning_rate": 3.4331797235023043e-06, "loss": 0.3676, "step": 299 }, { "epoch": 0.020739716557207053, "grad_norm": 1.1661914611408297, "learning_rate": 3.4447004608294932e-06, "loss": 0.4788, "step": 300 }, { "epoch": 0.020808848945731075, "grad_norm": 1.1653384361429857, "learning_rate": 3.4562211981566825e-06, "loss": 0.5039, "step": 301 }, { "epoch": 0.0208779813342551, "grad_norm": 1.4547365307789706, "learning_rate": 3.4677419354838714e-06, "loss": 0.402, "step": 302 }, { "epoch": 0.02094711372277912, "grad_norm": 0.956434048291889, "learning_rate": 3.4792626728110603e-06, "loss": 0.3409, "step": 303 }, { "epoch": 0.021016246111303146, "grad_norm": 1.096628155574155, "learning_rate": 3.490783410138249e-06, "loss": 0.5598, "step": 304 }, { "epoch": 0.021085378499827168, "grad_norm": 1.1548495224088238, "learning_rate": 3.502304147465438e-06, "loss": 0.5139, "step": 305 }, { "epoch": 0.021154510888351193, "grad_norm": 1.1160436097127184, "learning_rate": 3.513824884792627e-06, "loss": 0.4019, "step": 306 }, { "epoch": 0.021223643276875218, "grad_norm": 1.324086274640779, "learning_rate": 3.525345622119816e-06, "loss": 0.4089, "step": 307 }, { "epoch": 0.02129277566539924, "grad_norm": 1.1380723373846373, "learning_rate": 3.5368663594470047e-06, "loss": 0.4889, "step": 308 }, { "epoch": 0.021361908053923264, "grad_norm": 1.0237756400594207, "learning_rate": 3.548387096774194e-06, "loss": 0.3751, "step": 309 }, { "epoch": 0.021431040442447286, "grad_norm": 1.0336442882742292, "learning_rate": 3.5599078341013825e-06, "loss": 0.386, "step": 310 }, { "epoch": 0.02150017283097131, "grad_norm": 1.0481769414750322, "learning_rate": 3.5714285714285718e-06, "loss": 0.4741, "step": 311 }, { "epoch": 0.021569305219495332, "grad_norm": 1.092341290123691, "learning_rate": 3.5829493087557602e-06, "loss": 0.4371, "step": 312 }, { "epoch": 0.021638437608019357, "grad_norm": 1.0144604349811814, "learning_rate": 3.5944700460829495e-06, "loss": 0.3685, "step": 313 }, { "epoch": 0.021707569996543382, "grad_norm": 4.235482313625142, "learning_rate": 3.605990783410139e-06, "loss": 0.3881, "step": 314 }, { "epoch": 0.021776702385067404, "grad_norm": 1.1931862326807723, "learning_rate": 3.6175115207373273e-06, "loss": 0.4865, "step": 315 }, { "epoch": 0.02184583477359143, "grad_norm": 1.195733652271527, "learning_rate": 3.6290322580645166e-06, "loss": 0.4582, "step": 316 }, { "epoch": 0.02191496716211545, "grad_norm": 1.1039273987472102, "learning_rate": 3.640552995391705e-06, "loss": 0.4094, "step": 317 }, { "epoch": 0.021984099550639475, "grad_norm": 1.2432987578896761, "learning_rate": 3.6520737327188944e-06, "loss": 0.4966, "step": 318 }, { "epoch": 0.022053231939163497, "grad_norm": 1.1431893842636462, "learning_rate": 3.6635944700460837e-06, "loss": 0.5057, "step": 319 }, { "epoch": 0.02212236432768752, "grad_norm": 1.2151155240500455, "learning_rate": 3.675115207373272e-06, "loss": 0.3929, "step": 320 }, { "epoch": 0.022191496716211546, "grad_norm": 1.0700796869387839, "learning_rate": 3.6866359447004615e-06, "loss": 0.3306, "step": 321 }, { "epoch": 0.022260629104735568, "grad_norm": 1.3787226914691288, "learning_rate": 3.69815668202765e-06, "loss": 0.5418, "step": 322 }, { "epoch": 0.022329761493259593, "grad_norm": 1.2486804792275108, "learning_rate": 3.7096774193548392e-06, "loss": 0.5161, "step": 323 }, { "epoch": 0.022398893881783614, "grad_norm": 1.2279995581783258, "learning_rate": 3.721198156682028e-06, "loss": 0.3458, "step": 324 }, { "epoch": 0.02246802627030764, "grad_norm": 1.2145664339604818, "learning_rate": 3.732718894009217e-06, "loss": 0.3965, "step": 325 }, { "epoch": 0.02253715865883166, "grad_norm": 1.3245562008591067, "learning_rate": 3.744239631336406e-06, "loss": 0.4633, "step": 326 }, { "epoch": 0.022606291047355686, "grad_norm": 1.3321467654241774, "learning_rate": 3.7557603686635947e-06, "loss": 0.4102, "step": 327 }, { "epoch": 0.02267542343587971, "grad_norm": 1.3003789748912813, "learning_rate": 3.7672811059907836e-06, "loss": 0.3094, "step": 328 }, { "epoch": 0.022744555824403732, "grad_norm": 1.1351261247945634, "learning_rate": 3.7788018433179725e-06, "loss": 0.4836, "step": 329 }, { "epoch": 0.022813688212927757, "grad_norm": 1.1661197756104722, "learning_rate": 3.7903225806451614e-06, "loss": 0.4586, "step": 330 }, { "epoch": 0.02288282060145178, "grad_norm": 1.1427948682075821, "learning_rate": 3.8018433179723507e-06, "loss": 0.3753, "step": 331 }, { "epoch": 0.022951952989975804, "grad_norm": 1.0317548172434536, "learning_rate": 3.813364055299539e-06, "loss": 0.4258, "step": 332 }, { "epoch": 0.02302108537849983, "grad_norm": 1.1612900568699869, "learning_rate": 3.8248847926267285e-06, "loss": 0.5369, "step": 333 }, { "epoch": 0.02309021776702385, "grad_norm": 1.346369661474214, "learning_rate": 3.836405529953917e-06, "loss": 0.3792, "step": 334 }, { "epoch": 0.023159350155547875, "grad_norm": 1.396018643445313, "learning_rate": 3.847926267281106e-06, "loss": 0.4264, "step": 335 }, { "epoch": 0.023228482544071897, "grad_norm": 1.2319499656065642, "learning_rate": 3.8594470046082955e-06, "loss": 0.4429, "step": 336 }, { "epoch": 0.023297614932595922, "grad_norm": 1.6358664936256413, "learning_rate": 3.870967741935484e-06, "loss": 0.4619, "step": 337 }, { "epoch": 0.023366747321119943, "grad_norm": 1.077393311211583, "learning_rate": 3.882488479262673e-06, "loss": 0.3819, "step": 338 }, { "epoch": 0.02343587970964397, "grad_norm": 1.2117875648195366, "learning_rate": 3.894009216589862e-06, "loss": 0.4452, "step": 339 }, { "epoch": 0.023505012098167993, "grad_norm": 1.0843720378187003, "learning_rate": 3.905529953917051e-06, "loss": 0.5765, "step": 340 }, { "epoch": 0.023574144486692015, "grad_norm": 1.0288083482946964, "learning_rate": 3.91705069124424e-06, "loss": 0.3496, "step": 341 }, { "epoch": 0.02364327687521604, "grad_norm": 1.1498632654218335, "learning_rate": 3.928571428571429e-06, "loss": 0.4426, "step": 342 }, { "epoch": 0.02371240926374006, "grad_norm": 1.2113781850047785, "learning_rate": 3.940092165898618e-06, "loss": 0.4085, "step": 343 }, { "epoch": 0.023781541652264086, "grad_norm": 1.0726420046250202, "learning_rate": 3.951612903225807e-06, "loss": 0.4821, "step": 344 }, { "epoch": 0.023850674040788108, "grad_norm": 1.0183988838744094, "learning_rate": 3.963133640552996e-06, "loss": 0.3738, "step": 345 }, { "epoch": 0.023919806429312133, "grad_norm": 1.2079462372808007, "learning_rate": 3.974654377880185e-06, "loss": 0.3974, "step": 346 }, { "epoch": 0.023988938817836158, "grad_norm": 1.0787604668862503, "learning_rate": 3.986175115207374e-06, "loss": 0.4725, "step": 347 }, { "epoch": 0.02405807120636018, "grad_norm": 1.9996145788130864, "learning_rate": 3.997695852534563e-06, "loss": 0.425, "step": 348 }, { "epoch": 0.024127203594884204, "grad_norm": 1.0720894067789775, "learning_rate": 4.0092165898617514e-06, "loss": 0.4202, "step": 349 }, { "epoch": 0.024196335983408226, "grad_norm": 1.172087988451814, "learning_rate": 4.020737327188941e-06, "loss": 0.3942, "step": 350 }, { "epoch": 0.02426546837193225, "grad_norm": 1.2239948090924486, "learning_rate": 4.032258064516129e-06, "loss": 0.5625, "step": 351 }, { "epoch": 0.024334600760456272, "grad_norm": 1.212489868741383, "learning_rate": 4.0437788018433185e-06, "loss": 0.3847, "step": 352 }, { "epoch": 0.024403733148980297, "grad_norm": 1.0411607335840734, "learning_rate": 4.055299539170508e-06, "loss": 0.4227, "step": 353 }, { "epoch": 0.024472865537504322, "grad_norm": 1.1111175394603348, "learning_rate": 4.066820276497696e-06, "loss": 0.5251, "step": 354 }, { "epoch": 0.024541997926028344, "grad_norm": 1.0661326005670968, "learning_rate": 4.0783410138248856e-06, "loss": 0.4951, "step": 355 }, { "epoch": 0.02461113031455237, "grad_norm": 0.9548094201560263, "learning_rate": 4.089861751152074e-06, "loss": 0.3381, "step": 356 }, { "epoch": 0.02468026270307639, "grad_norm": 1.1385235590087472, "learning_rate": 4.101382488479263e-06, "loss": 0.4492, "step": 357 }, { "epoch": 0.024749395091600415, "grad_norm": 1.0819634694158966, "learning_rate": 4.112903225806452e-06, "loss": 0.5343, "step": 358 }, { "epoch": 0.024818527480124437, "grad_norm": 1.0842579727491615, "learning_rate": 4.124423963133641e-06, "loss": 0.3707, "step": 359 }, { "epoch": 0.02488765986864846, "grad_norm": 1.0045557849917068, "learning_rate": 4.1359447004608296e-06, "loss": 0.3557, "step": 360 }, { "epoch": 0.024956792257172487, "grad_norm": 1.135453709382368, "learning_rate": 4.147465437788019e-06, "loss": 0.4703, "step": 361 }, { "epoch": 0.025025924645696508, "grad_norm": 1.0434850849838886, "learning_rate": 4.158986175115207e-06, "loss": 0.4643, "step": 362 }, { "epoch": 0.025095057034220533, "grad_norm": 1.4053479869537957, "learning_rate": 4.170506912442397e-06, "loss": 0.4269, "step": 363 }, { "epoch": 0.025164189422744555, "grad_norm": 0.9448324257185977, "learning_rate": 4.182027649769586e-06, "loss": 0.3904, "step": 364 }, { "epoch": 0.02523332181126858, "grad_norm": 1.1789804425698407, "learning_rate": 4.193548387096774e-06, "loss": 0.5715, "step": 365 }, { "epoch": 0.0253024541997926, "grad_norm": 1.0590449625788887, "learning_rate": 4.205069124423964e-06, "loss": 0.4063, "step": 366 }, { "epoch": 0.025371586588316626, "grad_norm": 0.971661671687716, "learning_rate": 4.216589861751152e-06, "loss": 0.3417, "step": 367 }, { "epoch": 0.02544071897684065, "grad_norm": 1.0280114685003612, "learning_rate": 4.2281105990783415e-06, "loss": 0.4828, "step": 368 }, { "epoch": 0.025509851365364673, "grad_norm": 1.2103289517266138, "learning_rate": 4.23963133640553e-06, "loss": 0.5167, "step": 369 }, { "epoch": 0.025578983753888698, "grad_norm": 1.3473054218278377, "learning_rate": 4.251152073732719e-06, "loss": 0.3466, "step": 370 }, { "epoch": 0.02564811614241272, "grad_norm": 1.0888970169432495, "learning_rate": 4.262672811059908e-06, "loss": 0.3551, "step": 371 }, { "epoch": 0.025717248530936744, "grad_norm": 1.0804966562179734, "learning_rate": 4.274193548387097e-06, "loss": 0.576, "step": 372 }, { "epoch": 0.02578638091946077, "grad_norm": 1.00329516819645, "learning_rate": 4.2857142857142855e-06, "loss": 0.4409, "step": 373 }, { "epoch": 0.02585551330798479, "grad_norm": 1.3714769915227278, "learning_rate": 4.297235023041475e-06, "loss": 0.3375, "step": 374 }, { "epoch": 0.025924645696508816, "grad_norm": 1.0748458774392808, "learning_rate": 4.308755760368664e-06, "loss": 0.4883, "step": 375 }, { "epoch": 0.025993778085032837, "grad_norm": 1.2352954317561056, "learning_rate": 4.3202764976958525e-06, "loss": 0.562, "step": 376 }, { "epoch": 0.026062910473556862, "grad_norm": 1.1904055601256411, "learning_rate": 4.331797235023042e-06, "loss": 0.4105, "step": 377 }, { "epoch": 0.026132042862080884, "grad_norm": 3.314377435514267, "learning_rate": 4.34331797235023e-06, "loss": 0.3997, "step": 378 }, { "epoch": 0.02620117525060491, "grad_norm": 1.1044517292976377, "learning_rate": 4.35483870967742e-06, "loss": 0.542, "step": 379 }, { "epoch": 0.026270307639128933, "grad_norm": 1.0281749837600318, "learning_rate": 4.366359447004609e-06, "loss": 0.5186, "step": 380 }, { "epoch": 0.026339440027652955, "grad_norm": 0.9547275880408896, "learning_rate": 4.377880184331797e-06, "loss": 0.3632, "step": 381 }, { "epoch": 0.02640857241617698, "grad_norm": 1.170890854426869, "learning_rate": 4.389400921658987e-06, "loss": 0.4133, "step": 382 }, { "epoch": 0.026477704804701, "grad_norm": 1.1211542530946277, "learning_rate": 4.400921658986175e-06, "loss": 0.5549, "step": 383 }, { "epoch": 0.026546837193225026, "grad_norm": 1.0241085254643927, "learning_rate": 4.4124423963133644e-06, "loss": 0.3926, "step": 384 }, { "epoch": 0.026615969581749048, "grad_norm": 1.05667605715579, "learning_rate": 4.423963133640554e-06, "loss": 0.3611, "step": 385 }, { "epoch": 0.026685101970273073, "grad_norm": 1.2037064898864065, "learning_rate": 4.435483870967742e-06, "loss": 0.4369, "step": 386 }, { "epoch": 0.026754234358797098, "grad_norm": 1.0572335326217428, "learning_rate": 4.4470046082949315e-06, "loss": 0.3946, "step": 387 }, { "epoch": 0.02682336674732112, "grad_norm": 1.0513722922714657, "learning_rate": 4.45852534562212e-06, "loss": 0.4116, "step": 388 }, { "epoch": 0.026892499135845144, "grad_norm": 1.083219875797928, "learning_rate": 4.470046082949309e-06, "loss": 0.3788, "step": 389 }, { "epoch": 0.026961631524369166, "grad_norm": 1.0972904760923436, "learning_rate": 4.481566820276498e-06, "loss": 0.4912, "step": 390 }, { "epoch": 0.02703076391289319, "grad_norm": 1.2227674684818277, "learning_rate": 4.493087557603687e-06, "loss": 0.3795, "step": 391 }, { "epoch": 0.027099896301417212, "grad_norm": 1.0388592146910915, "learning_rate": 4.504608294930876e-06, "loss": 0.3735, "step": 392 }, { "epoch": 0.027169028689941237, "grad_norm": 1.1119521065395495, "learning_rate": 4.516129032258065e-06, "loss": 0.466, "step": 393 }, { "epoch": 0.027238161078465262, "grad_norm": 1.7935249951444974, "learning_rate": 4.527649769585254e-06, "loss": 0.467, "step": 394 }, { "epoch": 0.027307293466989284, "grad_norm": 2.5495051376592803, "learning_rate": 4.5391705069124426e-06, "loss": 0.4004, "step": 395 }, { "epoch": 0.02737642585551331, "grad_norm": 1.1462988098346545, "learning_rate": 4.550691244239632e-06, "loss": 0.3648, "step": 396 }, { "epoch": 0.02744555824403733, "grad_norm": 1.1085518394461817, "learning_rate": 4.562211981566821e-06, "loss": 0.5373, "step": 397 }, { "epoch": 0.027514690632561355, "grad_norm": 1.221864695115596, "learning_rate": 4.57373271889401e-06, "loss": 0.4361, "step": 398 }, { "epoch": 0.027583823021085377, "grad_norm": 2.100449992022917, "learning_rate": 4.585253456221199e-06, "loss": 0.3843, "step": 399 }, { "epoch": 0.027652955409609402, "grad_norm": 1.4472548989548064, "learning_rate": 4.596774193548387e-06, "loss": 0.4831, "step": 400 }, { "epoch": 0.027722087798133427, "grad_norm": 1.1377194421281138, "learning_rate": 4.608294930875577e-06, "loss": 0.5085, "step": 401 }, { "epoch": 0.02779122018665745, "grad_norm": 1.2429789681238088, "learning_rate": 4.619815668202766e-06, "loss": 0.2903, "step": 402 }, { "epoch": 0.027860352575181473, "grad_norm": 1.3588714470788665, "learning_rate": 4.6313364055299545e-06, "loss": 0.3706, "step": 403 }, { "epoch": 0.027929484963705495, "grad_norm": 1.0880589011855553, "learning_rate": 4.642857142857144e-06, "loss": 0.5183, "step": 404 }, { "epoch": 0.02799861735222952, "grad_norm": 1.2220084567928011, "learning_rate": 4.654377880184332e-06, "loss": 0.4664, "step": 405 }, { "epoch": 0.02806774974075354, "grad_norm": 1.5767741789929106, "learning_rate": 4.6658986175115215e-06, "loss": 0.3086, "step": 406 }, { "epoch": 0.028136882129277566, "grad_norm": 1.0812357389498446, "learning_rate": 4.67741935483871e-06, "loss": 0.4207, "step": 407 }, { "epoch": 0.02820601451780159, "grad_norm": 1.2769332758840122, "learning_rate": 4.688940092165899e-06, "loss": 0.4872, "step": 408 }, { "epoch": 0.028275146906325613, "grad_norm": 1.5683338589231162, "learning_rate": 4.700460829493088e-06, "loss": 0.3576, "step": 409 }, { "epoch": 0.028344279294849638, "grad_norm": 1.0735755849922715, "learning_rate": 4.711981566820277e-06, "loss": 0.3521, "step": 410 }, { "epoch": 0.02841341168337366, "grad_norm": 1.1878638711112415, "learning_rate": 4.7235023041474655e-06, "loss": 0.5379, "step": 411 }, { "epoch": 0.028482544071897684, "grad_norm": 1.1733408964202539, "learning_rate": 4.735023041474655e-06, "loss": 0.5069, "step": 412 }, { "epoch": 0.02855167646042171, "grad_norm": 0.9864135917427389, "learning_rate": 4.746543778801843e-06, "loss": 0.3735, "step": 413 }, { "epoch": 0.02862080884894573, "grad_norm": 1.1023174494661583, "learning_rate": 4.758064516129033e-06, "loss": 0.3863, "step": 414 }, { "epoch": 0.028689941237469756, "grad_norm": 1.0279960447492023, "learning_rate": 4.769585253456221e-06, "loss": 0.5228, "step": 415 }, { "epoch": 0.028759073625993777, "grad_norm": 1.09317731513656, "learning_rate": 4.78110599078341e-06, "loss": 0.402, "step": 416 }, { "epoch": 0.028828206014517802, "grad_norm": 1.2061242722108618, "learning_rate": 4.7926267281106e-06, "loss": 0.3811, "step": 417 }, { "epoch": 0.028897338403041824, "grad_norm": 1.9109470372554445, "learning_rate": 4.804147465437788e-06, "loss": 0.4673, "step": 418 }, { "epoch": 0.02896647079156585, "grad_norm": 0.9763995088760736, "learning_rate": 4.8156682027649774e-06, "loss": 0.4249, "step": 419 }, { "epoch": 0.029035603180089874, "grad_norm": 1.06572985332898, "learning_rate": 4.827188940092166e-06, "loss": 0.3798, "step": 420 }, { "epoch": 0.029104735568613895, "grad_norm": 1.591140576635712, "learning_rate": 4.838709677419355e-06, "loss": 0.3714, "step": 421 }, { "epoch": 0.02917386795713792, "grad_norm": 1.1611164721685463, "learning_rate": 4.850230414746544e-06, "loss": 0.4533, "step": 422 }, { "epoch": 0.02924300034566194, "grad_norm": 1.3280427930383554, "learning_rate": 4.861751152073733e-06, "loss": 0.4349, "step": 423 }, { "epoch": 0.029312132734185967, "grad_norm": 1.1469652543394955, "learning_rate": 4.873271889400922e-06, "loss": 0.3783, "step": 424 }, { "epoch": 0.029381265122709988, "grad_norm": 1.0518158529186095, "learning_rate": 4.884792626728111e-06, "loss": 0.4098, "step": 425 }, { "epoch": 0.029450397511234013, "grad_norm": 1.1484673468529352, "learning_rate": 4.8963133640553e-06, "loss": 0.5116, "step": 426 }, { "epoch": 0.029519529899758038, "grad_norm": 1.1644439647246236, "learning_rate": 4.9078341013824885e-06, "loss": 0.2995, "step": 427 }, { "epoch": 0.02958866228828206, "grad_norm": 1.9111590440172246, "learning_rate": 4.919354838709678e-06, "loss": 0.3036, "step": 428 }, { "epoch": 0.029657794676806085, "grad_norm": 1.0826644953996358, "learning_rate": 4.930875576036866e-06, "loss": 0.4944, "step": 429 }, { "epoch": 0.029726927065330106, "grad_norm": 1.288069602664274, "learning_rate": 4.9423963133640556e-06, "loss": 0.4021, "step": 430 }, { "epoch": 0.02979605945385413, "grad_norm": 1.2724800745429405, "learning_rate": 4.953917050691245e-06, "loss": 0.3607, "step": 431 }, { "epoch": 0.029865191842378153, "grad_norm": 2.3278130688021936, "learning_rate": 4.965437788018433e-06, "loss": 0.368, "step": 432 }, { "epoch": 0.029934324230902178, "grad_norm": 1.3526989457956613, "learning_rate": 4.976958525345623e-06, "loss": 0.4778, "step": 433 }, { "epoch": 0.030003456619426203, "grad_norm": 1.0917239527258276, "learning_rate": 4.988479262672811e-06, "loss": 0.3977, "step": 434 }, { "epoch": 0.030072589007950224, "grad_norm": 1.0179829069680157, "learning_rate": 5e-06, "loss": 0.3761, "step": 435 }, { "epoch": 0.03014172139647425, "grad_norm": 1.0496462349684643, "learning_rate": 5.011520737327189e-06, "loss": 0.4479, "step": 436 }, { "epoch": 0.03021085378499827, "grad_norm": 1.1081881922393253, "learning_rate": 5.023041474654379e-06, "loss": 0.4226, "step": 437 }, { "epoch": 0.030279986173522296, "grad_norm": 1.0615118318626002, "learning_rate": 5.0345622119815675e-06, "loss": 0.3632, "step": 438 }, { "epoch": 0.030349118562046317, "grad_norm": 1.230399450597272, "learning_rate": 5.046082949308756e-06, "loss": 0.3985, "step": 439 }, { "epoch": 0.030418250950570342, "grad_norm": 1.1315612191544149, "learning_rate": 5.057603686635944e-06, "loss": 0.518, "step": 440 }, { "epoch": 0.030487383339094367, "grad_norm": 2.4109733403992584, "learning_rate": 5.0691244239631346e-06, "loss": 0.4245, "step": 441 }, { "epoch": 0.03055651572761839, "grad_norm": 1.0907153105658498, "learning_rate": 5.080645161290323e-06, "loss": 0.3414, "step": 442 }, { "epoch": 0.030625648116142414, "grad_norm": 1.0265031722315197, "learning_rate": 5.0921658986175115e-06, "loss": 0.472, "step": 443 }, { "epoch": 0.030694780504666435, "grad_norm": 1.1534960391075628, "learning_rate": 5.103686635944702e-06, "loss": 0.447, "step": 444 }, { "epoch": 0.03076391289319046, "grad_norm": 1.035752793251534, "learning_rate": 5.11520737327189e-06, "loss": 0.3812, "step": 445 }, { "epoch": 0.030833045281714485, "grad_norm": 1.1047960346932835, "learning_rate": 5.1267281105990785e-06, "loss": 0.4083, "step": 446 }, { "epoch": 0.030902177670238506, "grad_norm": 1.1071400854707771, "learning_rate": 5.138248847926268e-06, "loss": 0.5025, "step": 447 }, { "epoch": 0.03097131005876253, "grad_norm": 1.1085147763933902, "learning_rate": 5.149769585253457e-06, "loss": 0.4717, "step": 448 }, { "epoch": 0.031040442447286553, "grad_norm": 1.1805398977908728, "learning_rate": 5.161290322580646e-06, "loss": 0.347, "step": 449 }, { "epoch": 0.031109574835810578, "grad_norm": 1.4633057898429351, "learning_rate": 5.172811059907834e-06, "loss": 0.4774, "step": 450 }, { "epoch": 0.0311787072243346, "grad_norm": 1.2616061855834568, "learning_rate": 5.184331797235023e-06, "loss": 0.4975, "step": 451 }, { "epoch": 0.031247839612858624, "grad_norm": 1.4247382011332885, "learning_rate": 5.195852534562213e-06, "loss": 0.3764, "step": 452 }, { "epoch": 0.03131697200138265, "grad_norm": 1.0771463135436101, "learning_rate": 5.207373271889401e-06, "loss": 0.3445, "step": 453 }, { "epoch": 0.031386104389906674, "grad_norm": 1.0961801463225942, "learning_rate": 5.2188940092165904e-06, "loss": 0.4746, "step": 454 }, { "epoch": 0.03145523677843069, "grad_norm": 1.0513140439094075, "learning_rate": 5.23041474654378e-06, "loss": 0.4435, "step": 455 }, { "epoch": 0.03152436916695472, "grad_norm": 1.0302965964600064, "learning_rate": 5.241935483870968e-06, "loss": 0.3508, "step": 456 }, { "epoch": 0.03159350155547874, "grad_norm": 1.050211624699588, "learning_rate": 5.253456221198157e-06, "loss": 0.441, "step": 457 }, { "epoch": 0.03166263394400277, "grad_norm": 1.1261461996530964, "learning_rate": 5.264976958525346e-06, "loss": 0.5307, "step": 458 }, { "epoch": 0.031731766332526785, "grad_norm": 1.0857431476853958, "learning_rate": 5.276497695852535e-06, "loss": 0.3814, "step": 459 }, { "epoch": 0.03180089872105081, "grad_norm": 1.3232126929547223, "learning_rate": 5.288018433179724e-06, "loss": 0.3632, "step": 460 }, { "epoch": 0.031870031109574835, "grad_norm": 1.0135981766200288, "learning_rate": 5.299539170506913e-06, "loss": 0.4145, "step": 461 }, { "epoch": 0.03193916349809886, "grad_norm": 1.8717853007074239, "learning_rate": 5.3110599078341015e-06, "loss": 0.443, "step": 462 }, { "epoch": 0.032008295886622885, "grad_norm": 1.4550606472378833, "learning_rate": 5.322580645161291e-06, "loss": 0.3435, "step": 463 }, { "epoch": 0.0320774282751469, "grad_norm": 1.108588450723349, "learning_rate": 5.33410138248848e-06, "loss": 0.4112, "step": 464 }, { "epoch": 0.03214656066367093, "grad_norm": 1.1441695426121183, "learning_rate": 5.345622119815669e-06, "loss": 0.5621, "step": 465 }, { "epoch": 0.03221569305219495, "grad_norm": 1.3659481411222172, "learning_rate": 5.357142857142857e-06, "loss": 0.3855, "step": 466 }, { "epoch": 0.03228482544071898, "grad_norm": 1.0565152104461206, "learning_rate": 5.368663594470046e-06, "loss": 0.3902, "step": 467 }, { "epoch": 0.032353957829243, "grad_norm": 1.1664439643838291, "learning_rate": 5.380184331797236e-06, "loss": 0.5573, "step": 468 }, { "epoch": 0.03242309021776702, "grad_norm": 1.1120140884819534, "learning_rate": 5.391705069124424e-06, "loss": 0.5536, "step": 469 }, { "epoch": 0.032492222606291046, "grad_norm": 1.202312486361394, "learning_rate": 5.4032258064516126e-06, "loss": 0.3461, "step": 470 }, { "epoch": 0.03256135499481507, "grad_norm": 1.069332859850439, "learning_rate": 5.414746543778803e-06, "loss": 0.4079, "step": 471 }, { "epoch": 0.032630487383339096, "grad_norm": 2.046478831565821, "learning_rate": 5.426267281105991e-06, "loss": 0.5395, "step": 472 }, { "epoch": 0.03269961977186312, "grad_norm": 1.1451679192561364, "learning_rate": 5.43778801843318e-06, "loss": 0.5589, "step": 473 }, { "epoch": 0.03276875216038714, "grad_norm": 0.9768543014740266, "learning_rate": 5.44930875576037e-06, "loss": 0.3591, "step": 474 }, { "epoch": 0.032837884548911164, "grad_norm": 1.1808965113358836, "learning_rate": 5.460829493087558e-06, "loss": 0.3943, "step": 475 }, { "epoch": 0.03290701693743519, "grad_norm": 1.0050482800477858, "learning_rate": 5.472350230414747e-06, "loss": 0.5231, "step": 476 }, { "epoch": 0.032976149325959214, "grad_norm": 1.1329861378916495, "learning_rate": 5.483870967741935e-06, "loss": 0.3336, "step": 477 }, { "epoch": 0.03304528171448323, "grad_norm": 1.252067461286509, "learning_rate": 5.495391705069125e-06, "loss": 0.4081, "step": 478 }, { "epoch": 0.03311441410300726, "grad_norm": 1.2953171527718492, "learning_rate": 5.506912442396314e-06, "loss": 0.4791, "step": 479 }, { "epoch": 0.03318354649153128, "grad_norm": 1.142099871478586, "learning_rate": 5.518433179723502e-06, "loss": 0.4744, "step": 480 }, { "epoch": 0.03325267888005531, "grad_norm": 1.346205377723496, "learning_rate": 5.529953917050692e-06, "loss": 0.3835, "step": 481 }, { "epoch": 0.03332181126857933, "grad_norm": 1.063122102685857, "learning_rate": 5.541474654377881e-06, "loss": 0.3812, "step": 482 }, { "epoch": 0.03339094365710335, "grad_norm": 2.489427083386171, "learning_rate": 5.552995391705069e-06, "loss": 0.4927, "step": 483 }, { "epoch": 0.033460076045627375, "grad_norm": 1.0516858064038022, "learning_rate": 5.564516129032258e-06, "loss": 0.3936, "step": 484 }, { "epoch": 0.0335292084341514, "grad_norm": 1.0986971491994515, "learning_rate": 5.576036866359448e-06, "loss": 0.3402, "step": 485 }, { "epoch": 0.033598340822675425, "grad_norm": 1.1187888189661575, "learning_rate": 5.587557603686636e-06, "loss": 0.5383, "step": 486 }, { "epoch": 0.03366747321119945, "grad_norm": 1.1953170407089242, "learning_rate": 5.599078341013825e-06, "loss": 0.4831, "step": 487 }, { "epoch": 0.03373660559972347, "grad_norm": 0.9925795092597788, "learning_rate": 5.610599078341015e-06, "loss": 0.3735, "step": 488 }, { "epoch": 0.03380573798824749, "grad_norm": 1.0641581350445042, "learning_rate": 5.6221198156682035e-06, "loss": 0.4192, "step": 489 }, { "epoch": 0.03387487037677152, "grad_norm": 1.1143646627439556, "learning_rate": 5.633640552995392e-06, "loss": 0.5346, "step": 490 }, { "epoch": 0.03394400276529554, "grad_norm": 1.1626799605075888, "learning_rate": 5.645161290322582e-06, "loss": 0.4367, "step": 491 }, { "epoch": 0.03401313515381956, "grad_norm": 1.004287621569033, "learning_rate": 5.6566820276497705e-06, "loss": 0.4143, "step": 492 }, { "epoch": 0.034082267542343586, "grad_norm": 1.226129566272293, "learning_rate": 5.668202764976959e-06, "loss": 0.482, "step": 493 }, { "epoch": 0.03415139993086761, "grad_norm": 1.0355139605835095, "learning_rate": 5.6797235023041474e-06, "loss": 0.5068, "step": 494 }, { "epoch": 0.034220532319391636, "grad_norm": 1.1141252580869638, "learning_rate": 5.691244239631338e-06, "loss": 0.3492, "step": 495 }, { "epoch": 0.03428966470791566, "grad_norm": 1.0895300085956248, "learning_rate": 5.702764976958526e-06, "loss": 0.3722, "step": 496 }, { "epoch": 0.03435879709643968, "grad_norm": 1.1437475461860387, "learning_rate": 5.7142857142857145e-06, "loss": 0.4788, "step": 497 }, { "epoch": 0.034427929484963704, "grad_norm": 1.1686087028570262, "learning_rate": 5.725806451612904e-06, "loss": 0.4055, "step": 498 }, { "epoch": 0.03449706187348773, "grad_norm": 1.3445781196397795, "learning_rate": 5.737327188940093e-06, "loss": 0.3687, "step": 499 }, { "epoch": 0.034566194262011754, "grad_norm": 1.1090588027532686, "learning_rate": 5.748847926267282e-06, "loss": 0.3963, "step": 500 }, { "epoch": 0.03463532665053578, "grad_norm": 1.2464165206828122, "learning_rate": 5.76036866359447e-06, "loss": 0.4767, "step": 501 }, { "epoch": 0.0347044590390598, "grad_norm": 1.0555621399114061, "learning_rate": 5.771889400921659e-06, "loss": 0.3393, "step": 502 }, { "epoch": 0.03477359142758382, "grad_norm": 1.6662722095972196, "learning_rate": 5.783410138248849e-06, "loss": 0.3407, "step": 503 }, { "epoch": 0.03484272381610785, "grad_norm": 1.110019857374856, "learning_rate": 5.794930875576037e-06, "loss": 0.5028, "step": 504 }, { "epoch": 0.03491185620463187, "grad_norm": 1.1210301912659302, "learning_rate": 5.806451612903226e-06, "loss": 0.4238, "step": 505 }, { "epoch": 0.0349809885931559, "grad_norm": 1.0899316597855069, "learning_rate": 5.817972350230415e-06, "loss": 0.3799, "step": 506 }, { "epoch": 0.035050120981679915, "grad_norm": 1.1460673252925682, "learning_rate": 5.829493087557604e-06, "loss": 0.3739, "step": 507 }, { "epoch": 0.03511925337020394, "grad_norm": 1.3137598799529346, "learning_rate": 5.8410138248847935e-06, "loss": 0.4804, "step": 508 }, { "epoch": 0.035188385758727965, "grad_norm": 1.1238337558804652, "learning_rate": 5.852534562211982e-06, "loss": 0.3657, "step": 509 }, { "epoch": 0.03525751814725199, "grad_norm": 0.9734550485813794, "learning_rate": 5.864055299539171e-06, "loss": 0.323, "step": 510 }, { "epoch": 0.03532665053577601, "grad_norm": 1.147588738968184, "learning_rate": 5.87557603686636e-06, "loss": 0.4981, "step": 511 }, { "epoch": 0.03539578292430003, "grad_norm": 1.0673444325396224, "learning_rate": 5.887096774193549e-06, "loss": 0.4414, "step": 512 }, { "epoch": 0.03546491531282406, "grad_norm": 1.0832396127620558, "learning_rate": 5.8986175115207375e-06, "loss": 0.3378, "step": 513 }, { "epoch": 0.03553404770134808, "grad_norm": 1.0585798410588358, "learning_rate": 5.910138248847927e-06, "loss": 0.4433, "step": 514 }, { "epoch": 0.03560318008987211, "grad_norm": 1.0780619826451177, "learning_rate": 5.921658986175116e-06, "loss": 0.5364, "step": 515 }, { "epoch": 0.035672312478396126, "grad_norm": 1.2593902083095634, "learning_rate": 5.9331797235023045e-06, "loss": 0.3989, "step": 516 }, { "epoch": 0.03574144486692015, "grad_norm": 0.943695224670955, "learning_rate": 5.944700460829493e-06, "loss": 0.3322, "step": 517 }, { "epoch": 0.035810577255444176, "grad_norm": 8.548322247993495, "learning_rate": 5.956221198156682e-06, "loss": 0.4117, "step": 518 }, { "epoch": 0.0358797096439682, "grad_norm": 1.1259876714944368, "learning_rate": 5.967741935483872e-06, "loss": 0.5141, "step": 519 }, { "epoch": 0.035948842032492226, "grad_norm": 1.084837631980456, "learning_rate": 5.97926267281106e-06, "loss": 0.4373, "step": 520 }, { "epoch": 0.036017974421016244, "grad_norm": 1.1327037872180468, "learning_rate": 5.9907834101382485e-06, "loss": 0.426, "step": 521 }, { "epoch": 0.03608710680954027, "grad_norm": 1.0816977293849912, "learning_rate": 6.002304147465439e-06, "loss": 0.4915, "step": 522 }, { "epoch": 0.036156239198064294, "grad_norm": 1.0326489994013348, "learning_rate": 6.013824884792627e-06, "loss": 0.3983, "step": 523 }, { "epoch": 0.03622537158658832, "grad_norm": 1.0509527279683237, "learning_rate": 6.025345622119816e-06, "loss": 0.3542, "step": 524 }, { "epoch": 0.03629450397511234, "grad_norm": 1.8745064318706661, "learning_rate": 6.036866359447006e-06, "loss": 0.494, "step": 525 }, { "epoch": 0.03636363636363636, "grad_norm": 1.114441342921861, "learning_rate": 6.048387096774194e-06, "loss": 0.4944, "step": 526 }, { "epoch": 0.03643276875216039, "grad_norm": 1.0780349413736487, "learning_rate": 6.059907834101383e-06, "loss": 0.3693, "step": 527 }, { "epoch": 0.03650190114068441, "grad_norm": 1.205909634035231, "learning_rate": 6.071428571428571e-06, "loss": 0.3844, "step": 528 }, { "epoch": 0.03657103352920844, "grad_norm": 1.0968950343960073, "learning_rate": 6.082949308755761e-06, "loss": 0.5245, "step": 529 }, { "epoch": 0.036640165917732455, "grad_norm": 1.192152426815109, "learning_rate": 6.09447004608295e-06, "loss": 0.4346, "step": 530 }, { "epoch": 0.03670929830625648, "grad_norm": 1.410200171900497, "learning_rate": 6.105990783410138e-06, "loss": 0.4234, "step": 531 }, { "epoch": 0.036778430694780505, "grad_norm": 1.0765150830674992, "learning_rate": 6.117511520737328e-06, "loss": 0.3891, "step": 532 }, { "epoch": 0.03684756308330453, "grad_norm": 1.105120230238703, "learning_rate": 6.129032258064517e-06, "loss": 0.4996, "step": 533 }, { "epoch": 0.036916695471828555, "grad_norm": 1.2814552388912308, "learning_rate": 6.140552995391705e-06, "loss": 0.3429, "step": 534 }, { "epoch": 0.03698582786035257, "grad_norm": 1.221609421643265, "learning_rate": 6.152073732718894e-06, "loss": 0.3704, "step": 535 }, { "epoch": 0.0370549602488766, "grad_norm": 1.167415102318681, "learning_rate": 6.163594470046084e-06, "loss": 0.4678, "step": 536 }, { "epoch": 0.03712409263740062, "grad_norm": 1.4054368595458293, "learning_rate": 6.175115207373272e-06, "loss": 0.453, "step": 537 }, { "epoch": 0.03719322502592465, "grad_norm": 1.5242602073746008, "learning_rate": 6.186635944700461e-06, "loss": 0.368, "step": 538 }, { "epoch": 0.037262357414448666, "grad_norm": 1.0700573680712182, "learning_rate": 6.198156682027651e-06, "loss": 0.3678, "step": 539 }, { "epoch": 0.03733148980297269, "grad_norm": 1.2716482548555055, "learning_rate": 6.209677419354839e-06, "loss": 0.5985, "step": 540 }, { "epoch": 0.037400622191496716, "grad_norm": 1.0359526492090625, "learning_rate": 6.221198156682028e-06, "loss": 0.3729, "step": 541 }, { "epoch": 0.03746975458002074, "grad_norm": 1.3525920263490583, "learning_rate": 6.232718894009217e-06, "loss": 0.387, "step": 542 }, { "epoch": 0.037538886968544766, "grad_norm": 1.1570947981960056, "learning_rate": 6.2442396313364065e-06, "loss": 0.4129, "step": 543 }, { "epoch": 0.037608019357068784, "grad_norm": 1.266236951788563, "learning_rate": 6.255760368663595e-06, "loss": 0.4805, "step": 544 }, { "epoch": 0.03767715174559281, "grad_norm": 1.0877349949528383, "learning_rate": 6.267281105990783e-06, "loss": 0.366, "step": 545 }, { "epoch": 0.037746284134116834, "grad_norm": 1.2207810298112842, "learning_rate": 6.2788018433179736e-06, "loss": 0.4488, "step": 546 }, { "epoch": 0.03781541652264086, "grad_norm": 1.3590726254213972, "learning_rate": 6.290322580645162e-06, "loss": 0.5383, "step": 547 }, { "epoch": 0.037884548911164884, "grad_norm": 1.1835038237288091, "learning_rate": 6.3018433179723505e-06, "loss": 0.3915, "step": 548 }, { "epoch": 0.0379536812996889, "grad_norm": 1.3001505743511272, "learning_rate": 6.31336405529954e-06, "loss": 0.3759, "step": 549 }, { "epoch": 0.03802281368821293, "grad_norm": 1.1434494107881603, "learning_rate": 6.324884792626729e-06, "loss": 0.419, "step": 550 }, { "epoch": 0.03809194607673695, "grad_norm": 1.0230888193188206, "learning_rate": 6.3364055299539176e-06, "loss": 0.4527, "step": 551 }, { "epoch": 0.03816107846526098, "grad_norm": 1.1095180193558074, "learning_rate": 6.347926267281107e-06, "loss": 0.3999, "step": 552 }, { "epoch": 0.038230210853785, "grad_norm": 1.045128170699684, "learning_rate": 6.359447004608295e-06, "loss": 0.3854, "step": 553 }, { "epoch": 0.03829934324230902, "grad_norm": 1.4072442419883855, "learning_rate": 6.370967741935485e-06, "loss": 0.4687, "step": 554 }, { "epoch": 0.038368475630833045, "grad_norm": 1.1864308055070407, "learning_rate": 6.382488479262673e-06, "loss": 0.4913, "step": 555 }, { "epoch": 0.03843760801935707, "grad_norm": 1.2058242571495739, "learning_rate": 6.394009216589862e-06, "loss": 0.3317, "step": 556 }, { "epoch": 0.038506740407881095, "grad_norm": 1.1727034432702241, "learning_rate": 6.405529953917051e-06, "loss": 0.4409, "step": 557 }, { "epoch": 0.03857587279640511, "grad_norm": 1.266446573695054, "learning_rate": 6.41705069124424e-06, "loss": 0.4618, "step": 558 }, { "epoch": 0.03864500518492914, "grad_norm": 1.8713648306070605, "learning_rate": 6.4285714285714295e-06, "loss": 0.3239, "step": 559 }, { "epoch": 0.03871413757345316, "grad_norm": 1.0947632689369486, "learning_rate": 6.440092165898618e-06, "loss": 0.3864, "step": 560 }, { "epoch": 0.03878326996197719, "grad_norm": 1.2662951813871925, "learning_rate": 6.451612903225806e-06, "loss": 0.4664, "step": 561 }, { "epoch": 0.03885240235050121, "grad_norm": 1.1256710079240961, "learning_rate": 6.463133640552996e-06, "loss": 0.4294, "step": 562 }, { "epoch": 0.03892153473902523, "grad_norm": 1.1164892600962764, "learning_rate": 6.474654377880185e-06, "loss": 0.368, "step": 563 }, { "epoch": 0.038990667127549256, "grad_norm": 1.1056412834834468, "learning_rate": 6.4861751152073734e-06, "loss": 0.3938, "step": 564 }, { "epoch": 0.03905979951607328, "grad_norm": 1.2145696851161, "learning_rate": 6.497695852534563e-06, "loss": 0.5077, "step": 565 }, { "epoch": 0.039128931904597306, "grad_norm": 1.4134741391433332, "learning_rate": 6.509216589861752e-06, "loss": 0.4014, "step": 566 }, { "epoch": 0.03919806429312133, "grad_norm": 1.2065711759625588, "learning_rate": 6.5207373271889405e-06, "loss": 0.3465, "step": 567 }, { "epoch": 0.03926719668164535, "grad_norm": 1.0841952502306536, "learning_rate": 6.532258064516129e-06, "loss": 0.4473, "step": 568 }, { "epoch": 0.039336329070169374, "grad_norm": 1.092755187735849, "learning_rate": 6.543778801843319e-06, "loss": 0.5279, "step": 569 }, { "epoch": 0.0394054614586934, "grad_norm": 1.1025988106964917, "learning_rate": 6.555299539170508e-06, "loss": 0.4152, "step": 570 }, { "epoch": 0.039474593847217423, "grad_norm": 1.130739036727914, "learning_rate": 6.566820276497696e-06, "loss": 0.3542, "step": 571 }, { "epoch": 0.03954372623574144, "grad_norm": 1.3306253453657948, "learning_rate": 6.5783410138248845e-06, "loss": 0.574, "step": 572 }, { "epoch": 0.039612858624265467, "grad_norm": 1.1213813711469078, "learning_rate": 6.589861751152075e-06, "loss": 0.4348, "step": 573 }, { "epoch": 0.03968199101278949, "grad_norm": 1.380078999154775, "learning_rate": 6.601382488479263e-06, "loss": 0.3464, "step": 574 }, { "epoch": 0.039751123401313516, "grad_norm": 1.0302787581529407, "learning_rate": 6.612903225806452e-06, "loss": 0.4617, "step": 575 }, { "epoch": 0.03982025578983754, "grad_norm": 1.0226821597794005, "learning_rate": 6.624423963133642e-06, "loss": 0.5817, "step": 576 }, { "epoch": 0.03988938817836156, "grad_norm": 1.187461870040535, "learning_rate": 6.63594470046083e-06, "loss": 0.3647, "step": 577 }, { "epoch": 0.039958520566885584, "grad_norm": 1.0209886205019418, "learning_rate": 6.647465437788019e-06, "loss": 0.3303, "step": 578 }, { "epoch": 0.04002765295540961, "grad_norm": 1.244114805655698, "learning_rate": 6.658986175115207e-06, "loss": 0.4462, "step": 579 }, { "epoch": 0.040096785343933634, "grad_norm": 1.2594553229653596, "learning_rate": 6.670506912442397e-06, "loss": 0.441, "step": 580 }, { "epoch": 0.04016591773245766, "grad_norm": 1.3464729154837114, "learning_rate": 6.682027649769586e-06, "loss": 0.3567, "step": 581 }, { "epoch": 0.04023505012098168, "grad_norm": 1.027928328540199, "learning_rate": 6.693548387096774e-06, "loss": 0.4139, "step": 582 }, { "epoch": 0.0403041825095057, "grad_norm": 2.008211233183156, "learning_rate": 6.705069124423964e-06, "loss": 0.5435, "step": 583 }, { "epoch": 0.04037331489802973, "grad_norm": 1.9801262827718746, "learning_rate": 6.716589861751153e-06, "loss": 0.3817, "step": 584 }, { "epoch": 0.04044244728655375, "grad_norm": 2.1833527816256897, "learning_rate": 6.728110599078341e-06, "loss": 0.3476, "step": 585 }, { "epoch": 0.04051157967507778, "grad_norm": 1.1238146042689647, "learning_rate": 6.739631336405531e-06, "loss": 0.4533, "step": 586 }, { "epoch": 0.040580712063601795, "grad_norm": 1.0025287860464047, "learning_rate": 6.75115207373272e-06, "loss": 0.3914, "step": 587 }, { "epoch": 0.04064984445212582, "grad_norm": 1.0185211132616883, "learning_rate": 6.762672811059908e-06, "loss": 0.3338, "step": 588 }, { "epoch": 0.040718976840649845, "grad_norm": 1.0077048630523096, "learning_rate": 6.774193548387097e-06, "loss": 0.4732, "step": 589 }, { "epoch": 0.04078810922917387, "grad_norm": 1.040270771895413, "learning_rate": 6.785714285714287e-06, "loss": 0.494, "step": 590 }, { "epoch": 0.04085724161769789, "grad_norm": 1.394064520999901, "learning_rate": 6.797235023041475e-06, "loss": 0.3951, "step": 591 }, { "epoch": 0.04092637400622191, "grad_norm": 1.3252576045307012, "learning_rate": 6.808755760368664e-06, "loss": 0.4028, "step": 592 }, { "epoch": 0.04099550639474594, "grad_norm": 1.0849624318305737, "learning_rate": 6.820276497695853e-06, "loss": 0.4492, "step": 593 }, { "epoch": 0.04106463878326996, "grad_norm": 1.0373793299593366, "learning_rate": 6.8317972350230425e-06, "loss": 0.4333, "step": 594 }, { "epoch": 0.04113377117179399, "grad_norm": 1.0009796930566788, "learning_rate": 6.843317972350231e-06, "loss": 0.3707, "step": 595 }, { "epoch": 0.041202903560318006, "grad_norm": 1.2588049068585727, "learning_rate": 6.854838709677419e-06, "loss": 0.3481, "step": 596 }, { "epoch": 0.04127203594884203, "grad_norm": 1.0560364792178565, "learning_rate": 6.866359447004609e-06, "loss": 0.4852, "step": 597 }, { "epoch": 0.041341168337366056, "grad_norm": 1.0287441309957697, "learning_rate": 6.877880184331798e-06, "loss": 0.4462, "step": 598 }, { "epoch": 0.04141030072589008, "grad_norm": 1.3674785568974954, "learning_rate": 6.8894009216589865e-06, "loss": 0.3434, "step": 599 }, { "epoch": 0.041479433114414106, "grad_norm": 1.1543927916788546, "learning_rate": 6.900921658986176e-06, "loss": 0.4773, "step": 600 }, { "epoch": 0.041548565502938124, "grad_norm": 1.0194234275405167, "learning_rate": 6.912442396313365e-06, "loss": 0.479, "step": 601 }, { "epoch": 0.04161769789146215, "grad_norm": 1.0972233102145053, "learning_rate": 6.9239631336405535e-06, "loss": 0.3797, "step": 602 }, { "epoch": 0.041686830279986174, "grad_norm": 1.0582194336826716, "learning_rate": 6.935483870967743e-06, "loss": 0.327, "step": 603 }, { "epoch": 0.0417559626685102, "grad_norm": 1.1620953242325935, "learning_rate": 6.947004608294931e-06, "loss": 0.4718, "step": 604 }, { "epoch": 0.04182509505703422, "grad_norm": 1.1489220973853191, "learning_rate": 6.958525345622121e-06, "loss": 0.4677, "step": 605 }, { "epoch": 0.04189422744555824, "grad_norm": 1.3750349287061157, "learning_rate": 6.970046082949309e-06, "loss": 0.3198, "step": 606 }, { "epoch": 0.04196335983408227, "grad_norm": 1.0103018895538223, "learning_rate": 6.981566820276498e-06, "loss": 0.424, "step": 607 }, { "epoch": 0.04203249222260629, "grad_norm": 1.0570234452009153, "learning_rate": 6.993087557603687e-06, "loss": 0.4876, "step": 608 }, { "epoch": 0.04210162461113032, "grad_norm": 1.1953953014615077, "learning_rate": 7.004608294930876e-06, "loss": 0.3162, "step": 609 }, { "epoch": 0.042170756999654335, "grad_norm": 1.118588236495786, "learning_rate": 7.0161290322580654e-06, "loss": 0.3014, "step": 610 }, { "epoch": 0.04223988938817836, "grad_norm": 1.10170635664337, "learning_rate": 7.027649769585254e-06, "loss": 0.4758, "step": 611 }, { "epoch": 0.042309021776702385, "grad_norm": 1.1421594922072746, "learning_rate": 7.039170506912442e-06, "loss": 0.4701, "step": 612 }, { "epoch": 0.04237815416522641, "grad_norm": 1.1360297551351017, "learning_rate": 7.050691244239632e-06, "loss": 0.3272, "step": 613 }, { "epoch": 0.042447286553750435, "grad_norm": 3.141016537666228, "learning_rate": 7.062211981566821e-06, "loss": 0.391, "step": 614 }, { "epoch": 0.04251641894227445, "grad_norm": 1.1241086965010554, "learning_rate": 7.073732718894009e-06, "loss": 0.491, "step": 615 }, { "epoch": 0.04258555133079848, "grad_norm": 1.539253685790194, "learning_rate": 7.085253456221199e-06, "loss": 0.3001, "step": 616 }, { "epoch": 0.0426546837193225, "grad_norm": 1.1958426228580061, "learning_rate": 7.096774193548388e-06, "loss": 0.393, "step": 617 }, { "epoch": 0.04272381610784653, "grad_norm": 1.0581195725923673, "learning_rate": 7.1082949308755765e-06, "loss": 0.4711, "step": 618 }, { "epoch": 0.042792948496370546, "grad_norm": 1.01524734158247, "learning_rate": 7.119815668202765e-06, "loss": 0.4674, "step": 619 }, { "epoch": 0.04286208088489457, "grad_norm": 1.3561077633821768, "learning_rate": 7.131336405529955e-06, "loss": 0.3673, "step": 620 }, { "epoch": 0.042931213273418596, "grad_norm": 1.4016666561695124, "learning_rate": 7.1428571428571436e-06, "loss": 0.3639, "step": 621 }, { "epoch": 0.04300034566194262, "grad_norm": 1.0380358015988704, "learning_rate": 7.154377880184332e-06, "loss": 0.5778, "step": 622 }, { "epoch": 0.043069478050466646, "grad_norm": 0.9990125467617879, "learning_rate": 7.1658986175115205e-06, "loss": 0.4196, "step": 623 }, { "epoch": 0.043138610438990664, "grad_norm": 1.0744123410827062, "learning_rate": 7.177419354838711e-06, "loss": 0.3926, "step": 624 }, { "epoch": 0.04320774282751469, "grad_norm": 1.1075359019704498, "learning_rate": 7.188940092165899e-06, "loss": 0.3875, "step": 625 }, { "epoch": 0.043276875216038714, "grad_norm": 1.0099245708328632, "learning_rate": 7.2004608294930876e-06, "loss": 0.4743, "step": 626 }, { "epoch": 0.04334600760456274, "grad_norm": 1.3791066961698824, "learning_rate": 7.211981566820278e-06, "loss": 0.4241, "step": 627 }, { "epoch": 0.043415139993086764, "grad_norm": 1.345809281843909, "learning_rate": 7.223502304147466e-06, "loss": 0.3548, "step": 628 }, { "epoch": 0.04348427238161078, "grad_norm": 1.115222149330893, "learning_rate": 7.235023041474655e-06, "loss": 0.485, "step": 629 }, { "epoch": 0.04355340477013481, "grad_norm": 1.084469204228023, "learning_rate": 7.246543778801845e-06, "loss": 0.4497, "step": 630 }, { "epoch": 0.04362253715865883, "grad_norm": 1.1170589311837285, "learning_rate": 7.258064516129033e-06, "loss": 0.3319, "step": 631 }, { "epoch": 0.04369166954718286, "grad_norm": 5.272359729929205, "learning_rate": 7.269585253456222e-06, "loss": 0.4473, "step": 632 }, { "epoch": 0.04376080193570688, "grad_norm": 1.5668626049611858, "learning_rate": 7.28110599078341e-06, "loss": 0.4912, "step": 633 }, { "epoch": 0.0438299343242309, "grad_norm": 1.1199063721236484, "learning_rate": 7.2926267281106e-06, "loss": 0.4284, "step": 634 }, { "epoch": 0.043899066712754925, "grad_norm": 1.217509948124475, "learning_rate": 7.304147465437789e-06, "loss": 0.3318, "step": 635 }, { "epoch": 0.04396819910127895, "grad_norm": 1.128318469098737, "learning_rate": 7.315668202764977e-06, "loss": 0.4668, "step": 636 }, { "epoch": 0.044037331489802975, "grad_norm": 1.2936631231217197, "learning_rate": 7.327188940092167e-06, "loss": 0.4658, "step": 637 }, { "epoch": 0.04410646387832699, "grad_norm": 0.9861137762349897, "learning_rate": 7.338709677419356e-06, "loss": 0.3496, "step": 638 }, { "epoch": 0.04417559626685102, "grad_norm": 1.2283590805140967, "learning_rate": 7.350230414746544e-06, "loss": 0.3782, "step": 639 }, { "epoch": 0.04424472865537504, "grad_norm": 1.47379401041021, "learning_rate": 7.361751152073733e-06, "loss": 0.4752, "step": 640 }, { "epoch": 0.04431386104389907, "grad_norm": 1.1042211209017172, "learning_rate": 7.373271889400923e-06, "loss": 0.4104, "step": 641 }, { "epoch": 0.04438299343242309, "grad_norm": 1.1406520654712708, "learning_rate": 7.384792626728111e-06, "loss": 0.3645, "step": 642 }, { "epoch": 0.04445212582094711, "grad_norm": 1.2099107711395893, "learning_rate": 7.3963133640553e-06, "loss": 0.4567, "step": 643 }, { "epoch": 0.044521258209471136, "grad_norm": 1.0603471135402838, "learning_rate": 7.407834101382489e-06, "loss": 0.4671, "step": 644 }, { "epoch": 0.04459039059799516, "grad_norm": 2.2337843017101986, "learning_rate": 7.4193548387096784e-06, "loss": 0.3849, "step": 645 }, { "epoch": 0.044659522986519186, "grad_norm": 0.9570495878821388, "learning_rate": 7.430875576036867e-06, "loss": 0.3458, "step": 646 }, { "epoch": 0.04472865537504321, "grad_norm": 1.1200971962784079, "learning_rate": 7.442396313364056e-06, "loss": 0.4588, "step": 647 }, { "epoch": 0.04479778776356723, "grad_norm": 1.237688468630536, "learning_rate": 7.453917050691245e-06, "loss": 0.4202, "step": 648 }, { "epoch": 0.044866920152091254, "grad_norm": 1.3733654450158999, "learning_rate": 7.465437788018434e-06, "loss": 0.4221, "step": 649 }, { "epoch": 0.04493605254061528, "grad_norm": 1.1745587662147379, "learning_rate": 7.4769585253456224e-06, "loss": 0.461, "step": 650 }, { "epoch": 0.045005184929139304, "grad_norm": 1.0737292133446277, "learning_rate": 7.488479262672812e-06, "loss": 0.4961, "step": 651 }, { "epoch": 0.04507431731766332, "grad_norm": 0.9954494138105, "learning_rate": 7.500000000000001e-06, "loss": 0.3465, "step": 652 }, { "epoch": 0.04514344970618735, "grad_norm": 1.141775383553379, "learning_rate": 7.5115207373271895e-06, "loss": 0.3702, "step": 653 }, { "epoch": 0.04521258209471137, "grad_norm": 1.215133743864997, "learning_rate": 7.523041474654379e-06, "loss": 0.49, "step": 654 }, { "epoch": 0.0452817144832354, "grad_norm": 1.5667775725139697, "learning_rate": 7.534562211981567e-06, "loss": 0.4027, "step": 655 }, { "epoch": 0.04535084687175942, "grad_norm": 1.5318776104887526, "learning_rate": 7.5460829493087566e-06, "loss": 0.3518, "step": 656 }, { "epoch": 0.04541997926028344, "grad_norm": 1.427195927135899, "learning_rate": 7.557603686635945e-06, "loss": 0.4624, "step": 657 }, { "epoch": 0.045489111648807465, "grad_norm": 1.128312726928394, "learning_rate": 7.569124423963134e-06, "loss": 0.5232, "step": 658 }, { "epoch": 0.04555824403733149, "grad_norm": 1.3034361315682008, "learning_rate": 7.580645161290323e-06, "loss": 0.4624, "step": 659 }, { "epoch": 0.045627376425855515, "grad_norm": 2.092299582594318, "learning_rate": 7.592165898617512e-06, "loss": 0.3543, "step": 660 }, { "epoch": 0.04569650881437954, "grad_norm": 0.9926290468898911, "learning_rate": 7.603686635944701e-06, "loss": 0.4654, "step": 661 }, { "epoch": 0.04576564120290356, "grad_norm": 1.0184460677667586, "learning_rate": 7.61520737327189e-06, "loss": 0.4308, "step": 662 }, { "epoch": 0.04583477359142758, "grad_norm": 1.1691103011599226, "learning_rate": 7.626728110599078e-06, "loss": 0.3821, "step": 663 }, { "epoch": 0.04590390597995161, "grad_norm": 1.098833885375508, "learning_rate": 7.638248847926268e-06, "loss": 0.396, "step": 664 }, { "epoch": 0.04597303836847563, "grad_norm": 1.5502335386579837, "learning_rate": 7.649769585253457e-06, "loss": 0.5491, "step": 665 }, { "epoch": 0.04604217075699966, "grad_norm": 0.9699526299905836, "learning_rate": 7.661290322580646e-06, "loss": 0.4049, "step": 666 }, { "epoch": 0.046111303145523676, "grad_norm": 0.9677868666009892, "learning_rate": 7.672811059907834e-06, "loss": 0.3252, "step": 667 }, { "epoch": 0.0461804355340477, "grad_norm": 1.0394026177578688, "learning_rate": 7.684331797235023e-06, "loss": 0.4265, "step": 668 }, { "epoch": 0.046249567922571726, "grad_norm": 1.059184492959927, "learning_rate": 7.695852534562212e-06, "loss": 0.5104, "step": 669 }, { "epoch": 0.04631870031109575, "grad_norm": 1.3215587738356942, "learning_rate": 7.707373271889402e-06, "loss": 0.3901, "step": 670 }, { "epoch": 0.04638783269961977, "grad_norm": 1.0686215012600309, "learning_rate": 7.718894009216591e-06, "loss": 0.3715, "step": 671 }, { "epoch": 0.046456965088143794, "grad_norm": 1.0608522369035067, "learning_rate": 7.73041474654378e-06, "loss": 0.5027, "step": 672 }, { "epoch": 0.04652609747666782, "grad_norm": 1.172337214519508, "learning_rate": 7.741935483870968e-06, "loss": 0.44, "step": 673 }, { "epoch": 0.046595229865191844, "grad_norm": 1.0224839132307475, "learning_rate": 7.753456221198157e-06, "loss": 0.4539, "step": 674 }, { "epoch": 0.04666436225371587, "grad_norm": 1.3130129161653858, "learning_rate": 7.764976958525347e-06, "loss": 0.4644, "step": 675 }, { "epoch": 0.04673349464223989, "grad_norm": 1.1516714744449719, "learning_rate": 7.776497695852536e-06, "loss": 0.4958, "step": 676 }, { "epoch": 0.04680262703076391, "grad_norm": 1.1812693906716296, "learning_rate": 7.788018433179724e-06, "loss": 0.3386, "step": 677 }, { "epoch": 0.04687175941928794, "grad_norm": 1.050463087042283, "learning_rate": 7.799539170506913e-06, "loss": 0.4124, "step": 678 }, { "epoch": 0.04694089180781196, "grad_norm": 1.2594932689586738, "learning_rate": 7.811059907834102e-06, "loss": 0.4677, "step": 679 }, { "epoch": 0.04701002419633599, "grad_norm": 1.2447971386269856, "learning_rate": 7.822580645161291e-06, "loss": 0.4575, "step": 680 }, { "epoch": 0.047079156584860005, "grad_norm": 1.0001243176168217, "learning_rate": 7.83410138248848e-06, "loss": 0.3078, "step": 681 }, { "epoch": 0.04714828897338403, "grad_norm": 1.0012186154217517, "learning_rate": 7.845622119815668e-06, "loss": 0.3921, "step": 682 }, { "epoch": 0.047217421361908055, "grad_norm": 1.2415916077695326, "learning_rate": 7.857142857142858e-06, "loss": 0.4998, "step": 683 }, { "epoch": 0.04728655375043208, "grad_norm": 1.1369914709588687, "learning_rate": 7.868663594470047e-06, "loss": 0.4198, "step": 684 }, { "epoch": 0.0473556861389561, "grad_norm": 1.1070580078636139, "learning_rate": 7.880184331797236e-06, "loss": 0.402, "step": 685 }, { "epoch": 0.04742481852748012, "grad_norm": 1.063022309697442, "learning_rate": 7.891705069124424e-06, "loss": 0.4346, "step": 686 }, { "epoch": 0.04749395091600415, "grad_norm": 1.156304520038796, "learning_rate": 7.903225806451613e-06, "loss": 0.4742, "step": 687 }, { "epoch": 0.04756308330452817, "grad_norm": 1.1687617633334555, "learning_rate": 7.914746543778803e-06, "loss": 0.3604, "step": 688 }, { "epoch": 0.0476322156930522, "grad_norm": 1.3742055444060404, "learning_rate": 7.926267281105992e-06, "loss": 0.4206, "step": 689 }, { "epoch": 0.047701348081576216, "grad_norm": 1.0626042075087043, "learning_rate": 7.93778801843318e-06, "loss": 0.4819, "step": 690 }, { "epoch": 0.04777048047010024, "grad_norm": 1.1641133080571267, "learning_rate": 7.94930875576037e-06, "loss": 0.3746, "step": 691 }, { "epoch": 0.047839612858624266, "grad_norm": 0.9249244694946985, "learning_rate": 7.960829493087558e-06, "loss": 0.3998, "step": 692 }, { "epoch": 0.04790874524714829, "grad_norm": 1.0989495719943811, "learning_rate": 7.972350230414747e-06, "loss": 0.3951, "step": 693 }, { "epoch": 0.047977877635672316, "grad_norm": 1.1210289309901038, "learning_rate": 7.983870967741935e-06, "loss": 0.4654, "step": 694 }, { "epoch": 0.048047010024196334, "grad_norm": 1.1457268015743174, "learning_rate": 7.995391705069126e-06, "loss": 0.3444, "step": 695 }, { "epoch": 0.04811614241272036, "grad_norm": 1.3454934846020947, "learning_rate": 8.006912442396314e-06, "loss": 0.3618, "step": 696 }, { "epoch": 0.048185274801244384, "grad_norm": 1.1072739370445268, "learning_rate": 8.018433179723503e-06, "loss": 0.496, "step": 697 }, { "epoch": 0.04825440718976841, "grad_norm": 1.2525417786763942, "learning_rate": 8.029953917050692e-06, "loss": 0.465, "step": 698 }, { "epoch": 0.04832353957829243, "grad_norm": 1.0396533378088566, "learning_rate": 8.041474654377881e-06, "loss": 0.3735, "step": 699 }, { "epoch": 0.04839267196681645, "grad_norm": 1.0560170227564993, "learning_rate": 8.052995391705069e-06, "loss": 0.4314, "step": 700 }, { "epoch": 0.048461804355340476, "grad_norm": 1.0863262200568873, "learning_rate": 8.064516129032258e-06, "loss": 0.4883, "step": 701 }, { "epoch": 0.0485309367438645, "grad_norm": 1.176950654610821, "learning_rate": 8.076036866359448e-06, "loss": 0.3914, "step": 702 }, { "epoch": 0.048600069132388526, "grad_norm": 1.0506689376436567, "learning_rate": 8.087557603686637e-06, "loss": 0.3887, "step": 703 }, { "epoch": 0.048669201520912544, "grad_norm": 1.0711873056476209, "learning_rate": 8.099078341013825e-06, "loss": 0.5363, "step": 704 }, { "epoch": 0.04873833390943657, "grad_norm": 1.813983930322138, "learning_rate": 8.110599078341016e-06, "loss": 0.4765, "step": 705 }, { "epoch": 0.048807466297960594, "grad_norm": 1.099136902158937, "learning_rate": 8.122119815668203e-06, "loss": 0.3904, "step": 706 }, { "epoch": 0.04887659868648462, "grad_norm": 1.1001332405274657, "learning_rate": 8.133640552995393e-06, "loss": 0.4428, "step": 707 }, { "epoch": 0.048945731075008644, "grad_norm": 1.0504037154377066, "learning_rate": 8.145161290322582e-06, "loss": 0.4992, "step": 708 }, { "epoch": 0.04901486346353266, "grad_norm": 1.0255441069729256, "learning_rate": 8.156682027649771e-06, "loss": 0.4016, "step": 709 }, { "epoch": 0.04908399585205669, "grad_norm": 0.9998772719235751, "learning_rate": 8.168202764976959e-06, "loss": 0.3429, "step": 710 }, { "epoch": 0.04915312824058071, "grad_norm": 1.0556159518233772, "learning_rate": 8.179723502304148e-06, "loss": 0.4751, "step": 711 }, { "epoch": 0.04922226062910474, "grad_norm": 1.1352318444614777, "learning_rate": 8.191244239631337e-06, "loss": 0.4598, "step": 712 }, { "epoch": 0.04929139301762876, "grad_norm": 1.0009193559373601, "learning_rate": 8.202764976958527e-06, "loss": 0.3995, "step": 713 }, { "epoch": 0.04936052540615278, "grad_norm": 1.0641519915423265, "learning_rate": 8.214285714285714e-06, "loss": 0.3819, "step": 714 }, { "epoch": 0.049429657794676805, "grad_norm": 0.9942270119926321, "learning_rate": 8.225806451612904e-06, "loss": 0.5139, "step": 715 }, { "epoch": 0.04949879018320083, "grad_norm": 1.1172082870448852, "learning_rate": 8.237327188940093e-06, "loss": 0.4185, "step": 716 }, { "epoch": 0.049567922571724855, "grad_norm": 0.9266306304687996, "learning_rate": 8.248847926267282e-06, "loss": 0.3575, "step": 717 }, { "epoch": 0.04963705496024887, "grad_norm": 1.010296374630366, "learning_rate": 8.26036866359447e-06, "loss": 0.4182, "step": 718 }, { "epoch": 0.0497061873487729, "grad_norm": 1.0510862116544748, "learning_rate": 8.271889400921659e-06, "loss": 0.4273, "step": 719 }, { "epoch": 0.04977531973729692, "grad_norm": 1.1581708031940374, "learning_rate": 8.283410138248848e-06, "loss": 0.441, "step": 720 }, { "epoch": 0.04984445212582095, "grad_norm": 1.266240618866391, "learning_rate": 8.294930875576038e-06, "loss": 0.3635, "step": 721 }, { "epoch": 0.04991358451434497, "grad_norm": 1.1712293935435567, "learning_rate": 8.306451612903227e-06, "loss": 0.5094, "step": 722 }, { "epoch": 0.04998271690286899, "grad_norm": 1.1805557647521532, "learning_rate": 8.317972350230415e-06, "loss": 0.5385, "step": 723 }, { "epoch": 0.050051849291393016, "grad_norm": 1.1047565646558473, "learning_rate": 8.329493087557604e-06, "loss": 0.3323, "step": 724 }, { "epoch": 0.05012098167991704, "grad_norm": 1.0220156130800133, "learning_rate": 8.341013824884793e-06, "loss": 0.4478, "step": 725 }, { "epoch": 0.050190114068441066, "grad_norm": 1.0876959030054296, "learning_rate": 8.352534562211983e-06, "loss": 0.4968, "step": 726 }, { "epoch": 0.05025924645696509, "grad_norm": 1.044845579428868, "learning_rate": 8.364055299539172e-06, "loss": 0.415, "step": 727 }, { "epoch": 0.05032837884548911, "grad_norm": 0.9134851458708203, "learning_rate": 8.37557603686636e-06, "loss": 0.3622, "step": 728 }, { "epoch": 0.050397511234013134, "grad_norm": 1.1339651385187166, "learning_rate": 8.387096774193549e-06, "loss": 0.4626, "step": 729 }, { "epoch": 0.05046664362253716, "grad_norm": 1.0446688869160417, "learning_rate": 8.398617511520738e-06, "loss": 0.4059, "step": 730 }, { "epoch": 0.050535776011061184, "grad_norm": 1.0173065151734793, "learning_rate": 8.410138248847927e-06, "loss": 0.391, "step": 731 }, { "epoch": 0.0506049083995852, "grad_norm": 1.6741590103917083, "learning_rate": 8.421658986175117e-06, "loss": 0.4748, "step": 732 }, { "epoch": 0.05067404078810923, "grad_norm": 1.0283147858249, "learning_rate": 8.433179723502304e-06, "loss": 0.4797, "step": 733 }, { "epoch": 0.05074317317663325, "grad_norm": 1.0566025917546462, "learning_rate": 8.444700460829494e-06, "loss": 0.3812, "step": 734 }, { "epoch": 0.05081230556515728, "grad_norm": 0.9102168340505532, "learning_rate": 8.456221198156683e-06, "loss": 0.3385, "step": 735 }, { "epoch": 0.0508814379536813, "grad_norm": 1.0670399416588319, "learning_rate": 8.467741935483872e-06, "loss": 0.5115, "step": 736 }, { "epoch": 0.05095057034220532, "grad_norm": 1.2171209658300046, "learning_rate": 8.47926267281106e-06, "loss": 0.4814, "step": 737 }, { "epoch": 0.051019702730729345, "grad_norm": 1.0020136269414603, "learning_rate": 8.49078341013825e-06, "loss": 0.3763, "step": 738 }, { "epoch": 0.05108883511925337, "grad_norm": 1.032419706598949, "learning_rate": 8.502304147465438e-06, "loss": 0.3895, "step": 739 }, { "epoch": 0.051157967507777395, "grad_norm": 1.0586241430338925, "learning_rate": 8.513824884792628e-06, "loss": 0.5081, "step": 740 }, { "epoch": 0.05122709989630142, "grad_norm": 1.0881854237626984, "learning_rate": 8.525345622119815e-06, "loss": 0.4503, "step": 741 }, { "epoch": 0.05129623228482544, "grad_norm": 1.1610993534587466, "learning_rate": 8.536866359447006e-06, "loss": 0.3453, "step": 742 }, { "epoch": 0.05136536467334946, "grad_norm": 0.9768479460451467, "learning_rate": 8.548387096774194e-06, "loss": 0.4305, "step": 743 }, { "epoch": 0.05143449706187349, "grad_norm": 1.1592569316197443, "learning_rate": 8.559907834101383e-06, "loss": 0.5573, "step": 744 }, { "epoch": 0.05150362945039751, "grad_norm": 1.0822725086135858, "learning_rate": 8.571428571428571e-06, "loss": 0.3772, "step": 745 }, { "epoch": 0.05157276183892154, "grad_norm": 0.9777960848121076, "learning_rate": 8.582949308755762e-06, "loss": 0.3518, "step": 746 }, { "epoch": 0.051641894227445556, "grad_norm": 1.1350552744302642, "learning_rate": 8.59447004608295e-06, "loss": 0.5817, "step": 747 }, { "epoch": 0.05171102661596958, "grad_norm": 1.0657637541138458, "learning_rate": 8.605990783410139e-06, "loss": 0.4535, "step": 748 }, { "epoch": 0.051780159004493606, "grad_norm": 1.1240039078590207, "learning_rate": 8.617511520737328e-06, "loss": 0.4183, "step": 749 }, { "epoch": 0.05184929139301763, "grad_norm": 1.1293749379423017, "learning_rate": 8.629032258064517e-06, "loss": 0.4015, "step": 750 }, { "epoch": 0.05191842378154165, "grad_norm": 1.0522655442342876, "learning_rate": 8.640552995391705e-06, "loss": 0.493, "step": 751 }, { "epoch": 0.051987556170065674, "grad_norm": 0.9632579525894813, "learning_rate": 8.652073732718894e-06, "loss": 0.3445, "step": 752 }, { "epoch": 0.0520566885585897, "grad_norm": 1.142461700310248, "learning_rate": 8.663594470046084e-06, "loss": 0.4117, "step": 753 }, { "epoch": 0.052125820947113724, "grad_norm": 1.048468996461167, "learning_rate": 8.675115207373273e-06, "loss": 0.4696, "step": 754 }, { "epoch": 0.05219495333563775, "grad_norm": 1.1867717876973551, "learning_rate": 8.68663594470046e-06, "loss": 0.3788, "step": 755 }, { "epoch": 0.05226408572416177, "grad_norm": 0.9195928568628164, "learning_rate": 8.698156682027652e-06, "loss": 0.3068, "step": 756 }, { "epoch": 0.05233321811268579, "grad_norm": 1.029255451403183, "learning_rate": 8.70967741935484e-06, "loss": 0.3991, "step": 757 }, { "epoch": 0.05240235050120982, "grad_norm": 1.1488095708463129, "learning_rate": 8.721198156682029e-06, "loss": 0.462, "step": 758 }, { "epoch": 0.05247148288973384, "grad_norm": 1.2208385442649217, "learning_rate": 8.732718894009218e-06, "loss": 0.3962, "step": 759 }, { "epoch": 0.05254061527825787, "grad_norm": 0.959443861625557, "learning_rate": 8.744239631336407e-06, "loss": 0.3568, "step": 760 }, { "epoch": 0.052609747666781885, "grad_norm": 1.0592888426907288, "learning_rate": 8.755760368663595e-06, "loss": 0.4986, "step": 761 }, { "epoch": 0.05267888005530591, "grad_norm": 1.0454210210936656, "learning_rate": 8.767281105990784e-06, "loss": 0.4668, "step": 762 }, { "epoch": 0.052748012443829935, "grad_norm": 1.0287383733934181, "learning_rate": 8.778801843317973e-06, "loss": 0.3797, "step": 763 }, { "epoch": 0.05281714483235396, "grad_norm": 0.9899660871026007, "learning_rate": 8.790322580645163e-06, "loss": 0.3803, "step": 764 }, { "epoch": 0.05288627722087798, "grad_norm": 2.2835539292670264, "learning_rate": 8.80184331797235e-06, "loss": 0.4718, "step": 765 }, { "epoch": 0.052955409609402, "grad_norm": 1.2499882649338383, "learning_rate": 8.81336405529954e-06, "loss": 0.3846, "step": 766 }, { "epoch": 0.05302454199792603, "grad_norm": 0.9264713890209841, "learning_rate": 8.824884792626729e-06, "loss": 0.3268, "step": 767 }, { "epoch": 0.05309367438645005, "grad_norm": 0.9806643025552246, "learning_rate": 8.836405529953918e-06, "loss": 0.4638, "step": 768 }, { "epoch": 0.05316280677497408, "grad_norm": 1.0902730255592363, "learning_rate": 8.847926267281107e-06, "loss": 0.523, "step": 769 }, { "epoch": 0.053231939163498096, "grad_norm": 1.1558865049665128, "learning_rate": 8.859447004608295e-06, "loss": 0.4042, "step": 770 }, { "epoch": 0.05330107155202212, "grad_norm": 1.6080339250912745, "learning_rate": 8.870967741935484e-06, "loss": 0.3883, "step": 771 }, { "epoch": 0.053370203940546146, "grad_norm": 1.0610067704055577, "learning_rate": 8.882488479262674e-06, "loss": 0.4527, "step": 772 }, { "epoch": 0.05343933632907017, "grad_norm": 1.1260430215621413, "learning_rate": 8.894009216589863e-06, "loss": 0.4248, "step": 773 }, { "epoch": 0.053508468717594196, "grad_norm": 0.986358794999526, "learning_rate": 8.90552995391705e-06, "loss": 0.3188, "step": 774 }, { "epoch": 0.053577601106118214, "grad_norm": 1.2885304091143452, "learning_rate": 8.91705069124424e-06, "loss": 0.4854, "step": 775 }, { "epoch": 0.05364673349464224, "grad_norm": 1.05861693824235, "learning_rate": 8.92857142857143e-06, "loss": 0.556, "step": 776 }, { "epoch": 0.053715865883166264, "grad_norm": 1.3825467585060875, "learning_rate": 8.940092165898619e-06, "loss": 0.426, "step": 777 }, { "epoch": 0.05378499827169029, "grad_norm": 1.0845002668943409, "learning_rate": 8.951612903225806e-06, "loss": 0.3287, "step": 778 }, { "epoch": 0.053854130660214314, "grad_norm": 1.1122637738438816, "learning_rate": 8.963133640552995e-06, "loss": 0.4709, "step": 779 }, { "epoch": 0.05392326304873833, "grad_norm": 1.0368281587308568, "learning_rate": 8.974654377880185e-06, "loss": 0.4759, "step": 780 }, { "epoch": 0.05399239543726236, "grad_norm": 1.1031824484042998, "learning_rate": 8.986175115207374e-06, "loss": 0.424, "step": 781 }, { "epoch": 0.05406152782578638, "grad_norm": 1.4067269163659775, "learning_rate": 8.997695852534563e-06, "loss": 0.4209, "step": 782 }, { "epoch": 0.05413066021431041, "grad_norm": 1.0334058055395483, "learning_rate": 9.009216589861753e-06, "loss": 0.4536, "step": 783 }, { "epoch": 0.054199792602834425, "grad_norm": 0.985293509829988, "learning_rate": 9.02073732718894e-06, "loss": 0.3512, "step": 784 }, { "epoch": 0.05426892499135845, "grad_norm": 1.129567165995278, "learning_rate": 9.03225806451613e-06, "loss": 0.3669, "step": 785 }, { "epoch": 0.054338057379882475, "grad_norm": 1.1562204776763614, "learning_rate": 9.043778801843319e-06, "loss": 0.3964, "step": 786 }, { "epoch": 0.0544071897684065, "grad_norm": 1.1190602413153465, "learning_rate": 9.055299539170508e-06, "loss": 0.4322, "step": 787 }, { "epoch": 0.054476322156930525, "grad_norm": 0.8981915763654561, "learning_rate": 9.066820276497696e-06, "loss": 0.3552, "step": 788 }, { "epoch": 0.05454545454545454, "grad_norm": 1.022867798770987, "learning_rate": 9.078341013824885e-06, "loss": 0.3456, "step": 789 }, { "epoch": 0.05461458693397857, "grad_norm": 1.031613396832528, "learning_rate": 9.089861751152074e-06, "loss": 0.494, "step": 790 }, { "epoch": 0.05468371932250259, "grad_norm": 1.1299313008658611, "learning_rate": 9.101382488479264e-06, "loss": 0.3643, "step": 791 }, { "epoch": 0.05475285171102662, "grad_norm": 1.0449835492039776, "learning_rate": 9.112903225806451e-06, "loss": 0.3945, "step": 792 }, { "epoch": 0.05482198409955064, "grad_norm": 0.9880189796302881, "learning_rate": 9.124423963133642e-06, "loss": 0.4273, "step": 793 }, { "epoch": 0.05489111648807466, "grad_norm": 1.0482834522807267, "learning_rate": 9.13594470046083e-06, "loss": 0.4627, "step": 794 }, { "epoch": 0.054960248876598686, "grad_norm": 1.1667375961682223, "learning_rate": 9.14746543778802e-06, "loss": 0.3236, "step": 795 }, { "epoch": 0.05502938126512271, "grad_norm": 0.9584632961761788, "learning_rate": 9.158986175115207e-06, "loss": 0.3699, "step": 796 }, { "epoch": 0.055098513653646736, "grad_norm": 1.019212764312032, "learning_rate": 9.170506912442398e-06, "loss": 0.4508, "step": 797 }, { "epoch": 0.055167646042170754, "grad_norm": 0.9727426013266098, "learning_rate": 9.182027649769586e-06, "loss": 0.4638, "step": 798 }, { "epoch": 0.05523677843069478, "grad_norm": 1.080945549110446, "learning_rate": 9.193548387096775e-06, "loss": 0.4423, "step": 799 }, { "epoch": 0.055305910819218804, "grad_norm": 1.3772530930273166, "learning_rate": 9.205069124423964e-06, "loss": 0.4401, "step": 800 }, { "epoch": 0.05537504320774283, "grad_norm": 1.193546594746614, "learning_rate": 9.216589861751153e-06, "loss": 0.5327, "step": 801 }, { "epoch": 0.055444175596266854, "grad_norm": 1.0275314841401928, "learning_rate": 9.228110599078341e-06, "loss": 0.3474, "step": 802 }, { "epoch": 0.05551330798479087, "grad_norm": 1.1864457575954914, "learning_rate": 9.239631336405532e-06, "loss": 0.3792, "step": 803 }, { "epoch": 0.0555824403733149, "grad_norm": 1.0366935765019294, "learning_rate": 9.25115207373272e-06, "loss": 0.4543, "step": 804 }, { "epoch": 0.05565157276183892, "grad_norm": 1.0944063339519743, "learning_rate": 9.262672811059909e-06, "loss": 0.4169, "step": 805 }, { "epoch": 0.05572070515036295, "grad_norm": 1.0505829665308106, "learning_rate": 9.274193548387097e-06, "loss": 0.3503, "step": 806 }, { "epoch": 0.05578983753888697, "grad_norm": 1.0333386139915561, "learning_rate": 9.285714285714288e-06, "loss": 0.4303, "step": 807 }, { "epoch": 0.05585896992741099, "grad_norm": 1.114269346308682, "learning_rate": 9.297235023041475e-06, "loss": 0.5058, "step": 808 }, { "epoch": 0.055928102315935015, "grad_norm": 1.3051747807318452, "learning_rate": 9.308755760368664e-06, "loss": 0.373, "step": 809 }, { "epoch": 0.05599723470445904, "grad_norm": 1.2111394907478181, "learning_rate": 9.320276497695854e-06, "loss": 0.3525, "step": 810 }, { "epoch": 0.056066367092983065, "grad_norm": 1.3355365350203288, "learning_rate": 9.331797235023043e-06, "loss": 0.4275, "step": 811 }, { "epoch": 0.05613549948150708, "grad_norm": 1.041745274373973, "learning_rate": 9.34331797235023e-06, "loss": 0.4911, "step": 812 }, { "epoch": 0.05620463187003111, "grad_norm": 1.14895385999336, "learning_rate": 9.35483870967742e-06, "loss": 0.3814, "step": 813 }, { "epoch": 0.05627376425855513, "grad_norm": 1.32889302038083, "learning_rate": 9.36635944700461e-06, "loss": 0.3954, "step": 814 }, { "epoch": 0.05634289664707916, "grad_norm": 1.0533047236289645, "learning_rate": 9.377880184331799e-06, "loss": 0.5084, "step": 815 }, { "epoch": 0.05641202903560318, "grad_norm": 1.0672803015302395, "learning_rate": 9.389400921658986e-06, "loss": 0.3464, "step": 816 }, { "epoch": 0.0564811614241272, "grad_norm": 1.1591786442409733, "learning_rate": 9.400921658986176e-06, "loss": 0.3356, "step": 817 }, { "epoch": 0.056550293812651226, "grad_norm": 1.1410915931350272, "learning_rate": 9.412442396313365e-06, "loss": 0.4238, "step": 818 }, { "epoch": 0.05661942620117525, "grad_norm": 1.241623815838935, "learning_rate": 9.423963133640554e-06, "loss": 0.5169, "step": 819 }, { "epoch": 0.056688558589699276, "grad_norm": 1.0466950550350904, "learning_rate": 9.435483870967743e-06, "loss": 0.3664, "step": 820 }, { "epoch": 0.0567576909782233, "grad_norm": 1.159819725198304, "learning_rate": 9.447004608294931e-06, "loss": 0.3653, "step": 821 }, { "epoch": 0.05682682336674732, "grad_norm": 1.1156039654749146, "learning_rate": 9.45852534562212e-06, "loss": 0.4651, "step": 822 }, { "epoch": 0.056895955755271344, "grad_norm": 1.1350880064622033, "learning_rate": 9.47004608294931e-06, "loss": 0.4656, "step": 823 }, { "epoch": 0.05696508814379537, "grad_norm": 1.1923655959837505, "learning_rate": 9.481566820276499e-06, "loss": 0.3876, "step": 824 }, { "epoch": 0.057034220532319393, "grad_norm": 0.9748453100737645, "learning_rate": 9.493087557603687e-06, "loss": 0.3769, "step": 825 }, { "epoch": 0.05710335292084342, "grad_norm": 0.9363117352097593, "learning_rate": 9.504608294930876e-06, "loss": 0.4648, "step": 826 }, { "epoch": 0.057172485309367437, "grad_norm": 1.1644172329632956, "learning_rate": 9.516129032258065e-06, "loss": 0.4345, "step": 827 }, { "epoch": 0.05724161769789146, "grad_norm": 1.1473222401358982, "learning_rate": 9.527649769585255e-06, "loss": 0.3503, "step": 828 }, { "epoch": 0.057310750086415486, "grad_norm": 1.1356274390321865, "learning_rate": 9.539170506912442e-06, "loss": 0.4922, "step": 829 }, { "epoch": 0.05737988247493951, "grad_norm": 1.142473357355253, "learning_rate": 9.550691244239631e-06, "loss": 0.4226, "step": 830 }, { "epoch": 0.05744901486346353, "grad_norm": 1.0088713779303593, "learning_rate": 9.56221198156682e-06, "loss": 0.4264, "step": 831 }, { "epoch": 0.057518147251987554, "grad_norm": 1.1667157621953734, "learning_rate": 9.57373271889401e-06, "loss": 0.4022, "step": 832 }, { "epoch": 0.05758727964051158, "grad_norm": 1.3296648996572136, "learning_rate": 9.5852534562212e-06, "loss": 0.483, "step": 833 }, { "epoch": 0.057656412029035604, "grad_norm": 0.9944532833210636, "learning_rate": 9.596774193548389e-06, "loss": 0.3471, "step": 834 }, { "epoch": 0.05772554441755963, "grad_norm": 1.045228173572255, "learning_rate": 9.608294930875576e-06, "loss": 0.3741, "step": 835 }, { "epoch": 0.05779467680608365, "grad_norm": 1.4958698789840879, "learning_rate": 9.619815668202766e-06, "loss": 0.479, "step": 836 }, { "epoch": 0.05786380919460767, "grad_norm": 1.1601590807432947, "learning_rate": 9.631336405529955e-06, "loss": 0.4301, "step": 837 }, { "epoch": 0.0579329415831317, "grad_norm": 0.9392407054854589, "learning_rate": 9.642857142857144e-06, "loss": 0.3508, "step": 838 }, { "epoch": 0.05800207397165572, "grad_norm": 1.0887367187832697, "learning_rate": 9.654377880184332e-06, "loss": 0.4062, "step": 839 }, { "epoch": 0.05807120636017975, "grad_norm": 1.1448930800479227, "learning_rate": 9.665898617511521e-06, "loss": 0.4723, "step": 840 }, { "epoch": 0.058140338748703765, "grad_norm": 1.0260472367091817, "learning_rate": 9.67741935483871e-06, "loss": 0.3799, "step": 841 }, { "epoch": 0.05820947113722779, "grad_norm": 0.994630607043847, "learning_rate": 9.6889400921659e-06, "loss": 0.3399, "step": 842 }, { "epoch": 0.058278603525751815, "grad_norm": 1.019753000121382, "learning_rate": 9.700460829493087e-06, "loss": 0.4454, "step": 843 }, { "epoch": 0.05834773591427584, "grad_norm": 1.1190110510377989, "learning_rate": 9.711981566820278e-06, "loss": 0.5462, "step": 844 }, { "epoch": 0.05841686830279986, "grad_norm": 1.220725897868157, "learning_rate": 9.723502304147466e-06, "loss": 0.375, "step": 845 }, { "epoch": 0.05848600069132388, "grad_norm": 1.428158198382116, "learning_rate": 9.735023041474655e-06, "loss": 0.34, "step": 846 }, { "epoch": 0.05855513307984791, "grad_norm": 1.047690260499134, "learning_rate": 9.746543778801845e-06, "loss": 0.5319, "step": 847 }, { "epoch": 0.05862426546837193, "grad_norm": 1.098615724413494, "learning_rate": 9.758064516129034e-06, "loss": 0.4157, "step": 848 }, { "epoch": 0.05869339785689596, "grad_norm": 1.046736194869589, "learning_rate": 9.769585253456221e-06, "loss": 0.4132, "step": 849 }, { "epoch": 0.058762530245419976, "grad_norm": 1.1025761865178012, "learning_rate": 9.78110599078341e-06, "loss": 0.4014, "step": 850 }, { "epoch": 0.058831662633944, "grad_norm": 1.0991413912479169, "learning_rate": 9.7926267281106e-06, "loss": 0.5183, "step": 851 }, { "epoch": 0.058900795022468026, "grad_norm": 1.6084161681277085, "learning_rate": 9.80414746543779e-06, "loss": 0.3622, "step": 852 }, { "epoch": 0.05896992741099205, "grad_norm": 1.1687205410552002, "learning_rate": 9.815668202764977e-06, "loss": 0.3934, "step": 853 }, { "epoch": 0.059039059799516076, "grad_norm": 1.1552265956382923, "learning_rate": 9.827188940092168e-06, "loss": 0.4732, "step": 854 }, { "epoch": 0.059108192188040094, "grad_norm": 1.158013177977087, "learning_rate": 9.838709677419356e-06, "loss": 0.431, "step": 855 }, { "epoch": 0.05917732457656412, "grad_norm": 1.041159545903445, "learning_rate": 9.850230414746545e-06, "loss": 0.3635, "step": 856 }, { "epoch": 0.059246456965088144, "grad_norm": 1.058343242790134, "learning_rate": 9.861751152073733e-06, "loss": 0.4086, "step": 857 }, { "epoch": 0.05931558935361217, "grad_norm": 1.2153021544947018, "learning_rate": 9.873271889400924e-06, "loss": 0.5211, "step": 858 }, { "epoch": 0.059384721742136194, "grad_norm": 1.0873507540488552, "learning_rate": 9.884792626728111e-06, "loss": 0.4269, "step": 859 }, { "epoch": 0.05945385413066021, "grad_norm": 1.1243545877820282, "learning_rate": 9.8963133640553e-06, "loss": 0.3423, "step": 860 }, { "epoch": 0.05952298651918424, "grad_norm": 1.1415313571130385, "learning_rate": 9.90783410138249e-06, "loss": 0.4339, "step": 861 }, { "epoch": 0.05959211890770826, "grad_norm": 1.040888248182213, "learning_rate": 9.919354838709679e-06, "loss": 0.4301, "step": 862 }, { "epoch": 0.05966125129623229, "grad_norm": 0.8986172894100071, "learning_rate": 9.930875576036867e-06, "loss": 0.329, "step": 863 }, { "epoch": 0.059730383684756305, "grad_norm": 1.7919256680989326, "learning_rate": 9.942396313364056e-06, "loss": 0.5058, "step": 864 }, { "epoch": 0.05979951607328033, "grad_norm": 1.0375416595934772, "learning_rate": 9.953917050691245e-06, "loss": 0.4862, "step": 865 }, { "epoch": 0.059868648461804355, "grad_norm": 1.0259232938775495, "learning_rate": 9.965437788018435e-06, "loss": 0.3928, "step": 866 }, { "epoch": 0.05993778085032838, "grad_norm": 1.0081441210415192, "learning_rate": 9.976958525345622e-06, "loss": 0.3108, "step": 867 }, { "epoch": 0.060006913238852405, "grad_norm": 1.234575326076125, "learning_rate": 9.988479262672812e-06, "loss": 0.4225, "step": 868 }, { "epoch": 0.06007604562737642, "grad_norm": 1.057956580527247, "learning_rate": 1e-05, "loss": 0.4856, "step": 869 }, { "epoch": 0.06014517801590045, "grad_norm": 1.1001183574103586, "learning_rate": 9.999999968666963e-06, "loss": 0.3478, "step": 870 }, { "epoch": 0.06021431040442447, "grad_norm": 1.038126269216243, "learning_rate": 9.999999874667848e-06, "loss": 0.3878, "step": 871 }, { "epoch": 0.0602834427929485, "grad_norm": 1.072227239112567, "learning_rate": 9.999999718002656e-06, "loss": 0.4773, "step": 872 }, { "epoch": 0.06035257518147252, "grad_norm": 0.9843027230929113, "learning_rate": 9.999999498671391e-06, "loss": 0.4131, "step": 873 }, { "epoch": 0.06042170756999654, "grad_norm": 0.9816982839079775, "learning_rate": 9.999999216674057e-06, "loss": 0.319, "step": 874 }, { "epoch": 0.060490839958520566, "grad_norm": 1.2183812757210424, "learning_rate": 9.999998872010655e-06, "loss": 0.3824, "step": 875 }, { "epoch": 0.06055997234704459, "grad_norm": 1.1464865039092598, "learning_rate": 9.99999846468119e-06, "loss": 0.459, "step": 876 }, { "epoch": 0.060629104735568616, "grad_norm": 1.043893387063084, "learning_rate": 9.999997994685667e-06, "loss": 0.3924, "step": 877 }, { "epoch": 0.060698237124092634, "grad_norm": 0.9978661175150462, "learning_rate": 9.99999746202409e-06, "loss": 0.3805, "step": 878 }, { "epoch": 0.06076736951261666, "grad_norm": 1.1145238995036744, "learning_rate": 9.99999686669647e-06, "loss": 0.5168, "step": 879 }, { "epoch": 0.060836501901140684, "grad_norm": 1.5021840144959393, "learning_rate": 9.999996208702811e-06, "loss": 0.4373, "step": 880 }, { "epoch": 0.06090563428966471, "grad_norm": 0.8589890310051079, "learning_rate": 9.999995488043124e-06, "loss": 0.3959, "step": 881 }, { "epoch": 0.060974766678188734, "grad_norm": 0.9893704336491554, "learning_rate": 9.999994704717415e-06, "loss": 0.4158, "step": 882 }, { "epoch": 0.06104389906671275, "grad_norm": 1.1622961610614027, "learning_rate": 9.999993858725697e-06, "loss": 0.4966, "step": 883 }, { "epoch": 0.06111303145523678, "grad_norm": 1.043537873869462, "learning_rate": 9.999992950067975e-06, "loss": 0.3911, "step": 884 }, { "epoch": 0.0611821638437608, "grad_norm": 1.1417063773466394, "learning_rate": 9.999991978744268e-06, "loss": 0.3529, "step": 885 }, { "epoch": 0.06125129623228483, "grad_norm": 0.9889592035753886, "learning_rate": 9.999990944754585e-06, "loss": 0.4454, "step": 886 }, { "epoch": 0.06132042862080885, "grad_norm": 1.2506142053674156, "learning_rate": 9.999989848098936e-06, "loss": 0.4011, "step": 887 }, { "epoch": 0.06138956100933287, "grad_norm": 0.9926604943434214, "learning_rate": 9.999988688777338e-06, "loss": 0.4008, "step": 888 }, { "epoch": 0.061458693397856895, "grad_norm": 1.0171567969920703, "learning_rate": 9.999987466789805e-06, "loss": 0.3366, "step": 889 }, { "epoch": 0.06152782578638092, "grad_norm": 1.062120515724462, "learning_rate": 9.99998618213635e-06, "loss": 0.4662, "step": 890 }, { "epoch": 0.061596958174904945, "grad_norm": 1.2157372829596458, "learning_rate": 9.999984834816994e-06, "loss": 0.4327, "step": 891 }, { "epoch": 0.06166609056342897, "grad_norm": 1.850573764855978, "learning_rate": 9.999983424831749e-06, "loss": 0.4298, "step": 892 }, { "epoch": 0.06173522295195299, "grad_norm": 1.1752232740250639, "learning_rate": 9.999981952180635e-06, "loss": 0.4903, "step": 893 }, { "epoch": 0.06180435534047701, "grad_norm": 1.1434006766495093, "learning_rate": 9.999980416863671e-06, "loss": 0.4485, "step": 894 }, { "epoch": 0.06187348772900104, "grad_norm": 1.139978593697853, "learning_rate": 9.999978818880873e-06, "loss": 0.4038, "step": 895 }, { "epoch": 0.06194262011752506, "grad_norm": 1.1244814656441227, "learning_rate": 9.999977158232266e-06, "loss": 0.358, "step": 896 }, { "epoch": 0.06201175250604908, "grad_norm": 1.1601342634887928, "learning_rate": 9.999975434917867e-06, "loss": 0.4836, "step": 897 }, { "epoch": 0.062080884894573106, "grad_norm": 1.2705663208163325, "learning_rate": 9.9999736489377e-06, "loss": 0.5259, "step": 898 }, { "epoch": 0.06215001728309713, "grad_norm": 1.0423943417131973, "learning_rate": 9.999971800291784e-06, "loss": 0.3517, "step": 899 }, { "epoch": 0.062219149671621156, "grad_norm": 1.088088653451799, "learning_rate": 9.999969888980146e-06, "loss": 0.4189, "step": 900 }, { "epoch": 0.06228828206014518, "grad_norm": 0.9890321102966461, "learning_rate": 9.999967915002807e-06, "loss": 0.4763, "step": 901 }, { "epoch": 0.0623574144486692, "grad_norm": 1.2237417152007597, "learning_rate": 9.999965878359793e-06, "loss": 0.4238, "step": 902 }, { "epoch": 0.062426546837193224, "grad_norm": 1.0230524999318094, "learning_rate": 9.999963779051131e-06, "loss": 0.3523, "step": 903 }, { "epoch": 0.06249567922571725, "grad_norm": 0.9724217069377136, "learning_rate": 9.999961617076845e-06, "loss": 0.4463, "step": 904 }, { "epoch": 0.06256481161424127, "grad_norm": 0.9920265885784284, "learning_rate": 9.999959392436965e-06, "loss": 0.4249, "step": 905 }, { "epoch": 0.0626339440027653, "grad_norm": 1.3798643187864115, "learning_rate": 9.999957105131515e-06, "loss": 0.3442, "step": 906 }, { "epoch": 0.06270307639128932, "grad_norm": 1.0041420170379272, "learning_rate": 9.999954755160525e-06, "loss": 0.3999, "step": 907 }, { "epoch": 0.06277220877981335, "grad_norm": 0.9625431424329801, "learning_rate": 9.999952342524025e-06, "loss": 0.4942, "step": 908 }, { "epoch": 0.06284134116833737, "grad_norm": 1.15171427900875, "learning_rate": 9.999949867222045e-06, "loss": 0.401, "step": 909 }, { "epoch": 0.06291047355686138, "grad_norm": 1.2582407593266254, "learning_rate": 9.999947329254618e-06, "loss": 0.4282, "step": 910 }, { "epoch": 0.06297960594538542, "grad_norm": 1.100472028457597, "learning_rate": 9.999944728621774e-06, "loss": 0.5078, "step": 911 }, { "epoch": 0.06304873833390943, "grad_norm": 1.0292795805143644, "learning_rate": 9.999942065323544e-06, "loss": 0.431, "step": 912 }, { "epoch": 0.06311787072243347, "grad_norm": 1.5639090463871124, "learning_rate": 9.999939339359963e-06, "loss": 0.4118, "step": 913 }, { "epoch": 0.06318700311095748, "grad_norm": 0.9150915049423055, "learning_rate": 9.999936550731067e-06, "loss": 0.3438, "step": 914 }, { "epoch": 0.0632561354994815, "grad_norm": 0.9520906593031205, "learning_rate": 9.999933699436886e-06, "loss": 0.4907, "step": 915 }, { "epoch": 0.06332526788800553, "grad_norm": 1.07650005181708, "learning_rate": 9.999930785477461e-06, "loss": 0.4032, "step": 916 }, { "epoch": 0.06339440027652955, "grad_norm": 0.9828049415844049, "learning_rate": 9.999927808852826e-06, "loss": 0.3991, "step": 917 }, { "epoch": 0.06346353266505357, "grad_norm": 1.143463420026287, "learning_rate": 9.999924769563019e-06, "loss": 0.416, "step": 918 }, { "epoch": 0.0635326650535776, "grad_norm": 1.0164354255739767, "learning_rate": 9.999921667608079e-06, "loss": 0.5079, "step": 919 }, { "epoch": 0.06360179744210162, "grad_norm": 0.9289551410756354, "learning_rate": 9.999918502988042e-06, "loss": 0.3988, "step": 920 }, { "epoch": 0.06367092983062565, "grad_norm": 1.0334316749666588, "learning_rate": 9.99991527570295e-06, "loss": 0.3268, "step": 921 }, { "epoch": 0.06374006221914967, "grad_norm": 1.0378492941061805, "learning_rate": 9.999911985752842e-06, "loss": 0.6167, "step": 922 }, { "epoch": 0.06380919460767369, "grad_norm": 1.1224238910715487, "learning_rate": 9.999908633137759e-06, "loss": 0.4137, "step": 923 }, { "epoch": 0.06387832699619772, "grad_norm": 1.1321629285147312, "learning_rate": 9.999905217857746e-06, "loss": 0.3743, "step": 924 }, { "epoch": 0.06394745938472174, "grad_norm": 1.0011246125179694, "learning_rate": 9.999901739912843e-06, "loss": 0.4249, "step": 925 }, { "epoch": 0.06401659177324577, "grad_norm": 0.9334469039360054, "learning_rate": 9.999898199303095e-06, "loss": 0.555, "step": 926 }, { "epoch": 0.06408572416176979, "grad_norm": 0.9247052204096256, "learning_rate": 9.999894596028546e-06, "loss": 0.3607, "step": 927 }, { "epoch": 0.0641548565502938, "grad_norm": 0.9008410340120262, "learning_rate": 9.99989093008924e-06, "loss": 0.3377, "step": 928 }, { "epoch": 0.06422398893881784, "grad_norm": 1.183465039798963, "learning_rate": 9.999887201485223e-06, "loss": 0.5677, "step": 929 }, { "epoch": 0.06429312132734186, "grad_norm": 0.9138280887645298, "learning_rate": 9.999883410216544e-06, "loss": 0.365, "step": 930 }, { "epoch": 0.06436225371586589, "grad_norm": 0.9690178480793651, "learning_rate": 9.99987955628325e-06, "loss": 0.3425, "step": 931 }, { "epoch": 0.0644313861043899, "grad_norm": 1.004775738849027, "learning_rate": 9.999875639685388e-06, "loss": 0.3877, "step": 932 }, { "epoch": 0.06450051849291392, "grad_norm": 0.9899212757340785, "learning_rate": 9.999871660423006e-06, "loss": 0.5287, "step": 933 }, { "epoch": 0.06456965088143796, "grad_norm": 1.1184623201128527, "learning_rate": 9.999867618496157e-06, "loss": 0.4414, "step": 934 }, { "epoch": 0.06463878326996197, "grad_norm": 0.8231440798655695, "learning_rate": 9.999863513904889e-06, "loss": 0.3202, "step": 935 }, { "epoch": 0.064707915658486, "grad_norm": 1.1452641470111755, "learning_rate": 9.999859346649254e-06, "loss": 0.3906, "step": 936 }, { "epoch": 0.06477704804701002, "grad_norm": 0.9574233579438406, "learning_rate": 9.999855116729306e-06, "loss": 0.4512, "step": 937 }, { "epoch": 0.06484618043553404, "grad_norm": 0.8983843212609535, "learning_rate": 9.999850824145098e-06, "loss": 0.3458, "step": 938 }, { "epoch": 0.06491531282405807, "grad_norm": 0.9936331361127062, "learning_rate": 9.999846468896682e-06, "loss": 0.3503, "step": 939 }, { "epoch": 0.06498444521258209, "grad_norm": 0.9657896801450934, "learning_rate": 9.99984205098411e-06, "loss": 0.4398, "step": 940 }, { "epoch": 0.06505357760110612, "grad_norm": 1.0584033003861297, "learning_rate": 9.999837570407443e-06, "loss": 0.4328, "step": 941 }, { "epoch": 0.06512270998963014, "grad_norm": 1.0734882472526945, "learning_rate": 9.999833027166734e-06, "loss": 0.3556, "step": 942 }, { "epoch": 0.06519184237815416, "grad_norm": 0.9370219943938416, "learning_rate": 9.999828421262042e-06, "loss": 0.4494, "step": 943 }, { "epoch": 0.06526097476667819, "grad_norm": 1.0182749965736968, "learning_rate": 9.999823752693421e-06, "loss": 0.448, "step": 944 }, { "epoch": 0.06533010715520221, "grad_norm": 0.9267120902769568, "learning_rate": 9.999819021460934e-06, "loss": 0.351, "step": 945 }, { "epoch": 0.06539923954372624, "grad_norm": 0.9315607562624151, "learning_rate": 9.999814227564638e-06, "loss": 0.3154, "step": 946 }, { "epoch": 0.06546837193225026, "grad_norm": 1.0812280285251148, "learning_rate": 9.999809371004592e-06, "loss": 0.4758, "step": 947 }, { "epoch": 0.06553750432077428, "grad_norm": 0.9524203759877388, "learning_rate": 9.999804451780858e-06, "loss": 0.4005, "step": 948 }, { "epoch": 0.06560663670929831, "grad_norm": 1.0825057018944486, "learning_rate": 9.999799469893499e-06, "loss": 0.4257, "step": 949 }, { "epoch": 0.06567576909782233, "grad_norm": 1.00837757595466, "learning_rate": 9.999794425342576e-06, "loss": 0.4072, "step": 950 }, { "epoch": 0.06574490148634635, "grad_norm": 0.9675972887140407, "learning_rate": 9.999789318128152e-06, "loss": 0.4356, "step": 951 }, { "epoch": 0.06581403387487038, "grad_norm": 1.0263078278292543, "learning_rate": 9.999784148250292e-06, "loss": 0.3659, "step": 952 }, { "epoch": 0.0658831662633944, "grad_norm": 0.9548851190407376, "learning_rate": 9.999778915709059e-06, "loss": 0.3534, "step": 953 }, { "epoch": 0.06595229865191843, "grad_norm": 1.024506076999301, "learning_rate": 9.99977362050452e-06, "loss": 0.5, "step": 954 }, { "epoch": 0.06602143104044245, "grad_norm": 0.9264958046896059, "learning_rate": 9.999768262636742e-06, "loss": 0.3845, "step": 955 }, { "epoch": 0.06609056342896646, "grad_norm": 0.9904822505821607, "learning_rate": 9.99976284210579e-06, "loss": 0.3604, "step": 956 }, { "epoch": 0.0661596958174905, "grad_norm": 1.0168209317190675, "learning_rate": 9.999757358911735e-06, "loss": 0.339, "step": 957 }, { "epoch": 0.06622882820601451, "grad_norm": 1.0298012715224618, "learning_rate": 9.999751813054642e-06, "loss": 0.4972, "step": 958 }, { "epoch": 0.06629796059453855, "grad_norm": 0.9475619411407971, "learning_rate": 9.999746204534583e-06, "loss": 0.3761, "step": 959 }, { "epoch": 0.06636709298306256, "grad_norm": 1.0594935130582959, "learning_rate": 9.99974053335163e-06, "loss": 0.3985, "step": 960 }, { "epoch": 0.06643622537158658, "grad_norm": 1.3080216410915613, "learning_rate": 9.999734799505849e-06, "loss": 0.5051, "step": 961 }, { "epoch": 0.06650535776011061, "grad_norm": 1.1295728342167979, "learning_rate": 9.999729002997315e-06, "loss": 0.4336, "step": 962 }, { "epoch": 0.06657449014863463, "grad_norm": 0.9986012006542015, "learning_rate": 9.999723143826103e-06, "loss": 0.2931, "step": 963 }, { "epoch": 0.06664362253715866, "grad_norm": 0.9388057357523713, "learning_rate": 9.999717221992282e-06, "loss": 0.4306, "step": 964 }, { "epoch": 0.06671275492568268, "grad_norm": 1.114978989084716, "learning_rate": 9.999711237495928e-06, "loss": 0.5165, "step": 965 }, { "epoch": 0.0667818873142067, "grad_norm": 1.1855154365897256, "learning_rate": 9.999705190337116e-06, "loss": 0.401, "step": 966 }, { "epoch": 0.06685101970273073, "grad_norm": 1.1287438544385706, "learning_rate": 9.999699080515922e-06, "loss": 0.3567, "step": 967 }, { "epoch": 0.06692015209125475, "grad_norm": 0.9447889917768751, "learning_rate": 9.999692908032423e-06, "loss": 0.4118, "step": 968 }, { "epoch": 0.06698928447977878, "grad_norm": 1.2768509576053377, "learning_rate": 9.999686672886695e-06, "loss": 0.4245, "step": 969 }, { "epoch": 0.0670584168683028, "grad_norm": 1.0861892935688133, "learning_rate": 9.999680375078817e-06, "loss": 0.3652, "step": 970 }, { "epoch": 0.06712754925682682, "grad_norm": 1.1266294301995667, "learning_rate": 9.999674014608866e-06, "loss": 0.3753, "step": 971 }, { "epoch": 0.06719668164535085, "grad_norm": 1.0997240004294742, "learning_rate": 9.999667591476927e-06, "loss": 0.4531, "step": 972 }, { "epoch": 0.06726581403387487, "grad_norm": 0.9470244829490974, "learning_rate": 9.999661105683074e-06, "loss": 0.3916, "step": 973 }, { "epoch": 0.0673349464223989, "grad_norm": 0.9380510560818653, "learning_rate": 9.999654557227393e-06, "loss": 0.3259, "step": 974 }, { "epoch": 0.06740407881092292, "grad_norm": 0.9520844231360566, "learning_rate": 9.999647946109964e-06, "loss": 0.3691, "step": 975 }, { "epoch": 0.06747321119944694, "grad_norm": 0.9605442183611288, "learning_rate": 9.999641272330871e-06, "loss": 0.53, "step": 976 }, { "epoch": 0.06754234358797097, "grad_norm": 1.0103463476274455, "learning_rate": 9.999634535890196e-06, "loss": 0.4195, "step": 977 }, { "epoch": 0.06761147597649499, "grad_norm": 0.8549052044747788, "learning_rate": 9.999627736788025e-06, "loss": 0.3623, "step": 978 }, { "epoch": 0.06768060836501902, "grad_norm": 0.9540260643052012, "learning_rate": 9.999620875024443e-06, "loss": 0.4343, "step": 979 }, { "epoch": 0.06774974075354304, "grad_norm": 0.9820580504663413, "learning_rate": 9.999613950599534e-06, "loss": 0.4746, "step": 980 }, { "epoch": 0.06781887314206705, "grad_norm": 0.9928668975268171, "learning_rate": 9.999606963513387e-06, "loss": 0.3866, "step": 981 }, { "epoch": 0.06788800553059109, "grad_norm": 1.3697516731538029, "learning_rate": 9.99959991376609e-06, "loss": 0.4273, "step": 982 }, { "epoch": 0.0679571379191151, "grad_norm": 1.0261411743824933, "learning_rate": 9.999592801357729e-06, "loss": 0.4528, "step": 983 }, { "epoch": 0.06802627030763912, "grad_norm": 1.0082444988068053, "learning_rate": 9.999585626288394e-06, "loss": 0.3888, "step": 984 }, { "epoch": 0.06809540269616315, "grad_norm": 0.9991503520699594, "learning_rate": 9.999578388558177e-06, "loss": 0.3187, "step": 985 }, { "epoch": 0.06816453508468717, "grad_norm": 0.991252947549112, "learning_rate": 9.999571088167165e-06, "loss": 0.4744, "step": 986 }, { "epoch": 0.0682336674732112, "grad_norm": 1.0125760572827534, "learning_rate": 9.999563725115454e-06, "loss": 0.3856, "step": 987 }, { "epoch": 0.06830279986173522, "grad_norm": 0.8846009712150107, "learning_rate": 9.999556299403133e-06, "loss": 0.3621, "step": 988 }, { "epoch": 0.06837193225025924, "grad_norm": 0.9562401116826124, "learning_rate": 9.999548811030296e-06, "loss": 0.3212, "step": 989 }, { "epoch": 0.06844106463878327, "grad_norm": 1.1971984773050262, "learning_rate": 9.999541259997034e-06, "loss": 0.4495, "step": 990 }, { "epoch": 0.06851019702730729, "grad_norm": 0.9843211712925136, "learning_rate": 9.999533646303447e-06, "loss": 0.3807, "step": 991 }, { "epoch": 0.06857932941583132, "grad_norm": 1.0391384444547425, "learning_rate": 9.999525969949627e-06, "loss": 0.3527, "step": 992 }, { "epoch": 0.06864846180435534, "grad_norm": 1.1861803409363858, "learning_rate": 9.999518230935672e-06, "loss": 0.46, "step": 993 }, { "epoch": 0.06871759419287936, "grad_norm": 0.9646713059886165, "learning_rate": 9.999510429261677e-06, "loss": 0.4528, "step": 994 }, { "epoch": 0.06878672658140339, "grad_norm": 0.9271225641335341, "learning_rate": 9.999502564927742e-06, "loss": 0.4279, "step": 995 }, { "epoch": 0.06885585896992741, "grad_norm": 1.1015436221926518, "learning_rate": 9.999494637933964e-06, "loss": 0.3403, "step": 996 }, { "epoch": 0.06892499135845144, "grad_norm": 1.2204150062794785, "learning_rate": 9.99948664828044e-06, "loss": 0.4907, "step": 997 }, { "epoch": 0.06899412374697546, "grad_norm": 1.0003160616801148, "learning_rate": 9.999478595967275e-06, "loss": 0.4571, "step": 998 }, { "epoch": 0.06906325613549948, "grad_norm": 1.1113440253170572, "learning_rate": 9.999470480994569e-06, "loss": 0.3347, "step": 999 }, { "epoch": 0.06913238852402351, "grad_norm": 1.0585895810249506, "learning_rate": 9.999462303362422e-06, "loss": 0.4244, "step": 1000 }, { "epoch": 0.06920152091254753, "grad_norm": 0.8641433316603874, "learning_rate": 9.999454063070934e-06, "loss": 0.4952, "step": 1001 }, { "epoch": 0.06927065330107156, "grad_norm": 1.3761722236540144, "learning_rate": 9.999445760120214e-06, "loss": 0.3818, "step": 1002 }, { "epoch": 0.06933978568959558, "grad_norm": 0.9860599350840393, "learning_rate": 9.999437394510365e-06, "loss": 0.3569, "step": 1003 }, { "epoch": 0.0694089180781196, "grad_norm": 1.024281211093185, "learning_rate": 9.999428966241487e-06, "loss": 0.4985, "step": 1004 }, { "epoch": 0.06947805046664363, "grad_norm": 0.9962553636531493, "learning_rate": 9.99942047531369e-06, "loss": 0.4513, "step": 1005 }, { "epoch": 0.06954718285516764, "grad_norm": 1.3494116735455475, "learning_rate": 9.999411921727079e-06, "loss": 0.3945, "step": 1006 }, { "epoch": 0.06961631524369168, "grad_norm": 1.2208854612035118, "learning_rate": 9.999403305481761e-06, "loss": 0.403, "step": 1007 }, { "epoch": 0.0696854476322157, "grad_norm": 1.338702735923687, "learning_rate": 9.999394626577845e-06, "loss": 0.4953, "step": 1008 }, { "epoch": 0.06975458002073971, "grad_norm": 1.0694107602238978, "learning_rate": 9.99938588501544e-06, "loss": 0.3561, "step": 1009 }, { "epoch": 0.06982371240926374, "grad_norm": 0.9382077741508901, "learning_rate": 9.999377080794653e-06, "loss": 0.4472, "step": 1010 }, { "epoch": 0.06989284479778776, "grad_norm": 1.4576502855170066, "learning_rate": 9.999368213915596e-06, "loss": 0.4054, "step": 1011 }, { "epoch": 0.0699619771863118, "grad_norm": 0.945622278875527, "learning_rate": 9.999359284378382e-06, "loss": 0.4419, "step": 1012 }, { "epoch": 0.07003110957483581, "grad_norm": 1.205955717077917, "learning_rate": 9.99935029218312e-06, "loss": 0.4312, "step": 1013 }, { "epoch": 0.07010024196335983, "grad_norm": 0.9928440710465665, "learning_rate": 9.999341237329924e-06, "loss": 0.4, "step": 1014 }, { "epoch": 0.07016937435188386, "grad_norm": 1.0608039841743477, "learning_rate": 9.999332119818907e-06, "loss": 0.4721, "step": 1015 }, { "epoch": 0.07023850674040788, "grad_norm": 0.9808105738853488, "learning_rate": 9.999322939650184e-06, "loss": 0.4009, "step": 1016 }, { "epoch": 0.0703076391289319, "grad_norm": 0.9118206070000798, "learning_rate": 9.99931369682387e-06, "loss": 0.4387, "step": 1017 }, { "epoch": 0.07037677151745593, "grad_norm": 1.1938809851931997, "learning_rate": 9.999304391340079e-06, "loss": 0.4053, "step": 1018 }, { "epoch": 0.07044590390597995, "grad_norm": 0.9565602507978196, "learning_rate": 9.99929502319893e-06, "loss": 0.475, "step": 1019 }, { "epoch": 0.07051503629450398, "grad_norm": 0.971971165879703, "learning_rate": 9.99928559240054e-06, "loss": 0.3583, "step": 1020 }, { "epoch": 0.070584168683028, "grad_norm": 0.9653703248512424, "learning_rate": 9.999276098945026e-06, "loss": 0.3683, "step": 1021 }, { "epoch": 0.07065330107155202, "grad_norm": 0.9482392592774449, "learning_rate": 9.999266542832509e-06, "loss": 0.4527, "step": 1022 }, { "epoch": 0.07072243346007605, "grad_norm": 0.9229661631502938, "learning_rate": 9.999256924063105e-06, "loss": 0.4315, "step": 1023 }, { "epoch": 0.07079156584860007, "grad_norm": 0.9266532535104791, "learning_rate": 9.99924724263694e-06, "loss": 0.3418, "step": 1024 }, { "epoch": 0.0708606982371241, "grad_norm": 1.0280480861097, "learning_rate": 9.99923749855413e-06, "loss": 0.3885, "step": 1025 }, { "epoch": 0.07092983062564812, "grad_norm": 0.8911021091144288, "learning_rate": 9.999227691814801e-06, "loss": 0.4999, "step": 1026 }, { "epoch": 0.07099896301417213, "grad_norm": 1.0342983536437584, "learning_rate": 9.999217822419073e-06, "loss": 0.4041, "step": 1027 }, { "epoch": 0.07106809540269617, "grad_norm": 1.1029515119759985, "learning_rate": 9.999207890367073e-06, "loss": 0.3801, "step": 1028 }, { "epoch": 0.07113722779122018, "grad_norm": 1.1331798764239784, "learning_rate": 9.999197895658923e-06, "loss": 0.4931, "step": 1029 }, { "epoch": 0.07120636017974422, "grad_norm": 0.9583283478340918, "learning_rate": 9.999187838294749e-06, "loss": 0.4345, "step": 1030 }, { "epoch": 0.07127549256826823, "grad_norm": 0.9768227226768681, "learning_rate": 9.999177718274676e-06, "loss": 0.3812, "step": 1031 }, { "epoch": 0.07134462495679225, "grad_norm": 1.068347837309202, "learning_rate": 9.99916753559883e-06, "loss": 0.4394, "step": 1032 }, { "epoch": 0.07141375734531628, "grad_norm": 1.0000620542967815, "learning_rate": 9.999157290267342e-06, "loss": 0.4379, "step": 1033 }, { "epoch": 0.0714828897338403, "grad_norm": 0.8687645212693739, "learning_rate": 9.999146982280341e-06, "loss": 0.346, "step": 1034 }, { "epoch": 0.07155202212236433, "grad_norm": 0.9706015370289571, "learning_rate": 9.999136611637952e-06, "loss": 0.3455, "step": 1035 }, { "epoch": 0.07162115451088835, "grad_norm": 0.9705274710021112, "learning_rate": 9.999126178340307e-06, "loss": 0.4401, "step": 1036 }, { "epoch": 0.07169028689941237, "grad_norm": 0.9082741712403924, "learning_rate": 9.999115682387537e-06, "loss": 0.4192, "step": 1037 }, { "epoch": 0.0717594192879364, "grad_norm": 0.8439146080207847, "learning_rate": 9.999105123779773e-06, "loss": 0.3721, "step": 1038 }, { "epoch": 0.07182855167646042, "grad_norm": 0.9608448587605329, "learning_rate": 9.999094502517147e-06, "loss": 0.384, "step": 1039 }, { "epoch": 0.07189768406498445, "grad_norm": 1.0008141044934797, "learning_rate": 9.999083818599793e-06, "loss": 0.4917, "step": 1040 }, { "epoch": 0.07196681645350847, "grad_norm": 1.1727894380636794, "learning_rate": 9.999073072027845e-06, "loss": 0.4442, "step": 1041 }, { "epoch": 0.07203594884203249, "grad_norm": 1.0417431934360828, "learning_rate": 9.999062262801438e-06, "loss": 0.3429, "step": 1042 }, { "epoch": 0.07210508123055652, "grad_norm": 1.4234578992698221, "learning_rate": 9.999051390920706e-06, "loss": 0.464, "step": 1043 }, { "epoch": 0.07217421361908054, "grad_norm": 1.0896513554351455, "learning_rate": 9.999040456385785e-06, "loss": 0.4847, "step": 1044 }, { "epoch": 0.07224334600760456, "grad_norm": 1.0017912494011367, "learning_rate": 9.999029459196815e-06, "loss": 0.3315, "step": 1045 }, { "epoch": 0.07231247839612859, "grad_norm": 1.0864470016898016, "learning_rate": 9.99901839935393e-06, "loss": 0.3801, "step": 1046 }, { "epoch": 0.0723816107846526, "grad_norm": 0.9272965184136405, "learning_rate": 9.99900727685727e-06, "loss": 0.4908, "step": 1047 }, { "epoch": 0.07245074317317664, "grad_norm": 0.9410296779529902, "learning_rate": 9.998996091706978e-06, "loss": 0.3487, "step": 1048 }, { "epoch": 0.07251987556170066, "grad_norm": 1.0168926045267566, "learning_rate": 9.99898484390319e-06, "loss": 0.3717, "step": 1049 }, { "epoch": 0.07258900795022467, "grad_norm": 1.0330221469873833, "learning_rate": 9.998973533446047e-06, "loss": 0.4224, "step": 1050 }, { "epoch": 0.0726581403387487, "grad_norm": 1.0233061447969, "learning_rate": 9.998962160335693e-06, "loss": 0.4786, "step": 1051 }, { "epoch": 0.07272727272727272, "grad_norm": 1.5723492634553098, "learning_rate": 9.998950724572268e-06, "loss": 0.4009, "step": 1052 }, { "epoch": 0.07279640511579676, "grad_norm": 0.8737817950247643, "learning_rate": 9.998939226155917e-06, "loss": 0.3129, "step": 1053 }, { "epoch": 0.07286553750432077, "grad_norm": 1.0041238811198072, "learning_rate": 9.998927665086784e-06, "loss": 0.4643, "step": 1054 }, { "epoch": 0.07293466989284479, "grad_norm": 1.0801788055418056, "learning_rate": 9.998916041365014e-06, "loss": 0.3735, "step": 1055 }, { "epoch": 0.07300380228136882, "grad_norm": 0.979077619767224, "learning_rate": 9.998904354990752e-06, "loss": 0.3505, "step": 1056 }, { "epoch": 0.07307293466989284, "grad_norm": 1.0952417682409725, "learning_rate": 9.998892605964145e-06, "loss": 0.3691, "step": 1057 }, { "epoch": 0.07314206705841687, "grad_norm": 0.9516844104181885, "learning_rate": 9.998880794285341e-06, "loss": 0.4699, "step": 1058 }, { "epoch": 0.07321119944694089, "grad_norm": 0.9884220903332859, "learning_rate": 9.998868919954486e-06, "loss": 0.3658, "step": 1059 }, { "epoch": 0.07328033183546491, "grad_norm": 0.9624218205456084, "learning_rate": 9.998856982971732e-06, "loss": 0.3168, "step": 1060 }, { "epoch": 0.07334946422398894, "grad_norm": 0.9783814929772076, "learning_rate": 9.998844983337223e-06, "loss": 0.4757, "step": 1061 }, { "epoch": 0.07341859661251296, "grad_norm": 0.8951597554966363, "learning_rate": 9.998832921051115e-06, "loss": 0.4232, "step": 1062 }, { "epoch": 0.07348772900103699, "grad_norm": 0.9868742932178194, "learning_rate": 9.998820796113558e-06, "loss": 0.3502, "step": 1063 }, { "epoch": 0.07355686138956101, "grad_norm": 0.9406051378983638, "learning_rate": 9.998808608524702e-06, "loss": 0.3761, "step": 1064 }, { "epoch": 0.07362599377808503, "grad_norm": 0.9497494488108041, "learning_rate": 9.998796358284702e-06, "loss": 0.5063, "step": 1065 }, { "epoch": 0.07369512616660906, "grad_norm": 1.0103384782162537, "learning_rate": 9.998784045393708e-06, "loss": 0.4369, "step": 1066 }, { "epoch": 0.07376425855513308, "grad_norm": 0.9521295206389665, "learning_rate": 9.998771669851878e-06, "loss": 0.3326, "step": 1067 }, { "epoch": 0.07383339094365711, "grad_norm": 0.9974981840936068, "learning_rate": 9.998759231659367e-06, "loss": 0.4147, "step": 1068 }, { "epoch": 0.07390252333218113, "grad_norm": 1.2850940668254887, "learning_rate": 9.99874673081633e-06, "loss": 0.4083, "step": 1069 }, { "epoch": 0.07397165572070515, "grad_norm": 1.2191652482988917, "learning_rate": 9.99873416732292e-06, "loss": 0.3765, "step": 1070 }, { "epoch": 0.07404078810922918, "grad_norm": 0.9942606402164368, "learning_rate": 9.9987215411793e-06, "loss": 0.3438, "step": 1071 }, { "epoch": 0.0741099204977532, "grad_norm": 1.0025194295982403, "learning_rate": 9.998708852385626e-06, "loss": 0.55, "step": 1072 }, { "epoch": 0.07417905288627723, "grad_norm": 1.1199285752965804, "learning_rate": 9.998696100942059e-06, "loss": 0.3983, "step": 1073 }, { "epoch": 0.07424818527480125, "grad_norm": 1.0429493487896022, "learning_rate": 9.998683286848754e-06, "loss": 0.3467, "step": 1074 }, { "epoch": 0.07431731766332526, "grad_norm": 0.9296980300750888, "learning_rate": 9.998670410105875e-06, "loss": 0.387, "step": 1075 }, { "epoch": 0.0743864500518493, "grad_norm": 0.9299873444853399, "learning_rate": 9.998657470713584e-06, "loss": 0.4786, "step": 1076 }, { "epoch": 0.07445558244037331, "grad_norm": 0.9964227417516645, "learning_rate": 9.998644468672042e-06, "loss": 0.3832, "step": 1077 }, { "epoch": 0.07452471482889733, "grad_norm": 0.8340728794491916, "learning_rate": 9.998631403981411e-06, "loss": 0.317, "step": 1078 }, { "epoch": 0.07459384721742136, "grad_norm": 1.100490960201524, "learning_rate": 9.998618276641857e-06, "loss": 0.4333, "step": 1079 }, { "epoch": 0.07466297960594538, "grad_norm": 0.8790700140567504, "learning_rate": 9.998605086653544e-06, "loss": 0.3847, "step": 1080 }, { "epoch": 0.07473211199446941, "grad_norm": 0.8740193566745099, "learning_rate": 9.998591834016634e-06, "loss": 0.2979, "step": 1081 }, { "epoch": 0.07480124438299343, "grad_norm": 1.0531714472467528, "learning_rate": 9.998578518731297e-06, "loss": 0.4169, "step": 1082 }, { "epoch": 0.07487037677151745, "grad_norm": 1.0004951082222484, "learning_rate": 9.9985651407977e-06, "loss": 0.4847, "step": 1083 }, { "epoch": 0.07493950916004148, "grad_norm": 0.940585705847675, "learning_rate": 9.998551700216009e-06, "loss": 0.3818, "step": 1084 }, { "epoch": 0.0750086415485655, "grad_norm": 1.0991040930348415, "learning_rate": 9.998538196986392e-06, "loss": 0.3758, "step": 1085 }, { "epoch": 0.07507777393708953, "grad_norm": 0.9151696096105958, "learning_rate": 9.99852463110902e-06, "loss": 0.5338, "step": 1086 }, { "epoch": 0.07514690632561355, "grad_norm": 0.9145717525340431, "learning_rate": 9.99851100258406e-06, "loss": 0.4296, "step": 1087 }, { "epoch": 0.07521603871413757, "grad_norm": 0.8965141497319112, "learning_rate": 9.998497311411686e-06, "loss": 0.3583, "step": 1088 }, { "epoch": 0.0752851711026616, "grad_norm": 0.9774637270467906, "learning_rate": 9.998483557592066e-06, "loss": 0.3612, "step": 1089 }, { "epoch": 0.07535430349118562, "grad_norm": 1.0287223596497188, "learning_rate": 9.998469741125378e-06, "loss": 0.4541, "step": 1090 }, { "epoch": 0.07542343587970965, "grad_norm": 0.9449068911300671, "learning_rate": 9.998455862011791e-06, "loss": 0.4491, "step": 1091 }, { "epoch": 0.07549256826823367, "grad_norm": 0.8262068044561527, "learning_rate": 9.998441920251481e-06, "loss": 0.2887, "step": 1092 }, { "epoch": 0.07556170065675769, "grad_norm": 0.9775716653923491, "learning_rate": 9.99842791584462e-06, "loss": 0.3924, "step": 1093 }, { "epoch": 0.07563083304528172, "grad_norm": 1.0636600452055773, "learning_rate": 9.998413848791386e-06, "loss": 0.4697, "step": 1094 }, { "epoch": 0.07569996543380574, "grad_norm": 0.9393265119136496, "learning_rate": 9.998399719091954e-06, "loss": 0.3625, "step": 1095 }, { "epoch": 0.07576909782232977, "grad_norm": 0.9458519590764772, "learning_rate": 9.998385526746503e-06, "loss": 0.3598, "step": 1096 }, { "epoch": 0.07583823021085379, "grad_norm": 1.2531720814936262, "learning_rate": 9.998371271755206e-06, "loss": 0.5203, "step": 1097 }, { "epoch": 0.0759073625993778, "grad_norm": 0.9285130655734729, "learning_rate": 9.99835695411825e-06, "loss": 0.4002, "step": 1098 }, { "epoch": 0.07597649498790184, "grad_norm": 0.9865987119085521, "learning_rate": 9.998342573835806e-06, "loss": 0.3819, "step": 1099 }, { "epoch": 0.07604562737642585, "grad_norm": 1.2972058403032882, "learning_rate": 9.99832813090806e-06, "loss": 0.4568, "step": 1100 }, { "epoch": 0.07611475976494989, "grad_norm": 0.9561339274952227, "learning_rate": 9.998313625335188e-06, "loss": 0.4394, "step": 1101 }, { "epoch": 0.0761838921534739, "grad_norm": 0.9082979353081906, "learning_rate": 9.998299057117376e-06, "loss": 0.3328, "step": 1102 }, { "epoch": 0.07625302454199792, "grad_norm": 1.0161757480197722, "learning_rate": 9.998284426254806e-06, "loss": 0.3467, "step": 1103 }, { "epoch": 0.07632215693052195, "grad_norm": 1.0337326150744206, "learning_rate": 9.998269732747658e-06, "loss": 0.4881, "step": 1104 }, { "epoch": 0.07639128931904597, "grad_norm": 0.9652131688220571, "learning_rate": 9.998254976596123e-06, "loss": 0.3973, "step": 1105 }, { "epoch": 0.07646042170757, "grad_norm": 1.2691100792294627, "learning_rate": 9.998240157800378e-06, "loss": 0.362, "step": 1106 }, { "epoch": 0.07652955409609402, "grad_norm": 0.9486398456523261, "learning_rate": 9.998225276360613e-06, "loss": 0.387, "step": 1107 }, { "epoch": 0.07659868648461804, "grad_norm": 1.0043359785885138, "learning_rate": 9.998210332277015e-06, "loss": 0.5087, "step": 1108 }, { "epoch": 0.07666781887314207, "grad_norm": 0.9399400131800728, "learning_rate": 9.99819532554977e-06, "loss": 0.434, "step": 1109 }, { "epoch": 0.07673695126166609, "grad_norm": 0.9465502149543069, "learning_rate": 9.998180256179066e-06, "loss": 0.3411, "step": 1110 }, { "epoch": 0.07680608365019011, "grad_norm": 1.023226703888972, "learning_rate": 9.99816512416509e-06, "loss": 0.4086, "step": 1111 }, { "epoch": 0.07687521603871414, "grad_norm": 1.0518019437766832, "learning_rate": 9.998149929508038e-06, "loss": 0.4313, "step": 1112 }, { "epoch": 0.07694434842723816, "grad_norm": 0.9314090257900993, "learning_rate": 9.998134672208094e-06, "loss": 0.3828, "step": 1113 }, { "epoch": 0.07701348081576219, "grad_norm": 0.9681605957243309, "learning_rate": 9.998119352265452e-06, "loss": 0.3744, "step": 1114 }, { "epoch": 0.07708261320428621, "grad_norm": 0.990347442182821, "learning_rate": 9.998103969680302e-06, "loss": 0.4727, "step": 1115 }, { "epoch": 0.07715174559281023, "grad_norm": 1.474396620756858, "learning_rate": 9.99808852445284e-06, "loss": 0.3747, "step": 1116 }, { "epoch": 0.07722087798133426, "grad_norm": 0.9663821429860595, "learning_rate": 9.998073016583256e-06, "loss": 0.3014, "step": 1117 }, { "epoch": 0.07729001036985828, "grad_norm": 0.9832188877793244, "learning_rate": 9.998057446071749e-06, "loss": 0.3951, "step": 1118 }, { "epoch": 0.07735914275838231, "grad_norm": 1.1766915164144736, "learning_rate": 9.99804181291851e-06, "loss": 0.4799, "step": 1119 }, { "epoch": 0.07742827514690633, "grad_norm": 1.0067839701079442, "learning_rate": 9.998026117123734e-06, "loss": 0.3568, "step": 1120 }, { "epoch": 0.07749740753543034, "grad_norm": 1.1855062270192414, "learning_rate": 9.998010358687623e-06, "loss": 0.3114, "step": 1121 }, { "epoch": 0.07756653992395438, "grad_norm": 1.0061204062166416, "learning_rate": 9.99799453761037e-06, "loss": 0.4347, "step": 1122 }, { "epoch": 0.0776356723124784, "grad_norm": 1.1001145728011343, "learning_rate": 9.997978653892176e-06, "loss": 0.3687, "step": 1123 }, { "epoch": 0.07770480470100243, "grad_norm": 1.1775781929504079, "learning_rate": 9.997962707533237e-06, "loss": 0.3298, "step": 1124 }, { "epoch": 0.07777393708952644, "grad_norm": 0.974865423418445, "learning_rate": 9.997946698533755e-06, "loss": 0.42, "step": 1125 }, { "epoch": 0.07784306947805046, "grad_norm": 1.0224390917183541, "learning_rate": 9.997930626893933e-06, "loss": 0.4628, "step": 1126 }, { "epoch": 0.0779122018665745, "grad_norm": 1.464455809108208, "learning_rate": 9.997914492613969e-06, "loss": 0.3837, "step": 1127 }, { "epoch": 0.07798133425509851, "grad_norm": 1.1890157450330916, "learning_rate": 9.997898295694065e-06, "loss": 0.3991, "step": 1128 }, { "epoch": 0.07805046664362254, "grad_norm": 1.02497615653495, "learning_rate": 9.997882036134424e-06, "loss": 0.4865, "step": 1129 }, { "epoch": 0.07811959903214656, "grad_norm": 0.9589910941859441, "learning_rate": 9.997865713935252e-06, "loss": 0.5004, "step": 1130 }, { "epoch": 0.07818873142067058, "grad_norm": 0.9484256313589982, "learning_rate": 9.997849329096752e-06, "loss": 0.3653, "step": 1131 }, { "epoch": 0.07825786380919461, "grad_norm": 1.0961763291912767, "learning_rate": 9.997832881619129e-06, "loss": 0.3941, "step": 1132 }, { "epoch": 0.07832699619771863, "grad_norm": 0.970427800896756, "learning_rate": 9.99781637150259e-06, "loss": 0.4696, "step": 1133 }, { "epoch": 0.07839612858624266, "grad_norm": 0.9782246793161251, "learning_rate": 9.997799798747345e-06, "loss": 0.4062, "step": 1134 }, { "epoch": 0.07846526097476668, "grad_norm": 0.8937789802225815, "learning_rate": 9.997783163353597e-06, "loss": 0.3756, "step": 1135 }, { "epoch": 0.0785343933632907, "grad_norm": 0.916038975683207, "learning_rate": 9.997766465321554e-06, "loss": 0.4358, "step": 1136 }, { "epoch": 0.07860352575181473, "grad_norm": 1.045961267696719, "learning_rate": 9.997749704651427e-06, "loss": 0.4915, "step": 1137 }, { "epoch": 0.07867265814033875, "grad_norm": 0.960552358986224, "learning_rate": 9.99773288134343e-06, "loss": 0.3724, "step": 1138 }, { "epoch": 0.07874179052886278, "grad_norm": 0.8999372829973054, "learning_rate": 9.997715995397766e-06, "loss": 0.345, "step": 1139 }, { "epoch": 0.0788109229173868, "grad_norm": 0.9808198156455734, "learning_rate": 9.997699046814654e-06, "loss": 0.5358, "step": 1140 }, { "epoch": 0.07888005530591082, "grad_norm": 0.9569511285104841, "learning_rate": 9.9976820355943e-06, "loss": 0.3972, "step": 1141 }, { "epoch": 0.07894918769443485, "grad_norm": 0.970485903834354, "learning_rate": 9.997664961736923e-06, "loss": 0.3709, "step": 1142 }, { "epoch": 0.07901832008295887, "grad_norm": 0.9401537156039208, "learning_rate": 9.997647825242734e-06, "loss": 0.4074, "step": 1143 }, { "epoch": 0.07908745247148288, "grad_norm": 0.9944070072596717, "learning_rate": 9.997630626111948e-06, "loss": 0.4147, "step": 1144 }, { "epoch": 0.07915658486000691, "grad_norm": 0.9114619667018251, "learning_rate": 9.997613364344781e-06, "loss": 0.3928, "step": 1145 }, { "epoch": 0.07922571724853093, "grad_norm": 4.560889492013093, "learning_rate": 9.997596039941449e-06, "loss": 0.3519, "step": 1146 }, { "epoch": 0.07929484963705496, "grad_norm": 0.9822600422564348, "learning_rate": 9.99757865290217e-06, "loss": 0.4557, "step": 1147 }, { "epoch": 0.07936398202557898, "grad_norm": 0.9738287645923075, "learning_rate": 9.99756120322716e-06, "loss": 0.4321, "step": 1148 }, { "epoch": 0.079433114414103, "grad_norm": 0.9600225450023792, "learning_rate": 9.997543690916638e-06, "loss": 0.3657, "step": 1149 }, { "epoch": 0.07950224680262703, "grad_norm": 0.9710027845607343, "learning_rate": 9.997526115970826e-06, "loss": 0.387, "step": 1150 }, { "epoch": 0.07957137919115105, "grad_norm": 0.9893559380020842, "learning_rate": 9.997508478389944e-06, "loss": 0.4865, "step": 1151 }, { "epoch": 0.07964051157967508, "grad_norm": 1.0809233411973012, "learning_rate": 9.997490778174208e-06, "loss": 0.3134, "step": 1152 }, { "epoch": 0.0797096439681991, "grad_norm": 0.9811302625332204, "learning_rate": 9.997473015323846e-06, "loss": 0.3829, "step": 1153 }, { "epoch": 0.07977877635672312, "grad_norm": 0.9883554464179812, "learning_rate": 9.997455189839078e-06, "loss": 0.3738, "step": 1154 }, { "epoch": 0.07984790874524715, "grad_norm": 1.0353285057756128, "learning_rate": 9.997437301720127e-06, "loss": 0.4846, "step": 1155 }, { "epoch": 0.07991704113377117, "grad_norm": 1.159665448615571, "learning_rate": 9.997419350967216e-06, "loss": 0.3771, "step": 1156 }, { "epoch": 0.0799861735222952, "grad_norm": 1.0541143282502976, "learning_rate": 9.997401337580574e-06, "loss": 0.4451, "step": 1157 }, { "epoch": 0.08005530591081922, "grad_norm": 0.9201139099865389, "learning_rate": 9.997383261560425e-06, "loss": 0.4728, "step": 1158 }, { "epoch": 0.08012443829934324, "grad_norm": 0.9391979368419673, "learning_rate": 9.997365122906994e-06, "loss": 0.3995, "step": 1159 }, { "epoch": 0.08019357068786727, "grad_norm": 1.029989543469751, "learning_rate": 9.99734692162051e-06, "loss": 0.3273, "step": 1160 }, { "epoch": 0.08026270307639129, "grad_norm": 0.9546580828250169, "learning_rate": 9.9973286577012e-06, "loss": 0.5047, "step": 1161 }, { "epoch": 0.08033183546491532, "grad_norm": 1.0598963479790597, "learning_rate": 9.997310331149293e-06, "loss": 0.4084, "step": 1162 }, { "epoch": 0.08040096785343934, "grad_norm": 0.9128197732742683, "learning_rate": 9.997291941965019e-06, "loss": 0.3198, "step": 1163 }, { "epoch": 0.08047010024196335, "grad_norm": 1.0672386494090684, "learning_rate": 9.99727349014861e-06, "loss": 0.3558, "step": 1164 }, { "epoch": 0.08053923263048739, "grad_norm": 0.9543502291327879, "learning_rate": 9.997254975700295e-06, "loss": 0.5065, "step": 1165 }, { "epoch": 0.0806083650190114, "grad_norm": 0.9638670718582977, "learning_rate": 9.997236398620307e-06, "loss": 0.45, "step": 1166 }, { "epoch": 0.08067749740753544, "grad_norm": 0.9903116621377496, "learning_rate": 9.997217758908877e-06, "loss": 0.371, "step": 1167 }, { "epoch": 0.08074662979605945, "grad_norm": 1.085905415896811, "learning_rate": 9.997199056566243e-06, "loss": 0.4059, "step": 1168 }, { "epoch": 0.08081576218458347, "grad_norm": 0.911613136277433, "learning_rate": 9.997180291592635e-06, "loss": 0.437, "step": 1169 }, { "epoch": 0.0808848945731075, "grad_norm": 0.8559205097661978, "learning_rate": 9.99716146398829e-06, "loss": 0.3362, "step": 1170 }, { "epoch": 0.08095402696163152, "grad_norm": 1.1905307625600976, "learning_rate": 9.997142573753445e-06, "loss": 0.3472, "step": 1171 }, { "epoch": 0.08102315935015555, "grad_norm": 1.021613716663006, "learning_rate": 9.997123620888336e-06, "loss": 0.4728, "step": 1172 }, { "epoch": 0.08109229173867957, "grad_norm": 1.0865017437636617, "learning_rate": 9.997104605393198e-06, "loss": 0.4175, "step": 1173 }, { "epoch": 0.08116142412720359, "grad_norm": 0.9033203588975269, "learning_rate": 9.997085527268272e-06, "loss": 0.3375, "step": 1174 }, { "epoch": 0.08123055651572762, "grad_norm": 0.9539858733462594, "learning_rate": 9.997066386513799e-06, "loss": 0.4184, "step": 1175 }, { "epoch": 0.08129968890425164, "grad_norm": 0.9233773915896305, "learning_rate": 9.997047183130014e-06, "loss": 0.4494, "step": 1176 }, { "epoch": 0.08136882129277566, "grad_norm": 1.0504246540801951, "learning_rate": 9.997027917117161e-06, "loss": 0.3428, "step": 1177 }, { "epoch": 0.08143795368129969, "grad_norm": 0.9447273895754222, "learning_rate": 9.99700858847548e-06, "loss": 0.384, "step": 1178 }, { "epoch": 0.08150708606982371, "grad_norm": 0.932547585703352, "learning_rate": 9.996989197205216e-06, "loss": 0.4945, "step": 1179 }, { "epoch": 0.08157621845834774, "grad_norm": 1.2291710982047641, "learning_rate": 9.996969743306608e-06, "loss": 0.4144, "step": 1180 }, { "epoch": 0.08164535084687176, "grad_norm": 1.0096502124536408, "learning_rate": 9.996950226779903e-06, "loss": 0.3837, "step": 1181 }, { "epoch": 0.08171448323539578, "grad_norm": 0.9358831669835388, "learning_rate": 9.996930647625343e-06, "loss": 0.4029, "step": 1182 }, { "epoch": 0.08178361562391981, "grad_norm": 1.0110061185106045, "learning_rate": 9.996911005843175e-06, "loss": 0.3971, "step": 1183 }, { "epoch": 0.08185274801244383, "grad_norm": 0.9700615224750345, "learning_rate": 9.996891301433645e-06, "loss": 0.3273, "step": 1184 }, { "epoch": 0.08192188040096786, "grad_norm": 0.9361215688827882, "learning_rate": 9.996871534397001e-06, "loss": 0.3604, "step": 1185 }, { "epoch": 0.08199101278949188, "grad_norm": 1.156359378309438, "learning_rate": 9.996851704733487e-06, "loss": 0.3847, "step": 1186 }, { "epoch": 0.0820601451780159, "grad_norm": 0.9164951882016928, "learning_rate": 9.996831812443358e-06, "loss": 0.5008, "step": 1187 }, { "epoch": 0.08212927756653993, "grad_norm": 0.9889926570123384, "learning_rate": 9.996811857526858e-06, "loss": 0.4025, "step": 1188 }, { "epoch": 0.08219840995506394, "grad_norm": 1.0093849654274327, "learning_rate": 9.996791839984236e-06, "loss": 0.4065, "step": 1189 }, { "epoch": 0.08226754234358798, "grad_norm": 1.730486115289571, "learning_rate": 9.996771759815748e-06, "loss": 0.5024, "step": 1190 }, { "epoch": 0.082336674732112, "grad_norm": 1.2926887901212623, "learning_rate": 9.996751617021643e-06, "loss": 0.4501, "step": 1191 }, { "epoch": 0.08240580712063601, "grad_norm": 0.9309815297418735, "learning_rate": 9.996731411602174e-06, "loss": 0.3805, "step": 1192 }, { "epoch": 0.08247493950916004, "grad_norm": 1.0417514869652444, "learning_rate": 9.996711143557592e-06, "loss": 0.4042, "step": 1193 }, { "epoch": 0.08254407189768406, "grad_norm": 0.944685314053884, "learning_rate": 9.996690812888154e-06, "loss": 0.4672, "step": 1194 }, { "epoch": 0.0826132042862081, "grad_norm": 1.7082656845334523, "learning_rate": 9.996670419594115e-06, "loss": 0.321, "step": 1195 }, { "epoch": 0.08268233667473211, "grad_norm": 0.9086050528990551, "learning_rate": 9.996649963675726e-06, "loss": 0.3477, "step": 1196 }, { "epoch": 0.08275146906325613, "grad_norm": 0.9946234142476881, "learning_rate": 9.99662944513325e-06, "loss": 0.5504, "step": 1197 }, { "epoch": 0.08282060145178016, "grad_norm": 1.170953815887587, "learning_rate": 9.996608863966939e-06, "loss": 0.4015, "step": 1198 }, { "epoch": 0.08288973384030418, "grad_norm": 0.8663505934107113, "learning_rate": 9.996588220177052e-06, "loss": 0.3288, "step": 1199 }, { "epoch": 0.08295886622882821, "grad_norm": 0.970404906263458, "learning_rate": 9.996567513763851e-06, "loss": 0.4299, "step": 1200 }, { "epoch": 0.08302799861735223, "grad_norm": 1.0052818485575503, "learning_rate": 9.99654674472759e-06, "loss": 0.4858, "step": 1201 }, { "epoch": 0.08309713100587625, "grad_norm": 1.07262641605287, "learning_rate": 9.996525913068534e-06, "loss": 0.3342, "step": 1202 }, { "epoch": 0.08316626339440028, "grad_norm": 0.8911188712430996, "learning_rate": 9.996505018786942e-06, "loss": 0.3439, "step": 1203 }, { "epoch": 0.0832353957829243, "grad_norm": 0.9248077491395994, "learning_rate": 9.996484061883076e-06, "loss": 0.4911, "step": 1204 }, { "epoch": 0.08330452817144833, "grad_norm": 0.9524673515654478, "learning_rate": 9.996463042357201e-06, "loss": 0.443, "step": 1205 }, { "epoch": 0.08337366055997235, "grad_norm": 0.9520502343312058, "learning_rate": 9.996441960209575e-06, "loss": 0.3653, "step": 1206 }, { "epoch": 0.08344279294849637, "grad_norm": 1.055603163091639, "learning_rate": 9.99642081544047e-06, "loss": 0.3935, "step": 1207 }, { "epoch": 0.0835119253370204, "grad_norm": 1.0870488342245401, "learning_rate": 9.996399608050142e-06, "loss": 0.462, "step": 1208 }, { "epoch": 0.08358105772554442, "grad_norm": 1.0983706638658524, "learning_rate": 9.996378338038863e-06, "loss": 0.3672, "step": 1209 }, { "epoch": 0.08365019011406843, "grad_norm": 1.0330132638231557, "learning_rate": 9.996357005406898e-06, "loss": 0.3081, "step": 1210 }, { "epoch": 0.08371932250259247, "grad_norm": 0.9259936364537277, "learning_rate": 9.996335610154515e-06, "loss": 0.4606, "step": 1211 }, { "epoch": 0.08378845489111648, "grad_norm": 1.0539126331645203, "learning_rate": 9.996314152281981e-06, "loss": 0.4731, "step": 1212 }, { "epoch": 0.08385758727964052, "grad_norm": 0.9311902585927091, "learning_rate": 9.996292631789566e-06, "loss": 0.3802, "step": 1213 }, { "epoch": 0.08392671966816453, "grad_norm": 1.0483819235663254, "learning_rate": 9.996271048677537e-06, "loss": 0.3467, "step": 1214 }, { "epoch": 0.08399585205668855, "grad_norm": 1.0034451758353082, "learning_rate": 9.996249402946168e-06, "loss": 0.5351, "step": 1215 }, { "epoch": 0.08406498444521258, "grad_norm": 0.8579337028382349, "learning_rate": 9.996227694595731e-06, "loss": 0.3293, "step": 1216 }, { "epoch": 0.0841341168337366, "grad_norm": 0.9240237831712707, "learning_rate": 9.996205923626492e-06, "loss": 0.3489, "step": 1217 }, { "epoch": 0.08420324922226063, "grad_norm": 1.0379217320718608, "learning_rate": 9.99618409003873e-06, "loss": 0.3918, "step": 1218 }, { "epoch": 0.08427238161078465, "grad_norm": 0.870334704831009, "learning_rate": 9.996162193832718e-06, "loss": 0.4937, "step": 1219 }, { "epoch": 0.08434151399930867, "grad_norm": 1.0666767558538781, "learning_rate": 9.996140235008727e-06, "loss": 0.3978, "step": 1220 }, { "epoch": 0.0844106463878327, "grad_norm": 0.8844944159833021, "learning_rate": 9.996118213567035e-06, "loss": 0.3861, "step": 1221 }, { "epoch": 0.08447977877635672, "grad_norm": 0.995331227436723, "learning_rate": 9.996096129507915e-06, "loss": 0.4813, "step": 1222 }, { "epoch": 0.08454891116488075, "grad_norm": 0.8911112188482574, "learning_rate": 9.996073982831648e-06, "loss": 0.3724, "step": 1223 }, { "epoch": 0.08461804355340477, "grad_norm": 1.134256130873072, "learning_rate": 9.99605177353851e-06, "loss": 0.3833, "step": 1224 }, { "epoch": 0.08468717594192879, "grad_norm": 0.9710776536151277, "learning_rate": 9.99602950162878e-06, "loss": 0.4561, "step": 1225 }, { "epoch": 0.08475630833045282, "grad_norm": 1.0641493210641064, "learning_rate": 9.996007167102732e-06, "loss": 0.5102, "step": 1226 }, { "epoch": 0.08482544071897684, "grad_norm": 0.9053373525516438, "learning_rate": 9.995984769960654e-06, "loss": 0.3379, "step": 1227 }, { "epoch": 0.08489457310750087, "grad_norm": 0.8793216508921184, "learning_rate": 9.99596231020282e-06, "loss": 0.3172, "step": 1228 }, { "epoch": 0.08496370549602489, "grad_norm": 0.963956386196712, "learning_rate": 9.995939787829515e-06, "loss": 0.4908, "step": 1229 }, { "epoch": 0.0850328378845489, "grad_norm": 1.002878876942591, "learning_rate": 9.99591720284102e-06, "loss": 0.383, "step": 1230 }, { "epoch": 0.08510197027307294, "grad_norm": 0.92941987494682, "learning_rate": 9.995894555237619e-06, "loss": 0.3471, "step": 1231 }, { "epoch": 0.08517110266159696, "grad_norm": 0.9831819514136324, "learning_rate": 9.995871845019595e-06, "loss": 0.4101, "step": 1232 }, { "epoch": 0.08524023505012099, "grad_norm": 0.9502407856892897, "learning_rate": 9.995849072187233e-06, "loss": 0.4696, "step": 1233 }, { "epoch": 0.085309367438645, "grad_norm": 0.9189837925428115, "learning_rate": 9.99582623674082e-06, "loss": 0.3408, "step": 1234 }, { "epoch": 0.08537849982716902, "grad_norm": 0.9573902982754489, "learning_rate": 9.995803338680637e-06, "loss": 0.3777, "step": 1235 }, { "epoch": 0.08544763221569306, "grad_norm": 0.9167596039490534, "learning_rate": 9.995780378006978e-06, "loss": 0.4602, "step": 1236 }, { "epoch": 0.08551676460421707, "grad_norm": 1.077494026509378, "learning_rate": 9.995757354720126e-06, "loss": 0.4374, "step": 1237 }, { "epoch": 0.08558589699274109, "grad_norm": 0.9392667378661237, "learning_rate": 9.99573426882037e-06, "loss": 0.3236, "step": 1238 }, { "epoch": 0.08565502938126512, "grad_norm": 0.9400325512355165, "learning_rate": 9.995711120308001e-06, "loss": 0.3977, "step": 1239 }, { "epoch": 0.08572416176978914, "grad_norm": 1.0039844294117448, "learning_rate": 9.99568790918331e-06, "loss": 0.4845, "step": 1240 }, { "epoch": 0.08579329415831317, "grad_norm": 0.901341517600172, "learning_rate": 9.995664635446583e-06, "loss": 0.3584, "step": 1241 }, { "epoch": 0.08586242654683719, "grad_norm": 1.013827236920727, "learning_rate": 9.995641299098117e-06, "loss": 0.3903, "step": 1242 }, { "epoch": 0.08593155893536121, "grad_norm": 0.9049228703013632, "learning_rate": 9.9956179001382e-06, "loss": 0.3275, "step": 1243 }, { "epoch": 0.08600069132388524, "grad_norm": 0.9880446232309988, "learning_rate": 9.995594438567131e-06, "loss": 0.5002, "step": 1244 }, { "epoch": 0.08606982371240926, "grad_norm": 0.892464564313617, "learning_rate": 9.9955709143852e-06, "loss": 0.3329, "step": 1245 }, { "epoch": 0.08613895610093329, "grad_norm": 0.9093853712184635, "learning_rate": 9.995547327592703e-06, "loss": 0.3352, "step": 1246 }, { "epoch": 0.08620808848945731, "grad_norm": 0.9853519825126292, "learning_rate": 9.995523678189934e-06, "loss": 0.5345, "step": 1247 }, { "epoch": 0.08627722087798133, "grad_norm": 0.8728225224579237, "learning_rate": 9.995499966177191e-06, "loss": 0.3134, "step": 1248 }, { "epoch": 0.08634635326650536, "grad_norm": 0.8503695755700212, "learning_rate": 9.99547619155477e-06, "loss": 0.306, "step": 1249 }, { "epoch": 0.08641548565502938, "grad_norm": 1.2607088838607399, "learning_rate": 9.99545235432297e-06, "loss": 0.3735, "step": 1250 }, { "epoch": 0.08648461804355341, "grad_norm": 0.8955351149052184, "learning_rate": 9.99542845448209e-06, "loss": 0.4793, "step": 1251 }, { "epoch": 0.08655375043207743, "grad_norm": 0.8789971518189889, "learning_rate": 9.99540449203243e-06, "loss": 0.368, "step": 1252 }, { "epoch": 0.08662288282060145, "grad_norm": 0.9935107191058143, "learning_rate": 9.995380466974288e-06, "loss": 0.3497, "step": 1253 }, { "epoch": 0.08669201520912548, "grad_norm": 0.9521275860935872, "learning_rate": 9.995356379307968e-06, "loss": 0.4472, "step": 1254 }, { "epoch": 0.0867611475976495, "grad_norm": 1.2676826371506265, "learning_rate": 9.995332229033768e-06, "loss": 0.486, "step": 1255 }, { "epoch": 0.08683027998617353, "grad_norm": 0.9414686120288236, "learning_rate": 9.995308016151994e-06, "loss": 0.3045, "step": 1256 }, { "epoch": 0.08689941237469755, "grad_norm": 0.9946825032659162, "learning_rate": 9.99528374066295e-06, "loss": 0.4364, "step": 1257 }, { "epoch": 0.08696854476322156, "grad_norm": 0.9895412889051587, "learning_rate": 9.995259402566937e-06, "loss": 0.4893, "step": 1258 }, { "epoch": 0.0870376771517456, "grad_norm": 1.1660404512783775, "learning_rate": 9.995235001864263e-06, "loss": 0.3555, "step": 1259 }, { "epoch": 0.08710680954026961, "grad_norm": 1.170353592640534, "learning_rate": 9.995210538555233e-06, "loss": 0.4154, "step": 1260 }, { "epoch": 0.08717594192879365, "grad_norm": 0.9488459574486133, "learning_rate": 9.995186012640153e-06, "loss": 0.4644, "step": 1261 }, { "epoch": 0.08724507431731766, "grad_norm": 0.981569689134902, "learning_rate": 9.995161424119328e-06, "loss": 0.4855, "step": 1262 }, { "epoch": 0.08731420670584168, "grad_norm": 1.212683771414972, "learning_rate": 9.995136772993073e-06, "loss": 0.3705, "step": 1263 }, { "epoch": 0.08738333909436571, "grad_norm": 1.1272728793199012, "learning_rate": 9.99511205926169e-06, "loss": 0.3601, "step": 1264 }, { "epoch": 0.08745247148288973, "grad_norm": 0.8939064631528871, "learning_rate": 9.995087282925492e-06, "loss": 0.4564, "step": 1265 }, { "epoch": 0.08752160387141376, "grad_norm": 1.0350602895545138, "learning_rate": 9.995062443984789e-06, "loss": 0.4, "step": 1266 }, { "epoch": 0.08759073625993778, "grad_norm": 1.016647753810208, "learning_rate": 9.995037542439894e-06, "loss": 0.3103, "step": 1267 }, { "epoch": 0.0876598686484618, "grad_norm": 0.8992524848108765, "learning_rate": 9.995012578291114e-06, "loss": 0.4367, "step": 1268 }, { "epoch": 0.08772900103698583, "grad_norm": 1.157533803460462, "learning_rate": 9.994987551538769e-06, "loss": 0.4573, "step": 1269 }, { "epoch": 0.08779813342550985, "grad_norm": 0.9470875578742115, "learning_rate": 9.994962462183166e-06, "loss": 0.3182, "step": 1270 }, { "epoch": 0.08786726581403387, "grad_norm": 0.9305764307452429, "learning_rate": 9.994937310224625e-06, "loss": 0.3506, "step": 1271 }, { "epoch": 0.0879363982025579, "grad_norm": 1.9000095594855737, "learning_rate": 9.994912095663457e-06, "loss": 0.4525, "step": 1272 }, { "epoch": 0.08800553059108192, "grad_norm": 0.9406797755367589, "learning_rate": 9.99488681849998e-06, "loss": 0.4653, "step": 1273 }, { "epoch": 0.08807466297960595, "grad_norm": 1.1363415914851174, "learning_rate": 9.99486147873451e-06, "loss": 0.3802, "step": 1274 }, { "epoch": 0.08814379536812997, "grad_norm": 0.9901995882850616, "learning_rate": 9.994836076367367e-06, "loss": 0.4522, "step": 1275 }, { "epoch": 0.08821292775665399, "grad_norm": 1.7614584332657948, "learning_rate": 9.994810611398866e-06, "loss": 0.4401, "step": 1276 }, { "epoch": 0.08828206014517802, "grad_norm": 0.9396538572414331, "learning_rate": 9.994785083829328e-06, "loss": 0.3487, "step": 1277 }, { "epoch": 0.08835119253370204, "grad_norm": 1.5666741545764262, "learning_rate": 9.994759493659073e-06, "loss": 0.3441, "step": 1278 }, { "epoch": 0.08842032492222607, "grad_norm": 0.8718162170700738, "learning_rate": 9.99473384088842e-06, "loss": 0.4788, "step": 1279 }, { "epoch": 0.08848945731075009, "grad_norm": 0.8840470557488733, "learning_rate": 9.994708125517694e-06, "loss": 0.4068, "step": 1280 }, { "epoch": 0.0885585896992741, "grad_norm": 0.88989126543233, "learning_rate": 9.994682347547214e-06, "loss": 0.3269, "step": 1281 }, { "epoch": 0.08862772208779814, "grad_norm": 1.128238775488948, "learning_rate": 9.994656506977302e-06, "loss": 0.3918, "step": 1282 }, { "epoch": 0.08869685447632215, "grad_norm": 0.9513283732154099, "learning_rate": 9.994630603808286e-06, "loss": 0.4915, "step": 1283 }, { "epoch": 0.08876598686484619, "grad_norm": 0.8903480046549512, "learning_rate": 9.99460463804049e-06, "loss": 0.3481, "step": 1284 }, { "epoch": 0.0888351192533702, "grad_norm": 0.8585951001711064, "learning_rate": 9.994578609674236e-06, "loss": 0.3643, "step": 1285 }, { "epoch": 0.08890425164189422, "grad_norm": 0.979843045772781, "learning_rate": 9.994552518709854e-06, "loss": 0.4339, "step": 1286 }, { "epoch": 0.08897338403041825, "grad_norm": 0.9364619164936354, "learning_rate": 9.994526365147667e-06, "loss": 0.3968, "step": 1287 }, { "epoch": 0.08904251641894227, "grad_norm": 0.892478916581723, "learning_rate": 9.994500148988007e-06, "loss": 0.3531, "step": 1288 }, { "epoch": 0.0891116488074663, "grad_norm": 0.9136329138732903, "learning_rate": 9.994473870231201e-06, "loss": 0.348, "step": 1289 }, { "epoch": 0.08918078119599032, "grad_norm": 1.0211087166347985, "learning_rate": 9.994447528877578e-06, "loss": 0.5041, "step": 1290 }, { "epoch": 0.08924991358451434, "grad_norm": 0.9286444050871783, "learning_rate": 9.994421124927468e-06, "loss": 0.3957, "step": 1291 }, { "epoch": 0.08931904597303837, "grad_norm": 0.9030236688848284, "learning_rate": 9.9943946583812e-06, "loss": 0.3198, "step": 1292 }, { "epoch": 0.08938817836156239, "grad_norm": 1.2586435251634434, "learning_rate": 9.994368129239111e-06, "loss": 0.4615, "step": 1293 }, { "epoch": 0.08945731075008642, "grad_norm": 0.9952573609324634, "learning_rate": 9.994341537501529e-06, "loss": 0.4642, "step": 1294 }, { "epoch": 0.08952644313861044, "grad_norm": 1.2597140176330364, "learning_rate": 9.994314883168787e-06, "loss": 0.3538, "step": 1295 }, { "epoch": 0.08959557552713446, "grad_norm": 0.9042179602893023, "learning_rate": 9.994288166241225e-06, "loss": 0.309, "step": 1296 }, { "epoch": 0.08966470791565849, "grad_norm": 1.0937954303635038, "learning_rate": 9.994261386719169e-06, "loss": 0.461, "step": 1297 }, { "epoch": 0.08973384030418251, "grad_norm": 1.1779405206468383, "learning_rate": 9.99423454460296e-06, "loss": 0.3838, "step": 1298 }, { "epoch": 0.08980297269270654, "grad_norm": 0.9067000807375226, "learning_rate": 9.994207639892936e-06, "loss": 0.3732, "step": 1299 }, { "epoch": 0.08987210508123056, "grad_norm": 0.9919066594191617, "learning_rate": 9.994180672589431e-06, "loss": 0.3739, "step": 1300 }, { "epoch": 0.08994123746975458, "grad_norm": 1.0516642979870354, "learning_rate": 9.994153642692782e-06, "loss": 0.5277, "step": 1301 }, { "epoch": 0.09001036985827861, "grad_norm": 0.9032892447005082, "learning_rate": 9.994126550203332e-06, "loss": 0.359, "step": 1302 }, { "epoch": 0.09007950224680263, "grad_norm": 1.043967518455724, "learning_rate": 9.994099395121417e-06, "loss": 0.3289, "step": 1303 }, { "epoch": 0.09014863463532664, "grad_norm": 0.9306256102614954, "learning_rate": 9.994072177447378e-06, "loss": 0.4689, "step": 1304 }, { "epoch": 0.09021776702385068, "grad_norm": 0.9401402407290489, "learning_rate": 9.994044897181557e-06, "loss": 0.4954, "step": 1305 }, { "epoch": 0.0902868994123747, "grad_norm": 0.8490917588601737, "learning_rate": 9.994017554324296e-06, "loss": 0.3221, "step": 1306 }, { "epoch": 0.09035603180089873, "grad_norm": 1.1396636066084582, "learning_rate": 9.993990148875936e-06, "loss": 0.4056, "step": 1307 }, { "epoch": 0.09042516418942274, "grad_norm": 0.9271657476467817, "learning_rate": 9.993962680836822e-06, "loss": 0.4503, "step": 1308 }, { "epoch": 0.09049429657794676, "grad_norm": 0.969393100987309, "learning_rate": 9.993935150207299e-06, "loss": 0.3456, "step": 1309 }, { "epoch": 0.0905634289664708, "grad_norm": 0.829014655143757, "learning_rate": 9.993907556987709e-06, "loss": 0.3914, "step": 1310 }, { "epoch": 0.09063256135499481, "grad_norm": 1.0265605018631458, "learning_rate": 9.993879901178402e-06, "loss": 0.513, "step": 1311 }, { "epoch": 0.09070169374351884, "grad_norm": 0.9207127270731305, "learning_rate": 9.993852182779721e-06, "loss": 0.4453, "step": 1312 }, { "epoch": 0.09077082613204286, "grad_norm": 1.090649671514249, "learning_rate": 9.993824401792015e-06, "loss": 0.3508, "step": 1313 }, { "epoch": 0.09083995852056688, "grad_norm": 0.7995760515589984, "learning_rate": 9.993796558215632e-06, "loss": 0.3431, "step": 1314 }, { "epoch": 0.09090909090909091, "grad_norm": 0.9837935999478162, "learning_rate": 9.993768652050921e-06, "loss": 0.495, "step": 1315 }, { "epoch": 0.09097822329761493, "grad_norm": 2.3509925750909098, "learning_rate": 9.993740683298233e-06, "loss": 0.4657, "step": 1316 }, { "epoch": 0.09104735568613896, "grad_norm": 0.9169058077326538, "learning_rate": 9.993712651957915e-06, "loss": 0.2972, "step": 1317 }, { "epoch": 0.09111648807466298, "grad_norm": 1.110514748587004, "learning_rate": 9.993684558030322e-06, "loss": 0.4329, "step": 1318 }, { "epoch": 0.091185620463187, "grad_norm": 1.034355198351695, "learning_rate": 9.993656401515804e-06, "loss": 0.515, "step": 1319 }, { "epoch": 0.09125475285171103, "grad_norm": 1.2564794316154522, "learning_rate": 9.993628182414717e-06, "loss": 0.3607, "step": 1320 }, { "epoch": 0.09132388524023505, "grad_norm": 0.9301897037667451, "learning_rate": 9.993599900727409e-06, "loss": 0.3725, "step": 1321 }, { "epoch": 0.09139301762875908, "grad_norm": 1.0476624374775587, "learning_rate": 9.993571556454238e-06, "loss": 0.4855, "step": 1322 }, { "epoch": 0.0914621500172831, "grad_norm": 1.111432855642023, "learning_rate": 9.993543149595562e-06, "loss": 0.3863, "step": 1323 }, { "epoch": 0.09153128240580712, "grad_norm": 0.9142198189477353, "learning_rate": 9.99351468015173e-06, "loss": 0.301, "step": 1324 }, { "epoch": 0.09160041479433115, "grad_norm": 1.427316029113324, "learning_rate": 9.993486148123106e-06, "loss": 0.4218, "step": 1325 }, { "epoch": 0.09166954718285517, "grad_norm": 0.9486582622081967, "learning_rate": 9.993457553510043e-06, "loss": 0.4751, "step": 1326 }, { "epoch": 0.0917386795713792, "grad_norm": 0.9131867127348563, "learning_rate": 9.9934288963129e-06, "loss": 0.3526, "step": 1327 }, { "epoch": 0.09180781195990322, "grad_norm": 1.1345746597314388, "learning_rate": 9.993400176532039e-06, "loss": 0.3366, "step": 1328 }, { "epoch": 0.09187694434842723, "grad_norm": 0.9883489456314897, "learning_rate": 9.993371394167816e-06, "loss": 0.464, "step": 1329 }, { "epoch": 0.09194607673695127, "grad_norm": 0.9339112271453729, "learning_rate": 9.993342549220594e-06, "loss": 0.4067, "step": 1330 }, { "epoch": 0.09201520912547528, "grad_norm": 1.0999618289227984, "learning_rate": 9.993313641690736e-06, "loss": 0.329, "step": 1331 }, { "epoch": 0.09208434151399932, "grad_norm": 1.4422892254291544, "learning_rate": 9.993284671578601e-06, "loss": 0.394, "step": 1332 }, { "epoch": 0.09215347390252333, "grad_norm": 0.8701591716019064, "learning_rate": 9.993255638884553e-06, "loss": 0.4197, "step": 1333 }, { "epoch": 0.09222260629104735, "grad_norm": 1.061540901307444, "learning_rate": 9.993226543608958e-06, "loss": 0.3811, "step": 1334 }, { "epoch": 0.09229173867957138, "grad_norm": 0.8828993214795289, "learning_rate": 9.993197385752178e-06, "loss": 0.3136, "step": 1335 }, { "epoch": 0.0923608710680954, "grad_norm": 1.592498634510321, "learning_rate": 9.99316816531458e-06, "loss": 0.4974, "step": 1336 }, { "epoch": 0.09243000345661942, "grad_norm": 0.9489588575520946, "learning_rate": 9.99313888229653e-06, "loss": 0.4084, "step": 1337 }, { "epoch": 0.09249913584514345, "grad_norm": 0.9521965829385999, "learning_rate": 9.993109536698394e-06, "loss": 0.3512, "step": 1338 }, { "epoch": 0.09256826823366747, "grad_norm": 0.8978250935652118, "learning_rate": 9.99308012852054e-06, "loss": 0.3689, "step": 1339 }, { "epoch": 0.0926374006221915, "grad_norm": 0.9329731375506387, "learning_rate": 9.99305065776334e-06, "loss": 0.4629, "step": 1340 }, { "epoch": 0.09270653301071552, "grad_norm": 1.1940333834902326, "learning_rate": 9.993021124427158e-06, "loss": 0.3702, "step": 1341 }, { "epoch": 0.09277566539923954, "grad_norm": 1.2206517507773729, "learning_rate": 9.992991528512366e-06, "loss": 0.3277, "step": 1342 }, { "epoch": 0.09284479778776357, "grad_norm": 0.9557967883373303, "learning_rate": 9.992961870019338e-06, "loss": 0.4441, "step": 1343 }, { "epoch": 0.09291393017628759, "grad_norm": 1.0284057776379683, "learning_rate": 9.992932148948441e-06, "loss": 0.473, "step": 1344 }, { "epoch": 0.09298306256481162, "grad_norm": 0.857487159162928, "learning_rate": 9.992902365300051e-06, "loss": 0.39, "step": 1345 }, { "epoch": 0.09305219495333564, "grad_norm": 1.1909582165623724, "learning_rate": 9.992872519074538e-06, "loss": 0.344, "step": 1346 }, { "epoch": 0.09312132734185966, "grad_norm": 1.0498972524101458, "learning_rate": 9.992842610272282e-06, "loss": 0.48, "step": 1347 }, { "epoch": 0.09319045973038369, "grad_norm": 0.972212720479562, "learning_rate": 9.992812638893652e-06, "loss": 0.4397, "step": 1348 }, { "epoch": 0.0932595921189077, "grad_norm": 1.013362569646985, "learning_rate": 9.992782604939026e-06, "loss": 0.3557, "step": 1349 }, { "epoch": 0.09332872450743174, "grad_norm": 0.9845181479919647, "learning_rate": 9.992752508408778e-06, "loss": 0.349, "step": 1350 }, { "epoch": 0.09339785689595576, "grad_norm": 0.9454263994614704, "learning_rate": 9.99272234930329e-06, "loss": 0.515, "step": 1351 }, { "epoch": 0.09346698928447977, "grad_norm": 0.9577614710168689, "learning_rate": 9.992692127622937e-06, "loss": 0.3755, "step": 1352 }, { "epoch": 0.0935361216730038, "grad_norm": 0.8723105566242058, "learning_rate": 9.992661843368097e-06, "loss": 0.3087, "step": 1353 }, { "epoch": 0.09360525406152782, "grad_norm": 0.9057429729685178, "learning_rate": 9.992631496539151e-06, "loss": 0.4526, "step": 1354 }, { "epoch": 0.09367438645005186, "grad_norm": 1.0043364237720969, "learning_rate": 9.992601087136479e-06, "loss": 0.35, "step": 1355 }, { "epoch": 0.09374351883857587, "grad_norm": 0.9830623601276742, "learning_rate": 9.992570615160462e-06, "loss": 0.3346, "step": 1356 }, { "epoch": 0.09381265122709989, "grad_norm": 0.9389493784041324, "learning_rate": 9.992540080611481e-06, "loss": 0.3774, "step": 1357 }, { "epoch": 0.09388178361562392, "grad_norm": 1.026707111776327, "learning_rate": 9.99250948348992e-06, "loss": 0.4615, "step": 1358 }, { "epoch": 0.09395091600414794, "grad_norm": 0.9044454012667368, "learning_rate": 9.992478823796163e-06, "loss": 0.4046, "step": 1359 }, { "epoch": 0.09402004839267197, "grad_norm": 1.0377583907109231, "learning_rate": 9.992448101530594e-06, "loss": 0.3556, "step": 1360 }, { "epoch": 0.09408918078119599, "grad_norm": 0.9538845929041091, "learning_rate": 9.992417316693597e-06, "loss": 0.4527, "step": 1361 }, { "epoch": 0.09415831316972001, "grad_norm": 1.006547643230916, "learning_rate": 9.992386469285558e-06, "loss": 0.4258, "step": 1362 }, { "epoch": 0.09422744555824404, "grad_norm": 1.0918547835815098, "learning_rate": 9.992355559306863e-06, "loss": 0.3781, "step": 1363 }, { "epoch": 0.09429657794676806, "grad_norm": 0.894011685870125, "learning_rate": 9.992324586757901e-06, "loss": 0.3726, "step": 1364 }, { "epoch": 0.09436571033529209, "grad_norm": 0.8786086895238441, "learning_rate": 9.99229355163906e-06, "loss": 0.4393, "step": 1365 }, { "epoch": 0.09443484272381611, "grad_norm": 0.8490626115515557, "learning_rate": 9.99226245395073e-06, "loss": 0.3743, "step": 1366 }, { "epoch": 0.09450397511234013, "grad_norm": 1.273702812221664, "learning_rate": 9.992231293693296e-06, "loss": 0.3292, "step": 1367 }, { "epoch": 0.09457310750086416, "grad_norm": 0.922093869702812, "learning_rate": 9.992200070867153e-06, "loss": 0.3703, "step": 1368 }, { "epoch": 0.09464223988938818, "grad_norm": 1.2291258400625862, "learning_rate": 9.992168785472691e-06, "loss": 0.5079, "step": 1369 }, { "epoch": 0.0947113722779122, "grad_norm": 0.8627560942825675, "learning_rate": 9.992137437510303e-06, "loss": 0.331, "step": 1370 }, { "epoch": 0.09478050466643623, "grad_norm": 1.1261982744095762, "learning_rate": 9.992106026980381e-06, "loss": 0.3158, "step": 1371 }, { "epoch": 0.09484963705496025, "grad_norm": 1.0139188546256164, "learning_rate": 9.992074553883319e-06, "loss": 0.505, "step": 1372 }, { "epoch": 0.09491876944348428, "grad_norm": 1.1256069845638532, "learning_rate": 9.99204301821951e-06, "loss": 0.4379, "step": 1373 }, { "epoch": 0.0949879018320083, "grad_norm": 1.0271431821938608, "learning_rate": 9.99201141998935e-06, "loss": 0.3429, "step": 1374 }, { "epoch": 0.09505703422053231, "grad_norm": 1.0469459261800376, "learning_rate": 9.991979759193238e-06, "loss": 0.3834, "step": 1375 }, { "epoch": 0.09512616660905635, "grad_norm": 0.8154530959349044, "learning_rate": 9.991948035831567e-06, "loss": 0.4215, "step": 1376 }, { "epoch": 0.09519529899758036, "grad_norm": 1.075046282755353, "learning_rate": 9.991916249904735e-06, "loss": 0.376, "step": 1377 }, { "epoch": 0.0952644313861044, "grad_norm": 1.4649718627542696, "learning_rate": 9.991884401413142e-06, "loss": 0.3123, "step": 1378 }, { "epoch": 0.09533356377462841, "grad_norm": 1.0702454460720527, "learning_rate": 9.991852490357187e-06, "loss": 0.4505, "step": 1379 }, { "epoch": 0.09540269616315243, "grad_norm": 0.9193222068342516, "learning_rate": 9.99182051673727e-06, "loss": 0.4447, "step": 1380 }, { "epoch": 0.09547182855167646, "grad_norm": 0.8245883755052017, "learning_rate": 9.991788480553787e-06, "loss": 0.3136, "step": 1381 }, { "epoch": 0.09554096094020048, "grad_norm": 1.0943395134369134, "learning_rate": 9.991756381807148e-06, "loss": 0.3974, "step": 1382 }, { "epoch": 0.09561009332872451, "grad_norm": 0.8969812643795291, "learning_rate": 9.99172422049775e-06, "loss": 0.4544, "step": 1383 }, { "epoch": 0.09567922571724853, "grad_norm": 0.8892431387891497, "learning_rate": 9.991691996625996e-06, "loss": 0.3614, "step": 1384 }, { "epoch": 0.09574835810577255, "grad_norm": 5.104338651079366, "learning_rate": 9.991659710192292e-06, "loss": 0.314, "step": 1385 }, { "epoch": 0.09581749049429658, "grad_norm": 0.9078634960225119, "learning_rate": 9.99162736119704e-06, "loss": 0.5138, "step": 1386 }, { "epoch": 0.0958866228828206, "grad_norm": 1.618972660968104, "learning_rate": 9.99159494964065e-06, "loss": 0.3982, "step": 1387 }, { "epoch": 0.09595575527134463, "grad_norm": 0.9541736561312554, "learning_rate": 9.991562475523521e-06, "loss": 0.3497, "step": 1388 }, { "epoch": 0.09602488765986865, "grad_norm": 1.0495070680749388, "learning_rate": 9.991529938846067e-06, "loss": 0.386, "step": 1389 }, { "epoch": 0.09609402004839267, "grad_norm": 0.9808161829740892, "learning_rate": 9.991497339608693e-06, "loss": 0.4952, "step": 1390 }, { "epoch": 0.0961631524369167, "grad_norm": 0.9610412359845105, "learning_rate": 9.991464677811808e-06, "loss": 0.4041, "step": 1391 }, { "epoch": 0.09623228482544072, "grad_norm": 0.8249774866778641, "learning_rate": 9.991431953455819e-06, "loss": 0.3658, "step": 1392 }, { "epoch": 0.09630141721396475, "grad_norm": 1.4557841073837845, "learning_rate": 9.991399166541141e-06, "loss": 0.437, "step": 1393 }, { "epoch": 0.09637054960248877, "grad_norm": 2.810897171829884, "learning_rate": 9.991366317068179e-06, "loss": 0.4998, "step": 1394 }, { "epoch": 0.09643968199101279, "grad_norm": 0.8304679273967321, "learning_rate": 9.99133340503735e-06, "loss": 0.376, "step": 1395 }, { "epoch": 0.09650881437953682, "grad_norm": 0.9159198686769942, "learning_rate": 9.991300430449061e-06, "loss": 0.4198, "step": 1396 }, { "epoch": 0.09657794676806084, "grad_norm": 1.040501222951447, "learning_rate": 9.99126739330373e-06, "loss": 0.4492, "step": 1397 }, { "epoch": 0.09664707915658487, "grad_norm": 1.0487760861190576, "learning_rate": 9.99123429360177e-06, "loss": 0.3739, "step": 1398 }, { "epoch": 0.09671621154510888, "grad_norm": 0.9312763792211907, "learning_rate": 9.991201131343597e-06, "loss": 0.3237, "step": 1399 }, { "epoch": 0.0967853439336329, "grad_norm": 0.9509255923985813, "learning_rate": 9.991167906529622e-06, "loss": 0.4159, "step": 1400 }, { "epoch": 0.09685447632215693, "grad_norm": 0.8634070278864897, "learning_rate": 9.991134619160266e-06, "loss": 0.4367, "step": 1401 }, { "epoch": 0.09692360871068095, "grad_norm": 0.9845113550148804, "learning_rate": 9.991101269235945e-06, "loss": 0.3756, "step": 1402 }, { "epoch": 0.09699274109920497, "grad_norm": 0.9205363709334476, "learning_rate": 9.991067856757075e-06, "loss": 0.3363, "step": 1403 }, { "epoch": 0.097061873487729, "grad_norm": 0.910305181487818, "learning_rate": 9.99103438172408e-06, "loss": 0.456, "step": 1404 }, { "epoch": 0.09713100587625302, "grad_norm": 0.9333976147660455, "learning_rate": 9.991000844137372e-06, "loss": 0.4487, "step": 1405 }, { "epoch": 0.09720013826477705, "grad_norm": 0.8802432418011846, "learning_rate": 9.990967243997376e-06, "loss": 0.3265, "step": 1406 }, { "epoch": 0.09726927065330107, "grad_norm": 1.0843511491488784, "learning_rate": 9.990933581304514e-06, "loss": 0.3995, "step": 1407 }, { "epoch": 0.09733840304182509, "grad_norm": 1.2393835174504129, "learning_rate": 9.990899856059205e-06, "loss": 0.4842, "step": 1408 }, { "epoch": 0.09740753543034912, "grad_norm": 0.8736924197932207, "learning_rate": 9.990866068261873e-06, "loss": 0.4702, "step": 1409 }, { "epoch": 0.09747666781887314, "grad_norm": 1.3418301211445922, "learning_rate": 9.990832217912941e-06, "loss": 0.4284, "step": 1410 }, { "epoch": 0.09754580020739717, "grad_norm": 1.2766026210490118, "learning_rate": 9.990798305012834e-06, "loss": 0.466, "step": 1411 }, { "epoch": 0.09761493259592119, "grad_norm": 0.9664017815379412, "learning_rate": 9.990764329561979e-06, "loss": 0.4128, "step": 1412 }, { "epoch": 0.0976840649844452, "grad_norm": 0.7587082592381507, "learning_rate": 9.990730291560797e-06, "loss": 0.3034, "step": 1413 }, { "epoch": 0.09775319737296924, "grad_norm": 1.1618696217265654, "learning_rate": 9.990696191009717e-06, "loss": 0.324, "step": 1414 }, { "epoch": 0.09782232976149326, "grad_norm": 1.985444919848497, "learning_rate": 9.990662027909168e-06, "loss": 0.5179, "step": 1415 }, { "epoch": 0.09789146215001729, "grad_norm": 1.714578547965601, "learning_rate": 9.990627802259576e-06, "loss": 0.4145, "step": 1416 }, { "epoch": 0.0979605945385413, "grad_norm": 1.018413590053367, "learning_rate": 9.990593514061371e-06, "loss": 0.351, "step": 1417 }, { "epoch": 0.09802972692706532, "grad_norm": 0.9208284183291916, "learning_rate": 9.990559163314981e-06, "loss": 0.4029, "step": 1418 }, { "epoch": 0.09809885931558936, "grad_norm": 0.9606625530580494, "learning_rate": 9.99052475002084e-06, "loss": 0.4383, "step": 1419 }, { "epoch": 0.09816799170411337, "grad_norm": 0.9188463499241577, "learning_rate": 9.990490274179376e-06, "loss": 0.3333, "step": 1420 }, { "epoch": 0.0982371240926374, "grad_norm": 1.1143946449149766, "learning_rate": 9.990455735791023e-06, "loss": 0.3235, "step": 1421 }, { "epoch": 0.09830625648116142, "grad_norm": 0.8699358513028022, "learning_rate": 9.990421134856213e-06, "loss": 0.4646, "step": 1422 }, { "epoch": 0.09837538886968544, "grad_norm": 0.8513573413110509, "learning_rate": 9.99038647137538e-06, "loss": 0.3602, "step": 1423 }, { "epoch": 0.09844452125820947, "grad_norm": 0.9174506754789623, "learning_rate": 9.990351745348959e-06, "loss": 0.376, "step": 1424 }, { "epoch": 0.09851365364673349, "grad_norm": 0.9367440847547674, "learning_rate": 9.990316956777382e-06, "loss": 0.4041, "step": 1425 }, { "epoch": 0.09858278603525752, "grad_norm": 1.0006289232057737, "learning_rate": 9.99028210566109e-06, "loss": 0.4236, "step": 1426 }, { "epoch": 0.09865191842378154, "grad_norm": 0.9475368701973893, "learning_rate": 9.990247192000516e-06, "loss": 0.3602, "step": 1427 }, { "epoch": 0.09872105081230556, "grad_norm": 1.0095461794574918, "learning_rate": 9.9902122157961e-06, "loss": 0.2846, "step": 1428 }, { "epoch": 0.09879018320082959, "grad_norm": 0.9488466702115955, "learning_rate": 9.990177177048279e-06, "loss": 0.4336, "step": 1429 }, { "epoch": 0.09885931558935361, "grad_norm": 1.0436029918386134, "learning_rate": 9.990142075757491e-06, "loss": 0.4482, "step": 1430 }, { "epoch": 0.09892844797787763, "grad_norm": 1.0729208287989613, "learning_rate": 9.99010691192418e-06, "loss": 0.4112, "step": 1431 }, { "epoch": 0.09899758036640166, "grad_norm": 0.9613394413475176, "learning_rate": 9.990071685548782e-06, "loss": 0.3632, "step": 1432 }, { "epoch": 0.09906671275492568, "grad_norm": 1.1691941033132636, "learning_rate": 9.990036396631742e-06, "loss": 0.4956, "step": 1433 }, { "epoch": 0.09913584514344971, "grad_norm": 0.8740899402878741, "learning_rate": 9.9900010451735e-06, "loss": 0.3331, "step": 1434 }, { "epoch": 0.09920497753197373, "grad_norm": 0.9144907545133037, "learning_rate": 9.9899656311745e-06, "loss": 0.3464, "step": 1435 }, { "epoch": 0.09927410992049775, "grad_norm": 0.9572298514778227, "learning_rate": 9.989930154635186e-06, "loss": 0.4755, "step": 1436 }, { "epoch": 0.09934324230902178, "grad_norm": 1.323136813883158, "learning_rate": 9.989894615556003e-06, "loss": 0.4551, "step": 1437 }, { "epoch": 0.0994123746975458, "grad_norm": 0.8780075015105552, "learning_rate": 9.989859013937395e-06, "loss": 0.3293, "step": 1438 }, { "epoch": 0.09948150708606983, "grad_norm": 0.9435566387759039, "learning_rate": 9.989823349779811e-06, "loss": 0.3022, "step": 1439 }, { "epoch": 0.09955063947459385, "grad_norm": 0.965433569827455, "learning_rate": 9.989787623083694e-06, "loss": 0.4661, "step": 1440 }, { "epoch": 0.09961977186311786, "grad_norm": 1.002437727057457, "learning_rate": 9.989751833849492e-06, "loss": 0.4228, "step": 1441 }, { "epoch": 0.0996889042516419, "grad_norm": 1.0704558187236228, "learning_rate": 9.98971598207766e-06, "loss": 0.3251, "step": 1442 }, { "epoch": 0.09975803664016591, "grad_norm": 1.6843364967976946, "learning_rate": 9.989680067768638e-06, "loss": 0.4309, "step": 1443 }, { "epoch": 0.09982716902868995, "grad_norm": 3.05868362370251, "learning_rate": 9.989644090922885e-06, "loss": 0.5319, "step": 1444 }, { "epoch": 0.09989630141721396, "grad_norm": 2.228596954431443, "learning_rate": 9.989608051540845e-06, "loss": 0.3537, "step": 1445 }, { "epoch": 0.09996543380573798, "grad_norm": 0.9424639769747717, "learning_rate": 9.989571949622973e-06, "loss": 0.359, "step": 1446 }, { "epoch": 0.10003456619426201, "grad_norm": 1.029787879838862, "learning_rate": 9.989535785169722e-06, "loss": 0.5099, "step": 1447 }, { "epoch": 0.10010369858278603, "grad_norm": 1.360778363019767, "learning_rate": 9.989499558181543e-06, "loss": 0.3792, "step": 1448 }, { "epoch": 0.10017283097131006, "grad_norm": 0.9171836109320647, "learning_rate": 9.989463268658892e-06, "loss": 0.3709, "step": 1449 }, { "epoch": 0.10024196335983408, "grad_norm": 1.0320798019581578, "learning_rate": 9.989426916602222e-06, "loss": 0.4422, "step": 1450 }, { "epoch": 0.1003110957483581, "grad_norm": 1.2343303100353804, "learning_rate": 9.989390502011992e-06, "loss": 0.4742, "step": 1451 }, { "epoch": 0.10038022813688213, "grad_norm": 1.0909615711724958, "learning_rate": 9.989354024888654e-06, "loss": 0.3448, "step": 1452 }, { "epoch": 0.10044936052540615, "grad_norm": 1.1750930824491461, "learning_rate": 9.989317485232667e-06, "loss": 0.3709, "step": 1453 }, { "epoch": 0.10051849291393018, "grad_norm": 0.9193772050641098, "learning_rate": 9.98928088304449e-06, "loss": 0.4406, "step": 1454 }, { "epoch": 0.1005876253024542, "grad_norm": 0.9252155893686941, "learning_rate": 9.989244218324581e-06, "loss": 0.4486, "step": 1455 }, { "epoch": 0.10065675769097822, "grad_norm": 1.5106598184559665, "learning_rate": 9.989207491073399e-06, "loss": 0.3311, "step": 1456 }, { "epoch": 0.10072589007950225, "grad_norm": 1.1063945421226087, "learning_rate": 9.989170701291403e-06, "loss": 0.4103, "step": 1457 }, { "epoch": 0.10079502246802627, "grad_norm": 1.6155700550583192, "learning_rate": 9.989133848979059e-06, "loss": 0.4687, "step": 1458 }, { "epoch": 0.1008641548565503, "grad_norm": 1.0719195412840792, "learning_rate": 9.989096934136825e-06, "loss": 0.3683, "step": 1459 }, { "epoch": 0.10093328724507432, "grad_norm": 0.9161382838108656, "learning_rate": 9.989059956765163e-06, "loss": 0.353, "step": 1460 }, { "epoch": 0.10100241963359834, "grad_norm": 0.8368748973545429, "learning_rate": 9.989022916864538e-06, "loss": 0.3975, "step": 1461 }, { "epoch": 0.10107155202212237, "grad_norm": 1.4842306428744663, "learning_rate": 9.988985814435414e-06, "loss": 0.4491, "step": 1462 }, { "epoch": 0.10114068441064639, "grad_norm": 1.1116332491087488, "learning_rate": 9.988948649478257e-06, "loss": 0.2796, "step": 1463 }, { "epoch": 0.1012098167991704, "grad_norm": 0.8985912601266614, "learning_rate": 9.988911421993532e-06, "loss": 0.3832, "step": 1464 }, { "epoch": 0.10127894918769444, "grad_norm": 1.0349426133439998, "learning_rate": 9.988874131981705e-06, "loss": 0.4909, "step": 1465 }, { "epoch": 0.10134808157621845, "grad_norm": 1.0245886477543005, "learning_rate": 9.988836779443244e-06, "loss": 0.4243, "step": 1466 }, { "epoch": 0.10141721396474249, "grad_norm": 1.0195580635805885, "learning_rate": 9.988799364378616e-06, "loss": 0.3932, "step": 1467 }, { "epoch": 0.1014863463532665, "grad_norm": 0.9956646463543867, "learning_rate": 9.988761886788291e-06, "loss": 0.4557, "step": 1468 }, { "epoch": 0.10155547874179052, "grad_norm": 0.9249239562878316, "learning_rate": 9.98872434667274e-06, "loss": 0.4288, "step": 1469 }, { "epoch": 0.10162461113031455, "grad_norm": 1.243189782603024, "learning_rate": 9.988686744032431e-06, "loss": 0.3669, "step": 1470 }, { "epoch": 0.10169374351883857, "grad_norm": 3.7499963661552154, "learning_rate": 9.988649078867836e-06, "loss": 0.3774, "step": 1471 }, { "epoch": 0.1017628759073626, "grad_norm": 1.197414028296773, "learning_rate": 9.98861135117943e-06, "loss": 0.4971, "step": 1472 }, { "epoch": 0.10183200829588662, "grad_norm": 0.9218301880946599, "learning_rate": 9.98857356096768e-06, "loss": 0.3923, "step": 1473 }, { "epoch": 0.10190114068441064, "grad_norm": 2.53707820427453, "learning_rate": 9.988535708233065e-06, "loss": 0.3599, "step": 1474 }, { "epoch": 0.10197027307293467, "grad_norm": 0.9499761524319336, "learning_rate": 9.988497792976054e-06, "loss": 0.4742, "step": 1475 }, { "epoch": 0.10203940546145869, "grad_norm": 1.0578202984783676, "learning_rate": 9.98845981519713e-06, "loss": 0.453, "step": 1476 }, { "epoch": 0.10210853784998272, "grad_norm": 0.9550645742717716, "learning_rate": 9.988421774896763e-06, "loss": 0.3501, "step": 1477 }, { "epoch": 0.10217767023850674, "grad_norm": 0.9379421254766634, "learning_rate": 9.98838367207543e-06, "loss": 0.3525, "step": 1478 }, { "epoch": 0.10224680262703076, "grad_norm": 2.355588072464702, "learning_rate": 9.988345506733612e-06, "loss": 0.4963, "step": 1479 }, { "epoch": 0.10231593501555479, "grad_norm": 0.8043556335937142, "learning_rate": 9.988307278871782e-06, "loss": 0.4059, "step": 1480 }, { "epoch": 0.10238506740407881, "grad_norm": 0.863856814521924, "learning_rate": 9.988268988490426e-06, "loss": 0.353, "step": 1481 }, { "epoch": 0.10245419979260284, "grad_norm": 1.0587302933929608, "learning_rate": 9.988230635590018e-06, "loss": 0.4192, "step": 1482 }, { "epoch": 0.10252333218112686, "grad_norm": 1.0581142713500948, "learning_rate": 9.988192220171041e-06, "loss": 0.4836, "step": 1483 }, { "epoch": 0.10259246456965088, "grad_norm": 0.9511931777486929, "learning_rate": 9.988153742233977e-06, "loss": 0.3597, "step": 1484 }, { "epoch": 0.10266159695817491, "grad_norm": 0.9798662622214169, "learning_rate": 9.988115201779307e-06, "loss": 0.4186, "step": 1485 }, { "epoch": 0.10273072934669893, "grad_norm": 1.1266759446701942, "learning_rate": 9.988076598807514e-06, "loss": 0.4334, "step": 1486 }, { "epoch": 0.10279986173522296, "grad_norm": 0.956209108830165, "learning_rate": 9.988037933319085e-06, "loss": 0.423, "step": 1487 }, { "epoch": 0.10286899412374698, "grad_norm": 1.0947918149917566, "learning_rate": 9.9879992053145e-06, "loss": 0.3134, "step": 1488 }, { "epoch": 0.102938126512271, "grad_norm": 0.8474230332647121, "learning_rate": 9.987960414794247e-06, "loss": 0.3722, "step": 1489 }, { "epoch": 0.10300725890079503, "grad_norm": 1.0844004707750494, "learning_rate": 9.98792156175881e-06, "loss": 0.4439, "step": 1490 }, { "epoch": 0.10307639128931904, "grad_norm": 0.8203044107125081, "learning_rate": 9.98788264620868e-06, "loss": 0.3597, "step": 1491 }, { "epoch": 0.10314552367784308, "grad_norm": 0.9862144864948214, "learning_rate": 9.987843668144341e-06, "loss": 0.3105, "step": 1492 }, { "epoch": 0.1032146560663671, "grad_norm": 1.0437943074839844, "learning_rate": 9.987804627566284e-06, "loss": 0.4219, "step": 1493 }, { "epoch": 0.10328378845489111, "grad_norm": 0.9540002853275963, "learning_rate": 9.987765524474997e-06, "loss": 0.4161, "step": 1494 }, { "epoch": 0.10335292084341514, "grad_norm": 0.9036693922316665, "learning_rate": 9.987726358870969e-06, "loss": 0.3465, "step": 1495 }, { "epoch": 0.10342205323193916, "grad_norm": 1.1979996318114787, "learning_rate": 9.987687130754693e-06, "loss": 0.3537, "step": 1496 }, { "epoch": 0.10349118562046318, "grad_norm": 1.3294814120530445, "learning_rate": 9.987647840126658e-06, "loss": 0.5214, "step": 1497 }, { "epoch": 0.10356031800898721, "grad_norm": 1.3834777596824035, "learning_rate": 9.98760848698736e-06, "loss": 0.4627, "step": 1498 }, { "epoch": 0.10362945039751123, "grad_norm": 0.9515082015341948, "learning_rate": 9.98756907133729e-06, "loss": 0.368, "step": 1499 }, { "epoch": 0.10369858278603526, "grad_norm": 0.9701666201251367, "learning_rate": 9.987529593176944e-06, "loss": 0.4588, "step": 1500 }, { "epoch": 0.10376771517455928, "grad_norm": 0.850212472667123, "learning_rate": 9.987490052506812e-06, "loss": 0.4397, "step": 1501 }, { "epoch": 0.1038368475630833, "grad_norm": 0.8299331471834376, "learning_rate": 9.987450449327394e-06, "loss": 0.3504, "step": 1502 }, { "epoch": 0.10390597995160733, "grad_norm": 0.9906980391269821, "learning_rate": 9.987410783639186e-06, "loss": 0.3914, "step": 1503 }, { "epoch": 0.10397511234013135, "grad_norm": 0.9812336806837441, "learning_rate": 9.987371055442683e-06, "loss": 0.4063, "step": 1504 }, { "epoch": 0.10404424472865538, "grad_norm": 0.9026652027701749, "learning_rate": 9.987331264738386e-06, "loss": 0.4361, "step": 1505 }, { "epoch": 0.1041133771171794, "grad_norm": 1.086356208832275, "learning_rate": 9.98729141152679e-06, "loss": 0.3675, "step": 1506 }, { "epoch": 0.10418250950570342, "grad_norm": 1.0763499462485595, "learning_rate": 9.987251495808397e-06, "loss": 0.3641, "step": 1507 }, { "epoch": 0.10425164189422745, "grad_norm": 1.0381375977877698, "learning_rate": 9.987211517583707e-06, "loss": 0.4963, "step": 1508 }, { "epoch": 0.10432077428275147, "grad_norm": 1.241926585113269, "learning_rate": 9.98717147685322e-06, "loss": 0.3748, "step": 1509 }, { "epoch": 0.1043899066712755, "grad_norm": 0.8899348746969463, "learning_rate": 9.98713137361744e-06, "loss": 0.3686, "step": 1510 }, { "epoch": 0.10445903905979952, "grad_norm": 1.05032837740411, "learning_rate": 9.987091207876866e-06, "loss": 0.4566, "step": 1511 }, { "epoch": 0.10452817144832353, "grad_norm": 1.0078540936723681, "learning_rate": 9.987050979632006e-06, "loss": 0.4174, "step": 1512 }, { "epoch": 0.10459730383684757, "grad_norm": 0.8855100406549847, "learning_rate": 9.987010688883361e-06, "loss": 0.3309, "step": 1513 }, { "epoch": 0.10466643622537158, "grad_norm": 0.9736047732317262, "learning_rate": 9.986970335631436e-06, "loss": 0.3836, "step": 1514 }, { "epoch": 0.10473556861389562, "grad_norm": 0.8950464934334125, "learning_rate": 9.986929919876737e-06, "loss": 0.475, "step": 1515 }, { "epoch": 0.10480470100241963, "grad_norm": 1.1027215957664573, "learning_rate": 9.986889441619772e-06, "loss": 0.4043, "step": 1516 }, { "epoch": 0.10487383339094365, "grad_norm": 1.0845347890088692, "learning_rate": 9.986848900861048e-06, "loss": 0.3097, "step": 1517 }, { "epoch": 0.10494296577946768, "grad_norm": 1.1728903311650551, "learning_rate": 9.986808297601071e-06, "loss": 0.4201, "step": 1518 }, { "epoch": 0.1050120981679917, "grad_norm": 0.8972914341649447, "learning_rate": 9.986767631840353e-06, "loss": 0.4813, "step": 1519 }, { "epoch": 0.10508123055651573, "grad_norm": 0.8796209336010109, "learning_rate": 9.9867269035794e-06, "loss": 0.3208, "step": 1520 }, { "epoch": 0.10515036294503975, "grad_norm": 0.9274635142295844, "learning_rate": 9.986686112818726e-06, "loss": 0.3773, "step": 1521 }, { "epoch": 0.10521949533356377, "grad_norm": 0.978254625846592, "learning_rate": 9.98664525955884e-06, "loss": 0.507, "step": 1522 }, { "epoch": 0.1052886277220878, "grad_norm": 0.8481763116387178, "learning_rate": 9.986604343800254e-06, "loss": 0.3591, "step": 1523 }, { "epoch": 0.10535776011061182, "grad_norm": 0.9158577940289321, "learning_rate": 9.986563365543483e-06, "loss": 0.367, "step": 1524 }, { "epoch": 0.10542689249913585, "grad_norm": 1.014893642809448, "learning_rate": 9.98652232478904e-06, "loss": 0.4022, "step": 1525 }, { "epoch": 0.10549602488765987, "grad_norm": 0.926610131863713, "learning_rate": 9.986481221537437e-06, "loss": 0.5144, "step": 1526 }, { "epoch": 0.10556515727618389, "grad_norm": 1.371535277752158, "learning_rate": 9.98644005578919e-06, "loss": 0.321, "step": 1527 }, { "epoch": 0.10563428966470792, "grad_norm": 0.9902829816954609, "learning_rate": 9.986398827544816e-06, "loss": 0.3383, "step": 1528 }, { "epoch": 0.10570342205323194, "grad_norm": 0.9651816962738824, "learning_rate": 9.986357536804832e-06, "loss": 0.43, "step": 1529 }, { "epoch": 0.10577255444175596, "grad_norm": 0.9910357605023896, "learning_rate": 9.986316183569755e-06, "loss": 0.4329, "step": 1530 }, { "epoch": 0.10584168683027999, "grad_norm": 0.8956797317566446, "learning_rate": 9.986274767840103e-06, "loss": 0.3751, "step": 1531 }, { "epoch": 0.105910819218804, "grad_norm": 1.1379234294812692, "learning_rate": 9.986233289616394e-06, "loss": 0.4226, "step": 1532 }, { "epoch": 0.10597995160732804, "grad_norm": 1.0399360124831138, "learning_rate": 9.98619174889915e-06, "loss": 0.43, "step": 1533 }, { "epoch": 0.10604908399585206, "grad_norm": 0.9015167661301005, "learning_rate": 9.98615014568889e-06, "loss": 0.3515, "step": 1534 }, { "epoch": 0.10611821638437607, "grad_norm": 1.1366568870714826, "learning_rate": 9.986108479986137e-06, "loss": 0.422, "step": 1535 }, { "epoch": 0.1061873487729001, "grad_norm": 0.9233945468762678, "learning_rate": 9.986066751791412e-06, "loss": 0.4402, "step": 1536 }, { "epoch": 0.10625648116142412, "grad_norm": 2.013488779558297, "learning_rate": 9.986024961105239e-06, "loss": 0.4728, "step": 1537 }, { "epoch": 0.10632561354994816, "grad_norm": 0.9804871485434129, "learning_rate": 9.98598310792814e-06, "loss": 0.3327, "step": 1538 }, { "epoch": 0.10639474593847217, "grad_norm": 0.9468443276018575, "learning_rate": 9.98594119226064e-06, "loss": 0.3078, "step": 1539 }, { "epoch": 0.10646387832699619, "grad_norm": 0.944767327476592, "learning_rate": 9.985899214103267e-06, "loss": 0.4114, "step": 1540 }, { "epoch": 0.10653301071552022, "grad_norm": 0.9601806093602883, "learning_rate": 9.985857173456543e-06, "loss": 0.3903, "step": 1541 }, { "epoch": 0.10660214310404424, "grad_norm": 0.8361784330135144, "learning_rate": 9.985815070320998e-06, "loss": 0.3394, "step": 1542 }, { "epoch": 0.10667127549256827, "grad_norm": 0.9582658556080277, "learning_rate": 9.98577290469716e-06, "loss": 0.4019, "step": 1543 }, { "epoch": 0.10674040788109229, "grad_norm": 1.0217460565699508, "learning_rate": 9.985730676585555e-06, "loss": 0.4484, "step": 1544 }, { "epoch": 0.10680954026961631, "grad_norm": 0.921238312039243, "learning_rate": 9.985688385986714e-06, "loss": 0.3687, "step": 1545 }, { "epoch": 0.10687867265814034, "grad_norm": 1.3188339740562962, "learning_rate": 9.985646032901166e-06, "loss": 0.3311, "step": 1546 }, { "epoch": 0.10694780504666436, "grad_norm": 0.9792257339189674, "learning_rate": 9.98560361732944e-06, "loss": 0.4662, "step": 1547 }, { "epoch": 0.10701693743518839, "grad_norm": 1.2136372051861406, "learning_rate": 9.985561139272072e-06, "loss": 0.418, "step": 1548 }, { "epoch": 0.10708606982371241, "grad_norm": 0.8370001945930152, "learning_rate": 9.985518598729593e-06, "loss": 0.4163, "step": 1549 }, { "epoch": 0.10715520221223643, "grad_norm": 1.031591376702481, "learning_rate": 9.985475995702533e-06, "loss": 0.3348, "step": 1550 }, { "epoch": 0.10722433460076046, "grad_norm": 1.1369366148760238, "learning_rate": 9.985433330191432e-06, "loss": 0.4145, "step": 1551 }, { "epoch": 0.10729346698928448, "grad_norm": 1.3235110475365468, "learning_rate": 9.985390602196819e-06, "loss": 0.3546, "step": 1552 }, { "epoch": 0.10736259937780851, "grad_norm": 1.4123856436712021, "learning_rate": 9.98534781171923e-06, "loss": 0.3578, "step": 1553 }, { "epoch": 0.10743173176633253, "grad_norm": 1.0694272710100243, "learning_rate": 9.985304958759205e-06, "loss": 0.436, "step": 1554 }, { "epoch": 0.10750086415485655, "grad_norm": 1.0516093423970003, "learning_rate": 9.985262043317279e-06, "loss": 0.4155, "step": 1555 }, { "epoch": 0.10756999654338058, "grad_norm": 0.9273811456675485, "learning_rate": 9.98521906539399e-06, "loss": 0.3123, "step": 1556 }, { "epoch": 0.1076391289319046, "grad_norm": 1.0034222669034323, "learning_rate": 9.985176024989876e-06, "loss": 0.4241, "step": 1557 }, { "epoch": 0.10770826132042863, "grad_norm": 1.2858303258868637, "learning_rate": 9.985132922105475e-06, "loss": 0.4801, "step": 1558 }, { "epoch": 0.10777739370895265, "grad_norm": 0.9366967463306574, "learning_rate": 9.985089756741333e-06, "loss": 0.4321, "step": 1559 }, { "epoch": 0.10784652609747666, "grad_norm": 0.7867666740474766, "learning_rate": 9.985046528897987e-06, "loss": 0.307, "step": 1560 }, { "epoch": 0.1079156584860007, "grad_norm": 0.9723990220708553, "learning_rate": 9.985003238575976e-06, "loss": 0.4468, "step": 1561 }, { "epoch": 0.10798479087452471, "grad_norm": 1.0284830229896598, "learning_rate": 9.984959885775846e-06, "loss": 0.4394, "step": 1562 }, { "epoch": 0.10805392326304873, "grad_norm": 1.1446290415320797, "learning_rate": 9.984916470498142e-06, "loss": 0.3559, "step": 1563 }, { "epoch": 0.10812305565157276, "grad_norm": 0.9486855221492888, "learning_rate": 9.984872992743404e-06, "loss": 0.3816, "step": 1564 }, { "epoch": 0.10819218804009678, "grad_norm": 0.9213539901803638, "learning_rate": 9.98482945251218e-06, "loss": 0.4395, "step": 1565 }, { "epoch": 0.10826132042862081, "grad_norm": 0.9447382242795943, "learning_rate": 9.984785849805014e-06, "loss": 0.398, "step": 1566 }, { "epoch": 0.10833045281714483, "grad_norm": 1.0866654335416595, "learning_rate": 9.984742184622453e-06, "loss": 0.3301, "step": 1567 }, { "epoch": 0.10839958520566885, "grad_norm": 0.8802138749093297, "learning_rate": 9.984698456965045e-06, "loss": 0.4359, "step": 1568 }, { "epoch": 0.10846871759419288, "grad_norm": 0.975696253331199, "learning_rate": 9.984654666833337e-06, "loss": 0.5268, "step": 1569 }, { "epoch": 0.1085378499827169, "grad_norm": 1.116105128823605, "learning_rate": 9.98461081422788e-06, "loss": 0.3107, "step": 1570 }, { "epoch": 0.10860698237124093, "grad_norm": 0.9214828085471422, "learning_rate": 9.98456689914922e-06, "loss": 0.383, "step": 1571 }, { "epoch": 0.10867611475976495, "grad_norm": 0.9188776295506299, "learning_rate": 9.98452292159791e-06, "loss": 0.5003, "step": 1572 }, { "epoch": 0.10874524714828897, "grad_norm": 0.908782258052724, "learning_rate": 9.984478881574499e-06, "loss": 0.4249, "step": 1573 }, { "epoch": 0.108814379536813, "grad_norm": 1.0068836900971654, "learning_rate": 9.984434779079545e-06, "loss": 0.3988, "step": 1574 }, { "epoch": 0.10888351192533702, "grad_norm": 0.9549056584280771, "learning_rate": 9.984390614113591e-06, "loss": 0.4042, "step": 1575 }, { "epoch": 0.10895264431386105, "grad_norm": 0.9416243068817045, "learning_rate": 9.984346386677198e-06, "loss": 0.4696, "step": 1576 }, { "epoch": 0.10902177670238507, "grad_norm": 1.0721393223454203, "learning_rate": 9.984302096770916e-06, "loss": 0.3391, "step": 1577 }, { "epoch": 0.10909090909090909, "grad_norm": 0.8611255916594527, "learning_rate": 9.984257744395305e-06, "loss": 0.3454, "step": 1578 }, { "epoch": 0.10916004147943312, "grad_norm": 0.8787557021576332, "learning_rate": 9.984213329550917e-06, "loss": 0.446, "step": 1579 }, { "epoch": 0.10922917386795714, "grad_norm": 1.0493845927783048, "learning_rate": 9.984168852238308e-06, "loss": 0.424, "step": 1580 }, { "epoch": 0.10929830625648117, "grad_norm": 0.8710360249708335, "learning_rate": 9.984124312458039e-06, "loss": 0.4256, "step": 1581 }, { "epoch": 0.10936743864500519, "grad_norm": 0.8995295454194074, "learning_rate": 9.984079710210666e-06, "loss": 0.372, "step": 1582 }, { "epoch": 0.1094365710335292, "grad_norm": 0.8937662005954916, "learning_rate": 9.984035045496747e-06, "loss": 0.4093, "step": 1583 }, { "epoch": 0.10950570342205324, "grad_norm": 0.8658943178841889, "learning_rate": 9.983990318316846e-06, "loss": 0.3077, "step": 1584 }, { "epoch": 0.10957483581057725, "grad_norm": 1.4812471039671373, "learning_rate": 9.983945528671517e-06, "loss": 0.3298, "step": 1585 }, { "epoch": 0.10964396819910129, "grad_norm": 0.9135249315601834, "learning_rate": 9.983900676561327e-06, "loss": 0.4045, "step": 1586 }, { "epoch": 0.1097131005876253, "grad_norm": 1.0411187144671337, "learning_rate": 9.983855761986836e-06, "loss": 0.3979, "step": 1587 }, { "epoch": 0.10978223297614932, "grad_norm": 0.9974811137487949, "learning_rate": 9.983810784948608e-06, "loss": 0.4818, "step": 1588 }, { "epoch": 0.10985136536467335, "grad_norm": 1.018460592385358, "learning_rate": 9.983765745447205e-06, "loss": 0.3782, "step": 1589 }, { "epoch": 0.10992049775319737, "grad_norm": 0.9714763874745519, "learning_rate": 9.983720643483192e-06, "loss": 0.4496, "step": 1590 }, { "epoch": 0.1099896301417214, "grad_norm": 0.8899131284768628, "learning_rate": 9.983675479057134e-06, "loss": 0.3983, "step": 1591 }, { "epoch": 0.11005876253024542, "grad_norm": 1.0770501680184317, "learning_rate": 9.9836302521696e-06, "loss": 0.3862, "step": 1592 }, { "epoch": 0.11012789491876944, "grad_norm": 0.9089213661129196, "learning_rate": 9.983584962821152e-06, "loss": 0.3894, "step": 1593 }, { "epoch": 0.11019702730729347, "grad_norm": 1.027111607060494, "learning_rate": 9.98353961101236e-06, "loss": 0.4109, "step": 1594 }, { "epoch": 0.11026615969581749, "grad_norm": 0.9185007735831285, "learning_rate": 9.983494196743796e-06, "loss": 0.3792, "step": 1595 }, { "epoch": 0.11033529208434151, "grad_norm": 0.8614854826056255, "learning_rate": 9.983448720016023e-06, "loss": 0.3757, "step": 1596 }, { "epoch": 0.11040442447286554, "grad_norm": 0.998813631278792, "learning_rate": 9.983403180829613e-06, "loss": 0.4328, "step": 1597 }, { "epoch": 0.11047355686138956, "grad_norm": 1.3088597629076801, "learning_rate": 9.983357579185138e-06, "loss": 0.4003, "step": 1598 }, { "epoch": 0.11054268924991359, "grad_norm": 0.930332631063675, "learning_rate": 9.98331191508317e-06, "loss": 0.3139, "step": 1599 }, { "epoch": 0.11061182163843761, "grad_norm": 1.060005833899148, "learning_rate": 9.983266188524279e-06, "loss": 0.3772, "step": 1600 }, { "epoch": 0.11068095402696163, "grad_norm": 1.0145031159214983, "learning_rate": 9.98322039950904e-06, "loss": 0.4752, "step": 1601 }, { "epoch": 0.11075008641548566, "grad_norm": 0.8380463898369048, "learning_rate": 9.983174548038027e-06, "loss": 0.3768, "step": 1602 }, { "epoch": 0.11081921880400968, "grad_norm": 1.229540022636904, "learning_rate": 9.983128634111811e-06, "loss": 0.3845, "step": 1603 }, { "epoch": 0.11088835119253371, "grad_norm": 1.0177850947801825, "learning_rate": 9.983082657730974e-06, "loss": 0.4897, "step": 1604 }, { "epoch": 0.11095748358105773, "grad_norm": 0.9569104429485814, "learning_rate": 9.983036618896088e-06, "loss": 0.4146, "step": 1605 }, { "epoch": 0.11102661596958174, "grad_norm": 1.0248492685702058, "learning_rate": 9.98299051760773e-06, "loss": 0.3672, "step": 1606 }, { "epoch": 0.11109574835810578, "grad_norm": 0.8276254419269184, "learning_rate": 9.982944353866477e-06, "loss": 0.3342, "step": 1607 }, { "epoch": 0.1111648807466298, "grad_norm": 0.9619274329283511, "learning_rate": 9.982898127672909e-06, "loss": 0.4636, "step": 1608 }, { "epoch": 0.11123401313515383, "grad_norm": 0.8780894253825459, "learning_rate": 9.982851839027608e-06, "loss": 0.3405, "step": 1609 }, { "epoch": 0.11130314552367784, "grad_norm": 0.9560335808405859, "learning_rate": 9.982805487931148e-06, "loss": 0.4158, "step": 1610 }, { "epoch": 0.11137227791220186, "grad_norm": 1.0036826633882348, "learning_rate": 9.982759074384116e-06, "loss": 0.3964, "step": 1611 }, { "epoch": 0.1114414103007259, "grad_norm": 0.9143047799732941, "learning_rate": 9.98271259838709e-06, "loss": 0.4315, "step": 1612 }, { "epoch": 0.11151054268924991, "grad_norm": 0.9158939841801406, "learning_rate": 9.982666059940655e-06, "loss": 0.3926, "step": 1613 }, { "epoch": 0.11157967507777394, "grad_norm": 0.9903606591736812, "learning_rate": 9.982619459045391e-06, "loss": 0.3706, "step": 1614 }, { "epoch": 0.11164880746629796, "grad_norm": 0.8759355276976335, "learning_rate": 9.982572795701887e-06, "loss": 0.4006, "step": 1615 }, { "epoch": 0.11171793985482198, "grad_norm": 0.8445442690359368, "learning_rate": 9.98252606991072e-06, "loss": 0.4255, "step": 1616 }, { "epoch": 0.11178707224334601, "grad_norm": 0.9373600334259984, "learning_rate": 9.982479281672484e-06, "loss": 0.3698, "step": 1617 }, { "epoch": 0.11185620463187003, "grad_norm": 0.885726181176253, "learning_rate": 9.98243243098776e-06, "loss": 0.4337, "step": 1618 }, { "epoch": 0.11192533702039406, "grad_norm": 0.9697790785088346, "learning_rate": 9.98238551785714e-06, "loss": 0.483, "step": 1619 }, { "epoch": 0.11199446940891808, "grad_norm": 0.8775170763591957, "learning_rate": 9.982338542281208e-06, "loss": 0.3343, "step": 1620 }, { "epoch": 0.1120636017974421, "grad_norm": 0.9864249558579214, "learning_rate": 9.982291504260551e-06, "loss": 0.345, "step": 1621 }, { "epoch": 0.11213273418596613, "grad_norm": 1.0199320571710717, "learning_rate": 9.982244403795764e-06, "loss": 0.4507, "step": 1622 }, { "epoch": 0.11220186657449015, "grad_norm": 1.1361381492692133, "learning_rate": 9.982197240887433e-06, "loss": 0.4004, "step": 1623 }, { "epoch": 0.11227099896301417, "grad_norm": 0.9356867852321239, "learning_rate": 9.982150015536152e-06, "loss": 0.3437, "step": 1624 }, { "epoch": 0.1123401313515382, "grad_norm": 1.342795762623547, "learning_rate": 9.982102727742512e-06, "loss": 0.411, "step": 1625 }, { "epoch": 0.11240926374006222, "grad_norm": 0.8639182271105721, "learning_rate": 9.982055377507103e-06, "loss": 0.4681, "step": 1626 }, { "epoch": 0.11247839612858625, "grad_norm": 0.8501090690095828, "learning_rate": 9.982007964830522e-06, "loss": 0.4262, "step": 1627 }, { "epoch": 0.11254752851711027, "grad_norm": 0.8554014143028618, "learning_rate": 9.981960489713361e-06, "loss": 0.3141, "step": 1628 }, { "epoch": 0.11261666090563428, "grad_norm": 0.9031396842970396, "learning_rate": 9.981912952156217e-06, "loss": 0.4056, "step": 1629 }, { "epoch": 0.11268579329415832, "grad_norm": 1.0650218705799126, "learning_rate": 9.981865352159685e-06, "loss": 0.4172, "step": 1630 }, { "epoch": 0.11275492568268233, "grad_norm": 0.8383861940337791, "learning_rate": 9.981817689724361e-06, "loss": 0.3485, "step": 1631 }, { "epoch": 0.11282405807120637, "grad_norm": 0.8296042487205494, "learning_rate": 9.981769964850843e-06, "loss": 0.3237, "step": 1632 }, { "epoch": 0.11289319045973038, "grad_norm": 0.9778457536307453, "learning_rate": 9.981722177539728e-06, "loss": 0.4475, "step": 1633 }, { "epoch": 0.1129623228482544, "grad_norm": 0.9296067178535318, "learning_rate": 9.981674327791617e-06, "loss": 0.3332, "step": 1634 }, { "epoch": 0.11303145523677843, "grad_norm": 0.8470673877951747, "learning_rate": 9.98162641560711e-06, "loss": 0.3608, "step": 1635 }, { "epoch": 0.11310058762530245, "grad_norm": 0.9217013112397412, "learning_rate": 9.981578440986802e-06, "loss": 0.4802, "step": 1636 }, { "epoch": 0.11316972001382648, "grad_norm": 0.8717584291762497, "learning_rate": 9.9815304039313e-06, "loss": 0.4582, "step": 1637 }, { "epoch": 0.1132388524023505, "grad_norm": 0.9228819446057691, "learning_rate": 9.981482304441205e-06, "loss": 0.3672, "step": 1638 }, { "epoch": 0.11330798479087452, "grad_norm": 0.9177145743854798, "learning_rate": 9.981434142517117e-06, "loss": 0.3157, "step": 1639 }, { "epoch": 0.11337711717939855, "grad_norm": 0.9718488171905503, "learning_rate": 9.981385918159645e-06, "loss": 0.5147, "step": 1640 }, { "epoch": 0.11344624956792257, "grad_norm": 1.5577768459169294, "learning_rate": 9.981337631369387e-06, "loss": 0.3573, "step": 1641 }, { "epoch": 0.1135153819564466, "grad_norm": 0.9269454664099619, "learning_rate": 9.981289282146955e-06, "loss": 0.3475, "step": 1642 }, { "epoch": 0.11358451434497062, "grad_norm": 1.2561502140958853, "learning_rate": 9.981240870492949e-06, "loss": 0.3657, "step": 1643 }, { "epoch": 0.11365364673349464, "grad_norm": 0.9363495521443147, "learning_rate": 9.981192396407979e-06, "loss": 0.4435, "step": 1644 }, { "epoch": 0.11372277912201867, "grad_norm": 0.9162828169999192, "learning_rate": 9.98114385989265e-06, "loss": 0.3445, "step": 1645 }, { "epoch": 0.11379191151054269, "grad_norm": 0.9092369885553242, "learning_rate": 9.981095260947573e-06, "loss": 0.3136, "step": 1646 }, { "epoch": 0.11386104389906672, "grad_norm": 2.473205439409222, "learning_rate": 9.981046599573357e-06, "loss": 0.4256, "step": 1647 }, { "epoch": 0.11393017628759074, "grad_norm": 0.9071994179683871, "learning_rate": 9.98099787577061e-06, "loss": 0.3313, "step": 1648 }, { "epoch": 0.11399930867611476, "grad_norm": 0.7692509451235044, "learning_rate": 9.980949089539945e-06, "loss": 0.298, "step": 1649 }, { "epoch": 0.11406844106463879, "grad_norm": 1.0435595185469588, "learning_rate": 9.98090024088197e-06, "loss": 0.437, "step": 1650 }, { "epoch": 0.1141375734531628, "grad_norm": 0.9632857384579859, "learning_rate": 9.9808513297973e-06, "loss": 0.4788, "step": 1651 }, { "epoch": 0.11420670584168684, "grad_norm": 1.0472113401052445, "learning_rate": 9.980802356286547e-06, "loss": 0.3925, "step": 1652 }, { "epoch": 0.11427583823021085, "grad_norm": 1.090412047270523, "learning_rate": 9.980753320350325e-06, "loss": 0.3941, "step": 1653 }, { "epoch": 0.11434497061873487, "grad_norm": 0.9714389647113015, "learning_rate": 9.98070422198925e-06, "loss": 0.3831, "step": 1654 }, { "epoch": 0.1144141030072589, "grad_norm": 1.0286847968977784, "learning_rate": 9.980655061203934e-06, "loss": 0.452, "step": 1655 }, { "epoch": 0.11448323539578292, "grad_norm": 1.1019076224299458, "learning_rate": 9.980605837994998e-06, "loss": 0.3109, "step": 1656 }, { "epoch": 0.11455236778430694, "grad_norm": 1.32132556655911, "learning_rate": 9.980556552363055e-06, "loss": 0.3687, "step": 1657 }, { "epoch": 0.11462150017283097, "grad_norm": 0.9465152189426508, "learning_rate": 9.980507204308722e-06, "loss": 0.4991, "step": 1658 }, { "epoch": 0.11469063256135499, "grad_norm": 1.1770644019008163, "learning_rate": 9.980457793832622e-06, "loss": 0.3427, "step": 1659 }, { "epoch": 0.11475976494987902, "grad_norm": 0.8667507450424412, "learning_rate": 9.980408320935368e-06, "loss": 0.3109, "step": 1660 }, { "epoch": 0.11482889733840304, "grad_norm": 0.9296063009534227, "learning_rate": 9.980358785617587e-06, "loss": 0.3975, "step": 1661 }, { "epoch": 0.11489802972692706, "grad_norm": 0.8573633694918432, "learning_rate": 9.980309187879895e-06, "loss": 0.404, "step": 1662 }, { "epoch": 0.11496716211545109, "grad_norm": 0.8945562117781951, "learning_rate": 9.980259527722915e-06, "loss": 0.3204, "step": 1663 }, { "epoch": 0.11503629450397511, "grad_norm": 1.1230837308771737, "learning_rate": 9.980209805147272e-06, "loss": 0.3242, "step": 1664 }, { "epoch": 0.11510542689249914, "grad_norm": 0.961231306101415, "learning_rate": 9.980160020153584e-06, "loss": 0.4311, "step": 1665 }, { "epoch": 0.11517455928102316, "grad_norm": 0.9088408600184229, "learning_rate": 9.980110172742478e-06, "loss": 0.3774, "step": 1666 }, { "epoch": 0.11524369166954718, "grad_norm": 0.9536331654228483, "learning_rate": 9.98006026291458e-06, "loss": 0.3458, "step": 1667 }, { "epoch": 0.11531282405807121, "grad_norm": 0.9438453422993882, "learning_rate": 9.980010290670511e-06, "loss": 0.4278, "step": 1668 }, { "epoch": 0.11538195644659523, "grad_norm": 0.8894896072003806, "learning_rate": 9.979960256010902e-06, "loss": 0.3944, "step": 1669 }, { "epoch": 0.11545108883511926, "grad_norm": 0.9424096529576427, "learning_rate": 9.97991015893638e-06, "loss": 0.3409, "step": 1670 }, { "epoch": 0.11552022122364328, "grad_norm": 0.8938790437477919, "learning_rate": 9.979859999447568e-06, "loss": 0.3856, "step": 1671 }, { "epoch": 0.1155893536121673, "grad_norm": 0.9245584015760226, "learning_rate": 9.979809777545099e-06, "loss": 0.4546, "step": 1672 }, { "epoch": 0.11565848600069133, "grad_norm": 0.9322736763621247, "learning_rate": 9.979759493229601e-06, "loss": 0.4239, "step": 1673 }, { "epoch": 0.11572761838921534, "grad_norm": 2.277769217339253, "learning_rate": 9.979709146501706e-06, "loss": 0.3558, "step": 1674 }, { "epoch": 0.11579675077773938, "grad_norm": 1.01975191596097, "learning_rate": 9.97965873736204e-06, "loss": 0.4035, "step": 1675 }, { "epoch": 0.1158658831662634, "grad_norm": 0.9385883770962156, "learning_rate": 9.979608265811243e-06, "loss": 0.5327, "step": 1676 }, { "epoch": 0.11593501555478741, "grad_norm": 1.0174192391275996, "learning_rate": 9.97955773184994e-06, "loss": 0.4218, "step": 1677 }, { "epoch": 0.11600414794331144, "grad_norm": 0.9319017296143122, "learning_rate": 9.979507135478768e-06, "loss": 0.3467, "step": 1678 }, { "epoch": 0.11607328033183546, "grad_norm": 1.05582363890358, "learning_rate": 9.979456476698362e-06, "loss": 0.4421, "step": 1679 }, { "epoch": 0.1161424127203595, "grad_norm": 0.9480734939546605, "learning_rate": 9.979405755509353e-06, "loss": 0.3938, "step": 1680 }, { "epoch": 0.11621154510888351, "grad_norm": 1.3841980597331345, "learning_rate": 9.979354971912379e-06, "loss": 0.3761, "step": 1681 }, { "epoch": 0.11628067749740753, "grad_norm": 0.9107208813348349, "learning_rate": 9.979304125908079e-06, "loss": 0.3734, "step": 1682 }, { "epoch": 0.11634980988593156, "grad_norm": 1.3085790541049567, "learning_rate": 9.979253217497084e-06, "loss": 0.4969, "step": 1683 }, { "epoch": 0.11641894227445558, "grad_norm": 0.8897757633022904, "learning_rate": 9.979202246680038e-06, "loss": 0.3463, "step": 1684 }, { "epoch": 0.11648807466297961, "grad_norm": 0.9315631701159501, "learning_rate": 9.979151213457578e-06, "loss": 0.3734, "step": 1685 }, { "epoch": 0.11655720705150363, "grad_norm": 0.9287023792356816, "learning_rate": 9.979100117830342e-06, "loss": 0.4121, "step": 1686 }, { "epoch": 0.11662633944002765, "grad_norm": 1.0170617024676196, "learning_rate": 9.979048959798974e-06, "loss": 0.4222, "step": 1687 }, { "epoch": 0.11669547182855168, "grad_norm": 1.1298166572616724, "learning_rate": 9.97899773936411e-06, "loss": 0.3897, "step": 1688 }, { "epoch": 0.1167646042170757, "grad_norm": 1.0047214845151962, "learning_rate": 9.978946456526396e-06, "loss": 0.3754, "step": 1689 }, { "epoch": 0.11683373660559972, "grad_norm": 0.9598295436435512, "learning_rate": 9.978895111286473e-06, "loss": 0.4022, "step": 1690 }, { "epoch": 0.11690286899412375, "grad_norm": 1.0164076723770348, "learning_rate": 9.978843703644987e-06, "loss": 0.3817, "step": 1691 }, { "epoch": 0.11697200138264777, "grad_norm": 1.0824651060337473, "learning_rate": 9.978792233602576e-06, "loss": 0.3375, "step": 1692 }, { "epoch": 0.1170411337711718, "grad_norm": 1.1463669047745813, "learning_rate": 9.978740701159894e-06, "loss": 0.4326, "step": 1693 }, { "epoch": 0.11711026615969582, "grad_norm": 1.014182724135384, "learning_rate": 9.978689106317581e-06, "loss": 0.4337, "step": 1694 }, { "epoch": 0.11717939854821983, "grad_norm": 0.858560594614022, "learning_rate": 9.978637449076283e-06, "loss": 0.3847, "step": 1695 }, { "epoch": 0.11724853093674387, "grad_norm": 0.8962375594654122, "learning_rate": 9.978585729436651e-06, "loss": 0.3429, "step": 1696 }, { "epoch": 0.11731766332526788, "grad_norm": 0.8656343926917511, "learning_rate": 9.978533947399331e-06, "loss": 0.4592, "step": 1697 }, { "epoch": 0.11738679571379192, "grad_norm": 0.8926700831477667, "learning_rate": 9.978482102964973e-06, "loss": 0.3633, "step": 1698 }, { "epoch": 0.11745592810231593, "grad_norm": 0.9755831390176962, "learning_rate": 9.978430196134226e-06, "loss": 0.3879, "step": 1699 }, { "epoch": 0.11752506049083995, "grad_norm": 0.9475627516197552, "learning_rate": 9.978378226907742e-06, "loss": 0.4487, "step": 1700 }, { "epoch": 0.11759419287936398, "grad_norm": 0.8558017490331041, "learning_rate": 9.97832619528617e-06, "loss": 0.4666, "step": 1701 }, { "epoch": 0.117663325267888, "grad_norm": 0.9590166132819881, "learning_rate": 9.978274101270163e-06, "loss": 0.3218, "step": 1702 }, { "epoch": 0.11773245765641203, "grad_norm": 0.9101250296303559, "learning_rate": 9.978221944860375e-06, "loss": 0.3461, "step": 1703 }, { "epoch": 0.11780159004493605, "grad_norm": 0.8996887702619882, "learning_rate": 9.978169726057458e-06, "loss": 0.4292, "step": 1704 }, { "epoch": 0.11787072243346007, "grad_norm": 0.9022876381693057, "learning_rate": 9.978117444862068e-06, "loss": 0.3559, "step": 1705 }, { "epoch": 0.1179398548219841, "grad_norm": 0.9613108471429932, "learning_rate": 9.97806510127486e-06, "loss": 0.3649, "step": 1706 }, { "epoch": 0.11800898721050812, "grad_norm": 0.8848123448602145, "learning_rate": 9.978012695296491e-06, "loss": 0.3615, "step": 1707 }, { "epoch": 0.11807811959903215, "grad_norm": 0.8223054161894406, "learning_rate": 9.977960226927615e-06, "loss": 0.4495, "step": 1708 }, { "epoch": 0.11814725198755617, "grad_norm": 1.1010099879900834, "learning_rate": 9.977907696168892e-06, "loss": 0.3432, "step": 1709 }, { "epoch": 0.11821638437608019, "grad_norm": 0.8758623641201319, "learning_rate": 9.977855103020977e-06, "loss": 0.3208, "step": 1710 }, { "epoch": 0.11828551676460422, "grad_norm": 0.9711379238603557, "learning_rate": 9.977802447484533e-06, "loss": 0.485, "step": 1711 }, { "epoch": 0.11835464915312824, "grad_norm": 0.8574877825796935, "learning_rate": 9.97774972956022e-06, "loss": 0.4126, "step": 1712 }, { "epoch": 0.11842378154165227, "grad_norm": 0.9431387580219669, "learning_rate": 9.977696949248696e-06, "loss": 0.3816, "step": 1713 }, { "epoch": 0.11849291393017629, "grad_norm": 0.9586863329190103, "learning_rate": 9.977644106550625e-06, "loss": 0.3062, "step": 1714 }, { "epoch": 0.1185620463187003, "grad_norm": 0.9700055282895383, "learning_rate": 9.977591201466667e-06, "loss": 0.4896, "step": 1715 }, { "epoch": 0.11863117870722434, "grad_norm": 0.8596917219619251, "learning_rate": 9.977538233997486e-06, "loss": 0.3384, "step": 1716 }, { "epoch": 0.11870031109574836, "grad_norm": 0.8449462385501583, "learning_rate": 9.977485204143746e-06, "loss": 0.3218, "step": 1717 }, { "epoch": 0.11876944348427239, "grad_norm": 1.1390172661944757, "learning_rate": 9.977432111906113e-06, "loss": 0.362, "step": 1718 }, { "epoch": 0.1188385758727964, "grad_norm": 0.9647833631567607, "learning_rate": 9.977378957285251e-06, "loss": 0.512, "step": 1719 }, { "epoch": 0.11890770826132042, "grad_norm": 0.9160065363331463, "learning_rate": 9.977325740281826e-06, "loss": 0.3127, "step": 1720 }, { "epoch": 0.11897684064984446, "grad_norm": 1.004217949119672, "learning_rate": 9.977272460896505e-06, "loss": 0.3732, "step": 1721 }, { "epoch": 0.11904597303836847, "grad_norm": 0.8544172482830258, "learning_rate": 9.977219119129955e-06, "loss": 0.4867, "step": 1722 }, { "epoch": 0.11911510542689249, "grad_norm": 0.9064669074260577, "learning_rate": 9.977165714982848e-06, "loss": 0.3583, "step": 1723 }, { "epoch": 0.11918423781541652, "grad_norm": 0.9335708118228503, "learning_rate": 9.977112248455849e-06, "loss": 0.3664, "step": 1724 }, { "epoch": 0.11925337020394054, "grad_norm": 0.8519679400377871, "learning_rate": 9.97705871954963e-06, "loss": 0.3984, "step": 1725 }, { "epoch": 0.11932250259246457, "grad_norm": 0.9038042019721476, "learning_rate": 9.977005128264862e-06, "loss": 0.4519, "step": 1726 }, { "epoch": 0.11939163498098859, "grad_norm": 0.9223428798937539, "learning_rate": 9.976951474602217e-06, "loss": 0.3548, "step": 1727 }, { "epoch": 0.11946076736951261, "grad_norm": 0.9117446115455996, "learning_rate": 9.976897758562368e-06, "loss": 0.3539, "step": 1728 }, { "epoch": 0.11952989975803664, "grad_norm": 1.2516297285098852, "learning_rate": 9.976843980145988e-06, "loss": 0.4499, "step": 1729 }, { "epoch": 0.11959903214656066, "grad_norm": 0.939201631103719, "learning_rate": 9.976790139353747e-06, "loss": 0.4362, "step": 1730 }, { "epoch": 0.11966816453508469, "grad_norm": 0.8367247538146013, "learning_rate": 9.976736236186324e-06, "loss": 0.3149, "step": 1731 }, { "epoch": 0.11973729692360871, "grad_norm": 0.8052472557745961, "learning_rate": 9.976682270644396e-06, "loss": 0.3683, "step": 1732 }, { "epoch": 0.11980642931213273, "grad_norm": 0.8224329419644614, "learning_rate": 9.976628242728635e-06, "loss": 0.4336, "step": 1733 }, { "epoch": 0.11987556170065676, "grad_norm": 0.9637746848340459, "learning_rate": 9.976574152439719e-06, "loss": 0.3333, "step": 1734 }, { "epoch": 0.11994469408918078, "grad_norm": 0.9479832763248518, "learning_rate": 9.97651999977833e-06, "loss": 0.3644, "step": 1735 }, { "epoch": 0.12001382647770481, "grad_norm": 1.5943104535537829, "learning_rate": 9.976465784745142e-06, "loss": 0.4073, "step": 1736 }, { "epoch": 0.12008295886622883, "grad_norm": 0.8749276892657729, "learning_rate": 9.976411507340838e-06, "loss": 0.4344, "step": 1737 }, { "epoch": 0.12015209125475285, "grad_norm": 1.0089675006245387, "learning_rate": 9.976357167566096e-06, "loss": 0.3645, "step": 1738 }, { "epoch": 0.12022122364327688, "grad_norm": 0.9314267917254354, "learning_rate": 9.976302765421596e-06, "loss": 0.4779, "step": 1739 }, { "epoch": 0.1202903560318009, "grad_norm": 1.5047941792528157, "learning_rate": 9.976248300908022e-06, "loss": 0.4866, "step": 1740 }, { "epoch": 0.12035948842032493, "grad_norm": 0.8374913648482847, "learning_rate": 9.976193774026056e-06, "loss": 0.3521, "step": 1741 }, { "epoch": 0.12042862080884895, "grad_norm": 0.8433930713417774, "learning_rate": 9.976139184776383e-06, "loss": 0.3439, "step": 1742 }, { "epoch": 0.12049775319737296, "grad_norm": 1.037921119983377, "learning_rate": 9.976084533159685e-06, "loss": 0.3718, "step": 1743 }, { "epoch": 0.120566885585897, "grad_norm": 1.0350961464998112, "learning_rate": 9.976029819176646e-06, "loss": 0.4676, "step": 1744 }, { "epoch": 0.12063601797442101, "grad_norm": 1.0817375721775897, "learning_rate": 9.975975042827955e-06, "loss": 0.3247, "step": 1745 }, { "epoch": 0.12070515036294505, "grad_norm": 0.869481681819335, "learning_rate": 9.975920204114295e-06, "loss": 0.3457, "step": 1746 }, { "epoch": 0.12077428275146906, "grad_norm": 0.8651895329890907, "learning_rate": 9.975865303036357e-06, "loss": 0.4345, "step": 1747 }, { "epoch": 0.12084341513999308, "grad_norm": 0.9033465709617848, "learning_rate": 9.975810339594827e-06, "loss": 0.4205, "step": 1748 }, { "epoch": 0.12091254752851711, "grad_norm": 1.1546248786709692, "learning_rate": 9.975755313790395e-06, "loss": 0.3433, "step": 1749 }, { "epoch": 0.12098167991704113, "grad_norm": 0.9208473396008747, "learning_rate": 9.975700225623748e-06, "loss": 0.3959, "step": 1750 }, { "epoch": 0.12105081230556516, "grad_norm": 0.8296990933052102, "learning_rate": 9.97564507509558e-06, "loss": 0.4356, "step": 1751 }, { "epoch": 0.12111994469408918, "grad_norm": 0.8559577917924861, "learning_rate": 9.975589862206578e-06, "loss": 0.3265, "step": 1752 }, { "epoch": 0.1211890770826132, "grad_norm": 0.850950963555227, "learning_rate": 9.975534586957438e-06, "loss": 0.3055, "step": 1753 }, { "epoch": 0.12125820947113723, "grad_norm": 0.9917865279957038, "learning_rate": 9.975479249348851e-06, "loss": 0.4877, "step": 1754 }, { "epoch": 0.12132734185966125, "grad_norm": 0.9056557814833631, "learning_rate": 9.975423849381511e-06, "loss": 0.4161, "step": 1755 }, { "epoch": 0.12139647424818527, "grad_norm": 0.8823677418007028, "learning_rate": 9.975368387056112e-06, "loss": 0.3082, "step": 1756 }, { "epoch": 0.1214656066367093, "grad_norm": 0.9898120887279593, "learning_rate": 9.975312862373348e-06, "loss": 0.4352, "step": 1757 }, { "epoch": 0.12153473902523332, "grad_norm": 0.9095028520444498, "learning_rate": 9.975257275333917e-06, "loss": 0.498, "step": 1758 }, { "epoch": 0.12160387141375735, "grad_norm": 1.1651220135270297, "learning_rate": 9.975201625938515e-06, "loss": 0.3594, "step": 1759 }, { "epoch": 0.12167300380228137, "grad_norm": 1.5019656175658718, "learning_rate": 9.975145914187839e-06, "loss": 0.3203, "step": 1760 }, { "epoch": 0.12174213619080539, "grad_norm": 0.8403616259406227, "learning_rate": 9.975090140082588e-06, "loss": 0.4154, "step": 1761 }, { "epoch": 0.12181126857932942, "grad_norm": 0.9957478720309958, "learning_rate": 9.975034303623461e-06, "loss": 0.4279, "step": 1762 }, { "epoch": 0.12188040096785344, "grad_norm": 0.8129962692971706, "learning_rate": 9.974978404811155e-06, "loss": 0.3766, "step": 1763 }, { "epoch": 0.12194953335637747, "grad_norm": 0.9425858318439596, "learning_rate": 9.974922443646376e-06, "loss": 0.3151, "step": 1764 }, { "epoch": 0.12201866574490149, "grad_norm": 0.9229733680136761, "learning_rate": 9.974866420129821e-06, "loss": 0.4339, "step": 1765 }, { "epoch": 0.1220877981334255, "grad_norm": 1.0289965491039612, "learning_rate": 9.974810334262193e-06, "loss": 0.3795, "step": 1766 }, { "epoch": 0.12215693052194954, "grad_norm": 1.4171246345019126, "learning_rate": 9.974754186044196e-06, "loss": 0.4073, "step": 1767 }, { "epoch": 0.12222606291047355, "grad_norm": 0.9895819620695436, "learning_rate": 9.974697975476533e-06, "loss": 0.487, "step": 1768 }, { "epoch": 0.12229519529899759, "grad_norm": 1.1345781374448398, "learning_rate": 9.974641702559907e-06, "loss": 0.4685, "step": 1769 }, { "epoch": 0.1223643276875216, "grad_norm": 2.320735837696488, "learning_rate": 9.974585367295026e-06, "loss": 0.4608, "step": 1770 }, { "epoch": 0.12243346007604562, "grad_norm": 0.9229981023945129, "learning_rate": 9.974528969682597e-06, "loss": 0.3721, "step": 1771 }, { "epoch": 0.12250259246456965, "grad_norm": 0.9056983667499056, "learning_rate": 9.974472509723324e-06, "loss": 0.4721, "step": 1772 }, { "epoch": 0.12257172485309367, "grad_norm": 1.077499140885208, "learning_rate": 9.974415987417914e-06, "loss": 0.4499, "step": 1773 }, { "epoch": 0.1226408572416177, "grad_norm": 1.2697823647983644, "learning_rate": 9.974359402767077e-06, "loss": 0.3127, "step": 1774 }, { "epoch": 0.12270998963014172, "grad_norm": 0.8795041249494328, "learning_rate": 9.974302755771521e-06, "loss": 0.404, "step": 1775 }, { "epoch": 0.12277912201866574, "grad_norm": 1.0105367498702031, "learning_rate": 9.974246046431959e-06, "loss": 0.4357, "step": 1776 }, { "epoch": 0.12284825440718977, "grad_norm": 1.6792372809945129, "learning_rate": 9.974189274749099e-06, "loss": 0.4356, "step": 1777 }, { "epoch": 0.12291738679571379, "grad_norm": 1.1327076057615404, "learning_rate": 9.974132440723654e-06, "loss": 0.3206, "step": 1778 }, { "epoch": 0.12298651918423782, "grad_norm": 1.3161649352179754, "learning_rate": 9.974075544356333e-06, "loss": 0.4806, "step": 1779 }, { "epoch": 0.12305565157276184, "grad_norm": 1.3982428015222828, "learning_rate": 9.974018585647855e-06, "loss": 0.4835, "step": 1780 }, { "epoch": 0.12312478396128586, "grad_norm": 0.8195469396407811, "learning_rate": 9.973961564598932e-06, "loss": 0.2942, "step": 1781 }, { "epoch": 0.12319391634980989, "grad_norm": 1.1210028216709902, "learning_rate": 9.973904481210273e-06, "loss": 0.3614, "step": 1782 }, { "epoch": 0.12326304873833391, "grad_norm": 1.2005589511583727, "learning_rate": 9.973847335482597e-06, "loss": 0.439, "step": 1783 }, { "epoch": 0.12333218112685794, "grad_norm": 1.0733421147283257, "learning_rate": 9.973790127416625e-06, "loss": 0.3499, "step": 1784 }, { "epoch": 0.12340131351538196, "grad_norm": 0.9657002625678374, "learning_rate": 9.973732857013067e-06, "loss": 0.341, "step": 1785 }, { "epoch": 0.12347044590390598, "grad_norm": 1.0244099302340952, "learning_rate": 9.973675524272646e-06, "loss": 0.3759, "step": 1786 }, { "epoch": 0.12353957829243001, "grad_norm": 1.6860347089099241, "learning_rate": 9.973618129196077e-06, "loss": 0.4044, "step": 1787 }, { "epoch": 0.12360871068095403, "grad_norm": 1.1295426679351257, "learning_rate": 9.97356067178408e-06, "loss": 0.3403, "step": 1788 }, { "epoch": 0.12367784306947804, "grad_norm": 0.965395635480987, "learning_rate": 9.973503152037376e-06, "loss": 0.4458, "step": 1789 }, { "epoch": 0.12374697545800208, "grad_norm": 0.968250103060812, "learning_rate": 9.973445569956686e-06, "loss": 0.457, "step": 1790 }, { "epoch": 0.1238161078465261, "grad_norm": 1.2207540345922139, "learning_rate": 9.97338792554273e-06, "loss": 0.3478, "step": 1791 }, { "epoch": 0.12388524023505013, "grad_norm": 0.8303605614384038, "learning_rate": 9.973330218796234e-06, "loss": 0.295, "step": 1792 }, { "epoch": 0.12395437262357414, "grad_norm": 0.8297858858845123, "learning_rate": 9.973272449717917e-06, "loss": 0.3759, "step": 1793 }, { "epoch": 0.12402350501209816, "grad_norm": 0.9521035875260878, "learning_rate": 9.973214618308506e-06, "loss": 0.4713, "step": 1794 }, { "epoch": 0.1240926374006222, "grad_norm": 0.9656223977303082, "learning_rate": 9.973156724568725e-06, "loss": 0.3903, "step": 1795 }, { "epoch": 0.12416176978914621, "grad_norm": 0.9121375708780994, "learning_rate": 9.973098768499299e-06, "loss": 0.3521, "step": 1796 }, { "epoch": 0.12423090217767024, "grad_norm": 0.8712763278208847, "learning_rate": 9.973040750100955e-06, "loss": 0.4849, "step": 1797 }, { "epoch": 0.12430003456619426, "grad_norm": 1.0698266943843295, "learning_rate": 9.97298266937442e-06, "loss": 0.3373, "step": 1798 }, { "epoch": 0.12436916695471828, "grad_norm": 0.9353027937569858, "learning_rate": 9.972924526320422e-06, "loss": 0.3584, "step": 1799 }, { "epoch": 0.12443829934324231, "grad_norm": 0.9835311682026829, "learning_rate": 9.972866320939689e-06, "loss": 0.4278, "step": 1800 }, { "epoch": 0.12450743173176633, "grad_norm": 0.8966858112967317, "learning_rate": 9.972808053232952e-06, "loss": 0.4914, "step": 1801 }, { "epoch": 0.12457656412029036, "grad_norm": 0.7611236402060657, "learning_rate": 9.972749723200941e-06, "loss": 0.3059, "step": 1802 }, { "epoch": 0.12464569650881438, "grad_norm": 0.91925188048905, "learning_rate": 9.972691330844383e-06, "loss": 0.3384, "step": 1803 }, { "epoch": 0.1247148288973384, "grad_norm": 0.9837079719433005, "learning_rate": 9.972632876164017e-06, "loss": 0.4544, "step": 1804 }, { "epoch": 0.12478396128586243, "grad_norm": 0.8324701438493127, "learning_rate": 9.972574359160571e-06, "loss": 0.3581, "step": 1805 }, { "epoch": 0.12485309367438645, "grad_norm": 0.9012374875061788, "learning_rate": 9.97251577983478e-06, "loss": 0.3856, "step": 1806 }, { "epoch": 0.12492222606291048, "grad_norm": 1.0895918562704898, "learning_rate": 9.972457138187376e-06, "loss": 0.399, "step": 1807 }, { "epoch": 0.1249913584514345, "grad_norm": 0.9289416187455848, "learning_rate": 9.972398434219097e-06, "loss": 0.4484, "step": 1808 }, { "epoch": 0.12506049083995852, "grad_norm": 0.8687507355067519, "learning_rate": 9.972339667930678e-06, "loss": 0.3656, "step": 1809 }, { "epoch": 0.12512962322848253, "grad_norm": 0.914406529657489, "learning_rate": 9.972280839322852e-06, "loss": 0.3399, "step": 1810 }, { "epoch": 0.12519875561700658, "grad_norm": 0.90207393838854, "learning_rate": 9.972221948396362e-06, "loss": 0.4326, "step": 1811 }, { "epoch": 0.1252678880055306, "grad_norm": 0.857345716306426, "learning_rate": 9.972162995151939e-06, "loss": 0.444, "step": 1812 }, { "epoch": 0.12533702039405462, "grad_norm": 0.9359071473632669, "learning_rate": 9.97210397959033e-06, "loss": 0.3265, "step": 1813 }, { "epoch": 0.12540615278257863, "grad_norm": 0.828350957931349, "learning_rate": 9.97204490171227e-06, "loss": 0.3478, "step": 1814 }, { "epoch": 0.12547528517110265, "grad_norm": 0.8407779998066713, "learning_rate": 9.9719857615185e-06, "loss": 0.4658, "step": 1815 }, { "epoch": 0.1255444175596267, "grad_norm": 0.9676736135292116, "learning_rate": 9.971926559009761e-06, "loss": 0.4526, "step": 1816 }, { "epoch": 0.12561354994815072, "grad_norm": 0.8656073843149811, "learning_rate": 9.971867294186794e-06, "loss": 0.3803, "step": 1817 }, { "epoch": 0.12568268233667473, "grad_norm": 0.8368010542752313, "learning_rate": 9.971807967050346e-06, "loss": 0.4046, "step": 1818 }, { "epoch": 0.12575181472519875, "grad_norm": 0.8120097171448277, "learning_rate": 9.971748577601156e-06, "loss": 0.452, "step": 1819 }, { "epoch": 0.12582094711372277, "grad_norm": 0.7875456176117989, "learning_rate": 9.97168912583997e-06, "loss": 0.386, "step": 1820 }, { "epoch": 0.12589007950224682, "grad_norm": 0.8397499798108676, "learning_rate": 9.971629611767534e-06, "loss": 0.3246, "step": 1821 }, { "epoch": 0.12595921189077083, "grad_norm": 0.9047765736371755, "learning_rate": 9.971570035384592e-06, "loss": 0.4379, "step": 1822 }, { "epoch": 0.12602834427929485, "grad_norm": 0.84334024321577, "learning_rate": 9.971510396691892e-06, "loss": 0.3918, "step": 1823 }, { "epoch": 0.12609747666781887, "grad_norm": 0.8996938825057011, "learning_rate": 9.971450695690182e-06, "loss": 0.3978, "step": 1824 }, { "epoch": 0.1261666090563429, "grad_norm": 0.935845870357917, "learning_rate": 9.971390932380209e-06, "loss": 0.485, "step": 1825 }, { "epoch": 0.12623574144486693, "grad_norm": 0.8413010386291835, "learning_rate": 9.971331106762723e-06, "loss": 0.5139, "step": 1826 }, { "epoch": 0.12630487383339095, "grad_norm": 1.0266439217031118, "learning_rate": 9.971271218838475e-06, "loss": 0.3777, "step": 1827 }, { "epoch": 0.12637400622191497, "grad_norm": 0.8116276423339018, "learning_rate": 9.97121126860821e-06, "loss": 0.3651, "step": 1828 }, { "epoch": 0.126443138610439, "grad_norm": 0.9259483143969363, "learning_rate": 9.971151256072686e-06, "loss": 0.5379, "step": 1829 }, { "epoch": 0.126512270998963, "grad_norm": 0.8888478349309858, "learning_rate": 9.97109118123265e-06, "loss": 0.4108, "step": 1830 }, { "epoch": 0.12658140338748705, "grad_norm": 0.8960522381522782, "learning_rate": 9.97103104408886e-06, "loss": 0.387, "step": 1831 }, { "epoch": 0.12665053577601107, "grad_norm": 0.9215225112579266, "learning_rate": 9.970970844642066e-06, "loss": 0.3659, "step": 1832 }, { "epoch": 0.1267196681645351, "grad_norm": 0.9018542986837869, "learning_rate": 9.970910582893023e-06, "loss": 0.5038, "step": 1833 }, { "epoch": 0.1267888005530591, "grad_norm": 0.8613913528890004, "learning_rate": 9.970850258842489e-06, "loss": 0.3728, "step": 1834 }, { "epoch": 0.12685793294158312, "grad_norm": 0.7832378712052986, "learning_rate": 9.970789872491215e-06, "loss": 0.3215, "step": 1835 }, { "epoch": 0.12692706533010714, "grad_norm": 0.9389604038334757, "learning_rate": 9.970729423839963e-06, "loss": 0.4842, "step": 1836 }, { "epoch": 0.1269961977186312, "grad_norm": 1.0188956575219785, "learning_rate": 9.970668912889488e-06, "loss": 0.4029, "step": 1837 }, { "epoch": 0.1270653301071552, "grad_norm": 0.7809934097235178, "learning_rate": 9.970608339640546e-06, "loss": 0.3721, "step": 1838 }, { "epoch": 0.12713446249567922, "grad_norm": 0.8121481422947303, "learning_rate": 9.970547704093902e-06, "loss": 0.3394, "step": 1839 }, { "epoch": 0.12720359488420324, "grad_norm": 0.9283640600871248, "learning_rate": 9.970487006250312e-06, "loss": 0.5242, "step": 1840 }, { "epoch": 0.12727272727272726, "grad_norm": 0.823564255702894, "learning_rate": 9.970426246110537e-06, "loss": 0.3367, "step": 1841 }, { "epoch": 0.1273418596612513, "grad_norm": 0.8886763284510315, "learning_rate": 9.97036542367534e-06, "loss": 0.3641, "step": 1842 }, { "epoch": 0.12741099204977532, "grad_norm": 0.9070572671746165, "learning_rate": 9.970304538945483e-06, "loss": 0.3802, "step": 1843 }, { "epoch": 0.12748012443829934, "grad_norm": 0.9735484381441706, "learning_rate": 9.970243591921727e-06, "loss": 0.4251, "step": 1844 }, { "epoch": 0.12754925682682336, "grad_norm": 0.9680603954389561, "learning_rate": 9.970182582604837e-06, "loss": 0.3216, "step": 1845 }, { "epoch": 0.12761838921534738, "grad_norm": 0.8601704918896539, "learning_rate": 9.970121510995579e-06, "loss": 0.3193, "step": 1846 }, { "epoch": 0.12768752160387142, "grad_norm": 1.0822299083188534, "learning_rate": 9.970060377094718e-06, "loss": 0.4507, "step": 1847 }, { "epoch": 0.12775665399239544, "grad_norm": 0.8085769882702964, "learning_rate": 9.969999180903019e-06, "loss": 0.3436, "step": 1848 }, { "epoch": 0.12782578638091946, "grad_norm": 0.8801863934230059, "learning_rate": 9.969937922421248e-06, "loss": 0.3577, "step": 1849 }, { "epoch": 0.12789491876944348, "grad_norm": 0.8985265118508993, "learning_rate": 9.969876601650178e-06, "loss": 0.4168, "step": 1850 }, { "epoch": 0.1279640511579675, "grad_norm": 1.415480859474153, "learning_rate": 9.969815218590572e-06, "loss": 0.4646, "step": 1851 }, { "epoch": 0.12803318354649154, "grad_norm": 1.3855234803758432, "learning_rate": 9.9697537732432e-06, "loss": 0.3432, "step": 1852 }, { "epoch": 0.12810231593501556, "grad_norm": 1.2669960345116442, "learning_rate": 9.969692265608836e-06, "loss": 0.357, "step": 1853 }, { "epoch": 0.12817144832353958, "grad_norm": 0.8613555348458524, "learning_rate": 9.969630695688247e-06, "loss": 0.4107, "step": 1854 }, { "epoch": 0.1282405807120636, "grad_norm": 0.9466562870793759, "learning_rate": 9.969569063482204e-06, "loss": 0.428, "step": 1855 }, { "epoch": 0.1283097131005876, "grad_norm": 0.8032120418200185, "learning_rate": 9.969507368991485e-06, "loss": 0.341, "step": 1856 }, { "epoch": 0.12837884548911166, "grad_norm": 0.837120129843123, "learning_rate": 9.969445612216857e-06, "loss": 0.3832, "step": 1857 }, { "epoch": 0.12844797787763568, "grad_norm": 1.230249254358559, "learning_rate": 9.969383793159097e-06, "loss": 0.4642, "step": 1858 }, { "epoch": 0.1285171102661597, "grad_norm": 0.8776711977872376, "learning_rate": 9.969321911818981e-06, "loss": 0.3096, "step": 1859 }, { "epoch": 0.1285862426546837, "grad_norm": 0.8740091443435011, "learning_rate": 9.969259968197282e-06, "loss": 0.3393, "step": 1860 }, { "epoch": 0.12865537504320773, "grad_norm": 0.8667757179990057, "learning_rate": 9.969197962294776e-06, "loss": 0.3708, "step": 1861 }, { "epoch": 0.12872450743173178, "grad_norm": 0.9039151997905025, "learning_rate": 9.969135894112243e-06, "loss": 0.4734, "step": 1862 }, { "epoch": 0.1287936398202558, "grad_norm": 0.8694696788264041, "learning_rate": 9.96907376365046e-06, "loss": 0.321, "step": 1863 }, { "epoch": 0.1288627722087798, "grad_norm": 1.1790236300359027, "learning_rate": 9.969011570910204e-06, "loss": 0.3815, "step": 1864 }, { "epoch": 0.12893190459730383, "grad_norm": 1.0039314663930488, "learning_rate": 9.968949315892255e-06, "loss": 0.5115, "step": 1865 }, { "epoch": 0.12900103698582785, "grad_norm": 1.3244025448916377, "learning_rate": 9.968886998597396e-06, "loss": 0.3174, "step": 1866 }, { "epoch": 0.1290701693743519, "grad_norm": 0.7993152724356412, "learning_rate": 9.968824619026402e-06, "loss": 0.3648, "step": 1867 }, { "epoch": 0.1291393017628759, "grad_norm": 0.8807123633574184, "learning_rate": 9.968762177180063e-06, "loss": 0.4411, "step": 1868 }, { "epoch": 0.12920843415139993, "grad_norm": 0.8878810873973914, "learning_rate": 9.968699673059155e-06, "loss": 0.4578, "step": 1869 }, { "epoch": 0.12927756653992395, "grad_norm": 0.75136873326563, "learning_rate": 9.968637106664465e-06, "loss": 0.3324, "step": 1870 }, { "epoch": 0.12934669892844797, "grad_norm": 0.7571515931142342, "learning_rate": 9.968574477996772e-06, "loss": 0.3222, "step": 1871 }, { "epoch": 0.129415831316972, "grad_norm": 0.8249020139636631, "learning_rate": 9.96851178705687e-06, "loss": 0.4207, "step": 1872 }, { "epoch": 0.12948496370549603, "grad_norm": 1.215417261709829, "learning_rate": 9.968449033845536e-06, "loss": 0.4398, "step": 1873 }, { "epoch": 0.12955409609402005, "grad_norm": 0.9089291528157334, "learning_rate": 9.968386218363562e-06, "loss": 0.37, "step": 1874 }, { "epoch": 0.12962322848254407, "grad_norm": 0.9448151568695863, "learning_rate": 9.96832334061173e-06, "loss": 0.4005, "step": 1875 }, { "epoch": 0.12969236087106809, "grad_norm": 0.9974378342197463, "learning_rate": 9.968260400590835e-06, "loss": 0.4758, "step": 1876 }, { "epoch": 0.12976149325959213, "grad_norm": 1.335247658174191, "learning_rate": 9.96819739830166e-06, "loss": 0.3085, "step": 1877 }, { "epoch": 0.12983062564811615, "grad_norm": 0.9832051480200356, "learning_rate": 9.968134333744996e-06, "loss": 0.3712, "step": 1878 }, { "epoch": 0.12989975803664017, "grad_norm": 0.9290095911341698, "learning_rate": 9.968071206921637e-06, "loss": 0.4806, "step": 1879 }, { "epoch": 0.12996889042516419, "grad_norm": 0.7781990441969703, "learning_rate": 9.96800801783237e-06, "loss": 0.3716, "step": 1880 }, { "epoch": 0.1300380228136882, "grad_norm": 0.910322138702726, "learning_rate": 9.967944766477988e-06, "loss": 0.3149, "step": 1881 }, { "epoch": 0.13010715520221225, "grad_norm": 0.8700418949409253, "learning_rate": 9.967881452859284e-06, "loss": 0.3674, "step": 1882 }, { "epoch": 0.13017628759073627, "grad_norm": 0.8550146641901359, "learning_rate": 9.967818076977052e-06, "loss": 0.4644, "step": 1883 }, { "epoch": 0.13024541997926029, "grad_norm": 1.3594992641954768, "learning_rate": 9.967754638832085e-06, "loss": 0.3287, "step": 1884 }, { "epoch": 0.1303145523677843, "grad_norm": 0.8637239297318003, "learning_rate": 9.96769113842518e-06, "loss": 0.3277, "step": 1885 }, { "epoch": 0.13038368475630832, "grad_norm": 0.8985909166467357, "learning_rate": 9.96762757575713e-06, "loss": 0.3403, "step": 1886 }, { "epoch": 0.13045281714483237, "grad_norm": 0.940176918784045, "learning_rate": 9.967563950828735e-06, "loss": 0.4703, "step": 1887 }, { "epoch": 0.13052194953335639, "grad_norm": 0.8073151162423117, "learning_rate": 9.967500263640793e-06, "loss": 0.3212, "step": 1888 }, { "epoch": 0.1305910819218804, "grad_norm": 1.0032165740189012, "learning_rate": 9.967436514194097e-06, "loss": 0.3568, "step": 1889 }, { "epoch": 0.13066021431040442, "grad_norm": 0.8577989942845546, "learning_rate": 9.96737270248945e-06, "loss": 0.4756, "step": 1890 }, { "epoch": 0.13072934669892844, "grad_norm": 1.4062643084083537, "learning_rate": 9.967308828527651e-06, "loss": 0.3605, "step": 1891 }, { "epoch": 0.13079847908745248, "grad_norm": 0.8910139645983253, "learning_rate": 9.9672448923095e-06, "loss": 0.3064, "step": 1892 }, { "epoch": 0.1308676114759765, "grad_norm": 0.9534712906159105, "learning_rate": 9.967180893835799e-06, "loss": 0.4251, "step": 1893 }, { "epoch": 0.13093674386450052, "grad_norm": 0.953639923455527, "learning_rate": 9.967116833107348e-06, "loss": 0.4404, "step": 1894 }, { "epoch": 0.13100587625302454, "grad_norm": 0.8756048903879087, "learning_rate": 9.967052710124956e-06, "loss": 0.3235, "step": 1895 }, { "epoch": 0.13107500864154856, "grad_norm": 0.8929122232015807, "learning_rate": 9.966988524889418e-06, "loss": 0.3952, "step": 1896 }, { "epoch": 0.13114414103007258, "grad_norm": 0.9331746273495802, "learning_rate": 9.966924277401545e-06, "loss": 0.4531, "step": 1897 }, { "epoch": 0.13121327341859662, "grad_norm": 0.9491240883876428, "learning_rate": 9.96685996766214e-06, "loss": 0.4526, "step": 1898 }, { "epoch": 0.13128240580712064, "grad_norm": 1.0420392618776118, "learning_rate": 9.966795595672009e-06, "loss": 0.3723, "step": 1899 }, { "epoch": 0.13135153819564466, "grad_norm": 0.9040134349986709, "learning_rate": 9.966731161431958e-06, "loss": 0.4119, "step": 1900 }, { "epoch": 0.13142067058416868, "grad_norm": 0.9634385132275611, "learning_rate": 9.966666664942796e-06, "loss": 0.4297, "step": 1901 }, { "epoch": 0.1314898029726927, "grad_norm": 0.7937521065566135, "learning_rate": 9.966602106205332e-06, "loss": 0.2948, "step": 1902 }, { "epoch": 0.13155893536121674, "grad_norm": 1.158792767733434, "learning_rate": 9.966537485220372e-06, "loss": 0.2987, "step": 1903 }, { "epoch": 0.13162806774974076, "grad_norm": 0.9249478005306437, "learning_rate": 9.96647280198873e-06, "loss": 0.4424, "step": 1904 }, { "epoch": 0.13169720013826478, "grad_norm": 0.8959389190513081, "learning_rate": 9.966408056511213e-06, "loss": 0.4294, "step": 1905 }, { "epoch": 0.1317663325267888, "grad_norm": 0.7902689769642263, "learning_rate": 9.966343248788633e-06, "loss": 0.32, "step": 1906 }, { "epoch": 0.1318354649153128, "grad_norm": 0.9670600117026507, "learning_rate": 9.966278378821804e-06, "loss": 0.4233, "step": 1907 }, { "epoch": 0.13190459730383686, "grad_norm": 0.8971892076690402, "learning_rate": 9.966213446611539e-06, "loss": 0.5139, "step": 1908 }, { "epoch": 0.13197372969236087, "grad_norm": 0.8233364422992642, "learning_rate": 9.966148452158651e-06, "loss": 0.347, "step": 1909 }, { "epoch": 0.1320428620808849, "grad_norm": 0.8095527629953191, "learning_rate": 9.966083395463953e-06, "loss": 0.3361, "step": 1910 }, { "epoch": 0.1321119944694089, "grad_norm": 0.8304815030258504, "learning_rate": 9.966018276528263e-06, "loss": 0.4956, "step": 1911 }, { "epoch": 0.13218112685793293, "grad_norm": 0.9323793422150555, "learning_rate": 9.965953095352397e-06, "loss": 0.4616, "step": 1912 }, { "epoch": 0.13225025924645697, "grad_norm": 0.9838770916276532, "learning_rate": 9.96588785193717e-06, "loss": 0.3392, "step": 1913 }, { "epoch": 0.132319391634981, "grad_norm": 0.9279535102081551, "learning_rate": 9.965822546283402e-06, "loss": 0.3711, "step": 1914 }, { "epoch": 0.132388524023505, "grad_norm": 0.9460890947099461, "learning_rate": 9.965757178391908e-06, "loss": 0.5059, "step": 1915 }, { "epoch": 0.13245765641202903, "grad_norm": 0.8072587755766216, "learning_rate": 9.965691748263512e-06, "loss": 0.3557, "step": 1916 }, { "epoch": 0.13252678880055305, "grad_norm": 1.032800482600776, "learning_rate": 9.96562625589903e-06, "loss": 0.2872, "step": 1917 }, { "epoch": 0.1325959211890771, "grad_norm": 1.1448107984650695, "learning_rate": 9.965560701299286e-06, "loss": 0.4152, "step": 1918 }, { "epoch": 0.1326650535776011, "grad_norm": 0.8617779059273446, "learning_rate": 9.965495084465098e-06, "loss": 0.4595, "step": 1919 }, { "epoch": 0.13273418596612513, "grad_norm": 0.9632059776001265, "learning_rate": 9.965429405397293e-06, "loss": 0.35, "step": 1920 }, { "epoch": 0.13280331835464915, "grad_norm": 0.9449452494995286, "learning_rate": 9.96536366409669e-06, "loss": 0.3861, "step": 1921 }, { "epoch": 0.13287245074317316, "grad_norm": 0.9261604516625107, "learning_rate": 9.965297860564113e-06, "loss": 0.4479, "step": 1922 }, { "epoch": 0.1329415831316972, "grad_norm": 0.8440045482179656, "learning_rate": 9.965231994800392e-06, "loss": 0.3506, "step": 1923 }, { "epoch": 0.13301071552022123, "grad_norm": 1.124138719829054, "learning_rate": 9.965166066806347e-06, "loss": 0.3778, "step": 1924 }, { "epoch": 0.13307984790874525, "grad_norm": 0.978566786998397, "learning_rate": 9.965100076582806e-06, "loss": 0.4619, "step": 1925 }, { "epoch": 0.13314898029726926, "grad_norm": 0.8020702347165781, "learning_rate": 9.965034024130595e-06, "loss": 0.4719, "step": 1926 }, { "epoch": 0.13321811268579328, "grad_norm": 0.9863853328878637, "learning_rate": 9.964967909450544e-06, "loss": 0.3708, "step": 1927 }, { "epoch": 0.13328724507431733, "grad_norm": 0.8336785145374581, "learning_rate": 9.964901732543481e-06, "loss": 0.373, "step": 1928 }, { "epoch": 0.13335637746284135, "grad_norm": 0.913932786607555, "learning_rate": 9.964835493410235e-06, "loss": 0.4372, "step": 1929 }, { "epoch": 0.13342550985136536, "grad_norm": 0.8585630002953153, "learning_rate": 9.964769192051636e-06, "loss": 0.3674, "step": 1930 }, { "epoch": 0.13349464223988938, "grad_norm": 0.9656033881664217, "learning_rate": 9.964702828468516e-06, "loss": 0.3196, "step": 1931 }, { "epoch": 0.1335637746284134, "grad_norm": 0.7738994033116171, "learning_rate": 9.964636402661706e-06, "loss": 0.367, "step": 1932 }, { "epoch": 0.13363290701693745, "grad_norm": 0.9369272740052547, "learning_rate": 9.964569914632036e-06, "loss": 0.5806, "step": 1933 }, { "epoch": 0.13370203940546146, "grad_norm": 1.139639454451517, "learning_rate": 9.964503364380344e-06, "loss": 0.3842, "step": 1934 }, { "epoch": 0.13377117179398548, "grad_norm": 2.732632420257232, "learning_rate": 9.964436751907462e-06, "loss": 0.3143, "step": 1935 }, { "epoch": 0.1338403041825095, "grad_norm": 1.1243543621753413, "learning_rate": 9.964370077214223e-06, "loss": 0.4085, "step": 1936 }, { "epoch": 0.13390943657103352, "grad_norm": 0.9912962985353634, "learning_rate": 9.964303340301467e-06, "loss": 0.513, "step": 1937 }, { "epoch": 0.13397856895955756, "grad_norm": 1.203873966854223, "learning_rate": 9.964236541170025e-06, "loss": 0.3264, "step": 1938 }, { "epoch": 0.13404770134808158, "grad_norm": 0.8159655290440043, "learning_rate": 9.96416967982074e-06, "loss": 0.4029, "step": 1939 }, { "epoch": 0.1341168337366056, "grad_norm": 0.8781667049436853, "learning_rate": 9.964102756254443e-06, "loss": 0.4194, "step": 1940 }, { "epoch": 0.13418596612512962, "grad_norm": 1.1008815093784305, "learning_rate": 9.964035770471981e-06, "loss": 0.3912, "step": 1941 }, { "epoch": 0.13425509851365364, "grad_norm": 0.9465308684824985, "learning_rate": 9.963968722474186e-06, "loss": 0.3729, "step": 1942 }, { "epoch": 0.13432423090217768, "grad_norm": 0.8578680480530685, "learning_rate": 9.963901612261903e-06, "loss": 0.4512, "step": 1943 }, { "epoch": 0.1343933632907017, "grad_norm": 0.8986590317674003, "learning_rate": 9.963834439835974e-06, "loss": 0.4882, "step": 1944 }, { "epoch": 0.13446249567922572, "grad_norm": 1.1135445720755697, "learning_rate": 9.963767205197236e-06, "loss": 0.2945, "step": 1945 }, { "epoch": 0.13453162806774974, "grad_norm": 0.9631802767765145, "learning_rate": 9.963699908346535e-06, "loss": 0.3593, "step": 1946 }, { "epoch": 0.13460076045627375, "grad_norm": 0.929411168838707, "learning_rate": 9.963632549284715e-06, "loss": 0.5936, "step": 1947 }, { "epoch": 0.1346698928447978, "grad_norm": 1.4136318077910213, "learning_rate": 9.963565128012616e-06, "loss": 0.375, "step": 1948 }, { "epoch": 0.13473902523332182, "grad_norm": 1.4417964488067243, "learning_rate": 9.963497644531089e-06, "loss": 0.3044, "step": 1949 }, { "epoch": 0.13480815762184584, "grad_norm": 0.8882834599723205, "learning_rate": 9.963430098840973e-06, "loss": 0.4058, "step": 1950 }, { "epoch": 0.13487729001036985, "grad_norm": 0.9436368753685558, "learning_rate": 9.963362490943123e-06, "loss": 0.4412, "step": 1951 }, { "epoch": 0.13494642239889387, "grad_norm": 0.8981744717057243, "learning_rate": 9.963294820838379e-06, "loss": 0.3699, "step": 1952 }, { "epoch": 0.13501555478741792, "grad_norm": 0.8806539174261762, "learning_rate": 9.963227088527591e-06, "loss": 0.3786, "step": 1953 }, { "epoch": 0.13508468717594194, "grad_norm": 0.9758790723810207, "learning_rate": 9.96315929401161e-06, "loss": 0.4774, "step": 1954 }, { "epoch": 0.13515381956446595, "grad_norm": 0.8679644670764585, "learning_rate": 9.963091437291286e-06, "loss": 0.3811, "step": 1955 }, { "epoch": 0.13522295195298997, "grad_norm": 3.2971838114182233, "learning_rate": 9.963023518367466e-06, "loss": 0.2863, "step": 1956 }, { "epoch": 0.135292084341514, "grad_norm": 0.8573389197961027, "learning_rate": 9.962955537241004e-06, "loss": 0.3537, "step": 1957 }, { "epoch": 0.13536121673003804, "grad_norm": 0.9474078198730971, "learning_rate": 9.96288749391275e-06, "loss": 0.4957, "step": 1958 }, { "epoch": 0.13543034911856205, "grad_norm": 0.8635950482635915, "learning_rate": 9.962819388383557e-06, "loss": 0.3577, "step": 1959 }, { "epoch": 0.13549948150708607, "grad_norm": 0.9139231690488784, "learning_rate": 9.962751220654283e-06, "loss": 0.3509, "step": 1960 }, { "epoch": 0.1355686138956101, "grad_norm": 0.9120464422516047, "learning_rate": 9.962682990725776e-06, "loss": 0.4243, "step": 1961 }, { "epoch": 0.1356377462841341, "grad_norm": 0.8713652162722731, "learning_rate": 9.962614698598895e-06, "loss": 0.4491, "step": 1962 }, { "epoch": 0.13570687867265813, "grad_norm": 0.8982679510008948, "learning_rate": 9.962546344274496e-06, "loss": 0.3892, "step": 1963 }, { "epoch": 0.13577601106118217, "grad_norm": 0.8173599172970631, "learning_rate": 9.962477927753434e-06, "loss": 0.2924, "step": 1964 }, { "epoch": 0.1358451434497062, "grad_norm": 0.9035791262881239, "learning_rate": 9.962409449036566e-06, "loss": 0.5297, "step": 1965 }, { "epoch": 0.1359142758382302, "grad_norm": 1.7323144870900018, "learning_rate": 9.962340908124752e-06, "loss": 0.4281, "step": 1966 }, { "epoch": 0.13598340822675423, "grad_norm": 1.8219438765233475, "learning_rate": 9.962272305018849e-06, "loss": 0.3816, "step": 1967 }, { "epoch": 0.13605254061527824, "grad_norm": 0.8976819342328033, "learning_rate": 9.962203639719722e-06, "loss": 0.3808, "step": 1968 }, { "epoch": 0.1361216730038023, "grad_norm": 0.922963543959527, "learning_rate": 9.962134912228225e-06, "loss": 0.4861, "step": 1969 }, { "epoch": 0.1361908053923263, "grad_norm": 0.788872224938736, "learning_rate": 9.962066122545221e-06, "loss": 0.3659, "step": 1970 }, { "epoch": 0.13625993778085033, "grad_norm": 0.9250519168086062, "learning_rate": 9.961997270671574e-06, "loss": 0.3132, "step": 1971 }, { "epoch": 0.13632907016937434, "grad_norm": 0.9647075175296552, "learning_rate": 9.961928356608149e-06, "loss": 0.5749, "step": 1972 }, { "epoch": 0.13639820255789836, "grad_norm": 0.8734412244328494, "learning_rate": 9.961859380355805e-06, "loss": 0.4253, "step": 1973 }, { "epoch": 0.1364673349464224, "grad_norm": 0.8685203887644062, "learning_rate": 9.96179034191541e-06, "loss": 0.3292, "step": 1974 }, { "epoch": 0.13653646733494643, "grad_norm": 1.0507312809662064, "learning_rate": 9.961721241287825e-06, "loss": 0.3594, "step": 1975 }, { "epoch": 0.13660559972347044, "grad_norm": 0.9436741340161898, "learning_rate": 9.961652078473922e-06, "loss": 0.4734, "step": 1976 }, { "epoch": 0.13667473211199446, "grad_norm": 0.9065298066849616, "learning_rate": 9.961582853474562e-06, "loss": 0.325, "step": 1977 }, { "epoch": 0.13674386450051848, "grad_norm": 1.9927580296011924, "learning_rate": 9.961513566290618e-06, "loss": 0.302, "step": 1978 }, { "epoch": 0.13681299688904253, "grad_norm": 0.8902730421590057, "learning_rate": 9.961444216922954e-06, "loss": 0.4793, "step": 1979 }, { "epoch": 0.13688212927756654, "grad_norm": 1.19061225982609, "learning_rate": 9.96137480537244e-06, "loss": 0.4009, "step": 1980 }, { "epoch": 0.13695126166609056, "grad_norm": 1.050429820338979, "learning_rate": 9.961305331639948e-06, "loss": 0.3061, "step": 1981 }, { "epoch": 0.13702039405461458, "grad_norm": 0.9468115487285946, "learning_rate": 9.961235795726347e-06, "loss": 0.3909, "step": 1982 }, { "epoch": 0.1370895264431386, "grad_norm": 1.026117364135588, "learning_rate": 9.961166197632508e-06, "loss": 0.4647, "step": 1983 }, { "epoch": 0.13715865883166264, "grad_norm": 0.8885198539932786, "learning_rate": 9.961096537359305e-06, "loss": 0.3404, "step": 1984 }, { "epoch": 0.13722779122018666, "grad_norm": 0.9473003272160929, "learning_rate": 9.961026814907613e-06, "loss": 0.3769, "step": 1985 }, { "epoch": 0.13729692360871068, "grad_norm": 0.8689829885465467, "learning_rate": 9.9609570302783e-06, "loss": 0.4627, "step": 1986 }, { "epoch": 0.1373660559972347, "grad_norm": 0.8820229280013474, "learning_rate": 9.960887183472245e-06, "loss": 0.4288, "step": 1987 }, { "epoch": 0.13743518838575872, "grad_norm": 0.8879491364058122, "learning_rate": 9.960817274490321e-06, "loss": 0.356, "step": 1988 }, { "epoch": 0.13750432077428276, "grad_norm": 0.9652912436956499, "learning_rate": 9.960747303333408e-06, "loss": 0.3873, "step": 1989 }, { "epoch": 0.13757345316280678, "grad_norm": 1.0655923249217358, "learning_rate": 9.960677270002379e-06, "loss": 0.3951, "step": 1990 }, { "epoch": 0.1376425855513308, "grad_norm": 0.9603515100926346, "learning_rate": 9.960607174498113e-06, "loss": 0.3607, "step": 1991 }, { "epoch": 0.13771171793985482, "grad_norm": 0.9697370013688111, "learning_rate": 9.96053701682149e-06, "loss": 0.3776, "step": 1992 }, { "epoch": 0.13778085032837883, "grad_norm": 1.0221383625917435, "learning_rate": 9.960466796973386e-06, "loss": 0.4399, "step": 1993 }, { "epoch": 0.13784998271690288, "grad_norm": 0.9006093993711984, "learning_rate": 9.960396514954684e-06, "loss": 0.4684, "step": 1994 }, { "epoch": 0.1379191151054269, "grad_norm": 1.0308332805722686, "learning_rate": 9.960326170766263e-06, "loss": 0.4756, "step": 1995 }, { "epoch": 0.13798824749395092, "grad_norm": 2.387666399078766, "learning_rate": 9.960255764409007e-06, "loss": 0.3548, "step": 1996 }, { "epoch": 0.13805737988247493, "grad_norm": 0.8648075919412815, "learning_rate": 9.960185295883796e-06, "loss": 0.4806, "step": 1997 }, { "epoch": 0.13812651227099895, "grad_norm": 0.9320673625229137, "learning_rate": 9.960114765191516e-06, "loss": 0.4463, "step": 1998 }, { "epoch": 0.138195644659523, "grad_norm": 0.8602962388057277, "learning_rate": 9.960044172333046e-06, "loss": 0.3588, "step": 1999 }, { "epoch": 0.13826477704804702, "grad_norm": 1.1780848089004183, "learning_rate": 9.959973517309276e-06, "loss": 0.4004, "step": 2000 }, { "epoch": 0.13833390943657103, "grad_norm": 0.8595722525719109, "learning_rate": 9.959902800121088e-06, "loss": 0.3563, "step": 2001 }, { "epoch": 0.13840304182509505, "grad_norm": 0.9307013689925492, "learning_rate": 9.959832020769371e-06, "loss": 0.3556, "step": 2002 }, { "epoch": 0.13847217421361907, "grad_norm": 1.0881644776172419, "learning_rate": 9.959761179255012e-06, "loss": 0.3409, "step": 2003 }, { "epoch": 0.13854130660214312, "grad_norm": 0.8592340010729977, "learning_rate": 9.959690275578895e-06, "loss": 0.4747, "step": 2004 }, { "epoch": 0.13861043899066713, "grad_norm": 0.9110198332341831, "learning_rate": 9.959619309741913e-06, "loss": 0.4314, "step": 2005 }, { "epoch": 0.13867957137919115, "grad_norm": 0.8623664736054635, "learning_rate": 9.959548281744952e-06, "loss": 0.3601, "step": 2006 }, { "epoch": 0.13874870376771517, "grad_norm": 0.8477151723899172, "learning_rate": 9.959477191588907e-06, "loss": 0.3503, "step": 2007 }, { "epoch": 0.1388178361562392, "grad_norm": 1.0153899276365985, "learning_rate": 9.959406039274664e-06, "loss": 0.4347, "step": 2008 }, { "epoch": 0.13888696854476323, "grad_norm": 0.9173197778202674, "learning_rate": 9.959334824803117e-06, "loss": 0.3922, "step": 2009 }, { "epoch": 0.13895610093328725, "grad_norm": 0.9093734708846483, "learning_rate": 9.95926354817516e-06, "loss": 0.3141, "step": 2010 }, { "epoch": 0.13902523332181127, "grad_norm": 1.150434528568557, "learning_rate": 9.959192209391682e-06, "loss": 0.4141, "step": 2011 }, { "epoch": 0.1390943657103353, "grad_norm": 0.9259923322088611, "learning_rate": 9.959120808453582e-06, "loss": 0.4383, "step": 2012 }, { "epoch": 0.1391634980988593, "grad_norm": 0.9372697715167703, "learning_rate": 9.959049345361751e-06, "loss": 0.3569, "step": 2013 }, { "epoch": 0.13923263048738335, "grad_norm": 0.8698626671677389, "learning_rate": 9.958977820117086e-06, "loss": 0.3501, "step": 2014 }, { "epoch": 0.13930176287590737, "grad_norm": 0.981150678778472, "learning_rate": 9.958906232720485e-06, "loss": 0.4196, "step": 2015 }, { "epoch": 0.1393708952644314, "grad_norm": 0.9621720998727565, "learning_rate": 9.958834583172843e-06, "loss": 0.3695, "step": 2016 }, { "epoch": 0.1394400276529554, "grad_norm": 0.8453442361000806, "learning_rate": 9.958762871475058e-06, "loss": 0.3343, "step": 2017 }, { "epoch": 0.13950916004147942, "grad_norm": 0.8447561660202406, "learning_rate": 9.958691097628032e-06, "loss": 0.3709, "step": 2018 }, { "epoch": 0.13957829243000347, "grad_norm": 0.8356967377708858, "learning_rate": 9.958619261632662e-06, "loss": 0.4248, "step": 2019 }, { "epoch": 0.1396474248185275, "grad_norm": 0.960922366426555, "learning_rate": 9.958547363489847e-06, "loss": 0.3804, "step": 2020 }, { "epoch": 0.1397165572070515, "grad_norm": 0.847683505676119, "learning_rate": 9.958475403200491e-06, "loss": 0.3112, "step": 2021 }, { "epoch": 0.13978568959557552, "grad_norm": 0.9076262528208336, "learning_rate": 9.958403380765494e-06, "loss": 0.4867, "step": 2022 }, { "epoch": 0.13985482198409954, "grad_norm": 0.8823155019297005, "learning_rate": 9.958331296185759e-06, "loss": 0.4002, "step": 2023 }, { "epoch": 0.1399239543726236, "grad_norm": 0.937308855370992, "learning_rate": 9.958259149462191e-06, "loss": 0.3322, "step": 2024 }, { "epoch": 0.1399930867611476, "grad_norm": 0.8348986180158062, "learning_rate": 9.958186940595692e-06, "loss": 0.4393, "step": 2025 }, { "epoch": 0.14006221914967162, "grad_norm": 1.395065894829357, "learning_rate": 9.958114669587168e-06, "loss": 0.5047, "step": 2026 }, { "epoch": 0.14013135153819564, "grad_norm": 0.8367940273649608, "learning_rate": 9.958042336437526e-06, "loss": 0.3346, "step": 2027 }, { "epoch": 0.14020048392671966, "grad_norm": 1.2003600952223146, "learning_rate": 9.95796994114767e-06, "loss": 0.3071, "step": 2028 }, { "epoch": 0.14026961631524368, "grad_norm": 0.9094792258825978, "learning_rate": 9.957897483718509e-06, "loss": 0.404, "step": 2029 }, { "epoch": 0.14033874870376772, "grad_norm": 1.7443758842109283, "learning_rate": 9.95782496415095e-06, "loss": 0.4241, "step": 2030 }, { "epoch": 0.14040788109229174, "grad_norm": 1.6362844764761233, "learning_rate": 9.957752382445904e-06, "loss": 0.3568, "step": 2031 }, { "epoch": 0.14047701348081576, "grad_norm": 1.035119856750417, "learning_rate": 9.957679738604279e-06, "loss": 0.3481, "step": 2032 }, { "epoch": 0.14054614586933978, "grad_norm": 0.8113454835032464, "learning_rate": 9.957607032626986e-06, "loss": 0.4379, "step": 2033 }, { "epoch": 0.1406152782578638, "grad_norm": 0.9158666820751569, "learning_rate": 9.957534264514936e-06, "loss": 0.3758, "step": 2034 }, { "epoch": 0.14068441064638784, "grad_norm": 0.973444714870114, "learning_rate": 9.95746143426904e-06, "loss": 0.3447, "step": 2035 }, { "epoch": 0.14075354303491186, "grad_norm": 0.9769503221294077, "learning_rate": 9.957388541890213e-06, "loss": 0.4074, "step": 2036 }, { "epoch": 0.14082267542343588, "grad_norm": 0.9309901347381304, "learning_rate": 9.957315587379368e-06, "loss": 0.4306, "step": 2037 }, { "epoch": 0.1408918078119599, "grad_norm": 1.0902272411645013, "learning_rate": 9.957242570737417e-06, "loss": 0.3252, "step": 2038 }, { "epoch": 0.14096094020048391, "grad_norm": 0.883357003049754, "learning_rate": 9.95716949196528e-06, "loss": 0.4139, "step": 2039 }, { "epoch": 0.14103007258900796, "grad_norm": 0.8807582233447695, "learning_rate": 9.957096351063868e-06, "loss": 0.4408, "step": 2040 }, { "epoch": 0.14109920497753198, "grad_norm": 0.8902938121074824, "learning_rate": 9.9570231480341e-06, "loss": 0.3929, "step": 2041 }, { "epoch": 0.141168337366056, "grad_norm": 0.8638898592275338, "learning_rate": 9.956949882876892e-06, "loss": 0.3119, "step": 2042 }, { "epoch": 0.14123746975458001, "grad_norm": 0.965347285557273, "learning_rate": 9.956876555593163e-06, "loss": 0.377, "step": 2043 }, { "epoch": 0.14130660214310403, "grad_norm": 1.000295429492598, "learning_rate": 9.956803166183833e-06, "loss": 0.4219, "step": 2044 }, { "epoch": 0.14137573453162808, "grad_norm": 0.7385267367711902, "learning_rate": 9.956729714649822e-06, "loss": 0.2901, "step": 2045 }, { "epoch": 0.1414448669201521, "grad_norm": 0.8287226667480018, "learning_rate": 9.95665620099205e-06, "loss": 0.3599, "step": 2046 }, { "epoch": 0.1415139993086761, "grad_norm": 0.883491962683463, "learning_rate": 9.956582625211437e-06, "loss": 0.413, "step": 2047 }, { "epoch": 0.14158313169720013, "grad_norm": 0.9481894842061076, "learning_rate": 9.956508987308905e-06, "loss": 0.3485, "step": 2048 }, { "epoch": 0.14165226408572415, "grad_norm": 1.042073729947016, "learning_rate": 9.95643528728538e-06, "loss": 0.3472, "step": 2049 }, { "epoch": 0.1417213964742482, "grad_norm": 0.8039830034382857, "learning_rate": 9.956361525141784e-06, "loss": 0.3903, "step": 2050 }, { "epoch": 0.1417905288627722, "grad_norm": 0.8830702702049126, "learning_rate": 9.956287700879041e-06, "loss": 0.4122, "step": 2051 }, { "epoch": 0.14185966125129623, "grad_norm": 0.9879996789566888, "learning_rate": 9.956213814498077e-06, "loss": 0.3765, "step": 2052 }, { "epoch": 0.14192879363982025, "grad_norm": 1.1505711486618073, "learning_rate": 9.956139865999817e-06, "loss": 0.3525, "step": 2053 }, { "epoch": 0.14199792602834427, "grad_norm": 0.8566498285163859, "learning_rate": 9.956065855385187e-06, "loss": 0.48, "step": 2054 }, { "epoch": 0.1420670584168683, "grad_norm": 0.8790763807577403, "learning_rate": 9.955991782655117e-06, "loss": 0.4256, "step": 2055 }, { "epoch": 0.14213619080539233, "grad_norm": 0.8983267165480593, "learning_rate": 9.955917647810536e-06, "loss": 0.33, "step": 2056 }, { "epoch": 0.14220532319391635, "grad_norm": 2.0499527555654313, "learning_rate": 9.95584345085237e-06, "loss": 0.3866, "step": 2057 }, { "epoch": 0.14227445558244037, "grad_norm": 0.8780296744951654, "learning_rate": 9.95576919178155e-06, "loss": 0.5105, "step": 2058 }, { "epoch": 0.14234358797096439, "grad_norm": 0.8480973847085548, "learning_rate": 9.955694870599007e-06, "loss": 0.4047, "step": 2059 }, { "epoch": 0.14241272035948843, "grad_norm": 0.8231346644353933, "learning_rate": 9.955620487305674e-06, "loss": 0.3414, "step": 2060 }, { "epoch": 0.14248185274801245, "grad_norm": 1.0313451387608727, "learning_rate": 9.95554604190248e-06, "loss": 0.4163, "step": 2061 }, { "epoch": 0.14255098513653647, "grad_norm": 0.8333589134181782, "learning_rate": 9.955471534390361e-06, "loss": 0.4777, "step": 2062 }, { "epoch": 0.14262011752506049, "grad_norm": 1.1051986562326408, "learning_rate": 9.955396964770251e-06, "loss": 0.3833, "step": 2063 }, { "epoch": 0.1426892499135845, "grad_norm": 1.0847909612417659, "learning_rate": 9.955322333043083e-06, "loss": 0.381, "step": 2064 }, { "epoch": 0.14275838230210855, "grad_norm": 0.9165505873487834, "learning_rate": 9.955247639209791e-06, "loss": 0.4635, "step": 2065 }, { "epoch": 0.14282751469063257, "grad_norm": 0.949394674312578, "learning_rate": 9.955172883271313e-06, "loss": 0.3637, "step": 2066 }, { "epoch": 0.14289664707915659, "grad_norm": 1.0924197706566596, "learning_rate": 9.955098065228588e-06, "loss": 0.3102, "step": 2067 }, { "epoch": 0.1429657794676806, "grad_norm": 1.09904300127173, "learning_rate": 9.95502318508255e-06, "loss": 0.3958, "step": 2068 }, { "epoch": 0.14303491185620462, "grad_norm": 0.8833132211543144, "learning_rate": 9.954948242834139e-06, "loss": 0.4139, "step": 2069 }, { "epoch": 0.14310404424472867, "grad_norm": 0.9445107066893563, "learning_rate": 9.954873238484294e-06, "loss": 0.3472, "step": 2070 }, { "epoch": 0.14317317663325269, "grad_norm": 0.7858329042678148, "learning_rate": 9.954798172033956e-06, "loss": 0.3496, "step": 2071 }, { "epoch": 0.1432423090217767, "grad_norm": 0.8113793734066824, "learning_rate": 9.954723043484065e-06, "loss": 0.4895, "step": 2072 }, { "epoch": 0.14331144141030072, "grad_norm": 0.9675832913851217, "learning_rate": 9.954647852835563e-06, "loss": 0.3974, "step": 2073 }, { "epoch": 0.14338057379882474, "grad_norm": 0.7968161852097555, "learning_rate": 9.954572600089392e-06, "loss": 0.3024, "step": 2074 }, { "epoch": 0.14344970618734879, "grad_norm": 0.9428766322206447, "learning_rate": 9.954497285246496e-06, "loss": 0.4666, "step": 2075 }, { "epoch": 0.1435188385758728, "grad_norm": 0.9286176318150839, "learning_rate": 9.954421908307817e-06, "loss": 0.4419, "step": 2076 }, { "epoch": 0.14358797096439682, "grad_norm": 0.8910777808900974, "learning_rate": 9.954346469274301e-06, "loss": 0.3875, "step": 2077 }, { "epoch": 0.14365710335292084, "grad_norm": 0.957297128763239, "learning_rate": 9.954270968146894e-06, "loss": 0.2892, "step": 2078 }, { "epoch": 0.14372623574144486, "grad_norm": 1.011349409723069, "learning_rate": 9.954195404926542e-06, "loss": 0.4383, "step": 2079 }, { "epoch": 0.1437953681299689, "grad_norm": 0.863977548534888, "learning_rate": 9.95411977961419e-06, "loss": 0.3678, "step": 2080 }, { "epoch": 0.14386450051849292, "grad_norm": 0.8817983535544256, "learning_rate": 9.95404409221079e-06, "loss": 0.4028, "step": 2081 }, { "epoch": 0.14393363290701694, "grad_norm": 0.8811815883876614, "learning_rate": 9.953968342717287e-06, "loss": 0.3838, "step": 2082 }, { "epoch": 0.14400276529554096, "grad_norm": 0.8128247385850691, "learning_rate": 9.953892531134631e-06, "loss": 0.4686, "step": 2083 }, { "epoch": 0.14407189768406498, "grad_norm": 0.9055274310707553, "learning_rate": 9.953816657463773e-06, "loss": 0.3457, "step": 2084 }, { "epoch": 0.14414103007258902, "grad_norm": 1.3856365842674891, "learning_rate": 9.953740721705665e-06, "loss": 0.3387, "step": 2085 }, { "epoch": 0.14421016246111304, "grad_norm": 0.9214055826339786, "learning_rate": 9.953664723861256e-06, "loss": 0.4348, "step": 2086 }, { "epoch": 0.14427929484963706, "grad_norm": 0.9250434818767452, "learning_rate": 9.9535886639315e-06, "loss": 0.4517, "step": 2087 }, { "epoch": 0.14434842723816108, "grad_norm": 0.9246388800470678, "learning_rate": 9.953512541917351e-06, "loss": 0.3475, "step": 2088 }, { "epoch": 0.1444175596266851, "grad_norm": 0.7995448889678178, "learning_rate": 9.953436357819762e-06, "loss": 0.3539, "step": 2089 }, { "epoch": 0.1444866920152091, "grad_norm": 0.8320479391968413, "learning_rate": 9.953360111639687e-06, "loss": 0.4941, "step": 2090 }, { "epoch": 0.14455582440373316, "grad_norm": 1.0207561173393584, "learning_rate": 9.953283803378084e-06, "loss": 0.4189, "step": 2091 }, { "epoch": 0.14462495679225718, "grad_norm": 1.039944284262919, "learning_rate": 9.953207433035907e-06, "loss": 0.2912, "step": 2092 }, { "epoch": 0.1446940891807812, "grad_norm": 0.8585820475475276, "learning_rate": 9.953131000614115e-06, "loss": 0.4049, "step": 2093 }, { "epoch": 0.1447632215693052, "grad_norm": 0.7651179952269105, "learning_rate": 9.953054506113664e-06, "loss": 0.4446, "step": 2094 }, { "epoch": 0.14483235395782923, "grad_norm": 0.9492855553299266, "learning_rate": 9.952977949535516e-06, "loss": 0.3377, "step": 2095 }, { "epoch": 0.14490148634635328, "grad_norm": 0.9683741867998191, "learning_rate": 9.952901330880626e-06, "loss": 0.3231, "step": 2096 }, { "epoch": 0.1449706187348773, "grad_norm": 1.9283380902433314, "learning_rate": 9.952824650149959e-06, "loss": 0.4408, "step": 2097 }, { "epoch": 0.1450397511234013, "grad_norm": 0.781321825020996, "learning_rate": 9.952747907344472e-06, "loss": 0.3321, "step": 2098 }, { "epoch": 0.14510888351192533, "grad_norm": 0.934958451680397, "learning_rate": 9.95267110246513e-06, "loss": 0.3315, "step": 2099 }, { "epoch": 0.14517801590044935, "grad_norm": 0.8621348725920823, "learning_rate": 9.952594235512892e-06, "loss": 0.3497, "step": 2100 }, { "epoch": 0.1452471482889734, "grad_norm": 0.8455186991343993, "learning_rate": 9.952517306488724e-06, "loss": 0.4729, "step": 2101 }, { "epoch": 0.1453162806774974, "grad_norm": 0.8392034136406386, "learning_rate": 9.952440315393593e-06, "loss": 0.3329, "step": 2102 }, { "epoch": 0.14538541306602143, "grad_norm": 0.827104914590434, "learning_rate": 9.952363262228456e-06, "loss": 0.3128, "step": 2103 }, { "epoch": 0.14545454545454545, "grad_norm": 0.9014224027002576, "learning_rate": 9.952286146994287e-06, "loss": 0.4639, "step": 2104 }, { "epoch": 0.14552367784306947, "grad_norm": 0.8792760952548173, "learning_rate": 9.952208969692047e-06, "loss": 0.4116, "step": 2105 }, { "epoch": 0.1455928102315935, "grad_norm": 0.9277461446927883, "learning_rate": 9.952131730322706e-06, "loss": 0.3114, "step": 2106 }, { "epoch": 0.14566194262011753, "grad_norm": 1.2584053357414033, "learning_rate": 9.95205442888723e-06, "loss": 0.3707, "step": 2107 }, { "epoch": 0.14573107500864155, "grad_norm": 0.8787053358759508, "learning_rate": 9.95197706538659e-06, "loss": 0.4777, "step": 2108 }, { "epoch": 0.14580020739716557, "grad_norm": 0.973588976240068, "learning_rate": 9.951899639821755e-06, "loss": 0.4035, "step": 2109 }, { "epoch": 0.14586933978568958, "grad_norm": 1.093934931254248, "learning_rate": 9.951822152193694e-06, "loss": 0.2638, "step": 2110 }, { "epoch": 0.14593847217421363, "grad_norm": 1.1843607346667737, "learning_rate": 9.951744602503382e-06, "loss": 0.4426, "step": 2111 }, { "epoch": 0.14600760456273765, "grad_norm": 0.8960861593051306, "learning_rate": 9.951666990751786e-06, "loss": 0.4251, "step": 2112 }, { "epoch": 0.14607673695126167, "grad_norm": 0.947803411836165, "learning_rate": 9.95158931693988e-06, "loss": 0.3373, "step": 2113 }, { "epoch": 0.14614586933978568, "grad_norm": 0.818077474563063, "learning_rate": 9.95151158106864e-06, "loss": 0.3486, "step": 2114 }, { "epoch": 0.1462150017283097, "grad_norm": 0.819350956893034, "learning_rate": 9.951433783139039e-06, "loss": 0.4789, "step": 2115 }, { "epoch": 0.14628413411683375, "grad_norm": 1.0536845617114505, "learning_rate": 9.95135592315205e-06, "loss": 0.4425, "step": 2116 }, { "epoch": 0.14635326650535777, "grad_norm": 0.946939535237416, "learning_rate": 9.951278001108652e-06, "loss": 0.2982, "step": 2117 }, { "epoch": 0.14642239889388178, "grad_norm": 1.0108913305323521, "learning_rate": 9.95120001700982e-06, "loss": 0.4052, "step": 2118 }, { "epoch": 0.1464915312824058, "grad_norm": 0.9473911225065307, "learning_rate": 9.951121970856533e-06, "loss": 0.4715, "step": 2119 }, { "epoch": 0.14656066367092982, "grad_norm": 1.232134880722149, "learning_rate": 9.951043862649766e-06, "loss": 0.3407, "step": 2120 }, { "epoch": 0.14662979605945387, "grad_norm": 0.8447191465234692, "learning_rate": 9.950965692390499e-06, "loss": 0.3362, "step": 2121 }, { "epoch": 0.14669892844797788, "grad_norm": 0.8134627805378373, "learning_rate": 9.950887460079712e-06, "loss": 0.4732, "step": 2122 }, { "epoch": 0.1467680608365019, "grad_norm": 0.8606515848618943, "learning_rate": 9.950809165718389e-06, "loss": 0.4038, "step": 2123 }, { "epoch": 0.14683719322502592, "grad_norm": 0.8683553798786959, "learning_rate": 9.950730809307506e-06, "loss": 0.3574, "step": 2124 }, { "epoch": 0.14690632561354994, "grad_norm": 0.9455320819508125, "learning_rate": 9.950652390848047e-06, "loss": 0.3615, "step": 2125 }, { "epoch": 0.14697545800207398, "grad_norm": 0.8960155518516777, "learning_rate": 9.950573910340996e-06, "loss": 0.5416, "step": 2126 }, { "epoch": 0.147044590390598, "grad_norm": 0.9039365916087589, "learning_rate": 9.950495367787335e-06, "loss": 0.4077, "step": 2127 }, { "epoch": 0.14711372277912202, "grad_norm": 1.1063674481188865, "learning_rate": 9.950416763188052e-06, "loss": 0.369, "step": 2128 }, { "epoch": 0.14718285516764604, "grad_norm": 0.878728730510617, "learning_rate": 9.950338096544125e-06, "loss": 0.4998, "step": 2129 }, { "epoch": 0.14725198755617006, "grad_norm": 0.821944349700974, "learning_rate": 9.950259367856548e-06, "loss": 0.3702, "step": 2130 }, { "epoch": 0.1473211199446941, "grad_norm": 0.9143671015747974, "learning_rate": 9.9501805771263e-06, "loss": 0.3584, "step": 2131 }, { "epoch": 0.14739025233321812, "grad_norm": 0.8502450318138844, "learning_rate": 9.950101724354375e-06, "loss": 0.3683, "step": 2132 }, { "epoch": 0.14745938472174214, "grad_norm": 0.8317573417881886, "learning_rate": 9.95002280954176e-06, "loss": 0.4402, "step": 2133 }, { "epoch": 0.14752851711026616, "grad_norm": 0.9254232778970776, "learning_rate": 9.94994383268944e-06, "loss": 0.355, "step": 2134 }, { "epoch": 0.14759764949879017, "grad_norm": 1.0061199508556595, "learning_rate": 9.94986479379841e-06, "loss": 0.3827, "step": 2135 }, { "epoch": 0.14766678188731422, "grad_norm": 1.0413321567579006, "learning_rate": 9.949785692869657e-06, "loss": 0.4046, "step": 2136 }, { "epoch": 0.14773591427583824, "grad_norm": 0.828880482923723, "learning_rate": 9.949706529904173e-06, "loss": 0.3887, "step": 2137 }, { "epoch": 0.14780504666436226, "grad_norm": 0.8754833132968451, "learning_rate": 9.949627304902951e-06, "loss": 0.3375, "step": 2138 }, { "epoch": 0.14787417905288627, "grad_norm": 0.9782816376747534, "learning_rate": 9.949548017866986e-06, "loss": 0.4164, "step": 2139 }, { "epoch": 0.1479433114414103, "grad_norm": 0.8852762354434355, "learning_rate": 9.949468668797267e-06, "loss": 0.4727, "step": 2140 }, { "epoch": 0.14801244382993434, "grad_norm": 1.0366117959732721, "learning_rate": 9.949389257694792e-06, "loss": 0.3768, "step": 2141 }, { "epoch": 0.14808157621845836, "grad_norm": 0.9113755638843117, "learning_rate": 9.949309784560555e-06, "loss": 0.3394, "step": 2142 }, { "epoch": 0.14815070860698237, "grad_norm": 0.9094434383814621, "learning_rate": 9.949230249395551e-06, "loss": 0.4423, "step": 2143 }, { "epoch": 0.1482198409955064, "grad_norm": 1.2528517749732289, "learning_rate": 9.949150652200781e-06, "loss": 0.4277, "step": 2144 }, { "epoch": 0.1482889733840304, "grad_norm": 0.8648322574127993, "learning_rate": 9.949070992977236e-06, "loss": 0.4008, "step": 2145 }, { "epoch": 0.14835810577255445, "grad_norm": 0.8233336169458044, "learning_rate": 9.948991271725921e-06, "loss": 0.34, "step": 2146 }, { "epoch": 0.14842723816107847, "grad_norm": 1.259812888163033, "learning_rate": 9.948911488447833e-06, "loss": 0.5083, "step": 2147 }, { "epoch": 0.1484963705496025, "grad_norm": 3.222533890897229, "learning_rate": 9.948831643143969e-06, "loss": 0.3585, "step": 2148 }, { "epoch": 0.1485655029381265, "grad_norm": 0.778283967469291, "learning_rate": 9.948751735815332e-06, "loss": 0.3501, "step": 2149 }, { "epoch": 0.14863463532665053, "grad_norm": 0.9675522636358552, "learning_rate": 9.948671766462923e-06, "loss": 0.4413, "step": 2150 }, { "epoch": 0.14870376771517457, "grad_norm": 0.9288647568944721, "learning_rate": 9.948591735087747e-06, "loss": 0.4274, "step": 2151 }, { "epoch": 0.1487729001036986, "grad_norm": 1.0612397247600478, "learning_rate": 9.948511641690804e-06, "loss": 0.3201, "step": 2152 }, { "epoch": 0.1488420324922226, "grad_norm": 1.011601298265559, "learning_rate": 9.948431486273098e-06, "loss": 0.396, "step": 2153 }, { "epoch": 0.14891116488074663, "grad_norm": 0.938403741196351, "learning_rate": 9.948351268835634e-06, "loss": 0.42, "step": 2154 }, { "epoch": 0.14898029726927065, "grad_norm": 1.151751755964325, "learning_rate": 9.948270989379418e-06, "loss": 0.4204, "step": 2155 }, { "epoch": 0.14904942965779466, "grad_norm": 1.280434561051462, "learning_rate": 9.948190647905456e-06, "loss": 0.3714, "step": 2156 }, { "epoch": 0.1491185620463187, "grad_norm": 0.9417614905297393, "learning_rate": 9.948110244414753e-06, "loss": 0.3873, "step": 2157 }, { "epoch": 0.14918769443484273, "grad_norm": 1.2987167003779665, "learning_rate": 9.94802977890832e-06, "loss": 0.4326, "step": 2158 }, { "epoch": 0.14925682682336675, "grad_norm": 1.1108464173869685, "learning_rate": 9.947949251387164e-06, "loss": 0.379, "step": 2159 }, { "epoch": 0.14932595921189076, "grad_norm": 1.4086166141346073, "learning_rate": 9.947868661852293e-06, "loss": 0.3342, "step": 2160 }, { "epoch": 0.14939509160041478, "grad_norm": 11.402236450143633, "learning_rate": 9.94778801030472e-06, "loss": 0.3909, "step": 2161 }, { "epoch": 0.14946422398893883, "grad_norm": 0.9274425137587645, "learning_rate": 9.947707296745452e-06, "loss": 0.4794, "step": 2162 }, { "epoch": 0.14953335637746284, "grad_norm": 1.466286578921824, "learning_rate": 9.947626521175504e-06, "loss": 0.3413, "step": 2163 }, { "epoch": 0.14960248876598686, "grad_norm": 0.8435332468439622, "learning_rate": 9.947545683595886e-06, "loss": 0.371, "step": 2164 }, { "epoch": 0.14967162115451088, "grad_norm": 1.7446647941850073, "learning_rate": 9.947464784007613e-06, "loss": 0.5038, "step": 2165 }, { "epoch": 0.1497407535430349, "grad_norm": 0.9487895175557338, "learning_rate": 9.947383822411698e-06, "loss": 0.3824, "step": 2166 }, { "epoch": 0.14980988593155894, "grad_norm": 0.9695154009193647, "learning_rate": 9.947302798809155e-06, "loss": 0.3664, "step": 2167 }, { "epoch": 0.14987901832008296, "grad_norm": 0.9858284763393972, "learning_rate": 9.947221713201e-06, "loss": 0.3819, "step": 2168 }, { "epoch": 0.14994815070860698, "grad_norm": 0.7966176612930537, "learning_rate": 9.947140565588252e-06, "loss": 0.417, "step": 2169 }, { "epoch": 0.150017283097131, "grad_norm": 1.1481010935975162, "learning_rate": 9.947059355971925e-06, "loss": 0.2728, "step": 2170 }, { "epoch": 0.15008641548565502, "grad_norm": 0.8674300145195574, "learning_rate": 9.946978084353034e-06, "loss": 0.3648, "step": 2171 }, { "epoch": 0.15015554787417906, "grad_norm": 1.0652031831325663, "learning_rate": 9.946896750732605e-06, "loss": 0.4481, "step": 2172 }, { "epoch": 0.15022468026270308, "grad_norm": 0.9492827802910396, "learning_rate": 9.94681535511165e-06, "loss": 0.3778, "step": 2173 }, { "epoch": 0.1502938126512271, "grad_norm": 0.9239188942024152, "learning_rate": 9.946733897491194e-06, "loss": 0.2897, "step": 2174 }, { "epoch": 0.15036294503975112, "grad_norm": 1.4848539254955053, "learning_rate": 9.946652377872256e-06, "loss": 0.3761, "step": 2175 }, { "epoch": 0.15043207742827513, "grad_norm": 0.8272007592308775, "learning_rate": 9.946570796255862e-06, "loss": 0.4536, "step": 2176 }, { "epoch": 0.15050120981679918, "grad_norm": 0.8798165656071425, "learning_rate": 9.946489152643027e-06, "loss": 0.3625, "step": 2177 }, { "epoch": 0.1505703422053232, "grad_norm": 0.9792040441855694, "learning_rate": 9.946407447034776e-06, "loss": 0.3127, "step": 2178 }, { "epoch": 0.15063947459384722, "grad_norm": 1.019079216475127, "learning_rate": 9.946325679432139e-06, "loss": 0.4966, "step": 2179 }, { "epoch": 0.15070860698237123, "grad_norm": 0.9411076585449757, "learning_rate": 9.946243849836134e-06, "loss": 0.4119, "step": 2180 }, { "epoch": 0.15077773937089525, "grad_norm": 0.9889183861599099, "learning_rate": 9.94616195824779e-06, "loss": 0.3975, "step": 2181 }, { "epoch": 0.1508468717594193, "grad_norm": 0.971098225156341, "learning_rate": 9.946080004668132e-06, "loss": 0.3024, "step": 2182 }, { "epoch": 0.15091600414794332, "grad_norm": 0.9469224011853368, "learning_rate": 9.94599798909819e-06, "loss": 0.4855, "step": 2183 }, { "epoch": 0.15098513653646733, "grad_norm": 0.8708010584560019, "learning_rate": 9.945915911538989e-06, "loss": 0.3132, "step": 2184 }, { "epoch": 0.15105426892499135, "grad_norm": 1.9235845025500304, "learning_rate": 9.945833771991558e-06, "loss": 0.3042, "step": 2185 }, { "epoch": 0.15112340131351537, "grad_norm": 0.9257547391138268, "learning_rate": 9.945751570456925e-06, "loss": 0.4177, "step": 2186 }, { "epoch": 0.15119253370203942, "grad_norm": 0.9424050068026241, "learning_rate": 9.945669306936126e-06, "loss": 0.4395, "step": 2187 }, { "epoch": 0.15126166609056343, "grad_norm": 0.8028514665462158, "learning_rate": 9.945586981430185e-06, "loss": 0.3467, "step": 2188 }, { "epoch": 0.15133079847908745, "grad_norm": 0.9441703180417664, "learning_rate": 9.945504593940139e-06, "loss": 0.3781, "step": 2189 }, { "epoch": 0.15139993086761147, "grad_norm": 1.70852188260307, "learning_rate": 9.945422144467017e-06, "loss": 0.4453, "step": 2190 }, { "epoch": 0.1514690632561355, "grad_norm": 0.9821922139239649, "learning_rate": 9.945339633011855e-06, "loss": 0.318, "step": 2191 }, { "epoch": 0.15153819564465953, "grad_norm": 1.015365967937453, "learning_rate": 9.945257059575685e-06, "loss": 0.3215, "step": 2192 }, { "epoch": 0.15160732803318355, "grad_norm": 0.8776304220723691, "learning_rate": 9.945174424159544e-06, "loss": 0.397, "step": 2193 }, { "epoch": 0.15167646042170757, "grad_norm": 1.2208087880524192, "learning_rate": 9.945091726764466e-06, "loss": 0.5008, "step": 2194 }, { "epoch": 0.1517455928102316, "grad_norm": 1.003485871720804, "learning_rate": 9.945008967391488e-06, "loss": 0.3527, "step": 2195 }, { "epoch": 0.1518147251987556, "grad_norm": 0.9517528992403361, "learning_rate": 9.944926146041648e-06, "loss": 0.3714, "step": 2196 }, { "epoch": 0.15188385758727965, "grad_norm": 0.8466466749541146, "learning_rate": 9.944843262715982e-06, "loss": 0.4368, "step": 2197 }, { "epoch": 0.15195298997580367, "grad_norm": 1.0503806414706245, "learning_rate": 9.944760317415531e-06, "loss": 0.3765, "step": 2198 }, { "epoch": 0.1520221223643277, "grad_norm": 1.03996097201477, "learning_rate": 9.944677310141336e-06, "loss": 0.3308, "step": 2199 }, { "epoch": 0.1520912547528517, "grad_norm": 1.065984392864292, "learning_rate": 9.944594240894432e-06, "loss": 0.4244, "step": 2200 }, { "epoch": 0.15216038714137572, "grad_norm": 0.8769444494049093, "learning_rate": 9.944511109675865e-06, "loss": 0.458, "step": 2201 }, { "epoch": 0.15222951952989977, "grad_norm": 1.2408971482479458, "learning_rate": 9.944427916486674e-06, "loss": 0.3398, "step": 2202 }, { "epoch": 0.1522986519184238, "grad_norm": 1.0284433708438314, "learning_rate": 9.944344661327903e-06, "loss": 0.3286, "step": 2203 }, { "epoch": 0.1523677843069478, "grad_norm": 0.9329385127097808, "learning_rate": 9.944261344200597e-06, "loss": 0.4893, "step": 2204 }, { "epoch": 0.15243691669547182, "grad_norm": 0.9709308390399274, "learning_rate": 9.944177965105795e-06, "loss": 0.3787, "step": 2205 }, { "epoch": 0.15250604908399584, "grad_norm": 0.8546053847673362, "learning_rate": 9.944094524044548e-06, "loss": 0.338, "step": 2206 }, { "epoch": 0.1525751814725199, "grad_norm": 0.9510154777136234, "learning_rate": 9.944011021017898e-06, "loss": 0.3658, "step": 2207 }, { "epoch": 0.1526443138610439, "grad_norm": 0.8887130479321558, "learning_rate": 9.943927456026894e-06, "loss": 0.4862, "step": 2208 }, { "epoch": 0.15271344624956792, "grad_norm": 0.9094054280371024, "learning_rate": 9.943843829072583e-06, "loss": 0.3746, "step": 2209 }, { "epoch": 0.15278257863809194, "grad_norm": 0.8759176750037259, "learning_rate": 9.943760140156011e-06, "loss": 0.3263, "step": 2210 }, { "epoch": 0.15285171102661596, "grad_norm": 0.8596955176923006, "learning_rate": 9.943676389278227e-06, "loss": 0.4102, "step": 2211 }, { "epoch": 0.15292084341514, "grad_norm": 0.8242906490359794, "learning_rate": 9.943592576440284e-06, "loss": 0.3955, "step": 2212 }, { "epoch": 0.15298997580366402, "grad_norm": 0.7580656827850676, "learning_rate": 9.943508701643229e-06, "loss": 0.3441, "step": 2213 }, { "epoch": 0.15305910819218804, "grad_norm": 0.9999363471545452, "learning_rate": 9.943424764888115e-06, "loss": 0.4542, "step": 2214 }, { "epoch": 0.15312824058071206, "grad_norm": 0.8864618312387853, "learning_rate": 9.943340766175993e-06, "loss": 0.4249, "step": 2215 }, { "epoch": 0.15319737296923608, "grad_norm": 0.7825079152774859, "learning_rate": 9.943256705507918e-06, "loss": 0.352, "step": 2216 }, { "epoch": 0.15326650535776012, "grad_norm": 0.9434542166852383, "learning_rate": 9.943172582884938e-06, "loss": 0.3469, "step": 2217 }, { "epoch": 0.15333563774628414, "grad_norm": 0.9291782735550429, "learning_rate": 9.943088398308114e-06, "loss": 0.4171, "step": 2218 }, { "epoch": 0.15340477013480816, "grad_norm": 1.146185607304691, "learning_rate": 9.9430041517785e-06, "loss": 0.5114, "step": 2219 }, { "epoch": 0.15347390252333218, "grad_norm": 0.8850442825184293, "learning_rate": 9.942919843297147e-06, "loss": 0.2807, "step": 2220 }, { "epoch": 0.1535430349118562, "grad_norm": 1.7787135293316965, "learning_rate": 9.942835472865116e-06, "loss": 0.3788, "step": 2221 }, { "epoch": 0.15361216730038021, "grad_norm": 0.8747602512019033, "learning_rate": 9.942751040483462e-06, "loss": 0.5301, "step": 2222 }, { "epoch": 0.15368129968890426, "grad_norm": 0.9763700114639959, "learning_rate": 9.942666546153247e-06, "loss": 0.3825, "step": 2223 }, { "epoch": 0.15375043207742828, "grad_norm": 1.2064114900690261, "learning_rate": 9.942581989875526e-06, "loss": 0.3295, "step": 2224 }, { "epoch": 0.1538195644659523, "grad_norm": 0.8382547577286499, "learning_rate": 9.942497371651359e-06, "loss": 0.3883, "step": 2225 }, { "epoch": 0.15388869685447631, "grad_norm": 0.8419742101341938, "learning_rate": 9.94241269148181e-06, "loss": 0.4525, "step": 2226 }, { "epoch": 0.15395782924300033, "grad_norm": 1.1640309993573534, "learning_rate": 9.942327949367936e-06, "loss": 0.3103, "step": 2227 }, { "epoch": 0.15402696163152438, "grad_norm": 0.8718640883452462, "learning_rate": 9.942243145310803e-06, "loss": 0.2834, "step": 2228 }, { "epoch": 0.1540960940200484, "grad_norm": 0.8220983036488487, "learning_rate": 9.942158279311472e-06, "loss": 0.4148, "step": 2229 }, { "epoch": 0.15416522640857241, "grad_norm": 0.8950913290112248, "learning_rate": 9.942073351371005e-06, "loss": 0.4149, "step": 2230 }, { "epoch": 0.15423435879709643, "grad_norm": 0.8654266967709343, "learning_rate": 9.94198836149047e-06, "loss": 0.3268, "step": 2231 }, { "epoch": 0.15430349118562045, "grad_norm": 0.8350606096292121, "learning_rate": 9.941903309670928e-06, "loss": 0.3739, "step": 2232 }, { "epoch": 0.1543726235741445, "grad_norm": 1.2627586244822715, "learning_rate": 9.941818195913448e-06, "loss": 0.459, "step": 2233 }, { "epoch": 0.15444175596266851, "grad_norm": 1.073409090722232, "learning_rate": 9.941733020219095e-06, "loss": 0.4244, "step": 2234 }, { "epoch": 0.15451088835119253, "grad_norm": 1.1679575319116169, "learning_rate": 9.94164778258894e-06, "loss": 0.3112, "step": 2235 }, { "epoch": 0.15458002073971655, "grad_norm": 0.9046403651506981, "learning_rate": 9.941562483024047e-06, "loss": 0.3962, "step": 2236 }, { "epoch": 0.15464915312824057, "grad_norm": 1.2279779864578733, "learning_rate": 9.941477121525487e-06, "loss": 0.415, "step": 2237 }, { "epoch": 0.15471828551676461, "grad_norm": 0.836033203401662, "learning_rate": 9.941391698094331e-06, "loss": 0.3445, "step": 2238 }, { "epoch": 0.15478741790528863, "grad_norm": 0.8958174696912606, "learning_rate": 9.941306212731648e-06, "loss": 0.3705, "step": 2239 }, { "epoch": 0.15485655029381265, "grad_norm": 1.9319840309770007, "learning_rate": 9.94122066543851e-06, "loss": 0.5188, "step": 2240 }, { "epoch": 0.15492568268233667, "grad_norm": 0.8942086512547877, "learning_rate": 9.941135056215987e-06, "loss": 0.3572, "step": 2241 }, { "epoch": 0.1549948150708607, "grad_norm": 0.9539690089264476, "learning_rate": 9.941049385065156e-06, "loss": 0.3354, "step": 2242 }, { "epoch": 0.15506394745938473, "grad_norm": 1.1411848602795474, "learning_rate": 9.940963651987088e-06, "loss": 0.3823, "step": 2243 }, { "epoch": 0.15513307984790875, "grad_norm": 1.1294410878271826, "learning_rate": 9.940877856982857e-06, "loss": 0.3997, "step": 2244 }, { "epoch": 0.15520221223643277, "grad_norm": 0.9800072665927968, "learning_rate": 9.94079200005354e-06, "loss": 0.3211, "step": 2245 }, { "epoch": 0.1552713446249568, "grad_norm": 0.9068282024033069, "learning_rate": 9.940706081200215e-06, "loss": 0.325, "step": 2246 }, { "epoch": 0.1553404770134808, "grad_norm": 0.9972361148435569, "learning_rate": 9.940620100423953e-06, "loss": 0.423, "step": 2247 }, { "epoch": 0.15540960940200485, "grad_norm": 1.2503335747517617, "learning_rate": 9.940534057725836e-06, "loss": 0.4702, "step": 2248 }, { "epoch": 0.15547874179052887, "grad_norm": 0.8471975002836135, "learning_rate": 9.940447953106942e-06, "loss": 0.3732, "step": 2249 }, { "epoch": 0.1555478741790529, "grad_norm": 1.0221674230236244, "learning_rate": 9.94036178656835e-06, "loss": 0.3562, "step": 2250 }, { "epoch": 0.1556170065675769, "grad_norm": 1.1484226091728689, "learning_rate": 9.94027555811114e-06, "loss": 0.4315, "step": 2251 }, { "epoch": 0.15568613895610092, "grad_norm": 0.8906853583727214, "learning_rate": 9.94018926773639e-06, "loss": 0.3145, "step": 2252 }, { "epoch": 0.15575527134462497, "grad_norm": 1.3152101507042675, "learning_rate": 9.940102915445185e-06, "loss": 0.3364, "step": 2253 }, { "epoch": 0.155824403733149, "grad_norm": 0.9860297204113696, "learning_rate": 9.940016501238606e-06, "loss": 0.468, "step": 2254 }, { "epoch": 0.155893536121673, "grad_norm": 0.9107355001302396, "learning_rate": 9.939930025117734e-06, "loss": 0.4094, "step": 2255 }, { "epoch": 0.15596266851019702, "grad_norm": 0.8130694648190379, "learning_rate": 9.939843487083657e-06, "loss": 0.3062, "step": 2256 }, { "epoch": 0.15603180089872104, "grad_norm": 1.1156137530089791, "learning_rate": 9.939756887137457e-06, "loss": 0.3964, "step": 2257 }, { "epoch": 0.15610093328724509, "grad_norm": 1.023043892690021, "learning_rate": 9.93967022528022e-06, "loss": 0.4397, "step": 2258 }, { "epoch": 0.1561700656757691, "grad_norm": 0.9546383258775542, "learning_rate": 9.93958350151303e-06, "loss": 0.3068, "step": 2259 }, { "epoch": 0.15623919806429312, "grad_norm": 0.8372750812074319, "learning_rate": 9.939496715836979e-06, "loss": 0.354, "step": 2260 }, { "epoch": 0.15630833045281714, "grad_norm": 1.1180398711965314, "learning_rate": 9.93940986825315e-06, "loss": 0.3708, "step": 2261 }, { "epoch": 0.15637746284134116, "grad_norm": 1.1333288339555878, "learning_rate": 9.939322958762634e-06, "loss": 0.4129, "step": 2262 }, { "epoch": 0.1564465952298652, "grad_norm": 0.8644137108207379, "learning_rate": 9.939235987366519e-06, "loss": 0.3617, "step": 2263 }, { "epoch": 0.15651572761838922, "grad_norm": 3.375489861026232, "learning_rate": 9.939148954065893e-06, "loss": 0.3473, "step": 2264 }, { "epoch": 0.15658486000691324, "grad_norm": 0.9055514659925528, "learning_rate": 9.939061858861852e-06, "loss": 0.4834, "step": 2265 }, { "epoch": 0.15665399239543726, "grad_norm": 1.3822375251318288, "learning_rate": 9.938974701755484e-06, "loss": 0.3822, "step": 2266 }, { "epoch": 0.15672312478396128, "grad_norm": 0.8372521611119905, "learning_rate": 9.938887482747882e-06, "loss": 0.3383, "step": 2267 }, { "epoch": 0.15679225717248532, "grad_norm": 1.3027527773612586, "learning_rate": 9.938800201840139e-06, "loss": 0.4688, "step": 2268 }, { "epoch": 0.15686138956100934, "grad_norm": 0.8517090788016538, "learning_rate": 9.938712859033349e-06, "loss": 0.4203, "step": 2269 }, { "epoch": 0.15693052194953336, "grad_norm": 0.8786561333608197, "learning_rate": 9.938625454328608e-06, "loss": 0.3441, "step": 2270 }, { "epoch": 0.15699965433805738, "grad_norm": 0.8036008085745466, "learning_rate": 9.938537987727008e-06, "loss": 0.2926, "step": 2271 }, { "epoch": 0.1570687867265814, "grad_norm": 0.9293284466841026, "learning_rate": 9.938450459229649e-06, "loss": 0.4094, "step": 2272 }, { "epoch": 0.15713791911510544, "grad_norm": 0.9009179929888945, "learning_rate": 9.938362868837627e-06, "loss": 0.4555, "step": 2273 }, { "epoch": 0.15720705150362946, "grad_norm": 0.9938297133735956, "learning_rate": 9.93827521655204e-06, "loss": 0.3968, "step": 2274 }, { "epoch": 0.15727618389215348, "grad_norm": 0.8810083982760225, "learning_rate": 9.938187502373984e-06, "loss": 0.4262, "step": 2275 }, { "epoch": 0.1573453162806775, "grad_norm": 0.886299945461153, "learning_rate": 9.938099726304562e-06, "loss": 0.477, "step": 2276 }, { "epoch": 0.1574144486692015, "grad_norm": 0.9522092557785992, "learning_rate": 9.93801188834487e-06, "loss": 0.3562, "step": 2277 }, { "epoch": 0.15748358105772556, "grad_norm": 0.8905009879917956, "learning_rate": 9.937923988496014e-06, "loss": 0.3236, "step": 2278 }, { "epoch": 0.15755271344624958, "grad_norm": 0.8310949240030493, "learning_rate": 9.93783602675909e-06, "loss": 0.4709, "step": 2279 }, { "epoch": 0.1576218458347736, "grad_norm": 0.9360243314349969, "learning_rate": 9.937748003135205e-06, "loss": 0.4594, "step": 2280 }, { "epoch": 0.1576909782232976, "grad_norm": 0.7983175150256617, "learning_rate": 9.937659917625462e-06, "loss": 0.4118, "step": 2281 }, { "epoch": 0.15776011061182163, "grad_norm": 1.1052265521058593, "learning_rate": 9.937571770230962e-06, "loss": 0.3753, "step": 2282 }, { "epoch": 0.15782924300034565, "grad_norm": 1.4783686906788143, "learning_rate": 9.937483560952812e-06, "loss": 0.48, "step": 2283 }, { "epoch": 0.1578983753888697, "grad_norm": 0.8603680276014567, "learning_rate": 9.937395289792115e-06, "loss": 0.3421, "step": 2284 }, { "epoch": 0.1579675077773937, "grad_norm": 0.8537084012585194, "learning_rate": 9.93730695674998e-06, "loss": 0.3711, "step": 2285 }, { "epoch": 0.15803664016591773, "grad_norm": 0.9189272719398635, "learning_rate": 9.937218561827516e-06, "loss": 0.4041, "step": 2286 }, { "epoch": 0.15810577255444175, "grad_norm": 1.098208005329363, "learning_rate": 9.937130105025825e-06, "loss": 0.4504, "step": 2287 }, { "epoch": 0.15817490494296577, "grad_norm": 0.9554334132119576, "learning_rate": 9.937041586346022e-06, "loss": 0.4108, "step": 2288 }, { "epoch": 0.1582440373314898, "grad_norm": 0.9869166206284543, "learning_rate": 9.936953005789209e-06, "loss": 0.3061, "step": 2289 }, { "epoch": 0.15831316972001383, "grad_norm": 0.8849157884269467, "learning_rate": 9.936864363356504e-06, "loss": 0.4645, "step": 2290 }, { "epoch": 0.15838230210853785, "grad_norm": 0.946761024139192, "learning_rate": 9.936775659049012e-06, "loss": 0.441, "step": 2291 }, { "epoch": 0.15845143449706187, "grad_norm": 0.88019211583882, "learning_rate": 9.93668689286785e-06, "loss": 0.363, "step": 2292 }, { "epoch": 0.15852056688558588, "grad_norm": 1.009982480227393, "learning_rate": 9.936598064814124e-06, "loss": 0.4373, "step": 2293 }, { "epoch": 0.15858969927410993, "grad_norm": 0.8494628673471848, "learning_rate": 9.936509174888954e-06, "loss": 0.4986, "step": 2294 }, { "epoch": 0.15865883166263395, "grad_norm": 0.9173482297654316, "learning_rate": 9.93642022309345e-06, "loss": 0.366, "step": 2295 }, { "epoch": 0.15872796405115797, "grad_norm": 0.8442645473554644, "learning_rate": 9.936331209428728e-06, "loss": 0.2982, "step": 2296 }, { "epoch": 0.15879709643968198, "grad_norm": 0.9282269452697947, "learning_rate": 9.936242133895903e-06, "loss": 0.4696, "step": 2297 }, { "epoch": 0.158866228828206, "grad_norm": 0.8365913186482182, "learning_rate": 9.936152996496094e-06, "loss": 0.4516, "step": 2298 }, { "epoch": 0.15893536121673005, "grad_norm": 0.867979142765249, "learning_rate": 9.936063797230415e-06, "loss": 0.3084, "step": 2299 }, { "epoch": 0.15900449360525407, "grad_norm": 0.7890094972617437, "learning_rate": 9.935974536099985e-06, "loss": 0.3497, "step": 2300 }, { "epoch": 0.15907362599377808, "grad_norm": 0.8435413493779716, "learning_rate": 9.935885213105922e-06, "loss": 0.4183, "step": 2301 }, { "epoch": 0.1591427583823021, "grad_norm": 0.8635781418638291, "learning_rate": 9.935795828249347e-06, "loss": 0.3698, "step": 2302 }, { "epoch": 0.15921189077082612, "grad_norm": 0.8104917939586048, "learning_rate": 9.93570638153138e-06, "loss": 0.3163, "step": 2303 }, { "epoch": 0.15928102315935017, "grad_norm": 0.8578783532226539, "learning_rate": 9.935616872953142e-06, "loss": 0.4592, "step": 2304 }, { "epoch": 0.15935015554787418, "grad_norm": 0.8345919680525129, "learning_rate": 9.935527302515753e-06, "loss": 0.3432, "step": 2305 }, { "epoch": 0.1594192879363982, "grad_norm": 0.8393526225903748, "learning_rate": 9.935437670220338e-06, "loss": 0.355, "step": 2306 }, { "epoch": 0.15948842032492222, "grad_norm": 0.8952752618400737, "learning_rate": 9.93534797606802e-06, "loss": 0.3305, "step": 2307 }, { "epoch": 0.15955755271344624, "grad_norm": 0.9147561212349135, "learning_rate": 9.935258220059922e-06, "loss": 0.4801, "step": 2308 }, { "epoch": 0.15962668510197028, "grad_norm": 0.9281808649230769, "learning_rate": 9.93516840219717e-06, "loss": 0.3338, "step": 2309 }, { "epoch": 0.1596958174904943, "grad_norm": 0.9268822803478655, "learning_rate": 9.935078522480888e-06, "loss": 0.3234, "step": 2310 }, { "epoch": 0.15976494987901832, "grad_norm": 1.1901979416125503, "learning_rate": 9.934988580912205e-06, "loss": 0.4142, "step": 2311 }, { "epoch": 0.15983408226754234, "grad_norm": 2.1650826066974904, "learning_rate": 9.934898577492247e-06, "loss": 0.4436, "step": 2312 }, { "epoch": 0.15990321465606636, "grad_norm": 0.9165680920001008, "learning_rate": 9.934808512222142e-06, "loss": 0.3955, "step": 2313 }, { "epoch": 0.1599723470445904, "grad_norm": 0.8150563857086788, "learning_rate": 9.934718385103019e-06, "loss": 0.3343, "step": 2314 }, { "epoch": 0.16004147943311442, "grad_norm": 0.9102187490556551, "learning_rate": 9.934628196136006e-06, "loss": 0.4436, "step": 2315 }, { "epoch": 0.16011061182163844, "grad_norm": 1.0755798662548317, "learning_rate": 9.934537945322236e-06, "loss": 0.4, "step": 2316 }, { "epoch": 0.16017974421016246, "grad_norm": 0.7947148904165497, "learning_rate": 9.93444763266284e-06, "loss": 0.328, "step": 2317 }, { "epoch": 0.16024887659868647, "grad_norm": 0.8706290470341391, "learning_rate": 9.934357258158947e-06, "loss": 0.3777, "step": 2318 }, { "epoch": 0.16031800898721052, "grad_norm": 0.9797612163851098, "learning_rate": 9.934266821811692e-06, "loss": 0.4818, "step": 2319 }, { "epoch": 0.16038714137573454, "grad_norm": 0.9943288131495982, "learning_rate": 9.934176323622209e-06, "loss": 0.3287, "step": 2320 }, { "epoch": 0.16045627376425856, "grad_norm": 1.00063861753859, "learning_rate": 9.93408576359163e-06, "loss": 0.2872, "step": 2321 }, { "epoch": 0.16052540615278257, "grad_norm": 1.0224666323243197, "learning_rate": 9.93399514172109e-06, "loss": 0.4432, "step": 2322 }, { "epoch": 0.1605945385413066, "grad_norm": 1.1387751358205795, "learning_rate": 9.933904458011729e-06, "loss": 0.4372, "step": 2323 }, { "epoch": 0.16066367092983064, "grad_norm": 1.0365096538102383, "learning_rate": 9.93381371246468e-06, "loss": 0.3456, "step": 2324 }, { "epoch": 0.16073280331835466, "grad_norm": 1.1543607930940833, "learning_rate": 9.933722905081079e-06, "loss": 0.3847, "step": 2325 }, { "epoch": 0.16080193570687867, "grad_norm": 0.8654249286726036, "learning_rate": 9.933632035862066e-06, "loss": 0.44, "step": 2326 }, { "epoch": 0.1608710680954027, "grad_norm": 0.8287639681527963, "learning_rate": 9.933541104808782e-06, "loss": 0.3319, "step": 2327 }, { "epoch": 0.1609402004839267, "grad_norm": 0.9024879310368636, "learning_rate": 9.933450111922364e-06, "loss": 0.3054, "step": 2328 }, { "epoch": 0.16100933287245076, "grad_norm": 0.9301800087499686, "learning_rate": 9.933359057203953e-06, "loss": 0.4764, "step": 2329 }, { "epoch": 0.16107846526097477, "grad_norm": 0.9126500531419757, "learning_rate": 9.933267940654689e-06, "loss": 0.3974, "step": 2330 }, { "epoch": 0.1611475976494988, "grad_norm": 0.9992992006510947, "learning_rate": 9.933176762275716e-06, "loss": 0.371, "step": 2331 }, { "epoch": 0.1612167300380228, "grad_norm": 2.1952878332226944, "learning_rate": 9.933085522068175e-06, "loss": 0.384, "step": 2332 }, { "epoch": 0.16128586242654683, "grad_norm": 0.8876033699405846, "learning_rate": 9.932994220033212e-06, "loss": 0.4105, "step": 2333 }, { "epoch": 0.16135499481507087, "grad_norm": 1.1051775867776157, "learning_rate": 9.932902856171968e-06, "loss": 0.4149, "step": 2334 }, { "epoch": 0.1614241272035949, "grad_norm": 1.227588145125398, "learning_rate": 9.932811430485592e-06, "loss": 0.3297, "step": 2335 }, { "epoch": 0.1614932595921189, "grad_norm": 0.9805311580181286, "learning_rate": 9.932719942975227e-06, "loss": 0.5588, "step": 2336 }, { "epoch": 0.16156239198064293, "grad_norm": 0.8513406172965532, "learning_rate": 9.932628393642018e-06, "loss": 0.4515, "step": 2337 }, { "epoch": 0.16163152436916695, "grad_norm": 0.8197592023509714, "learning_rate": 9.93253678248712e-06, "loss": 0.2691, "step": 2338 }, { "epoch": 0.161700656757691, "grad_norm": 0.9524451798484028, "learning_rate": 9.93244510951167e-06, "loss": 0.2424, "step": 2339 }, { "epoch": 0.161769789146215, "grad_norm": 0.8217155599947434, "learning_rate": 9.932353374716827e-06, "loss": 0.4364, "step": 2340 }, { "epoch": 0.16183892153473903, "grad_norm": 0.8492993816551113, "learning_rate": 9.932261578103735e-06, "loss": 0.3856, "step": 2341 }, { "epoch": 0.16190805392326305, "grad_norm": 1.1229273803315114, "learning_rate": 9.932169719673546e-06, "loss": 0.3755, "step": 2342 }, { "epoch": 0.16197718631178706, "grad_norm": 0.8364513909815205, "learning_rate": 9.932077799427412e-06, "loss": 0.3526, "step": 2343 }, { "epoch": 0.1620463187003111, "grad_norm": 0.8122442859336959, "learning_rate": 9.931985817366484e-06, "loss": 0.3704, "step": 2344 }, { "epoch": 0.16211545108883513, "grad_norm": 0.8461462378778238, "learning_rate": 9.931893773491917e-06, "loss": 0.3177, "step": 2345 }, { "epoch": 0.16218458347735915, "grad_norm": 1.0594058270268927, "learning_rate": 9.93180166780486e-06, "loss": 0.3491, "step": 2346 }, { "epoch": 0.16225371586588316, "grad_norm": 0.8573741194650399, "learning_rate": 9.931709500306473e-06, "loss": 0.4948, "step": 2347 }, { "epoch": 0.16232284825440718, "grad_norm": 0.7927072697214654, "learning_rate": 9.931617270997907e-06, "loss": 0.4009, "step": 2348 }, { "epoch": 0.1623919806429312, "grad_norm": 0.8512970078332456, "learning_rate": 9.931524979880318e-06, "loss": 0.3249, "step": 2349 }, { "epoch": 0.16246111303145525, "grad_norm": 0.8786733567065554, "learning_rate": 9.931432626954867e-06, "loss": 0.4141, "step": 2350 }, { "epoch": 0.16253024541997926, "grad_norm": 0.8366029571435332, "learning_rate": 9.931340212222706e-06, "loss": 0.4437, "step": 2351 }, { "epoch": 0.16259937780850328, "grad_norm": 2.4353073272962598, "learning_rate": 9.931247735684997e-06, "loss": 0.3576, "step": 2352 }, { "epoch": 0.1626685101970273, "grad_norm": 0.7693064740850055, "learning_rate": 9.931155197342899e-06, "loss": 0.3059, "step": 2353 }, { "epoch": 0.16273764258555132, "grad_norm": 0.775768410912278, "learning_rate": 9.931062597197569e-06, "loss": 0.3821, "step": 2354 }, { "epoch": 0.16280677497407536, "grad_norm": 0.8100006355530694, "learning_rate": 9.93096993525017e-06, "loss": 0.3589, "step": 2355 }, { "epoch": 0.16287590736259938, "grad_norm": 0.8209191137022268, "learning_rate": 9.93087721150186e-06, "loss": 0.3341, "step": 2356 }, { "epoch": 0.1629450397511234, "grad_norm": 0.8973857148302887, "learning_rate": 9.930784425953807e-06, "loss": 0.3367, "step": 2357 }, { "epoch": 0.16301417213964742, "grad_norm": 0.8074328033181748, "learning_rate": 9.93069157860717e-06, "loss": 0.4385, "step": 2358 }, { "epoch": 0.16308330452817144, "grad_norm": 0.8470764793571528, "learning_rate": 9.930598669463112e-06, "loss": 0.347, "step": 2359 }, { "epoch": 0.16315243691669548, "grad_norm": 0.825752819582706, "learning_rate": 9.9305056985228e-06, "loss": 0.3045, "step": 2360 }, { "epoch": 0.1632215693052195, "grad_norm": 0.9138181785500744, "learning_rate": 9.930412665787399e-06, "loss": 0.4538, "step": 2361 }, { "epoch": 0.16329070169374352, "grad_norm": 0.859372364956485, "learning_rate": 9.930319571258072e-06, "loss": 0.42, "step": 2362 }, { "epoch": 0.16335983408226754, "grad_norm": 1.0166366813031629, "learning_rate": 9.930226414935989e-06, "loss": 0.3415, "step": 2363 }, { "epoch": 0.16342896647079155, "grad_norm": 0.9540173736927079, "learning_rate": 9.930133196822314e-06, "loss": 0.3904, "step": 2364 }, { "epoch": 0.1634980988593156, "grad_norm": 0.822672559481457, "learning_rate": 9.930039916918219e-06, "loss": 0.4698, "step": 2365 }, { "epoch": 0.16356723124783962, "grad_norm": 0.9153849750783989, "learning_rate": 9.929946575224872e-06, "loss": 0.4074, "step": 2366 }, { "epoch": 0.16363636363636364, "grad_norm": 1.478049401072593, "learning_rate": 9.929853171743443e-06, "loss": 0.3129, "step": 2367 }, { "epoch": 0.16370549602488765, "grad_norm": 0.8586694795695871, "learning_rate": 9.929759706475101e-06, "loss": 0.3671, "step": 2368 }, { "epoch": 0.16377462841341167, "grad_norm": 0.9359662850898409, "learning_rate": 9.929666179421019e-06, "loss": 0.4316, "step": 2369 }, { "epoch": 0.16384376080193572, "grad_norm": 0.9484082438532152, "learning_rate": 9.92957259058237e-06, "loss": 0.3913, "step": 2370 }, { "epoch": 0.16391289319045974, "grad_norm": 0.9230306271624433, "learning_rate": 9.929478939960324e-06, "loss": 0.2919, "step": 2371 }, { "epoch": 0.16398202557898375, "grad_norm": 0.8692624465038225, "learning_rate": 9.929385227556058e-06, "loss": 0.4399, "step": 2372 }, { "epoch": 0.16405115796750777, "grad_norm": 0.8240575887526013, "learning_rate": 9.929291453370745e-06, "loss": 0.3868, "step": 2373 }, { "epoch": 0.1641202903560318, "grad_norm": 0.9828396030358378, "learning_rate": 9.92919761740556e-06, "loss": 0.392, "step": 2374 }, { "epoch": 0.16418942274455584, "grad_norm": 0.8776535620245866, "learning_rate": 9.929103719661678e-06, "loss": 0.3518, "step": 2375 }, { "epoch": 0.16425855513307985, "grad_norm": 0.8653053296772344, "learning_rate": 9.929009760140278e-06, "loss": 0.5241, "step": 2376 }, { "epoch": 0.16432768752160387, "grad_norm": 0.7909627234642114, "learning_rate": 9.928915738842537e-06, "loss": 0.3155, "step": 2377 }, { "epoch": 0.1643968199101279, "grad_norm": 0.8643897538062382, "learning_rate": 9.928821655769634e-06, "loss": 0.3643, "step": 2378 }, { "epoch": 0.1644659522986519, "grad_norm": 0.9302036839119637, "learning_rate": 9.928727510922746e-06, "loss": 0.4101, "step": 2379 }, { "epoch": 0.16453508468717595, "grad_norm": 1.2431972490599894, "learning_rate": 9.928633304303056e-06, "loss": 0.3792, "step": 2380 }, { "epoch": 0.16460421707569997, "grad_norm": 0.8335257729305287, "learning_rate": 9.928539035911742e-06, "loss": 0.3502, "step": 2381 }, { "epoch": 0.164673349464224, "grad_norm": 0.8212528899095348, "learning_rate": 9.928444705749988e-06, "loss": 0.402, "step": 2382 }, { "epoch": 0.164742481852748, "grad_norm": 1.4222949368598636, "learning_rate": 9.928350313818973e-06, "loss": 0.4606, "step": 2383 }, { "epoch": 0.16481161424127203, "grad_norm": 1.0089140164449426, "learning_rate": 9.928255860119884e-06, "loss": 0.3619, "step": 2384 }, { "epoch": 0.16488074662979607, "grad_norm": 0.9396988440842541, "learning_rate": 9.9281613446539e-06, "loss": 0.2547, "step": 2385 }, { "epoch": 0.1649498790183201, "grad_norm": 1.0210542659755382, "learning_rate": 9.92806676742221e-06, "loss": 0.4115, "step": 2386 }, { "epoch": 0.1650190114068441, "grad_norm": 1.1224612238300569, "learning_rate": 9.927972128425996e-06, "loss": 0.4078, "step": 2387 }, { "epoch": 0.16508814379536813, "grad_norm": 0.776807995366039, "learning_rate": 9.927877427666448e-06, "loss": 0.3697, "step": 2388 }, { "epoch": 0.16515727618389214, "grad_norm": 1.0386017255125515, "learning_rate": 9.927782665144749e-06, "loss": 0.3717, "step": 2389 }, { "epoch": 0.1652264085724162, "grad_norm": 0.9231385056422784, "learning_rate": 9.92768784086209e-06, "loss": 0.5118, "step": 2390 }, { "epoch": 0.1652955409609402, "grad_norm": 0.9738583491538179, "learning_rate": 9.927592954819656e-06, "loss": 0.3349, "step": 2391 }, { "epoch": 0.16536467334946423, "grad_norm": 0.9950803020542716, "learning_rate": 9.92749800701864e-06, "loss": 0.3005, "step": 2392 }, { "epoch": 0.16543380573798824, "grad_norm": 0.8790492941355507, "learning_rate": 9.927402997460227e-06, "loss": 0.437, "step": 2393 }, { "epoch": 0.16550293812651226, "grad_norm": 0.8350764457771427, "learning_rate": 9.927307926145613e-06, "loss": 0.423, "step": 2394 }, { "epoch": 0.1655720705150363, "grad_norm": 1.0151885680042618, "learning_rate": 9.927212793075986e-06, "loss": 0.3206, "step": 2395 }, { "epoch": 0.16564120290356033, "grad_norm": 0.7917808598896355, "learning_rate": 9.92711759825254e-06, "loss": 0.3339, "step": 2396 }, { "epoch": 0.16571033529208434, "grad_norm": 0.9167196091737362, "learning_rate": 9.927022341676469e-06, "loss": 0.4657, "step": 2397 }, { "epoch": 0.16577946768060836, "grad_norm": 0.9220201102261238, "learning_rate": 9.926927023348965e-06, "loss": 0.4067, "step": 2398 }, { "epoch": 0.16584860006913238, "grad_norm": 0.8454589166849437, "learning_rate": 9.926831643271223e-06, "loss": 0.3548, "step": 2399 }, { "epoch": 0.16591773245765642, "grad_norm": 0.8867737301828976, "learning_rate": 9.926736201444438e-06, "loss": 0.3545, "step": 2400 }, { "epoch": 0.16598686484618044, "grad_norm": 0.8312322449576478, "learning_rate": 9.926640697869807e-06, "loss": 0.4866, "step": 2401 }, { "epoch": 0.16605599723470446, "grad_norm": 0.8127509294948085, "learning_rate": 9.926545132548527e-06, "loss": 0.3493, "step": 2402 }, { "epoch": 0.16612512962322848, "grad_norm": 0.7654798558745033, "learning_rate": 9.926449505481796e-06, "loss": 0.3186, "step": 2403 }, { "epoch": 0.1661942620117525, "grad_norm": 0.8491889340357324, "learning_rate": 9.926353816670813e-06, "loss": 0.5092, "step": 2404 }, { "epoch": 0.16626339440027654, "grad_norm": 0.8929882327304002, "learning_rate": 9.926258066116775e-06, "loss": 0.4699, "step": 2405 }, { "epoch": 0.16633252678880056, "grad_norm": 0.7798853299139953, "learning_rate": 9.926162253820884e-06, "loss": 0.2866, "step": 2406 }, { "epoch": 0.16640165917732458, "grad_norm": 1.576021124910506, "learning_rate": 9.926066379784339e-06, "loss": 0.4281, "step": 2407 }, { "epoch": 0.1664707915658486, "grad_norm": 0.86636475450864, "learning_rate": 9.925970444008344e-06, "loss": 0.4931, "step": 2408 }, { "epoch": 0.16653992395437262, "grad_norm": 0.8401706523766624, "learning_rate": 9.9258744464941e-06, "loss": 0.3591, "step": 2409 }, { "epoch": 0.16660905634289666, "grad_norm": 0.8427694029742349, "learning_rate": 9.92577838724281e-06, "loss": 0.308, "step": 2410 }, { "epoch": 0.16667818873142068, "grad_norm": 0.9946439092140078, "learning_rate": 9.92568226625568e-06, "loss": 0.4128, "step": 2411 }, { "epoch": 0.1667473211199447, "grad_norm": 0.9747248604234363, "learning_rate": 9.925586083533912e-06, "loss": 0.4616, "step": 2412 }, { "epoch": 0.16681645350846872, "grad_norm": 0.8429545648346137, "learning_rate": 9.925489839078713e-06, "loss": 0.3929, "step": 2413 }, { "epoch": 0.16688558589699273, "grad_norm": 2.335077924089308, "learning_rate": 9.925393532891288e-06, "loss": 0.3923, "step": 2414 }, { "epoch": 0.16695471828551675, "grad_norm": 0.8485806393915821, "learning_rate": 9.925297164972846e-06, "loss": 0.4654, "step": 2415 }, { "epoch": 0.1670238506740408, "grad_norm": 0.8606190660299324, "learning_rate": 9.925200735324592e-06, "loss": 0.3437, "step": 2416 }, { "epoch": 0.16709298306256481, "grad_norm": 0.8225194421494952, "learning_rate": 9.925104243947737e-06, "loss": 0.3012, "step": 2417 }, { "epoch": 0.16716211545108883, "grad_norm": 0.8944359359429045, "learning_rate": 9.92500769084349e-06, "loss": 0.3865, "step": 2418 }, { "epoch": 0.16723124783961285, "grad_norm": 0.8841726587620888, "learning_rate": 9.92491107601306e-06, "loss": 0.4751, "step": 2419 }, { "epoch": 0.16730038022813687, "grad_norm": 1.0981842373850568, "learning_rate": 9.92481439945766e-06, "loss": 0.3936, "step": 2420 }, { "epoch": 0.16736951261666091, "grad_norm": 0.8875108985019889, "learning_rate": 9.924717661178497e-06, "loss": 0.3186, "step": 2421 }, { "epoch": 0.16743864500518493, "grad_norm": 1.555051134048125, "learning_rate": 9.924620861176791e-06, "loss": 0.4255, "step": 2422 }, { "epoch": 0.16750777739370895, "grad_norm": 0.7790305276864006, "learning_rate": 9.924523999453749e-06, "loss": 0.3411, "step": 2423 }, { "epoch": 0.16757690978223297, "grad_norm": 0.813808889633638, "learning_rate": 9.924427076010585e-06, "loss": 0.3105, "step": 2424 }, { "epoch": 0.167646042170757, "grad_norm": 1.0405457198590469, "learning_rate": 9.924330090848517e-06, "loss": 0.3708, "step": 2425 }, { "epoch": 0.16771517455928103, "grad_norm": 0.8465776683421505, "learning_rate": 9.924233043968761e-06, "loss": 0.4265, "step": 2426 }, { "epoch": 0.16778430694780505, "grad_norm": 1.7599187219700252, "learning_rate": 9.924135935372528e-06, "loss": 0.4261, "step": 2427 }, { "epoch": 0.16785343933632907, "grad_norm": 0.8350879360448276, "learning_rate": 9.924038765061042e-06, "loss": 0.3898, "step": 2428 }, { "epoch": 0.1679225717248531, "grad_norm": 0.9040626125475776, "learning_rate": 9.923941533035515e-06, "loss": 0.4193, "step": 2429 }, { "epoch": 0.1679917041133771, "grad_norm": 0.8408943955341646, "learning_rate": 9.923844239297169e-06, "loss": 0.4609, "step": 2430 }, { "epoch": 0.16806083650190115, "grad_norm": 0.7820681751363094, "learning_rate": 9.923746883847222e-06, "loss": 0.3159, "step": 2431 }, { "epoch": 0.16812996889042517, "grad_norm": 1.1883984392888007, "learning_rate": 9.923649466686895e-06, "loss": 0.4002, "step": 2432 }, { "epoch": 0.1681991012789492, "grad_norm": 0.8191323559766739, "learning_rate": 9.923551987817408e-06, "loss": 0.4305, "step": 2433 }, { "epoch": 0.1682682336674732, "grad_norm": 0.8261714232780881, "learning_rate": 9.923454447239984e-06, "loss": 0.3658, "step": 2434 }, { "epoch": 0.16833736605599722, "grad_norm": 0.7474099620327185, "learning_rate": 9.923356844955845e-06, "loss": 0.2588, "step": 2435 }, { "epoch": 0.16840649844452127, "grad_norm": 0.8572284703471669, "learning_rate": 9.923259180966214e-06, "loss": 0.4219, "step": 2436 }, { "epoch": 0.1684756308330453, "grad_norm": 0.8740746952629185, "learning_rate": 9.923161455272315e-06, "loss": 0.4521, "step": 2437 }, { "epoch": 0.1685447632215693, "grad_norm": 0.8211352239136063, "learning_rate": 9.923063667875372e-06, "loss": 0.3593, "step": 2438 }, { "epoch": 0.16861389561009332, "grad_norm": 0.8060084026557586, "learning_rate": 9.922965818776613e-06, "loss": 0.366, "step": 2439 }, { "epoch": 0.16868302799861734, "grad_norm": 1.5112204952408639, "learning_rate": 9.922867907977262e-06, "loss": 0.4766, "step": 2440 }, { "epoch": 0.1687521603871414, "grad_norm": 1.1705754871804162, "learning_rate": 9.922769935478546e-06, "loss": 0.4139, "step": 2441 }, { "epoch": 0.1688212927756654, "grad_norm": 0.8156576009281519, "learning_rate": 9.922671901281696e-06, "loss": 0.3605, "step": 2442 }, { "epoch": 0.16889042516418942, "grad_norm": 0.7615238186556027, "learning_rate": 9.922573805387939e-06, "loss": 0.3875, "step": 2443 }, { "epoch": 0.16895955755271344, "grad_norm": 0.8676232477921842, "learning_rate": 9.922475647798501e-06, "loss": 0.4115, "step": 2444 }, { "epoch": 0.16902868994123746, "grad_norm": 0.8607818506455104, "learning_rate": 9.922377428514617e-06, "loss": 0.3288, "step": 2445 }, { "epoch": 0.1690978223297615, "grad_norm": 0.9922696909040076, "learning_rate": 9.922279147537517e-06, "loss": 0.3825, "step": 2446 }, { "epoch": 0.16916695471828552, "grad_norm": 0.8643648946538176, "learning_rate": 9.922180804868431e-06, "loss": 0.4716, "step": 2447 }, { "epoch": 0.16923608710680954, "grad_norm": 1.0985309109413621, "learning_rate": 9.922082400508594e-06, "loss": 0.4096, "step": 2448 }, { "epoch": 0.16930521949533356, "grad_norm": 0.852847392896029, "learning_rate": 9.921983934459237e-06, "loss": 0.3177, "step": 2449 }, { "epoch": 0.16937435188385758, "grad_norm": 1.107735111900301, "learning_rate": 9.921885406721595e-06, "loss": 0.3433, "step": 2450 }, { "epoch": 0.16944348427238162, "grad_norm": 0.7984037423713468, "learning_rate": 9.921786817296904e-06, "loss": 0.4882, "step": 2451 }, { "epoch": 0.16951261666090564, "grad_norm": 1.1876493228877973, "learning_rate": 9.921688166186398e-06, "loss": 0.3284, "step": 2452 }, { "epoch": 0.16958174904942966, "grad_norm": 0.7677689458920806, "learning_rate": 9.921589453391312e-06, "loss": 0.343, "step": 2453 }, { "epoch": 0.16965088143795368, "grad_norm": 0.817783989947749, "learning_rate": 9.921490678912888e-06, "loss": 0.3991, "step": 2454 }, { "epoch": 0.1697200138264777, "grad_norm": 0.9033816577806124, "learning_rate": 9.92139184275236e-06, "loss": 0.3723, "step": 2455 }, { "epoch": 0.16978914621500174, "grad_norm": 1.0586224335035124, "learning_rate": 9.921292944910968e-06, "loss": 0.3662, "step": 2456 }, { "epoch": 0.16985827860352576, "grad_norm": 0.8819458827847342, "learning_rate": 9.921193985389952e-06, "loss": 0.456, "step": 2457 }, { "epoch": 0.16992741099204978, "grad_norm": 0.807694592497585, "learning_rate": 9.921094964190549e-06, "loss": 0.4488, "step": 2458 }, { "epoch": 0.1699965433805738, "grad_norm": 0.8568033114078887, "learning_rate": 9.920995881314005e-06, "loss": 0.3695, "step": 2459 }, { "epoch": 0.1700656757690978, "grad_norm": 0.7944898741975055, "learning_rate": 9.920896736761559e-06, "loss": 0.3379, "step": 2460 }, { "epoch": 0.17013480815762186, "grad_norm": 0.9283229850135113, "learning_rate": 9.920797530534453e-06, "loss": 0.4074, "step": 2461 }, { "epoch": 0.17020394054614588, "grad_norm": 0.8128619026140776, "learning_rate": 9.920698262633935e-06, "loss": 0.4884, "step": 2462 }, { "epoch": 0.1702730729346699, "grad_norm": 0.8103082436250365, "learning_rate": 9.920598933061243e-06, "loss": 0.329, "step": 2463 }, { "epoch": 0.1703422053231939, "grad_norm": 0.8667030996258159, "learning_rate": 9.920499541817625e-06, "loss": 0.3624, "step": 2464 }, { "epoch": 0.17041133771171793, "grad_norm": 0.797488421055223, "learning_rate": 9.920400088904326e-06, "loss": 0.4056, "step": 2465 }, { "epoch": 0.17048047010024198, "grad_norm": 0.8748077369403139, "learning_rate": 9.920300574322593e-06, "loss": 0.3768, "step": 2466 }, { "epoch": 0.170549602488766, "grad_norm": 1.1448742604412359, "learning_rate": 9.920200998073672e-06, "loss": 0.3435, "step": 2467 }, { "epoch": 0.17061873487729, "grad_norm": 0.8197689093929134, "learning_rate": 9.920101360158814e-06, "loss": 0.3526, "step": 2468 }, { "epoch": 0.17068786726581403, "grad_norm": 0.8517074028223807, "learning_rate": 9.920001660579265e-06, "loss": 0.4525, "step": 2469 }, { "epoch": 0.17075699965433805, "grad_norm": 0.8294616891642878, "learning_rate": 9.919901899336275e-06, "loss": 0.3706, "step": 2470 }, { "epoch": 0.1708261320428621, "grad_norm": 0.8280652128554403, "learning_rate": 9.919802076431095e-06, "loss": 0.3611, "step": 2471 }, { "epoch": 0.1708952644313861, "grad_norm": 0.8449494533288807, "learning_rate": 9.919702191864975e-06, "loss": 0.473, "step": 2472 }, { "epoch": 0.17096439681991013, "grad_norm": 0.8679753790574313, "learning_rate": 9.919602245639168e-06, "loss": 0.369, "step": 2473 }, { "epoch": 0.17103352920843415, "grad_norm": 0.7485143974284291, "learning_rate": 9.919502237754926e-06, "loss": 0.2914, "step": 2474 }, { "epoch": 0.17110266159695817, "grad_norm": 0.8536895273239321, "learning_rate": 9.919402168213504e-06, "loss": 0.4033, "step": 2475 }, { "epoch": 0.17117179398548218, "grad_norm": 0.8417063766579639, "learning_rate": 9.919302037016153e-06, "loss": 0.4905, "step": 2476 }, { "epoch": 0.17124092637400623, "grad_norm": 0.9669195756511342, "learning_rate": 9.91920184416413e-06, "loss": 0.4246, "step": 2477 }, { "epoch": 0.17131005876253025, "grad_norm": 1.155079034965693, "learning_rate": 9.919101589658694e-06, "loss": 0.3621, "step": 2478 }, { "epoch": 0.17137919115105427, "grad_norm": 0.9424797147015653, "learning_rate": 9.919001273501095e-06, "loss": 0.406, "step": 2479 }, { "epoch": 0.17144832353957828, "grad_norm": 1.2110705289719481, "learning_rate": 9.918900895692594e-06, "loss": 0.3715, "step": 2480 }, { "epoch": 0.1715174559281023, "grad_norm": 0.8231218982373577, "learning_rate": 9.918800456234448e-06, "loss": 0.2895, "step": 2481 }, { "epoch": 0.17158658831662635, "grad_norm": 0.809420863293698, "learning_rate": 9.918699955127917e-06, "loss": 0.3849, "step": 2482 }, { "epoch": 0.17165572070515037, "grad_norm": 0.8593207250052779, "learning_rate": 9.91859939237426e-06, "loss": 0.4653, "step": 2483 }, { "epoch": 0.17172485309367438, "grad_norm": 0.7724899901921367, "learning_rate": 9.918498767974735e-06, "loss": 0.2963, "step": 2484 }, { "epoch": 0.1717939854821984, "grad_norm": 0.8519049184818137, "learning_rate": 9.918398081930606e-06, "loss": 0.3489, "step": 2485 }, { "epoch": 0.17186311787072242, "grad_norm": 0.9151478562262539, "learning_rate": 9.918297334243136e-06, "loss": 0.4338, "step": 2486 }, { "epoch": 0.17193225025924647, "grad_norm": 0.8124494650561241, "learning_rate": 9.918196524913585e-06, "loss": 0.5388, "step": 2487 }, { "epoch": 0.17200138264777048, "grad_norm": 0.8246416095063529, "learning_rate": 9.918095653943219e-06, "loss": 0.3014, "step": 2488 }, { "epoch": 0.1720705150362945, "grad_norm": 0.7758957788706868, "learning_rate": 9.917994721333301e-06, "loss": 0.4388, "step": 2489 }, { "epoch": 0.17213964742481852, "grad_norm": 0.8817939759699779, "learning_rate": 9.917893727085094e-06, "loss": 0.4652, "step": 2490 }, { "epoch": 0.17220877981334254, "grad_norm": 0.9640541915779056, "learning_rate": 9.917792671199866e-06, "loss": 0.3268, "step": 2491 }, { "epoch": 0.17227791220186658, "grad_norm": 0.8023553222761736, "learning_rate": 9.917691553678883e-06, "loss": 0.3352, "step": 2492 }, { "epoch": 0.1723470445903906, "grad_norm": 0.8981035529685311, "learning_rate": 9.917590374523411e-06, "loss": 0.3398, "step": 2493 }, { "epoch": 0.17241617697891462, "grad_norm": 0.8338338185552243, "learning_rate": 9.917489133734721e-06, "loss": 0.436, "step": 2494 }, { "epoch": 0.17248530936743864, "grad_norm": 0.818886051752515, "learning_rate": 9.91738783131408e-06, "loss": 0.3356, "step": 2495 }, { "epoch": 0.17255444175596266, "grad_norm": 0.8389848650773414, "learning_rate": 9.917286467262758e-06, "loss": 0.3283, "step": 2496 }, { "epoch": 0.1726235741444867, "grad_norm": 0.8643710605878716, "learning_rate": 9.917185041582028e-06, "loss": 0.4705, "step": 2497 }, { "epoch": 0.17269270653301072, "grad_norm": 0.8262749373742488, "learning_rate": 9.917083554273155e-06, "loss": 0.4221, "step": 2498 }, { "epoch": 0.17276183892153474, "grad_norm": 0.953527492840217, "learning_rate": 9.916982005337416e-06, "loss": 0.3611, "step": 2499 }, { "epoch": 0.17283097131005876, "grad_norm": 0.9067127017465069, "learning_rate": 9.916880394776083e-06, "loss": 0.3881, "step": 2500 }, { "epoch": 0.17290010369858277, "grad_norm": 1.1452253369832734, "learning_rate": 9.916778722590427e-06, "loss": 0.4251, "step": 2501 }, { "epoch": 0.17296923608710682, "grad_norm": 0.9433388043077524, "learning_rate": 9.916676988781724e-06, "loss": 0.3685, "step": 2502 }, { "epoch": 0.17303836847563084, "grad_norm": 0.8585598908160188, "learning_rate": 9.916575193351251e-06, "loss": 0.361, "step": 2503 }, { "epoch": 0.17310750086415486, "grad_norm": 0.897281841082227, "learning_rate": 9.916473336300281e-06, "loss": 0.4424, "step": 2504 }, { "epoch": 0.17317663325267887, "grad_norm": 0.9062871722461415, "learning_rate": 9.916371417630092e-06, "loss": 0.4351, "step": 2505 }, { "epoch": 0.1732457656412029, "grad_norm": 2.322196818118664, "learning_rate": 9.916269437341961e-06, "loss": 0.2968, "step": 2506 }, { "epoch": 0.17331489802972694, "grad_norm": 0.8196290979135473, "learning_rate": 9.916167395437167e-06, "loss": 0.4036, "step": 2507 }, { "epoch": 0.17338403041825096, "grad_norm": 1.057749193361415, "learning_rate": 9.916065291916985e-06, "loss": 0.4477, "step": 2508 }, { "epoch": 0.17345316280677497, "grad_norm": 0.7707863145094023, "learning_rate": 9.915963126782699e-06, "loss": 0.3055, "step": 2509 }, { "epoch": 0.173522295195299, "grad_norm": 0.7584751327635614, "learning_rate": 9.91586090003559e-06, "loss": 0.3576, "step": 2510 }, { "epoch": 0.173591427583823, "grad_norm": 0.8678950386405256, "learning_rate": 9.915758611676936e-06, "loss": 0.4194, "step": 2511 }, { "epoch": 0.17366055997234706, "grad_norm": 0.8554019634152028, "learning_rate": 9.91565626170802e-06, "loss": 0.3798, "step": 2512 }, { "epoch": 0.17372969236087107, "grad_norm": 0.8173026647041782, "learning_rate": 9.915553850130125e-06, "loss": 0.3044, "step": 2513 }, { "epoch": 0.1737988247493951, "grad_norm": 0.8321708670744824, "learning_rate": 9.915451376944535e-06, "loss": 0.4585, "step": 2514 }, { "epoch": 0.1738679571379191, "grad_norm": 0.9805462443530797, "learning_rate": 9.915348842152534e-06, "loss": 0.4196, "step": 2515 }, { "epoch": 0.17393708952644313, "grad_norm": 0.9053752292276606, "learning_rate": 9.915246245755407e-06, "loss": 0.3736, "step": 2516 }, { "epoch": 0.17400622191496717, "grad_norm": 0.7872817937531279, "learning_rate": 9.915143587754441e-06, "loss": 0.3469, "step": 2517 }, { "epoch": 0.1740753543034912, "grad_norm": 0.8692423879938466, "learning_rate": 9.91504086815092e-06, "loss": 0.3558, "step": 2518 }, { "epoch": 0.1741444866920152, "grad_norm": 0.8116444118665309, "learning_rate": 9.914938086946134e-06, "loss": 0.4746, "step": 2519 }, { "epoch": 0.17421361908053923, "grad_norm": 0.9173602610963739, "learning_rate": 9.91483524414137e-06, "loss": 0.3453, "step": 2520 }, { "epoch": 0.17428275146906325, "grad_norm": 0.7417770637589368, "learning_rate": 9.914732339737918e-06, "loss": 0.352, "step": 2521 }, { "epoch": 0.1743518838575873, "grad_norm": 0.8255540758371245, "learning_rate": 9.914629373737067e-06, "loss": 0.4506, "step": 2522 }, { "epoch": 0.1744210162461113, "grad_norm": 2.3698012595807083, "learning_rate": 9.914526346140105e-06, "loss": 0.372, "step": 2523 }, { "epoch": 0.17449014863463533, "grad_norm": 0.9764233248362071, "learning_rate": 9.914423256948326e-06, "loss": 0.3828, "step": 2524 }, { "epoch": 0.17455928102315935, "grad_norm": 0.815955658081978, "learning_rate": 9.914320106163024e-06, "loss": 0.3649, "step": 2525 }, { "epoch": 0.17462841341168336, "grad_norm": 0.8995680010182856, "learning_rate": 9.914216893785486e-06, "loss": 0.4235, "step": 2526 }, { "epoch": 0.1746975458002074, "grad_norm": 0.8931065104985484, "learning_rate": 9.914113619817012e-06, "loss": 0.3491, "step": 2527 }, { "epoch": 0.17476667818873143, "grad_norm": 0.7886579706884839, "learning_rate": 9.914010284258893e-06, "loss": 0.2771, "step": 2528 }, { "epoch": 0.17483581057725545, "grad_norm": 0.8168289526955728, "learning_rate": 9.913906887112423e-06, "loss": 0.4576, "step": 2529 }, { "epoch": 0.17490494296577946, "grad_norm": 0.8313670107645584, "learning_rate": 9.913803428378902e-06, "loss": 0.4044, "step": 2530 }, { "epoch": 0.17497407535430348, "grad_norm": 0.7873187274642032, "learning_rate": 9.913699908059622e-06, "loss": 0.3299, "step": 2531 }, { "epoch": 0.17504320774282753, "grad_norm": 0.9598781411241699, "learning_rate": 9.913596326155883e-06, "loss": 0.4153, "step": 2532 }, { "epoch": 0.17511234013135155, "grad_norm": 0.90950680306098, "learning_rate": 9.913492682668984e-06, "loss": 0.5052, "step": 2533 }, { "epoch": 0.17518147251987556, "grad_norm": 0.8523860651917068, "learning_rate": 9.91338897760022e-06, "loss": 0.4301, "step": 2534 }, { "epoch": 0.17525060490839958, "grad_norm": 0.8626615022638012, "learning_rate": 9.913285210950895e-06, "loss": 0.3171, "step": 2535 }, { "epoch": 0.1753197372969236, "grad_norm": 0.9002234769928683, "learning_rate": 9.91318138272231e-06, "loss": 0.4186, "step": 2536 }, { "epoch": 0.17538886968544765, "grad_norm": 1.1497130358168377, "learning_rate": 9.913077492915762e-06, "loss": 0.4075, "step": 2537 }, { "epoch": 0.17545800207397166, "grad_norm": 0.8706817839520585, "learning_rate": 9.912973541532555e-06, "loss": 0.3794, "step": 2538 }, { "epoch": 0.17552713446249568, "grad_norm": 0.8396579235092663, "learning_rate": 9.912869528573994e-06, "loss": 0.3394, "step": 2539 }, { "epoch": 0.1755962668510197, "grad_norm": 0.9858089079037373, "learning_rate": 9.91276545404138e-06, "loss": 0.5597, "step": 2540 }, { "epoch": 0.17566539923954372, "grad_norm": 0.8670898718170045, "learning_rate": 9.912661317936018e-06, "loss": 0.3802, "step": 2541 }, { "epoch": 0.17573453162806774, "grad_norm": 0.8520285681000853, "learning_rate": 9.912557120259213e-06, "loss": 0.3721, "step": 2542 }, { "epoch": 0.17580366401659178, "grad_norm": 0.8253328853351617, "learning_rate": 9.912452861012271e-06, "loss": 0.4405, "step": 2543 }, { "epoch": 0.1758727964051158, "grad_norm": 0.865486667088832, "learning_rate": 9.912348540196502e-06, "loss": 0.4918, "step": 2544 }, { "epoch": 0.17594192879363982, "grad_norm": 0.7801733927050378, "learning_rate": 9.912244157813209e-06, "loss": 0.3312, "step": 2545 }, { "epoch": 0.17601106118216384, "grad_norm": 0.8102629421519368, "learning_rate": 9.912139713863699e-06, "loss": 0.3252, "step": 2546 }, { "epoch": 0.17608019357068785, "grad_norm": 0.9196044405342038, "learning_rate": 9.912035208349288e-06, "loss": 0.4054, "step": 2547 }, { "epoch": 0.1761493259592119, "grad_norm": 0.9756804960735564, "learning_rate": 9.91193064127128e-06, "loss": 0.4296, "step": 2548 }, { "epoch": 0.17621845834773592, "grad_norm": 0.9496551669073293, "learning_rate": 9.911826012630986e-06, "loss": 0.3721, "step": 2549 }, { "epoch": 0.17628759073625994, "grad_norm": 0.8983901599756391, "learning_rate": 9.91172132242972e-06, "loss": 0.3898, "step": 2550 }, { "epoch": 0.17635672312478395, "grad_norm": 0.8539449032495771, "learning_rate": 9.911616570668793e-06, "loss": 0.472, "step": 2551 }, { "epoch": 0.17642585551330797, "grad_norm": 0.9197813106224562, "learning_rate": 9.911511757349514e-06, "loss": 0.3664, "step": 2552 }, { "epoch": 0.17649498790183202, "grad_norm": 0.8038292251398202, "learning_rate": 9.911406882473205e-06, "loss": 0.3246, "step": 2553 }, { "epoch": 0.17656412029035604, "grad_norm": 0.8721059196315661, "learning_rate": 9.911301946041172e-06, "loss": 0.4487, "step": 2554 }, { "epoch": 0.17663325267888005, "grad_norm": 0.8365109537448328, "learning_rate": 9.911196948054735e-06, "loss": 0.4054, "step": 2555 }, { "epoch": 0.17670238506740407, "grad_norm": 0.7637507003855247, "learning_rate": 9.911091888515208e-06, "loss": 0.3026, "step": 2556 }, { "epoch": 0.1767715174559281, "grad_norm": 0.7595159579335441, "learning_rate": 9.910986767423909e-06, "loss": 0.3085, "step": 2557 }, { "epoch": 0.17684064984445214, "grad_norm": 0.9046294506370124, "learning_rate": 9.910881584782155e-06, "loss": 0.4595, "step": 2558 }, { "epoch": 0.17690978223297615, "grad_norm": 0.9602891576200396, "learning_rate": 9.910776340591263e-06, "loss": 0.2989, "step": 2559 }, { "epoch": 0.17697891462150017, "grad_norm": 0.8751836788329617, "learning_rate": 9.910671034852555e-06, "loss": 0.3439, "step": 2560 }, { "epoch": 0.1770480470100242, "grad_norm": 0.7963463477675932, "learning_rate": 9.910565667567349e-06, "loss": 0.3737, "step": 2561 }, { "epoch": 0.1771171793985482, "grad_norm": 0.8377345002253552, "learning_rate": 9.910460238736965e-06, "loss": 0.4353, "step": 2562 }, { "epoch": 0.17718631178707225, "grad_norm": 0.8322146599102589, "learning_rate": 9.910354748362725e-06, "loss": 0.3536, "step": 2563 }, { "epoch": 0.17725544417559627, "grad_norm": 0.8835851349767285, "learning_rate": 9.91024919644595e-06, "loss": 0.3977, "step": 2564 }, { "epoch": 0.1773245765641203, "grad_norm": 0.8779178047298901, "learning_rate": 9.910143582987966e-06, "loss": 0.4286, "step": 2565 }, { "epoch": 0.1773937089526443, "grad_norm": 0.8405464812141148, "learning_rate": 9.910037907990093e-06, "loss": 0.3807, "step": 2566 }, { "epoch": 0.17746284134116833, "grad_norm": 0.9726909437344798, "learning_rate": 9.909932171453659e-06, "loss": 0.3424, "step": 2567 }, { "epoch": 0.17753197372969237, "grad_norm": 0.8522008754557494, "learning_rate": 9.909826373379988e-06, "loss": 0.3438, "step": 2568 }, { "epoch": 0.1776011061182164, "grad_norm": 0.8908835873217885, "learning_rate": 9.909720513770401e-06, "loss": 0.5064, "step": 2569 }, { "epoch": 0.1776702385067404, "grad_norm": 0.8215989774723628, "learning_rate": 9.909614592626233e-06, "loss": 0.4025, "step": 2570 }, { "epoch": 0.17773937089526443, "grad_norm": 1.1409691477725812, "learning_rate": 9.909508609948807e-06, "loss": 0.3374, "step": 2571 }, { "epoch": 0.17780850328378844, "grad_norm": 0.8307683339035093, "learning_rate": 9.909402565739452e-06, "loss": 0.4315, "step": 2572 }, { "epoch": 0.1778776356723125, "grad_norm": 0.9064032393109246, "learning_rate": 9.909296459999496e-06, "loss": 0.3947, "step": 2573 }, { "epoch": 0.1779467680608365, "grad_norm": 0.923552613004057, "learning_rate": 9.909190292730271e-06, "loss": 0.3236, "step": 2574 }, { "epoch": 0.17801590044936053, "grad_norm": 0.810527018668327, "learning_rate": 9.909084063933106e-06, "loss": 0.381, "step": 2575 }, { "epoch": 0.17808503283788454, "grad_norm": 0.8443123005282573, "learning_rate": 9.908977773609334e-06, "loss": 0.4728, "step": 2576 }, { "epoch": 0.17815416522640856, "grad_norm": 0.7665062630404598, "learning_rate": 9.908871421760285e-06, "loss": 0.3303, "step": 2577 }, { "epoch": 0.1782232976149326, "grad_norm": 0.794329258916725, "learning_rate": 9.908765008387293e-06, "loss": 0.3382, "step": 2578 }, { "epoch": 0.17829243000345663, "grad_norm": 0.9015377536036088, "learning_rate": 9.908658533491692e-06, "loss": 0.5057, "step": 2579 }, { "epoch": 0.17836156239198064, "grad_norm": 0.8059764941489594, "learning_rate": 9.908551997074816e-06, "loss": 0.4171, "step": 2580 }, { "epoch": 0.17843069478050466, "grad_norm": 0.8971050767339658, "learning_rate": 9.908445399138001e-06, "loss": 0.2997, "step": 2581 }, { "epoch": 0.17849982716902868, "grad_norm": 0.7397949137817191, "learning_rate": 9.908338739682582e-06, "loss": 0.3224, "step": 2582 }, { "epoch": 0.17856895955755273, "grad_norm": 0.8442954567380082, "learning_rate": 9.908232018709896e-06, "loss": 0.4257, "step": 2583 }, { "epoch": 0.17863809194607674, "grad_norm": 0.8051997399060514, "learning_rate": 9.90812523622128e-06, "loss": 0.3199, "step": 2584 }, { "epoch": 0.17870722433460076, "grad_norm": 0.9666538023152029, "learning_rate": 9.908018392218074e-06, "loss": 0.3238, "step": 2585 }, { "epoch": 0.17877635672312478, "grad_norm": 0.8390354232756038, "learning_rate": 9.907911486701617e-06, "loss": 0.4085, "step": 2586 }, { "epoch": 0.1788454891116488, "grad_norm": 0.8510148629130029, "learning_rate": 9.907804519673247e-06, "loss": 0.4573, "step": 2587 }, { "epoch": 0.17891462150017284, "grad_norm": 0.7373650966116079, "learning_rate": 9.907697491134306e-06, "loss": 0.2966, "step": 2588 }, { "epoch": 0.17898375388869686, "grad_norm": 0.8191805816522452, "learning_rate": 9.907590401086136e-06, "loss": 0.3579, "step": 2589 }, { "epoch": 0.17905288627722088, "grad_norm": 1.0165467671206656, "learning_rate": 9.907483249530077e-06, "loss": 0.4506, "step": 2590 }, { "epoch": 0.1791220186657449, "grad_norm": 0.7699495592024165, "learning_rate": 9.907376036467473e-06, "loss": 0.3309, "step": 2591 }, { "epoch": 0.17919115105426892, "grad_norm": 0.9295786286030622, "learning_rate": 9.907268761899668e-06, "loss": 0.3617, "step": 2592 }, { "epoch": 0.17926028344279296, "grad_norm": 0.8436548559944327, "learning_rate": 9.907161425828009e-06, "loss": 0.4182, "step": 2593 }, { "epoch": 0.17932941583131698, "grad_norm": 0.9182115291806526, "learning_rate": 9.907054028253836e-06, "loss": 0.4235, "step": 2594 }, { "epoch": 0.179398548219841, "grad_norm": 0.8656486852548527, "learning_rate": 9.9069465691785e-06, "loss": 0.3214, "step": 2595 }, { "epoch": 0.17946768060836502, "grad_norm": 0.76339242453839, "learning_rate": 9.906839048603343e-06, "loss": 0.2941, "step": 2596 }, { "epoch": 0.17953681299688903, "grad_norm": 0.8806219862411047, "learning_rate": 9.906731466529717e-06, "loss": 0.4089, "step": 2597 }, { "epoch": 0.17960594538541308, "grad_norm": 0.8359004202528274, "learning_rate": 9.906623822958967e-06, "loss": 0.3523, "step": 2598 }, { "epoch": 0.1796750777739371, "grad_norm": 0.7598350032485603, "learning_rate": 9.906516117892446e-06, "loss": 0.36, "step": 2599 }, { "epoch": 0.17974421016246112, "grad_norm": 0.9107013339321404, "learning_rate": 9.9064083513315e-06, "loss": 0.3394, "step": 2600 }, { "epoch": 0.17981334255098513, "grad_norm": 0.8063116432393144, "learning_rate": 9.906300523277482e-06, "loss": 0.462, "step": 2601 }, { "epoch": 0.17988247493950915, "grad_norm": 0.8577168614036111, "learning_rate": 9.906192633731742e-06, "loss": 0.4126, "step": 2602 }, { "epoch": 0.1799516073280332, "grad_norm": 0.8932018545714671, "learning_rate": 9.906084682695633e-06, "loss": 0.3455, "step": 2603 }, { "epoch": 0.18002073971655722, "grad_norm": 0.7844912390579262, "learning_rate": 9.905976670170507e-06, "loss": 0.4202, "step": 2604 }, { "epoch": 0.18008987210508123, "grad_norm": 0.9208021586760372, "learning_rate": 9.90586859615772e-06, "loss": 0.4683, "step": 2605 }, { "epoch": 0.18015900449360525, "grad_norm": 0.8367671659391072, "learning_rate": 9.905760460658623e-06, "loss": 0.2704, "step": 2606 }, { "epoch": 0.18022813688212927, "grad_norm": 0.9066983557206374, "learning_rate": 9.905652263674575e-06, "loss": 0.3956, "step": 2607 }, { "epoch": 0.1802972692706533, "grad_norm": 0.8212202517556573, "learning_rate": 9.90554400520693e-06, "loss": 0.5005, "step": 2608 }, { "epoch": 0.18036640165917733, "grad_norm": 0.816988549543827, "learning_rate": 9.905435685257045e-06, "loss": 0.3196, "step": 2609 }, { "epoch": 0.18043553404770135, "grad_norm": 1.5257127708228646, "learning_rate": 9.905327303826278e-06, "loss": 0.2916, "step": 2610 }, { "epoch": 0.18050466643622537, "grad_norm": 0.9405126210010664, "learning_rate": 9.905218860915989e-06, "loss": 0.4259, "step": 2611 }, { "epoch": 0.1805737988247494, "grad_norm": 0.8737724443006559, "learning_rate": 9.905110356527531e-06, "loss": 0.4389, "step": 2612 }, { "epoch": 0.1806429312132734, "grad_norm": 0.7789210969565674, "learning_rate": 9.905001790662271e-06, "loss": 0.3241, "step": 2613 }, { "epoch": 0.18071206360179745, "grad_norm": 0.9167683562711851, "learning_rate": 9.904893163321567e-06, "loss": 0.355, "step": 2614 }, { "epoch": 0.18078119599032147, "grad_norm": 0.7837541579422433, "learning_rate": 9.904784474506781e-06, "loss": 0.4104, "step": 2615 }, { "epoch": 0.1808503283788455, "grad_norm": 0.903642076061371, "learning_rate": 9.904675724219274e-06, "loss": 0.3369, "step": 2616 }, { "epoch": 0.1809194607673695, "grad_norm": 0.7511577093979612, "learning_rate": 9.90456691246041e-06, "loss": 0.2967, "step": 2617 }, { "epoch": 0.18098859315589352, "grad_norm": 0.8384780065189803, "learning_rate": 9.90445803923155e-06, "loss": 0.3709, "step": 2618 }, { "epoch": 0.18105772554441757, "grad_norm": 0.953854672549729, "learning_rate": 9.904349104534064e-06, "loss": 0.4632, "step": 2619 }, { "epoch": 0.1811268579329416, "grad_norm": 0.8201565953549628, "learning_rate": 9.904240108369312e-06, "loss": 0.3283, "step": 2620 }, { "epoch": 0.1811959903214656, "grad_norm": 0.8861817529884338, "learning_rate": 9.904131050738663e-06, "loss": 0.3953, "step": 2621 }, { "epoch": 0.18126512270998962, "grad_norm": 0.8531310129738856, "learning_rate": 9.904021931643485e-06, "loss": 0.4103, "step": 2622 }, { "epoch": 0.18133425509851364, "grad_norm": 0.7755686079850213, "learning_rate": 9.90391275108514e-06, "loss": 0.3016, "step": 2623 }, { "epoch": 0.1814033874870377, "grad_norm": 0.8625030367799826, "learning_rate": 9.903803509065002e-06, "loss": 0.3207, "step": 2624 }, { "epoch": 0.1814725198755617, "grad_norm": 0.8659424008026646, "learning_rate": 9.903694205584437e-06, "loss": 0.3928, "step": 2625 }, { "epoch": 0.18154165226408572, "grad_norm": 0.908323742412328, "learning_rate": 9.903584840644819e-06, "loss": 0.4663, "step": 2626 }, { "epoch": 0.18161078465260974, "grad_norm": 0.917361871906909, "learning_rate": 9.903475414247515e-06, "loss": 0.3387, "step": 2627 }, { "epoch": 0.18167991704113376, "grad_norm": 0.9635685439626223, "learning_rate": 9.903365926393895e-06, "loss": 0.3742, "step": 2628 }, { "epoch": 0.1817490494296578, "grad_norm": 0.8420072033215343, "learning_rate": 9.903256377085334e-06, "loss": 0.4617, "step": 2629 }, { "epoch": 0.18181818181818182, "grad_norm": 0.8955585567963102, "learning_rate": 9.903146766323205e-06, "loss": 0.4462, "step": 2630 }, { "epoch": 0.18188731420670584, "grad_norm": 0.7493304394788767, "learning_rate": 9.903037094108884e-06, "loss": 0.2953, "step": 2631 }, { "epoch": 0.18195644659522986, "grad_norm": 0.8116342704457852, "learning_rate": 9.902927360443741e-06, "loss": 0.3538, "step": 2632 }, { "epoch": 0.18202557898375388, "grad_norm": 0.8814812951056387, "learning_rate": 9.902817565329153e-06, "loss": 0.5153, "step": 2633 }, { "epoch": 0.18209471137227792, "grad_norm": 0.8962808616935741, "learning_rate": 9.902707708766496e-06, "loss": 0.375, "step": 2634 }, { "epoch": 0.18216384376080194, "grad_norm": 2.0602711848694772, "learning_rate": 9.902597790757146e-06, "loss": 0.4966, "step": 2635 }, { "epoch": 0.18223297614932596, "grad_norm": 0.8210799076228595, "learning_rate": 9.902487811302485e-06, "loss": 0.3756, "step": 2636 }, { "epoch": 0.18230210853784998, "grad_norm": 1.0032641676944503, "learning_rate": 9.902377770403886e-06, "loss": 0.4193, "step": 2637 }, { "epoch": 0.182371240926374, "grad_norm": 0.8182012469612269, "learning_rate": 9.90226766806273e-06, "loss": 0.3322, "step": 2638 }, { "epoch": 0.18244037331489804, "grad_norm": 0.8263028406397623, "learning_rate": 9.902157504280398e-06, "loss": 0.3354, "step": 2639 }, { "epoch": 0.18250950570342206, "grad_norm": 0.9271420063238857, "learning_rate": 9.902047279058272e-06, "loss": 0.4997, "step": 2640 }, { "epoch": 0.18257863809194608, "grad_norm": 0.9964205437124799, "learning_rate": 9.901936992397729e-06, "loss": 0.3596, "step": 2641 }, { "epoch": 0.1826477704804701, "grad_norm": 0.742865275798771, "learning_rate": 9.901826644300155e-06, "loss": 0.3397, "step": 2642 }, { "epoch": 0.1827169028689941, "grad_norm": 0.8704222468876575, "learning_rate": 9.901716234766929e-06, "loss": 0.3796, "step": 2643 }, { "epoch": 0.18278603525751816, "grad_norm": 0.9235509840293487, "learning_rate": 9.90160576379944e-06, "loss": 0.4499, "step": 2644 }, { "epoch": 0.18285516764604218, "grad_norm": 0.9350012747438485, "learning_rate": 9.90149523139907e-06, "loss": 0.3384, "step": 2645 }, { "epoch": 0.1829243000345662, "grad_norm": 0.8450391320740949, "learning_rate": 9.901384637567203e-06, "loss": 0.4187, "step": 2646 }, { "epoch": 0.1829934324230902, "grad_norm": 0.8124159472684158, "learning_rate": 9.901273982305228e-06, "loss": 0.4195, "step": 2647 }, { "epoch": 0.18306256481161423, "grad_norm": 0.8049007705353322, "learning_rate": 9.901163265614528e-06, "loss": 0.4075, "step": 2648 }, { "epoch": 0.18313169720013828, "grad_norm": 0.9991367879887728, "learning_rate": 9.901052487496497e-06, "loss": 0.3042, "step": 2649 }, { "epoch": 0.1832008295886623, "grad_norm": 0.8746333605757705, "learning_rate": 9.900941647952516e-06, "loss": 0.4081, "step": 2650 }, { "epoch": 0.1832699619771863, "grad_norm": 1.1745133543168167, "learning_rate": 9.900830746983977e-06, "loss": 0.5258, "step": 2651 }, { "epoch": 0.18333909436571033, "grad_norm": 0.8400716115040483, "learning_rate": 9.900719784592274e-06, "loss": 0.3225, "step": 2652 }, { "epoch": 0.18340822675423435, "grad_norm": 0.8363392759931704, "learning_rate": 9.900608760778791e-06, "loss": 0.3291, "step": 2653 }, { "epoch": 0.1834773591427584, "grad_norm": 0.9223745755646396, "learning_rate": 9.900497675544925e-06, "loss": 0.429, "step": 2654 }, { "epoch": 0.1835464915312824, "grad_norm": 0.8945108579742374, "learning_rate": 9.900386528892066e-06, "loss": 0.4414, "step": 2655 }, { "epoch": 0.18361562391980643, "grad_norm": 0.8205531471448115, "learning_rate": 9.900275320821605e-06, "loss": 0.3363, "step": 2656 }, { "epoch": 0.18368475630833045, "grad_norm": 0.9247598532432875, "learning_rate": 9.90016405133494e-06, "loss": 0.3642, "step": 2657 }, { "epoch": 0.18375388869685447, "grad_norm": 0.9270747906695858, "learning_rate": 9.90005272043346e-06, "loss": 0.4258, "step": 2658 }, { "epoch": 0.1838230210853785, "grad_norm": 0.9945104079887703, "learning_rate": 9.899941328118567e-06, "loss": 0.3336, "step": 2659 }, { "epoch": 0.18389215347390253, "grad_norm": 1.1089883356131958, "learning_rate": 9.899829874391651e-06, "loss": 0.3307, "step": 2660 }, { "epoch": 0.18396128586242655, "grad_norm": 0.7651465734420361, "learning_rate": 9.899718359254115e-06, "loss": 0.3878, "step": 2661 }, { "epoch": 0.18403041825095057, "grad_norm": 3.0056907154323484, "learning_rate": 9.899606782707352e-06, "loss": 0.5113, "step": 2662 }, { "epoch": 0.18409955063947459, "grad_norm": 0.9707683753541682, "learning_rate": 9.899495144752759e-06, "loss": 0.3622, "step": 2663 }, { "epoch": 0.18416868302799863, "grad_norm": 0.8775202910362442, "learning_rate": 9.899383445391742e-06, "loss": 0.3957, "step": 2664 }, { "epoch": 0.18423781541652265, "grad_norm": 0.9184834769408541, "learning_rate": 9.899271684625695e-06, "loss": 0.4395, "step": 2665 }, { "epoch": 0.18430694780504667, "grad_norm": 0.934475908365153, "learning_rate": 9.89915986245602e-06, "loss": 0.3616, "step": 2666 }, { "epoch": 0.18437608019357069, "grad_norm": 0.7245150631484341, "learning_rate": 9.89904797888412e-06, "loss": 0.3172, "step": 2667 }, { "epoch": 0.1844452125820947, "grad_norm": 0.8181438355349786, "learning_rate": 9.898936033911396e-06, "loss": 0.4147, "step": 2668 }, { "epoch": 0.18451434497061872, "grad_norm": 0.8161739141892552, "learning_rate": 9.898824027539252e-06, "loss": 0.4137, "step": 2669 }, { "epoch": 0.18458347735914277, "grad_norm": 0.8154654728124332, "learning_rate": 9.89871195976909e-06, "loss": 0.3491, "step": 2670 }, { "epoch": 0.18465260974766678, "grad_norm": 0.8312221675196697, "learning_rate": 9.898599830602315e-06, "loss": 0.3074, "step": 2671 }, { "epoch": 0.1847217421361908, "grad_norm": 0.8612690817003595, "learning_rate": 9.898487640040336e-06, "loss": 0.4229, "step": 2672 }, { "epoch": 0.18479087452471482, "grad_norm": 0.8854848860376456, "learning_rate": 9.898375388084555e-06, "loss": 0.4561, "step": 2673 }, { "epoch": 0.18486000691323884, "grad_norm": 0.8148421124814489, "learning_rate": 9.898263074736379e-06, "loss": 0.3114, "step": 2674 }, { "epoch": 0.18492913930176288, "grad_norm": 0.8829760024710116, "learning_rate": 9.898150699997217e-06, "loss": 0.4213, "step": 2675 }, { "epoch": 0.1849982716902869, "grad_norm": 0.8775701423974929, "learning_rate": 9.898038263868478e-06, "loss": 0.4138, "step": 2676 }, { "epoch": 0.18506740407881092, "grad_norm": 0.8856346168784334, "learning_rate": 9.897925766351569e-06, "loss": 0.3648, "step": 2677 }, { "epoch": 0.18513653646733494, "grad_norm": 0.9523769620446842, "learning_rate": 9.897813207447902e-06, "loss": 0.3136, "step": 2678 }, { "epoch": 0.18520566885585896, "grad_norm": 1.1275500531077833, "learning_rate": 9.897700587158887e-06, "loss": 0.4147, "step": 2679 }, { "epoch": 0.185274801244383, "grad_norm": 1.0550493096529137, "learning_rate": 9.897587905485936e-06, "loss": 0.3502, "step": 2680 }, { "epoch": 0.18534393363290702, "grad_norm": 0.9280881044025656, "learning_rate": 9.89747516243046e-06, "loss": 0.3623, "step": 2681 }, { "epoch": 0.18541306602143104, "grad_norm": 0.8028219477687647, "learning_rate": 9.897362357993873e-06, "loss": 0.3494, "step": 2682 }, { "epoch": 0.18548219840995506, "grad_norm": 0.8600282415563656, "learning_rate": 9.89724949217759e-06, "loss": 0.4205, "step": 2683 }, { "epoch": 0.18555133079847907, "grad_norm": 0.9294529473551554, "learning_rate": 9.897136564983021e-06, "loss": 0.4354, "step": 2684 }, { "epoch": 0.18562046318700312, "grad_norm": 0.8449345649142164, "learning_rate": 9.897023576411587e-06, "loss": 0.3783, "step": 2685 }, { "epoch": 0.18568959557552714, "grad_norm": 0.7600253777624604, "learning_rate": 9.8969105264647e-06, "loss": 0.3959, "step": 2686 }, { "epoch": 0.18575872796405116, "grad_norm": 0.7858606427974616, "learning_rate": 9.89679741514378e-06, "loss": 0.3699, "step": 2687 }, { "epoch": 0.18582786035257517, "grad_norm": 0.8007887721237139, "learning_rate": 9.896684242450241e-06, "loss": 0.2788, "step": 2688 }, { "epoch": 0.1858969927410992, "grad_norm": 0.7774807840745805, "learning_rate": 9.896571008385504e-06, "loss": 0.3213, "step": 2689 }, { "epoch": 0.18596612512962324, "grad_norm": 0.8772412331514526, "learning_rate": 9.896457712950991e-06, "loss": 0.4418, "step": 2690 }, { "epoch": 0.18603525751814726, "grad_norm": 0.8870471288585388, "learning_rate": 9.896344356148116e-06, "loss": 0.3561, "step": 2691 }, { "epoch": 0.18610438990667127, "grad_norm": 0.8604068260355493, "learning_rate": 9.896230937978301e-06, "loss": 0.3218, "step": 2692 }, { "epoch": 0.1861735222951953, "grad_norm": 0.8808370545913922, "learning_rate": 9.896117458442972e-06, "loss": 0.3921, "step": 2693 }, { "epoch": 0.1862426546837193, "grad_norm": 0.9202748178223012, "learning_rate": 9.896003917543546e-06, "loss": 0.4447, "step": 2694 }, { "epoch": 0.18631178707224336, "grad_norm": 0.9447997711391195, "learning_rate": 9.89589031528145e-06, "loss": 0.373, "step": 2695 }, { "epoch": 0.18638091946076737, "grad_norm": 1.24099852061731, "learning_rate": 9.895776651658105e-06, "loss": 0.3031, "step": 2696 }, { "epoch": 0.1864500518492914, "grad_norm": 0.7963730862005745, "learning_rate": 9.895662926674936e-06, "loss": 0.4859, "step": 2697 }, { "epoch": 0.1865191842378154, "grad_norm": 1.0969802551979817, "learning_rate": 9.89554914033337e-06, "loss": 0.4007, "step": 2698 }, { "epoch": 0.18658831662633943, "grad_norm": 0.8529980025537788, "learning_rate": 9.895435292634832e-06, "loss": 0.3031, "step": 2699 }, { "epoch": 0.18665744901486347, "grad_norm": 0.7967450771275454, "learning_rate": 9.895321383580748e-06, "loss": 0.3362, "step": 2700 }, { "epoch": 0.1867265814033875, "grad_norm": 0.9004219707824782, "learning_rate": 9.895207413172547e-06, "loss": 0.4979, "step": 2701 }, { "epoch": 0.1867957137919115, "grad_norm": 0.846960560057461, "learning_rate": 9.895093381411657e-06, "loss": 0.3026, "step": 2702 }, { "epoch": 0.18686484618043553, "grad_norm": 0.7751168307768663, "learning_rate": 9.894979288299507e-06, "loss": 0.3626, "step": 2703 }, { "epoch": 0.18693397856895955, "grad_norm": 0.9759442818484149, "learning_rate": 9.894865133837528e-06, "loss": 0.4484, "step": 2704 }, { "epoch": 0.1870031109574836, "grad_norm": 0.8766361228208809, "learning_rate": 9.89475091802715e-06, "loss": 0.3633, "step": 2705 }, { "epoch": 0.1870722433460076, "grad_norm": 0.8121481749506065, "learning_rate": 9.894636640869802e-06, "loss": 0.3559, "step": 2706 }, { "epoch": 0.18714137573453163, "grad_norm": 0.8227128328154744, "learning_rate": 9.89452230236692e-06, "loss": 0.3573, "step": 2707 }, { "epoch": 0.18721050812305565, "grad_norm": 1.2035351381639265, "learning_rate": 9.894407902519937e-06, "loss": 0.4197, "step": 2708 }, { "epoch": 0.18727964051157966, "grad_norm": 0.7505666999633819, "learning_rate": 9.894293441330283e-06, "loss": 0.3194, "step": 2709 }, { "epoch": 0.1873487729001037, "grad_norm": 0.8702641943787042, "learning_rate": 9.894178918799395e-06, "loss": 0.3238, "step": 2710 }, { "epoch": 0.18741790528862773, "grad_norm": 0.8936905765363217, "learning_rate": 9.89406433492871e-06, "loss": 0.39, "step": 2711 }, { "epoch": 0.18748703767715175, "grad_norm": 2.0763264545068, "learning_rate": 9.893949689719662e-06, "loss": 0.4492, "step": 2712 }, { "epoch": 0.18755617006567576, "grad_norm": 0.8646539805778405, "learning_rate": 9.893834983173686e-06, "loss": 0.3494, "step": 2713 }, { "epoch": 0.18762530245419978, "grad_norm": 0.8507474347296125, "learning_rate": 9.893720215292226e-06, "loss": 0.4112, "step": 2714 }, { "epoch": 0.18769443484272383, "grad_norm": 0.9786948566519474, "learning_rate": 9.893605386076713e-06, "loss": 0.5251, "step": 2715 }, { "epoch": 0.18776356723124785, "grad_norm": 0.8488123717110088, "learning_rate": 9.893490495528591e-06, "loss": 0.4123, "step": 2716 }, { "epoch": 0.18783269961977186, "grad_norm": 0.7760104054975988, "learning_rate": 9.893375543649299e-06, "loss": 0.3306, "step": 2717 }, { "epoch": 0.18790183200829588, "grad_norm": 0.7789164289956395, "learning_rate": 9.893260530440276e-06, "loss": 0.3666, "step": 2718 }, { "epoch": 0.1879709643968199, "grad_norm": 0.8564235846945591, "learning_rate": 9.893145455902965e-06, "loss": 0.5161, "step": 2719 }, { "epoch": 0.18804009678534395, "grad_norm": 0.9430411459086875, "learning_rate": 9.893030320038808e-06, "loss": 0.3504, "step": 2720 }, { "epoch": 0.18810922917386796, "grad_norm": 0.831099410600297, "learning_rate": 9.892915122849248e-06, "loss": 0.4017, "step": 2721 }, { "epoch": 0.18817836156239198, "grad_norm": 0.7797904051330358, "learning_rate": 9.89279986433573e-06, "loss": 0.4927, "step": 2722 }, { "epoch": 0.188247493950916, "grad_norm": 0.790318604356134, "learning_rate": 9.892684544499694e-06, "loss": 0.4054, "step": 2723 }, { "epoch": 0.18831662633944002, "grad_norm": 0.9331832553899962, "learning_rate": 9.892569163342592e-06, "loss": 0.3196, "step": 2724 }, { "epoch": 0.18838575872796406, "grad_norm": 0.9509251160962412, "learning_rate": 9.892453720865864e-06, "loss": 0.3414, "step": 2725 }, { "epoch": 0.18845489111648808, "grad_norm": 0.7952218862488066, "learning_rate": 9.892338217070962e-06, "loss": 0.4482, "step": 2726 }, { "epoch": 0.1885240235050121, "grad_norm": 0.8139058213125727, "learning_rate": 9.892222651959328e-06, "loss": 0.3081, "step": 2727 }, { "epoch": 0.18859315589353612, "grad_norm": 1.1379024050683102, "learning_rate": 9.892107025532416e-06, "loss": 0.3354, "step": 2728 }, { "epoch": 0.18866228828206014, "grad_norm": 0.8549465279763888, "learning_rate": 9.891991337791674e-06, "loss": 0.4418, "step": 2729 }, { "epoch": 0.18873142067058418, "grad_norm": 0.790207034914138, "learning_rate": 9.891875588738548e-06, "loss": 0.3846, "step": 2730 }, { "epoch": 0.1888005530591082, "grad_norm": 0.8068879109809913, "learning_rate": 9.891759778374493e-06, "loss": 0.39, "step": 2731 }, { "epoch": 0.18886968544763222, "grad_norm": 0.822264732251305, "learning_rate": 9.891643906700958e-06, "loss": 0.3847, "step": 2732 }, { "epoch": 0.18893881783615624, "grad_norm": 0.8614848524666947, "learning_rate": 9.891527973719398e-06, "loss": 0.4407, "step": 2733 }, { "epoch": 0.18900795022468025, "grad_norm": 0.9074179884623922, "learning_rate": 9.891411979431263e-06, "loss": 0.3471, "step": 2734 }, { "epoch": 0.18907708261320427, "grad_norm": 0.7534845756798427, "learning_rate": 9.891295923838006e-06, "loss": 0.3283, "step": 2735 }, { "epoch": 0.18914621500172832, "grad_norm": 0.8796688391405144, "learning_rate": 9.891179806941088e-06, "loss": 0.3819, "step": 2736 }, { "epoch": 0.18921534739025234, "grad_norm": 0.817808358541113, "learning_rate": 9.891063628741957e-06, "loss": 0.3924, "step": 2737 }, { "epoch": 0.18928447977877635, "grad_norm": 0.7657760974816264, "learning_rate": 9.890947389242072e-06, "loss": 0.2856, "step": 2738 }, { "epoch": 0.18935361216730037, "grad_norm": 0.7889575522113267, "learning_rate": 9.89083108844289e-06, "loss": 0.3361, "step": 2739 }, { "epoch": 0.1894227445558244, "grad_norm": 0.7915415340642945, "learning_rate": 9.890714726345867e-06, "loss": 0.4448, "step": 2740 }, { "epoch": 0.18949187694434844, "grad_norm": 0.955120343643206, "learning_rate": 9.890598302952465e-06, "loss": 0.3493, "step": 2741 }, { "epoch": 0.18956100933287245, "grad_norm": 0.861371007073603, "learning_rate": 9.89048181826414e-06, "loss": 0.3474, "step": 2742 }, { "epoch": 0.18963014172139647, "grad_norm": 0.8396419200530646, "learning_rate": 9.890365272282351e-06, "loss": 0.3702, "step": 2743 }, { "epoch": 0.1896992741099205, "grad_norm": 0.8046950184924625, "learning_rate": 9.890248665008564e-06, "loss": 0.4784, "step": 2744 }, { "epoch": 0.1897684064984445, "grad_norm": 0.8044080494760135, "learning_rate": 9.890131996444235e-06, "loss": 0.3246, "step": 2745 }, { "epoch": 0.18983753888696855, "grad_norm": 0.843679881920289, "learning_rate": 9.89001526659083e-06, "loss": 0.3537, "step": 2746 }, { "epoch": 0.18990667127549257, "grad_norm": 0.8364929161416197, "learning_rate": 9.889898475449808e-06, "loss": 0.4534, "step": 2747 }, { "epoch": 0.1899758036640166, "grad_norm": 0.9867689309042577, "learning_rate": 9.889781623022636e-06, "loss": 0.4016, "step": 2748 }, { "epoch": 0.1900449360525406, "grad_norm": 0.8240216726633225, "learning_rate": 9.889664709310778e-06, "loss": 0.298, "step": 2749 }, { "epoch": 0.19011406844106463, "grad_norm": 0.9857799754711263, "learning_rate": 9.889547734315697e-06, "loss": 0.3747, "step": 2750 }, { "epoch": 0.19018320082958867, "grad_norm": 0.9269560215237014, "learning_rate": 9.889430698038863e-06, "loss": 0.4807, "step": 2751 }, { "epoch": 0.1902523332181127, "grad_norm": 0.8923896662328485, "learning_rate": 9.88931360048174e-06, "loss": 0.3256, "step": 2752 }, { "epoch": 0.1903214656066367, "grad_norm": 0.9208520146533747, "learning_rate": 9.889196441645797e-06, "loss": 0.2762, "step": 2753 }, { "epoch": 0.19039059799516073, "grad_norm": 0.7958801092526764, "learning_rate": 9.8890792215325e-06, "loss": 0.4398, "step": 2754 }, { "epoch": 0.19045973038368474, "grad_norm": 0.7781151384546563, "learning_rate": 9.888961940143323e-06, "loss": 0.3917, "step": 2755 }, { "epoch": 0.1905288627722088, "grad_norm": 0.8890926619315437, "learning_rate": 9.88884459747973e-06, "loss": 0.3561, "step": 2756 }, { "epoch": 0.1905979951607328, "grad_norm": 0.9010483102824808, "learning_rate": 9.888727193543194e-06, "loss": 0.3595, "step": 2757 }, { "epoch": 0.19066712754925683, "grad_norm": 0.8694679275196957, "learning_rate": 9.88860972833519e-06, "loss": 0.4695, "step": 2758 }, { "epoch": 0.19073625993778084, "grad_norm": 0.7428355921584993, "learning_rate": 9.888492201857186e-06, "loss": 0.3196, "step": 2759 }, { "epoch": 0.19080539232630486, "grad_norm": 0.7656960291473626, "learning_rate": 9.888374614110656e-06, "loss": 0.3095, "step": 2760 }, { "epoch": 0.1908745247148289, "grad_norm": 0.77237532157132, "learning_rate": 9.888256965097075e-06, "loss": 0.3441, "step": 2761 }, { "epoch": 0.19094365710335293, "grad_norm": 0.8240675081645225, "learning_rate": 9.888139254817913e-06, "loss": 0.3985, "step": 2762 }, { "epoch": 0.19101278949187694, "grad_norm": 0.870509324157878, "learning_rate": 9.888021483274651e-06, "loss": 0.3262, "step": 2763 }, { "epoch": 0.19108192188040096, "grad_norm": 0.803851666148973, "learning_rate": 9.887903650468763e-06, "loss": 0.3255, "step": 2764 }, { "epoch": 0.19115105426892498, "grad_norm": 1.158597932514195, "learning_rate": 9.887785756401726e-06, "loss": 0.4415, "step": 2765 }, { "epoch": 0.19122018665744903, "grad_norm": 0.8338884493097961, "learning_rate": 9.887667801075016e-06, "loss": 0.3852, "step": 2766 }, { "epoch": 0.19128931904597304, "grad_norm": 0.7675230938493102, "learning_rate": 9.887549784490112e-06, "loss": 0.3377, "step": 2767 }, { "epoch": 0.19135845143449706, "grad_norm": 0.9722138919543706, "learning_rate": 9.887431706648496e-06, "loss": 0.3617, "step": 2768 }, { "epoch": 0.19142758382302108, "grad_norm": 1.1274594658641626, "learning_rate": 9.887313567551645e-06, "loss": 0.4541, "step": 2769 }, { "epoch": 0.1914967162115451, "grad_norm": 0.8434243004577564, "learning_rate": 9.887195367201038e-06, "loss": 0.414, "step": 2770 }, { "epoch": 0.19156584860006914, "grad_norm": 0.9054475966509552, "learning_rate": 9.88707710559816e-06, "loss": 0.3052, "step": 2771 }, { "epoch": 0.19163498098859316, "grad_norm": 0.8404549146314225, "learning_rate": 9.88695878274449e-06, "loss": 0.4532, "step": 2772 }, { "epoch": 0.19170411337711718, "grad_norm": 1.0401868076990959, "learning_rate": 9.886840398641517e-06, "loss": 0.4288, "step": 2773 }, { "epoch": 0.1917732457656412, "grad_norm": 0.885230541524216, "learning_rate": 9.886721953290719e-06, "loss": 0.3296, "step": 2774 }, { "epoch": 0.19184237815416522, "grad_norm": 0.8702758090050615, "learning_rate": 9.886603446693581e-06, "loss": 0.3455, "step": 2775 }, { "epoch": 0.19191151054268926, "grad_norm": 0.8233057063704151, "learning_rate": 9.88648487885159e-06, "loss": 0.4597, "step": 2776 }, { "epoch": 0.19198064293121328, "grad_norm": 1.053616219518699, "learning_rate": 9.886366249766231e-06, "loss": 0.3193, "step": 2777 }, { "epoch": 0.1920497753197373, "grad_norm": 0.7423387513677343, "learning_rate": 9.88624755943899e-06, "loss": 0.3416, "step": 2778 }, { "epoch": 0.19211890770826132, "grad_norm": 0.8488924210988482, "learning_rate": 9.886128807871358e-06, "loss": 0.469, "step": 2779 }, { "epoch": 0.19218804009678533, "grad_norm": 0.8282000178871424, "learning_rate": 9.88600999506482e-06, "loss": 0.3776, "step": 2780 }, { "epoch": 0.19225717248530938, "grad_norm": 0.8029801771500505, "learning_rate": 9.885891121020866e-06, "loss": 0.3098, "step": 2781 }, { "epoch": 0.1923263048738334, "grad_norm": 1.2683893207413417, "learning_rate": 9.885772185740987e-06, "loss": 0.4073, "step": 2782 }, { "epoch": 0.19239543726235742, "grad_norm": 0.7830574128130317, "learning_rate": 9.885653189226672e-06, "loss": 0.427, "step": 2783 }, { "epoch": 0.19246456965088143, "grad_norm": 0.7699777019281159, "learning_rate": 9.885534131479412e-06, "loss": 0.3241, "step": 2784 }, { "epoch": 0.19253370203940545, "grad_norm": 0.8529200465786371, "learning_rate": 9.885415012500702e-06, "loss": 0.2979, "step": 2785 }, { "epoch": 0.1926028344279295, "grad_norm": 0.7805780462246084, "learning_rate": 9.885295832292033e-06, "loss": 0.3247, "step": 2786 }, { "epoch": 0.19267196681645352, "grad_norm": 0.8534497134918941, "learning_rate": 9.8851765908549e-06, "loss": 0.4599, "step": 2787 }, { "epoch": 0.19274109920497753, "grad_norm": 0.6974748031538838, "learning_rate": 9.885057288190794e-06, "loss": 0.3669, "step": 2788 }, { "epoch": 0.19281023159350155, "grad_norm": 0.8769983435026358, "learning_rate": 9.884937924301214e-06, "loss": 0.3867, "step": 2789 }, { "epoch": 0.19287936398202557, "grad_norm": 1.1939201653556015, "learning_rate": 9.884818499187654e-06, "loss": 0.4656, "step": 2790 }, { "epoch": 0.19294849637054962, "grad_norm": 0.8465743898727441, "learning_rate": 9.884699012851612e-06, "loss": 0.3979, "step": 2791 }, { "epoch": 0.19301762875907363, "grad_norm": 0.8677962856858484, "learning_rate": 9.884579465294584e-06, "loss": 0.321, "step": 2792 }, { "epoch": 0.19308676114759765, "grad_norm": 0.8264841060313068, "learning_rate": 9.88445985651807e-06, "loss": 0.4004, "step": 2793 }, { "epoch": 0.19315589353612167, "grad_norm": 0.7729165239376685, "learning_rate": 9.884340186523569e-06, "loss": 0.4188, "step": 2794 }, { "epoch": 0.1932250259246457, "grad_norm": 0.8152970213371105, "learning_rate": 9.88422045531258e-06, "loss": 0.2969, "step": 2795 }, { "epoch": 0.19329415831316973, "grad_norm": 0.8071255984747487, "learning_rate": 9.884100662886605e-06, "loss": 0.3297, "step": 2796 }, { "epoch": 0.19336329070169375, "grad_norm": 0.8790847275386605, "learning_rate": 9.883980809247142e-06, "loss": 0.4962, "step": 2797 }, { "epoch": 0.19343242309021777, "grad_norm": 0.89253711897666, "learning_rate": 9.883860894395697e-06, "loss": 0.3452, "step": 2798 }, { "epoch": 0.1935015554787418, "grad_norm": 0.7747800622332898, "learning_rate": 9.883740918333771e-06, "loss": 0.3372, "step": 2799 }, { "epoch": 0.1935706878672658, "grad_norm": 0.8807504713329143, "learning_rate": 9.883620881062868e-06, "loss": 0.359, "step": 2800 }, { "epoch": 0.19363982025578982, "grad_norm": 0.8547047375744758, "learning_rate": 9.883500782584492e-06, "loss": 0.4666, "step": 2801 }, { "epoch": 0.19370895264431387, "grad_norm": 0.8531287181165529, "learning_rate": 9.883380622900148e-06, "loss": 0.3062, "step": 2802 }, { "epoch": 0.1937780850328379, "grad_norm": 0.7474301072697442, "learning_rate": 9.883260402011343e-06, "loss": 0.3015, "step": 2803 }, { "epoch": 0.1938472174213619, "grad_norm": 0.788243360306426, "learning_rate": 9.883140119919585e-06, "loss": 0.412, "step": 2804 }, { "epoch": 0.19391634980988592, "grad_norm": 0.8014053116273713, "learning_rate": 9.88301977662638e-06, "loss": 0.3925, "step": 2805 }, { "epoch": 0.19398548219840994, "grad_norm": 0.7106549564695581, "learning_rate": 9.882899372133235e-06, "loss": 0.2708, "step": 2806 }, { "epoch": 0.194054614586934, "grad_norm": 0.9964431628100008, "learning_rate": 9.882778906441659e-06, "loss": 0.3997, "step": 2807 }, { "epoch": 0.194123746975458, "grad_norm": 0.8886582166797955, "learning_rate": 9.882658379553164e-06, "loss": 0.4582, "step": 2808 }, { "epoch": 0.19419287936398202, "grad_norm": 0.7263483287654454, "learning_rate": 9.882537791469262e-06, "loss": 0.2865, "step": 2809 }, { "epoch": 0.19426201175250604, "grad_norm": 0.9212668802848771, "learning_rate": 9.88241714219146e-06, "loss": 0.3437, "step": 2810 }, { "epoch": 0.19433114414103006, "grad_norm": 0.8406451385627478, "learning_rate": 9.882296431721273e-06, "loss": 0.4108, "step": 2811 }, { "epoch": 0.1944002765295541, "grad_norm": 0.9079490446664561, "learning_rate": 9.882175660060212e-06, "loss": 0.4719, "step": 2812 }, { "epoch": 0.19446940891807812, "grad_norm": 0.9583585004577112, "learning_rate": 9.882054827209793e-06, "loss": 0.3262, "step": 2813 }, { "epoch": 0.19453854130660214, "grad_norm": 0.8652462482760154, "learning_rate": 9.881933933171529e-06, "loss": 0.389, "step": 2814 }, { "epoch": 0.19460767369512616, "grad_norm": 0.8207569723291378, "learning_rate": 9.881812977946934e-06, "loss": 0.4835, "step": 2815 }, { "epoch": 0.19467680608365018, "grad_norm": 0.8128057110543351, "learning_rate": 9.881691961537527e-06, "loss": 0.3992, "step": 2816 }, { "epoch": 0.19474593847217422, "grad_norm": 1.0028484319419446, "learning_rate": 9.881570883944822e-06, "loss": 0.3539, "step": 2817 }, { "epoch": 0.19481507086069824, "grad_norm": 0.8558136038657665, "learning_rate": 9.881449745170338e-06, "loss": 0.3937, "step": 2818 }, { "epoch": 0.19488420324922226, "grad_norm": 0.7727704899946777, "learning_rate": 9.881328545215592e-06, "loss": 0.4267, "step": 2819 }, { "epoch": 0.19495333563774628, "grad_norm": 0.8146151777321669, "learning_rate": 9.881207284082105e-06, "loss": 0.3004, "step": 2820 }, { "epoch": 0.1950224680262703, "grad_norm": 0.8282625024028998, "learning_rate": 9.881085961771395e-06, "loss": 0.329, "step": 2821 }, { "epoch": 0.19509160041479434, "grad_norm": 0.8759247859445194, "learning_rate": 9.880964578284983e-06, "loss": 0.4767, "step": 2822 }, { "epoch": 0.19516073280331836, "grad_norm": 1.0106595746221234, "learning_rate": 9.88084313362439e-06, "loss": 0.4115, "step": 2823 }, { "epoch": 0.19522986519184238, "grad_norm": 0.9338910488638343, "learning_rate": 9.88072162779114e-06, "loss": 0.3049, "step": 2824 }, { "epoch": 0.1952989975803664, "grad_norm": 1.5098873715891585, "learning_rate": 9.880600060786754e-06, "loss": 0.4082, "step": 2825 }, { "epoch": 0.1953681299688904, "grad_norm": 0.8094566247800561, "learning_rate": 9.880478432612755e-06, "loss": 0.3961, "step": 2826 }, { "epoch": 0.19543726235741446, "grad_norm": 0.898141494584076, "learning_rate": 9.88035674327067e-06, "loss": 0.3491, "step": 2827 }, { "epoch": 0.19550639474593848, "grad_norm": 0.8252091291609137, "learning_rate": 9.880234992762022e-06, "loss": 0.3828, "step": 2828 }, { "epoch": 0.1955755271344625, "grad_norm": 0.8535547752922213, "learning_rate": 9.880113181088335e-06, "loss": 0.4545, "step": 2829 }, { "epoch": 0.1956446595229865, "grad_norm": 0.9811983437988021, "learning_rate": 9.879991308251142e-06, "loss": 0.4097, "step": 2830 }, { "epoch": 0.19571379191151053, "grad_norm": 0.920696635360915, "learning_rate": 9.879869374251965e-06, "loss": 0.3489, "step": 2831 }, { "epoch": 0.19578292430003458, "grad_norm": 0.9221289772693312, "learning_rate": 9.879747379092335e-06, "loss": 0.3486, "step": 2832 }, { "epoch": 0.1958520566885586, "grad_norm": 0.8686013863420627, "learning_rate": 9.879625322773778e-06, "loss": 0.4958, "step": 2833 }, { "epoch": 0.1959211890770826, "grad_norm": 0.7759041451214089, "learning_rate": 9.879503205297827e-06, "loss": 0.2878, "step": 2834 }, { "epoch": 0.19599032146560663, "grad_norm": 0.8249217779448453, "learning_rate": 9.87938102666601e-06, "loss": 0.3127, "step": 2835 }, { "epoch": 0.19605945385413065, "grad_norm": 0.8889887430917688, "learning_rate": 9.879258786879861e-06, "loss": 0.4102, "step": 2836 }, { "epoch": 0.1961285862426547, "grad_norm": 0.8713669818689611, "learning_rate": 9.87913648594091e-06, "loss": 0.4179, "step": 2837 }, { "epoch": 0.1961977186311787, "grad_norm": 0.8525943031509008, "learning_rate": 9.879014123850689e-06, "loss": 0.3811, "step": 2838 }, { "epoch": 0.19626685101970273, "grad_norm": 0.8782701482192967, "learning_rate": 9.878891700610736e-06, "loss": 0.3854, "step": 2839 }, { "epoch": 0.19633598340822675, "grad_norm": 0.7973789116218313, "learning_rate": 9.87876921622258e-06, "loss": 0.4452, "step": 2840 }, { "epoch": 0.19640511579675077, "grad_norm": 0.903323698749238, "learning_rate": 9.87864667068776e-06, "loss": 0.3432, "step": 2841 }, { "epoch": 0.1964742481852748, "grad_norm": 0.8752904594384219, "learning_rate": 9.87852406400781e-06, "loss": 0.3225, "step": 2842 }, { "epoch": 0.19654338057379883, "grad_norm": 1.038863739898258, "learning_rate": 9.878401396184265e-06, "loss": 0.4249, "step": 2843 }, { "epoch": 0.19661251296232285, "grad_norm": 0.8322041864223487, "learning_rate": 9.878278667218666e-06, "loss": 0.4507, "step": 2844 }, { "epoch": 0.19668164535084687, "grad_norm": 0.7910689374056931, "learning_rate": 9.878155877112547e-06, "loss": 0.2985, "step": 2845 }, { "epoch": 0.19675077773937089, "grad_norm": 0.7600212431217642, "learning_rate": 9.878033025867452e-06, "loss": 0.369, "step": 2846 }, { "epoch": 0.19681991012789493, "grad_norm": 0.8196619221723979, "learning_rate": 9.877910113484918e-06, "loss": 0.4975, "step": 2847 }, { "epoch": 0.19688904251641895, "grad_norm": 1.2449962515844932, "learning_rate": 9.877787139966485e-06, "loss": 0.3662, "step": 2848 }, { "epoch": 0.19695817490494297, "grad_norm": 0.8232434639183928, "learning_rate": 9.877664105313699e-06, "loss": 0.2902, "step": 2849 }, { "epoch": 0.19702730729346699, "grad_norm": 1.090963358620298, "learning_rate": 9.877541009528093e-06, "loss": 0.3735, "step": 2850 }, { "epoch": 0.197096439681991, "grad_norm": 0.7814301445873485, "learning_rate": 9.877417852611218e-06, "loss": 0.3947, "step": 2851 }, { "epoch": 0.19716557207051505, "grad_norm": 0.8229424354739505, "learning_rate": 9.877294634564612e-06, "loss": 0.3365, "step": 2852 }, { "epoch": 0.19723470445903907, "grad_norm": 0.8397671201871404, "learning_rate": 9.877171355389824e-06, "loss": 0.3067, "step": 2853 }, { "epoch": 0.19730383684756309, "grad_norm": 0.7751421446557312, "learning_rate": 9.877048015088395e-06, "loss": 0.4518, "step": 2854 }, { "epoch": 0.1973729692360871, "grad_norm": 0.8033898516843719, "learning_rate": 9.876924613661871e-06, "loss": 0.4192, "step": 2855 }, { "epoch": 0.19744210162461112, "grad_norm": 1.1130244804120344, "learning_rate": 9.876801151111804e-06, "loss": 0.3984, "step": 2856 }, { "epoch": 0.19751123401313517, "grad_norm": 0.8215046544256829, "learning_rate": 9.876677627439734e-06, "loss": 0.3478, "step": 2857 }, { "epoch": 0.19758036640165919, "grad_norm": 0.8216246253102808, "learning_rate": 9.876554042647215e-06, "loss": 0.4819, "step": 2858 }, { "epoch": 0.1976494987901832, "grad_norm": 0.8628692540952915, "learning_rate": 9.876430396735795e-06, "loss": 0.4155, "step": 2859 }, { "epoch": 0.19771863117870722, "grad_norm": 0.8713417301319177, "learning_rate": 9.87630668970702e-06, "loss": 0.3274, "step": 2860 }, { "epoch": 0.19778776356723124, "grad_norm": 0.8394171955195615, "learning_rate": 9.876182921562442e-06, "loss": 0.4644, "step": 2861 }, { "epoch": 0.19785689595575526, "grad_norm": 1.003496907952008, "learning_rate": 9.876059092303616e-06, "loss": 0.4029, "step": 2862 }, { "epoch": 0.1979260283442793, "grad_norm": 0.8035978563877573, "learning_rate": 9.87593520193209e-06, "loss": 0.2869, "step": 2863 }, { "epoch": 0.19799516073280332, "grad_norm": 0.7757902254203507, "learning_rate": 9.875811250449416e-06, "loss": 0.3089, "step": 2864 }, { "epoch": 0.19806429312132734, "grad_norm": 0.8124023689908108, "learning_rate": 9.87568723785715e-06, "loss": 0.4638, "step": 2865 }, { "epoch": 0.19813342550985136, "grad_norm": 0.7848443209034988, "learning_rate": 9.875563164156847e-06, "loss": 0.3644, "step": 2866 }, { "epoch": 0.19820255789837538, "grad_norm": 0.8876707676187486, "learning_rate": 9.87543902935006e-06, "loss": 0.2905, "step": 2867 }, { "epoch": 0.19827169028689942, "grad_norm": 0.8731290831735573, "learning_rate": 9.875314833438345e-06, "loss": 0.3537, "step": 2868 }, { "epoch": 0.19834082267542344, "grad_norm": 0.9114055410914287, "learning_rate": 9.875190576423259e-06, "loss": 0.3323, "step": 2869 }, { "epoch": 0.19840995506394746, "grad_norm": 0.8288004069297665, "learning_rate": 9.875066258306358e-06, "loss": 0.3407, "step": 2870 }, { "epoch": 0.19847908745247148, "grad_norm": 0.8839133564140497, "learning_rate": 9.874941879089202e-06, "loss": 0.3368, "step": 2871 }, { "epoch": 0.1985482198409955, "grad_norm": 0.8559384615448098, "learning_rate": 9.87481743877335e-06, "loss": 0.4374, "step": 2872 }, { "epoch": 0.19861735222951954, "grad_norm": 0.90877039735719, "learning_rate": 9.874692937360361e-06, "loss": 0.387, "step": 2873 }, { "epoch": 0.19868648461804356, "grad_norm": 0.8063225674867263, "learning_rate": 9.874568374851796e-06, "loss": 0.3252, "step": 2874 }, { "epoch": 0.19875561700656758, "grad_norm": 0.9078390505507701, "learning_rate": 9.874443751249214e-06, "loss": 0.411, "step": 2875 }, { "epoch": 0.1988247493950916, "grad_norm": 0.7855479059094899, "learning_rate": 9.874319066554179e-06, "loss": 0.4719, "step": 2876 }, { "epoch": 0.1988938817836156, "grad_norm": 0.8728941760815846, "learning_rate": 9.874194320768253e-06, "loss": 0.3493, "step": 2877 }, { "epoch": 0.19896301417213966, "grad_norm": 0.8013528591079397, "learning_rate": 9.874069513893e-06, "loss": 0.3367, "step": 2878 }, { "epoch": 0.19903214656066368, "grad_norm": 0.8057916136436174, "learning_rate": 9.873944645929983e-06, "loss": 0.4454, "step": 2879 }, { "epoch": 0.1991012789491877, "grad_norm": 0.8526923891149046, "learning_rate": 9.873819716880769e-06, "loss": 0.4483, "step": 2880 }, { "epoch": 0.1991704113377117, "grad_norm": 1.5726408819676545, "learning_rate": 9.87369472674692e-06, "loss": 0.3638, "step": 2881 }, { "epoch": 0.19923954372623573, "grad_norm": 0.9152730666051092, "learning_rate": 9.873569675530009e-06, "loss": 0.3876, "step": 2882 }, { "epoch": 0.19930867611475978, "grad_norm": 0.9067321483577253, "learning_rate": 9.873444563231597e-06, "loss": 0.4652, "step": 2883 }, { "epoch": 0.1993778085032838, "grad_norm": 0.8619362532227127, "learning_rate": 9.873319389853254e-06, "loss": 0.3759, "step": 2884 }, { "epoch": 0.1994469408918078, "grad_norm": 0.8654469366299488, "learning_rate": 9.87319415539655e-06, "loss": 0.3094, "step": 2885 }, { "epoch": 0.19951607328033183, "grad_norm": 0.8515127464278647, "learning_rate": 9.873068859863055e-06, "loss": 0.4165, "step": 2886 }, { "epoch": 0.19958520566885585, "grad_norm": 0.8330671232305936, "learning_rate": 9.872943503254337e-06, "loss": 0.4464, "step": 2887 }, { "epoch": 0.1996543380573799, "grad_norm": 0.811522063203312, "learning_rate": 9.872818085571968e-06, "loss": 0.3427, "step": 2888 }, { "epoch": 0.1997234704459039, "grad_norm": 0.8192770694033111, "learning_rate": 9.872692606817521e-06, "loss": 0.3172, "step": 2889 }, { "epoch": 0.19979260283442793, "grad_norm": 0.8261264004261667, "learning_rate": 9.872567066992568e-06, "loss": 0.4586, "step": 2890 }, { "epoch": 0.19986173522295195, "grad_norm": 0.8126613584142843, "learning_rate": 9.872441466098681e-06, "loss": 0.3553, "step": 2891 }, { "epoch": 0.19993086761147597, "grad_norm": 0.8555367763196361, "learning_rate": 9.872315804137437e-06, "loss": 0.3171, "step": 2892 }, { "epoch": 0.2, "grad_norm": 0.8944525820911624, "learning_rate": 9.872190081110409e-06, "loss": 0.356, "step": 2893 }, { "epoch": 0.20006913238852403, "grad_norm": 0.8197478311469008, "learning_rate": 9.872064297019172e-06, "loss": 0.4297, "step": 2894 }, { "epoch": 0.20013826477704805, "grad_norm": 0.9339931797362139, "learning_rate": 9.871938451865304e-06, "loss": 0.3152, "step": 2895 }, { "epoch": 0.20020739716557207, "grad_norm": 0.8036998729739074, "learning_rate": 9.871812545650383e-06, "loss": 0.3204, "step": 2896 }, { "epoch": 0.20027652955409608, "grad_norm": 0.8346680309372756, "learning_rate": 9.871686578375986e-06, "loss": 0.4434, "step": 2897 }, { "epoch": 0.20034566194262013, "grad_norm": 0.8656987211853759, "learning_rate": 9.87156055004369e-06, "loss": 0.4529, "step": 2898 }, { "epoch": 0.20041479433114415, "grad_norm": 0.7969282892964098, "learning_rate": 9.871434460655078e-06, "loss": 0.2955, "step": 2899 }, { "epoch": 0.20048392671966817, "grad_norm": 0.8400206278595524, "learning_rate": 9.871308310211727e-06, "loss": 0.3751, "step": 2900 }, { "epoch": 0.20055305910819218, "grad_norm": 0.7443912889726919, "learning_rate": 9.871182098715222e-06, "loss": 0.4919, "step": 2901 }, { "epoch": 0.2006221914967162, "grad_norm": 0.7771916386809679, "learning_rate": 9.871055826167139e-06, "loss": 0.3663, "step": 2902 }, { "epoch": 0.20069132388524025, "grad_norm": 0.7697021309282966, "learning_rate": 9.870929492569066e-06, "loss": 0.3142, "step": 2903 }, { "epoch": 0.20076045627376427, "grad_norm": 0.7958458214692922, "learning_rate": 9.870803097922586e-06, "loss": 0.3994, "step": 2904 }, { "epoch": 0.20082958866228828, "grad_norm": 0.8288693652382803, "learning_rate": 9.870676642229277e-06, "loss": 0.4042, "step": 2905 }, { "epoch": 0.2008987210508123, "grad_norm": 0.8300283403143867, "learning_rate": 9.870550125490731e-06, "loss": 0.3512, "step": 2906 }, { "epoch": 0.20096785343933632, "grad_norm": 0.7572636568502019, "learning_rate": 9.870423547708532e-06, "loss": 0.3426, "step": 2907 }, { "epoch": 0.20103698582786036, "grad_norm": 0.7823738024503225, "learning_rate": 9.870296908884266e-06, "loss": 0.4277, "step": 2908 }, { "epoch": 0.20110611821638438, "grad_norm": 0.855050911180245, "learning_rate": 9.870170209019517e-06, "loss": 0.3292, "step": 2909 }, { "epoch": 0.2011752506049084, "grad_norm": 0.8616922752122069, "learning_rate": 9.870043448115877e-06, "loss": 0.3384, "step": 2910 }, { "epoch": 0.20124438299343242, "grad_norm": 0.885781419677216, "learning_rate": 9.869916626174933e-06, "loss": 0.4579, "step": 2911 }, { "epoch": 0.20131351538195644, "grad_norm": 0.7767392780438337, "learning_rate": 9.869789743198275e-06, "loss": 0.4713, "step": 2912 }, { "epoch": 0.20138264777048048, "grad_norm": 0.9099990931641572, "learning_rate": 9.869662799187494e-06, "loss": 0.3025, "step": 2913 }, { "epoch": 0.2014517801590045, "grad_norm": 0.8461454903877946, "learning_rate": 9.869535794144178e-06, "loss": 0.3837, "step": 2914 }, { "epoch": 0.20152091254752852, "grad_norm": 0.8614086584222903, "learning_rate": 9.869408728069922e-06, "loss": 0.4803, "step": 2915 }, { "epoch": 0.20159004493605254, "grad_norm": 0.7678024028242911, "learning_rate": 9.869281600966319e-06, "loss": 0.3205, "step": 2916 }, { "epoch": 0.20165917732457656, "grad_norm": 0.7874512605906849, "learning_rate": 9.869154412834959e-06, "loss": 0.3549, "step": 2917 }, { "epoch": 0.2017283097131006, "grad_norm": 0.8458835864026879, "learning_rate": 9.869027163677439e-06, "loss": 0.3656, "step": 2918 }, { "epoch": 0.20179744210162462, "grad_norm": 0.7998513426224289, "learning_rate": 9.86889985349535e-06, "loss": 0.3978, "step": 2919 }, { "epoch": 0.20186657449014864, "grad_norm": 0.9461674100133227, "learning_rate": 9.868772482290294e-06, "loss": 0.3712, "step": 2920 }, { "epoch": 0.20193570687867266, "grad_norm": 0.8830878942715296, "learning_rate": 9.86864505006386e-06, "loss": 0.4324, "step": 2921 }, { "epoch": 0.20200483926719667, "grad_norm": 0.8695639123237059, "learning_rate": 9.868517556817652e-06, "loss": 0.4283, "step": 2922 }, { "epoch": 0.20207397165572072, "grad_norm": 0.9388631809988675, "learning_rate": 9.868390002553263e-06, "loss": 0.3253, "step": 2923 }, { "epoch": 0.20214310404424474, "grad_norm": 0.89739132142217, "learning_rate": 9.868262387272294e-06, "loss": 0.3759, "step": 2924 }, { "epoch": 0.20221223643276875, "grad_norm": 0.8881976299230518, "learning_rate": 9.868134710976343e-06, "loss": 0.3852, "step": 2925 }, { "epoch": 0.20228136882129277, "grad_norm": 0.7138188411495566, "learning_rate": 9.868006973667011e-06, "loss": 0.4133, "step": 2926 }, { "epoch": 0.2023505012098168, "grad_norm": 1.0426377080151414, "learning_rate": 9.8678791753459e-06, "loss": 0.4021, "step": 2927 }, { "epoch": 0.2024196335983408, "grad_norm": 1.001730570552374, "learning_rate": 9.86775131601461e-06, "loss": 0.3558, "step": 2928 }, { "epoch": 0.20248876598686485, "grad_norm": 0.8437498212767781, "learning_rate": 9.867623395674744e-06, "loss": 0.4204, "step": 2929 }, { "epoch": 0.20255789837538887, "grad_norm": 0.809807446187721, "learning_rate": 9.867495414327906e-06, "loss": 0.4035, "step": 2930 }, { "epoch": 0.2026270307639129, "grad_norm": 0.768923909053465, "learning_rate": 9.867367371975697e-06, "loss": 0.2847, "step": 2931 }, { "epoch": 0.2026961631524369, "grad_norm": 0.7956606815231724, "learning_rate": 9.867239268619729e-06, "loss": 0.3566, "step": 2932 }, { "epoch": 0.20276529554096093, "grad_norm": 0.8065512132897878, "learning_rate": 9.867111104261598e-06, "loss": 0.4934, "step": 2933 }, { "epoch": 0.20283442792948497, "grad_norm": 1.1955075420970178, "learning_rate": 9.866982878902917e-06, "loss": 0.3467, "step": 2934 }, { "epoch": 0.202903560318009, "grad_norm": 0.7756521631812956, "learning_rate": 9.866854592545293e-06, "loss": 0.3353, "step": 2935 }, { "epoch": 0.202972692706533, "grad_norm": 1.099866329755539, "learning_rate": 9.866726245190328e-06, "loss": 0.4566, "step": 2936 }, { "epoch": 0.20304182509505703, "grad_norm": 0.850134262894843, "learning_rate": 9.866597836839637e-06, "loss": 0.3921, "step": 2937 }, { "epoch": 0.20311095748358104, "grad_norm": 0.8588872373280265, "learning_rate": 9.866469367494827e-06, "loss": 0.3541, "step": 2938 }, { "epoch": 0.2031800898721051, "grad_norm": 0.7635706993377374, "learning_rate": 9.866340837157507e-06, "loss": 0.3131, "step": 2939 }, { "epoch": 0.2032492222606291, "grad_norm": 0.8119445740912697, "learning_rate": 9.86621224582929e-06, "loss": 0.4719, "step": 2940 }, { "epoch": 0.20331835464915313, "grad_norm": 0.7884472775591544, "learning_rate": 9.866083593511785e-06, "loss": 0.3465, "step": 2941 }, { "epoch": 0.20338748703767714, "grad_norm": 0.8244551473961232, "learning_rate": 9.865954880206609e-06, "loss": 0.3008, "step": 2942 }, { "epoch": 0.20345661942620116, "grad_norm": 0.974934120245011, "learning_rate": 9.86582610591537e-06, "loss": 0.4007, "step": 2943 }, { "epoch": 0.2035257518147252, "grad_norm": 0.8253765440956294, "learning_rate": 9.865697270639684e-06, "loss": 0.3764, "step": 2944 }, { "epoch": 0.20359488420324923, "grad_norm": 0.86661884471663, "learning_rate": 9.865568374381168e-06, "loss": 0.3363, "step": 2945 }, { "epoch": 0.20366401659177324, "grad_norm": 0.7381157219489949, "learning_rate": 9.865439417141433e-06, "loss": 0.3224, "step": 2946 }, { "epoch": 0.20373314898029726, "grad_norm": 0.7944573373506081, "learning_rate": 9.8653103989221e-06, "loss": 0.4335, "step": 2947 }, { "epoch": 0.20380228136882128, "grad_norm": 0.8144792451249969, "learning_rate": 9.865181319724783e-06, "loss": 0.3909, "step": 2948 }, { "epoch": 0.20387141375734533, "grad_norm": 0.9341498586707678, "learning_rate": 9.8650521795511e-06, "loss": 0.3168, "step": 2949 }, { "epoch": 0.20394054614586934, "grad_norm": 0.804696512049786, "learning_rate": 9.86492297840267e-06, "loss": 0.3337, "step": 2950 }, { "epoch": 0.20400967853439336, "grad_norm": 0.8421964746425669, "learning_rate": 9.864793716281111e-06, "loss": 0.4773, "step": 2951 }, { "epoch": 0.20407881092291738, "grad_norm": 0.860515090648972, "learning_rate": 9.864664393188046e-06, "loss": 0.3564, "step": 2952 }, { "epoch": 0.2041479433114414, "grad_norm": 0.7937691615671795, "learning_rate": 9.864535009125095e-06, "loss": 0.3167, "step": 2953 }, { "epoch": 0.20421707569996544, "grad_norm": 0.7653824133745606, "learning_rate": 9.864405564093877e-06, "loss": 0.4138, "step": 2954 }, { "epoch": 0.20428620808848946, "grad_norm": 0.7970677054941214, "learning_rate": 9.864276058096017e-06, "loss": 0.3602, "step": 2955 }, { "epoch": 0.20435534047701348, "grad_norm": 0.8007143872621718, "learning_rate": 9.864146491133137e-06, "loss": 0.3431, "step": 2956 }, { "epoch": 0.2044244728655375, "grad_norm": 0.7772065195389228, "learning_rate": 9.864016863206863e-06, "loss": 0.3737, "step": 2957 }, { "epoch": 0.20449360525406152, "grad_norm": 0.7276326550711785, "learning_rate": 9.863887174318816e-06, "loss": 0.432, "step": 2958 }, { "epoch": 0.20456273764258556, "grad_norm": 0.8248917670576416, "learning_rate": 9.863757424470623e-06, "loss": 0.3843, "step": 2959 }, { "epoch": 0.20463187003110958, "grad_norm": 0.7362007179082848, "learning_rate": 9.86362761366391e-06, "loss": 0.3073, "step": 2960 }, { "epoch": 0.2047010024196336, "grad_norm": 1.0304994491462942, "learning_rate": 9.863497741900307e-06, "loss": 0.3779, "step": 2961 }, { "epoch": 0.20477013480815762, "grad_norm": 0.8391653231405192, "learning_rate": 9.863367809181436e-06, "loss": 0.4685, "step": 2962 }, { "epoch": 0.20483926719668163, "grad_norm": 0.7281674856075169, "learning_rate": 9.86323781550893e-06, "loss": 0.3107, "step": 2963 }, { "epoch": 0.20490839958520568, "grad_norm": 0.7846741099411897, "learning_rate": 9.863107760884419e-06, "loss": 0.393, "step": 2964 }, { "epoch": 0.2049775319737297, "grad_norm": 0.8502605023468326, "learning_rate": 9.862977645309527e-06, "loss": 0.4099, "step": 2965 }, { "epoch": 0.20504666436225372, "grad_norm": 0.87400892062583, "learning_rate": 9.86284746878589e-06, "loss": 0.3509, "step": 2966 }, { "epoch": 0.20511579675077773, "grad_norm": 0.7900204919444666, "learning_rate": 9.86271723131514e-06, "loss": 0.3335, "step": 2967 }, { "epoch": 0.20518492913930175, "grad_norm": 0.7839768558226897, "learning_rate": 9.862586932898906e-06, "loss": 0.358, "step": 2968 }, { "epoch": 0.2052540615278258, "grad_norm": 0.9120620214248837, "learning_rate": 9.862456573538821e-06, "loss": 0.4213, "step": 2969 }, { "epoch": 0.20532319391634982, "grad_norm": 0.7834081624391168, "learning_rate": 9.86232615323652e-06, "loss": 0.3382, "step": 2970 }, { "epoch": 0.20539232630487383, "grad_norm": 0.7858100194254303, "learning_rate": 9.862195671993641e-06, "loss": 0.2855, "step": 2971 }, { "epoch": 0.20546145869339785, "grad_norm": 0.9457627058830047, "learning_rate": 9.862065129811815e-06, "loss": 0.4937, "step": 2972 }, { "epoch": 0.20553059108192187, "grad_norm": 0.8198220890027356, "learning_rate": 9.861934526692679e-06, "loss": 0.3603, "step": 2973 }, { "epoch": 0.20559972347044592, "grad_norm": 0.769926834419788, "learning_rate": 9.861803862637872e-06, "loss": 0.3284, "step": 2974 }, { "epoch": 0.20566885585896993, "grad_norm": 0.9957002375272564, "learning_rate": 9.861673137649026e-06, "loss": 0.3523, "step": 2975 }, { "epoch": 0.20573798824749395, "grad_norm": 0.8784032979541464, "learning_rate": 9.861542351727787e-06, "loss": 0.49, "step": 2976 }, { "epoch": 0.20580712063601797, "grad_norm": 0.7660100242573046, "learning_rate": 9.86141150487579e-06, "loss": 0.3373, "step": 2977 }, { "epoch": 0.205876253024542, "grad_norm": 0.7349705219403135, "learning_rate": 9.861280597094672e-06, "loss": 0.3165, "step": 2978 }, { "epoch": 0.20594538541306603, "grad_norm": 0.8074658287078851, "learning_rate": 9.86114962838608e-06, "loss": 0.3954, "step": 2979 }, { "epoch": 0.20601451780159005, "grad_norm": 0.7206488689688144, "learning_rate": 9.861018598751651e-06, "loss": 0.366, "step": 2980 }, { "epoch": 0.20608365019011407, "grad_norm": 0.7308660497675679, "learning_rate": 9.86088750819303e-06, "loss": 0.2826, "step": 2981 }, { "epoch": 0.2061527825786381, "grad_norm": 0.7840379181893425, "learning_rate": 9.860756356711858e-06, "loss": 0.3702, "step": 2982 }, { "epoch": 0.2062219149671621, "grad_norm": 0.818063563615216, "learning_rate": 9.86062514430978e-06, "loss": 0.4072, "step": 2983 }, { "epoch": 0.20629104735568615, "grad_norm": 0.9918233617272245, "learning_rate": 9.860493870988439e-06, "loss": 0.3048, "step": 2984 }, { "epoch": 0.20636017974421017, "grad_norm": 0.7470333181586654, "learning_rate": 9.860362536749482e-06, "loss": 0.3575, "step": 2985 }, { "epoch": 0.2064293121327342, "grad_norm": 0.8428980358515683, "learning_rate": 9.860231141594555e-06, "loss": 0.4126, "step": 2986 }, { "epoch": 0.2064984445212582, "grad_norm": 0.839527996870145, "learning_rate": 9.860099685525302e-06, "loss": 0.4822, "step": 2987 }, { "epoch": 0.20656757690978222, "grad_norm": 1.4725277314674061, "learning_rate": 9.859968168543374e-06, "loss": 0.3246, "step": 2988 }, { "epoch": 0.20663670929830627, "grad_norm": 0.8980000131504826, "learning_rate": 9.85983659065042e-06, "loss": 0.3055, "step": 2989 }, { "epoch": 0.2067058416868303, "grad_norm": 0.7857254104091885, "learning_rate": 9.859704951848083e-06, "loss": 0.4057, "step": 2990 }, { "epoch": 0.2067749740753543, "grad_norm": 0.9181166623339154, "learning_rate": 9.859573252138021e-06, "loss": 0.3342, "step": 2991 }, { "epoch": 0.20684410646387832, "grad_norm": 0.876073120104209, "learning_rate": 9.859441491521879e-06, "loss": 0.3509, "step": 2992 }, { "epoch": 0.20691323885240234, "grad_norm": 0.8642719837399786, "learning_rate": 9.85930967000131e-06, "loss": 0.3865, "step": 2993 }, { "epoch": 0.20698237124092636, "grad_norm": 0.8031108673727789, "learning_rate": 9.859177787577964e-06, "loss": 0.4019, "step": 2994 }, { "epoch": 0.2070515036294504, "grad_norm": 0.8647193827865626, "learning_rate": 9.859045844253498e-06, "loss": 0.2952, "step": 2995 }, { "epoch": 0.20712063601797442, "grad_norm": 0.8314569775897009, "learning_rate": 9.858913840029564e-06, "loss": 0.3171, "step": 2996 }, { "epoch": 0.20718976840649844, "grad_norm": 0.778386017328798, "learning_rate": 9.858781774907815e-06, "loss": 0.4612, "step": 2997 }, { "epoch": 0.20725890079502246, "grad_norm": 0.7869685909494611, "learning_rate": 9.85864964888991e-06, "loss": 0.4124, "step": 2998 }, { "epoch": 0.20732803318354648, "grad_norm": 0.7388799865392691, "learning_rate": 9.8585174619775e-06, "loss": 0.3105, "step": 2999 }, { "epoch": 0.20739716557207052, "grad_norm": 0.9329702340333895, "learning_rate": 9.858385214172244e-06, "loss": 0.4611, "step": 3000 }, { "epoch": 0.20746629796059454, "grad_norm": 0.76381992902989, "learning_rate": 9.8582529054758e-06, "loss": 0.3762, "step": 3001 }, { "epoch": 0.20753543034911856, "grad_norm": 0.7667732755409531, "learning_rate": 9.858120535889825e-06, "loss": 0.2942, "step": 3002 }, { "epoch": 0.20760456273764258, "grad_norm": 0.7852430258377023, "learning_rate": 9.85798810541598e-06, "loss": 0.368, "step": 3003 }, { "epoch": 0.2076736951261666, "grad_norm": 0.7687849953684558, "learning_rate": 9.857855614055923e-06, "loss": 0.4123, "step": 3004 }, { "epoch": 0.20774282751469064, "grad_norm": 0.7923445272102504, "learning_rate": 9.857723061811315e-06, "loss": 0.4234, "step": 3005 }, { "epoch": 0.20781195990321466, "grad_norm": 0.7917975566093793, "learning_rate": 9.857590448683818e-06, "loss": 0.2945, "step": 3006 }, { "epoch": 0.20788109229173868, "grad_norm": 0.7703145086037283, "learning_rate": 9.857457774675092e-06, "loss": 0.3574, "step": 3007 }, { "epoch": 0.2079502246802627, "grad_norm": 0.8729137495341384, "learning_rate": 9.857325039786802e-06, "loss": 0.4865, "step": 3008 }, { "epoch": 0.20801935706878671, "grad_norm": 0.8318603393849724, "learning_rate": 9.857192244020613e-06, "loss": 0.349, "step": 3009 }, { "epoch": 0.20808848945731076, "grad_norm": 0.8118218482771811, "learning_rate": 9.857059387378184e-06, "loss": 0.34, "step": 3010 }, { "epoch": 0.20815762184583478, "grad_norm": 0.8646923881085556, "learning_rate": 9.856926469861185e-06, "loss": 0.4309, "step": 3011 }, { "epoch": 0.2082267542343588, "grad_norm": 0.872809577318202, "learning_rate": 9.85679349147128e-06, "loss": 0.4484, "step": 3012 }, { "epoch": 0.20829588662288281, "grad_norm": 0.8062388733782152, "learning_rate": 9.856660452210135e-06, "loss": 0.3295, "step": 3013 }, { "epoch": 0.20836501901140683, "grad_norm": 0.8676517720656662, "learning_rate": 9.856527352079421e-06, "loss": 0.3391, "step": 3014 }, { "epoch": 0.20843415139993088, "grad_norm": 0.8386624514761742, "learning_rate": 9.856394191080802e-06, "loss": 0.4364, "step": 3015 }, { "epoch": 0.2085032837884549, "grad_norm": 0.9040088378742708, "learning_rate": 9.856260969215949e-06, "loss": 0.4139, "step": 3016 }, { "epoch": 0.20857241617697891, "grad_norm": 0.8563559296494607, "learning_rate": 9.85612768648653e-06, "loss": 0.2557, "step": 3017 }, { "epoch": 0.20864154856550293, "grad_norm": 0.7887319860280798, "learning_rate": 9.855994342894218e-06, "loss": 0.4164, "step": 3018 }, { "epoch": 0.20871068095402695, "grad_norm": 0.7638486138492837, "learning_rate": 9.85586093844068e-06, "loss": 0.4387, "step": 3019 }, { "epoch": 0.208779813342551, "grad_norm": 1.1994989831105531, "learning_rate": 9.855727473127595e-06, "loss": 0.3058, "step": 3020 }, { "epoch": 0.20884894573107501, "grad_norm": 1.0350813886450074, "learning_rate": 9.855593946956628e-06, "loss": 0.2946, "step": 3021 }, { "epoch": 0.20891807811959903, "grad_norm": 0.8297859068384451, "learning_rate": 9.855460359929458e-06, "loss": 0.4768, "step": 3022 }, { "epoch": 0.20898721050812305, "grad_norm": 0.9091988095921318, "learning_rate": 9.855326712047756e-06, "loss": 0.3587, "step": 3023 }, { "epoch": 0.20905634289664707, "grad_norm": 0.8754087113621777, "learning_rate": 9.855193003313199e-06, "loss": 0.3589, "step": 3024 }, { "epoch": 0.20912547528517111, "grad_norm": 0.8232160291782452, "learning_rate": 9.855059233727462e-06, "loss": 0.2814, "step": 3025 }, { "epoch": 0.20919460767369513, "grad_norm": 0.7746189375003878, "learning_rate": 9.854925403292223e-06, "loss": 0.4146, "step": 3026 }, { "epoch": 0.20926374006221915, "grad_norm": 0.9627135660459349, "learning_rate": 9.854791512009156e-06, "loss": 0.3362, "step": 3027 }, { "epoch": 0.20933287245074317, "grad_norm": 0.9524087998215945, "learning_rate": 9.854657559879943e-06, "loss": 0.3258, "step": 3028 }, { "epoch": 0.2094020048392672, "grad_norm": 0.8448330345037983, "learning_rate": 9.85452354690626e-06, "loss": 0.4456, "step": 3029 }, { "epoch": 0.20947113722779123, "grad_norm": 0.9643181915565888, "learning_rate": 9.854389473089788e-06, "loss": 0.4589, "step": 3030 }, { "epoch": 0.20954026961631525, "grad_norm": 0.8757683908494108, "learning_rate": 9.854255338432207e-06, "loss": 0.3097, "step": 3031 }, { "epoch": 0.20960940200483927, "grad_norm": 0.7998967469384415, "learning_rate": 9.854121142935197e-06, "loss": 0.3994, "step": 3032 }, { "epoch": 0.2096785343933633, "grad_norm": 0.8620496561304066, "learning_rate": 9.853986886600442e-06, "loss": 0.4424, "step": 3033 }, { "epoch": 0.2097476667818873, "grad_norm": 0.9417906989327555, "learning_rate": 9.853852569429624e-06, "loss": 0.3416, "step": 3034 }, { "epoch": 0.20981679917041135, "grad_norm": 0.7362907999131314, "learning_rate": 9.853718191424426e-06, "loss": 0.3477, "step": 3035 }, { "epoch": 0.20988593155893537, "grad_norm": 0.9210949491817937, "learning_rate": 9.853583752586533e-06, "loss": 0.5359, "step": 3036 }, { "epoch": 0.20995506394745939, "grad_norm": 0.8483467160442533, "learning_rate": 9.853449252917628e-06, "loss": 0.4084, "step": 3037 }, { "epoch": 0.2100241963359834, "grad_norm": 0.7311151665124798, "learning_rate": 9.8533146924194e-06, "loss": 0.3393, "step": 3038 }, { "epoch": 0.21009332872450742, "grad_norm": 0.8019171506660371, "learning_rate": 9.853180071093531e-06, "loss": 0.3409, "step": 3039 }, { "epoch": 0.21016246111303147, "grad_norm": 0.862387025082726, "learning_rate": 9.85304538894171e-06, "loss": 0.4416, "step": 3040 }, { "epoch": 0.21023159350155549, "grad_norm": 0.8542597082348482, "learning_rate": 9.852910645965629e-06, "loss": 0.4138, "step": 3041 }, { "epoch": 0.2103007258900795, "grad_norm": 0.7880535767512186, "learning_rate": 9.852775842166972e-06, "loss": 0.2992, "step": 3042 }, { "epoch": 0.21036985827860352, "grad_norm": 0.8011708428894214, "learning_rate": 9.85264097754743e-06, "loss": 0.3675, "step": 3043 }, { "epoch": 0.21043899066712754, "grad_norm": 0.7831597928185862, "learning_rate": 9.852506052108692e-06, "loss": 0.4158, "step": 3044 }, { "epoch": 0.21050812305565159, "grad_norm": 0.872829582957569, "learning_rate": 9.852371065852453e-06, "loss": 0.2791, "step": 3045 }, { "epoch": 0.2105772554441756, "grad_norm": 0.7503492698757616, "learning_rate": 9.8522360187804e-06, "loss": 0.3279, "step": 3046 }, { "epoch": 0.21064638783269962, "grad_norm": 0.7607133936169171, "learning_rate": 9.852100910894227e-06, "loss": 0.414, "step": 3047 }, { "epoch": 0.21071552022122364, "grad_norm": 0.7842607760759948, "learning_rate": 9.851965742195631e-06, "loss": 0.3661, "step": 3048 }, { "epoch": 0.21078465260974766, "grad_norm": 0.9208009601489477, "learning_rate": 9.8518305126863e-06, "loss": 0.3164, "step": 3049 }, { "epoch": 0.2108537849982717, "grad_norm": 0.7894929742912625, "learning_rate": 9.851695222367935e-06, "loss": 0.4187, "step": 3050 }, { "epoch": 0.21092291738679572, "grad_norm": 1.161358356864983, "learning_rate": 9.851559871242226e-06, "loss": 0.4541, "step": 3051 }, { "epoch": 0.21099204977531974, "grad_norm": 0.7549626659918027, "learning_rate": 9.851424459310874e-06, "loss": 0.3375, "step": 3052 }, { "epoch": 0.21106118216384376, "grad_norm": 0.716899875240571, "learning_rate": 9.851288986575573e-06, "loss": 0.3128, "step": 3053 }, { "epoch": 0.21113031455236778, "grad_norm": 0.7784569115571435, "learning_rate": 9.851153453038021e-06, "loss": 0.5007, "step": 3054 }, { "epoch": 0.2111994469408918, "grad_norm": 0.8572372481212083, "learning_rate": 9.85101785869992e-06, "loss": 0.4047, "step": 3055 }, { "epoch": 0.21126857932941584, "grad_norm": 1.5486667738033704, "learning_rate": 9.850882203562966e-06, "loss": 0.3579, "step": 3056 }, { "epoch": 0.21133771171793986, "grad_norm": 0.8503605688375775, "learning_rate": 9.850746487628861e-06, "loss": 0.3445, "step": 3057 }, { "epoch": 0.21140684410646388, "grad_norm": 0.8467018652825279, "learning_rate": 9.850610710899304e-06, "loss": 0.5207, "step": 3058 }, { "epoch": 0.2114759764949879, "grad_norm": 0.8103311740622484, "learning_rate": 9.850474873376e-06, "loss": 0.3461, "step": 3059 }, { "epoch": 0.2115451088835119, "grad_norm": 0.7303497120797667, "learning_rate": 9.850338975060648e-06, "loss": 0.3273, "step": 3060 }, { "epoch": 0.21161424127203596, "grad_norm": 0.7620136783159123, "learning_rate": 9.850203015954954e-06, "loss": 0.3854, "step": 3061 }, { "epoch": 0.21168337366055998, "grad_norm": 0.8905547417862354, "learning_rate": 9.850066996060618e-06, "loss": 0.4698, "step": 3062 }, { "epoch": 0.211752506049084, "grad_norm": 0.9248141156189026, "learning_rate": 9.84993091537935e-06, "loss": 0.3392, "step": 3063 }, { "epoch": 0.211821638437608, "grad_norm": 0.718793103991136, "learning_rate": 9.849794773912852e-06, "loss": 0.3137, "step": 3064 }, { "epoch": 0.21189077082613203, "grad_norm": 0.8574334531504163, "learning_rate": 9.849658571662834e-06, "loss": 0.416, "step": 3065 }, { "epoch": 0.21195990321465608, "grad_norm": 0.8717210583058298, "learning_rate": 9.849522308630998e-06, "loss": 0.3281, "step": 3066 }, { "epoch": 0.2120290356031801, "grad_norm": 0.9004016033299866, "learning_rate": 9.849385984819055e-06, "loss": 0.3209, "step": 3067 }, { "epoch": 0.2120981679917041, "grad_norm": 0.7893134817813284, "learning_rate": 9.849249600228712e-06, "loss": 0.4013, "step": 3068 }, { "epoch": 0.21216730038022813, "grad_norm": 0.7744344966966911, "learning_rate": 9.84911315486168e-06, "loss": 0.4337, "step": 3069 }, { "epoch": 0.21223643276875215, "grad_norm": 0.8127288497198801, "learning_rate": 9.848976648719667e-06, "loss": 0.3564, "step": 3070 }, { "epoch": 0.2123055651572762, "grad_norm": 0.8553308850960779, "learning_rate": 9.848840081804384e-06, "loss": 0.3272, "step": 3071 }, { "epoch": 0.2123746975458002, "grad_norm": 0.7173152359400703, "learning_rate": 9.848703454117547e-06, "loss": 0.3369, "step": 3072 }, { "epoch": 0.21244382993432423, "grad_norm": 0.8019613863892349, "learning_rate": 9.848566765660865e-06, "loss": 0.4101, "step": 3073 }, { "epoch": 0.21251296232284825, "grad_norm": 0.7837382079048397, "learning_rate": 9.848430016436048e-06, "loss": 0.3308, "step": 3074 }, { "epoch": 0.21258209471137227, "grad_norm": 0.8424015480873461, "learning_rate": 9.848293206444817e-06, "loss": 0.4105, "step": 3075 }, { "epoch": 0.2126512270998963, "grad_norm": 0.8421463668163707, "learning_rate": 9.84815633568888e-06, "loss": 0.4599, "step": 3076 }, { "epoch": 0.21272035948842033, "grad_norm": 0.7188816207729659, "learning_rate": 9.848019404169956e-06, "loss": 0.3002, "step": 3077 }, { "epoch": 0.21278949187694435, "grad_norm": 0.776956483289278, "learning_rate": 9.84788241188976e-06, "loss": 0.3026, "step": 3078 }, { "epoch": 0.21285862426546837, "grad_norm": 0.8819622532857585, "learning_rate": 9.847745358850011e-06, "loss": 0.4036, "step": 3079 }, { "epoch": 0.21292775665399238, "grad_norm": 0.8238804240650259, "learning_rate": 9.847608245052424e-06, "loss": 0.3975, "step": 3080 }, { "epoch": 0.21299688904251643, "grad_norm": 0.7911092253776929, "learning_rate": 9.847471070498719e-06, "loss": 0.3831, "step": 3081 }, { "epoch": 0.21306602143104045, "grad_norm": 0.9304921324777403, "learning_rate": 9.847333835190613e-06, "loss": 0.4884, "step": 3082 }, { "epoch": 0.21313515381956447, "grad_norm": 0.7528255069340838, "learning_rate": 9.84719653912983e-06, "loss": 0.4396, "step": 3083 }, { "epoch": 0.21320428620808848, "grad_norm": 0.866635018697317, "learning_rate": 9.847059182318088e-06, "loss": 0.3855, "step": 3084 }, { "epoch": 0.2132734185966125, "grad_norm": 0.8382614747562357, "learning_rate": 9.84692176475711e-06, "loss": 0.3048, "step": 3085 }, { "epoch": 0.21334255098513655, "grad_norm": 1.068056872782824, "learning_rate": 9.846784286448616e-06, "loss": 0.3726, "step": 3086 }, { "epoch": 0.21341168337366057, "grad_norm": 2.040455361575307, "learning_rate": 9.846646747394331e-06, "loss": 0.4577, "step": 3087 }, { "epoch": 0.21348081576218458, "grad_norm": 0.801049044133708, "learning_rate": 9.846509147595978e-06, "loss": 0.3186, "step": 3088 }, { "epoch": 0.2135499481507086, "grad_norm": 0.8174030681230218, "learning_rate": 9.846371487055281e-06, "loss": 0.4061, "step": 3089 }, { "epoch": 0.21361908053923262, "grad_norm": 0.8445137305420275, "learning_rate": 9.846233765773969e-06, "loss": 0.4349, "step": 3090 }, { "epoch": 0.21368821292775667, "grad_norm": 1.5968927095170857, "learning_rate": 9.846095983753764e-06, "loss": 0.3832, "step": 3091 }, { "epoch": 0.21375734531628068, "grad_norm": 1.1573295061665534, "learning_rate": 9.845958140996392e-06, "loss": 0.3295, "step": 3092 }, { "epoch": 0.2138264777048047, "grad_norm": 0.8631548260305221, "learning_rate": 9.845820237503584e-06, "loss": 0.4218, "step": 3093 }, { "epoch": 0.21389561009332872, "grad_norm": 0.7728255846312452, "learning_rate": 9.845682273277068e-06, "loss": 0.382, "step": 3094 }, { "epoch": 0.21396474248185274, "grad_norm": 0.756174272856675, "learning_rate": 9.845544248318572e-06, "loss": 0.327, "step": 3095 }, { "epoch": 0.21403387487037678, "grad_norm": 0.9580885925014242, "learning_rate": 9.845406162629825e-06, "loss": 0.3491, "step": 3096 }, { "epoch": 0.2141030072589008, "grad_norm": 0.781033305838007, "learning_rate": 9.845268016212559e-06, "loss": 0.4474, "step": 3097 }, { "epoch": 0.21417213964742482, "grad_norm": 0.7045762120235403, "learning_rate": 9.845129809068505e-06, "loss": 0.3536, "step": 3098 }, { "epoch": 0.21424127203594884, "grad_norm": 0.8205288394365012, "learning_rate": 9.844991541199396e-06, "loss": 0.2679, "step": 3099 }, { "epoch": 0.21431040442447286, "grad_norm": 1.2750675567348162, "learning_rate": 9.844853212606965e-06, "loss": 0.4013, "step": 3100 }, { "epoch": 0.2143795368129969, "grad_norm": 0.7885592837950182, "learning_rate": 9.844714823292944e-06, "loss": 0.4243, "step": 3101 }, { "epoch": 0.21444866920152092, "grad_norm": 0.7353075802648306, "learning_rate": 9.84457637325907e-06, "loss": 0.3512, "step": 3102 }, { "epoch": 0.21451780159004494, "grad_norm": 0.6764238570445186, "learning_rate": 9.844437862507075e-06, "loss": 0.2922, "step": 3103 }, { "epoch": 0.21458693397856896, "grad_norm": 0.8391230123534341, "learning_rate": 9.844299291038697e-06, "loss": 0.4082, "step": 3104 }, { "epoch": 0.21465606636709297, "grad_norm": 0.8033525915417138, "learning_rate": 9.844160658855671e-06, "loss": 0.4465, "step": 3105 }, { "epoch": 0.21472519875561702, "grad_norm": 0.8149203248368132, "learning_rate": 9.844021965959738e-06, "loss": 0.3244, "step": 3106 }, { "epoch": 0.21479433114414104, "grad_norm": 0.8363599369334451, "learning_rate": 9.843883212352635e-06, "loss": 0.3628, "step": 3107 }, { "epoch": 0.21486346353266506, "grad_norm": 0.7933721427572428, "learning_rate": 9.843744398036097e-06, "loss": 0.5119, "step": 3108 }, { "epoch": 0.21493259592118907, "grad_norm": 0.8005182317977577, "learning_rate": 9.84360552301187e-06, "loss": 0.2969, "step": 3109 }, { "epoch": 0.2150017283097131, "grad_norm": 1.3477902421469048, "learning_rate": 9.84346658728169e-06, "loss": 0.3101, "step": 3110 }, { "epoch": 0.21507086069823714, "grad_norm": 0.8701942187367371, "learning_rate": 9.8433275908473e-06, "loss": 0.3839, "step": 3111 }, { "epoch": 0.21513999308676116, "grad_norm": 0.7462322662207287, "learning_rate": 9.843188533710442e-06, "loss": 0.3607, "step": 3112 }, { "epoch": 0.21520912547528517, "grad_norm": 0.7225206625458699, "learning_rate": 9.84304941587286e-06, "loss": 0.3188, "step": 3113 }, { "epoch": 0.2152782578638092, "grad_norm": 0.9773523431944329, "learning_rate": 9.842910237336295e-06, "loss": 0.3333, "step": 3114 }, { "epoch": 0.2153473902523332, "grad_norm": 0.8384081464495363, "learning_rate": 9.842770998102493e-06, "loss": 0.4559, "step": 3115 }, { "epoch": 0.21541652264085726, "grad_norm": 0.7151902921473381, "learning_rate": 9.8426316981732e-06, "loss": 0.3287, "step": 3116 }, { "epoch": 0.21548565502938127, "grad_norm": 0.7622822693849403, "learning_rate": 9.842492337550159e-06, "loss": 0.3122, "step": 3117 }, { "epoch": 0.2155547874179053, "grad_norm": 0.9574247958788108, "learning_rate": 9.84235291623512e-06, "loss": 0.4069, "step": 3118 }, { "epoch": 0.2156239198064293, "grad_norm": 0.8609790369246147, "learning_rate": 9.842213434229829e-06, "loss": 0.38, "step": 3119 }, { "epoch": 0.21569305219495333, "grad_norm": 0.882272743296657, "learning_rate": 9.84207389153603e-06, "loss": 0.3124, "step": 3120 }, { "epoch": 0.21576218458347735, "grad_norm": 0.9787542462210924, "learning_rate": 9.841934288155478e-06, "loss": 0.3752, "step": 3121 }, { "epoch": 0.2158313169720014, "grad_norm": 1.2513383832868739, "learning_rate": 9.841794624089923e-06, "loss": 0.4682, "step": 3122 }, { "epoch": 0.2159004493605254, "grad_norm": 0.8125080235705067, "learning_rate": 9.841654899341111e-06, "loss": 0.3635, "step": 3123 }, { "epoch": 0.21596958174904943, "grad_norm": 0.8298592903395959, "learning_rate": 9.841515113910795e-06, "loss": 0.28, "step": 3124 }, { "epoch": 0.21603871413757345, "grad_norm": 0.9741696615915499, "learning_rate": 9.841375267800729e-06, "loss": 0.3834, "step": 3125 }, { "epoch": 0.21610784652609746, "grad_norm": 0.960254550939453, "learning_rate": 9.841235361012661e-06, "loss": 0.4719, "step": 3126 }, { "epoch": 0.2161769789146215, "grad_norm": 0.823531406665501, "learning_rate": 9.841095393548349e-06, "loss": 0.3286, "step": 3127 }, { "epoch": 0.21624611130314553, "grad_norm": 0.743497860544046, "learning_rate": 9.840955365409546e-06, "loss": 0.3159, "step": 3128 }, { "epoch": 0.21631524369166955, "grad_norm": 1.0496696273587536, "learning_rate": 9.840815276598004e-06, "loss": 0.3785, "step": 3129 }, { "epoch": 0.21638437608019356, "grad_norm": 0.7877567913410144, "learning_rate": 9.840675127115485e-06, "loss": 0.3923, "step": 3130 }, { "epoch": 0.21645350846871758, "grad_norm": 0.8525411032804252, "learning_rate": 9.840534916963738e-06, "loss": 0.3177, "step": 3131 }, { "epoch": 0.21652264085724163, "grad_norm": 0.823532445328727, "learning_rate": 9.840394646144527e-06, "loss": 0.3289, "step": 3132 }, { "epoch": 0.21659177324576565, "grad_norm": 0.8799812864606193, "learning_rate": 9.840254314659605e-06, "loss": 0.5266, "step": 3133 }, { "epoch": 0.21666090563428966, "grad_norm": 0.9763480101973472, "learning_rate": 9.840113922510735e-06, "loss": 0.379, "step": 3134 }, { "epoch": 0.21673003802281368, "grad_norm": 0.8488833114303265, "learning_rate": 9.839973469699675e-06, "loss": 0.3293, "step": 3135 }, { "epoch": 0.2167991704113377, "grad_norm": 0.9588027870887359, "learning_rate": 9.839832956228182e-06, "loss": 0.4112, "step": 3136 }, { "epoch": 0.21686830279986175, "grad_norm": 0.7922384242385218, "learning_rate": 9.839692382098023e-06, "loss": 0.3983, "step": 3137 }, { "epoch": 0.21693743518838576, "grad_norm": 1.9442394101514475, "learning_rate": 9.839551747310954e-06, "loss": 0.3775, "step": 3138 }, { "epoch": 0.21700656757690978, "grad_norm": 0.793724234684763, "learning_rate": 9.839411051868743e-06, "loss": 0.2957, "step": 3139 }, { "epoch": 0.2170756999654338, "grad_norm": 0.927167557509538, "learning_rate": 9.839270295773148e-06, "loss": 0.4564, "step": 3140 }, { "epoch": 0.21714483235395782, "grad_norm": 0.8538899906871553, "learning_rate": 9.839129479025939e-06, "loss": 0.3402, "step": 3141 }, { "epoch": 0.21721396474248186, "grad_norm": 0.6801382460935522, "learning_rate": 9.838988601628877e-06, "loss": 0.2865, "step": 3142 }, { "epoch": 0.21728309713100588, "grad_norm": 0.8350429157394289, "learning_rate": 9.838847663583725e-06, "loss": 0.3966, "step": 3143 }, { "epoch": 0.2173522295195299, "grad_norm": 0.7972383337314499, "learning_rate": 9.838706664892257e-06, "loss": 0.4, "step": 3144 }, { "epoch": 0.21742136190805392, "grad_norm": 0.8108691724700592, "learning_rate": 9.838565605556234e-06, "loss": 0.3141, "step": 3145 }, { "epoch": 0.21749049429657794, "grad_norm": 0.8057699157339467, "learning_rate": 9.838424485577427e-06, "loss": 0.3706, "step": 3146 }, { "epoch": 0.21755962668510198, "grad_norm": 0.7945412955883966, "learning_rate": 9.838283304957603e-06, "loss": 0.4057, "step": 3147 }, { "epoch": 0.217628759073626, "grad_norm": 0.8299578632087699, "learning_rate": 9.838142063698533e-06, "loss": 0.3591, "step": 3148 }, { "epoch": 0.21769789146215002, "grad_norm": 0.7179731664225101, "learning_rate": 9.838000761801985e-06, "loss": 0.3078, "step": 3149 }, { "epoch": 0.21776702385067404, "grad_norm": 0.8168829906449451, "learning_rate": 9.837859399269731e-06, "loss": 0.4064, "step": 3150 }, { "epoch": 0.21783615623919805, "grad_norm": 0.7743390007782485, "learning_rate": 9.837717976103545e-06, "loss": 0.487, "step": 3151 }, { "epoch": 0.2179052886277221, "grad_norm": 0.7067954720548928, "learning_rate": 9.837576492305196e-06, "loss": 0.3595, "step": 3152 }, { "epoch": 0.21797442101624612, "grad_norm": 0.9599955869790014, "learning_rate": 9.837434947876458e-06, "loss": 0.3182, "step": 3153 }, { "epoch": 0.21804355340477014, "grad_norm": 0.8360779416850056, "learning_rate": 9.837293342819107e-06, "loss": 0.497, "step": 3154 }, { "epoch": 0.21811268579329415, "grad_norm": 1.0353403871882032, "learning_rate": 9.837151677134917e-06, "loss": 0.3892, "step": 3155 }, { "epoch": 0.21818181818181817, "grad_norm": 1.0750620893003544, "learning_rate": 9.837009950825663e-06, "loss": 0.3301, "step": 3156 }, { "epoch": 0.21825095057034222, "grad_norm": 0.7407283065892771, "learning_rate": 9.83686816389312e-06, "loss": 0.4138, "step": 3157 }, { "epoch": 0.21832008295886624, "grad_norm": 0.9171386327264945, "learning_rate": 9.836726316339067e-06, "loss": 0.3943, "step": 3158 }, { "epoch": 0.21838921534739025, "grad_norm": 0.874832072335666, "learning_rate": 9.83658440816528e-06, "loss": 0.3658, "step": 3159 }, { "epoch": 0.21845834773591427, "grad_norm": 1.0934015130015617, "learning_rate": 9.83644243937354e-06, "loss": 0.3668, "step": 3160 }, { "epoch": 0.2185274801244383, "grad_norm": 0.8347202073688536, "learning_rate": 9.836300409965625e-06, "loss": 0.447, "step": 3161 }, { "epoch": 0.21859661251296233, "grad_norm": 0.8759448743039451, "learning_rate": 9.836158319943314e-06, "loss": 0.4008, "step": 3162 }, { "epoch": 0.21866574490148635, "grad_norm": 1.0521497911884294, "learning_rate": 9.83601616930839e-06, "loss": 0.3314, "step": 3163 }, { "epoch": 0.21873487729001037, "grad_norm": 0.939749348578903, "learning_rate": 9.835873958062634e-06, "loss": 0.3679, "step": 3164 }, { "epoch": 0.2188040096785344, "grad_norm": 0.863293392410104, "learning_rate": 9.835731686207828e-06, "loss": 0.5079, "step": 3165 }, { "epoch": 0.2188731420670584, "grad_norm": 1.2836967616893669, "learning_rate": 9.835589353745754e-06, "loss": 0.3474, "step": 3166 }, { "epoch": 0.21894227445558245, "grad_norm": 0.8595309609607537, "learning_rate": 9.835446960678197e-06, "loss": 0.3452, "step": 3167 }, { "epoch": 0.21901140684410647, "grad_norm": 0.7645482480467581, "learning_rate": 9.835304507006942e-06, "loss": 0.3328, "step": 3168 }, { "epoch": 0.2190805392326305, "grad_norm": 1.130411978518501, "learning_rate": 9.835161992733773e-06, "loss": 0.4835, "step": 3169 }, { "epoch": 0.2191496716211545, "grad_norm": 0.9631379480539038, "learning_rate": 9.835019417860479e-06, "loss": 0.3688, "step": 3170 }, { "epoch": 0.21921880400967853, "grad_norm": 0.6959770946192495, "learning_rate": 9.834876782388845e-06, "loss": 0.3242, "step": 3171 }, { "epoch": 0.21928793639820257, "grad_norm": 0.828836048607069, "learning_rate": 9.834734086320656e-06, "loss": 0.4562, "step": 3172 }, { "epoch": 0.2193570687867266, "grad_norm": 0.9821578309128438, "learning_rate": 9.834591329657706e-06, "loss": 0.4292, "step": 3173 }, { "epoch": 0.2194262011752506, "grad_norm": 0.9487546574153823, "learning_rate": 9.83444851240178e-06, "loss": 0.337, "step": 3174 }, { "epoch": 0.21949533356377463, "grad_norm": 0.8542553863223591, "learning_rate": 9.83430563455467e-06, "loss": 0.3729, "step": 3175 }, { "epoch": 0.21956446595229864, "grad_norm": 1.1744012530146397, "learning_rate": 9.834162696118165e-06, "loss": 0.4967, "step": 3176 }, { "epoch": 0.2196335983408227, "grad_norm": 0.8416589361829804, "learning_rate": 9.83401969709406e-06, "loss": 0.3178, "step": 3177 }, { "epoch": 0.2197027307293467, "grad_norm": 0.7075573327602779, "learning_rate": 9.833876637484144e-06, "loss": 0.3126, "step": 3178 }, { "epoch": 0.21977186311787072, "grad_norm": 0.9005608677453438, "learning_rate": 9.83373351729021e-06, "loss": 0.432, "step": 3179 }, { "epoch": 0.21984099550639474, "grad_norm": 0.8406999095524519, "learning_rate": 9.833590336514051e-06, "loss": 0.405, "step": 3180 }, { "epoch": 0.21991012789491876, "grad_norm": 0.9043602709890121, "learning_rate": 9.833447095157465e-06, "loss": 0.415, "step": 3181 }, { "epoch": 0.2199792602834428, "grad_norm": 1.083472824276575, "learning_rate": 9.833303793222246e-06, "loss": 0.3738, "step": 3182 }, { "epoch": 0.22004839267196682, "grad_norm": 0.8475025571561987, "learning_rate": 9.833160430710189e-06, "loss": 0.4219, "step": 3183 }, { "epoch": 0.22011752506049084, "grad_norm": 0.8267417618572326, "learning_rate": 9.833017007623089e-06, "loss": 0.3935, "step": 3184 }, { "epoch": 0.22018665744901486, "grad_norm": 0.8828074663091786, "learning_rate": 9.832873523962748e-06, "loss": 0.3873, "step": 3185 }, { "epoch": 0.22025578983753888, "grad_norm": 0.8348368179783059, "learning_rate": 9.832729979730961e-06, "loss": 0.3996, "step": 3186 }, { "epoch": 0.2203249222260629, "grad_norm": 0.7639491277247275, "learning_rate": 9.832586374929528e-06, "loss": 0.399, "step": 3187 }, { "epoch": 0.22039405461458694, "grad_norm": 1.055099184325823, "learning_rate": 9.832442709560251e-06, "loss": 0.3545, "step": 3188 }, { "epoch": 0.22046318700311096, "grad_norm": 0.824992099422881, "learning_rate": 9.832298983624927e-06, "loss": 0.3355, "step": 3189 }, { "epoch": 0.22053231939163498, "grad_norm": 2.0950908168154174, "learning_rate": 9.832155197125357e-06, "loss": 0.4419, "step": 3190 }, { "epoch": 0.220601451780159, "grad_norm": 0.7733173174789489, "learning_rate": 9.832011350063347e-06, "loss": 0.3583, "step": 3191 }, { "epoch": 0.22067058416868301, "grad_norm": 0.8372624071654725, "learning_rate": 9.831867442440699e-06, "loss": 0.3132, "step": 3192 }, { "epoch": 0.22073971655720706, "grad_norm": 0.9036703161700607, "learning_rate": 9.831723474259213e-06, "loss": 0.4195, "step": 3193 }, { "epoch": 0.22080884894573108, "grad_norm": 0.9344727045154493, "learning_rate": 9.831579445520697e-06, "loss": 0.4876, "step": 3194 }, { "epoch": 0.2208779813342551, "grad_norm": 0.8307010422001656, "learning_rate": 9.831435356226954e-06, "loss": 0.391, "step": 3195 }, { "epoch": 0.22094711372277911, "grad_norm": 0.7506326806812272, "learning_rate": 9.831291206379792e-06, "loss": 0.3514, "step": 3196 }, { "epoch": 0.22101624611130313, "grad_norm": 0.8251301145521766, "learning_rate": 9.831146995981017e-06, "loss": 0.4418, "step": 3197 }, { "epoch": 0.22108537849982718, "grad_norm": 0.8240309970342129, "learning_rate": 9.831002725032435e-06, "loss": 0.3868, "step": 3198 }, { "epoch": 0.2211545108883512, "grad_norm": 0.8899279508651201, "learning_rate": 9.830858393535855e-06, "loss": 0.3072, "step": 3199 }, { "epoch": 0.22122364327687521, "grad_norm": 0.7692502578654712, "learning_rate": 9.830714001493088e-06, "loss": 0.3561, "step": 3200 }, { "epoch": 0.22129277566539923, "grad_norm": 0.876109379244575, "learning_rate": 9.83056954890594e-06, "loss": 0.4624, "step": 3201 }, { "epoch": 0.22136190805392325, "grad_norm": 0.8662777269465445, "learning_rate": 9.830425035776224e-06, "loss": 0.3205, "step": 3202 }, { "epoch": 0.2214310404424473, "grad_norm": 0.9849373990938054, "learning_rate": 9.830280462105748e-06, "loss": 0.3804, "step": 3203 }, { "epoch": 0.22150017283097131, "grad_norm": 0.8519415687790138, "learning_rate": 9.830135827896328e-06, "loss": 0.5252, "step": 3204 }, { "epoch": 0.22156930521949533, "grad_norm": 0.8187376133624882, "learning_rate": 9.829991133149777e-06, "loss": 0.4024, "step": 3205 }, { "epoch": 0.22163843760801935, "grad_norm": 0.7281889601247636, "learning_rate": 9.829846377867904e-06, "loss": 0.4106, "step": 3206 }, { "epoch": 0.22170756999654337, "grad_norm": 0.799302590757484, "learning_rate": 9.829701562052527e-06, "loss": 0.3984, "step": 3207 }, { "epoch": 0.22177670238506741, "grad_norm": 0.963207892043523, "learning_rate": 9.829556685705461e-06, "loss": 0.422, "step": 3208 }, { "epoch": 0.22184583477359143, "grad_norm": 0.8001522715205142, "learning_rate": 9.82941174882852e-06, "loss": 0.3204, "step": 3209 }, { "epoch": 0.22191496716211545, "grad_norm": 0.9149162504836628, "learning_rate": 9.82926675142352e-06, "loss": 0.3832, "step": 3210 }, { "epoch": 0.22198409955063947, "grad_norm": 0.7855491105695478, "learning_rate": 9.82912169349228e-06, "loss": 0.418, "step": 3211 }, { "epoch": 0.2220532319391635, "grad_norm": 0.7531894579126525, "learning_rate": 9.82897657503662e-06, "loss": 0.3965, "step": 3212 }, { "epoch": 0.22212236432768753, "grad_norm": 0.8291255404554314, "learning_rate": 9.828831396058354e-06, "loss": 0.3642, "step": 3213 }, { "epoch": 0.22219149671621155, "grad_norm": 0.7894711311938468, "learning_rate": 9.828686156559303e-06, "loss": 0.3425, "step": 3214 }, { "epoch": 0.22226062910473557, "grad_norm": 0.8206765103977384, "learning_rate": 9.82854085654129e-06, "loss": 0.5167, "step": 3215 }, { "epoch": 0.2223297614932596, "grad_norm": 1.3260511687036227, "learning_rate": 9.828395496006133e-06, "loss": 0.3498, "step": 3216 }, { "epoch": 0.2223988938817836, "grad_norm": 0.7535067148116444, "learning_rate": 9.828250074955657e-06, "loss": 0.2617, "step": 3217 }, { "epoch": 0.22246802627030765, "grad_norm": 0.8002605170906276, "learning_rate": 9.828104593391682e-06, "loss": 0.3555, "step": 3218 }, { "epoch": 0.22253715865883167, "grad_norm": 0.8413471375684264, "learning_rate": 9.82795905131603e-06, "loss": 0.4395, "step": 3219 }, { "epoch": 0.2226062910473557, "grad_norm": 0.8200839597525094, "learning_rate": 9.827813448730528e-06, "loss": 0.3414, "step": 3220 }, { "epoch": 0.2226754234358797, "grad_norm": 1.0009833668141146, "learning_rate": 9.827667785637001e-06, "loss": 0.3299, "step": 3221 }, { "epoch": 0.22274455582440372, "grad_norm": 0.8973858685692135, "learning_rate": 9.827522062037273e-06, "loss": 0.5081, "step": 3222 }, { "epoch": 0.22281368821292777, "grad_norm": 0.9043704601440353, "learning_rate": 9.827376277933171e-06, "loss": 0.3845, "step": 3223 }, { "epoch": 0.2228828206014518, "grad_norm": 1.160881712095496, "learning_rate": 9.827230433326525e-06, "loss": 0.3017, "step": 3224 }, { "epoch": 0.2229519529899758, "grad_norm": 0.9416919929358207, "learning_rate": 9.827084528219157e-06, "loss": 0.3711, "step": 3225 }, { "epoch": 0.22302108537849982, "grad_norm": 1.1990741870722756, "learning_rate": 9.8269385626129e-06, "loss": 0.4097, "step": 3226 }, { "epoch": 0.22309021776702384, "grad_norm": 0.9399610730407216, "learning_rate": 9.826792536509581e-06, "loss": 0.3216, "step": 3227 }, { "epoch": 0.2231593501555479, "grad_norm": 0.8369092964177622, "learning_rate": 9.826646449911033e-06, "loss": 0.3243, "step": 3228 }, { "epoch": 0.2232284825440719, "grad_norm": 1.4696077940096557, "learning_rate": 9.826500302819083e-06, "loss": 0.4094, "step": 3229 }, { "epoch": 0.22329761493259592, "grad_norm": 1.1657339452249538, "learning_rate": 9.826354095235565e-06, "loss": 0.4228, "step": 3230 }, { "epoch": 0.22336674732111994, "grad_norm": 0.8218193812763714, "learning_rate": 9.826207827162315e-06, "loss": 0.3304, "step": 3231 }, { "epoch": 0.22343587970964396, "grad_norm": 11.138524100000938, "learning_rate": 9.826061498601161e-06, "loss": 0.3039, "step": 3232 }, { "epoch": 0.223505012098168, "grad_norm": 0.8472482267969285, "learning_rate": 9.825915109553937e-06, "loss": 0.4394, "step": 3233 }, { "epoch": 0.22357414448669202, "grad_norm": 0.9653411913059539, "learning_rate": 9.82576866002248e-06, "loss": 0.3687, "step": 3234 }, { "epoch": 0.22364327687521604, "grad_norm": 1.0534065341407692, "learning_rate": 9.825622150008625e-06, "loss": 0.3495, "step": 3235 }, { "epoch": 0.22371240926374006, "grad_norm": 0.8804948385509499, "learning_rate": 9.825475579514207e-06, "loss": 0.3908, "step": 3236 }, { "epoch": 0.22378154165226408, "grad_norm": 0.8297976040644177, "learning_rate": 9.825328948541066e-06, "loss": 0.4385, "step": 3237 }, { "epoch": 0.22385067404078812, "grad_norm": 0.9130560992396372, "learning_rate": 9.825182257091036e-06, "loss": 0.3224, "step": 3238 }, { "epoch": 0.22391980642931214, "grad_norm": 0.8776995892640466, "learning_rate": 9.825035505165961e-06, "loss": 0.4549, "step": 3239 }, { "epoch": 0.22398893881783616, "grad_norm": 0.7890335613676595, "learning_rate": 9.824888692767673e-06, "loss": 0.4475, "step": 3240 }, { "epoch": 0.22405807120636018, "grad_norm": 0.8080723655052465, "learning_rate": 9.824741819898016e-06, "loss": 0.2802, "step": 3241 }, { "epoch": 0.2241272035948842, "grad_norm": 0.910961253034727, "learning_rate": 9.824594886558831e-06, "loss": 0.3458, "step": 3242 }, { "epoch": 0.22419633598340824, "grad_norm": 0.9331053289876581, "learning_rate": 9.824447892751961e-06, "loss": 0.4585, "step": 3243 }, { "epoch": 0.22426546837193226, "grad_norm": 0.8028202424608658, "learning_rate": 9.824300838479244e-06, "loss": 0.3777, "step": 3244 }, { "epoch": 0.22433460076045628, "grad_norm": 0.7842488682531591, "learning_rate": 9.824153723742524e-06, "loss": 0.3295, "step": 3245 }, { "epoch": 0.2244037331489803, "grad_norm": 0.7890318915898391, "learning_rate": 9.824006548543649e-06, "loss": 0.3798, "step": 3246 }, { "epoch": 0.2244728655375043, "grad_norm": 0.9291625471760547, "learning_rate": 9.823859312884459e-06, "loss": 0.4749, "step": 3247 }, { "epoch": 0.22454199792602833, "grad_norm": 1.138782857940782, "learning_rate": 9.823712016766802e-06, "loss": 0.3225, "step": 3248 }, { "epoch": 0.22461113031455238, "grad_norm": 0.903745559755807, "learning_rate": 9.823564660192523e-06, "loss": 0.3308, "step": 3249 }, { "epoch": 0.2246802627030764, "grad_norm": 0.7770268175494972, "learning_rate": 9.823417243163469e-06, "loss": 0.3769, "step": 3250 }, { "epoch": 0.2247493950916004, "grad_norm": 0.789970915109616, "learning_rate": 9.823269765681487e-06, "loss": 0.4028, "step": 3251 }, { "epoch": 0.22481852748012443, "grad_norm": 0.8401637853971349, "learning_rate": 9.823122227748426e-06, "loss": 0.3774, "step": 3252 }, { "epoch": 0.22488765986864845, "grad_norm": 0.8207839234356742, "learning_rate": 9.822974629366135e-06, "loss": 0.3085, "step": 3253 }, { "epoch": 0.2249567922571725, "grad_norm": 1.26986302646888, "learning_rate": 9.822826970536466e-06, "loss": 0.4633, "step": 3254 }, { "epoch": 0.2250259246456965, "grad_norm": 0.8605156975894392, "learning_rate": 9.822679251261266e-06, "loss": 0.4176, "step": 3255 }, { "epoch": 0.22509505703422053, "grad_norm": 0.9277255284539634, "learning_rate": 9.822531471542387e-06, "loss": 0.3881, "step": 3256 }, { "epoch": 0.22516418942274455, "grad_norm": 0.9554974396123388, "learning_rate": 9.822383631381684e-06, "loss": 0.4021, "step": 3257 }, { "epoch": 0.22523332181126857, "grad_norm": 0.905126274093579, "learning_rate": 9.822235730781007e-06, "loss": 0.4593, "step": 3258 }, { "epoch": 0.2253024541997926, "grad_norm": 0.7745835975116733, "learning_rate": 9.822087769742212e-06, "loss": 0.3174, "step": 3259 }, { "epoch": 0.22537158658831663, "grad_norm": 1.427343557100313, "learning_rate": 9.821939748267152e-06, "loss": 0.2775, "step": 3260 }, { "epoch": 0.22544071897684065, "grad_norm": 0.8698208894022124, "learning_rate": 9.821791666357682e-06, "loss": 0.3901, "step": 3261 }, { "epoch": 0.22550985136536467, "grad_norm": 0.7822974463528596, "learning_rate": 9.821643524015657e-06, "loss": 0.4911, "step": 3262 }, { "epoch": 0.22557898375388868, "grad_norm": 0.9632363301716679, "learning_rate": 9.821495321242938e-06, "loss": 0.3821, "step": 3263 }, { "epoch": 0.22564811614241273, "grad_norm": 0.8852291326635733, "learning_rate": 9.821347058041376e-06, "loss": 0.3463, "step": 3264 }, { "epoch": 0.22571724853093675, "grad_norm": 0.8197353659354741, "learning_rate": 9.821198734412835e-06, "loss": 0.4401, "step": 3265 }, { "epoch": 0.22578638091946077, "grad_norm": 0.8396407009181697, "learning_rate": 9.821050350359173e-06, "loss": 0.3507, "step": 3266 }, { "epoch": 0.22585551330798478, "grad_norm": 0.7159572672060724, "learning_rate": 9.820901905882246e-06, "loss": 0.289, "step": 3267 }, { "epoch": 0.2259246456965088, "grad_norm": 0.8616996301917648, "learning_rate": 9.820753400983917e-06, "loss": 0.393, "step": 3268 }, { "epoch": 0.22599377808503285, "grad_norm": 0.762198395885136, "learning_rate": 9.820604835666048e-06, "loss": 0.4338, "step": 3269 }, { "epoch": 0.22606291047355687, "grad_norm": 0.8878735878282331, "learning_rate": 9.8204562099305e-06, "loss": 0.348, "step": 3270 }, { "epoch": 0.22613204286208088, "grad_norm": 0.8848477417233666, "learning_rate": 9.820307523779136e-06, "loss": 0.3014, "step": 3271 }, { "epoch": 0.2262011752506049, "grad_norm": 0.8534313687910527, "learning_rate": 9.82015877721382e-06, "loss": 0.4677, "step": 3272 }, { "epoch": 0.22627030763912892, "grad_norm": 0.8266613923388526, "learning_rate": 9.820009970236414e-06, "loss": 0.4016, "step": 3273 }, { "epoch": 0.22633944002765297, "grad_norm": 1.0307457056243043, "learning_rate": 9.819861102848784e-06, "loss": 0.3258, "step": 3274 }, { "epoch": 0.22640857241617698, "grad_norm": 0.8802454892136222, "learning_rate": 9.819712175052799e-06, "loss": 0.3968, "step": 3275 }, { "epoch": 0.226477704804701, "grad_norm": 0.8479480111590963, "learning_rate": 9.819563186850322e-06, "loss": 0.4466, "step": 3276 }, { "epoch": 0.22654683719322502, "grad_norm": 0.8119038968807988, "learning_rate": 9.819414138243222e-06, "loss": 0.3292, "step": 3277 }, { "epoch": 0.22661596958174904, "grad_norm": 1.02690266979076, "learning_rate": 9.819265029233365e-06, "loss": 0.3153, "step": 3278 }, { "epoch": 0.22668510197027308, "grad_norm": 0.9356090750506402, "learning_rate": 9.819115859822623e-06, "loss": 0.4203, "step": 3279 }, { "epoch": 0.2267542343587971, "grad_norm": 0.7443943976185733, "learning_rate": 9.818966630012863e-06, "loss": 0.3831, "step": 3280 }, { "epoch": 0.22682336674732112, "grad_norm": 0.8198221300812081, "learning_rate": 9.818817339805955e-06, "loss": 0.3494, "step": 3281 }, { "epoch": 0.22689249913584514, "grad_norm": 0.8484497868427738, "learning_rate": 9.818667989203771e-06, "loss": 0.3385, "step": 3282 }, { "epoch": 0.22696163152436916, "grad_norm": 0.9567093328808871, "learning_rate": 9.818518578208187e-06, "loss": 0.486, "step": 3283 }, { "epoch": 0.2270307639128932, "grad_norm": 0.8035739431460585, "learning_rate": 9.818369106821069e-06, "loss": 0.3767, "step": 3284 }, { "epoch": 0.22709989630141722, "grad_norm": 0.8778291954622899, "learning_rate": 9.818219575044292e-06, "loss": 0.3343, "step": 3285 }, { "epoch": 0.22716902868994124, "grad_norm": 0.7682451083945271, "learning_rate": 9.818069982879732e-06, "loss": 0.4348, "step": 3286 }, { "epoch": 0.22723816107846526, "grad_norm": 3.7872425232292684, "learning_rate": 9.817920330329262e-06, "loss": 0.465, "step": 3287 }, { "epoch": 0.22730729346698927, "grad_norm": 0.8122306156867134, "learning_rate": 9.817770617394759e-06, "loss": 0.2919, "step": 3288 }, { "epoch": 0.22737642585551332, "grad_norm": 0.7839883927140985, "learning_rate": 9.817620844078102e-06, "loss": 0.3428, "step": 3289 }, { "epoch": 0.22744555824403734, "grad_norm": 0.8500375170103571, "learning_rate": 9.817471010381162e-06, "loss": 0.4315, "step": 3290 }, { "epoch": 0.22751469063256136, "grad_norm": 0.7803117681961996, "learning_rate": 9.81732111630582e-06, "loss": 0.3405, "step": 3291 }, { "epoch": 0.22758382302108537, "grad_norm": 0.80548357930654, "learning_rate": 9.817171161853956e-06, "loss": 0.2496, "step": 3292 }, { "epoch": 0.2276529554096094, "grad_norm": 0.7789757471836964, "learning_rate": 9.817021147027447e-06, "loss": 0.4293, "step": 3293 }, { "epoch": 0.22772208779813344, "grad_norm": 1.0892756417109752, "learning_rate": 9.816871071828175e-06, "loss": 0.4208, "step": 3294 }, { "epoch": 0.22779122018665746, "grad_norm": 0.8988539191863365, "learning_rate": 9.816720936258022e-06, "loss": 0.3296, "step": 3295 }, { "epoch": 0.22786035257518147, "grad_norm": 0.7676504850321949, "learning_rate": 9.816570740318864e-06, "loss": 0.3066, "step": 3296 }, { "epoch": 0.2279294849637055, "grad_norm": 0.9449119995542474, "learning_rate": 9.81642048401259e-06, "loss": 0.4466, "step": 3297 }, { "epoch": 0.2279986173522295, "grad_norm": 1.082269163592521, "learning_rate": 9.81627016734108e-06, "loss": 0.3979, "step": 3298 }, { "epoch": 0.22806774974075356, "grad_norm": 0.8708549752354495, "learning_rate": 9.816119790306218e-06, "loss": 0.2913, "step": 3299 }, { "epoch": 0.22813688212927757, "grad_norm": 0.7932568701847472, "learning_rate": 9.815969352909892e-06, "loss": 0.3573, "step": 3300 }, { "epoch": 0.2282060145178016, "grad_norm": 0.9190872322603997, "learning_rate": 9.81581885515398e-06, "loss": 0.4616, "step": 3301 }, { "epoch": 0.2282751469063256, "grad_norm": 0.9420401818895733, "learning_rate": 9.815668297040374e-06, "loss": 0.3729, "step": 3302 }, { "epoch": 0.22834427929484963, "grad_norm": 0.7642976639517011, "learning_rate": 9.815517678570963e-06, "loss": 0.3141, "step": 3303 }, { "epoch": 0.22841341168337367, "grad_norm": 0.8097227458188516, "learning_rate": 9.815366999747629e-06, "loss": 0.4412, "step": 3304 }, { "epoch": 0.2284825440718977, "grad_norm": 0.910668794696256, "learning_rate": 9.815216260572264e-06, "loss": 0.439, "step": 3305 }, { "epoch": 0.2285516764604217, "grad_norm": 0.7886261524691379, "learning_rate": 9.815065461046755e-06, "loss": 0.3371, "step": 3306 }, { "epoch": 0.22862080884894573, "grad_norm": 0.8319732496002626, "learning_rate": 9.814914601172994e-06, "loss": 0.3624, "step": 3307 }, { "epoch": 0.22868994123746975, "grad_norm": 0.7996319581700368, "learning_rate": 9.814763680952873e-06, "loss": 0.4587, "step": 3308 }, { "epoch": 0.2287590736259938, "grad_norm": 0.8243153779458423, "learning_rate": 9.81461270038828e-06, "loss": 0.3799, "step": 3309 }, { "epoch": 0.2288282060145178, "grad_norm": 0.8448307786432498, "learning_rate": 9.814461659481107e-06, "loss": 0.3111, "step": 3310 }, { "epoch": 0.22889733840304183, "grad_norm": 0.9129321599777885, "learning_rate": 9.814310558233251e-06, "loss": 0.4158, "step": 3311 }, { "epoch": 0.22896647079156585, "grad_norm": 1.089088118856781, "learning_rate": 9.814159396646603e-06, "loss": 0.38, "step": 3312 }, { "epoch": 0.22903560318008986, "grad_norm": 0.819736599835782, "learning_rate": 9.81400817472306e-06, "loss": 0.3657, "step": 3313 }, { "epoch": 0.22910473556861388, "grad_norm": 1.2367295603234025, "learning_rate": 9.813856892464512e-06, "loss": 0.3154, "step": 3314 }, { "epoch": 0.22917386795713793, "grad_norm": 1.3141836299399043, "learning_rate": 9.813705549872862e-06, "loss": 0.5022, "step": 3315 }, { "epoch": 0.22924300034566195, "grad_norm": 0.8058146434331882, "learning_rate": 9.813554146950002e-06, "loss": 0.3366, "step": 3316 }, { "epoch": 0.22931213273418596, "grad_norm": 0.8934134403114544, "learning_rate": 9.81340268369783e-06, "loss": 0.2977, "step": 3317 }, { "epoch": 0.22938126512270998, "grad_norm": 0.8768174177620199, "learning_rate": 9.813251160118247e-06, "loss": 0.4421, "step": 3318 }, { "epoch": 0.229450397511234, "grad_norm": 0.7689289026567555, "learning_rate": 9.81309957621315e-06, "loss": 0.4216, "step": 3319 }, { "epoch": 0.22951952989975805, "grad_norm": 0.7679931215083304, "learning_rate": 9.812947931984439e-06, "loss": 0.3224, "step": 3320 }, { "epoch": 0.22958866228828206, "grad_norm": 0.9620892743490578, "learning_rate": 9.812796227434014e-06, "loss": 0.3363, "step": 3321 }, { "epoch": 0.22965779467680608, "grad_norm": 0.8234851553715198, "learning_rate": 9.812644462563776e-06, "loss": 0.492, "step": 3322 }, { "epoch": 0.2297269270653301, "grad_norm": 0.8791388180948405, "learning_rate": 9.81249263737563e-06, "loss": 0.3402, "step": 3323 }, { "epoch": 0.22979605945385412, "grad_norm": 0.8373005777729039, "learning_rate": 9.812340751871478e-06, "loss": 0.3674, "step": 3324 }, { "epoch": 0.22986519184237816, "grad_norm": 0.8639860246582203, "learning_rate": 9.81218880605322e-06, "loss": 0.4234, "step": 3325 }, { "epoch": 0.22993432423090218, "grad_norm": 0.8240218785214617, "learning_rate": 9.812036799922766e-06, "loss": 0.4283, "step": 3326 }, { "epoch": 0.2300034566194262, "grad_norm": 0.8109856972110095, "learning_rate": 9.811884733482016e-06, "loss": 0.3119, "step": 3327 }, { "epoch": 0.23007258900795022, "grad_norm": 0.7179913626667737, "learning_rate": 9.811732606732879e-06, "loss": 0.3257, "step": 3328 }, { "epoch": 0.23014172139647424, "grad_norm": 0.8838766764622883, "learning_rate": 9.811580419677261e-06, "loss": 0.4885, "step": 3329 }, { "epoch": 0.23021085378499828, "grad_norm": 0.8526422813422433, "learning_rate": 9.811428172317068e-06, "loss": 0.3835, "step": 3330 }, { "epoch": 0.2302799861735223, "grad_norm": 1.0635934516944818, "learning_rate": 9.811275864654211e-06, "loss": 0.3431, "step": 3331 }, { "epoch": 0.23034911856204632, "grad_norm": 0.8911633892321148, "learning_rate": 9.811123496690596e-06, "loss": 0.3264, "step": 3332 }, { "epoch": 0.23041825095057034, "grad_norm": 0.7534176199231989, "learning_rate": 9.810971068428133e-06, "loss": 0.3876, "step": 3333 }, { "epoch": 0.23048738333909435, "grad_norm": 1.3581429300915615, "learning_rate": 9.810818579868732e-06, "loss": 0.3814, "step": 3334 }, { "epoch": 0.2305565157276184, "grad_norm": 0.683804284530479, "learning_rate": 9.810666031014309e-06, "loss": 0.2954, "step": 3335 }, { "epoch": 0.23062564811614242, "grad_norm": 1.1825457389769434, "learning_rate": 9.810513421866771e-06, "loss": 0.4349, "step": 3336 }, { "epoch": 0.23069478050466644, "grad_norm": 0.8697096999773262, "learning_rate": 9.810360752428031e-06, "loss": 0.4066, "step": 3337 }, { "epoch": 0.23076391289319045, "grad_norm": 0.819054409096422, "learning_rate": 9.810208022700005e-06, "loss": 0.3432, "step": 3338 }, { "epoch": 0.23083304528171447, "grad_norm": 0.7802922062619199, "learning_rate": 9.810055232684604e-06, "loss": 0.3019, "step": 3339 }, { "epoch": 0.23090217767023852, "grad_norm": 0.9834405276508835, "learning_rate": 9.809902382383745e-06, "loss": 0.5221, "step": 3340 }, { "epoch": 0.23097131005876254, "grad_norm": 0.8795306251750442, "learning_rate": 9.809749471799343e-06, "loss": 0.3247, "step": 3341 }, { "epoch": 0.23104044244728655, "grad_norm": 0.7096943924195364, "learning_rate": 9.809596500933316e-06, "loss": 0.2728, "step": 3342 }, { "epoch": 0.23110957483581057, "grad_norm": 0.7794435463705086, "learning_rate": 9.80944346978758e-06, "loss": 0.3714, "step": 3343 }, { "epoch": 0.2311787072243346, "grad_norm": 0.8413812470761461, "learning_rate": 9.809290378364053e-06, "loss": 0.4127, "step": 3344 }, { "epoch": 0.23124783961285864, "grad_norm": 0.7819071052216955, "learning_rate": 9.809137226664651e-06, "loss": 0.347, "step": 3345 }, { "epoch": 0.23131697200138265, "grad_norm": 0.8531186763997483, "learning_rate": 9.8089840146913e-06, "loss": 0.3137, "step": 3346 }, { "epoch": 0.23138610438990667, "grad_norm": 0.8520016878541071, "learning_rate": 9.808830742445914e-06, "loss": 0.4323, "step": 3347 }, { "epoch": 0.2314552367784307, "grad_norm": 0.8467186335153023, "learning_rate": 9.808677409930418e-06, "loss": 0.3657, "step": 3348 }, { "epoch": 0.2315243691669547, "grad_norm": 0.7678468466819481, "learning_rate": 9.80852401714673e-06, "loss": 0.3444, "step": 3349 }, { "epoch": 0.23159350155547875, "grad_norm": 0.8225341148366275, "learning_rate": 9.808370564096776e-06, "loss": 0.4321, "step": 3350 }, { "epoch": 0.23166263394400277, "grad_norm": 0.8071699941871888, "learning_rate": 9.808217050782478e-06, "loss": 0.4887, "step": 3351 }, { "epoch": 0.2317317663325268, "grad_norm": 0.8286200236232403, "learning_rate": 9.80806347720576e-06, "loss": 0.3643, "step": 3352 }, { "epoch": 0.2318008987210508, "grad_norm": 0.8242934809629802, "learning_rate": 9.807909843368547e-06, "loss": 0.3613, "step": 3353 }, { "epoch": 0.23187003110957483, "grad_norm": 0.8917468289089883, "learning_rate": 9.807756149272763e-06, "loss": 0.4964, "step": 3354 }, { "epoch": 0.23193916349809887, "grad_norm": 0.9363960716355078, "learning_rate": 9.807602394920334e-06, "loss": 0.3877, "step": 3355 }, { "epoch": 0.2320082958866229, "grad_norm": 1.7166246936753462, "learning_rate": 9.807448580313192e-06, "loss": 0.3189, "step": 3356 }, { "epoch": 0.2320774282751469, "grad_norm": 1.5901893100068543, "learning_rate": 9.807294705453259e-06, "loss": 0.4351, "step": 3357 }, { "epoch": 0.23214656066367093, "grad_norm": 0.7831642392441701, "learning_rate": 9.807140770342466e-06, "loss": 0.4313, "step": 3358 }, { "epoch": 0.23221569305219494, "grad_norm": 0.8177773261834445, "learning_rate": 9.806986774982744e-06, "loss": 0.3383, "step": 3359 }, { "epoch": 0.232284825440719, "grad_norm": 0.8364019478175279, "learning_rate": 9.80683271937602e-06, "loss": 0.2973, "step": 3360 }, { "epoch": 0.232353957829243, "grad_norm": 0.9351308174413643, "learning_rate": 9.806678603524226e-06, "loss": 0.3355, "step": 3361 }, { "epoch": 0.23242309021776703, "grad_norm": 1.0662417340301678, "learning_rate": 9.806524427429294e-06, "loss": 0.3948, "step": 3362 }, { "epoch": 0.23249222260629104, "grad_norm": 0.7983990730362441, "learning_rate": 9.806370191093155e-06, "loss": 0.287, "step": 3363 }, { "epoch": 0.23256135499481506, "grad_norm": 0.8967836483997559, "learning_rate": 9.806215894517743e-06, "loss": 0.3323, "step": 3364 }, { "epoch": 0.2326304873833391, "grad_norm": 0.9813742199900336, "learning_rate": 9.806061537704995e-06, "loss": 0.4065, "step": 3365 }, { "epoch": 0.23269961977186313, "grad_norm": 0.7968271425987478, "learning_rate": 9.80590712065684e-06, "loss": 0.3293, "step": 3366 }, { "epoch": 0.23276875216038714, "grad_norm": 0.7761613647708077, "learning_rate": 9.805752643375214e-06, "loss": 0.3487, "step": 3367 }, { "epoch": 0.23283788454891116, "grad_norm": 0.8284757472469583, "learning_rate": 9.805598105862057e-06, "loss": 0.3844, "step": 3368 }, { "epoch": 0.23290701693743518, "grad_norm": 0.8955898182595449, "learning_rate": 9.805443508119304e-06, "loss": 0.4225, "step": 3369 }, { "epoch": 0.23297614932595923, "grad_norm": 0.6913753942779365, "learning_rate": 9.805288850148891e-06, "loss": 0.3089, "step": 3370 }, { "epoch": 0.23304528171448324, "grad_norm": 0.9629621491915471, "learning_rate": 9.805134131952758e-06, "loss": 0.4176, "step": 3371 }, { "epoch": 0.23311441410300726, "grad_norm": 0.7762568436856806, "learning_rate": 9.804979353532844e-06, "loss": 0.3907, "step": 3372 }, { "epoch": 0.23318354649153128, "grad_norm": 0.7182923399363181, "learning_rate": 9.80482451489109e-06, "loss": 0.3256, "step": 3373 }, { "epoch": 0.2332526788800553, "grad_norm": 0.7569469159994968, "learning_rate": 9.804669616029432e-06, "loss": 0.329, "step": 3374 }, { "epoch": 0.23332181126857934, "grad_norm": 0.7622540297192244, "learning_rate": 9.804514656949817e-06, "loss": 0.3279, "step": 3375 }, { "epoch": 0.23339094365710336, "grad_norm": 0.7791103591258288, "learning_rate": 9.804359637654185e-06, "loss": 0.4654, "step": 3376 }, { "epoch": 0.23346007604562738, "grad_norm": 0.792398624489333, "learning_rate": 9.804204558144479e-06, "loss": 0.335, "step": 3377 }, { "epoch": 0.2335292084341514, "grad_norm": 0.9973971341561719, "learning_rate": 9.804049418422641e-06, "loss": 0.3262, "step": 3378 }, { "epoch": 0.23359834082267542, "grad_norm": 1.0739475654818733, "learning_rate": 9.803894218490617e-06, "loss": 0.438, "step": 3379 }, { "epoch": 0.23366747321119943, "grad_norm": 0.7960606900026033, "learning_rate": 9.803738958350352e-06, "loss": 0.4018, "step": 3380 }, { "epoch": 0.23373660559972348, "grad_norm": 0.9129325462857063, "learning_rate": 9.803583638003794e-06, "loss": 0.3602, "step": 3381 }, { "epoch": 0.2338057379882475, "grad_norm": 0.7305407867635481, "learning_rate": 9.803428257452884e-06, "loss": 0.3125, "step": 3382 }, { "epoch": 0.23387487037677152, "grad_norm": 0.82216304552278, "learning_rate": 9.803272816699574e-06, "loss": 0.4658, "step": 3383 }, { "epoch": 0.23394400276529553, "grad_norm": 0.8403069470505202, "learning_rate": 9.803117315745811e-06, "loss": 0.3889, "step": 3384 }, { "epoch": 0.23401313515381955, "grad_norm": 0.8236505593779867, "learning_rate": 9.802961754593546e-06, "loss": 0.3716, "step": 3385 }, { "epoch": 0.2340822675423436, "grad_norm": 0.9043912114188097, "learning_rate": 9.802806133244724e-06, "loss": 0.4369, "step": 3386 }, { "epoch": 0.23415139993086762, "grad_norm": 0.7448636462997088, "learning_rate": 9.8026504517013e-06, "loss": 0.4039, "step": 3387 }, { "epoch": 0.23422053231939163, "grad_norm": 0.7557898633719556, "learning_rate": 9.802494709965222e-06, "loss": 0.3222, "step": 3388 }, { "epoch": 0.23428966470791565, "grad_norm": 0.8629377415430161, "learning_rate": 9.802338908038445e-06, "loss": 0.3117, "step": 3389 }, { "epoch": 0.23435879709643967, "grad_norm": 0.7935961997357261, "learning_rate": 9.802183045922919e-06, "loss": 0.3975, "step": 3390 }, { "epoch": 0.23442792948496372, "grad_norm": 0.8246728526182349, "learning_rate": 9.802027123620597e-06, "loss": 0.3664, "step": 3391 }, { "epoch": 0.23449706187348773, "grad_norm": 0.9099088545063606, "learning_rate": 9.801871141133437e-06, "loss": 0.2974, "step": 3392 }, { "epoch": 0.23456619426201175, "grad_norm": 0.8565862624520613, "learning_rate": 9.80171509846339e-06, "loss": 0.3825, "step": 3393 }, { "epoch": 0.23463532665053577, "grad_norm": 0.8176947247634503, "learning_rate": 9.801558995612414e-06, "loss": 0.3804, "step": 3394 }, { "epoch": 0.2347044590390598, "grad_norm": 0.7801436061643554, "learning_rate": 9.801402832582465e-06, "loss": 0.3361, "step": 3395 }, { "epoch": 0.23477359142758383, "grad_norm": 0.8499871692976039, "learning_rate": 9.801246609375498e-06, "loss": 0.3403, "step": 3396 }, { "epoch": 0.23484272381610785, "grad_norm": 0.8609380352429394, "learning_rate": 9.801090325993474e-06, "loss": 0.4067, "step": 3397 }, { "epoch": 0.23491185620463187, "grad_norm": 0.746859320856658, "learning_rate": 9.800933982438352e-06, "loss": 0.3584, "step": 3398 }, { "epoch": 0.2349809885931559, "grad_norm": 0.9907261869998129, "learning_rate": 9.800777578712088e-06, "loss": 0.3371, "step": 3399 }, { "epoch": 0.2350501209816799, "grad_norm": 0.869134025442274, "learning_rate": 9.800621114816646e-06, "loss": 0.3823, "step": 3400 }, { "epoch": 0.23511925337020395, "grad_norm": 0.7595701755693859, "learning_rate": 9.800464590753986e-06, "loss": 0.4346, "step": 3401 }, { "epoch": 0.23518838575872797, "grad_norm": 1.0281930730340918, "learning_rate": 9.800308006526069e-06, "loss": 0.3357, "step": 3402 }, { "epoch": 0.235257518147252, "grad_norm": 0.7399112972967254, "learning_rate": 9.800151362134855e-06, "loss": 0.3053, "step": 3403 }, { "epoch": 0.235326650535776, "grad_norm": 0.7974279491491547, "learning_rate": 9.799994657582312e-06, "loss": 0.4011, "step": 3404 }, { "epoch": 0.23539578292430002, "grad_norm": 0.7898034473579347, "learning_rate": 9.799837892870403e-06, "loss": 0.3526, "step": 3405 }, { "epoch": 0.23546491531282407, "grad_norm": 0.8884504973781256, "learning_rate": 9.799681068001091e-06, "loss": 0.3219, "step": 3406 }, { "epoch": 0.2355340477013481, "grad_norm": 0.8022117101053599, "learning_rate": 9.79952418297634e-06, "loss": 0.4034, "step": 3407 }, { "epoch": 0.2356031800898721, "grad_norm": 0.8474373166626913, "learning_rate": 9.799367237798121e-06, "loss": 0.4741, "step": 3408 }, { "epoch": 0.23567231247839612, "grad_norm": 0.7912578085679769, "learning_rate": 9.799210232468397e-06, "loss": 0.3004, "step": 3409 }, { "epoch": 0.23574144486692014, "grad_norm": 0.7487545294238318, "learning_rate": 9.79905316698914e-06, "loss": 0.3073, "step": 3410 }, { "epoch": 0.2358105772554442, "grad_norm": 0.753313989337246, "learning_rate": 9.798896041362314e-06, "loss": 0.4284, "step": 3411 }, { "epoch": 0.2358797096439682, "grad_norm": 0.7491473665656924, "learning_rate": 9.798738855589888e-06, "loss": 0.4019, "step": 3412 }, { "epoch": 0.23594884203249222, "grad_norm": 0.8882713290555811, "learning_rate": 9.798581609673837e-06, "loss": 0.3289, "step": 3413 }, { "epoch": 0.23601797442101624, "grad_norm": 0.7893319694124089, "learning_rate": 9.798424303616129e-06, "loss": 0.2964, "step": 3414 }, { "epoch": 0.23608710680954026, "grad_norm": 0.7942675205899351, "learning_rate": 9.798266937418735e-06, "loss": 0.426, "step": 3415 }, { "epoch": 0.2361562391980643, "grad_norm": 0.7930340782102984, "learning_rate": 9.798109511083627e-06, "loss": 0.3861, "step": 3416 }, { "epoch": 0.23622537158658832, "grad_norm": 0.8553988421478123, "learning_rate": 9.79795202461278e-06, "loss": 0.3242, "step": 3417 }, { "epoch": 0.23629450397511234, "grad_norm": 0.8449914044121953, "learning_rate": 9.797794478008166e-06, "loss": 0.385, "step": 3418 }, { "epoch": 0.23636363636363636, "grad_norm": 0.7998247912045369, "learning_rate": 9.79763687127176e-06, "loss": 0.4262, "step": 3419 }, { "epoch": 0.23643276875216038, "grad_norm": 0.9087624496898796, "learning_rate": 9.797479204405538e-06, "loss": 0.3863, "step": 3420 }, { "epoch": 0.23650190114068442, "grad_norm": 0.8491215770822428, "learning_rate": 9.797321477411475e-06, "loss": 0.4137, "step": 3421 }, { "epoch": 0.23657103352920844, "grad_norm": 1.2035120052849686, "learning_rate": 9.79716369029155e-06, "loss": 0.4645, "step": 3422 }, { "epoch": 0.23664016591773246, "grad_norm": 0.7846985511799491, "learning_rate": 9.797005843047738e-06, "loss": 0.3474, "step": 3423 }, { "epoch": 0.23670929830625648, "grad_norm": 0.734332653917169, "learning_rate": 9.796847935682019e-06, "loss": 0.3799, "step": 3424 }, { "epoch": 0.2367784306947805, "grad_norm": 0.8932647799534258, "learning_rate": 9.79668996819637e-06, "loss": 0.4209, "step": 3425 }, { "epoch": 0.23684756308330454, "grad_norm": 0.8146636150832347, "learning_rate": 9.796531940592774e-06, "loss": 0.4015, "step": 3426 }, { "epoch": 0.23691669547182856, "grad_norm": 0.917849525353404, "learning_rate": 9.79637385287321e-06, "loss": 0.3446, "step": 3427 }, { "epoch": 0.23698582786035258, "grad_norm": 0.7825900986838613, "learning_rate": 9.796215705039658e-06, "loss": 0.3744, "step": 3428 }, { "epoch": 0.2370549602488766, "grad_norm": 0.760461347473448, "learning_rate": 9.796057497094102e-06, "loss": 0.3938, "step": 3429 }, { "epoch": 0.2371240926374006, "grad_norm": 0.7815325146900142, "learning_rate": 9.795899229038525e-06, "loss": 0.417, "step": 3430 }, { "epoch": 0.23719322502592466, "grad_norm": 0.8038427283296735, "learning_rate": 9.795740900874908e-06, "loss": 0.3166, "step": 3431 }, { "epoch": 0.23726235741444868, "grad_norm": 0.816404089070647, "learning_rate": 9.79558251260524e-06, "loss": 0.3685, "step": 3432 }, { "epoch": 0.2373314898029727, "grad_norm": 0.7715469246087369, "learning_rate": 9.7954240642315e-06, "loss": 0.4237, "step": 3433 }, { "epoch": 0.2374006221914967, "grad_norm": 0.7353184118694743, "learning_rate": 9.79526555575568e-06, "loss": 0.3191, "step": 3434 }, { "epoch": 0.23746975458002073, "grad_norm": 0.7479562213750796, "learning_rate": 9.795106987179761e-06, "loss": 0.3328, "step": 3435 }, { "epoch": 0.23753888696854478, "grad_norm": 0.8810886203439261, "learning_rate": 9.794948358505736e-06, "loss": 0.3481, "step": 3436 }, { "epoch": 0.2376080193570688, "grad_norm": 0.9881341499764535, "learning_rate": 9.794789669735588e-06, "loss": 0.3475, "step": 3437 }, { "epoch": 0.2376771517455928, "grad_norm": 0.7772141850705393, "learning_rate": 9.79463092087131e-06, "loss": 0.2559, "step": 3438 }, { "epoch": 0.23774628413411683, "grad_norm": 0.8096796647849142, "learning_rate": 9.794472111914888e-06, "loss": 0.338, "step": 3439 }, { "epoch": 0.23781541652264085, "grad_norm": 0.7816401530528471, "learning_rate": 9.794313242868316e-06, "loss": 0.4071, "step": 3440 }, { "epoch": 0.23788454891116487, "grad_norm": 0.9539863961401251, "learning_rate": 9.794154313733581e-06, "loss": 0.3453, "step": 3441 }, { "epoch": 0.2379536812996889, "grad_norm": 0.7824998994091829, "learning_rate": 9.793995324512679e-06, "loss": 0.3328, "step": 3442 }, { "epoch": 0.23802281368821293, "grad_norm": 0.8885583747276994, "learning_rate": 9.793836275207602e-06, "loss": 0.3435, "step": 3443 }, { "epoch": 0.23809194607673695, "grad_norm": 0.840637194896619, "learning_rate": 9.793677165820341e-06, "loss": 0.4557, "step": 3444 }, { "epoch": 0.23816107846526097, "grad_norm": 0.7644637073842656, "learning_rate": 9.79351799635289e-06, "loss": 0.324, "step": 3445 }, { "epoch": 0.23823021085378498, "grad_norm": 0.7265127404860832, "learning_rate": 9.793358766807247e-06, "loss": 0.3256, "step": 3446 }, { "epoch": 0.23829934324230903, "grad_norm": 0.8155057966652256, "learning_rate": 9.793199477185406e-06, "loss": 0.417, "step": 3447 }, { "epoch": 0.23836847563083305, "grad_norm": 0.929396303963109, "learning_rate": 9.793040127489364e-06, "loss": 0.3482, "step": 3448 }, { "epoch": 0.23843760801935707, "grad_norm": 0.8827882021778024, "learning_rate": 9.792880717721116e-06, "loss": 0.3093, "step": 3449 }, { "epoch": 0.23850674040788108, "grad_norm": 0.8639834267226475, "learning_rate": 9.792721247882662e-06, "loss": 0.3855, "step": 3450 }, { "epoch": 0.2385758727964051, "grad_norm": 0.7438499077965235, "learning_rate": 9.792561717976e-06, "loss": 0.4491, "step": 3451 }, { "epoch": 0.23864500518492915, "grad_norm": 0.9195717293393643, "learning_rate": 9.79240212800313e-06, "loss": 0.4092, "step": 3452 }, { "epoch": 0.23871413757345317, "grad_norm": 0.8207948663775126, "learning_rate": 9.792242477966051e-06, "loss": 0.3577, "step": 3453 }, { "epoch": 0.23878326996197718, "grad_norm": 0.8246918295018867, "learning_rate": 9.792082767866766e-06, "loss": 0.3436, "step": 3454 }, { "epoch": 0.2388524023505012, "grad_norm": 0.8243132151464603, "learning_rate": 9.791922997707275e-06, "loss": 0.4007, "step": 3455 }, { "epoch": 0.23892153473902522, "grad_norm": 0.8617797077654592, "learning_rate": 9.79176316748958e-06, "loss": 0.3716, "step": 3456 }, { "epoch": 0.23899066712754927, "grad_norm": 0.8698478616256742, "learning_rate": 9.791603277215684e-06, "loss": 0.3656, "step": 3457 }, { "epoch": 0.23905979951607328, "grad_norm": 0.8775449825607567, "learning_rate": 9.791443326887594e-06, "loss": 0.4615, "step": 3458 }, { "epoch": 0.2391289319045973, "grad_norm": 0.8151628816469901, "learning_rate": 9.791283316507312e-06, "loss": 0.3892, "step": 3459 }, { "epoch": 0.23919806429312132, "grad_norm": 0.8087859327477817, "learning_rate": 9.791123246076844e-06, "loss": 0.3283, "step": 3460 }, { "epoch": 0.23926719668164534, "grad_norm": 0.7772521892100567, "learning_rate": 9.790963115598197e-06, "loss": 0.3705, "step": 3461 }, { "epoch": 0.23933632907016938, "grad_norm": 0.7766696371693663, "learning_rate": 9.790802925073376e-06, "loss": 0.4019, "step": 3462 }, { "epoch": 0.2394054614586934, "grad_norm": 0.7687895807658607, "learning_rate": 9.79064267450439e-06, "loss": 0.3287, "step": 3463 }, { "epoch": 0.23947459384721742, "grad_norm": 0.8494897392499502, "learning_rate": 9.790482363893248e-06, "loss": 0.3866, "step": 3464 }, { "epoch": 0.23954372623574144, "grad_norm": 1.034844406533194, "learning_rate": 9.790321993241958e-06, "loss": 0.4599, "step": 3465 }, { "epoch": 0.23961285862426546, "grad_norm": 0.8027509760458834, "learning_rate": 9.790161562552532e-06, "loss": 0.3509, "step": 3466 }, { "epoch": 0.2396819910127895, "grad_norm": 0.7968152910073644, "learning_rate": 9.790001071826978e-06, "loss": 0.3108, "step": 3467 }, { "epoch": 0.23975112340131352, "grad_norm": 1.2217139717819852, "learning_rate": 9.789840521067307e-06, "loss": 0.3567, "step": 3468 }, { "epoch": 0.23982025578983754, "grad_norm": 0.797588299170629, "learning_rate": 9.789679910275535e-06, "loss": 0.4146, "step": 3469 }, { "epoch": 0.23988938817836156, "grad_norm": 0.7319342803705703, "learning_rate": 9.789519239453673e-06, "loss": 0.3378, "step": 3470 }, { "epoch": 0.23995852056688557, "grad_norm": 0.792460644198425, "learning_rate": 9.789358508603735e-06, "loss": 0.3294, "step": 3471 }, { "epoch": 0.24002765295540962, "grad_norm": 0.8153908380827092, "learning_rate": 9.789197717727735e-06, "loss": 0.4714, "step": 3472 }, { "epoch": 0.24009678534393364, "grad_norm": 0.8285576704016318, "learning_rate": 9.789036866827686e-06, "loss": 0.3773, "step": 3473 }, { "epoch": 0.24016591773245766, "grad_norm": 0.7855295818740609, "learning_rate": 9.788875955905608e-06, "loss": 0.3984, "step": 3474 }, { "epoch": 0.24023505012098167, "grad_norm": 0.7482366312980129, "learning_rate": 9.788714984963516e-06, "loss": 0.3707, "step": 3475 }, { "epoch": 0.2403041825095057, "grad_norm": 0.8668664864592838, "learning_rate": 9.788553954003428e-06, "loss": 0.4409, "step": 3476 }, { "epoch": 0.24037331489802974, "grad_norm": 1.3695131468591177, "learning_rate": 9.788392863027361e-06, "loss": 0.2724, "step": 3477 }, { "epoch": 0.24044244728655376, "grad_norm": 0.7987366490101911, "learning_rate": 9.788231712037334e-06, "loss": 0.3501, "step": 3478 }, { "epoch": 0.24051157967507777, "grad_norm": 0.9006872718392736, "learning_rate": 9.788070501035369e-06, "loss": 0.4833, "step": 3479 }, { "epoch": 0.2405807120636018, "grad_norm": 0.837777878203119, "learning_rate": 9.787909230023485e-06, "loss": 0.3556, "step": 3480 }, { "epoch": 0.2406498444521258, "grad_norm": 0.7748173959048018, "learning_rate": 9.7877478990037e-06, "loss": 0.3725, "step": 3481 }, { "epoch": 0.24071897684064986, "grad_norm": 0.7832658520132929, "learning_rate": 9.787586507978043e-06, "loss": 0.356, "step": 3482 }, { "epoch": 0.24078810922917387, "grad_norm": 0.849077033397378, "learning_rate": 9.787425056948531e-06, "loss": 0.4434, "step": 3483 }, { "epoch": 0.2408572416176979, "grad_norm": 0.8673212234759412, "learning_rate": 9.787263545917192e-06, "loss": 0.3287, "step": 3484 }, { "epoch": 0.2409263740062219, "grad_norm": 0.7452512862701361, "learning_rate": 9.787101974886044e-06, "loss": 0.3448, "step": 3485 }, { "epoch": 0.24099550639474593, "grad_norm": 0.843166218480835, "learning_rate": 9.786940343857118e-06, "loss": 0.4067, "step": 3486 }, { "epoch": 0.24106463878326997, "grad_norm": 0.984613990588441, "learning_rate": 9.786778652832439e-06, "loss": 0.4194, "step": 3487 }, { "epoch": 0.241133771171794, "grad_norm": 0.8572648595944461, "learning_rate": 9.786616901814028e-06, "loss": 0.3829, "step": 3488 }, { "epoch": 0.241202903560318, "grad_norm": 0.8205306773407774, "learning_rate": 9.78645509080392e-06, "loss": 0.3831, "step": 3489 }, { "epoch": 0.24127203594884203, "grad_norm": 0.8893485670221524, "learning_rate": 9.786293219804136e-06, "loss": 0.4844, "step": 3490 }, { "epoch": 0.24134116833736605, "grad_norm": 0.8340711345910957, "learning_rate": 9.78613128881671e-06, "loss": 0.3129, "step": 3491 }, { "epoch": 0.2414103007258901, "grad_norm": 0.9268480271023383, "learning_rate": 9.78596929784367e-06, "loss": 0.3238, "step": 3492 }, { "epoch": 0.2414794331144141, "grad_norm": 0.8416510393820738, "learning_rate": 9.785807246887044e-06, "loss": 0.3745, "step": 3493 }, { "epoch": 0.24154856550293813, "grad_norm": 0.7501466301322205, "learning_rate": 9.785645135948864e-06, "loss": 0.3726, "step": 3494 }, { "epoch": 0.24161769789146215, "grad_norm": 0.8797923876894732, "learning_rate": 9.785482965031166e-06, "loss": 0.2933, "step": 3495 }, { "epoch": 0.24168683027998616, "grad_norm": 0.9555676461518887, "learning_rate": 9.785320734135977e-06, "loss": 0.3424, "step": 3496 }, { "epoch": 0.2417559626685102, "grad_norm": 0.7984471339990551, "learning_rate": 9.785158443265332e-06, "loss": 0.4647, "step": 3497 }, { "epoch": 0.24182509505703423, "grad_norm": 0.8723878462202762, "learning_rate": 9.784996092421267e-06, "loss": 0.3867, "step": 3498 }, { "epoch": 0.24189422744555825, "grad_norm": 1.1395477232968179, "learning_rate": 9.784833681605814e-06, "loss": 0.2828, "step": 3499 }, { "epoch": 0.24196335983408226, "grad_norm": 0.7996438546585692, "learning_rate": 9.78467121082101e-06, "loss": 0.3855, "step": 3500 }, { "epoch": 0.24203249222260628, "grad_norm": 0.7442031861054834, "learning_rate": 9.784508680068892e-06, "loss": 0.4519, "step": 3501 }, { "epoch": 0.24210162461113033, "grad_norm": 0.7600544186818263, "learning_rate": 9.784346089351494e-06, "loss": 0.3483, "step": 3502 }, { "epoch": 0.24217075699965435, "grad_norm": 1.0913537853166166, "learning_rate": 9.784183438670857e-06, "loss": 0.3352, "step": 3503 }, { "epoch": 0.24223988938817836, "grad_norm": 0.8606440598682518, "learning_rate": 9.784020728029017e-06, "loss": 0.3821, "step": 3504 }, { "epoch": 0.24230902177670238, "grad_norm": 0.7304653773283379, "learning_rate": 9.783857957428017e-06, "loss": 0.3744, "step": 3505 }, { "epoch": 0.2423781541652264, "grad_norm": 0.8047585275945648, "learning_rate": 9.783695126869895e-06, "loss": 0.3395, "step": 3506 }, { "epoch": 0.24244728655375042, "grad_norm": 0.8193409340690995, "learning_rate": 9.78353223635669e-06, "loss": 0.3751, "step": 3507 }, { "epoch": 0.24251641894227446, "grad_norm": 1.0742508257162153, "learning_rate": 9.783369285890446e-06, "loss": 0.4386, "step": 3508 }, { "epoch": 0.24258555133079848, "grad_norm": 0.7516444701748485, "learning_rate": 9.783206275473202e-06, "loss": 0.3666, "step": 3509 }, { "epoch": 0.2426546837193225, "grad_norm": 0.757086376746113, "learning_rate": 9.783043205107004e-06, "loss": 0.3292, "step": 3510 }, { "epoch": 0.24272381610784652, "grad_norm": 0.8375118988953135, "learning_rate": 9.782880074793896e-06, "loss": 0.477, "step": 3511 }, { "epoch": 0.24279294849637054, "grad_norm": 0.8013689780423722, "learning_rate": 9.782716884535921e-06, "loss": 0.4314, "step": 3512 }, { "epoch": 0.24286208088489458, "grad_norm": 0.9077705458408202, "learning_rate": 9.782553634335126e-06, "loss": 0.3104, "step": 3513 }, { "epoch": 0.2429312132734186, "grad_norm": 0.9325378684697365, "learning_rate": 9.782390324193555e-06, "loss": 0.435, "step": 3514 }, { "epoch": 0.24300034566194262, "grad_norm": 0.9743733711431116, "learning_rate": 9.782226954113257e-06, "loss": 0.424, "step": 3515 }, { "epoch": 0.24306947805046664, "grad_norm": 0.9120321495563607, "learning_rate": 9.782063524096279e-06, "loss": 0.3215, "step": 3516 }, { "epoch": 0.24313861043899065, "grad_norm": 0.7732682031767294, "learning_rate": 9.781900034144668e-06, "loss": 0.3014, "step": 3517 }, { "epoch": 0.2432077428275147, "grad_norm": 0.9633210304833596, "learning_rate": 9.781736484260472e-06, "loss": 0.4383, "step": 3518 }, { "epoch": 0.24327687521603872, "grad_norm": 1.6457991650446744, "learning_rate": 9.781572874445744e-06, "loss": 0.4127, "step": 3519 }, { "epoch": 0.24334600760456274, "grad_norm": 0.7391150293048335, "learning_rate": 9.781409204702533e-06, "loss": 0.3098, "step": 3520 }, { "epoch": 0.24341513999308675, "grad_norm": 1.0661423618826635, "learning_rate": 9.78124547503289e-06, "loss": 0.2986, "step": 3521 }, { "epoch": 0.24348427238161077, "grad_norm": 0.9225260791529338, "learning_rate": 9.78108168543887e-06, "loss": 0.4816, "step": 3522 }, { "epoch": 0.24355340477013482, "grad_norm": 1.2443724717978748, "learning_rate": 9.780917835922519e-06, "loss": 0.3827, "step": 3523 }, { "epoch": 0.24362253715865884, "grad_norm": 0.7536021693635988, "learning_rate": 9.780753926485897e-06, "loss": 0.3078, "step": 3524 }, { "epoch": 0.24369166954718285, "grad_norm": 0.8107005257878177, "learning_rate": 9.780589957131056e-06, "loss": 0.3809, "step": 3525 }, { "epoch": 0.24376080193570687, "grad_norm": 0.8075650093445468, "learning_rate": 9.780425927860054e-06, "loss": 0.4086, "step": 3526 }, { "epoch": 0.2438299343242309, "grad_norm": 0.7381139673057348, "learning_rate": 9.78026183867494e-06, "loss": 0.3247, "step": 3527 }, { "epoch": 0.24389906671275494, "grad_norm": 0.7882776315405761, "learning_rate": 9.780097689577777e-06, "loss": 0.3259, "step": 3528 }, { "epoch": 0.24396819910127895, "grad_norm": 0.8462648857422338, "learning_rate": 9.779933480570619e-06, "loss": 0.4416, "step": 3529 }, { "epoch": 0.24403733148980297, "grad_norm": 1.0895589480970982, "learning_rate": 9.779769211655525e-06, "loss": 0.4163, "step": 3530 }, { "epoch": 0.244106463878327, "grad_norm": 0.762357235381129, "learning_rate": 9.779604882834553e-06, "loss": 0.2856, "step": 3531 }, { "epoch": 0.244175596266851, "grad_norm": 1.2353041226430836, "learning_rate": 9.779440494109764e-06, "loss": 0.3788, "step": 3532 }, { "epoch": 0.24424472865537505, "grad_norm": 0.9788140016209076, "learning_rate": 9.779276045483219e-06, "loss": 0.4503, "step": 3533 }, { "epoch": 0.24431386104389907, "grad_norm": 1.0203782036732834, "learning_rate": 9.779111536956975e-06, "loss": 0.3781, "step": 3534 }, { "epoch": 0.2443829934324231, "grad_norm": 0.7616661804781425, "learning_rate": 9.7789469685331e-06, "loss": 0.2789, "step": 3535 }, { "epoch": 0.2444521258209471, "grad_norm": 1.0355730044536917, "learning_rate": 9.77878234021365e-06, "loss": 0.4295, "step": 3536 }, { "epoch": 0.24452125820947113, "grad_norm": 0.85724389352106, "learning_rate": 9.778617652000694e-06, "loss": 0.4329, "step": 3537 }, { "epoch": 0.24459039059799517, "grad_norm": 1.0782203805921744, "learning_rate": 9.778452903896292e-06, "loss": 0.3995, "step": 3538 }, { "epoch": 0.2446595229865192, "grad_norm": 0.8093954618421438, "learning_rate": 9.77828809590251e-06, "loss": 0.4345, "step": 3539 }, { "epoch": 0.2447286553750432, "grad_norm": 0.8411336956671064, "learning_rate": 9.778123228021417e-06, "loss": 0.4858, "step": 3540 }, { "epoch": 0.24479778776356723, "grad_norm": 0.732300883265372, "learning_rate": 9.777958300255072e-06, "loss": 0.3893, "step": 3541 }, { "epoch": 0.24486692015209124, "grad_norm": 0.8514277851538455, "learning_rate": 9.77779331260555e-06, "loss": 0.2949, "step": 3542 }, { "epoch": 0.2449360525406153, "grad_norm": 0.749656401203541, "learning_rate": 9.777628265074913e-06, "loss": 0.3535, "step": 3543 }, { "epoch": 0.2450051849291393, "grad_norm": 0.7977807533803086, "learning_rate": 9.777463157665232e-06, "loss": 0.3796, "step": 3544 }, { "epoch": 0.24507431731766333, "grad_norm": 0.7770868637447773, "learning_rate": 9.777297990378578e-06, "loss": 0.4285, "step": 3545 }, { "epoch": 0.24514344970618734, "grad_norm": 0.8721078903812757, "learning_rate": 9.777132763217017e-06, "loss": 0.3564, "step": 3546 }, { "epoch": 0.24521258209471136, "grad_norm": 0.7924172609619057, "learning_rate": 9.776967476182624e-06, "loss": 0.4006, "step": 3547 }, { "epoch": 0.2452817144832354, "grad_norm": 0.7770845654831046, "learning_rate": 9.776802129277467e-06, "loss": 0.4043, "step": 3548 }, { "epoch": 0.24535084687175943, "grad_norm": 0.7910201401705885, "learning_rate": 9.77663672250362e-06, "loss": 0.2944, "step": 3549 }, { "epoch": 0.24541997926028344, "grad_norm": 1.0761418473189859, "learning_rate": 9.776471255863155e-06, "loss": 0.3622, "step": 3550 }, { "epoch": 0.24548911164880746, "grad_norm": 0.8491423700367172, "learning_rate": 9.776305729358148e-06, "loss": 0.464, "step": 3551 }, { "epoch": 0.24555824403733148, "grad_norm": 0.7386041513636168, "learning_rate": 9.776140142990673e-06, "loss": 0.3769, "step": 3552 }, { "epoch": 0.24562737642585553, "grad_norm": 0.8260585515339367, "learning_rate": 9.775974496762804e-06, "loss": 0.344, "step": 3553 }, { "epoch": 0.24569650881437954, "grad_norm": 0.9782243454559455, "learning_rate": 9.775808790676617e-06, "loss": 0.3503, "step": 3554 }, { "epoch": 0.24576564120290356, "grad_norm": 0.8301462370061576, "learning_rate": 9.775643024734191e-06, "loss": 0.422, "step": 3555 }, { "epoch": 0.24583477359142758, "grad_norm": 0.8237262637158457, "learning_rate": 9.775477198937602e-06, "loss": 0.2965, "step": 3556 }, { "epoch": 0.2459039059799516, "grad_norm": 0.8063108700416537, "learning_rate": 9.775311313288928e-06, "loss": 0.4359, "step": 3557 }, { "epoch": 0.24597303836847564, "grad_norm": 0.7335265100384226, "learning_rate": 9.77514536779025e-06, "loss": 0.4143, "step": 3558 }, { "epoch": 0.24604217075699966, "grad_norm": 0.8592869546080422, "learning_rate": 9.774979362443644e-06, "loss": 0.3527, "step": 3559 }, { "epoch": 0.24611130314552368, "grad_norm": 0.7263802927733607, "learning_rate": 9.774813297251194e-06, "loss": 0.3118, "step": 3560 }, { "epoch": 0.2461804355340477, "grad_norm": 0.8945292713358112, "learning_rate": 9.774647172214981e-06, "loss": 0.421, "step": 3561 }, { "epoch": 0.24624956792257172, "grad_norm": 0.7737298143515889, "learning_rate": 9.774480987337085e-06, "loss": 0.4129, "step": 3562 }, { "epoch": 0.24631870031109576, "grad_norm": 0.7160471085364656, "learning_rate": 9.774314742619591e-06, "loss": 0.3168, "step": 3563 }, { "epoch": 0.24638783269961978, "grad_norm": 1.013890869015541, "learning_rate": 9.774148438064583e-06, "loss": 0.3888, "step": 3564 }, { "epoch": 0.2464569650881438, "grad_norm": 0.8007520051615323, "learning_rate": 9.773982073674144e-06, "loss": 0.4505, "step": 3565 }, { "epoch": 0.24652609747666782, "grad_norm": 0.771031311559987, "learning_rate": 9.773815649450358e-06, "loss": 0.3782, "step": 3566 }, { "epoch": 0.24659522986519183, "grad_norm": 0.7788412899214221, "learning_rate": 9.773649165395312e-06, "loss": 0.31, "step": 3567 }, { "epoch": 0.24666436225371588, "grad_norm": 0.8309637647891872, "learning_rate": 9.773482621511094e-06, "loss": 0.3214, "step": 3568 }, { "epoch": 0.2467334946422399, "grad_norm": 0.7881607947607956, "learning_rate": 9.77331601779979e-06, "loss": 0.4849, "step": 3569 }, { "epoch": 0.24680262703076392, "grad_norm": 0.7036476975101301, "learning_rate": 9.773149354263487e-06, "loss": 0.295, "step": 3570 }, { "epoch": 0.24687175941928793, "grad_norm": 0.7456919818786462, "learning_rate": 9.772982630904275e-06, "loss": 0.385, "step": 3571 }, { "epoch": 0.24694089180781195, "grad_norm": 0.7874367044515971, "learning_rate": 9.772815847724245e-06, "loss": 0.463, "step": 3572 }, { "epoch": 0.24701002419633597, "grad_norm": 0.9417830339217524, "learning_rate": 9.772649004725483e-06, "loss": 0.3793, "step": 3573 }, { "epoch": 0.24707915658486002, "grad_norm": 0.8535173132082624, "learning_rate": 9.772482101910085e-06, "loss": 0.3505, "step": 3574 }, { "epoch": 0.24714828897338403, "grad_norm": 0.6892094092070797, "learning_rate": 9.77231513928014e-06, "loss": 0.3021, "step": 3575 }, { "epoch": 0.24721742136190805, "grad_norm": 0.7867728356921004, "learning_rate": 9.772148116837741e-06, "loss": 0.4684, "step": 3576 }, { "epoch": 0.24728655375043207, "grad_norm": 0.820115596886023, "learning_rate": 9.771981034584982e-06, "loss": 0.3663, "step": 3577 }, { "epoch": 0.2473556861389561, "grad_norm": 0.7675168692876989, "learning_rate": 9.771813892523959e-06, "loss": 0.3555, "step": 3578 }, { "epoch": 0.24742481852748013, "grad_norm": 0.7947175819763963, "learning_rate": 9.771646690656763e-06, "loss": 0.379, "step": 3579 }, { "epoch": 0.24749395091600415, "grad_norm": 0.8144680944021347, "learning_rate": 9.771479428985488e-06, "loss": 0.354, "step": 3580 }, { "epoch": 0.24756308330452817, "grad_norm": 0.7767988631939149, "learning_rate": 9.771312107512237e-06, "loss": 0.3017, "step": 3581 }, { "epoch": 0.2476322156930522, "grad_norm": 0.8244788017602049, "learning_rate": 9.771144726239104e-06, "loss": 0.3478, "step": 3582 }, { "epoch": 0.2477013480815762, "grad_norm": 0.8120316619553577, "learning_rate": 9.770977285168184e-06, "loss": 0.466, "step": 3583 }, { "epoch": 0.24777048047010025, "grad_norm": 1.0314775316818203, "learning_rate": 9.770809784301578e-06, "loss": 0.3078, "step": 3584 }, { "epoch": 0.24783961285862427, "grad_norm": 1.2422418473945787, "learning_rate": 9.770642223641387e-06, "loss": 0.3899, "step": 3585 }, { "epoch": 0.2479087452471483, "grad_norm": 0.929199715501915, "learning_rate": 9.770474603189708e-06, "loss": 0.4268, "step": 3586 }, { "epoch": 0.2479778776356723, "grad_norm": 0.7212371602491424, "learning_rate": 9.770306922948643e-06, "loss": 0.3759, "step": 3587 }, { "epoch": 0.24804701002419632, "grad_norm": 0.8048498410169982, "learning_rate": 9.770139182920295e-06, "loss": 0.3005, "step": 3588 }, { "epoch": 0.24811614241272037, "grad_norm": 0.7766762837097113, "learning_rate": 9.769971383106765e-06, "loss": 0.3688, "step": 3589 }, { "epoch": 0.2481852748012444, "grad_norm": 0.7766875215416293, "learning_rate": 9.769803523510155e-06, "loss": 0.4148, "step": 3590 }, { "epoch": 0.2482544071897684, "grad_norm": 1.0062241487462145, "learning_rate": 9.76963560413257e-06, "loss": 0.3483, "step": 3591 }, { "epoch": 0.24832353957829242, "grad_norm": 0.7481196354057943, "learning_rate": 9.769467624976115e-06, "loss": 0.303, "step": 3592 }, { "epoch": 0.24839267196681644, "grad_norm": 0.8269754688582979, "learning_rate": 9.769299586042894e-06, "loss": 0.3449, "step": 3593 }, { "epoch": 0.2484618043553405, "grad_norm": 0.8572488906110811, "learning_rate": 9.769131487335015e-06, "loss": 0.4276, "step": 3594 }, { "epoch": 0.2485309367438645, "grad_norm": 0.8320366184918871, "learning_rate": 9.768963328854583e-06, "loss": 0.3181, "step": 3595 }, { "epoch": 0.24860006913238852, "grad_norm": 0.9032516540884606, "learning_rate": 9.768795110603707e-06, "loss": 0.3536, "step": 3596 }, { "epoch": 0.24866920152091254, "grad_norm": 0.8162459628505258, "learning_rate": 9.768626832584494e-06, "loss": 0.4419, "step": 3597 }, { "epoch": 0.24873833390943656, "grad_norm": 0.8831220861113201, "learning_rate": 9.768458494799053e-06, "loss": 0.4039, "step": 3598 }, { "epoch": 0.2488074662979606, "grad_norm": 0.9209349256584837, "learning_rate": 9.768290097249496e-06, "loss": 0.2619, "step": 3599 }, { "epoch": 0.24887659868648462, "grad_norm": 0.8425809971921822, "learning_rate": 9.768121639937931e-06, "loss": 0.3494, "step": 3600 }, { "epoch": 0.24894573107500864, "grad_norm": 0.8623346750329062, "learning_rate": 9.767953122866471e-06, "loss": 0.4457, "step": 3601 }, { "epoch": 0.24901486346353266, "grad_norm": 1.1073162068634725, "learning_rate": 9.767784546037227e-06, "loss": 0.3668, "step": 3602 }, { "epoch": 0.24908399585205668, "grad_norm": 0.7459904496573736, "learning_rate": 9.767615909452313e-06, "loss": 0.3276, "step": 3603 }, { "epoch": 0.24915312824058072, "grad_norm": 0.8477345325560625, "learning_rate": 9.767447213113841e-06, "loss": 0.4421, "step": 3604 }, { "epoch": 0.24922226062910474, "grad_norm": 0.7623967457961868, "learning_rate": 9.767278457023925e-06, "loss": 0.4082, "step": 3605 }, { "epoch": 0.24929139301762876, "grad_norm": 0.7361734138879601, "learning_rate": 9.767109641184683e-06, "loss": 0.2984, "step": 3606 }, { "epoch": 0.24936052540615278, "grad_norm": 0.7971572830802541, "learning_rate": 9.766940765598228e-06, "loss": 0.3608, "step": 3607 }, { "epoch": 0.2494296577946768, "grad_norm": 0.792869916269362, "learning_rate": 9.766771830266677e-06, "loss": 0.4896, "step": 3608 }, { "epoch": 0.24949879018320084, "grad_norm": 1.0563675686384264, "learning_rate": 9.766602835192148e-06, "loss": 0.3439, "step": 3609 }, { "epoch": 0.24956792257172486, "grad_norm": 0.7210328854956815, "learning_rate": 9.766433780376759e-06, "loss": 0.3198, "step": 3610 }, { "epoch": 0.24963705496024888, "grad_norm": 0.9065954870908478, "learning_rate": 9.766264665822628e-06, "loss": 0.4751, "step": 3611 }, { "epoch": 0.2497061873487729, "grad_norm": 0.9414083162346799, "learning_rate": 9.766095491531877e-06, "loss": 0.445, "step": 3612 }, { "epoch": 0.2497753197372969, "grad_norm": 0.7367033412503898, "learning_rate": 9.765926257506624e-06, "loss": 0.3661, "step": 3613 }, { "epoch": 0.24984445212582096, "grad_norm": 0.832284777193329, "learning_rate": 9.76575696374899e-06, "loss": 0.2801, "step": 3614 }, { "epoch": 0.24991358451434498, "grad_norm": 0.7910121028201683, "learning_rate": 9.765587610261099e-06, "loss": 0.3818, "step": 3615 }, { "epoch": 0.249982716902869, "grad_norm": 0.9740387399504035, "learning_rate": 9.765418197045068e-06, "loss": 0.3517, "step": 3616 }, { "epoch": 0.250051849291393, "grad_norm": 0.819497781838136, "learning_rate": 9.765248724103027e-06, "loss": 0.3175, "step": 3617 }, { "epoch": 0.25012098167991703, "grad_norm": 1.5764920621539307, "learning_rate": 9.765079191437096e-06, "loss": 0.3688, "step": 3618 }, { "epoch": 0.25019011406844105, "grad_norm": 0.7957867982196906, "learning_rate": 9.764909599049403e-06, "loss": 0.3998, "step": 3619 }, { "epoch": 0.25025924645696507, "grad_norm": 0.7717062760825053, "learning_rate": 9.76473994694207e-06, "loss": 0.3501, "step": 3620 }, { "epoch": 0.2503283788454891, "grad_norm": 0.7278714853183343, "learning_rate": 9.764570235117226e-06, "loss": 0.3196, "step": 3621 }, { "epoch": 0.25039751123401316, "grad_norm": 1.053042142177066, "learning_rate": 9.764400463576994e-06, "loss": 0.4876, "step": 3622 }, { "epoch": 0.2504666436225372, "grad_norm": 1.0608358829380518, "learning_rate": 9.764230632323508e-06, "loss": 0.3825, "step": 3623 }, { "epoch": 0.2505357760110612, "grad_norm": 0.7590856077707174, "learning_rate": 9.764060741358891e-06, "loss": 0.3291, "step": 3624 }, { "epoch": 0.2506049083995852, "grad_norm": 0.8312423910027475, "learning_rate": 9.763890790685276e-06, "loss": 0.3094, "step": 3625 }, { "epoch": 0.25067404078810923, "grad_norm": 0.7597133901846296, "learning_rate": 9.76372078030479e-06, "loss": 0.4592, "step": 3626 }, { "epoch": 0.25074317317663325, "grad_norm": 0.7619550131938511, "learning_rate": 9.763550710219566e-06, "loss": 0.3833, "step": 3627 }, { "epoch": 0.25081230556515727, "grad_norm": 0.7913708516515063, "learning_rate": 9.763380580431735e-06, "loss": 0.3142, "step": 3628 }, { "epoch": 0.2508814379536813, "grad_norm": 0.7913017653927628, "learning_rate": 9.76321039094343e-06, "loss": 0.4424, "step": 3629 }, { "epoch": 0.2509505703422053, "grad_norm": 0.7482734565567083, "learning_rate": 9.763040141756782e-06, "loss": 0.3331, "step": 3630 }, { "epoch": 0.2510197027307293, "grad_norm": 0.8237604128097051, "learning_rate": 9.762869832873925e-06, "loss": 0.3444, "step": 3631 }, { "epoch": 0.2510888351192534, "grad_norm": 0.8818478079136183, "learning_rate": 9.762699464296995e-06, "loss": 0.3286, "step": 3632 }, { "epoch": 0.2511579675077774, "grad_norm": 0.743230795141543, "learning_rate": 9.762529036028127e-06, "loss": 0.4369, "step": 3633 }, { "epoch": 0.25122709989630143, "grad_norm": 0.7936826670614027, "learning_rate": 9.762358548069457e-06, "loss": 0.3541, "step": 3634 }, { "epoch": 0.25129623228482545, "grad_norm": 0.8361906579002275, "learning_rate": 9.76218800042312e-06, "loss": 0.3662, "step": 3635 }, { "epoch": 0.25136536467334947, "grad_norm": 0.8366042622433292, "learning_rate": 9.762017393091256e-06, "loss": 0.4129, "step": 3636 }, { "epoch": 0.2514344970618735, "grad_norm": 0.7976257594488189, "learning_rate": 9.761846726076001e-06, "loss": 0.422, "step": 3637 }, { "epoch": 0.2515036294503975, "grad_norm": 0.7712473699082784, "learning_rate": 9.761675999379497e-06, "loss": 0.335, "step": 3638 }, { "epoch": 0.2515727618389215, "grad_norm": 0.8356406027405748, "learning_rate": 9.761505213003882e-06, "loss": 0.3597, "step": 3639 }, { "epoch": 0.25164189422744554, "grad_norm": 0.8059364376292559, "learning_rate": 9.761334366951294e-06, "loss": 0.3972, "step": 3640 }, { "epoch": 0.25171102661596956, "grad_norm": 0.8479753936373179, "learning_rate": 9.761163461223878e-06, "loss": 0.3223, "step": 3641 }, { "epoch": 0.25178015900449363, "grad_norm": 0.8022829804652106, "learning_rate": 9.760992495823775e-06, "loss": 0.323, "step": 3642 }, { "epoch": 0.25184929139301765, "grad_norm": 0.7951341186011927, "learning_rate": 9.760821470753128e-06, "loss": 0.3807, "step": 3643 }, { "epoch": 0.25191842378154167, "grad_norm": 0.883895051580567, "learning_rate": 9.76065038601408e-06, "loss": 0.4706, "step": 3644 }, { "epoch": 0.2519875561700657, "grad_norm": 0.8121415215576917, "learning_rate": 9.760479241608774e-06, "loss": 0.3682, "step": 3645 }, { "epoch": 0.2520566885585897, "grad_norm": 0.7175090009105844, "learning_rate": 9.760308037539356e-06, "loss": 0.316, "step": 3646 }, { "epoch": 0.2521258209471137, "grad_norm": 0.7756267984353242, "learning_rate": 9.76013677380797e-06, "loss": 0.39, "step": 3647 }, { "epoch": 0.25219495333563774, "grad_norm": 0.7411477255405079, "learning_rate": 9.759965450416769e-06, "loss": 0.359, "step": 3648 }, { "epoch": 0.25226408572416176, "grad_norm": 0.744247845694343, "learning_rate": 9.759794067367894e-06, "loss": 0.2985, "step": 3649 }, { "epoch": 0.2523332181126858, "grad_norm": 0.8195145347452516, "learning_rate": 9.759622624663492e-06, "loss": 0.4227, "step": 3650 }, { "epoch": 0.2524023505012098, "grad_norm": 0.8475546944157542, "learning_rate": 9.759451122305718e-06, "loss": 0.4081, "step": 3651 }, { "epoch": 0.25247148288973387, "grad_norm": 0.7926277528907834, "learning_rate": 9.759279560296715e-06, "loss": 0.2992, "step": 3652 }, { "epoch": 0.2525406152782579, "grad_norm": 0.770124949588463, "learning_rate": 9.759107938638638e-06, "loss": 0.3672, "step": 3653 }, { "epoch": 0.2526097476667819, "grad_norm": 0.7963404216681064, "learning_rate": 9.758936257333635e-06, "loss": 0.4719, "step": 3654 }, { "epoch": 0.2526788800553059, "grad_norm": 0.855301392877646, "learning_rate": 9.758764516383858e-06, "loss": 0.3548, "step": 3655 }, { "epoch": 0.25274801244382994, "grad_norm": 0.8438820464859286, "learning_rate": 9.758592715791462e-06, "loss": 0.3951, "step": 3656 }, { "epoch": 0.25281714483235396, "grad_norm": 1.3511047467981012, "learning_rate": 9.758420855558597e-06, "loss": 0.3297, "step": 3657 }, { "epoch": 0.252886277220878, "grad_norm": 0.7802164075006699, "learning_rate": 9.75824893568742e-06, "loss": 0.4701, "step": 3658 }, { "epoch": 0.252955409609402, "grad_norm": 0.8380634861879477, "learning_rate": 9.758076956180083e-06, "loss": 0.3389, "step": 3659 }, { "epoch": 0.253024541997926, "grad_norm": 0.9964783014599099, "learning_rate": 9.757904917038743e-06, "loss": 0.3625, "step": 3660 }, { "epoch": 0.25309367438645003, "grad_norm": 0.9672519780337345, "learning_rate": 9.757732818265554e-06, "loss": 0.4567, "step": 3661 }, { "epoch": 0.2531628067749741, "grad_norm": 0.841147690836037, "learning_rate": 9.757560659862676e-06, "loss": 0.4341, "step": 3662 }, { "epoch": 0.2532319391634981, "grad_norm": 1.1043121399855413, "learning_rate": 9.757388441832265e-06, "loss": 0.3023, "step": 3663 }, { "epoch": 0.25330107155202214, "grad_norm": 0.8047194179475002, "learning_rate": 9.75721616417648e-06, "loss": 0.3682, "step": 3664 }, { "epoch": 0.25337020394054616, "grad_norm": 0.8623609892784366, "learning_rate": 9.75704382689748e-06, "loss": 0.5276, "step": 3665 }, { "epoch": 0.2534393363290702, "grad_norm": 0.7943391547756875, "learning_rate": 9.756871429997425e-06, "loss": 0.3971, "step": 3666 }, { "epoch": 0.2535084687175942, "grad_norm": 0.9726205019417075, "learning_rate": 9.756698973478475e-06, "loss": 0.3192, "step": 3667 }, { "epoch": 0.2535776011061182, "grad_norm": 0.777073642478931, "learning_rate": 9.756526457342794e-06, "loss": 0.3808, "step": 3668 }, { "epoch": 0.25364673349464223, "grad_norm": 0.8119649795558659, "learning_rate": 9.75635388159254e-06, "loss": 0.4153, "step": 3669 }, { "epoch": 0.25371586588316625, "grad_norm": 0.814062120804629, "learning_rate": 9.75618124622988e-06, "loss": 0.3313, "step": 3670 }, { "epoch": 0.25378499827169027, "grad_norm": 0.996785687389183, "learning_rate": 9.756008551256975e-06, "loss": 0.3592, "step": 3671 }, { "epoch": 0.2538541306602143, "grad_norm": 0.8548289419776858, "learning_rate": 9.75583579667599e-06, "loss": 0.4474, "step": 3672 }, { "epoch": 0.25392326304873836, "grad_norm": 0.7992811968664738, "learning_rate": 9.75566298248909e-06, "loss": 0.3515, "step": 3673 }, { "epoch": 0.2539923954372624, "grad_norm": 0.944675126202402, "learning_rate": 9.755490108698443e-06, "loss": 0.3138, "step": 3674 }, { "epoch": 0.2540615278257864, "grad_norm": 0.9206089900482615, "learning_rate": 9.755317175306214e-06, "loss": 0.4079, "step": 3675 }, { "epoch": 0.2541306602143104, "grad_norm": 0.7711781134761375, "learning_rate": 9.755144182314568e-06, "loss": 0.442, "step": 3676 }, { "epoch": 0.25419979260283443, "grad_norm": 0.7974257823054953, "learning_rate": 9.754971129725679e-06, "loss": 0.3745, "step": 3677 }, { "epoch": 0.25426892499135845, "grad_norm": 1.1423023673331378, "learning_rate": 9.754798017541712e-06, "loss": 0.3343, "step": 3678 }, { "epoch": 0.25433805737988247, "grad_norm": 0.8416847566694718, "learning_rate": 9.754624845764834e-06, "loss": 0.4545, "step": 3679 }, { "epoch": 0.2544071897684065, "grad_norm": 1.3148433233357217, "learning_rate": 9.75445161439722e-06, "loss": 0.3612, "step": 3680 }, { "epoch": 0.2544763221569305, "grad_norm": 0.9253150756524005, "learning_rate": 9.754278323441043e-06, "loss": 0.3051, "step": 3681 }, { "epoch": 0.2545454545454545, "grad_norm": 1.156897385782307, "learning_rate": 9.754104972898467e-06, "loss": 0.3702, "step": 3682 }, { "epoch": 0.2546145869339786, "grad_norm": 0.731730579376521, "learning_rate": 9.753931562771671e-06, "loss": 0.4311, "step": 3683 }, { "epoch": 0.2546837193225026, "grad_norm": 1.0803366445284124, "learning_rate": 9.753758093062827e-06, "loss": 0.3296, "step": 3684 }, { "epoch": 0.25475285171102663, "grad_norm": 1.0824046741873568, "learning_rate": 9.753584563774109e-06, "loss": 0.3504, "step": 3685 }, { "epoch": 0.25482198409955065, "grad_norm": 0.8025068811435437, "learning_rate": 9.753410974907691e-06, "loss": 0.3967, "step": 3686 }, { "epoch": 0.25489111648807466, "grad_norm": 1.0428090375973786, "learning_rate": 9.75323732646575e-06, "loss": 0.4141, "step": 3687 }, { "epoch": 0.2549602488765987, "grad_norm": 1.0365576482271455, "learning_rate": 9.75306361845046e-06, "loss": 0.3304, "step": 3688 }, { "epoch": 0.2550293812651227, "grad_norm": 0.9603237748834584, "learning_rate": 9.752889850864002e-06, "loss": 0.3696, "step": 3689 }, { "epoch": 0.2550985136536467, "grad_norm": 0.7797718491840911, "learning_rate": 9.75271602370855e-06, "loss": 0.3762, "step": 3690 }, { "epoch": 0.25516764604217074, "grad_norm": 0.8556686310544804, "learning_rate": 9.752542136986285e-06, "loss": 0.289, "step": 3691 }, { "epoch": 0.25523677843069476, "grad_norm": 0.812986680665352, "learning_rate": 9.752368190699384e-06, "loss": 0.3397, "step": 3692 }, { "epoch": 0.25530591081921883, "grad_norm": 1.1297275786171297, "learning_rate": 9.75219418485003e-06, "loss": 0.3917, "step": 3693 }, { "epoch": 0.25537504320774285, "grad_norm": 0.7580851307355563, "learning_rate": 9.752020119440404e-06, "loss": 0.4732, "step": 3694 }, { "epoch": 0.25544417559626686, "grad_norm": 1.0446141562062454, "learning_rate": 9.751845994472684e-06, "loss": 0.3678, "step": 3695 }, { "epoch": 0.2555133079847909, "grad_norm": 0.7826068466600294, "learning_rate": 9.751671809949055e-06, "loss": 0.3496, "step": 3696 }, { "epoch": 0.2555824403733149, "grad_norm": 0.8446642029183856, "learning_rate": 9.7514975658717e-06, "loss": 0.4355, "step": 3697 }, { "epoch": 0.2556515727618389, "grad_norm": 0.7513455941296422, "learning_rate": 9.751323262242802e-06, "loss": 0.3853, "step": 3698 }, { "epoch": 0.25572070515036294, "grad_norm": 0.7923265813882361, "learning_rate": 9.751148899064547e-06, "loss": 0.2672, "step": 3699 }, { "epoch": 0.25578983753888695, "grad_norm": 0.9165359047865438, "learning_rate": 9.75097447633912e-06, "loss": 0.4637, "step": 3700 }, { "epoch": 0.255858969927411, "grad_norm": 0.7884606384394057, "learning_rate": 9.750799994068704e-06, "loss": 0.4608, "step": 3701 }, { "epoch": 0.255928102315935, "grad_norm": 0.8296259050926147, "learning_rate": 9.75062545225549e-06, "loss": 0.4145, "step": 3702 }, { "epoch": 0.25599723470445906, "grad_norm": 0.8599916127646048, "learning_rate": 9.750450850901663e-06, "loss": 0.2915, "step": 3703 }, { "epoch": 0.2560663670929831, "grad_norm": 0.8346197311574194, "learning_rate": 9.75027619000941e-06, "loss": 0.4327, "step": 3704 }, { "epoch": 0.2561354994815071, "grad_norm": 0.7135462332345766, "learning_rate": 9.750101469580927e-06, "loss": 0.4165, "step": 3705 }, { "epoch": 0.2562046318700311, "grad_norm": 1.180374727367639, "learning_rate": 9.749926689618397e-06, "loss": 0.3307, "step": 3706 }, { "epoch": 0.25627376425855514, "grad_norm": 0.845266814846547, "learning_rate": 9.749751850124011e-06, "loss": 0.4222, "step": 3707 }, { "epoch": 0.25634289664707915, "grad_norm": 0.8385385268260193, "learning_rate": 9.749576951099964e-06, "loss": 0.4512, "step": 3708 }, { "epoch": 0.2564120290356032, "grad_norm": 0.8298740403905496, "learning_rate": 9.749401992548443e-06, "loss": 0.3552, "step": 3709 }, { "epoch": 0.2564811614241272, "grad_norm": 0.7034874549620758, "learning_rate": 9.749226974471645e-06, "loss": 0.27, "step": 3710 }, { "epoch": 0.2565502938126512, "grad_norm": 0.764897766535123, "learning_rate": 9.749051896871763e-06, "loss": 0.4133, "step": 3711 }, { "epoch": 0.2566194262011752, "grad_norm": 0.7675233311070566, "learning_rate": 9.748876759750991e-06, "loss": 0.519, "step": 3712 }, { "epoch": 0.2566885585896993, "grad_norm": 0.8321065250363565, "learning_rate": 9.748701563111525e-06, "loss": 0.335, "step": 3713 }, { "epoch": 0.2567576909782233, "grad_norm": 0.6863487842179888, "learning_rate": 9.748526306955558e-06, "loss": 0.2833, "step": 3714 }, { "epoch": 0.25682682336674734, "grad_norm": 0.7586319314333633, "learning_rate": 9.748350991285285e-06, "loss": 0.4273, "step": 3715 }, { "epoch": 0.25689595575527135, "grad_norm": 0.7559812693202521, "learning_rate": 9.748175616102908e-06, "loss": 0.3609, "step": 3716 }, { "epoch": 0.2569650881437954, "grad_norm": 0.8369873182241034, "learning_rate": 9.748000181410624e-06, "loss": 0.3668, "step": 3717 }, { "epoch": 0.2570342205323194, "grad_norm": 0.8374057513210746, "learning_rate": 9.747824687210631e-06, "loss": 0.3977, "step": 3718 }, { "epoch": 0.2571033529208434, "grad_norm": 0.7508290165913309, "learning_rate": 9.747649133505129e-06, "loss": 0.3917, "step": 3719 }, { "epoch": 0.2571724853093674, "grad_norm": 0.8853925697812792, "learning_rate": 9.747473520296316e-06, "loss": 0.3595, "step": 3720 }, { "epoch": 0.25724161769789144, "grad_norm": 1.430413242121493, "learning_rate": 9.747297847586394e-06, "loss": 0.3373, "step": 3721 }, { "epoch": 0.25731075008641546, "grad_norm": 0.8955004314636645, "learning_rate": 9.747122115377567e-06, "loss": 0.4844, "step": 3722 }, { "epoch": 0.25737988247493954, "grad_norm": 0.7995697374624541, "learning_rate": 9.746946323672035e-06, "loss": 0.3933, "step": 3723 }, { "epoch": 0.25744901486346355, "grad_norm": 2.986306613217133, "learning_rate": 9.746770472472004e-06, "loss": 0.3761, "step": 3724 }, { "epoch": 0.2575181472519876, "grad_norm": 0.7641054395579114, "learning_rate": 9.746594561779674e-06, "loss": 0.3408, "step": 3725 }, { "epoch": 0.2575872796405116, "grad_norm": 0.7798953561976053, "learning_rate": 9.746418591597253e-06, "loss": 0.4431, "step": 3726 }, { "epoch": 0.2576564120290356, "grad_norm": 0.6989762626130769, "learning_rate": 9.746242561926944e-06, "loss": 0.3221, "step": 3727 }, { "epoch": 0.2577255444175596, "grad_norm": 0.7689925202239042, "learning_rate": 9.746066472770957e-06, "loss": 0.2943, "step": 3728 }, { "epoch": 0.25779467680608364, "grad_norm": 0.752871086751888, "learning_rate": 9.745890324131494e-06, "loss": 0.3756, "step": 3729 }, { "epoch": 0.25786380919460766, "grad_norm": 1.0715373060629536, "learning_rate": 9.745714116010766e-06, "loss": 0.3632, "step": 3730 }, { "epoch": 0.2579329415831317, "grad_norm": 0.7873028958682797, "learning_rate": 9.745537848410981e-06, "loss": 0.3616, "step": 3731 }, { "epoch": 0.2580020739716557, "grad_norm": 0.8236529772926224, "learning_rate": 9.745361521334347e-06, "loss": 0.3637, "step": 3732 }, { "epoch": 0.2580712063601797, "grad_norm": 0.8088100225142933, "learning_rate": 9.745185134783076e-06, "loss": 0.4512, "step": 3733 }, { "epoch": 0.2581403387487038, "grad_norm": 0.7953539164151285, "learning_rate": 9.745008688759377e-06, "loss": 0.3806, "step": 3734 }, { "epoch": 0.2582094711372278, "grad_norm": 1.6306393449825207, "learning_rate": 9.744832183265462e-06, "loss": 0.312, "step": 3735 }, { "epoch": 0.2582786035257518, "grad_norm": 0.8429808618737485, "learning_rate": 9.744655618303543e-06, "loss": 0.3713, "step": 3736 }, { "epoch": 0.25834773591427584, "grad_norm": 0.784783924108202, "learning_rate": 9.744478993875834e-06, "loss": 0.4141, "step": 3737 }, { "epoch": 0.25841686830279986, "grad_norm": 1.0331858430049932, "learning_rate": 9.744302309984546e-06, "loss": 0.3183, "step": 3738 }, { "epoch": 0.2584860006913239, "grad_norm": 0.7909816527807808, "learning_rate": 9.744125566631896e-06, "loss": 0.4099, "step": 3739 }, { "epoch": 0.2585551330798479, "grad_norm": 0.7338232497680417, "learning_rate": 9.7439487638201e-06, "loss": 0.4703, "step": 3740 }, { "epoch": 0.2586242654683719, "grad_norm": 0.7411800226050632, "learning_rate": 9.74377190155137e-06, "loss": 0.373, "step": 3741 }, { "epoch": 0.25869339785689593, "grad_norm": 0.7773264137628724, "learning_rate": 9.743594979827925e-06, "loss": 0.303, "step": 3742 }, { "epoch": 0.25876253024541995, "grad_norm": 0.8771243786419093, "learning_rate": 9.743417998651982e-06, "loss": 0.4204, "step": 3743 }, { "epoch": 0.258831662633944, "grad_norm": 0.8392226065883766, "learning_rate": 9.743240958025759e-06, "loss": 0.4177, "step": 3744 }, { "epoch": 0.25890079502246804, "grad_norm": 0.8207247130262338, "learning_rate": 9.743063857951477e-06, "loss": 0.3285, "step": 3745 }, { "epoch": 0.25896992741099206, "grad_norm": 0.9027474728293514, "learning_rate": 9.742886698431353e-06, "loss": 0.3405, "step": 3746 }, { "epoch": 0.2590390597995161, "grad_norm": 1.0013461130016061, "learning_rate": 9.74270947946761e-06, "loss": 0.4477, "step": 3747 }, { "epoch": 0.2591081921880401, "grad_norm": 0.7471966502007997, "learning_rate": 9.742532201062465e-06, "loss": 0.3935, "step": 3748 }, { "epoch": 0.2591773245765641, "grad_norm": 0.8721351481763585, "learning_rate": 9.742354863218142e-06, "loss": 0.3988, "step": 3749 }, { "epoch": 0.25924645696508813, "grad_norm": 0.7728487270715032, "learning_rate": 9.742177465936866e-06, "loss": 0.3799, "step": 3750 }, { "epoch": 0.25931558935361215, "grad_norm": 0.8870447649941035, "learning_rate": 9.742000009220857e-06, "loss": 0.4388, "step": 3751 }, { "epoch": 0.25938472174213617, "grad_norm": 0.9953099604087573, "learning_rate": 9.741822493072341e-06, "loss": 0.3161, "step": 3752 }, { "epoch": 0.2594538541306602, "grad_norm": 0.9167065347956993, "learning_rate": 9.741644917493543e-06, "loss": 0.3817, "step": 3753 }, { "epoch": 0.25952298651918426, "grad_norm": 0.8948779239185419, "learning_rate": 9.741467282486685e-06, "loss": 0.4576, "step": 3754 }, { "epoch": 0.2595921189077083, "grad_norm": 0.8813826280719371, "learning_rate": 9.741289588053999e-06, "loss": 0.3374, "step": 3755 }, { "epoch": 0.2596612512962323, "grad_norm": 0.7710134017181606, "learning_rate": 9.741111834197707e-06, "loss": 0.3493, "step": 3756 }, { "epoch": 0.2597303836847563, "grad_norm": 0.8714225712501485, "learning_rate": 9.740934020920042e-06, "loss": 0.3102, "step": 3757 }, { "epoch": 0.25979951607328033, "grad_norm": 0.8399432462477879, "learning_rate": 9.740756148223227e-06, "loss": 0.4386, "step": 3758 }, { "epoch": 0.25986864846180435, "grad_norm": 0.7979774866176682, "learning_rate": 9.740578216109494e-06, "loss": 0.3684, "step": 3759 }, { "epoch": 0.25993778085032837, "grad_norm": 0.8449245361359695, "learning_rate": 9.740400224581073e-06, "loss": 0.3048, "step": 3760 }, { "epoch": 0.2600069132388524, "grad_norm": 0.881449925514056, "learning_rate": 9.740222173640196e-06, "loss": 0.4091, "step": 3761 }, { "epoch": 0.2600760456273764, "grad_norm": 0.7907099617995891, "learning_rate": 9.740044063289094e-06, "loss": 0.4097, "step": 3762 }, { "epoch": 0.2601451780159004, "grad_norm": 0.7118564386308988, "learning_rate": 9.739865893529997e-06, "loss": 0.3453, "step": 3763 }, { "epoch": 0.2602143104044245, "grad_norm": 0.7557075526395346, "learning_rate": 9.73968766436514e-06, "loss": 0.3434, "step": 3764 }, { "epoch": 0.2602834427929485, "grad_norm": 0.8459626584593725, "learning_rate": 9.739509375796757e-06, "loss": 0.4686, "step": 3765 }, { "epoch": 0.26035257518147253, "grad_norm": 0.7517035006203219, "learning_rate": 9.739331027827082e-06, "loss": 0.3482, "step": 3766 }, { "epoch": 0.26042170756999655, "grad_norm": 0.8206155071507018, "learning_rate": 9.73915262045835e-06, "loss": 0.2888, "step": 3767 }, { "epoch": 0.26049083995852057, "grad_norm": 2.104078492891995, "learning_rate": 9.738974153692797e-06, "loss": 0.3247, "step": 3768 }, { "epoch": 0.2605599723470446, "grad_norm": 0.7979956796231094, "learning_rate": 9.73879562753266e-06, "loss": 0.4734, "step": 3769 }, { "epoch": 0.2606291047355686, "grad_norm": 0.8991759486954586, "learning_rate": 9.738617041980176e-06, "loss": 0.3902, "step": 3770 }, { "epoch": 0.2606982371240926, "grad_norm": 0.751041702442587, "learning_rate": 9.738438397037585e-06, "loss": 0.3944, "step": 3771 }, { "epoch": 0.26076736951261664, "grad_norm": 0.8874464362189315, "learning_rate": 9.738259692707125e-06, "loss": 0.4703, "step": 3772 }, { "epoch": 0.26083650190114066, "grad_norm": 0.7506219808382901, "learning_rate": 9.738080928991035e-06, "loss": 0.4531, "step": 3773 }, { "epoch": 0.26090563428966473, "grad_norm": 1.1053524955419067, "learning_rate": 9.737902105891557e-06, "loss": 0.3303, "step": 3774 }, { "epoch": 0.26097476667818875, "grad_norm": 0.8415207851964875, "learning_rate": 9.737723223410928e-06, "loss": 0.3713, "step": 3775 }, { "epoch": 0.26104389906671277, "grad_norm": 0.8399927781843713, "learning_rate": 9.737544281551397e-06, "loss": 0.4621, "step": 3776 }, { "epoch": 0.2611130314552368, "grad_norm": 0.7610628544330246, "learning_rate": 9.737365280315202e-06, "loss": 0.3598, "step": 3777 }, { "epoch": 0.2611821638437608, "grad_norm": 0.8463329089710614, "learning_rate": 9.737186219704586e-06, "loss": 0.327, "step": 3778 }, { "epoch": 0.2612512962322848, "grad_norm": 0.882609002347954, "learning_rate": 9.737007099721796e-06, "loss": 0.4314, "step": 3779 }, { "epoch": 0.26132042862080884, "grad_norm": 0.8611891894639875, "learning_rate": 9.736827920369077e-06, "loss": 0.4121, "step": 3780 }, { "epoch": 0.26138956100933286, "grad_norm": 0.7227102312873237, "learning_rate": 9.73664868164867e-06, "loss": 0.2986, "step": 3781 }, { "epoch": 0.2614586933978569, "grad_norm": 0.8633062660563828, "learning_rate": 9.736469383562827e-06, "loss": 0.4157, "step": 3782 }, { "epoch": 0.2615278257863809, "grad_norm": 0.8430014365036196, "learning_rate": 9.736290026113792e-06, "loss": 0.4467, "step": 3783 }, { "epoch": 0.26159695817490497, "grad_norm": 0.6839236873727802, "learning_rate": 9.736110609303813e-06, "loss": 0.3309, "step": 3784 }, { "epoch": 0.261666090563429, "grad_norm": 0.7670838182374684, "learning_rate": 9.735931133135142e-06, "loss": 0.3331, "step": 3785 }, { "epoch": 0.261735222951953, "grad_norm": 0.8561525600606187, "learning_rate": 9.735751597610023e-06, "loss": 0.3948, "step": 3786 }, { "epoch": 0.261804355340477, "grad_norm": 0.74298115159789, "learning_rate": 9.735572002730711e-06, "loss": 0.3891, "step": 3787 }, { "epoch": 0.26187348772900104, "grad_norm": 0.7342239270486791, "learning_rate": 9.735392348499454e-06, "loss": 0.3364, "step": 3788 }, { "epoch": 0.26194262011752506, "grad_norm": 0.7757153404868289, "learning_rate": 9.735212634918505e-06, "loss": 0.3332, "step": 3789 }, { "epoch": 0.2620117525060491, "grad_norm": 0.918488377178445, "learning_rate": 9.735032861990114e-06, "loss": 0.457, "step": 3790 }, { "epoch": 0.2620808848945731, "grad_norm": 0.7656640278721193, "learning_rate": 9.73485302971654e-06, "loss": 0.335, "step": 3791 }, { "epoch": 0.2621500172830971, "grad_norm": 0.7538999011846664, "learning_rate": 9.734673138100031e-06, "loss": 0.3163, "step": 3792 }, { "epoch": 0.26221914967162113, "grad_norm": 0.7988452227998155, "learning_rate": 9.734493187142843e-06, "loss": 0.3795, "step": 3793 }, { "epoch": 0.26228828206014515, "grad_norm": 0.8362166174983828, "learning_rate": 9.734313176847232e-06, "loss": 0.4335, "step": 3794 }, { "epoch": 0.2623574144486692, "grad_norm": 0.8467639130797602, "learning_rate": 9.734133107215456e-06, "loss": 0.354, "step": 3795 }, { "epoch": 0.26242654683719324, "grad_norm": 1.0413089969303044, "learning_rate": 9.733952978249768e-06, "loss": 0.2848, "step": 3796 }, { "epoch": 0.26249567922571726, "grad_norm": 0.8422443056262442, "learning_rate": 9.733772789952429e-06, "loss": 0.4005, "step": 3797 }, { "epoch": 0.2625648116142413, "grad_norm": 0.7618022547794209, "learning_rate": 9.733592542325693e-06, "loss": 0.3751, "step": 3798 }, { "epoch": 0.2626339440027653, "grad_norm": 0.7530842107060349, "learning_rate": 9.733412235371825e-06, "loss": 0.3754, "step": 3799 }, { "epoch": 0.2627030763912893, "grad_norm": 0.7697808975725502, "learning_rate": 9.73323186909308e-06, "loss": 0.3096, "step": 3800 }, { "epoch": 0.26277220877981333, "grad_norm": 0.8187397669028678, "learning_rate": 9.733051443491722e-06, "loss": 0.4591, "step": 3801 }, { "epoch": 0.26284134116833735, "grad_norm": 0.7415471916227029, "learning_rate": 9.73287095857001e-06, "loss": 0.2974, "step": 3802 }, { "epoch": 0.26291047355686137, "grad_norm": 0.8499363226826103, "learning_rate": 9.732690414330207e-06, "loss": 0.3312, "step": 3803 }, { "epoch": 0.2629796059453854, "grad_norm": 0.801179204914978, "learning_rate": 9.732509810774575e-06, "loss": 0.4304, "step": 3804 }, { "epoch": 0.26304873833390946, "grad_norm": 0.7711519284057595, "learning_rate": 9.732329147905377e-06, "loss": 0.4161, "step": 3805 }, { "epoch": 0.2631178707224335, "grad_norm": 0.7355629275356693, "learning_rate": 9.732148425724879e-06, "loss": 0.3139, "step": 3806 }, { "epoch": 0.2631870031109575, "grad_norm": 0.7807471682216414, "learning_rate": 9.731967644235347e-06, "loss": 0.3675, "step": 3807 }, { "epoch": 0.2632561354994815, "grad_norm": 0.7444229062487351, "learning_rate": 9.731786803439044e-06, "loss": 0.4342, "step": 3808 }, { "epoch": 0.26332526788800553, "grad_norm": 1.641739111129251, "learning_rate": 9.731605903338236e-06, "loss": 0.3701, "step": 3809 }, { "epoch": 0.26339440027652955, "grad_norm": 0.797125032988392, "learning_rate": 9.731424943935196e-06, "loss": 0.3315, "step": 3810 }, { "epoch": 0.26346353266505357, "grad_norm": 0.9411346125291511, "learning_rate": 9.731243925232185e-06, "loss": 0.3718, "step": 3811 }, { "epoch": 0.2635326650535776, "grad_norm": 0.8211810472516602, "learning_rate": 9.731062847231475e-06, "loss": 0.3699, "step": 3812 }, { "epoch": 0.2636017974421016, "grad_norm": 0.8156822596616711, "learning_rate": 9.730881709935335e-06, "loss": 0.3277, "step": 3813 }, { "epoch": 0.2636709298306256, "grad_norm": 0.7851753246625199, "learning_rate": 9.730700513346037e-06, "loss": 0.3212, "step": 3814 }, { "epoch": 0.2637400622191497, "grad_norm": 0.8258947082525138, "learning_rate": 9.730519257465849e-06, "loss": 0.5071, "step": 3815 }, { "epoch": 0.2638091946076737, "grad_norm": 0.8049915895551999, "learning_rate": 9.730337942297044e-06, "loss": 0.421, "step": 3816 }, { "epoch": 0.26387832699619773, "grad_norm": 0.7917462173935959, "learning_rate": 9.730156567841894e-06, "loss": 0.3427, "step": 3817 }, { "epoch": 0.26394745938472175, "grad_norm": 0.7264748113434735, "learning_rate": 9.729975134102675e-06, "loss": 0.3443, "step": 3818 }, { "epoch": 0.26401659177324577, "grad_norm": 0.7526413891842545, "learning_rate": 9.729793641081657e-06, "loss": 0.4196, "step": 3819 }, { "epoch": 0.2640857241617698, "grad_norm": 0.8924197950596121, "learning_rate": 9.729612088781115e-06, "loss": 0.3644, "step": 3820 }, { "epoch": 0.2641548565502938, "grad_norm": 0.7266109605373535, "learning_rate": 9.729430477203328e-06, "loss": 0.3098, "step": 3821 }, { "epoch": 0.2642239889388178, "grad_norm": 0.7890562953459073, "learning_rate": 9.72924880635057e-06, "loss": 0.466, "step": 3822 }, { "epoch": 0.26429312132734184, "grad_norm": 0.7954873814155382, "learning_rate": 9.729067076225118e-06, "loss": 0.3586, "step": 3823 }, { "epoch": 0.26436225371586586, "grad_norm": 0.7125516147549384, "learning_rate": 9.728885286829249e-06, "loss": 0.3273, "step": 3824 }, { "epoch": 0.26443138610438993, "grad_norm": 0.8063460295903864, "learning_rate": 9.728703438165241e-06, "loss": 0.5059, "step": 3825 }, { "epoch": 0.26450051849291395, "grad_norm": 0.774721472222705, "learning_rate": 9.728521530235375e-06, "loss": 0.5083, "step": 3826 }, { "epoch": 0.26456965088143797, "grad_norm": 0.7755884871169658, "learning_rate": 9.72833956304193e-06, "loss": 0.3526, "step": 3827 }, { "epoch": 0.264638783269962, "grad_norm": 0.7634881597026666, "learning_rate": 9.728157536587188e-06, "loss": 0.3509, "step": 3828 }, { "epoch": 0.264707915658486, "grad_norm": 0.8797671030905613, "learning_rate": 9.727975450873428e-06, "loss": 0.4333, "step": 3829 }, { "epoch": 0.26477704804701, "grad_norm": 0.7943878414127623, "learning_rate": 9.727793305902933e-06, "loss": 0.4178, "step": 3830 }, { "epoch": 0.26484618043553404, "grad_norm": 0.7586260134447007, "learning_rate": 9.727611101677987e-06, "loss": 0.3241, "step": 3831 }, { "epoch": 0.26491531282405806, "grad_norm": 0.6502610771987449, "learning_rate": 9.727428838200871e-06, "loss": 0.3076, "step": 3832 }, { "epoch": 0.2649844452125821, "grad_norm": 0.7452362625159227, "learning_rate": 9.727246515473872e-06, "loss": 0.4657, "step": 3833 }, { "epoch": 0.2650535776011061, "grad_norm": 0.8992453483911187, "learning_rate": 9.727064133499274e-06, "loss": 0.3618, "step": 3834 }, { "epoch": 0.26512270998963017, "grad_norm": 0.8536876225048526, "learning_rate": 9.726881692279363e-06, "loss": 0.2787, "step": 3835 }, { "epoch": 0.2651918423781542, "grad_norm": 0.7491436548354412, "learning_rate": 9.726699191816425e-06, "loss": 0.43, "step": 3836 }, { "epoch": 0.2652609747666782, "grad_norm": 0.8231385821769088, "learning_rate": 9.726516632112748e-06, "loss": 0.4749, "step": 3837 }, { "epoch": 0.2653301071552022, "grad_norm": 0.7998624057712883, "learning_rate": 9.72633401317062e-06, "loss": 0.4031, "step": 3838 }, { "epoch": 0.26539923954372624, "grad_norm": 0.7852347325029249, "learning_rate": 9.726151334992331e-06, "loss": 0.3665, "step": 3839 }, { "epoch": 0.26546837193225026, "grad_norm": 0.7473720600507762, "learning_rate": 9.725968597580166e-06, "loss": 0.4137, "step": 3840 }, { "epoch": 0.2655375043207743, "grad_norm": 0.7623638984977219, "learning_rate": 9.725785800936421e-06, "loss": 0.3391, "step": 3841 }, { "epoch": 0.2656066367092983, "grad_norm": 0.7868589869083693, "learning_rate": 9.725602945063383e-06, "loss": 0.3639, "step": 3842 }, { "epoch": 0.2656757690978223, "grad_norm": 0.778683213745328, "learning_rate": 9.725420029963348e-06, "loss": 0.4227, "step": 3843 }, { "epoch": 0.26574490148634633, "grad_norm": 0.8364726357928325, "learning_rate": 9.725237055638603e-06, "loss": 0.4936, "step": 3844 }, { "epoch": 0.2658140338748704, "grad_norm": 0.7248639877575158, "learning_rate": 9.725054022091446e-06, "loss": 0.3441, "step": 3845 }, { "epoch": 0.2658831662633944, "grad_norm": 0.9816345614655073, "learning_rate": 9.724870929324169e-06, "loss": 0.3148, "step": 3846 }, { "epoch": 0.26595229865191844, "grad_norm": 0.7987949339109912, "learning_rate": 9.724687777339066e-06, "loss": 0.4661, "step": 3847 }, { "epoch": 0.26602143104044246, "grad_norm": 0.6765220343804106, "learning_rate": 9.724504566138434e-06, "loss": 0.3531, "step": 3848 }, { "epoch": 0.2660905634289665, "grad_norm": 2.0423585740998216, "learning_rate": 9.724321295724566e-06, "loss": 0.2696, "step": 3849 }, { "epoch": 0.2661596958174905, "grad_norm": 0.7530000534022357, "learning_rate": 9.724137966099762e-06, "loss": 0.3502, "step": 3850 }, { "epoch": 0.2662288282060145, "grad_norm": 0.7951186057452669, "learning_rate": 9.723954577266321e-06, "loss": 0.468, "step": 3851 }, { "epoch": 0.26629796059453853, "grad_norm": 0.7731412010483749, "learning_rate": 9.72377112922654e-06, "loss": 0.3114, "step": 3852 }, { "epoch": 0.26636709298306255, "grad_norm": 0.8743970773268228, "learning_rate": 9.723587621982718e-06, "loss": 0.252, "step": 3853 }, { "epoch": 0.26643622537158657, "grad_norm": 0.7510956786981616, "learning_rate": 9.723404055537154e-06, "loss": 0.3845, "step": 3854 }, { "epoch": 0.26650535776011064, "grad_norm": 0.8190441172240095, "learning_rate": 9.723220429892148e-06, "loss": 0.4396, "step": 3855 }, { "epoch": 0.26657449014863466, "grad_norm": 0.9365874032318866, "learning_rate": 9.723036745050004e-06, "loss": 0.3409, "step": 3856 }, { "epoch": 0.2666436225371587, "grad_norm": 0.7604687514991902, "learning_rate": 9.722853001013024e-06, "loss": 0.3619, "step": 3857 }, { "epoch": 0.2667127549256827, "grad_norm": 0.8780043668689919, "learning_rate": 9.722669197783509e-06, "loss": 0.4637, "step": 3858 }, { "epoch": 0.2667818873142067, "grad_norm": 0.7851810830988641, "learning_rate": 9.722485335363762e-06, "loss": 0.3786, "step": 3859 }, { "epoch": 0.26685101970273073, "grad_norm": 1.2064468838617908, "learning_rate": 9.722301413756092e-06, "loss": 0.3257, "step": 3860 }, { "epoch": 0.26692015209125475, "grad_norm": 0.8958109349450608, "learning_rate": 9.722117432962798e-06, "loss": 0.376, "step": 3861 }, { "epoch": 0.26698928447977877, "grad_norm": 0.8082656875179877, "learning_rate": 9.721933392986192e-06, "loss": 0.4316, "step": 3862 }, { "epoch": 0.2670584168683028, "grad_norm": 0.8659497655400679, "learning_rate": 9.721749293828577e-06, "loss": 0.3634, "step": 3863 }, { "epoch": 0.2671275492568268, "grad_norm": 0.810555453322585, "learning_rate": 9.721565135492259e-06, "loss": 0.3636, "step": 3864 }, { "epoch": 0.2671966816453508, "grad_norm": 0.8125368707773692, "learning_rate": 9.721380917979551e-06, "loss": 0.4583, "step": 3865 }, { "epoch": 0.2672658140338749, "grad_norm": 0.9599884696284359, "learning_rate": 9.721196641292756e-06, "loss": 0.3414, "step": 3866 }, { "epoch": 0.2673349464223989, "grad_norm": 0.7211760384368986, "learning_rate": 9.721012305434187e-06, "loss": 0.3109, "step": 3867 }, { "epoch": 0.26740407881092293, "grad_norm": 0.7945175839136144, "learning_rate": 9.720827910406156e-06, "loss": 0.4635, "step": 3868 }, { "epoch": 0.26747321119944695, "grad_norm": 0.8038613915546243, "learning_rate": 9.72064345621097e-06, "loss": 0.4114, "step": 3869 }, { "epoch": 0.26754234358797097, "grad_norm": 0.7786231686187257, "learning_rate": 9.720458942850942e-06, "loss": 0.3489, "step": 3870 }, { "epoch": 0.267611475976495, "grad_norm": 0.7464193860868243, "learning_rate": 9.720274370328386e-06, "loss": 0.3354, "step": 3871 }, { "epoch": 0.267680608365019, "grad_norm": 0.8774927367044287, "learning_rate": 9.720089738645615e-06, "loss": 0.5544, "step": 3872 }, { "epoch": 0.267749740753543, "grad_norm": 0.7777848673884277, "learning_rate": 9.719905047804942e-06, "loss": 0.3509, "step": 3873 }, { "epoch": 0.26781887314206704, "grad_norm": 0.9502923042921564, "learning_rate": 9.719720297808682e-06, "loss": 0.2601, "step": 3874 }, { "epoch": 0.26788800553059106, "grad_norm": 0.8308975816966205, "learning_rate": 9.719535488659152e-06, "loss": 0.4016, "step": 3875 }, { "epoch": 0.26795713791911513, "grad_norm": 0.7908516578525014, "learning_rate": 9.719350620358666e-06, "loss": 0.4549, "step": 3876 }, { "epoch": 0.26802627030763915, "grad_norm": 0.7803290436474143, "learning_rate": 9.719165692909542e-06, "loss": 0.3478, "step": 3877 }, { "epoch": 0.26809540269616317, "grad_norm": 0.8441397223748016, "learning_rate": 9.718980706314096e-06, "loss": 0.3654, "step": 3878 }, { "epoch": 0.2681645350846872, "grad_norm": 0.7313353672814729, "learning_rate": 9.718795660574652e-06, "loss": 0.4162, "step": 3879 }, { "epoch": 0.2682336674732112, "grad_norm": 0.729786930229315, "learning_rate": 9.718610555693523e-06, "loss": 0.3325, "step": 3880 }, { "epoch": 0.2683027998617352, "grad_norm": 0.7923522565686321, "learning_rate": 9.718425391673032e-06, "loss": 0.2922, "step": 3881 }, { "epoch": 0.26837193225025924, "grad_norm": 0.934327953397426, "learning_rate": 9.7182401685155e-06, "loss": 0.4381, "step": 3882 }, { "epoch": 0.26844106463878326, "grad_norm": 1.2330562801878975, "learning_rate": 9.718054886223248e-06, "loss": 0.4205, "step": 3883 }, { "epoch": 0.2685101970273073, "grad_norm": 0.8933851862438098, "learning_rate": 9.717869544798596e-06, "loss": 0.3424, "step": 3884 }, { "epoch": 0.2685793294158313, "grad_norm": 0.7538499561608695, "learning_rate": 9.717684144243868e-06, "loss": 0.3002, "step": 3885 }, { "epoch": 0.26864846180435537, "grad_norm": 0.8068764695117276, "learning_rate": 9.71749868456139e-06, "loss": 0.3824, "step": 3886 }, { "epoch": 0.2687175941928794, "grad_norm": 0.8035516094246599, "learning_rate": 9.717313165753484e-06, "loss": 0.3954, "step": 3887 }, { "epoch": 0.2687867265814034, "grad_norm": 0.8603227749531329, "learning_rate": 9.717127587822477e-06, "loss": 0.3335, "step": 3888 }, { "epoch": 0.2688558589699274, "grad_norm": 0.8144934723291646, "learning_rate": 9.716941950770692e-06, "loss": 0.3618, "step": 3889 }, { "epoch": 0.26892499135845144, "grad_norm": 0.7777250612930322, "learning_rate": 9.71675625460046e-06, "loss": 0.4305, "step": 3890 }, { "epoch": 0.26899412374697546, "grad_norm": 0.8761834229857633, "learning_rate": 9.716570499314102e-06, "loss": 0.3904, "step": 3891 }, { "epoch": 0.2690632561354995, "grad_norm": 0.7940520745134442, "learning_rate": 9.716384684913953e-06, "loss": 0.3309, "step": 3892 }, { "epoch": 0.2691323885240235, "grad_norm": 0.8990490482096782, "learning_rate": 9.716198811402338e-06, "loss": 0.4227, "step": 3893 }, { "epoch": 0.2692015209125475, "grad_norm": 0.83761933204784, "learning_rate": 9.716012878781586e-06, "loss": 0.4615, "step": 3894 }, { "epoch": 0.2692706533010715, "grad_norm": 1.359966152150947, "learning_rate": 9.71582688705403e-06, "loss": 0.3365, "step": 3895 }, { "epoch": 0.2693397856895956, "grad_norm": 0.7451050724107285, "learning_rate": 9.715640836222e-06, "loss": 0.3208, "step": 3896 }, { "epoch": 0.2694089180781196, "grad_norm": 0.8377979688916495, "learning_rate": 9.715454726287828e-06, "loss": 0.4462, "step": 3897 }, { "epoch": 0.26947805046664364, "grad_norm": 0.8532897917622511, "learning_rate": 9.715268557253844e-06, "loss": 0.3745, "step": 3898 }, { "epoch": 0.26954718285516766, "grad_norm": 0.8007514479211651, "learning_rate": 9.715082329122385e-06, "loss": 0.3164, "step": 3899 }, { "epoch": 0.2696163152436917, "grad_norm": 0.7385767517860595, "learning_rate": 9.714896041895783e-06, "loss": 0.3313, "step": 3900 }, { "epoch": 0.2696854476322157, "grad_norm": 0.7260011214073576, "learning_rate": 9.714709695576372e-06, "loss": 0.472, "step": 3901 }, { "epoch": 0.2697545800207397, "grad_norm": 0.8618493748565532, "learning_rate": 9.71452329016649e-06, "loss": 0.3246, "step": 3902 }, { "epoch": 0.2698237124092637, "grad_norm": 0.9294428048024099, "learning_rate": 9.714336825668474e-06, "loss": 0.3958, "step": 3903 }, { "epoch": 0.26989284479778775, "grad_norm": 0.8132665636421823, "learning_rate": 9.714150302084655e-06, "loss": 0.4153, "step": 3904 }, { "epoch": 0.26996197718631176, "grad_norm": 0.871568960301109, "learning_rate": 9.713963719417377e-06, "loss": 0.3983, "step": 3905 }, { "epoch": 0.27003110957483584, "grad_norm": 0.7926673637525081, "learning_rate": 9.713777077668976e-06, "loss": 0.3247, "step": 3906 }, { "epoch": 0.27010024196335986, "grad_norm": 0.8111154427623849, "learning_rate": 9.713590376841791e-06, "loss": 0.3276, "step": 3907 }, { "epoch": 0.2701693743518839, "grad_norm": 0.7903068437954056, "learning_rate": 9.713403616938164e-06, "loss": 0.4864, "step": 3908 }, { "epoch": 0.2702385067404079, "grad_norm": 0.8619270362890105, "learning_rate": 9.713216797960431e-06, "loss": 0.3737, "step": 3909 }, { "epoch": 0.2703076391289319, "grad_norm": 0.8222288516334408, "learning_rate": 9.71302991991094e-06, "loss": 0.4041, "step": 3910 }, { "epoch": 0.2703767715174559, "grad_norm": 1.1365781737563387, "learning_rate": 9.712842982792028e-06, "loss": 0.3868, "step": 3911 }, { "epoch": 0.27044590390597995, "grad_norm": 0.838578228901226, "learning_rate": 9.71265598660604e-06, "loss": 0.3979, "step": 3912 }, { "epoch": 0.27051503629450396, "grad_norm": 0.7239958575095232, "learning_rate": 9.712468931355319e-06, "loss": 0.3604, "step": 3913 }, { "epoch": 0.270584168683028, "grad_norm": 0.932714999946707, "learning_rate": 9.71228181704221e-06, "loss": 0.3497, "step": 3914 }, { "epoch": 0.270653301071552, "grad_norm": 0.9897274617002045, "learning_rate": 9.712094643669058e-06, "loss": 0.4939, "step": 3915 }, { "epoch": 0.2707224334600761, "grad_norm": 0.7584822488446873, "learning_rate": 9.711907411238208e-06, "loss": 0.3483, "step": 3916 }, { "epoch": 0.2707915658486001, "grad_norm": 0.7662957537392329, "learning_rate": 9.711720119752009e-06, "loss": 0.4089, "step": 3917 }, { "epoch": 0.2708606982371241, "grad_norm": 0.924979609187734, "learning_rate": 9.711532769212804e-06, "loss": 0.3472, "step": 3918 }, { "epoch": 0.2709298306256481, "grad_norm": 0.9068738592758595, "learning_rate": 9.711345359622945e-06, "loss": 0.5115, "step": 3919 }, { "epoch": 0.27099896301417215, "grad_norm": 0.7410703186100329, "learning_rate": 9.71115789098478e-06, "loss": 0.2529, "step": 3920 }, { "epoch": 0.27106809540269616, "grad_norm": 0.7148458882643418, "learning_rate": 9.71097036330066e-06, "loss": 0.3642, "step": 3921 }, { "epoch": 0.2711372277912202, "grad_norm": 0.8075412393080749, "learning_rate": 9.710782776572933e-06, "loss": 0.4368, "step": 3922 }, { "epoch": 0.2712063601797442, "grad_norm": 1.5155637816141942, "learning_rate": 9.710595130803948e-06, "loss": 0.4008, "step": 3923 }, { "epoch": 0.2712754925682682, "grad_norm": 0.788095251065274, "learning_rate": 9.710407425996063e-06, "loss": 0.3363, "step": 3924 }, { "epoch": 0.27134462495679224, "grad_norm": 0.8522188687102479, "learning_rate": 9.710219662151624e-06, "loss": 0.4035, "step": 3925 }, { "epoch": 0.27141375734531625, "grad_norm": 0.7655013853541525, "learning_rate": 9.710031839272989e-06, "loss": 0.4433, "step": 3926 }, { "epoch": 0.2714828897338403, "grad_norm": 0.7514694279012656, "learning_rate": 9.70984395736251e-06, "loss": 0.3263, "step": 3927 }, { "epoch": 0.27155202212236434, "grad_norm": 1.0237893518376013, "learning_rate": 9.709656016422542e-06, "loss": 0.3381, "step": 3928 }, { "epoch": 0.27162115451088836, "grad_norm": 3.4469247045432696, "learning_rate": 9.70946801645544e-06, "loss": 0.4228, "step": 3929 }, { "epoch": 0.2716902868994124, "grad_norm": 2.6024406548522374, "learning_rate": 9.709279957463563e-06, "loss": 0.3832, "step": 3930 }, { "epoch": 0.2717594192879364, "grad_norm": 16.339588059232142, "learning_rate": 9.709091839449263e-06, "loss": 0.3911, "step": 3931 }, { "epoch": 0.2718285516764604, "grad_norm": 7.790951594657908, "learning_rate": 9.7089036624149e-06, "loss": 0.4199, "step": 3932 }, { "epoch": 0.27189768406498444, "grad_norm": 18.133566034771327, "learning_rate": 9.708715426362836e-06, "loss": 0.5466, "step": 3933 }, { "epoch": 0.27196681645350845, "grad_norm": 3.476992759360596, "learning_rate": 9.708527131295427e-06, "loss": 0.3783, "step": 3934 }, { "epoch": 0.27203594884203247, "grad_norm": 8.764615391448402, "learning_rate": 9.708338777215032e-06, "loss": 0.3744, "step": 3935 }, { "epoch": 0.2721050812305565, "grad_norm": 9.162501239929695, "learning_rate": 9.708150364124012e-06, "loss": 0.4255, "step": 3936 }, { "epoch": 0.27217421361908056, "grad_norm": 2.021784921340199, "learning_rate": 9.707961892024731e-06, "loss": 0.3737, "step": 3937 }, { "epoch": 0.2722433460076046, "grad_norm": 7.136249364848225, "learning_rate": 9.707773360919548e-06, "loss": 0.3335, "step": 3938 }, { "epoch": 0.2723124783961286, "grad_norm": 1.0521718636518196, "learning_rate": 9.707584770810827e-06, "loss": 0.3458, "step": 3939 }, { "epoch": 0.2723816107846526, "grad_norm": 1.0224479995487046, "learning_rate": 9.707396121700934e-06, "loss": 0.4564, "step": 3940 }, { "epoch": 0.27245074317317663, "grad_norm": 0.8013124852227759, "learning_rate": 9.70720741359223e-06, "loss": 0.3886, "step": 3941 }, { "epoch": 0.27251987556170065, "grad_norm": 1.120825187859804, "learning_rate": 9.70701864648708e-06, "loss": 0.3187, "step": 3942 }, { "epoch": 0.27258900795022467, "grad_norm": 1.138919900250676, "learning_rate": 9.706829820387855e-06, "loss": 0.3779, "step": 3943 }, { "epoch": 0.2726581403387487, "grad_norm": 0.8506856064830651, "learning_rate": 9.706640935296917e-06, "loss": 0.4435, "step": 3944 }, { "epoch": 0.2727272727272727, "grad_norm": 1.3149641686634694, "learning_rate": 9.706451991216632e-06, "loss": 0.3126, "step": 3945 }, { "epoch": 0.2727964051157967, "grad_norm": 0.9350056823928821, "learning_rate": 9.706262988149371e-06, "loss": 0.3412, "step": 3946 }, { "epoch": 0.2728655375043208, "grad_norm": 1.1083527917011655, "learning_rate": 9.706073926097503e-06, "loss": 0.4601, "step": 3947 }, { "epoch": 0.2729346698928448, "grad_norm": 0.9602998369858321, "learning_rate": 9.705884805063396e-06, "loss": 0.4112, "step": 3948 }, { "epoch": 0.27300380228136883, "grad_norm": 0.7896681730047487, "learning_rate": 9.70569562504942e-06, "loss": 0.3265, "step": 3949 }, { "epoch": 0.27307293466989285, "grad_norm": 0.985661509501759, "learning_rate": 9.70550638605795e-06, "loss": 0.3603, "step": 3950 }, { "epoch": 0.27314206705841687, "grad_norm": 1.0165894442653471, "learning_rate": 9.705317088091354e-06, "loss": 0.3932, "step": 3951 }, { "epoch": 0.2732111994469409, "grad_norm": 0.8464772495141362, "learning_rate": 9.705127731152003e-06, "loss": 0.3227, "step": 3952 }, { "epoch": 0.2732803318354649, "grad_norm": 0.755696869128254, "learning_rate": 9.704938315242275e-06, "loss": 0.372, "step": 3953 }, { "epoch": 0.2733494642239889, "grad_norm": 0.8031050182552465, "learning_rate": 9.70474884036454e-06, "loss": 0.4452, "step": 3954 }, { "epoch": 0.27341859661251294, "grad_norm": 0.9778147878836211, "learning_rate": 9.704559306521176e-06, "loss": 0.4341, "step": 3955 }, { "epoch": 0.27348772900103696, "grad_norm": 0.8582972089563307, "learning_rate": 9.704369713714554e-06, "loss": 0.3315, "step": 3956 }, { "epoch": 0.27355686138956103, "grad_norm": 0.9533115431939342, "learning_rate": 9.704180061947055e-06, "loss": 0.3664, "step": 3957 }, { "epoch": 0.27362599377808505, "grad_norm": 0.9990580468247159, "learning_rate": 9.703990351221053e-06, "loss": 0.471, "step": 3958 }, { "epoch": 0.27369512616660907, "grad_norm": 0.8371644617524873, "learning_rate": 9.703800581538929e-06, "loss": 0.3586, "step": 3959 }, { "epoch": 0.2737642585551331, "grad_norm": 0.7850596926018616, "learning_rate": 9.703610752903057e-06, "loss": 0.3698, "step": 3960 }, { "epoch": 0.2738333909436571, "grad_norm": 0.8358935700478303, "learning_rate": 9.703420865315819e-06, "loss": 0.4192, "step": 3961 }, { "epoch": 0.2739025233321811, "grad_norm": 0.7564653349386696, "learning_rate": 9.703230918779592e-06, "loss": 0.374, "step": 3962 }, { "epoch": 0.27397165572070514, "grad_norm": 0.8334378970566954, "learning_rate": 9.70304091329676e-06, "loss": 0.3199, "step": 3963 }, { "epoch": 0.27404078810922916, "grad_norm": 0.9233156479688326, "learning_rate": 9.702850848869702e-06, "loss": 0.333, "step": 3964 }, { "epoch": 0.2741099204977532, "grad_norm": 0.8392038935975186, "learning_rate": 9.702660725500804e-06, "loss": 0.4205, "step": 3965 }, { "epoch": 0.2741790528862772, "grad_norm": 0.8063772569054839, "learning_rate": 9.702470543192444e-06, "loss": 0.2945, "step": 3966 }, { "epoch": 0.27424818527480127, "grad_norm": 0.8803012125246387, "learning_rate": 9.702280301947009e-06, "loss": 0.3125, "step": 3967 }, { "epoch": 0.2743173176633253, "grad_norm": 0.9738412430906003, "learning_rate": 9.702090001766881e-06, "loss": 0.3997, "step": 3968 }, { "epoch": 0.2743864500518493, "grad_norm": 0.8918914352933685, "learning_rate": 9.701899642654446e-06, "loss": 0.4566, "step": 3969 }, { "epoch": 0.2744555824403733, "grad_norm": 1.2399837434652858, "learning_rate": 9.70170922461209e-06, "loss": 0.2822, "step": 3970 }, { "epoch": 0.27452471482889734, "grad_norm": 0.8830046424928871, "learning_rate": 9.7015187476422e-06, "loss": 0.305, "step": 3971 }, { "epoch": 0.27459384721742136, "grad_norm": 0.8440796170391771, "learning_rate": 9.701328211747162e-06, "loss": 0.4009, "step": 3972 }, { "epoch": 0.2746629796059454, "grad_norm": 0.7752967602553906, "learning_rate": 9.701137616929366e-06, "loss": 0.3471, "step": 3973 }, { "epoch": 0.2747321119944694, "grad_norm": 0.8293418882943625, "learning_rate": 9.700946963191198e-06, "loss": 0.3615, "step": 3974 }, { "epoch": 0.2748012443829934, "grad_norm": 0.7809217071205508, "learning_rate": 9.700756250535051e-06, "loss": 0.4291, "step": 3975 }, { "epoch": 0.27487037677151743, "grad_norm": 1.0992338120020035, "learning_rate": 9.70056547896331e-06, "loss": 0.5121, "step": 3976 }, { "epoch": 0.2749395091600415, "grad_norm": 0.8430205084139035, "learning_rate": 9.700374648478372e-06, "loss": 0.3125, "step": 3977 }, { "epoch": 0.2750086415485655, "grad_norm": 0.9254181605705049, "learning_rate": 9.700183759082625e-06, "loss": 0.2732, "step": 3978 }, { "epoch": 0.27507777393708954, "grad_norm": 1.298669314541705, "learning_rate": 9.699992810778462e-06, "loss": 0.408, "step": 3979 }, { "epoch": 0.27514690632561356, "grad_norm": 0.9957246383051526, "learning_rate": 9.69980180356828e-06, "loss": 0.407, "step": 3980 }, { "epoch": 0.2752160387141376, "grad_norm": 0.8515252193481541, "learning_rate": 9.699610737454465e-06, "loss": 0.3675, "step": 3981 }, { "epoch": 0.2752851711026616, "grad_norm": 0.9492240972159857, "learning_rate": 9.699419612439418e-06, "loss": 0.3165, "step": 3982 }, { "epoch": 0.2753543034911856, "grad_norm": 0.7746226369692271, "learning_rate": 9.699228428525534e-06, "loss": 0.4262, "step": 3983 }, { "epoch": 0.27542343587970963, "grad_norm": 0.7605179411520337, "learning_rate": 9.699037185715205e-06, "loss": 0.3049, "step": 3984 }, { "epoch": 0.27549256826823365, "grad_norm": 0.8738442281397552, "learning_rate": 9.698845884010832e-06, "loss": 0.3135, "step": 3985 }, { "epoch": 0.27556170065675767, "grad_norm": 0.9161581442818466, "learning_rate": 9.698654523414813e-06, "loss": 0.4377, "step": 3986 }, { "epoch": 0.2756308330452817, "grad_norm": 2.1855722296933453, "learning_rate": 9.698463103929542e-06, "loss": 0.3834, "step": 3987 }, { "epoch": 0.27569996543380576, "grad_norm": 0.8501177101375195, "learning_rate": 9.698271625557422e-06, "loss": 0.3306, "step": 3988 }, { "epoch": 0.2757690978223298, "grad_norm": 2.104925379108187, "learning_rate": 9.698080088300852e-06, "loss": 0.3138, "step": 3989 }, { "epoch": 0.2758382302108538, "grad_norm": 0.7411643672046427, "learning_rate": 9.697888492162232e-06, "loss": 0.4443, "step": 3990 }, { "epoch": 0.2759073625993778, "grad_norm": 0.8691570151979426, "learning_rate": 9.697696837143964e-06, "loss": 0.402, "step": 3991 }, { "epoch": 0.27597649498790183, "grad_norm": 1.1970475277609545, "learning_rate": 9.697505123248449e-06, "loss": 0.2898, "step": 3992 }, { "epoch": 0.27604562737642585, "grad_norm": 0.8758732174068193, "learning_rate": 9.69731335047809e-06, "loss": 0.3414, "step": 3993 }, { "epoch": 0.27611475976494987, "grad_norm": 0.9677969282283332, "learning_rate": 9.69712151883529e-06, "loss": 0.4282, "step": 3994 }, { "epoch": 0.2761838921534739, "grad_norm": 1.183610855075553, "learning_rate": 9.696929628322455e-06, "loss": 0.36, "step": 3995 }, { "epoch": 0.2762530245419979, "grad_norm": 1.1308713490783089, "learning_rate": 9.696737678941989e-06, "loss": 0.336, "step": 3996 }, { "epoch": 0.2763221569305219, "grad_norm": 0.8681540959781241, "learning_rate": 9.6965456706963e-06, "loss": 0.4307, "step": 3997 }, { "epoch": 0.276391289319046, "grad_norm": 0.9323949043203797, "learning_rate": 9.69635360358779e-06, "loss": 0.3587, "step": 3998 }, { "epoch": 0.27646042170757, "grad_norm": 0.7530363573204698, "learning_rate": 9.696161477618869e-06, "loss": 0.4347, "step": 3999 }, { "epoch": 0.27652955409609403, "grad_norm": 1.0315490297949912, "learning_rate": 9.695969292791946e-06, "loss": 0.3483, "step": 4000 }, { "epoch": 0.27659868648461805, "grad_norm": 0.7857215469242885, "learning_rate": 9.695777049109427e-06, "loss": 0.4619, "step": 4001 }, { "epoch": 0.27666781887314207, "grad_norm": 0.9281065016248637, "learning_rate": 9.695584746573724e-06, "loss": 0.3452, "step": 4002 }, { "epoch": 0.2767369512616661, "grad_norm": 0.8731049059012164, "learning_rate": 9.695392385187244e-06, "loss": 0.3594, "step": 4003 }, { "epoch": 0.2768060836501901, "grad_norm": 0.8135264288715084, "learning_rate": 9.695199964952399e-06, "loss": 0.3889, "step": 4004 }, { "epoch": 0.2768752160387141, "grad_norm": 0.8059133246966269, "learning_rate": 9.695007485871604e-06, "loss": 0.3937, "step": 4005 }, { "epoch": 0.27694434842723814, "grad_norm": 0.7594768495054182, "learning_rate": 9.694814947947266e-06, "loss": 0.3144, "step": 4006 }, { "epoch": 0.27701348081576216, "grad_norm": 0.8816228367263541, "learning_rate": 9.694622351181804e-06, "loss": 0.3469, "step": 4007 }, { "epoch": 0.27708261320428623, "grad_norm": 0.9886450848165625, "learning_rate": 9.694429695577628e-06, "loss": 0.4377, "step": 4008 }, { "epoch": 0.27715174559281025, "grad_norm": 0.9436386950923458, "learning_rate": 9.694236981137151e-06, "loss": 0.3452, "step": 4009 }, { "epoch": 0.27722087798133427, "grad_norm": 0.8716143113689319, "learning_rate": 9.694044207862791e-06, "loss": 0.328, "step": 4010 }, { "epoch": 0.2772900103698583, "grad_norm": 0.8737171634976784, "learning_rate": 9.693851375756966e-06, "loss": 0.4073, "step": 4011 }, { "epoch": 0.2773591427583823, "grad_norm": 1.282778043756337, "learning_rate": 9.693658484822091e-06, "loss": 0.4493, "step": 4012 }, { "epoch": 0.2774282751469063, "grad_norm": 0.952848314819436, "learning_rate": 9.69346553506058e-06, "loss": 0.3525, "step": 4013 }, { "epoch": 0.27749740753543034, "grad_norm": 1.3219773078988302, "learning_rate": 9.693272526474857e-06, "loss": 0.3389, "step": 4014 }, { "epoch": 0.27756653992395436, "grad_norm": 0.8724547141587581, "learning_rate": 9.693079459067337e-06, "loss": 0.4677, "step": 4015 }, { "epoch": 0.2776356723124784, "grad_norm": 0.7795794719800282, "learning_rate": 9.692886332840441e-06, "loss": 0.3156, "step": 4016 }, { "epoch": 0.2777048047010024, "grad_norm": 0.7858730210002451, "learning_rate": 9.69269314779659e-06, "loss": 0.3623, "step": 4017 }, { "epoch": 0.27777393708952647, "grad_norm": 0.9578011801306865, "learning_rate": 9.692499903938206e-06, "loss": 0.3705, "step": 4018 }, { "epoch": 0.2778430694780505, "grad_norm": 0.8806429909132255, "learning_rate": 9.69230660126771e-06, "loss": 0.4404, "step": 4019 }, { "epoch": 0.2779122018665745, "grad_norm": 0.820453595008792, "learning_rate": 9.692113239787523e-06, "loss": 0.367, "step": 4020 }, { "epoch": 0.2779813342550985, "grad_norm": 0.7631786413552711, "learning_rate": 9.69191981950007e-06, "loss": 0.2893, "step": 4021 }, { "epoch": 0.27805046664362254, "grad_norm": 0.9366778492942128, "learning_rate": 9.691726340407776e-06, "loss": 0.4128, "step": 4022 }, { "epoch": 0.27811959903214656, "grad_norm": 0.8375628675470367, "learning_rate": 9.691532802513064e-06, "loss": 0.4405, "step": 4023 }, { "epoch": 0.2781887314206706, "grad_norm": 0.750963093683633, "learning_rate": 9.691339205818362e-06, "loss": 0.3082, "step": 4024 }, { "epoch": 0.2782578638091946, "grad_norm": 0.7856878547973103, "learning_rate": 9.691145550326096e-06, "loss": 0.3888, "step": 4025 }, { "epoch": 0.2783269961977186, "grad_norm": 0.9522142931154609, "learning_rate": 9.69095183603869e-06, "loss": 0.4464, "step": 4026 }, { "epoch": 0.27839612858624263, "grad_norm": 0.7147881094885935, "learning_rate": 9.690758062958575e-06, "loss": 0.3441, "step": 4027 }, { "epoch": 0.2784652609747667, "grad_norm": 0.8338094430553348, "learning_rate": 9.690564231088179e-06, "loss": 0.3765, "step": 4028 }, { "epoch": 0.2785343933632907, "grad_norm": 0.7514792133466289, "learning_rate": 9.69037034042993e-06, "loss": 0.394, "step": 4029 }, { "epoch": 0.27860352575181474, "grad_norm": 0.8149589171895524, "learning_rate": 9.690176390986262e-06, "loss": 0.4156, "step": 4030 }, { "epoch": 0.27867265814033876, "grad_norm": 0.8533963635116524, "learning_rate": 9.689982382759599e-06, "loss": 0.3733, "step": 4031 }, { "epoch": 0.2787417905288628, "grad_norm": 0.9098760983074827, "learning_rate": 9.689788315752379e-06, "loss": 0.329, "step": 4032 }, { "epoch": 0.2788109229173868, "grad_norm": 0.8952308970512729, "learning_rate": 9.689594189967032e-06, "loss": 0.3904, "step": 4033 }, { "epoch": 0.2788800553059108, "grad_norm": 0.7995478041114498, "learning_rate": 9.689400005405989e-06, "loss": 0.3942, "step": 4034 }, { "epoch": 0.27894918769443483, "grad_norm": 0.7945458774859497, "learning_rate": 9.689205762071686e-06, "loss": 0.3077, "step": 4035 }, { "epoch": 0.27901832008295885, "grad_norm": 0.8295504744955853, "learning_rate": 9.689011459966558e-06, "loss": 0.3451, "step": 4036 }, { "epoch": 0.27908745247148287, "grad_norm": 0.7637664537624241, "learning_rate": 9.688817099093039e-06, "loss": 0.4107, "step": 4037 }, { "epoch": 0.27915658486000694, "grad_norm": 1.4577984309138472, "learning_rate": 9.688622679453565e-06, "loss": 0.2951, "step": 4038 }, { "epoch": 0.27922571724853096, "grad_norm": 0.769090447212773, "learning_rate": 9.688428201050573e-06, "loss": 0.2942, "step": 4039 }, { "epoch": 0.279294849637055, "grad_norm": 0.8058064562334049, "learning_rate": 9.6882336638865e-06, "loss": 0.4018, "step": 4040 }, { "epoch": 0.279363982025579, "grad_norm": 0.7905375454426468, "learning_rate": 9.688039067963784e-06, "loss": 0.274, "step": 4041 }, { "epoch": 0.279433114414103, "grad_norm": 0.6801719771842734, "learning_rate": 9.687844413284866e-06, "loss": 0.2835, "step": 4042 }, { "epoch": 0.27950224680262703, "grad_norm": 0.7814996528099556, "learning_rate": 9.687649699852183e-06, "loss": 0.3712, "step": 4043 }, { "epoch": 0.27957137919115105, "grad_norm": 0.7876233421907995, "learning_rate": 9.687454927668176e-06, "loss": 0.423, "step": 4044 }, { "epoch": 0.27964051157967507, "grad_norm": 0.7939380699510898, "learning_rate": 9.68726009673529e-06, "loss": 0.3312, "step": 4045 }, { "epoch": 0.2797096439681991, "grad_norm": 0.8515246279196936, "learning_rate": 9.687065207055959e-06, "loss": 0.3715, "step": 4046 }, { "epoch": 0.2797787763567231, "grad_norm": 0.8120707808151573, "learning_rate": 9.686870258632632e-06, "loss": 0.488, "step": 4047 }, { "epoch": 0.2798479087452472, "grad_norm": 0.8160467028915692, "learning_rate": 9.686675251467751e-06, "loss": 0.3988, "step": 4048 }, { "epoch": 0.2799170411337712, "grad_norm": 0.8516828355294175, "learning_rate": 9.68648018556376e-06, "loss": 0.3963, "step": 4049 }, { "epoch": 0.2799861735222952, "grad_norm": 0.8800184510089115, "learning_rate": 9.686285060923102e-06, "loss": 0.3103, "step": 4050 }, { "epoch": 0.28005530591081923, "grad_norm": 0.8445051371993303, "learning_rate": 9.686089877548224e-06, "loss": 0.5233, "step": 4051 }, { "epoch": 0.28012443829934325, "grad_norm": 0.7700671055638455, "learning_rate": 9.685894635441571e-06, "loss": 0.3174, "step": 4052 }, { "epoch": 0.28019357068786727, "grad_norm": 0.9891073204941553, "learning_rate": 9.685699334605592e-06, "loss": 0.3348, "step": 4053 }, { "epoch": 0.2802627030763913, "grad_norm": 0.7769395085865409, "learning_rate": 9.685503975042735e-06, "loss": 0.4098, "step": 4054 }, { "epoch": 0.2803318354649153, "grad_norm": 1.0125848334826906, "learning_rate": 9.685308556755446e-06, "loss": 0.3893, "step": 4055 }, { "epoch": 0.2804009678534393, "grad_norm": 0.718016233572319, "learning_rate": 9.685113079746175e-06, "loss": 0.324, "step": 4056 }, { "epoch": 0.28047010024196334, "grad_norm": 0.7590651014545026, "learning_rate": 9.684917544017374e-06, "loss": 0.3178, "step": 4057 }, { "epoch": 0.28053923263048736, "grad_norm": 0.8107333736658698, "learning_rate": 9.684721949571491e-06, "loss": 0.468, "step": 4058 }, { "epoch": 0.28060836501901143, "grad_norm": 0.8547217575695607, "learning_rate": 9.68452629641098e-06, "loss": 0.3236, "step": 4059 }, { "epoch": 0.28067749740753545, "grad_norm": 0.8558512619348481, "learning_rate": 9.684330584538291e-06, "loss": 0.2967, "step": 4060 }, { "epoch": 0.28074662979605947, "grad_norm": 0.8834340984018838, "learning_rate": 9.68413481395588e-06, "loss": 0.4624, "step": 4061 }, { "epoch": 0.2808157621845835, "grad_norm": 0.8209636691546451, "learning_rate": 9.683938984666195e-06, "loss": 0.4175, "step": 4062 }, { "epoch": 0.2808848945731075, "grad_norm": 0.8803759252492248, "learning_rate": 9.683743096671694e-06, "loss": 0.3427, "step": 4063 }, { "epoch": 0.2809540269616315, "grad_norm": 0.8949368752278802, "learning_rate": 9.683547149974832e-06, "loss": 0.3838, "step": 4064 }, { "epoch": 0.28102315935015554, "grad_norm": 0.8234415932485584, "learning_rate": 9.683351144578065e-06, "loss": 0.3955, "step": 4065 }, { "epoch": 0.28109229173867956, "grad_norm": 0.9621301662542058, "learning_rate": 9.68315508048385e-06, "loss": 0.3571, "step": 4066 }, { "epoch": 0.2811614241272036, "grad_norm": 0.7773039190360075, "learning_rate": 9.682958957694644e-06, "loss": 0.3757, "step": 4067 }, { "epoch": 0.2812305565157276, "grad_norm": 1.0057171354113894, "learning_rate": 9.682762776212903e-06, "loss": 0.3612, "step": 4068 }, { "epoch": 0.28129968890425167, "grad_norm": 0.763984477010353, "learning_rate": 9.682566536041087e-06, "loss": 0.4096, "step": 4069 }, { "epoch": 0.2813688212927757, "grad_norm": 0.8423933207366835, "learning_rate": 9.682370237181658e-06, "loss": 0.3124, "step": 4070 }, { "epoch": 0.2814379536812997, "grad_norm": 0.7837763998982995, "learning_rate": 9.682173879637073e-06, "loss": 0.3179, "step": 4071 }, { "epoch": 0.2815070860698237, "grad_norm": 0.7953771763035392, "learning_rate": 9.681977463409793e-06, "loss": 0.4601, "step": 4072 }, { "epoch": 0.28157621845834774, "grad_norm": 0.7955073305141777, "learning_rate": 9.681780988502282e-06, "loss": 0.339, "step": 4073 }, { "epoch": 0.28164535084687176, "grad_norm": 0.7946300294529848, "learning_rate": 9.681584454917002e-06, "loss": 0.3328, "step": 4074 }, { "epoch": 0.2817144832353958, "grad_norm": 0.8938277899366472, "learning_rate": 9.681387862656414e-06, "loss": 0.4083, "step": 4075 }, { "epoch": 0.2817836156239198, "grad_norm": 0.9485089103288895, "learning_rate": 9.681191211722985e-06, "loss": 0.3952, "step": 4076 }, { "epoch": 0.2818527480124438, "grad_norm": 0.7297827107431879, "learning_rate": 9.680994502119176e-06, "loss": 0.3125, "step": 4077 }, { "epoch": 0.28192188040096783, "grad_norm": 0.7401686544454495, "learning_rate": 9.680797733847457e-06, "loss": 0.2924, "step": 4078 }, { "epoch": 0.2819910127894919, "grad_norm": 0.7989646467967088, "learning_rate": 9.680600906910291e-06, "loss": 0.4587, "step": 4079 }, { "epoch": 0.2820601451780159, "grad_norm": 0.8350210501413671, "learning_rate": 9.680404021310143e-06, "loss": 0.3849, "step": 4080 }, { "epoch": 0.28212927756653994, "grad_norm": 0.7922822302678328, "learning_rate": 9.680207077049486e-06, "loss": 0.3186, "step": 4081 }, { "epoch": 0.28219840995506396, "grad_norm": 1.3403586688040487, "learning_rate": 9.680010074130785e-06, "loss": 0.3957, "step": 4082 }, { "epoch": 0.282267542343588, "grad_norm": 0.8694815856906822, "learning_rate": 9.679813012556509e-06, "loss": 0.4739, "step": 4083 }, { "epoch": 0.282336674732112, "grad_norm": 0.8937418273182735, "learning_rate": 9.679615892329129e-06, "loss": 0.4005, "step": 4084 }, { "epoch": 0.282405807120636, "grad_norm": 0.8225036080654001, "learning_rate": 9.679418713451115e-06, "loss": 0.3234, "step": 4085 }, { "epoch": 0.28247493950916003, "grad_norm": 1.2325723486521385, "learning_rate": 9.679221475924938e-06, "loss": 0.4114, "step": 4086 }, { "epoch": 0.28254407189768405, "grad_norm": 0.7958208510790858, "learning_rate": 9.679024179753069e-06, "loss": 0.4053, "step": 4087 }, { "epoch": 0.28261320428620806, "grad_norm": 0.82843338532745, "learning_rate": 9.678826824937983e-06, "loss": 0.2772, "step": 4088 }, { "epoch": 0.28268233667473214, "grad_norm": 0.7964250003435096, "learning_rate": 9.678629411482153e-06, "loss": 0.3516, "step": 4089 }, { "epoch": 0.28275146906325616, "grad_norm": 0.7945012618983407, "learning_rate": 9.678431939388053e-06, "loss": 0.4589, "step": 4090 }, { "epoch": 0.2828206014517802, "grad_norm": 0.8116958060582701, "learning_rate": 9.678234408658157e-06, "loss": 0.4658, "step": 4091 }, { "epoch": 0.2828897338403042, "grad_norm": 0.8765849635028452, "learning_rate": 9.67803681929494e-06, "loss": 0.3233, "step": 4092 }, { "epoch": 0.2829588662288282, "grad_norm": 0.7168842881062805, "learning_rate": 9.677839171300883e-06, "loss": 0.3577, "step": 4093 }, { "epoch": 0.2830279986173522, "grad_norm": 0.8714603573308899, "learning_rate": 9.677641464678457e-06, "loss": 0.4556, "step": 4094 }, { "epoch": 0.28309713100587625, "grad_norm": 0.7258247543519091, "learning_rate": 9.677443699430145e-06, "loss": 0.2994, "step": 4095 }, { "epoch": 0.28316626339440026, "grad_norm": 0.7940835037679765, "learning_rate": 9.67724587555842e-06, "loss": 0.32, "step": 4096 }, { "epoch": 0.2832353957829243, "grad_norm": 0.8658815565955293, "learning_rate": 9.677047993065768e-06, "loss": 0.4242, "step": 4097 }, { "epoch": 0.2833045281714483, "grad_norm": 1.5762700206660878, "learning_rate": 9.676850051954665e-06, "loss": 0.3943, "step": 4098 }, { "epoch": 0.2833736605599724, "grad_norm": 0.7152562647666876, "learning_rate": 9.676652052227595e-06, "loss": 0.2776, "step": 4099 }, { "epoch": 0.2834427929484964, "grad_norm": 1.1276698286085027, "learning_rate": 9.676453993887035e-06, "loss": 0.3796, "step": 4100 }, { "epoch": 0.2835119253370204, "grad_norm": 0.8271882168276502, "learning_rate": 9.67625587693547e-06, "loss": 0.4706, "step": 4101 }, { "epoch": 0.2835810577255444, "grad_norm": 0.9541897702257409, "learning_rate": 9.676057701375382e-06, "loss": 0.2988, "step": 4102 }, { "epoch": 0.28365019011406845, "grad_norm": 0.7987513391549705, "learning_rate": 9.675859467209258e-06, "loss": 0.3442, "step": 4103 }, { "epoch": 0.28371932250259246, "grad_norm": 0.8490695852307141, "learning_rate": 9.675661174439577e-06, "loss": 0.4678, "step": 4104 }, { "epoch": 0.2837884548911165, "grad_norm": 0.9734343844839575, "learning_rate": 9.67546282306883e-06, "loss": 0.4523, "step": 4105 }, { "epoch": 0.2838575872796405, "grad_norm": 0.7240156551326289, "learning_rate": 9.675264413099498e-06, "loss": 0.2961, "step": 4106 }, { "epoch": 0.2839267196681645, "grad_norm": 0.8695028797777448, "learning_rate": 9.675065944534072e-06, "loss": 0.4184, "step": 4107 }, { "epoch": 0.28399585205668854, "grad_norm": 0.8821001768891665, "learning_rate": 9.674867417375035e-06, "loss": 0.4107, "step": 4108 }, { "epoch": 0.2840649844452126, "grad_norm": 0.8748024314663048, "learning_rate": 9.67466883162488e-06, "loss": 0.3089, "step": 4109 }, { "epoch": 0.2841341168337366, "grad_norm": 0.7210235469914436, "learning_rate": 9.674470187286093e-06, "loss": 0.251, "step": 4110 }, { "epoch": 0.28420324922226065, "grad_norm": 0.7682860605207267, "learning_rate": 9.674271484361163e-06, "loss": 0.3846, "step": 4111 }, { "epoch": 0.28427238161078466, "grad_norm": 0.7989384988019279, "learning_rate": 9.674072722852582e-06, "loss": 0.3826, "step": 4112 }, { "epoch": 0.2843415139993087, "grad_norm": 0.8425110186792113, "learning_rate": 9.673873902762841e-06, "loss": 0.3469, "step": 4113 }, { "epoch": 0.2844106463878327, "grad_norm": 0.7272078458439167, "learning_rate": 9.673675024094432e-06, "loss": 0.3003, "step": 4114 }, { "epoch": 0.2844797787763567, "grad_norm": 0.8348847620805144, "learning_rate": 9.673476086849846e-06, "loss": 0.4035, "step": 4115 }, { "epoch": 0.28454891116488074, "grad_norm": 1.5639704495652398, "learning_rate": 9.67327709103158e-06, "loss": 0.3479, "step": 4116 }, { "epoch": 0.28461804355340475, "grad_norm": 0.6994857580602922, "learning_rate": 9.673078036642122e-06, "loss": 0.3113, "step": 4117 }, { "epoch": 0.28468717594192877, "grad_norm": 1.0279057030128247, "learning_rate": 9.67287892368397e-06, "loss": 0.3412, "step": 4118 }, { "epoch": 0.2847563083304528, "grad_norm": 0.7536877552412572, "learning_rate": 9.672679752159623e-06, "loss": 0.4505, "step": 4119 }, { "epoch": 0.28482544071897686, "grad_norm": 0.7659617745248958, "learning_rate": 9.672480522071572e-06, "loss": 0.3704, "step": 4120 }, { "epoch": 0.2848945731075009, "grad_norm": 1.383150160313137, "learning_rate": 9.672281233422318e-06, "loss": 0.3082, "step": 4121 }, { "epoch": 0.2849637054960249, "grad_norm": 0.7483898546716558, "learning_rate": 9.672081886214355e-06, "loss": 0.3747, "step": 4122 }, { "epoch": 0.2850328378845489, "grad_norm": 0.7774162879803832, "learning_rate": 9.671882480450183e-06, "loss": 0.4556, "step": 4123 }, { "epoch": 0.28510197027307294, "grad_norm": 0.7628013135031423, "learning_rate": 9.671683016132303e-06, "loss": 0.375, "step": 4124 }, { "epoch": 0.28517110266159695, "grad_norm": 0.7612094047128611, "learning_rate": 9.671483493263214e-06, "loss": 0.3741, "step": 4125 }, { "epoch": 0.28524023505012097, "grad_norm": 0.7545329295461473, "learning_rate": 9.671283911845413e-06, "loss": 0.4419, "step": 4126 }, { "epoch": 0.285309367438645, "grad_norm": 0.7854525096943232, "learning_rate": 9.671084271881407e-06, "loss": 0.3398, "step": 4127 }, { "epoch": 0.285378499827169, "grad_norm": 0.8223663737634459, "learning_rate": 9.670884573373694e-06, "loss": 0.3038, "step": 4128 }, { "epoch": 0.285447632215693, "grad_norm": 0.7920478702317726, "learning_rate": 9.67068481632478e-06, "loss": 0.4542, "step": 4129 }, { "epoch": 0.2855167646042171, "grad_norm": 0.7776636065440322, "learning_rate": 9.670485000737167e-06, "loss": 0.4234, "step": 4130 }, { "epoch": 0.2855858969927411, "grad_norm": 0.7523675071408481, "learning_rate": 9.670285126613357e-06, "loss": 0.3052, "step": 4131 }, { "epoch": 0.28565502938126514, "grad_norm": 0.8695333454233989, "learning_rate": 9.670085193955859e-06, "loss": 0.2959, "step": 4132 }, { "epoch": 0.28572416176978915, "grad_norm": 0.7779295700962731, "learning_rate": 9.669885202767178e-06, "loss": 0.4641, "step": 4133 }, { "epoch": 0.28579329415831317, "grad_norm": 0.7210582517315978, "learning_rate": 9.66968515304982e-06, "loss": 0.357, "step": 4134 }, { "epoch": 0.2858624265468372, "grad_norm": 0.6948108722600009, "learning_rate": 9.66948504480629e-06, "loss": 0.3001, "step": 4135 }, { "epoch": 0.2859315589353612, "grad_norm": 0.7783815287930417, "learning_rate": 9.669284878039099e-06, "loss": 0.3711, "step": 4136 }, { "epoch": 0.2860006913238852, "grad_norm": 0.9400600624452634, "learning_rate": 9.669084652750755e-06, "loss": 0.4442, "step": 4137 }, { "epoch": 0.28606982371240924, "grad_norm": 0.8285431059777209, "learning_rate": 9.668884368943767e-06, "loss": 0.3805, "step": 4138 }, { "epoch": 0.28613895610093326, "grad_norm": 0.7641610172180779, "learning_rate": 9.668684026620645e-06, "loss": 0.3572, "step": 4139 }, { "epoch": 0.28620808848945734, "grad_norm": 0.8965786684411031, "learning_rate": 9.668483625783902e-06, "loss": 0.4544, "step": 4140 }, { "epoch": 0.28627722087798135, "grad_norm": 0.9275419177266448, "learning_rate": 9.668283166436045e-06, "loss": 0.365, "step": 4141 }, { "epoch": 0.28634635326650537, "grad_norm": 0.8884703190848047, "learning_rate": 9.668082648579593e-06, "loss": 0.3459, "step": 4142 }, { "epoch": 0.2864154856550294, "grad_norm": 1.159598581712493, "learning_rate": 9.667882072217053e-06, "loss": 0.3112, "step": 4143 }, { "epoch": 0.2864846180435534, "grad_norm": 0.8001987560756466, "learning_rate": 9.667681437350943e-06, "loss": 0.4589, "step": 4144 }, { "epoch": 0.2865537504320774, "grad_norm": 0.7519173749889069, "learning_rate": 9.667480743983775e-06, "loss": 0.3121, "step": 4145 }, { "epoch": 0.28662288282060144, "grad_norm": 0.9895434978720562, "learning_rate": 9.667279992118065e-06, "loss": 0.3302, "step": 4146 }, { "epoch": 0.28669201520912546, "grad_norm": 0.8394920775703921, "learning_rate": 9.667079181756332e-06, "loss": 0.4726, "step": 4147 }, { "epoch": 0.2867611475976495, "grad_norm": 0.9655971663993316, "learning_rate": 9.666878312901086e-06, "loss": 0.3108, "step": 4148 }, { "epoch": 0.2868302799861735, "grad_norm": 0.7627627564529629, "learning_rate": 9.666677385554853e-06, "loss": 0.3565, "step": 4149 }, { "epoch": 0.28689941237469757, "grad_norm": 0.9461448294935473, "learning_rate": 9.666476399720144e-06, "loss": 0.3238, "step": 4150 }, { "epoch": 0.2869685447632216, "grad_norm": 0.7518127309243177, "learning_rate": 9.666275355399484e-06, "loss": 0.4196, "step": 4151 }, { "epoch": 0.2870376771517456, "grad_norm": 0.8307880537356047, "learning_rate": 9.666074252595387e-06, "loss": 0.3542, "step": 4152 }, { "epoch": 0.2871068095402696, "grad_norm": 0.7095384271086065, "learning_rate": 9.665873091310378e-06, "loss": 0.3221, "step": 4153 }, { "epoch": 0.28717594192879364, "grad_norm": 0.8171548437947681, "learning_rate": 9.665671871546976e-06, "loss": 0.3689, "step": 4154 }, { "epoch": 0.28724507431731766, "grad_norm": 0.8162267813662555, "learning_rate": 9.665470593307701e-06, "loss": 0.3709, "step": 4155 }, { "epoch": 0.2873142067058417, "grad_norm": 0.864579406631212, "learning_rate": 9.665269256595081e-06, "loss": 0.3473, "step": 4156 }, { "epoch": 0.2873833390943657, "grad_norm": 0.7703335181138212, "learning_rate": 9.665067861411636e-06, "loss": 0.2949, "step": 4157 }, { "epoch": 0.2874524714828897, "grad_norm": 0.7975407156040769, "learning_rate": 9.66486640775989e-06, "loss": 0.4837, "step": 4158 }, { "epoch": 0.28752160387141373, "grad_norm": 0.7244749638343139, "learning_rate": 9.664664895642368e-06, "loss": 0.3334, "step": 4159 }, { "epoch": 0.2875907362599378, "grad_norm": 0.7608394534609785, "learning_rate": 9.664463325061596e-06, "loss": 0.3397, "step": 4160 }, { "epoch": 0.2876598686484618, "grad_norm": 0.7925154423534236, "learning_rate": 9.6642616960201e-06, "loss": 0.3752, "step": 4161 }, { "epoch": 0.28772900103698584, "grad_norm": 0.7714818897701317, "learning_rate": 9.66406000852041e-06, "loss": 0.441, "step": 4162 }, { "epoch": 0.28779813342550986, "grad_norm": 0.7473685082350595, "learning_rate": 9.66385826256505e-06, "loss": 0.3741, "step": 4163 }, { "epoch": 0.2878672658140339, "grad_norm": 0.8379171134119043, "learning_rate": 9.663656458156547e-06, "loss": 0.3042, "step": 4164 }, { "epoch": 0.2879363982025579, "grad_norm": 0.8328495734085098, "learning_rate": 9.663454595297436e-06, "loss": 0.4409, "step": 4165 }, { "epoch": 0.2880055305910819, "grad_norm": 0.76368951762892, "learning_rate": 9.663252673990242e-06, "loss": 0.3406, "step": 4166 }, { "epoch": 0.28807466297960593, "grad_norm": 0.7311885818783493, "learning_rate": 9.6630506942375e-06, "loss": 0.2891, "step": 4167 }, { "epoch": 0.28814379536812995, "grad_norm": 0.8767611144180081, "learning_rate": 9.662848656041737e-06, "loss": 0.4156, "step": 4168 }, { "epoch": 0.28821292775665397, "grad_norm": 0.7369120945476356, "learning_rate": 9.662646559405489e-06, "loss": 0.388, "step": 4169 }, { "epoch": 0.28828206014517804, "grad_norm": 0.9283577834175524, "learning_rate": 9.662444404331285e-06, "loss": 0.3324, "step": 4170 }, { "epoch": 0.28835119253370206, "grad_norm": 0.7785262248252912, "learning_rate": 9.662242190821663e-06, "loss": 0.2796, "step": 4171 }, { "epoch": 0.2884203249222261, "grad_norm": 0.7860886445639372, "learning_rate": 9.662039918879154e-06, "loss": 0.379, "step": 4172 }, { "epoch": 0.2884894573107501, "grad_norm": 0.9889169357404068, "learning_rate": 9.661837588506295e-06, "loss": 0.3411, "step": 4173 }, { "epoch": 0.2885585896992741, "grad_norm": 0.8397431342264035, "learning_rate": 9.661635199705622e-06, "loss": 0.2912, "step": 4174 }, { "epoch": 0.28862772208779813, "grad_norm": 0.7315256443513043, "learning_rate": 9.661432752479669e-06, "loss": 0.3619, "step": 4175 }, { "epoch": 0.28869685447632215, "grad_norm": 0.7637476367899895, "learning_rate": 9.661230246830976e-06, "loss": 0.4024, "step": 4176 }, { "epoch": 0.28876598686484617, "grad_norm": 0.8363155881566849, "learning_rate": 9.661027682762081e-06, "loss": 0.3502, "step": 4177 }, { "epoch": 0.2888351192533702, "grad_norm": 0.8404000506263588, "learning_rate": 9.660825060275521e-06, "loss": 0.3484, "step": 4178 }, { "epoch": 0.2889042516418942, "grad_norm": 0.7961623901305563, "learning_rate": 9.660622379373837e-06, "loss": 0.3844, "step": 4179 }, { "epoch": 0.2889733840304182, "grad_norm": 0.9976069975273342, "learning_rate": 9.66041964005957e-06, "loss": 0.4213, "step": 4180 }, { "epoch": 0.2890425164189423, "grad_norm": 0.9738995544390666, "learning_rate": 9.660216842335257e-06, "loss": 0.344, "step": 4181 }, { "epoch": 0.2891116488074663, "grad_norm": 0.7343116428359969, "learning_rate": 9.660013986203443e-06, "loss": 0.3813, "step": 4182 }, { "epoch": 0.28918078119599033, "grad_norm": 0.7518940376546875, "learning_rate": 9.659811071666672e-06, "loss": 0.4439, "step": 4183 }, { "epoch": 0.28924991358451435, "grad_norm": 0.8151281554344905, "learning_rate": 9.659608098727482e-06, "loss": 0.3541, "step": 4184 }, { "epoch": 0.28931904597303837, "grad_norm": 0.703087978086373, "learning_rate": 9.659405067388421e-06, "loss": 0.2839, "step": 4185 }, { "epoch": 0.2893881783615624, "grad_norm": 0.804067276370578, "learning_rate": 9.659201977652035e-06, "loss": 0.4342, "step": 4186 }, { "epoch": 0.2894573107500864, "grad_norm": 0.7995949630675103, "learning_rate": 9.658998829520864e-06, "loss": 0.5098, "step": 4187 }, { "epoch": 0.2895264431386104, "grad_norm": 0.7616296869751467, "learning_rate": 9.658795622997459e-06, "loss": 0.3396, "step": 4188 }, { "epoch": 0.28959557552713444, "grad_norm": 0.7697994969190157, "learning_rate": 9.658592358084362e-06, "loss": 0.3555, "step": 4189 }, { "epoch": 0.28966470791565846, "grad_norm": 0.7790209820488955, "learning_rate": 9.658389034784125e-06, "loss": 0.42, "step": 4190 }, { "epoch": 0.28973384030418253, "grad_norm": 0.7779516126103183, "learning_rate": 9.658185653099296e-06, "loss": 0.4183, "step": 4191 }, { "epoch": 0.28980297269270655, "grad_norm": 0.8241930362461026, "learning_rate": 9.657982213032422e-06, "loss": 0.3369, "step": 4192 }, { "epoch": 0.28987210508123057, "grad_norm": 0.8546529400533929, "learning_rate": 9.657778714586052e-06, "loss": 0.3803, "step": 4193 }, { "epoch": 0.2899412374697546, "grad_norm": 0.8160224948189868, "learning_rate": 9.65757515776274e-06, "loss": 0.4473, "step": 4194 }, { "epoch": 0.2900103698582786, "grad_norm": 0.7803936443668893, "learning_rate": 9.657371542565036e-06, "loss": 0.3057, "step": 4195 }, { "epoch": 0.2900795022468026, "grad_norm": 2.4921515302566077, "learning_rate": 9.657167868995488e-06, "loss": 0.3647, "step": 4196 }, { "epoch": 0.29014863463532664, "grad_norm": 0.8366101124299048, "learning_rate": 9.656964137056652e-06, "loss": 0.4953, "step": 4197 }, { "epoch": 0.29021776702385066, "grad_norm": 0.7672554682460848, "learning_rate": 9.656760346751085e-06, "loss": 0.3704, "step": 4198 }, { "epoch": 0.2902868994123747, "grad_norm": 0.7810451302353411, "learning_rate": 9.656556498081334e-06, "loss": 0.3028, "step": 4199 }, { "epoch": 0.2903560318008987, "grad_norm": 0.8241700513003226, "learning_rate": 9.656352591049958e-06, "loss": 0.4287, "step": 4200 }, { "epoch": 0.29042516418942277, "grad_norm": 0.7390077763087864, "learning_rate": 9.656148625659513e-06, "loss": 0.4416, "step": 4201 }, { "epoch": 0.2904942965779468, "grad_norm": 6.493575580855817, "learning_rate": 9.655944601912553e-06, "loss": 0.2641, "step": 4202 }, { "epoch": 0.2905634289664708, "grad_norm": 0.7998746186433523, "learning_rate": 9.655740519811637e-06, "loss": 0.3057, "step": 4203 }, { "epoch": 0.2906325613549948, "grad_norm": 1.2615089984340044, "learning_rate": 9.655536379359322e-06, "loss": 0.4264, "step": 4204 }, { "epoch": 0.29070169374351884, "grad_norm": 0.6962771490709186, "learning_rate": 9.655332180558168e-06, "loss": 0.342, "step": 4205 }, { "epoch": 0.29077082613204286, "grad_norm": 1.2237107308106383, "learning_rate": 9.65512792341073e-06, "loss": 0.3232, "step": 4206 }, { "epoch": 0.2908399585205669, "grad_norm": 0.7616400459577928, "learning_rate": 9.654923607919572e-06, "loss": 0.3623, "step": 4207 }, { "epoch": 0.2909090909090909, "grad_norm": 1.4855647294791974, "learning_rate": 9.654719234087255e-06, "loss": 0.4114, "step": 4208 }, { "epoch": 0.2909782232976149, "grad_norm": 1.3054064941704218, "learning_rate": 9.65451480191634e-06, "loss": 0.2997, "step": 4209 }, { "epoch": 0.29104735568613893, "grad_norm": 1.1686158263337147, "learning_rate": 9.654310311409385e-06, "loss": 0.31, "step": 4210 }, { "epoch": 0.291116488074663, "grad_norm": 0.8169939004459552, "learning_rate": 9.654105762568958e-06, "loss": 0.3198, "step": 4211 }, { "epoch": 0.291185620463187, "grad_norm": 0.9532178846168813, "learning_rate": 9.65390115539762e-06, "loss": 0.396, "step": 4212 }, { "epoch": 0.29125475285171104, "grad_norm": 0.9861400602538355, "learning_rate": 9.653696489897939e-06, "loss": 0.3146, "step": 4213 }, { "epoch": 0.29132388524023506, "grad_norm": 1.6884344506713012, "learning_rate": 9.653491766072473e-06, "loss": 0.3985, "step": 4214 }, { "epoch": 0.2913930176287591, "grad_norm": 1.0199320543207238, "learning_rate": 9.653286983923795e-06, "loss": 0.4501, "step": 4215 }, { "epoch": 0.2914621500172831, "grad_norm": 0.8128432369592038, "learning_rate": 9.653082143454469e-06, "loss": 0.2958, "step": 4216 }, { "epoch": 0.2915312824058071, "grad_norm": 0.9445114159815126, "learning_rate": 9.652877244667063e-06, "loss": 0.3416, "step": 4217 }, { "epoch": 0.29160041479433113, "grad_norm": 1.4964093637739924, "learning_rate": 9.652672287564142e-06, "loss": 0.3308, "step": 4218 }, { "epoch": 0.29166954718285515, "grad_norm": 0.786674892779174, "learning_rate": 9.652467272148278e-06, "loss": 0.4806, "step": 4219 }, { "epoch": 0.29173867957137917, "grad_norm": 1.108529091247624, "learning_rate": 9.652262198422039e-06, "loss": 0.3764, "step": 4220 }, { "epoch": 0.29180781195990324, "grad_norm": 0.8248050869001732, "learning_rate": 9.652057066387996e-06, "loss": 0.3563, "step": 4221 }, { "epoch": 0.29187694434842726, "grad_norm": 0.9253834106974465, "learning_rate": 9.651851876048719e-06, "loss": 0.4997, "step": 4222 }, { "epoch": 0.2919460767369513, "grad_norm": 0.8640695548845699, "learning_rate": 9.65164662740678e-06, "loss": 0.3964, "step": 4223 }, { "epoch": 0.2920152091254753, "grad_norm": 0.9142413217183238, "learning_rate": 9.651441320464754e-06, "loss": 0.2877, "step": 4224 }, { "epoch": 0.2920843415139993, "grad_norm": 0.8103313192410438, "learning_rate": 9.65123595522521e-06, "loss": 0.3879, "step": 4225 }, { "epoch": 0.29215347390252333, "grad_norm": 0.8280853785606783, "learning_rate": 9.651030531690724e-06, "loss": 0.5455, "step": 4226 }, { "epoch": 0.29222260629104735, "grad_norm": 0.8702648255660865, "learning_rate": 9.650825049863872e-06, "loss": 0.3343, "step": 4227 }, { "epoch": 0.29229173867957137, "grad_norm": 0.8314108581106605, "learning_rate": 9.650619509747225e-06, "loss": 0.3206, "step": 4228 }, { "epoch": 0.2923608710680954, "grad_norm": 0.855293274765684, "learning_rate": 9.650413911343364e-06, "loss": 0.3894, "step": 4229 }, { "epoch": 0.2924300034566194, "grad_norm": 1.2015390974162863, "learning_rate": 9.650208254654863e-06, "loss": 0.4172, "step": 4230 }, { "epoch": 0.2924991358451435, "grad_norm": 0.7730783184272993, "learning_rate": 9.6500025396843e-06, "loss": 0.3256, "step": 4231 }, { "epoch": 0.2925682682336675, "grad_norm": 0.8430440902889231, "learning_rate": 9.649796766434253e-06, "loss": 0.356, "step": 4232 }, { "epoch": 0.2926374006221915, "grad_norm": 0.8033078881613981, "learning_rate": 9.649590934907302e-06, "loss": 0.398, "step": 4233 }, { "epoch": 0.29270653301071553, "grad_norm": 1.5198568277664737, "learning_rate": 9.649385045106028e-06, "loss": 0.2999, "step": 4234 }, { "epoch": 0.29277566539923955, "grad_norm": 0.939823130027716, "learning_rate": 9.649179097033008e-06, "loss": 0.2947, "step": 4235 }, { "epoch": 0.29284479778776357, "grad_norm": 0.9413384112928684, "learning_rate": 9.648973090690825e-06, "loss": 0.3744, "step": 4236 }, { "epoch": 0.2929139301762876, "grad_norm": 0.9745405654782211, "learning_rate": 9.64876702608206e-06, "loss": 0.4926, "step": 4237 }, { "epoch": 0.2929830625648116, "grad_norm": 1.049305843115899, "learning_rate": 9.648560903209297e-06, "loss": 0.3107, "step": 4238 }, { "epoch": 0.2930521949533356, "grad_norm": 0.9491205196940307, "learning_rate": 9.648354722075118e-06, "loss": 0.3094, "step": 4239 }, { "epoch": 0.29312132734185964, "grad_norm": 0.7869597973040641, "learning_rate": 9.64814848268211e-06, "loss": 0.4311, "step": 4240 }, { "epoch": 0.2931904597303837, "grad_norm": 0.9214310576490664, "learning_rate": 9.647942185032854e-06, "loss": 0.3461, "step": 4241 }, { "epoch": 0.29325959211890773, "grad_norm": 0.7868029818185881, "learning_rate": 9.647735829129938e-06, "loss": 0.3105, "step": 4242 }, { "epoch": 0.29332872450743175, "grad_norm": 0.9168382016512561, "learning_rate": 9.647529414975947e-06, "loss": 0.3847, "step": 4243 }, { "epoch": 0.29339785689595577, "grad_norm": 1.295317912445497, "learning_rate": 9.64732294257347e-06, "loss": 0.4106, "step": 4244 }, { "epoch": 0.2934669892844798, "grad_norm": 0.9123322061679401, "learning_rate": 9.647116411925092e-06, "loss": 0.3841, "step": 4245 }, { "epoch": 0.2935361216730038, "grad_norm": 0.9773518432459819, "learning_rate": 9.646909823033404e-06, "loss": 0.3527, "step": 4246 }, { "epoch": 0.2936052540615278, "grad_norm": 0.8270000353553888, "learning_rate": 9.646703175900993e-06, "loss": 0.4964, "step": 4247 }, { "epoch": 0.29367438645005184, "grad_norm": 1.0481670928594575, "learning_rate": 9.64649647053045e-06, "loss": 0.3984, "step": 4248 }, { "epoch": 0.29374351883857586, "grad_norm": 1.0897840435028205, "learning_rate": 9.646289706924368e-06, "loss": 0.2745, "step": 4249 }, { "epoch": 0.2938126512270999, "grad_norm": 1.0021190838218599, "learning_rate": 9.646082885085335e-06, "loss": 0.3732, "step": 4250 }, { "epoch": 0.2938817836156239, "grad_norm": 0.8221526912567251, "learning_rate": 9.645876005015944e-06, "loss": 0.4435, "step": 4251 }, { "epoch": 0.29395091600414797, "grad_norm": 0.812483872880637, "learning_rate": 9.645669066718788e-06, "loss": 0.3505, "step": 4252 }, { "epoch": 0.294020048392672, "grad_norm": 0.7034224406807706, "learning_rate": 9.64546207019646e-06, "loss": 0.3698, "step": 4253 }, { "epoch": 0.294089180781196, "grad_norm": 0.7468558243174187, "learning_rate": 9.645255015451557e-06, "loss": 0.3893, "step": 4254 }, { "epoch": 0.29415831316972, "grad_norm": 0.8728143129068633, "learning_rate": 9.645047902486673e-06, "loss": 0.342, "step": 4255 }, { "epoch": 0.29422744555824404, "grad_norm": 0.7928415586856472, "learning_rate": 9.644840731304402e-06, "loss": 0.3065, "step": 4256 }, { "epoch": 0.29429657794676806, "grad_norm": 0.8490480949766, "learning_rate": 9.64463350190734e-06, "loss": 0.3661, "step": 4257 }, { "epoch": 0.2943657103352921, "grad_norm": 0.7848862776234283, "learning_rate": 9.644426214298085e-06, "loss": 0.4881, "step": 4258 }, { "epoch": 0.2944348427238161, "grad_norm": 1.233013344209552, "learning_rate": 9.64421886847924e-06, "loss": 0.3085, "step": 4259 }, { "epoch": 0.2945039751123401, "grad_norm": 0.8572419082944516, "learning_rate": 9.644011464453396e-06, "loss": 0.305, "step": 4260 }, { "epoch": 0.29457310750086413, "grad_norm": 0.8244627369313113, "learning_rate": 9.643804002223157e-06, "loss": 0.4146, "step": 4261 }, { "epoch": 0.2946422398893882, "grad_norm": 0.789508817451598, "learning_rate": 9.643596481791122e-06, "loss": 0.3883, "step": 4262 }, { "epoch": 0.2947113722779122, "grad_norm": 0.6913570746735173, "learning_rate": 9.643388903159892e-06, "loss": 0.3868, "step": 4263 }, { "epoch": 0.29478050466643624, "grad_norm": 0.9182011717617932, "learning_rate": 9.643181266332069e-06, "loss": 0.3754, "step": 4264 }, { "epoch": 0.29484963705496026, "grad_norm": 0.7625833752490329, "learning_rate": 9.642973571310255e-06, "loss": 0.4871, "step": 4265 }, { "epoch": 0.2949187694434843, "grad_norm": 0.8536739884164493, "learning_rate": 9.642765818097051e-06, "loss": 0.3846, "step": 4266 }, { "epoch": 0.2949879018320083, "grad_norm": 0.8516879758226312, "learning_rate": 9.642558006695065e-06, "loss": 0.2956, "step": 4267 }, { "epoch": 0.2950570342205323, "grad_norm": 0.797426122042388, "learning_rate": 9.6423501371069e-06, "loss": 0.3921, "step": 4268 }, { "epoch": 0.29512616660905633, "grad_norm": 0.7845861571228745, "learning_rate": 9.642142209335159e-06, "loss": 0.4831, "step": 4269 }, { "epoch": 0.29519529899758035, "grad_norm": 1.0801238806962916, "learning_rate": 9.641934223382449e-06, "loss": 0.2818, "step": 4270 }, { "epoch": 0.29526443138610436, "grad_norm": 0.809085124413914, "learning_rate": 9.64172617925138e-06, "loss": 0.3355, "step": 4271 }, { "epoch": 0.29533356377462844, "grad_norm": 0.839300440086212, "learning_rate": 9.641518076944554e-06, "loss": 0.3896, "step": 4272 }, { "epoch": 0.29540269616315246, "grad_norm": 0.9306531559166801, "learning_rate": 9.641309916464585e-06, "loss": 0.3506, "step": 4273 }, { "epoch": 0.2954718285516765, "grad_norm": 1.0350012679175358, "learning_rate": 9.641101697814076e-06, "loss": 0.2755, "step": 4274 }, { "epoch": 0.2955409609402005, "grad_norm": 0.9103407406998638, "learning_rate": 9.640893420995643e-06, "loss": 0.4286, "step": 4275 }, { "epoch": 0.2956100933287245, "grad_norm": 0.7894872768948723, "learning_rate": 9.64068508601189e-06, "loss": 0.4134, "step": 4276 }, { "epoch": 0.29567922571724853, "grad_norm": 0.8896294890987657, "learning_rate": 9.640476692865432e-06, "loss": 0.3472, "step": 4277 }, { "epoch": 0.29574835810577255, "grad_norm": 1.0628025455481587, "learning_rate": 9.640268241558881e-06, "loss": 0.2893, "step": 4278 }, { "epoch": 0.29581749049429656, "grad_norm": 0.8970825985693098, "learning_rate": 9.64005973209485e-06, "loss": 0.3868, "step": 4279 }, { "epoch": 0.2958866228828206, "grad_norm": 0.9203749592398823, "learning_rate": 9.639851164475946e-06, "loss": 0.3918, "step": 4280 }, { "epoch": 0.2959557552713446, "grad_norm": 1.0349951750365762, "learning_rate": 9.63964253870479e-06, "loss": 0.3237, "step": 4281 }, { "epoch": 0.2960248876598687, "grad_norm": 0.8232221311425141, "learning_rate": 9.639433854783997e-06, "loss": 0.3636, "step": 4282 }, { "epoch": 0.2960940200483927, "grad_norm": 0.8471061243074134, "learning_rate": 9.639225112716178e-06, "loss": 0.4582, "step": 4283 }, { "epoch": 0.2961631524369167, "grad_norm": 0.8173264795998614, "learning_rate": 9.63901631250395e-06, "loss": 0.2898, "step": 4284 }, { "epoch": 0.29623228482544073, "grad_norm": 1.1066795950372625, "learning_rate": 9.638807454149933e-06, "loss": 0.3218, "step": 4285 }, { "epoch": 0.29630141721396475, "grad_norm": 0.8232389700195343, "learning_rate": 9.638598537656742e-06, "loss": 0.3491, "step": 4286 }, { "epoch": 0.29637054960248876, "grad_norm": 0.7490708146240654, "learning_rate": 9.638389563026997e-06, "loss": 0.4613, "step": 4287 }, { "epoch": 0.2964396819910128, "grad_norm": 1.0206659224220211, "learning_rate": 9.638180530263317e-06, "loss": 0.3696, "step": 4288 }, { "epoch": 0.2965088143795368, "grad_norm": 0.8886385527694202, "learning_rate": 9.637971439368321e-06, "loss": 0.3224, "step": 4289 }, { "epoch": 0.2965779467680608, "grad_norm": 0.7574013647587678, "learning_rate": 9.637762290344628e-06, "loss": 0.4441, "step": 4290 }, { "epoch": 0.29664707915658484, "grad_norm": 0.9677910955243189, "learning_rate": 9.637553083194863e-06, "loss": 0.3278, "step": 4291 }, { "epoch": 0.2967162115451089, "grad_norm": 1.116948602217082, "learning_rate": 9.637343817921646e-06, "loss": 0.3142, "step": 4292 }, { "epoch": 0.29678534393363293, "grad_norm": 0.8041103898545886, "learning_rate": 9.637134494527599e-06, "loss": 0.3795, "step": 4293 }, { "epoch": 0.29685447632215695, "grad_norm": 0.9696925568882879, "learning_rate": 9.636925113015348e-06, "loss": 0.4979, "step": 4294 }, { "epoch": 0.29692360871068096, "grad_norm": 1.0920848026104697, "learning_rate": 9.636715673387513e-06, "loss": 0.3816, "step": 4295 }, { "epoch": 0.296992741099205, "grad_norm": 0.7703713697160129, "learning_rate": 9.636506175646725e-06, "loss": 0.3468, "step": 4296 }, { "epoch": 0.297061873487729, "grad_norm": 0.8403950192116276, "learning_rate": 9.636296619795604e-06, "loss": 0.4417, "step": 4297 }, { "epoch": 0.297131005876253, "grad_norm": 1.1146745578907287, "learning_rate": 9.636087005836779e-06, "loss": 0.4026, "step": 4298 }, { "epoch": 0.29720013826477704, "grad_norm": 0.7865329824778039, "learning_rate": 9.635877333772878e-06, "loss": 0.3216, "step": 4299 }, { "epoch": 0.29726927065330105, "grad_norm": 0.7902679340990969, "learning_rate": 9.635667603606527e-06, "loss": 0.3673, "step": 4300 }, { "epoch": 0.2973384030418251, "grad_norm": 0.7738459771098725, "learning_rate": 9.635457815340354e-06, "loss": 0.4272, "step": 4301 }, { "epoch": 0.29740753543034915, "grad_norm": 1.210077521679649, "learning_rate": 9.63524796897699e-06, "loss": 0.3583, "step": 4302 }, { "epoch": 0.29747666781887316, "grad_norm": 0.8772637090262392, "learning_rate": 9.635038064519066e-06, "loss": 0.3441, "step": 4303 }, { "epoch": 0.2975458002073972, "grad_norm": 0.9815872670508454, "learning_rate": 9.634828101969212e-06, "loss": 0.4296, "step": 4304 }, { "epoch": 0.2976149325959212, "grad_norm": 1.259849521811663, "learning_rate": 9.634618081330059e-06, "loss": 0.4615, "step": 4305 }, { "epoch": 0.2976840649844452, "grad_norm": 0.8136422244610935, "learning_rate": 9.634408002604236e-06, "loss": 0.3207, "step": 4306 }, { "epoch": 0.29775319737296924, "grad_norm": 0.7324369314909269, "learning_rate": 9.634197865794382e-06, "loss": 0.3094, "step": 4307 }, { "epoch": 0.29782232976149325, "grad_norm": 1.0314913615805812, "learning_rate": 9.633987670903128e-06, "loss": 0.4236, "step": 4308 }, { "epoch": 0.29789146215001727, "grad_norm": 0.8678090051616482, "learning_rate": 9.633777417933106e-06, "loss": 0.3897, "step": 4309 }, { "epoch": 0.2979605945385413, "grad_norm": 0.7499576023095385, "learning_rate": 9.633567106886956e-06, "loss": 0.3239, "step": 4310 }, { "epoch": 0.2980297269270653, "grad_norm": 0.8047419123329659, "learning_rate": 9.63335673776731e-06, "loss": 0.417, "step": 4311 }, { "epoch": 0.2980988593155893, "grad_norm": 1.1607402140379526, "learning_rate": 9.633146310576807e-06, "loss": 0.389, "step": 4312 }, { "epoch": 0.2981679917041134, "grad_norm": 0.8397474495832086, "learning_rate": 9.632935825318082e-06, "loss": 0.344, "step": 4313 }, { "epoch": 0.2982371240926374, "grad_norm": 0.9068735941494012, "learning_rate": 9.632725281993774e-06, "loss": 0.3493, "step": 4314 }, { "epoch": 0.29830625648116144, "grad_norm": 0.8378697800896671, "learning_rate": 9.632514680606521e-06, "loss": 0.4058, "step": 4315 }, { "epoch": 0.29837538886968545, "grad_norm": 0.8127975911981513, "learning_rate": 9.632304021158966e-06, "loss": 0.3132, "step": 4316 }, { "epoch": 0.29844452125820947, "grad_norm": 0.9403217865612308, "learning_rate": 9.632093303653747e-06, "loss": 0.2656, "step": 4317 }, { "epoch": 0.2985136536467335, "grad_norm": 0.8335640228652311, "learning_rate": 9.631882528093503e-06, "loss": 0.3885, "step": 4318 }, { "epoch": 0.2985827860352575, "grad_norm": 0.9310991711041381, "learning_rate": 9.631671694480879e-06, "loss": 0.5356, "step": 4319 }, { "epoch": 0.2986519184237815, "grad_norm": 0.7949939334252395, "learning_rate": 9.631460802818515e-06, "loss": 0.3293, "step": 4320 }, { "epoch": 0.29872105081230554, "grad_norm": 0.8830622556555509, "learning_rate": 9.631249853109054e-06, "loss": 0.296, "step": 4321 }, { "epoch": 0.29879018320082956, "grad_norm": 0.7856823566359804, "learning_rate": 9.631038845355143e-06, "loss": 0.4546, "step": 4322 }, { "epoch": 0.29885931558935364, "grad_norm": 1.0006832967091661, "learning_rate": 9.630827779559424e-06, "loss": 0.4101, "step": 4323 }, { "epoch": 0.29892844797787765, "grad_norm": 0.8242479511628893, "learning_rate": 9.630616655724542e-06, "loss": 0.2904, "step": 4324 }, { "epoch": 0.29899758036640167, "grad_norm": 0.8206537982985208, "learning_rate": 9.630405473853143e-06, "loss": 0.3947, "step": 4325 }, { "epoch": 0.2990667127549257, "grad_norm": 0.7984417605470878, "learning_rate": 9.630194233947878e-06, "loss": 0.419, "step": 4326 }, { "epoch": 0.2991358451434497, "grad_norm": 1.0540218910623644, "learning_rate": 9.629982936011388e-06, "loss": 0.3217, "step": 4327 }, { "epoch": 0.2992049775319737, "grad_norm": 0.7439067475278204, "learning_rate": 9.629771580046327e-06, "loss": 0.3555, "step": 4328 }, { "epoch": 0.29927410992049774, "grad_norm": 0.8057308183141597, "learning_rate": 9.62956016605534e-06, "loss": 0.4249, "step": 4329 }, { "epoch": 0.29934324230902176, "grad_norm": 1.2832602880844668, "learning_rate": 9.629348694041078e-06, "loss": 0.3987, "step": 4330 }, { "epoch": 0.2994123746975458, "grad_norm": 3.3582067243701395, "learning_rate": 9.629137164006192e-06, "loss": 0.3612, "step": 4331 }, { "epoch": 0.2994815070860698, "grad_norm": 0.9431817761451659, "learning_rate": 9.628925575953333e-06, "loss": 0.3055, "step": 4332 }, { "epoch": 0.29955063947459387, "grad_norm": 0.8965623894884146, "learning_rate": 9.628713929885152e-06, "loss": 0.4451, "step": 4333 }, { "epoch": 0.2996197718631179, "grad_norm": 0.8325968329061714, "learning_rate": 9.628502225804303e-06, "loss": 0.328, "step": 4334 }, { "epoch": 0.2996889042516419, "grad_norm": 0.9735645194176723, "learning_rate": 9.628290463713439e-06, "loss": 0.3271, "step": 4335 }, { "epoch": 0.2997580366401659, "grad_norm": 0.806600663338806, "learning_rate": 9.628078643615212e-06, "loss": 0.4296, "step": 4336 }, { "epoch": 0.29982716902868994, "grad_norm": 0.7959450793076566, "learning_rate": 9.627866765512279e-06, "loss": 0.4001, "step": 4337 }, { "epoch": 0.29989630141721396, "grad_norm": 0.8834810410554866, "learning_rate": 9.627654829407295e-06, "loss": 0.3158, "step": 4338 }, { "epoch": 0.299965433805738, "grad_norm": 2.655745566655083, "learning_rate": 9.627442835302915e-06, "loss": 0.2993, "step": 4339 }, { "epoch": 0.300034566194262, "grad_norm": 0.7787358069439269, "learning_rate": 9.627230783201798e-06, "loss": 0.4587, "step": 4340 }, { "epoch": 0.300103698582786, "grad_norm": 0.8915504612177901, "learning_rate": 9.627018673106602e-06, "loss": 0.3183, "step": 4341 }, { "epoch": 0.30017283097131003, "grad_norm": 0.7602278101959018, "learning_rate": 9.626806505019983e-06, "loss": 0.3452, "step": 4342 }, { "epoch": 0.3002419633598341, "grad_norm": 0.7780726197558949, "learning_rate": 9.626594278944601e-06, "loss": 0.368, "step": 4343 }, { "epoch": 0.3003110957483581, "grad_norm": 0.7989503467660836, "learning_rate": 9.626381994883117e-06, "loss": 0.4636, "step": 4344 }, { "epoch": 0.30038022813688214, "grad_norm": 0.7680648691292925, "learning_rate": 9.62616965283819e-06, "loss": 0.3369, "step": 4345 }, { "epoch": 0.30044936052540616, "grad_norm": 0.7277555197764131, "learning_rate": 9.625957252812484e-06, "loss": 0.3704, "step": 4346 }, { "epoch": 0.3005184929139302, "grad_norm": 0.8961829798237905, "learning_rate": 9.625744794808659e-06, "loss": 0.4762, "step": 4347 }, { "epoch": 0.3005876253024542, "grad_norm": 0.8301622067085569, "learning_rate": 9.625532278829374e-06, "loss": 0.3233, "step": 4348 }, { "epoch": 0.3006567576909782, "grad_norm": 0.8668727259944297, "learning_rate": 9.625319704877301e-06, "loss": 0.2886, "step": 4349 }, { "epoch": 0.30072589007950223, "grad_norm": 0.7635570307234736, "learning_rate": 9.625107072955098e-06, "loss": 0.3513, "step": 4350 }, { "epoch": 0.30079502246802625, "grad_norm": 0.7856364699037828, "learning_rate": 9.624894383065431e-06, "loss": 0.3941, "step": 4351 }, { "epoch": 0.30086415485655027, "grad_norm": 0.7625383815814418, "learning_rate": 9.624681635210969e-06, "loss": 0.3072, "step": 4352 }, { "epoch": 0.30093328724507434, "grad_norm": 0.8045749151435938, "learning_rate": 9.624468829394374e-06, "loss": 0.3041, "step": 4353 }, { "epoch": 0.30100241963359836, "grad_norm": 0.8626995334219866, "learning_rate": 9.624255965618314e-06, "loss": 0.4572, "step": 4354 }, { "epoch": 0.3010715520221224, "grad_norm": 0.7850916892172589, "learning_rate": 9.62404304388546e-06, "loss": 0.3667, "step": 4355 }, { "epoch": 0.3011406844106464, "grad_norm": 0.8159632290272905, "learning_rate": 9.623830064198477e-06, "loss": 0.3613, "step": 4356 }, { "epoch": 0.3012098167991704, "grad_norm": 2.3588145199125203, "learning_rate": 9.623617026560036e-06, "loss": 0.3914, "step": 4357 }, { "epoch": 0.30127894918769443, "grad_norm": 0.7962887037057191, "learning_rate": 9.623403930972808e-06, "loss": 0.4156, "step": 4358 }, { "epoch": 0.30134808157621845, "grad_norm": 0.9991348413395965, "learning_rate": 9.623190777439462e-06, "loss": 0.3789, "step": 4359 }, { "epoch": 0.30141721396474247, "grad_norm": 0.7455605023242283, "learning_rate": 9.622977565962668e-06, "loss": 0.2852, "step": 4360 }, { "epoch": 0.3014863463532665, "grad_norm": 0.810348682480744, "learning_rate": 9.622764296545102e-06, "loss": 0.3603, "step": 4361 }, { "epoch": 0.3015554787417905, "grad_norm": 0.8878580122680382, "learning_rate": 9.622550969189435e-06, "loss": 0.3912, "step": 4362 }, { "epoch": 0.3016246111303146, "grad_norm": 0.830696022721305, "learning_rate": 9.622337583898341e-06, "loss": 0.3854, "step": 4363 }, { "epoch": 0.3016937435188386, "grad_norm": 0.9076413497831538, "learning_rate": 9.622124140674495e-06, "loss": 0.3456, "step": 4364 }, { "epoch": 0.3017628759073626, "grad_norm": 0.8356421744525837, "learning_rate": 9.62191063952057e-06, "loss": 0.4285, "step": 4365 }, { "epoch": 0.30183200829588663, "grad_norm": 0.8000605297764231, "learning_rate": 9.621697080439244e-06, "loss": 0.3363, "step": 4366 }, { "epoch": 0.30190114068441065, "grad_norm": 0.8054615812858672, "learning_rate": 9.621483463433194e-06, "loss": 0.3117, "step": 4367 }, { "epoch": 0.30197027307293467, "grad_norm": 0.9270778846907697, "learning_rate": 9.621269788505094e-06, "loss": 0.3091, "step": 4368 }, { "epoch": 0.3020394054614587, "grad_norm": 0.8903582608969938, "learning_rate": 9.621056055657627e-06, "loss": 0.4504, "step": 4369 }, { "epoch": 0.3021085378499827, "grad_norm": 0.849680803782777, "learning_rate": 9.620842264893467e-06, "loss": 0.3924, "step": 4370 }, { "epoch": 0.3021776702385067, "grad_norm": 0.8828057306135407, "learning_rate": 9.620628416215296e-06, "loss": 0.3467, "step": 4371 }, { "epoch": 0.30224680262703074, "grad_norm": 0.840157695804803, "learning_rate": 9.620414509625792e-06, "loss": 0.4586, "step": 4372 }, { "epoch": 0.30231593501555476, "grad_norm": 0.8649668321418764, "learning_rate": 9.62020054512764e-06, "loss": 0.3223, "step": 4373 }, { "epoch": 0.30238506740407883, "grad_norm": 1.4352037973650256, "learning_rate": 9.61998652272352e-06, "loss": 0.2856, "step": 4374 }, { "epoch": 0.30245419979260285, "grad_norm": 1.0299576574319755, "learning_rate": 9.61977244241611e-06, "loss": 0.3535, "step": 4375 }, { "epoch": 0.30252333218112687, "grad_norm": 0.8719263341931104, "learning_rate": 9.619558304208099e-06, "loss": 0.441, "step": 4376 }, { "epoch": 0.3025924645696509, "grad_norm": 0.7762621179795499, "learning_rate": 9.619344108102168e-06, "loss": 0.3173, "step": 4377 }, { "epoch": 0.3026615969581749, "grad_norm": 0.9189379643728343, "learning_rate": 9.619129854101004e-06, "loss": 0.2815, "step": 4378 }, { "epoch": 0.3027307293466989, "grad_norm": 0.7516173334078227, "learning_rate": 9.618915542207289e-06, "loss": 0.411, "step": 4379 }, { "epoch": 0.30279986173522294, "grad_norm": 0.8354666470527677, "learning_rate": 9.61870117242371e-06, "loss": 0.3536, "step": 4380 }, { "epoch": 0.30286899412374696, "grad_norm": 0.8770888163106784, "learning_rate": 9.618486744752954e-06, "loss": 0.3357, "step": 4381 }, { "epoch": 0.302938126512271, "grad_norm": 0.8603642138973547, "learning_rate": 9.618272259197712e-06, "loss": 0.3193, "step": 4382 }, { "epoch": 0.303007258900795, "grad_norm": 0.7266169310723849, "learning_rate": 9.618057715760666e-06, "loss": 0.374, "step": 4383 }, { "epoch": 0.30307639128931907, "grad_norm": 0.8091904910754879, "learning_rate": 9.617843114444508e-06, "loss": 0.3209, "step": 4384 }, { "epoch": 0.3031455236778431, "grad_norm": 0.9197201261181113, "learning_rate": 9.617628455251928e-06, "loss": 0.2756, "step": 4385 }, { "epoch": 0.3032146560663671, "grad_norm": 1.1657066473928968, "learning_rate": 9.617413738185616e-06, "loss": 0.4005, "step": 4386 }, { "epoch": 0.3032837884548911, "grad_norm": 0.8335529417698565, "learning_rate": 9.617198963248263e-06, "loss": 0.4136, "step": 4387 }, { "epoch": 0.30335292084341514, "grad_norm": 0.7186937149412711, "learning_rate": 9.61698413044256e-06, "loss": 0.3885, "step": 4388 }, { "epoch": 0.30342205323193916, "grad_norm": 0.9513702727301434, "learning_rate": 9.6167692397712e-06, "loss": 0.3867, "step": 4389 }, { "epoch": 0.3034911856204632, "grad_norm": 0.7514634559813234, "learning_rate": 9.616554291236878e-06, "loss": 0.4124, "step": 4390 }, { "epoch": 0.3035603180089872, "grad_norm": 0.8827722353933665, "learning_rate": 9.616339284842286e-06, "loss": 0.3368, "step": 4391 }, { "epoch": 0.3036294503975112, "grad_norm": 0.7224938369132152, "learning_rate": 9.616124220590119e-06, "loss": 0.3152, "step": 4392 }, { "epoch": 0.30369858278603523, "grad_norm": 0.9411948656440259, "learning_rate": 9.615909098483073e-06, "loss": 0.4266, "step": 4393 }, { "epoch": 0.3037677151745593, "grad_norm": 1.138080692832357, "learning_rate": 9.615693918523844e-06, "loss": 0.4097, "step": 4394 }, { "epoch": 0.3038368475630833, "grad_norm": 0.8582817814533861, "learning_rate": 9.615478680715127e-06, "loss": 0.3445, "step": 4395 }, { "epoch": 0.30390597995160734, "grad_norm": 1.1989201691897533, "learning_rate": 9.615263385059624e-06, "loss": 0.3173, "step": 4396 }, { "epoch": 0.30397511234013136, "grad_norm": 0.8656852596298623, "learning_rate": 9.61504803156003e-06, "loss": 0.4546, "step": 4397 }, { "epoch": 0.3040442447286554, "grad_norm": 0.8702172943718717, "learning_rate": 9.614832620219044e-06, "loss": 0.4288, "step": 4398 }, { "epoch": 0.3041133771171794, "grad_norm": 0.7789256961284712, "learning_rate": 9.614617151039367e-06, "loss": 0.2626, "step": 4399 }, { "epoch": 0.3041825095057034, "grad_norm": 0.8631872880137211, "learning_rate": 9.6144016240237e-06, "loss": 0.3974, "step": 4400 }, { "epoch": 0.30425164189422743, "grad_norm": 0.8804622687177729, "learning_rate": 9.614186039174744e-06, "loss": 0.4489, "step": 4401 }, { "epoch": 0.30432077428275145, "grad_norm": 0.9656577912697593, "learning_rate": 9.613970396495199e-06, "loss": 0.3318, "step": 4402 }, { "epoch": 0.30438990667127547, "grad_norm": 0.8501110409777541, "learning_rate": 9.613754695987769e-06, "loss": 0.3101, "step": 4403 }, { "epoch": 0.30445903905979954, "grad_norm": 1.5674158861929453, "learning_rate": 9.613538937655158e-06, "loss": 0.487, "step": 4404 }, { "epoch": 0.30452817144832356, "grad_norm": 0.8835258317260035, "learning_rate": 9.613323121500069e-06, "loss": 0.4472, "step": 4405 }, { "epoch": 0.3045973038368476, "grad_norm": 0.7406271495069265, "learning_rate": 9.613107247525208e-06, "loss": 0.2848, "step": 4406 }, { "epoch": 0.3046664362253716, "grad_norm": 0.7236992558795374, "learning_rate": 9.612891315733279e-06, "loss": 0.3102, "step": 4407 }, { "epoch": 0.3047355686138956, "grad_norm": 0.8885096604501781, "learning_rate": 9.612675326126992e-06, "loss": 0.3948, "step": 4408 }, { "epoch": 0.30480470100241963, "grad_norm": 1.568752805393599, "learning_rate": 9.61245927870905e-06, "loss": 0.3393, "step": 4409 }, { "epoch": 0.30487383339094365, "grad_norm": 0.7839072166130658, "learning_rate": 9.612243173482162e-06, "loss": 0.3245, "step": 4410 }, { "epoch": 0.30494296577946767, "grad_norm": 0.7856023903227513, "learning_rate": 9.612027010449037e-06, "loss": 0.3896, "step": 4411 }, { "epoch": 0.3050120981679917, "grad_norm": 0.8378814510257497, "learning_rate": 9.611810789612383e-06, "loss": 0.3625, "step": 4412 }, { "epoch": 0.3050812305565157, "grad_norm": 0.7699600996830739, "learning_rate": 9.611594510974914e-06, "loss": 0.3499, "step": 4413 }, { "epoch": 0.3051503629450398, "grad_norm": 0.8022697784049239, "learning_rate": 9.611378174539334e-06, "loss": 0.3234, "step": 4414 }, { "epoch": 0.3052194953335638, "grad_norm": 1.173936895118061, "learning_rate": 9.611161780308361e-06, "loss": 0.4982, "step": 4415 }, { "epoch": 0.3052886277220878, "grad_norm": 0.8449287067199699, "learning_rate": 9.610945328284701e-06, "loss": 0.3077, "step": 4416 }, { "epoch": 0.30535776011061183, "grad_norm": 0.7425746749424319, "learning_rate": 9.610728818471072e-06, "loss": 0.3044, "step": 4417 }, { "epoch": 0.30542689249913585, "grad_norm": 0.9025869146408675, "learning_rate": 9.610512250870185e-06, "loss": 0.4611, "step": 4418 }, { "epoch": 0.30549602488765987, "grad_norm": 0.854125806838428, "learning_rate": 9.610295625484754e-06, "loss": 0.4434, "step": 4419 }, { "epoch": 0.3055651572761839, "grad_norm": 0.8788026447826721, "learning_rate": 9.610078942317496e-06, "loss": 0.3453, "step": 4420 }, { "epoch": 0.3056342896647079, "grad_norm": 0.7407096704465254, "learning_rate": 9.609862201371125e-06, "loss": 0.2786, "step": 4421 }, { "epoch": 0.3057034220532319, "grad_norm": 0.9655767462433287, "learning_rate": 9.609645402648358e-06, "loss": 0.4359, "step": 4422 }, { "epoch": 0.30577255444175594, "grad_norm": 0.8461353885371187, "learning_rate": 9.609428546151911e-06, "loss": 0.4229, "step": 4423 }, { "epoch": 0.30584168683028, "grad_norm": 0.811820996803356, "learning_rate": 9.609211631884505e-06, "loss": 0.3185, "step": 4424 }, { "epoch": 0.30591081921880403, "grad_norm": 0.8146298363889902, "learning_rate": 9.608994659848855e-06, "loss": 0.3552, "step": 4425 }, { "epoch": 0.30597995160732805, "grad_norm": 0.8535426832957784, "learning_rate": 9.608777630047683e-06, "loss": 0.4228, "step": 4426 }, { "epoch": 0.30604908399585207, "grad_norm": 0.7077823617730411, "learning_rate": 9.608560542483707e-06, "loss": 0.3103, "step": 4427 }, { "epoch": 0.3061182163843761, "grad_norm": 0.8121409735778296, "learning_rate": 9.608343397159651e-06, "loss": 0.3295, "step": 4428 }, { "epoch": 0.3061873487729001, "grad_norm": 0.7749994191304813, "learning_rate": 9.608126194078233e-06, "loss": 0.3997, "step": 4429 }, { "epoch": 0.3062564811614241, "grad_norm": 0.7968156809415488, "learning_rate": 9.607908933242177e-06, "loss": 0.3939, "step": 4430 }, { "epoch": 0.30632561354994814, "grad_norm": 0.928152062478075, "learning_rate": 9.607691614654207e-06, "loss": 0.3963, "step": 4431 }, { "epoch": 0.30639474593847216, "grad_norm": 0.8190013010350726, "learning_rate": 9.607474238317043e-06, "loss": 0.3064, "step": 4432 }, { "epoch": 0.3064638783269962, "grad_norm": 0.77579626716913, "learning_rate": 9.607256804233414e-06, "loss": 0.4221, "step": 4433 }, { "epoch": 0.30653301071552025, "grad_norm": 0.8106673685537508, "learning_rate": 9.60703931240604e-06, "loss": 0.3631, "step": 4434 }, { "epoch": 0.30660214310404427, "grad_norm": 0.8263096068589579, "learning_rate": 9.606821762837652e-06, "loss": 0.3043, "step": 4435 }, { "epoch": 0.3066712754925683, "grad_norm": 0.8288546052986779, "learning_rate": 9.606604155530977e-06, "loss": 0.3848, "step": 4436 }, { "epoch": 0.3067404078810923, "grad_norm": 0.8606997368374507, "learning_rate": 9.606386490488736e-06, "loss": 0.4199, "step": 4437 }, { "epoch": 0.3068095402696163, "grad_norm": 0.7507832673231326, "learning_rate": 9.606168767713662e-06, "loss": 0.3508, "step": 4438 }, { "epoch": 0.30687867265814034, "grad_norm": 0.7807470523607913, "learning_rate": 9.605950987208483e-06, "loss": 0.4061, "step": 4439 }, { "epoch": 0.30694780504666436, "grad_norm": 0.9501780385391926, "learning_rate": 9.605733148975927e-06, "loss": 0.4484, "step": 4440 }, { "epoch": 0.3070169374351884, "grad_norm": 0.8755290988503094, "learning_rate": 9.605515253018725e-06, "loss": 0.3474, "step": 4441 }, { "epoch": 0.3070860698237124, "grad_norm": 0.8531525432429353, "learning_rate": 9.60529729933961e-06, "loss": 0.3331, "step": 4442 }, { "epoch": 0.3071552022122364, "grad_norm": 0.777993576434498, "learning_rate": 9.60507928794131e-06, "loss": 0.3653, "step": 4443 }, { "epoch": 0.30722433460076043, "grad_norm": 0.7783849239657671, "learning_rate": 9.604861218826561e-06, "loss": 0.3736, "step": 4444 }, { "epoch": 0.3072934669892845, "grad_norm": 0.9011127358303729, "learning_rate": 9.604643091998094e-06, "loss": 0.3109, "step": 4445 }, { "epoch": 0.3073625993778085, "grad_norm": 0.6934189842365257, "learning_rate": 9.604424907458643e-06, "loss": 0.286, "step": 4446 }, { "epoch": 0.30743173176633254, "grad_norm": 0.7918725268101537, "learning_rate": 9.604206665210942e-06, "loss": 0.388, "step": 4447 }, { "epoch": 0.30750086415485656, "grad_norm": 0.7624954660780603, "learning_rate": 9.603988365257727e-06, "loss": 0.3504, "step": 4448 }, { "epoch": 0.3075699965433806, "grad_norm": 1.0917979251225276, "learning_rate": 9.603770007601735e-06, "loss": 0.3268, "step": 4449 }, { "epoch": 0.3076391289319046, "grad_norm": 0.7874267863282463, "learning_rate": 9.603551592245701e-06, "loss": 0.4727, "step": 4450 }, { "epoch": 0.3077082613204286, "grad_norm": 0.8229534325459903, "learning_rate": 9.603333119192364e-06, "loss": 0.4396, "step": 4451 }, { "epoch": 0.30777739370895263, "grad_norm": 0.7483606663157857, "learning_rate": 9.60311458844446e-06, "loss": 0.3337, "step": 4452 }, { "epoch": 0.30784652609747665, "grad_norm": 0.7785768344295998, "learning_rate": 9.602896000004731e-06, "loss": 0.2928, "step": 4453 }, { "epoch": 0.30791565848600067, "grad_norm": 0.7993196416521196, "learning_rate": 9.602677353875912e-06, "loss": 0.4206, "step": 4454 }, { "epoch": 0.30798479087452474, "grad_norm": 0.9110754882611993, "learning_rate": 9.602458650060751e-06, "loss": 0.4523, "step": 4455 }, { "epoch": 0.30805392326304876, "grad_norm": 0.7779389018785132, "learning_rate": 9.602239888561981e-06, "loss": 0.3245, "step": 4456 }, { "epoch": 0.3081230556515728, "grad_norm": 0.7091253462579082, "learning_rate": 9.602021069382347e-06, "loss": 0.3041, "step": 4457 }, { "epoch": 0.3081921880400968, "grad_norm": 0.799360681604005, "learning_rate": 9.601802192524593e-06, "loss": 0.4672, "step": 4458 }, { "epoch": 0.3082613204286208, "grad_norm": 0.732682956365471, "learning_rate": 9.60158325799146e-06, "loss": 0.2596, "step": 4459 }, { "epoch": 0.30833045281714483, "grad_norm": 0.7539482103783973, "learning_rate": 9.601364265785692e-06, "loss": 0.3003, "step": 4460 }, { "epoch": 0.30839958520566885, "grad_norm": 0.8812131952739621, "learning_rate": 9.601145215910035e-06, "loss": 0.3741, "step": 4461 }, { "epoch": 0.30846871759419286, "grad_norm": 0.9065212043211709, "learning_rate": 9.600926108367235e-06, "loss": 0.5069, "step": 4462 }, { "epoch": 0.3085378499827169, "grad_norm": 1.6290867681211014, "learning_rate": 9.600706943160034e-06, "loss": 0.3196, "step": 4463 }, { "epoch": 0.3086069823712409, "grad_norm": 1.1175814521341265, "learning_rate": 9.600487720291186e-06, "loss": 0.3465, "step": 4464 }, { "epoch": 0.308676114759765, "grad_norm": 0.8150154268287729, "learning_rate": 9.60026843976343e-06, "loss": 0.4758, "step": 4465 }, { "epoch": 0.308745247148289, "grad_norm": 0.8366972114581211, "learning_rate": 9.60004910157952e-06, "loss": 0.3752, "step": 4466 }, { "epoch": 0.308814379536813, "grad_norm": 0.7604053583566223, "learning_rate": 9.599829705742205e-06, "loss": 0.35, "step": 4467 }, { "epoch": 0.30888351192533703, "grad_norm": 0.8283170209078001, "learning_rate": 9.599610252254233e-06, "loss": 0.4222, "step": 4468 }, { "epoch": 0.30895264431386105, "grad_norm": 0.8715190730471902, "learning_rate": 9.599390741118354e-06, "loss": 0.4588, "step": 4469 }, { "epoch": 0.30902177670238506, "grad_norm": 0.7103640939640011, "learning_rate": 9.599171172337322e-06, "loss": 0.3304, "step": 4470 }, { "epoch": 0.3090909090909091, "grad_norm": 0.7842629166740531, "learning_rate": 9.598951545913887e-06, "loss": 0.3748, "step": 4471 }, { "epoch": 0.3091600414794331, "grad_norm": 0.8432665512617854, "learning_rate": 9.5987318618508e-06, "loss": 0.5004, "step": 4472 }, { "epoch": 0.3092291738679571, "grad_norm": 0.741874521065279, "learning_rate": 9.598512120150816e-06, "loss": 0.3559, "step": 4473 }, { "epoch": 0.30929830625648114, "grad_norm": 0.6740622259908414, "learning_rate": 9.598292320816689e-06, "loss": 0.2831, "step": 4474 }, { "epoch": 0.3093674386450052, "grad_norm": 0.7463939958287266, "learning_rate": 9.598072463851173e-06, "loss": 0.3244, "step": 4475 }, { "epoch": 0.30943657103352923, "grad_norm": 0.890537826974623, "learning_rate": 9.597852549257026e-06, "loss": 0.397, "step": 4476 }, { "epoch": 0.30950570342205325, "grad_norm": 0.7687357168093031, "learning_rate": 9.597632577037001e-06, "loss": 0.3435, "step": 4477 }, { "epoch": 0.30957483581057726, "grad_norm": 0.8421470167157592, "learning_rate": 9.597412547193857e-06, "loss": 0.3501, "step": 4478 }, { "epoch": 0.3096439681991013, "grad_norm": 0.7565303171374709, "learning_rate": 9.59719245973035e-06, "loss": 0.4024, "step": 4479 }, { "epoch": 0.3097131005876253, "grad_norm": 0.8316980478574816, "learning_rate": 9.596972314649241e-06, "loss": 0.3996, "step": 4480 }, { "epoch": 0.3097822329761493, "grad_norm": 0.6954343231120628, "learning_rate": 9.59675211195329e-06, "loss": 0.3242, "step": 4481 }, { "epoch": 0.30985136536467334, "grad_norm": 0.8855401487298628, "learning_rate": 9.596531851645252e-06, "loss": 0.3542, "step": 4482 }, { "epoch": 0.30992049775319735, "grad_norm": 0.8668191675602465, "learning_rate": 9.59631153372789e-06, "loss": 0.4426, "step": 4483 }, { "epoch": 0.3099896301417214, "grad_norm": 0.8864236864284314, "learning_rate": 9.596091158203967e-06, "loss": 0.3365, "step": 4484 }, { "epoch": 0.31005876253024545, "grad_norm": 0.7241963160536045, "learning_rate": 9.595870725076241e-06, "loss": 0.2885, "step": 4485 }, { "epoch": 0.31012789491876946, "grad_norm": 0.8296140048100367, "learning_rate": 9.595650234347477e-06, "loss": 0.376, "step": 4486 }, { "epoch": 0.3101970273072935, "grad_norm": 0.7313080822206218, "learning_rate": 9.595429686020441e-06, "loss": 0.3682, "step": 4487 }, { "epoch": 0.3102661596958175, "grad_norm": 0.712104925024475, "learning_rate": 9.595209080097894e-06, "loss": 0.3442, "step": 4488 }, { "epoch": 0.3103352920843415, "grad_norm": 0.773241445725044, "learning_rate": 9.594988416582602e-06, "loss": 0.3578, "step": 4489 }, { "epoch": 0.31040442447286554, "grad_norm": 0.77923651346579, "learning_rate": 9.59476769547733e-06, "loss": 0.4197, "step": 4490 }, { "epoch": 0.31047355686138955, "grad_norm": 0.7886805871822238, "learning_rate": 9.594546916784844e-06, "loss": 0.3439, "step": 4491 }, { "epoch": 0.3105426892499136, "grad_norm": 0.7755819784934226, "learning_rate": 9.594326080507912e-06, "loss": 0.3792, "step": 4492 }, { "epoch": 0.3106118216384376, "grad_norm": 0.8574351493112148, "learning_rate": 9.594105186649301e-06, "loss": 0.3988, "step": 4493 }, { "epoch": 0.3106809540269616, "grad_norm": 0.8831551465536626, "learning_rate": 9.59388423521178e-06, "loss": 0.4363, "step": 4494 }, { "epoch": 0.3107500864154857, "grad_norm": 0.9049662819763302, "learning_rate": 9.59366322619812e-06, "loss": 0.3771, "step": 4495 }, { "epoch": 0.3108192188040097, "grad_norm": 0.8189468772649332, "learning_rate": 9.593442159611089e-06, "loss": 0.3283, "step": 4496 }, { "epoch": 0.3108883511925337, "grad_norm": 0.77462568795937, "learning_rate": 9.593221035453457e-06, "loss": 0.4706, "step": 4497 }, { "epoch": 0.31095748358105774, "grad_norm": 0.9858824621181342, "learning_rate": 9.592999853727996e-06, "loss": 0.3611, "step": 4498 }, { "epoch": 0.31102661596958175, "grad_norm": 0.7851229908789115, "learning_rate": 9.592778614437479e-06, "loss": 0.3401, "step": 4499 }, { "epoch": 0.3110957483581058, "grad_norm": 0.7758458386897353, "learning_rate": 9.592557317584678e-06, "loss": 0.3599, "step": 4500 }, { "epoch": 0.3111648807466298, "grad_norm": 0.7642973667902375, "learning_rate": 9.592335963172366e-06, "loss": 0.4968, "step": 4501 }, { "epoch": 0.3112340131351538, "grad_norm": 0.7146972804951657, "learning_rate": 9.59211455120332e-06, "loss": 0.303, "step": 4502 }, { "epoch": 0.3113031455236778, "grad_norm": 0.9676904675541399, "learning_rate": 9.591893081680312e-06, "loss": 0.2756, "step": 4503 }, { "epoch": 0.31137227791220184, "grad_norm": 0.805935638651009, "learning_rate": 9.591671554606119e-06, "loss": 0.4214, "step": 4504 }, { "epoch": 0.31144141030072586, "grad_norm": 0.8425340400404435, "learning_rate": 9.591449969983517e-06, "loss": 0.4691, "step": 4505 }, { "epoch": 0.31151054268924994, "grad_norm": 0.8814952739643881, "learning_rate": 9.591228327815283e-06, "loss": 0.338, "step": 4506 }, { "epoch": 0.31157967507777395, "grad_norm": 1.5602596190988731, "learning_rate": 9.591006628104195e-06, "loss": 0.3869, "step": 4507 }, { "epoch": 0.311648807466298, "grad_norm": 0.9630291017186823, "learning_rate": 9.590784870853034e-06, "loss": 0.4115, "step": 4508 }, { "epoch": 0.311717939854822, "grad_norm": 0.7404375461226593, "learning_rate": 9.590563056064577e-06, "loss": 0.3111, "step": 4509 }, { "epoch": 0.311787072243346, "grad_norm": 0.8416059004433211, "learning_rate": 9.590341183741603e-06, "loss": 0.3195, "step": 4510 }, { "epoch": 0.31185620463187, "grad_norm": 0.7500116213622888, "learning_rate": 9.590119253886894e-06, "loss": 0.4361, "step": 4511 }, { "epoch": 0.31192533702039404, "grad_norm": 0.8175277318099181, "learning_rate": 9.589897266503233e-06, "loss": 0.3464, "step": 4512 }, { "epoch": 0.31199446940891806, "grad_norm": 0.7403196897745675, "learning_rate": 9.5896752215934e-06, "loss": 0.3176, "step": 4513 }, { "epoch": 0.3120636017974421, "grad_norm": 0.7549698790915578, "learning_rate": 9.589453119160178e-06, "loss": 0.2994, "step": 4514 }, { "epoch": 0.3121327341859661, "grad_norm": 0.9104972125448143, "learning_rate": 9.589230959206352e-06, "loss": 0.4083, "step": 4515 }, { "epoch": 0.31220186657449017, "grad_norm": 0.8112460608511567, "learning_rate": 9.589008741734706e-06, "loss": 0.3703, "step": 4516 }, { "epoch": 0.3122709989630142, "grad_norm": 0.8095133265418611, "learning_rate": 9.588786466748024e-06, "loss": 0.3662, "step": 4517 }, { "epoch": 0.3123401313515382, "grad_norm": 0.9175781483054936, "learning_rate": 9.588564134249094e-06, "loss": 0.4276, "step": 4518 }, { "epoch": 0.3124092637400622, "grad_norm": 0.7729895821419936, "learning_rate": 9.588341744240701e-06, "loss": 0.4517, "step": 4519 }, { "epoch": 0.31247839612858624, "grad_norm": 1.2632647719357504, "learning_rate": 9.588119296725632e-06, "loss": 0.3152, "step": 4520 }, { "epoch": 0.31254752851711026, "grad_norm": 0.7643086288158685, "learning_rate": 9.587896791706675e-06, "loss": 0.31, "step": 4521 }, { "epoch": 0.3126166609056343, "grad_norm": 0.7904825122137579, "learning_rate": 9.587674229186621e-06, "loss": 0.4814, "step": 4522 }, { "epoch": 0.3126857932941583, "grad_norm": 0.7603591299969892, "learning_rate": 9.587451609168256e-06, "loss": 0.4106, "step": 4523 }, { "epoch": 0.3127549256826823, "grad_norm": 1.054104680243611, "learning_rate": 9.587228931654371e-06, "loss": 0.3157, "step": 4524 }, { "epoch": 0.31282405807120633, "grad_norm": 0.8308434153314233, "learning_rate": 9.58700619664776e-06, "loss": 0.397, "step": 4525 }, { "epoch": 0.3128931904597304, "grad_norm": 0.8035781168238523, "learning_rate": 9.586783404151211e-06, "loss": 0.4232, "step": 4526 }, { "epoch": 0.3129623228482544, "grad_norm": 0.6750391009631912, "learning_rate": 9.586560554167517e-06, "loss": 0.292, "step": 4527 }, { "epoch": 0.31303145523677844, "grad_norm": 0.8784174294459475, "learning_rate": 9.586337646699473e-06, "loss": 0.3008, "step": 4528 }, { "epoch": 0.31310058762530246, "grad_norm": 0.785585146392669, "learning_rate": 9.58611468174987e-06, "loss": 0.4419, "step": 4529 }, { "epoch": 0.3131697200138265, "grad_norm": 0.8154203365691494, "learning_rate": 9.585891659321504e-06, "loss": 0.4153, "step": 4530 }, { "epoch": 0.3132388524023505, "grad_norm": 0.6891345014803097, "learning_rate": 9.58566857941717e-06, "loss": 0.2575, "step": 4531 }, { "epoch": 0.3133079847908745, "grad_norm": 0.7462830149661814, "learning_rate": 9.585445442039664e-06, "loss": 0.3562, "step": 4532 }, { "epoch": 0.31337711717939853, "grad_norm": 0.8853443714029812, "learning_rate": 9.585222247191783e-06, "loss": 0.4406, "step": 4533 }, { "epoch": 0.31344624956792255, "grad_norm": 0.8209652770180013, "learning_rate": 9.584998994876323e-06, "loss": 0.3746, "step": 4534 }, { "epoch": 0.31351538195644657, "grad_norm": 0.8018365789338857, "learning_rate": 9.584775685096082e-06, "loss": 0.2915, "step": 4535 }, { "epoch": 0.31358451434497064, "grad_norm": 0.8132575538192772, "learning_rate": 9.584552317853862e-06, "loss": 0.4504, "step": 4536 }, { "epoch": 0.31365364673349466, "grad_norm": 0.8454121794452585, "learning_rate": 9.584328893152459e-06, "loss": 0.4264, "step": 4537 }, { "epoch": 0.3137227791220187, "grad_norm": 0.8260700741876896, "learning_rate": 9.584105410994673e-06, "loss": 0.3472, "step": 4538 }, { "epoch": 0.3137919115105427, "grad_norm": 0.8391216117441915, "learning_rate": 9.58388187138331e-06, "loss": 0.3834, "step": 4539 }, { "epoch": 0.3138610438990667, "grad_norm": 1.671485449339165, "learning_rate": 9.583658274321164e-06, "loss": 0.4579, "step": 4540 }, { "epoch": 0.31393017628759073, "grad_norm": 0.8812278864865598, "learning_rate": 9.583434619811043e-06, "loss": 0.3549, "step": 4541 }, { "epoch": 0.31399930867611475, "grad_norm": 0.873008975919318, "learning_rate": 9.58321090785575e-06, "loss": 0.349, "step": 4542 }, { "epoch": 0.31406844106463877, "grad_norm": 1.5523847862937565, "learning_rate": 9.582987138458087e-06, "loss": 0.3777, "step": 4543 }, { "epoch": 0.3141375734531628, "grad_norm": 0.8864552692642911, "learning_rate": 9.582763311620858e-06, "loss": 0.4261, "step": 4544 }, { "epoch": 0.3142067058416868, "grad_norm": 0.731713686262479, "learning_rate": 9.582539427346869e-06, "loss": 0.3464, "step": 4545 }, { "epoch": 0.3142758382302109, "grad_norm": 0.7488877520027739, "learning_rate": 9.582315485638926e-06, "loss": 0.3415, "step": 4546 }, { "epoch": 0.3143449706187349, "grad_norm": 0.8044512588979088, "learning_rate": 9.582091486499836e-06, "loss": 0.4528, "step": 4547 }, { "epoch": 0.3144141030072589, "grad_norm": 0.8044647137252493, "learning_rate": 9.581867429932406e-06, "loss": 0.3771, "step": 4548 }, { "epoch": 0.31448323539578293, "grad_norm": 0.769513040266757, "learning_rate": 9.581643315939446e-06, "loss": 0.3176, "step": 4549 }, { "epoch": 0.31455236778430695, "grad_norm": 0.9243359476630385, "learning_rate": 9.581419144523763e-06, "loss": 0.3482, "step": 4550 }, { "epoch": 0.31462150017283097, "grad_norm": 0.8906228378505535, "learning_rate": 9.581194915688165e-06, "loss": 0.446, "step": 4551 }, { "epoch": 0.314690632561355, "grad_norm": 0.7574230129819046, "learning_rate": 9.580970629435467e-06, "loss": 0.3254, "step": 4552 }, { "epoch": 0.314759764949879, "grad_norm": 0.7825563509910878, "learning_rate": 9.580746285768476e-06, "loss": 0.2997, "step": 4553 }, { "epoch": 0.314828897338403, "grad_norm": 1.1462433085470003, "learning_rate": 9.580521884690004e-06, "loss": 0.4263, "step": 4554 }, { "epoch": 0.31489802972692704, "grad_norm": 0.8363921235798389, "learning_rate": 9.580297426202866e-06, "loss": 0.3984, "step": 4555 }, { "epoch": 0.3149671621154511, "grad_norm": 1.012055487092333, "learning_rate": 9.580072910309873e-06, "loss": 0.3424, "step": 4556 }, { "epoch": 0.31503629450397513, "grad_norm": 0.713886120823294, "learning_rate": 9.57984833701384e-06, "loss": 0.3358, "step": 4557 }, { "epoch": 0.31510542689249915, "grad_norm": 0.7515847030268953, "learning_rate": 9.579623706317581e-06, "loss": 0.4221, "step": 4558 }, { "epoch": 0.31517455928102317, "grad_norm": 1.0265848857443094, "learning_rate": 9.579399018223911e-06, "loss": 0.3875, "step": 4559 }, { "epoch": 0.3152436916695472, "grad_norm": 0.892528713300903, "learning_rate": 9.579174272735648e-06, "loss": 0.3467, "step": 4560 }, { "epoch": 0.3153128240580712, "grad_norm": 0.75037827662751, "learning_rate": 9.578949469855606e-06, "loss": 0.4259, "step": 4561 }, { "epoch": 0.3153819564465952, "grad_norm": 0.8100678842609549, "learning_rate": 9.578724609586604e-06, "loss": 0.4177, "step": 4562 }, { "epoch": 0.31545108883511924, "grad_norm": 0.7939553374090739, "learning_rate": 9.57849969193146e-06, "loss": 0.3103, "step": 4563 }, { "epoch": 0.31552022122364326, "grad_norm": 0.8979695092563119, "learning_rate": 9.578274716892994e-06, "loss": 0.3222, "step": 4564 }, { "epoch": 0.3155893536121673, "grad_norm": 0.830616876804393, "learning_rate": 9.578049684474024e-06, "loss": 0.4956, "step": 4565 }, { "epoch": 0.3156584860006913, "grad_norm": 0.7146722412702293, "learning_rate": 9.577824594677372e-06, "loss": 0.3344, "step": 4566 }, { "epoch": 0.31572761838921537, "grad_norm": 0.6898926207725701, "learning_rate": 9.577599447505858e-06, "loss": 0.3259, "step": 4567 }, { "epoch": 0.3157967507777394, "grad_norm": 0.7873797721506309, "learning_rate": 9.577374242962303e-06, "loss": 0.379, "step": 4568 }, { "epoch": 0.3158658831662634, "grad_norm": 1.1482294963115665, "learning_rate": 9.577148981049531e-06, "loss": 0.461, "step": 4569 }, { "epoch": 0.3159350155547874, "grad_norm": 0.7337048209593655, "learning_rate": 9.576923661770366e-06, "loss": 0.3067, "step": 4570 }, { "epoch": 0.31600414794331144, "grad_norm": 0.7830352598616955, "learning_rate": 9.57669828512763e-06, "loss": 0.2944, "step": 4571 }, { "epoch": 0.31607328033183546, "grad_norm": 0.7463522782930753, "learning_rate": 9.576472851124148e-06, "loss": 0.4095, "step": 4572 }, { "epoch": 0.3161424127203595, "grad_norm": 0.764263842075805, "learning_rate": 9.576247359762747e-06, "loss": 0.3348, "step": 4573 }, { "epoch": 0.3162115451088835, "grad_norm": 0.8528413063913293, "learning_rate": 9.57602181104625e-06, "loss": 0.3101, "step": 4574 }, { "epoch": 0.3162806774974075, "grad_norm": 0.7897628339149837, "learning_rate": 9.575796204977487e-06, "loss": 0.4347, "step": 4575 }, { "epoch": 0.31634980988593153, "grad_norm": 0.8592278045017312, "learning_rate": 9.575570541559285e-06, "loss": 0.4523, "step": 4576 }, { "epoch": 0.3164189422744556, "grad_norm": 0.6933266854587148, "learning_rate": 9.57534482079447e-06, "loss": 0.3438, "step": 4577 }, { "epoch": 0.3164880746629796, "grad_norm": 0.7762809968942234, "learning_rate": 9.575119042685874e-06, "loss": 0.3487, "step": 4578 }, { "epoch": 0.31655720705150364, "grad_norm": 0.971261756772753, "learning_rate": 9.574893207236327e-06, "loss": 0.4494, "step": 4579 }, { "epoch": 0.31662633944002766, "grad_norm": 0.8359429773758412, "learning_rate": 9.574667314448655e-06, "loss": 0.4421, "step": 4580 }, { "epoch": 0.3166954718285517, "grad_norm": 0.7549160671790168, "learning_rate": 9.574441364325694e-06, "loss": 0.3616, "step": 4581 }, { "epoch": 0.3167646042170757, "grad_norm": 0.8328658778884616, "learning_rate": 9.574215356870273e-06, "loss": 0.3546, "step": 4582 }, { "epoch": 0.3168337366055997, "grad_norm": 0.7759929011998865, "learning_rate": 9.573989292085226e-06, "loss": 0.438, "step": 4583 }, { "epoch": 0.31690286899412373, "grad_norm": 0.6845352357582772, "learning_rate": 9.573763169973387e-06, "loss": 0.2905, "step": 4584 }, { "epoch": 0.31697200138264775, "grad_norm": 0.7799685159091483, "learning_rate": 9.573536990537587e-06, "loss": 0.2699, "step": 4585 }, { "epoch": 0.31704113377117177, "grad_norm": 0.8136540938306233, "learning_rate": 9.573310753780664e-06, "loss": 0.4492, "step": 4586 }, { "epoch": 0.31711026615969584, "grad_norm": 0.7100521084392489, "learning_rate": 9.573084459705451e-06, "loss": 0.3761, "step": 4587 }, { "epoch": 0.31717939854821986, "grad_norm": 0.7975705701419434, "learning_rate": 9.572858108314787e-06, "loss": 0.3288, "step": 4588 }, { "epoch": 0.3172485309367439, "grad_norm": 0.7322138849287712, "learning_rate": 9.572631699611505e-06, "loss": 0.2964, "step": 4589 }, { "epoch": 0.3173176633252679, "grad_norm": 0.7239766462015714, "learning_rate": 9.572405233598445e-06, "loss": 0.514, "step": 4590 }, { "epoch": 0.3173867957137919, "grad_norm": 0.8990523373606694, "learning_rate": 9.572178710278447e-06, "loss": 0.3922, "step": 4591 }, { "epoch": 0.31745592810231593, "grad_norm": 0.7502768047441345, "learning_rate": 9.571952129654347e-06, "loss": 0.2872, "step": 4592 }, { "epoch": 0.31752506049083995, "grad_norm": 1.0846037085670042, "learning_rate": 9.571725491728987e-06, "loss": 0.3573, "step": 4593 }, { "epoch": 0.31759419287936397, "grad_norm": 0.7490937473297926, "learning_rate": 9.571498796505207e-06, "loss": 0.4886, "step": 4594 }, { "epoch": 0.317663325267888, "grad_norm": 0.9196063987669247, "learning_rate": 9.571272043985846e-06, "loss": 0.2942, "step": 4595 }, { "epoch": 0.317732457656412, "grad_norm": 0.8046539457053189, "learning_rate": 9.571045234173748e-06, "loss": 0.3657, "step": 4596 }, { "epoch": 0.3178015900449361, "grad_norm": 0.8016915613726212, "learning_rate": 9.570818367071757e-06, "loss": 0.4406, "step": 4597 }, { "epoch": 0.3178707224334601, "grad_norm": 0.7710748317973376, "learning_rate": 9.570591442682713e-06, "loss": 0.4398, "step": 4598 }, { "epoch": 0.3179398548219841, "grad_norm": 0.7750232311494724, "learning_rate": 9.570364461009463e-06, "loss": 0.3081, "step": 4599 }, { "epoch": 0.31800898721050813, "grad_norm": 0.8081610350410026, "learning_rate": 9.57013742205485e-06, "loss": 0.3501, "step": 4600 }, { "epoch": 0.31807811959903215, "grad_norm": 0.7225380544509569, "learning_rate": 9.56991032582172e-06, "loss": 0.4387, "step": 4601 }, { "epoch": 0.31814725198755617, "grad_norm": 0.7536769754705616, "learning_rate": 9.56968317231292e-06, "loss": 0.3286, "step": 4602 }, { "epoch": 0.3182163843760802, "grad_norm": 0.7231331667113937, "learning_rate": 9.569455961531297e-06, "loss": 0.3153, "step": 4603 }, { "epoch": 0.3182855167646042, "grad_norm": 0.8231617029357476, "learning_rate": 9.569228693479697e-06, "loss": 0.4042, "step": 4604 }, { "epoch": 0.3183546491531282, "grad_norm": 0.7787665923046516, "learning_rate": 9.56900136816097e-06, "loss": 0.3524, "step": 4605 }, { "epoch": 0.31842378154165224, "grad_norm": 0.7395548941719374, "learning_rate": 9.568773985577964e-06, "loss": 0.3957, "step": 4606 }, { "epoch": 0.3184929139301763, "grad_norm": 0.7392364273322758, "learning_rate": 9.56854654573353e-06, "loss": 0.3391, "step": 4607 }, { "epoch": 0.31856204631870033, "grad_norm": 0.8925251383004766, "learning_rate": 9.56831904863052e-06, "loss": 0.494, "step": 4608 }, { "epoch": 0.31863117870722435, "grad_norm": 0.7658971982149835, "learning_rate": 9.568091494271781e-06, "loss": 0.3645, "step": 4609 }, { "epoch": 0.31870031109574837, "grad_norm": 0.7396398190530141, "learning_rate": 9.567863882660167e-06, "loss": 0.3267, "step": 4610 }, { "epoch": 0.3187694434842724, "grad_norm": 0.775909065542673, "learning_rate": 9.56763621379853e-06, "loss": 0.3998, "step": 4611 }, { "epoch": 0.3188385758727964, "grad_norm": 1.3418360567178054, "learning_rate": 9.567408487689727e-06, "loss": 0.4409, "step": 4612 }, { "epoch": 0.3189077082613204, "grad_norm": 0.8222762515234313, "learning_rate": 9.567180704336607e-06, "loss": 0.3637, "step": 4613 }, { "epoch": 0.31897684064984444, "grad_norm": 0.8731366286746229, "learning_rate": 9.56695286374203e-06, "loss": 0.3761, "step": 4614 }, { "epoch": 0.31904597303836846, "grad_norm": 0.823904675150371, "learning_rate": 9.566724965908846e-06, "loss": 0.4531, "step": 4615 }, { "epoch": 0.3191151054268925, "grad_norm": 1.0137356834708362, "learning_rate": 9.566497010839917e-06, "loss": 0.4083, "step": 4616 }, { "epoch": 0.31918423781541655, "grad_norm": 0.6705884316369337, "learning_rate": 9.566268998538096e-06, "loss": 0.3128, "step": 4617 }, { "epoch": 0.31925337020394057, "grad_norm": 0.7600077537355631, "learning_rate": 9.566040929006242e-06, "loss": 0.3564, "step": 4618 }, { "epoch": 0.3193225025924646, "grad_norm": 0.7390002220017091, "learning_rate": 9.565812802247214e-06, "loss": 0.3989, "step": 4619 }, { "epoch": 0.3193916349809886, "grad_norm": 0.7689520010162746, "learning_rate": 9.56558461826387e-06, "loss": 0.3407, "step": 4620 }, { "epoch": 0.3194607673695126, "grad_norm": 0.8914006169709023, "learning_rate": 9.565356377059069e-06, "loss": 0.3632, "step": 4621 }, { "epoch": 0.31952989975803664, "grad_norm": 0.7961838415020299, "learning_rate": 9.565128078635677e-06, "loss": 0.4209, "step": 4622 }, { "epoch": 0.31959903214656066, "grad_norm": 0.8431069869082702, "learning_rate": 9.564899722996547e-06, "loss": 0.3107, "step": 4623 }, { "epoch": 0.3196681645350847, "grad_norm": 1.3129627229869825, "learning_rate": 9.564671310144548e-06, "loss": 0.3678, "step": 4624 }, { "epoch": 0.3197372969236087, "grad_norm": 0.7718543089374279, "learning_rate": 9.56444284008254e-06, "loss": 0.3494, "step": 4625 }, { "epoch": 0.3198064293121327, "grad_norm": 0.9193444905086239, "learning_rate": 9.564214312813386e-06, "loss": 0.4764, "step": 4626 }, { "epoch": 0.3198755617006568, "grad_norm": 0.827806357448994, "learning_rate": 9.563985728339951e-06, "loss": 0.3872, "step": 4627 }, { "epoch": 0.3199446940891808, "grad_norm": 0.7552203516396727, "learning_rate": 9.5637570866651e-06, "loss": 0.322, "step": 4628 }, { "epoch": 0.3200138264777048, "grad_norm": 0.7242581573263633, "learning_rate": 9.563528387791696e-06, "loss": 0.3653, "step": 4629 }, { "epoch": 0.32008295886622884, "grad_norm": 0.7532722163984739, "learning_rate": 9.563299631722611e-06, "loss": 0.4518, "step": 4630 }, { "epoch": 0.32015209125475286, "grad_norm": 0.7451622671708356, "learning_rate": 9.563070818460706e-06, "loss": 0.2764, "step": 4631 }, { "epoch": 0.3202212236432769, "grad_norm": 0.8493454450355761, "learning_rate": 9.562841948008854e-06, "loss": 0.4345, "step": 4632 }, { "epoch": 0.3202903560318009, "grad_norm": 0.8684994699267321, "learning_rate": 9.56261302036992e-06, "loss": 0.4413, "step": 4633 }, { "epoch": 0.3203594884203249, "grad_norm": 0.844557108354662, "learning_rate": 9.562384035546772e-06, "loss": 0.3225, "step": 4634 }, { "epoch": 0.32042862080884893, "grad_norm": 0.7457723501439701, "learning_rate": 9.562154993542285e-06, "loss": 0.2794, "step": 4635 }, { "epoch": 0.32049775319737295, "grad_norm": 0.9769917450936338, "learning_rate": 9.561925894359325e-06, "loss": 0.4474, "step": 4636 }, { "epoch": 0.32056688558589697, "grad_norm": 0.7856085396494393, "learning_rate": 9.561696738000765e-06, "loss": 0.393, "step": 4637 }, { "epoch": 0.32063601797442104, "grad_norm": 0.7715718813467871, "learning_rate": 9.561467524469478e-06, "loss": 0.2779, "step": 4638 }, { "epoch": 0.32070515036294506, "grad_norm": 0.6682823544690438, "learning_rate": 9.561238253768335e-06, "loss": 0.3262, "step": 4639 }, { "epoch": 0.3207742827514691, "grad_norm": 0.7914615542136744, "learning_rate": 9.561008925900211e-06, "loss": 0.4322, "step": 4640 }, { "epoch": 0.3208434151399931, "grad_norm": 1.003937772604062, "learning_rate": 9.56077954086798e-06, "loss": 0.361, "step": 4641 }, { "epoch": 0.3209125475285171, "grad_norm": 0.7602979486256909, "learning_rate": 9.560550098674517e-06, "loss": 0.4034, "step": 4642 }, { "epoch": 0.32098167991704113, "grad_norm": 0.8922370701536696, "learning_rate": 9.560320599322695e-06, "loss": 0.4376, "step": 4643 }, { "epoch": 0.32105081230556515, "grad_norm": 0.7911097085532263, "learning_rate": 9.560091042815395e-06, "loss": 0.4661, "step": 4644 }, { "epoch": 0.32111994469408917, "grad_norm": 1.3037036994730824, "learning_rate": 9.55986142915549e-06, "loss": 0.3014, "step": 4645 }, { "epoch": 0.3211890770826132, "grad_norm": 0.7973042397718774, "learning_rate": 9.559631758345859e-06, "loss": 0.3714, "step": 4646 }, { "epoch": 0.3212582094711372, "grad_norm": 0.7264719195486002, "learning_rate": 9.559402030389381e-06, "loss": 0.4236, "step": 4647 }, { "epoch": 0.3213273418596613, "grad_norm": 0.7754557793993994, "learning_rate": 9.559172245288937e-06, "loss": 0.4306, "step": 4648 }, { "epoch": 0.3213964742481853, "grad_norm": 1.109069498305677, "learning_rate": 9.558942403047405e-06, "loss": 0.3132, "step": 4649 }, { "epoch": 0.3214656066367093, "grad_norm": 0.7728590999329202, "learning_rate": 9.558712503667664e-06, "loss": 0.3205, "step": 4650 }, { "epoch": 0.32153473902523333, "grad_norm": 0.8459195943299842, "learning_rate": 9.558482547152598e-06, "loss": 0.4644, "step": 4651 }, { "epoch": 0.32160387141375735, "grad_norm": 1.0980725925925159, "learning_rate": 9.558252533505088e-06, "loss": 0.3548, "step": 4652 }, { "epoch": 0.32167300380228137, "grad_norm": 0.9179258044380616, "learning_rate": 9.558022462728018e-06, "loss": 0.3147, "step": 4653 }, { "epoch": 0.3217421361908054, "grad_norm": 0.803361463270193, "learning_rate": 9.55779233482427e-06, "loss": 0.3732, "step": 4654 }, { "epoch": 0.3218112685793294, "grad_norm": 0.859938767005537, "learning_rate": 9.557562149796729e-06, "loss": 0.3543, "step": 4655 }, { "epoch": 0.3218804009678534, "grad_norm": 0.7862344740719301, "learning_rate": 9.557331907648279e-06, "loss": 0.2575, "step": 4656 }, { "epoch": 0.32194953335637744, "grad_norm": 0.8670608597422903, "learning_rate": 9.557101608381807e-06, "loss": 0.3519, "step": 4657 }, { "epoch": 0.3220186657449015, "grad_norm": 0.7769836611592567, "learning_rate": 9.556871252000198e-06, "loss": 0.4508, "step": 4658 }, { "epoch": 0.32208779813342553, "grad_norm": 0.7777074834808818, "learning_rate": 9.556640838506341e-06, "loss": 0.4027, "step": 4659 }, { "epoch": 0.32215693052194955, "grad_norm": 0.7330886686127572, "learning_rate": 9.556410367903122e-06, "loss": 0.2879, "step": 4660 }, { "epoch": 0.32222606291047357, "grad_norm": 0.8230622901409942, "learning_rate": 9.556179840193432e-06, "loss": 0.4666, "step": 4661 }, { "epoch": 0.3222951952989976, "grad_norm": 0.834961827792565, "learning_rate": 9.555949255380158e-06, "loss": 0.4242, "step": 4662 }, { "epoch": 0.3223643276875216, "grad_norm": 0.8869316930743807, "learning_rate": 9.555718613466189e-06, "loss": 0.3523, "step": 4663 }, { "epoch": 0.3224334600760456, "grad_norm": 0.8410371307426222, "learning_rate": 9.555487914454418e-06, "loss": 0.339, "step": 4664 }, { "epoch": 0.32250259246456964, "grad_norm": 0.8912835355661287, "learning_rate": 9.555257158347736e-06, "loss": 0.4315, "step": 4665 }, { "epoch": 0.32257172485309366, "grad_norm": 0.7832174233580541, "learning_rate": 9.555026345149035e-06, "loss": 0.3399, "step": 4666 }, { "epoch": 0.3226408572416177, "grad_norm": 0.7081657562502317, "learning_rate": 9.554795474861207e-06, "loss": 0.3228, "step": 4667 }, { "epoch": 0.32270998963014175, "grad_norm": 0.8649180964087744, "learning_rate": 9.554564547487147e-06, "loss": 0.3415, "step": 4668 }, { "epoch": 0.32277912201866577, "grad_norm": 0.8083424643608078, "learning_rate": 9.554333563029747e-06, "loss": 0.4298, "step": 4669 }, { "epoch": 0.3228482544071898, "grad_norm": 1.4449404564157733, "learning_rate": 9.554102521491903e-06, "loss": 0.2981, "step": 4670 }, { "epoch": 0.3229173867957138, "grad_norm": 4.467107118681821, "learning_rate": 9.553871422876512e-06, "loss": 0.3104, "step": 4671 }, { "epoch": 0.3229865191842378, "grad_norm": 0.7808192078629328, "learning_rate": 9.553640267186471e-06, "loss": 0.4499, "step": 4672 }, { "epoch": 0.32305565157276184, "grad_norm": 0.735818862504435, "learning_rate": 9.553409054424674e-06, "loss": 0.4224, "step": 4673 }, { "epoch": 0.32312478396128586, "grad_norm": 0.7672636124854764, "learning_rate": 9.55317778459402e-06, "loss": 0.3317, "step": 4674 }, { "epoch": 0.3231939163498099, "grad_norm": 0.8165008278135761, "learning_rate": 9.552946457697408e-06, "loss": 0.4202, "step": 4675 }, { "epoch": 0.3232630487383339, "grad_norm": 0.8757126610243327, "learning_rate": 9.552715073737737e-06, "loss": 0.3723, "step": 4676 }, { "epoch": 0.3233321811268579, "grad_norm": 0.6929704537655323, "learning_rate": 9.552483632717908e-06, "loss": 0.3312, "step": 4677 }, { "epoch": 0.323401313515382, "grad_norm": 0.7943468288197576, "learning_rate": 9.552252134640822e-06, "loss": 0.3142, "step": 4678 }, { "epoch": 0.323470445903906, "grad_norm": 0.8182871673101431, "learning_rate": 9.552020579509377e-06, "loss": 0.3958, "step": 4679 }, { "epoch": 0.32353957829243, "grad_norm": 0.9740971992650778, "learning_rate": 9.55178896732648e-06, "loss": 0.4209, "step": 4680 }, { "epoch": 0.32360871068095404, "grad_norm": 0.8169310673333176, "learning_rate": 9.55155729809503e-06, "loss": 0.3781, "step": 4681 }, { "epoch": 0.32367784306947806, "grad_norm": 0.8762661219132417, "learning_rate": 9.551325571817932e-06, "loss": 0.3937, "step": 4682 }, { "epoch": 0.3237469754580021, "grad_norm": 0.7302437834806123, "learning_rate": 9.55109378849809e-06, "loss": 0.389, "step": 4683 }, { "epoch": 0.3238161078465261, "grad_norm": 0.7082396342841574, "learning_rate": 9.55086194813841e-06, "loss": 0.3538, "step": 4684 }, { "epoch": 0.3238852402350501, "grad_norm": 0.7142306811949262, "learning_rate": 9.550630050741797e-06, "loss": 0.3254, "step": 4685 }, { "epoch": 0.3239543726235741, "grad_norm": 0.774944106223892, "learning_rate": 9.550398096311155e-06, "loss": 0.4263, "step": 4686 }, { "epoch": 0.32402350501209815, "grad_norm": 0.8144878450949232, "learning_rate": 9.550166084849395e-06, "loss": 0.4216, "step": 4687 }, { "epoch": 0.3240926374006222, "grad_norm": 0.8003695624183792, "learning_rate": 9.549934016359423e-06, "loss": 0.314, "step": 4688 }, { "epoch": 0.32416176978914624, "grad_norm": 0.7631040955848698, "learning_rate": 9.549701890844148e-06, "loss": 0.3019, "step": 4689 }, { "epoch": 0.32423090217767025, "grad_norm": 0.8635605702629702, "learning_rate": 9.54946970830648e-06, "loss": 0.4647, "step": 4690 }, { "epoch": 0.3243000345661943, "grad_norm": 0.6932875159091859, "learning_rate": 9.549237468749327e-06, "loss": 0.3177, "step": 4691 }, { "epoch": 0.3243691669547183, "grad_norm": 0.80597197964628, "learning_rate": 9.549005172175601e-06, "loss": 0.3294, "step": 4692 }, { "epoch": 0.3244382993432423, "grad_norm": 0.7858426513736735, "learning_rate": 9.548772818588213e-06, "loss": 0.334, "step": 4693 }, { "epoch": 0.3245074317317663, "grad_norm": 0.8067637232166162, "learning_rate": 9.548540407990075e-06, "loss": 0.407, "step": 4694 }, { "epoch": 0.32457656412029035, "grad_norm": 1.111581254016381, "learning_rate": 9.5483079403841e-06, "loss": 0.3075, "step": 4695 }, { "epoch": 0.32464569650881436, "grad_norm": 0.7413773525583276, "learning_rate": 9.548075415773204e-06, "loss": 0.2814, "step": 4696 }, { "epoch": 0.3247148288973384, "grad_norm": 0.8407925394284328, "learning_rate": 9.547842834160296e-06, "loss": 0.4744, "step": 4697 }, { "epoch": 0.3247839612858624, "grad_norm": 0.8114631239810649, "learning_rate": 9.547610195548295e-06, "loss": 0.3426, "step": 4698 }, { "epoch": 0.3248530936743865, "grad_norm": 0.7616475713032678, "learning_rate": 9.547377499940118e-06, "loss": 0.3424, "step": 4699 }, { "epoch": 0.3249222260629105, "grad_norm": 0.9825813255594615, "learning_rate": 9.547144747338676e-06, "loss": 0.3926, "step": 4700 }, { "epoch": 0.3249913584514345, "grad_norm": 0.8474713519086271, "learning_rate": 9.546911937746893e-06, "loss": 0.4342, "step": 4701 }, { "epoch": 0.3250604908399585, "grad_norm": 0.7466414028166451, "learning_rate": 9.54667907116768e-06, "loss": 0.3344, "step": 4702 }, { "epoch": 0.32512962322848254, "grad_norm": 0.7250405838472583, "learning_rate": 9.546446147603961e-06, "loss": 0.3073, "step": 4703 }, { "epoch": 0.32519875561700656, "grad_norm": 0.7625186805262005, "learning_rate": 9.54621316705865e-06, "loss": 0.3829, "step": 4704 }, { "epoch": 0.3252678880055306, "grad_norm": 0.7640159312341358, "learning_rate": 9.545980129534673e-06, "loss": 0.4603, "step": 4705 }, { "epoch": 0.3253370203940546, "grad_norm": 0.760374020246332, "learning_rate": 9.545747035034947e-06, "loss": 0.2955, "step": 4706 }, { "epoch": 0.3254061527825786, "grad_norm": 0.8151301542524366, "learning_rate": 9.545513883562394e-06, "loss": 0.337, "step": 4707 }, { "epoch": 0.32547528517110264, "grad_norm": 1.008212532017342, "learning_rate": 9.545280675119937e-06, "loss": 0.4695, "step": 4708 }, { "epoch": 0.3255444175596267, "grad_norm": 0.9309296645781261, "learning_rate": 9.545047409710497e-06, "loss": 0.3427, "step": 4709 }, { "epoch": 0.3256135499481507, "grad_norm": 0.795196425638925, "learning_rate": 9.544814087336998e-06, "loss": 0.2901, "step": 4710 }, { "epoch": 0.32568268233667474, "grad_norm": 1.0330947912293504, "learning_rate": 9.544580708002367e-06, "loss": 0.426, "step": 4711 }, { "epoch": 0.32575181472519876, "grad_norm": 0.8419193677044051, "learning_rate": 9.544347271709526e-06, "loss": 0.423, "step": 4712 }, { "epoch": 0.3258209471137228, "grad_norm": 0.805226018228183, "learning_rate": 9.544113778461401e-06, "loss": 0.3109, "step": 4713 }, { "epoch": 0.3258900795022468, "grad_norm": 0.7904227166419229, "learning_rate": 9.54388022826092e-06, "loss": 0.3727, "step": 4714 }, { "epoch": 0.3259592118907708, "grad_norm": 0.8322703823628895, "learning_rate": 9.543646621111009e-06, "loss": 0.4172, "step": 4715 }, { "epoch": 0.32602834427929483, "grad_norm": 0.8915474707322744, "learning_rate": 9.543412957014597e-06, "loss": 0.3747, "step": 4716 }, { "epoch": 0.32609747666781885, "grad_norm": 0.7660200414186449, "learning_rate": 9.543179235974612e-06, "loss": 0.3599, "step": 4717 }, { "epoch": 0.32616660905634287, "grad_norm": 0.7971166058714667, "learning_rate": 9.54294545799398e-06, "loss": 0.4022, "step": 4718 }, { "epoch": 0.32623574144486694, "grad_norm": 0.7222088257605286, "learning_rate": 9.542711623075636e-06, "loss": 0.3957, "step": 4719 }, { "epoch": 0.32630487383339096, "grad_norm": 0.8039912115828688, "learning_rate": 9.542477731222509e-06, "loss": 0.3139, "step": 4720 }, { "epoch": 0.326374006221915, "grad_norm": 0.8935308517469497, "learning_rate": 9.54224378243753e-06, "loss": 0.3616, "step": 4721 }, { "epoch": 0.326443138610439, "grad_norm": 0.7421415138784584, "learning_rate": 9.542009776723631e-06, "loss": 0.4281, "step": 4722 }, { "epoch": 0.326512270998963, "grad_norm": 0.7981877912065585, "learning_rate": 9.541775714083744e-06, "loss": 0.4526, "step": 4723 }, { "epoch": 0.32658140338748703, "grad_norm": 0.8067423052276569, "learning_rate": 9.541541594520805e-06, "loss": 0.3356, "step": 4724 }, { "epoch": 0.32665053577601105, "grad_norm": 0.7522649552612888, "learning_rate": 9.541307418037746e-06, "loss": 0.3155, "step": 4725 }, { "epoch": 0.32671966816453507, "grad_norm": 0.7841789433938815, "learning_rate": 9.541073184637503e-06, "loss": 0.4138, "step": 4726 }, { "epoch": 0.3267888005530591, "grad_norm": 0.7619662586315125, "learning_rate": 9.540838894323011e-06, "loss": 0.3439, "step": 4727 }, { "epoch": 0.3268579329415831, "grad_norm": 0.7077666784540515, "learning_rate": 9.540604547097207e-06, "loss": 0.3686, "step": 4728 }, { "epoch": 0.3269270653301072, "grad_norm": 0.7555705084693433, "learning_rate": 9.54037014296303e-06, "loss": 0.4326, "step": 4729 }, { "epoch": 0.3269961977186312, "grad_norm": 0.8179948244396683, "learning_rate": 9.540135681923413e-06, "loss": 0.389, "step": 4730 }, { "epoch": 0.3270653301071552, "grad_norm": 0.8954793689015392, "learning_rate": 9.5399011639813e-06, "loss": 0.2944, "step": 4731 }, { "epoch": 0.32713446249567923, "grad_norm": 0.7621012923605405, "learning_rate": 9.539666589139626e-06, "loss": 0.3923, "step": 4732 }, { "epoch": 0.32720359488420325, "grad_norm": 0.7656922864801723, "learning_rate": 9.539431957401334e-06, "loss": 0.4508, "step": 4733 }, { "epoch": 0.32727272727272727, "grad_norm": 0.811412983206125, "learning_rate": 9.539197268769362e-06, "loss": 0.329, "step": 4734 }, { "epoch": 0.3273418596612513, "grad_norm": 0.7387298403952207, "learning_rate": 9.538962523246654e-06, "loss": 0.3062, "step": 4735 }, { "epoch": 0.3274109920497753, "grad_norm": 0.8530762724974627, "learning_rate": 9.53872772083615e-06, "loss": 0.4411, "step": 4736 }, { "epoch": 0.3274801244382993, "grad_norm": 0.7375329000937945, "learning_rate": 9.538492861540794e-06, "loss": 0.3802, "step": 4737 }, { "epoch": 0.32754925682682334, "grad_norm": 2.0785588705062583, "learning_rate": 9.53825794536353e-06, "loss": 0.3115, "step": 4738 }, { "epoch": 0.3276183892153474, "grad_norm": 0.72831167286397, "learning_rate": 9.538022972307302e-06, "loss": 0.3532, "step": 4739 }, { "epoch": 0.32768752160387143, "grad_norm": 0.7532547341102381, "learning_rate": 9.537787942375054e-06, "loss": 0.4699, "step": 4740 }, { "epoch": 0.32775665399239545, "grad_norm": 0.7478082829795243, "learning_rate": 9.53755285556973e-06, "loss": 0.4089, "step": 4741 }, { "epoch": 0.32782578638091947, "grad_norm": 0.7270751007029065, "learning_rate": 9.537317711894281e-06, "loss": 0.2472, "step": 4742 }, { "epoch": 0.3278949187694435, "grad_norm": 0.8028399444754764, "learning_rate": 9.537082511351652e-06, "loss": 0.4102, "step": 4743 }, { "epoch": 0.3279640511579675, "grad_norm": 0.7220523695924412, "learning_rate": 9.536847253944789e-06, "loss": 0.4643, "step": 4744 }, { "epoch": 0.3280331835464915, "grad_norm": 0.793168115831717, "learning_rate": 9.536611939676642e-06, "loss": 0.328, "step": 4745 }, { "epoch": 0.32810231593501554, "grad_norm": 0.8464200202173929, "learning_rate": 9.536376568550161e-06, "loss": 0.2987, "step": 4746 }, { "epoch": 0.32817144832353956, "grad_norm": 0.8248756092523337, "learning_rate": 9.536141140568296e-06, "loss": 0.3975, "step": 4747 }, { "epoch": 0.3282405807120636, "grad_norm": 0.7530181994984386, "learning_rate": 9.535905655733995e-06, "loss": 0.326, "step": 4748 }, { "epoch": 0.32830971310058765, "grad_norm": 0.7096699806014656, "learning_rate": 9.535670114050214e-06, "loss": 0.3375, "step": 4749 }, { "epoch": 0.32837884548911167, "grad_norm": 0.7451185013813562, "learning_rate": 9.5354345155199e-06, "loss": 0.3166, "step": 4750 }, { "epoch": 0.3284479778776357, "grad_norm": 0.7165367006719608, "learning_rate": 9.535198860146008e-06, "loss": 0.4636, "step": 4751 }, { "epoch": 0.3285171102661597, "grad_norm": 0.6401385922296585, "learning_rate": 9.534963147931494e-06, "loss": 0.3133, "step": 4752 }, { "epoch": 0.3285862426546837, "grad_norm": 1.079244503652805, "learning_rate": 9.534727378879307e-06, "loss": 0.2876, "step": 4753 }, { "epoch": 0.32865537504320774, "grad_norm": 0.7662735505563859, "learning_rate": 9.534491552992407e-06, "loss": 0.3786, "step": 4754 }, { "epoch": 0.32872450743173176, "grad_norm": 0.8366495605655306, "learning_rate": 9.534255670273747e-06, "loss": 0.3391, "step": 4755 }, { "epoch": 0.3287936398202558, "grad_norm": 0.8505137211714355, "learning_rate": 9.534019730726283e-06, "loss": 0.3184, "step": 4756 }, { "epoch": 0.3288627722087798, "grad_norm": 0.7680797363212459, "learning_rate": 9.533783734352974e-06, "loss": 0.328, "step": 4757 }, { "epoch": 0.3289319045973038, "grad_norm": 0.7481039817567137, "learning_rate": 9.533547681156776e-06, "loss": 0.4211, "step": 4758 }, { "epoch": 0.32900103698582783, "grad_norm": 0.9845812004444546, "learning_rate": 9.53331157114065e-06, "loss": 0.3517, "step": 4759 }, { "epoch": 0.3290701693743519, "grad_norm": 0.75312637858697, "learning_rate": 9.533075404307552e-06, "loss": 0.2864, "step": 4760 }, { "epoch": 0.3291393017628759, "grad_norm": 1.0512907028521514, "learning_rate": 9.532839180660445e-06, "loss": 0.3586, "step": 4761 }, { "epoch": 0.32920843415139994, "grad_norm": 0.735228046294096, "learning_rate": 9.532602900202286e-06, "loss": 0.3842, "step": 4762 }, { "epoch": 0.32927756653992396, "grad_norm": 0.7662144290669891, "learning_rate": 9.53236656293604e-06, "loss": 0.2381, "step": 4763 }, { "epoch": 0.329346698928448, "grad_norm": 0.7653163846056199, "learning_rate": 9.532130168864666e-06, "loss": 0.4016, "step": 4764 }, { "epoch": 0.329415831316972, "grad_norm": 0.83154380775249, "learning_rate": 9.53189371799113e-06, "loss": 0.4488, "step": 4765 }, { "epoch": 0.329484963705496, "grad_norm": 0.7294037007457443, "learning_rate": 9.531657210318392e-06, "loss": 0.3751, "step": 4766 }, { "epoch": 0.32955409609402003, "grad_norm": 0.7151988203941304, "learning_rate": 9.53142064584942e-06, "loss": 0.3264, "step": 4767 }, { "epoch": 0.32962322848254405, "grad_norm": 0.7330821369624628, "learning_rate": 9.531184024587175e-06, "loss": 0.311, "step": 4768 }, { "epoch": 0.32969236087106807, "grad_norm": 0.9095183119189597, "learning_rate": 9.530947346534627e-06, "loss": 0.4202, "step": 4769 }, { "epoch": 0.32976149325959214, "grad_norm": 0.8099710113199612, "learning_rate": 9.530710611694739e-06, "loss": 0.2793, "step": 4770 }, { "epoch": 0.32983062564811616, "grad_norm": 0.7982320710348259, "learning_rate": 9.530473820070476e-06, "loss": 0.3222, "step": 4771 }, { "epoch": 0.3298997580366402, "grad_norm": 0.8291108135511601, "learning_rate": 9.530236971664812e-06, "loss": 0.519, "step": 4772 }, { "epoch": 0.3299688904251642, "grad_norm": 1.0355805383555754, "learning_rate": 9.530000066480711e-06, "loss": 0.329, "step": 4773 }, { "epoch": 0.3300380228136882, "grad_norm": 0.9213279657094824, "learning_rate": 9.529763104521142e-06, "loss": 0.3133, "step": 4774 }, { "epoch": 0.33010715520221223, "grad_norm": 0.8105969942041559, "learning_rate": 9.529526085789078e-06, "loss": 0.4399, "step": 4775 }, { "epoch": 0.33017628759073625, "grad_norm": 0.7340501078594082, "learning_rate": 9.529289010287488e-06, "loss": 0.4162, "step": 4776 }, { "epoch": 0.33024541997926027, "grad_norm": 0.7635415548018551, "learning_rate": 9.529051878019341e-06, "loss": 0.3742, "step": 4777 }, { "epoch": 0.3303145523677843, "grad_norm": 0.8436661174067377, "learning_rate": 9.528814688987613e-06, "loss": 0.3162, "step": 4778 }, { "epoch": 0.3303836847563083, "grad_norm": 0.7936452416916702, "learning_rate": 9.528577443195274e-06, "loss": 0.3935, "step": 4779 }, { "epoch": 0.3304528171448324, "grad_norm": 0.7229818074398471, "learning_rate": 9.5283401406453e-06, "loss": 0.3564, "step": 4780 }, { "epoch": 0.3305219495333564, "grad_norm": 0.7394358962241022, "learning_rate": 9.528102781340662e-06, "loss": 0.3537, "step": 4781 }, { "epoch": 0.3305910819218804, "grad_norm": 0.6423848546147627, "learning_rate": 9.527865365284337e-06, "loss": 0.3326, "step": 4782 }, { "epoch": 0.33066021431040443, "grad_norm": 0.7879178536049168, "learning_rate": 9.527627892479298e-06, "loss": 0.5185, "step": 4783 }, { "epoch": 0.33072934669892845, "grad_norm": 0.7744834866962379, "learning_rate": 9.527390362928526e-06, "loss": 0.3085, "step": 4784 }, { "epoch": 0.33079847908745247, "grad_norm": 0.8199140936655336, "learning_rate": 9.527152776634995e-06, "loss": 0.2811, "step": 4785 }, { "epoch": 0.3308676114759765, "grad_norm": 0.7185256872288217, "learning_rate": 9.526915133601681e-06, "loss": 0.4004, "step": 4786 }, { "epoch": 0.3309367438645005, "grad_norm": 0.7344916628114332, "learning_rate": 9.526677433831567e-06, "loss": 0.4167, "step": 4787 }, { "epoch": 0.3310058762530245, "grad_norm": 0.7213221319013529, "learning_rate": 9.526439677327629e-06, "loss": 0.3012, "step": 4788 }, { "epoch": 0.33107500864154854, "grad_norm": 1.108855620264869, "learning_rate": 9.526201864092846e-06, "loss": 0.3733, "step": 4789 }, { "epoch": 0.3311441410300726, "grad_norm": 0.8302905800417689, "learning_rate": 9.5259639941302e-06, "loss": 0.5008, "step": 4790 }, { "epoch": 0.33121327341859663, "grad_norm": 1.066274829514482, "learning_rate": 9.525726067442673e-06, "loss": 0.3599, "step": 4791 }, { "epoch": 0.33128240580712065, "grad_norm": 0.7442495357406966, "learning_rate": 9.525488084033247e-06, "loss": 0.3299, "step": 4792 }, { "epoch": 0.33135153819564467, "grad_norm": 0.8127375214680894, "learning_rate": 9.525250043904904e-06, "loss": 0.3458, "step": 4793 }, { "epoch": 0.3314206705841687, "grad_norm": 0.7357098344488289, "learning_rate": 9.525011947060626e-06, "loss": 0.4655, "step": 4794 }, { "epoch": 0.3314898029726927, "grad_norm": 0.8097273075777439, "learning_rate": 9.5247737935034e-06, "loss": 0.3473, "step": 4795 }, { "epoch": 0.3315589353612167, "grad_norm": 0.7314414488228415, "learning_rate": 9.524535583236207e-06, "loss": 0.28, "step": 4796 }, { "epoch": 0.33162806774974074, "grad_norm": 0.8395823421563396, "learning_rate": 9.524297316262036e-06, "loss": 0.4145, "step": 4797 }, { "epoch": 0.33169720013826476, "grad_norm": 0.797724012248134, "learning_rate": 9.524058992583873e-06, "loss": 0.3564, "step": 4798 }, { "epoch": 0.3317663325267888, "grad_norm": 0.9694024982147156, "learning_rate": 9.523820612204704e-06, "loss": 0.2887, "step": 4799 }, { "epoch": 0.33183546491531285, "grad_norm": 0.7970682709244459, "learning_rate": 9.523582175127517e-06, "loss": 0.3243, "step": 4800 }, { "epoch": 0.33190459730383687, "grad_norm": 0.8669161804998796, "learning_rate": 9.523343681355298e-06, "loss": 0.4268, "step": 4801 }, { "epoch": 0.3319737296923609, "grad_norm": 0.8037354111509768, "learning_rate": 9.52310513089104e-06, "loss": 0.3426, "step": 4802 }, { "epoch": 0.3320428620808849, "grad_norm": 0.7794717869658422, "learning_rate": 9.52286652373773e-06, "loss": 0.2929, "step": 4803 }, { "epoch": 0.3321119944694089, "grad_norm": 0.8364309468071262, "learning_rate": 9.522627859898361e-06, "loss": 0.4178, "step": 4804 }, { "epoch": 0.33218112685793294, "grad_norm": 0.9546959934883821, "learning_rate": 9.522389139375923e-06, "loss": 0.4046, "step": 4805 }, { "epoch": 0.33225025924645696, "grad_norm": 0.7327084395885465, "learning_rate": 9.522150362173406e-06, "loss": 0.346, "step": 4806 }, { "epoch": 0.332319391634981, "grad_norm": 0.763997412579491, "learning_rate": 9.521911528293805e-06, "loss": 0.3696, "step": 4807 }, { "epoch": 0.332388524023505, "grad_norm": 0.7371038880769112, "learning_rate": 9.521672637740112e-06, "loss": 0.3875, "step": 4808 }, { "epoch": 0.332457656412029, "grad_norm": 1.648192979462204, "learning_rate": 9.521433690515323e-06, "loss": 0.2802, "step": 4809 }, { "epoch": 0.3325267888005531, "grad_norm": 0.7321577167648584, "learning_rate": 9.521194686622432e-06, "loss": 0.3371, "step": 4810 }, { "epoch": 0.3325959211890771, "grad_norm": 0.7604504099449529, "learning_rate": 9.520955626064433e-06, "loss": 0.427, "step": 4811 }, { "epoch": 0.3326650535776011, "grad_norm": 0.9041727997757555, "learning_rate": 9.520716508844324e-06, "loss": 0.3865, "step": 4812 }, { "epoch": 0.33273418596612514, "grad_norm": 0.787420420554255, "learning_rate": 9.5204773349651e-06, "loss": 0.3672, "step": 4813 }, { "epoch": 0.33280331835464916, "grad_norm": 0.7124113474343279, "learning_rate": 9.52023810442976e-06, "loss": 0.341, "step": 4814 }, { "epoch": 0.3328724507431732, "grad_norm": 0.7507638776343349, "learning_rate": 9.519998817241302e-06, "loss": 0.4659, "step": 4815 }, { "epoch": 0.3329415831316972, "grad_norm": 0.8333094789294057, "learning_rate": 9.519759473402726e-06, "loss": 0.3442, "step": 4816 }, { "epoch": 0.3330107155202212, "grad_norm": 0.6786471778007095, "learning_rate": 9.51952007291703e-06, "loss": 0.3132, "step": 4817 }, { "epoch": 0.33307984790874523, "grad_norm": 0.8398588276151625, "learning_rate": 9.519280615787216e-06, "loss": 0.3717, "step": 4818 }, { "epoch": 0.33314898029726925, "grad_norm": 0.9011009000347123, "learning_rate": 9.519041102016283e-06, "loss": 0.3951, "step": 4819 }, { "epoch": 0.3332181126857933, "grad_norm": 2.3584718417734996, "learning_rate": 9.518801531607236e-06, "loss": 0.3482, "step": 4820 }, { "epoch": 0.33328724507431734, "grad_norm": 0.9440338128011597, "learning_rate": 9.518561904563075e-06, "loss": 0.3358, "step": 4821 }, { "epoch": 0.33335637746284136, "grad_norm": 0.8071201853632931, "learning_rate": 9.518322220886806e-06, "loss": 0.4648, "step": 4822 }, { "epoch": 0.3334255098513654, "grad_norm": 0.7404846030231688, "learning_rate": 9.51808248058143e-06, "loss": 0.3886, "step": 4823 }, { "epoch": 0.3334946422398894, "grad_norm": 0.6839070820765404, "learning_rate": 9.517842683649952e-06, "loss": 0.2851, "step": 4824 }, { "epoch": 0.3335637746284134, "grad_norm": 0.7511371374718901, "learning_rate": 9.51760283009538e-06, "loss": 0.4034, "step": 4825 }, { "epoch": 0.33363290701693743, "grad_norm": 0.7598945214410295, "learning_rate": 9.517362919920718e-06, "loss": 0.4461, "step": 4826 }, { "epoch": 0.33370203940546145, "grad_norm": 0.9695080181493543, "learning_rate": 9.517122953128971e-06, "loss": 0.2998, "step": 4827 }, { "epoch": 0.33377117179398547, "grad_norm": 1.073543043696138, "learning_rate": 9.516882929723152e-06, "loss": 0.3346, "step": 4828 }, { "epoch": 0.3338403041825095, "grad_norm": 0.8084229881356452, "learning_rate": 9.516642849706267e-06, "loss": 0.3771, "step": 4829 }, { "epoch": 0.3339094365710335, "grad_norm": 0.8417794736214859, "learning_rate": 9.516402713081324e-06, "loss": 0.3796, "step": 4830 }, { "epoch": 0.3339785689595576, "grad_norm": 0.7640350249523576, "learning_rate": 9.516162519851331e-06, "loss": 0.2995, "step": 4831 }, { "epoch": 0.3340477013480816, "grad_norm": 0.8602627198265606, "learning_rate": 9.515922270019299e-06, "loss": 0.4052, "step": 4832 }, { "epoch": 0.3341168337366056, "grad_norm": 0.777392269816077, "learning_rate": 9.515681963588244e-06, "loss": 0.4914, "step": 4833 }, { "epoch": 0.33418596612512963, "grad_norm": 0.7159308621371361, "learning_rate": 9.515441600561172e-06, "loss": 0.2951, "step": 4834 }, { "epoch": 0.33425509851365365, "grad_norm": 0.7477787483983277, "learning_rate": 9.515201180941099e-06, "loss": 0.2775, "step": 4835 }, { "epoch": 0.33432423090217767, "grad_norm": 0.742873512442509, "learning_rate": 9.514960704731036e-06, "loss": 0.4113, "step": 4836 }, { "epoch": 0.3343933632907017, "grad_norm": 0.7022084119501258, "learning_rate": 9.514720171933997e-06, "loss": 0.3914, "step": 4837 }, { "epoch": 0.3344624956792257, "grad_norm": 0.736132383147262, "learning_rate": 9.514479582552997e-06, "loss": 0.3479, "step": 4838 }, { "epoch": 0.3345316280677497, "grad_norm": 0.7174505429582508, "learning_rate": 9.514238936591054e-06, "loss": 0.3173, "step": 4839 }, { "epoch": 0.33460076045627374, "grad_norm": 0.768651107441562, "learning_rate": 9.513998234051182e-06, "loss": 0.4266, "step": 4840 }, { "epoch": 0.3346698928447978, "grad_norm": 0.6780251377655585, "learning_rate": 9.513757474936397e-06, "loss": 0.2793, "step": 4841 }, { "epoch": 0.33473902523332183, "grad_norm": 0.7391074653524736, "learning_rate": 9.513516659249717e-06, "loss": 0.3365, "step": 4842 }, { "epoch": 0.33480815762184585, "grad_norm": 0.8456086880298178, "learning_rate": 9.51327578699416e-06, "loss": 0.3886, "step": 4843 }, { "epoch": 0.33487729001036987, "grad_norm": 0.7570561063164254, "learning_rate": 9.513034858172747e-06, "loss": 0.3976, "step": 4844 }, { "epoch": 0.3349464223988939, "grad_norm": 0.7860900976620088, "learning_rate": 9.512793872788496e-06, "loss": 0.3431, "step": 4845 }, { "epoch": 0.3350155547874179, "grad_norm": 0.7780504326384964, "learning_rate": 9.512552830844424e-06, "loss": 0.3159, "step": 4846 }, { "epoch": 0.3350846871759419, "grad_norm": 0.8513413067149738, "learning_rate": 9.512311732343558e-06, "loss": 0.3911, "step": 4847 }, { "epoch": 0.33515381956446594, "grad_norm": 0.7461740121776351, "learning_rate": 9.512070577288916e-06, "loss": 0.4351, "step": 4848 }, { "epoch": 0.33522295195298996, "grad_norm": 0.7605404294163567, "learning_rate": 9.511829365683522e-06, "loss": 0.3153, "step": 4849 }, { "epoch": 0.335292084341514, "grad_norm": 0.794255762751574, "learning_rate": 9.511588097530399e-06, "loss": 0.3709, "step": 4850 }, { "epoch": 0.33536121673003805, "grad_norm": 0.9515946863164259, "learning_rate": 9.51134677283257e-06, "loss": 0.4264, "step": 4851 }, { "epoch": 0.33543034911856207, "grad_norm": 0.6991938632378585, "learning_rate": 9.51110539159306e-06, "loss": 0.3219, "step": 4852 }, { "epoch": 0.3354994815070861, "grad_norm": 0.7668323661605618, "learning_rate": 9.510863953814896e-06, "loss": 0.2828, "step": 4853 }, { "epoch": 0.3355686138956101, "grad_norm": 0.7292929706814819, "learning_rate": 9.5106224595011e-06, "loss": 0.3702, "step": 4854 }, { "epoch": 0.3356377462841341, "grad_norm": 0.7885008598726366, "learning_rate": 9.510380908654702e-06, "loss": 0.4199, "step": 4855 }, { "epoch": 0.33570687867265814, "grad_norm": 0.7585185654319995, "learning_rate": 9.510139301278727e-06, "loss": 0.3554, "step": 4856 }, { "epoch": 0.33577601106118216, "grad_norm": 1.4559806901932415, "learning_rate": 9.509897637376206e-06, "loss": 0.2978, "step": 4857 }, { "epoch": 0.3358451434497062, "grad_norm": 0.7595049632815514, "learning_rate": 9.509655916950167e-06, "loss": 0.4516, "step": 4858 }, { "epoch": 0.3359142758382302, "grad_norm": 0.7411650345925321, "learning_rate": 9.509414140003637e-06, "loss": 0.3474, "step": 4859 }, { "epoch": 0.3359834082267542, "grad_norm": 0.8134340608159769, "learning_rate": 9.509172306539649e-06, "loss": 0.3019, "step": 4860 }, { "epoch": 0.3360525406152783, "grad_norm": 0.837695432354032, "learning_rate": 9.508930416561234e-06, "loss": 0.4462, "step": 4861 }, { "epoch": 0.3361216730038023, "grad_norm": 0.7226247412584144, "learning_rate": 9.508688470071422e-06, "loss": 0.4073, "step": 4862 }, { "epoch": 0.3361908053923263, "grad_norm": 0.733175919375155, "learning_rate": 9.508446467073247e-06, "loss": 0.3224, "step": 4863 }, { "epoch": 0.33625993778085034, "grad_norm": 0.7336928871579363, "learning_rate": 9.50820440756974e-06, "loss": 0.3359, "step": 4864 }, { "epoch": 0.33632907016937436, "grad_norm": 0.8384774460839831, "learning_rate": 9.507962291563937e-06, "loss": 0.5259, "step": 4865 }, { "epoch": 0.3363982025578984, "grad_norm": 0.7114394751106684, "learning_rate": 9.50772011905887e-06, "loss": 0.2906, "step": 4866 }, { "epoch": 0.3364673349464224, "grad_norm": 0.8759911646597094, "learning_rate": 9.507477890057576e-06, "loss": 0.287, "step": 4867 }, { "epoch": 0.3365364673349464, "grad_norm": 1.2569378513834182, "learning_rate": 9.507235604563092e-06, "loss": 0.3445, "step": 4868 }, { "epoch": 0.33660559972347043, "grad_norm": 0.8085680498514934, "learning_rate": 9.506993262578452e-06, "loss": 0.3913, "step": 4869 }, { "epoch": 0.33667473211199445, "grad_norm": 0.862391037005754, "learning_rate": 9.506750864106695e-06, "loss": 0.3282, "step": 4870 }, { "epoch": 0.3367438645005185, "grad_norm": 0.6686214958177623, "learning_rate": 9.506508409150859e-06, "loss": 0.282, "step": 4871 }, { "epoch": 0.33681299688904254, "grad_norm": 0.7394232520123302, "learning_rate": 9.506265897713983e-06, "loss": 0.473, "step": 4872 }, { "epoch": 0.33688212927756656, "grad_norm": 0.8283362756010114, "learning_rate": 9.506023329799105e-06, "loss": 0.4172, "step": 4873 }, { "epoch": 0.3369512616660906, "grad_norm": 0.7168502105904263, "learning_rate": 9.505780705409264e-06, "loss": 0.3942, "step": 4874 }, { "epoch": 0.3370203940546146, "grad_norm": 0.8686829985080102, "learning_rate": 9.505538024547504e-06, "loss": 0.3359, "step": 4875 }, { "epoch": 0.3370895264431386, "grad_norm": 0.7305838938438355, "learning_rate": 9.505295287216867e-06, "loss": 0.4333, "step": 4876 }, { "epoch": 0.3371586588316626, "grad_norm": 0.7593536159949217, "learning_rate": 9.505052493420392e-06, "loss": 0.348, "step": 4877 }, { "epoch": 0.33722779122018665, "grad_norm": 0.6904963653115167, "learning_rate": 9.504809643161126e-06, "loss": 0.298, "step": 4878 }, { "epoch": 0.33729692360871066, "grad_norm": 0.7713283978702812, "learning_rate": 9.504566736442107e-06, "loss": 0.3981, "step": 4879 }, { "epoch": 0.3373660559972347, "grad_norm": 0.7706232243733525, "learning_rate": 9.504323773266384e-06, "loss": 0.3673, "step": 4880 }, { "epoch": 0.33743518838575876, "grad_norm": 1.195247213795995, "learning_rate": 9.504080753637002e-06, "loss": 0.3934, "step": 4881 }, { "epoch": 0.3375043207742828, "grad_norm": 0.802071470291418, "learning_rate": 9.503837677557004e-06, "loss": 0.3156, "step": 4882 }, { "epoch": 0.3375734531628068, "grad_norm": 0.8017154073482136, "learning_rate": 9.503594545029439e-06, "loss": 0.4347, "step": 4883 }, { "epoch": 0.3376425855513308, "grad_norm": 1.5669116034936146, "learning_rate": 9.503351356057354e-06, "loss": 0.3355, "step": 4884 }, { "epoch": 0.3377117179398548, "grad_norm": 0.949960427106903, "learning_rate": 9.503108110643795e-06, "loss": 0.3299, "step": 4885 }, { "epoch": 0.33778085032837885, "grad_norm": 0.7666610255125189, "learning_rate": 9.502864808791812e-06, "loss": 0.3689, "step": 4886 }, { "epoch": 0.33784998271690286, "grad_norm": 0.9160921649964895, "learning_rate": 9.502621450504455e-06, "loss": 0.438, "step": 4887 }, { "epoch": 0.3379191151054269, "grad_norm": 0.9100344685138112, "learning_rate": 9.502378035784773e-06, "loss": 0.3239, "step": 4888 }, { "epoch": 0.3379882474939509, "grad_norm": 0.723420145220756, "learning_rate": 9.502134564635818e-06, "loss": 0.317, "step": 4889 }, { "epoch": 0.3380573798824749, "grad_norm": 0.7863901232080843, "learning_rate": 9.501891037060639e-06, "loss": 0.4545, "step": 4890 }, { "epoch": 0.33812651227099894, "grad_norm": 0.7915911079286099, "learning_rate": 9.50164745306229e-06, "loss": 0.3307, "step": 4891 }, { "epoch": 0.338195644659523, "grad_norm": 0.8500983799580487, "learning_rate": 9.501403812643826e-06, "loss": 0.3215, "step": 4892 }, { "epoch": 0.338264777048047, "grad_norm": 0.7595188766730591, "learning_rate": 9.501160115808296e-06, "loss": 0.3285, "step": 4893 }, { "epoch": 0.33833390943657105, "grad_norm": 0.7941059943451323, "learning_rate": 9.500916362558758e-06, "loss": 0.4394, "step": 4894 }, { "epoch": 0.33840304182509506, "grad_norm": 0.7697954438987236, "learning_rate": 9.500672552898266e-06, "loss": 0.3455, "step": 4895 }, { "epoch": 0.3384721742136191, "grad_norm": 0.9532992388679296, "learning_rate": 9.500428686829875e-06, "loss": 0.3678, "step": 4896 }, { "epoch": 0.3385413066021431, "grad_norm": 0.7193534043373548, "learning_rate": 9.500184764356642e-06, "loss": 0.3811, "step": 4897 }, { "epoch": 0.3386104389906671, "grad_norm": 0.7702355904237511, "learning_rate": 9.499940785481622e-06, "loss": 0.3259, "step": 4898 }, { "epoch": 0.33867957137919114, "grad_norm": 0.7873347260275064, "learning_rate": 9.499696750207876e-06, "loss": 0.3286, "step": 4899 }, { "epoch": 0.33874870376771515, "grad_norm": 0.7635433994991807, "learning_rate": 9.499452658538462e-06, "loss": 0.3491, "step": 4900 }, { "epoch": 0.33881783615623917, "grad_norm": 0.6852095450740675, "learning_rate": 9.499208510476436e-06, "loss": 0.3719, "step": 4901 }, { "epoch": 0.33888696854476325, "grad_norm": 0.9770848502485157, "learning_rate": 9.498964306024864e-06, "loss": 0.3324, "step": 4902 }, { "epoch": 0.33895610093328726, "grad_norm": 0.8192811099924905, "learning_rate": 9.498720045186801e-06, "loss": 0.3401, "step": 4903 }, { "epoch": 0.3390252333218113, "grad_norm": 0.8146414747615318, "learning_rate": 9.498475727965311e-06, "loss": 0.3791, "step": 4904 }, { "epoch": 0.3390943657103353, "grad_norm": 0.8244012686423259, "learning_rate": 9.498231354363457e-06, "loss": 0.3796, "step": 4905 }, { "epoch": 0.3391634980988593, "grad_norm": 0.8246174393413247, "learning_rate": 9.497986924384299e-06, "loss": 0.2895, "step": 4906 }, { "epoch": 0.33923263048738334, "grad_norm": 0.7696866156457451, "learning_rate": 9.497742438030902e-06, "loss": 0.3418, "step": 4907 }, { "epoch": 0.33930176287590735, "grad_norm": 0.922177645835024, "learning_rate": 9.49749789530633e-06, "loss": 0.3964, "step": 4908 }, { "epoch": 0.33937089526443137, "grad_norm": 0.7650886343039996, "learning_rate": 9.497253296213648e-06, "loss": 0.3253, "step": 4909 }, { "epoch": 0.3394400276529554, "grad_norm": 0.9033166426225351, "learning_rate": 9.497008640755923e-06, "loss": 0.3346, "step": 4910 }, { "epoch": 0.3395091600414794, "grad_norm": 0.7820068000957737, "learning_rate": 9.49676392893622e-06, "loss": 0.4209, "step": 4911 }, { "epoch": 0.3395782924300035, "grad_norm": 0.8626949674341081, "learning_rate": 9.496519160757602e-06, "loss": 0.4014, "step": 4912 }, { "epoch": 0.3396474248185275, "grad_norm": 0.8153110507395507, "learning_rate": 9.496274336223145e-06, "loss": 0.3271, "step": 4913 }, { "epoch": 0.3397165572070515, "grad_norm": 0.9276151173948365, "learning_rate": 9.496029455335911e-06, "loss": 0.3579, "step": 4914 }, { "epoch": 0.33978568959557554, "grad_norm": 0.8059649391788191, "learning_rate": 9.495784518098972e-06, "loss": 0.4431, "step": 4915 }, { "epoch": 0.33985482198409955, "grad_norm": 0.9209019706760684, "learning_rate": 9.495539524515397e-06, "loss": 0.3445, "step": 4916 }, { "epoch": 0.33992395437262357, "grad_norm": 0.8647385256071208, "learning_rate": 9.495294474588256e-06, "loss": 0.2973, "step": 4917 }, { "epoch": 0.3399930867611476, "grad_norm": 0.780077338307759, "learning_rate": 9.495049368320624e-06, "loss": 0.4214, "step": 4918 }, { "epoch": 0.3400622191496716, "grad_norm": 0.832655434227134, "learning_rate": 9.494804205715568e-06, "loss": 0.416, "step": 4919 }, { "epoch": 0.3401313515381956, "grad_norm": 0.8088574401598358, "learning_rate": 9.49455898677616e-06, "loss": 0.3298, "step": 4920 }, { "epoch": 0.34020048392671964, "grad_norm": 0.7756120896373864, "learning_rate": 9.49431371150548e-06, "loss": 0.2695, "step": 4921 }, { "epoch": 0.3402696163152437, "grad_norm": 0.8967626530211804, "learning_rate": 9.494068379906596e-06, "loss": 0.4498, "step": 4922 }, { "epoch": 0.34033874870376774, "grad_norm": 0.846366591798463, "learning_rate": 9.493822991982586e-06, "loss": 0.3611, "step": 4923 }, { "epoch": 0.34040788109229175, "grad_norm": 0.75515552375691, "learning_rate": 9.493577547736524e-06, "loss": 0.2761, "step": 4924 }, { "epoch": 0.34047701348081577, "grad_norm": 0.6919032623399188, "learning_rate": 9.493332047171487e-06, "loss": 0.3419, "step": 4925 }, { "epoch": 0.3405461458693398, "grad_norm": 0.7706811160835589, "learning_rate": 9.49308649029055e-06, "loss": 0.4087, "step": 4926 }, { "epoch": 0.3406152782578638, "grad_norm": 0.8074927771451604, "learning_rate": 9.492840877096794e-06, "loss": 0.3066, "step": 4927 }, { "epoch": 0.3406844106463878, "grad_norm": 0.8031712306474355, "learning_rate": 9.492595207593294e-06, "loss": 0.3634, "step": 4928 }, { "epoch": 0.34075354303491184, "grad_norm": 0.7527464710041253, "learning_rate": 9.492349481783131e-06, "loss": 0.4356, "step": 4929 }, { "epoch": 0.34082267542343586, "grad_norm": 0.7562613229447496, "learning_rate": 9.492103699669384e-06, "loss": 0.3759, "step": 4930 }, { "epoch": 0.3408918078119599, "grad_norm": 0.7352043993938238, "learning_rate": 9.491857861255134e-06, "loss": 0.2864, "step": 4931 }, { "epoch": 0.34096094020048395, "grad_norm": 0.7738815475338982, "learning_rate": 9.491611966543461e-06, "loss": 0.2856, "step": 4932 }, { "epoch": 0.34103007258900797, "grad_norm": 0.8283550819880263, "learning_rate": 9.49136601553745e-06, "loss": 0.3926, "step": 4933 }, { "epoch": 0.341099204977532, "grad_norm": 0.8829734688168391, "learning_rate": 9.491120008240178e-06, "loss": 0.3495, "step": 4934 }, { "epoch": 0.341168337366056, "grad_norm": 1.0032263397521923, "learning_rate": 9.490873944654733e-06, "loss": 0.387, "step": 4935 }, { "epoch": 0.34123746975458, "grad_norm": 0.728144631754632, "learning_rate": 9.490627824784198e-06, "loss": 0.3691, "step": 4936 }, { "epoch": 0.34130660214310404, "grad_norm": 0.7126130085992286, "learning_rate": 9.490381648631655e-06, "loss": 0.3897, "step": 4937 }, { "epoch": 0.34137573453162806, "grad_norm": 0.6715177751941908, "learning_rate": 9.490135416200194e-06, "loss": 0.3291, "step": 4938 }, { "epoch": 0.3414448669201521, "grad_norm": 0.8454463442897817, "learning_rate": 9.489889127492897e-06, "loss": 0.2886, "step": 4939 }, { "epoch": 0.3415139993086761, "grad_norm": 0.8236112149772334, "learning_rate": 9.489642782512853e-06, "loss": 0.4711, "step": 4940 }, { "epoch": 0.3415831316972001, "grad_norm": 0.7480410855506207, "learning_rate": 9.489396381263148e-06, "loss": 0.3784, "step": 4941 }, { "epoch": 0.3416522640857242, "grad_norm": 0.7160223473279459, "learning_rate": 9.489149923746872e-06, "loss": 0.2998, "step": 4942 }, { "epoch": 0.3417213964742482, "grad_norm": 0.8171376636032092, "learning_rate": 9.488903409967114e-06, "loss": 0.3464, "step": 4943 }, { "epoch": 0.3417905288627722, "grad_norm": 0.7625521032400352, "learning_rate": 9.48865683992696e-06, "loss": 0.5108, "step": 4944 }, { "epoch": 0.34185966125129624, "grad_norm": 0.8199744103398838, "learning_rate": 9.488410213629505e-06, "loss": 0.3652, "step": 4945 }, { "epoch": 0.34192879363982026, "grad_norm": 0.7897126741144831, "learning_rate": 9.488163531077836e-06, "loss": 0.3555, "step": 4946 }, { "epoch": 0.3419979260283443, "grad_norm": 0.9332530604800965, "learning_rate": 9.487916792275049e-06, "loss": 0.424, "step": 4947 }, { "epoch": 0.3420670584168683, "grad_norm": 0.7279442496632464, "learning_rate": 9.487669997224231e-06, "loss": 0.3685, "step": 4948 }, { "epoch": 0.3421361908053923, "grad_norm": 0.9145815002741384, "learning_rate": 9.48742314592848e-06, "loss": 0.3129, "step": 4949 }, { "epoch": 0.34220532319391633, "grad_norm": 0.8507251419776939, "learning_rate": 9.48717623839089e-06, "loss": 0.3252, "step": 4950 }, { "epoch": 0.34227445558244035, "grad_norm": 0.905538823102909, "learning_rate": 9.486929274614551e-06, "loss": 0.4702, "step": 4951 }, { "epoch": 0.34234358797096437, "grad_norm": 0.8248828553503488, "learning_rate": 9.48668225460256e-06, "loss": 0.2839, "step": 4952 }, { "epoch": 0.34241272035948844, "grad_norm": 0.8056746490348138, "learning_rate": 9.486435178358015e-06, "loss": 0.3291, "step": 4953 }, { "epoch": 0.34248185274801246, "grad_norm": 0.7017505869569208, "learning_rate": 9.486188045884013e-06, "loss": 0.3763, "step": 4954 }, { "epoch": 0.3425509851365365, "grad_norm": 0.7696569012199598, "learning_rate": 9.485940857183647e-06, "loss": 0.4087, "step": 4955 }, { "epoch": 0.3426201175250605, "grad_norm": 0.7466290879317962, "learning_rate": 9.48569361226002e-06, "loss": 0.296, "step": 4956 }, { "epoch": 0.3426892499135845, "grad_norm": 0.7991420704375739, "learning_rate": 9.485446311116228e-06, "loss": 0.3638, "step": 4957 }, { "epoch": 0.34275838230210853, "grad_norm": 0.6820279788900308, "learning_rate": 9.48519895375537e-06, "loss": 0.3562, "step": 4958 }, { "epoch": 0.34282751469063255, "grad_norm": 0.7312228472525322, "learning_rate": 9.48495154018055e-06, "loss": 0.3099, "step": 4959 }, { "epoch": 0.34289664707915657, "grad_norm": 0.6840645451445325, "learning_rate": 9.484704070394864e-06, "loss": 0.2638, "step": 4960 }, { "epoch": 0.3429657794676806, "grad_norm": 0.7898780526204756, "learning_rate": 9.484456544401416e-06, "loss": 0.3547, "step": 4961 }, { "epoch": 0.3430349118562046, "grad_norm": 0.7306351079751918, "learning_rate": 9.484208962203307e-06, "loss": 0.4041, "step": 4962 }, { "epoch": 0.3431040442447287, "grad_norm": 0.7373320438277599, "learning_rate": 9.483961323803643e-06, "loss": 0.3241, "step": 4963 }, { "epoch": 0.3431731766332527, "grad_norm": 0.7800585987712915, "learning_rate": 9.483713629205527e-06, "loss": 0.2799, "step": 4964 }, { "epoch": 0.3432423090217767, "grad_norm": 0.7650739033194338, "learning_rate": 9.48346587841206e-06, "loss": 0.486, "step": 4965 }, { "epoch": 0.34331144141030073, "grad_norm": 0.748192152882137, "learning_rate": 9.48321807142635e-06, "loss": 0.312, "step": 4966 }, { "epoch": 0.34338057379882475, "grad_norm": 0.7287601680114411, "learning_rate": 9.4829702082515e-06, "loss": 0.3096, "step": 4967 }, { "epoch": 0.34344970618734877, "grad_norm": 0.6801445657683207, "learning_rate": 9.482722288890622e-06, "loss": 0.3274, "step": 4968 }, { "epoch": 0.3435188385758728, "grad_norm": 0.805706436936006, "learning_rate": 9.482474313346818e-06, "loss": 0.5429, "step": 4969 }, { "epoch": 0.3435879709643968, "grad_norm": 1.0454051406512157, "learning_rate": 9.4822262816232e-06, "loss": 0.3837, "step": 4970 }, { "epoch": 0.3436571033529208, "grad_norm": 0.7315879226684502, "learning_rate": 9.481978193722872e-06, "loss": 0.3703, "step": 4971 }, { "epoch": 0.34372623574144484, "grad_norm": 0.7726392774790086, "learning_rate": 9.481730049648944e-06, "loss": 0.4274, "step": 4972 }, { "epoch": 0.3437953681299689, "grad_norm": 0.7370522185657591, "learning_rate": 9.481481849404533e-06, "loss": 0.3823, "step": 4973 }, { "epoch": 0.34386450051849293, "grad_norm": 0.8136803510105517, "learning_rate": 9.481233592992742e-06, "loss": 0.3189, "step": 4974 }, { "epoch": 0.34393363290701695, "grad_norm": 0.7138556397817757, "learning_rate": 9.480985280416683e-06, "loss": 0.3284, "step": 4975 }, { "epoch": 0.34400276529554097, "grad_norm": 0.8007489127242026, "learning_rate": 9.480736911679473e-06, "loss": 0.491, "step": 4976 }, { "epoch": 0.344071897684065, "grad_norm": 0.8640710790703088, "learning_rate": 9.480488486784221e-06, "loss": 0.3018, "step": 4977 }, { "epoch": 0.344141030072589, "grad_norm": 0.7374700208310273, "learning_rate": 9.480240005734043e-06, "loss": 0.3268, "step": 4978 }, { "epoch": 0.344210162461113, "grad_norm": 0.836912584324832, "learning_rate": 9.479991468532051e-06, "loss": 0.436, "step": 4979 }, { "epoch": 0.34427929484963704, "grad_norm": 1.0251599983957902, "learning_rate": 9.47974287518136e-06, "loss": 0.3827, "step": 4980 }, { "epoch": 0.34434842723816106, "grad_norm": 0.7750992665840665, "learning_rate": 9.479494225685086e-06, "loss": 0.3297, "step": 4981 }, { "epoch": 0.3444175596266851, "grad_norm": 0.7841608842831, "learning_rate": 9.479245520046348e-06, "loss": 0.3858, "step": 4982 }, { "epoch": 0.34448669201520915, "grad_norm": 0.6993874725723035, "learning_rate": 9.47899675826826e-06, "loss": 0.4412, "step": 4983 }, { "epoch": 0.34455582440373317, "grad_norm": 0.7779305655478487, "learning_rate": 9.478747940353941e-06, "loss": 0.3339, "step": 4984 }, { "epoch": 0.3446249567922572, "grad_norm": 0.8666641281537255, "learning_rate": 9.478499066306508e-06, "loss": 0.3165, "step": 4985 }, { "epoch": 0.3446940891807812, "grad_norm": 0.7518407982507622, "learning_rate": 9.478250136129084e-06, "loss": 0.4105, "step": 4986 }, { "epoch": 0.3447632215693052, "grad_norm": 0.9070896823723307, "learning_rate": 9.478001149824784e-06, "loss": 0.4196, "step": 4987 }, { "epoch": 0.34483235395782924, "grad_norm": 1.0846220825553081, "learning_rate": 9.477752107396732e-06, "loss": 0.3171, "step": 4988 }, { "epoch": 0.34490148634635326, "grad_norm": 0.7495504992418396, "learning_rate": 9.477503008848049e-06, "loss": 0.3026, "step": 4989 }, { "epoch": 0.3449706187348773, "grad_norm": 1.1923119038561907, "learning_rate": 9.477253854181854e-06, "loss": 0.4254, "step": 4990 }, { "epoch": 0.3450397511234013, "grad_norm": 0.7525739796052636, "learning_rate": 9.477004643401273e-06, "loss": 0.3528, "step": 4991 }, { "epoch": 0.3451088835119253, "grad_norm": 1.4677746781636032, "learning_rate": 9.476755376509427e-06, "loss": 0.3166, "step": 4992 }, { "epoch": 0.3451780159004494, "grad_norm": 0.8041995443507789, "learning_rate": 9.476506053509444e-06, "loss": 0.386, "step": 4993 }, { "epoch": 0.3452471482889734, "grad_norm": 0.7966025001297682, "learning_rate": 9.476256674404445e-06, "loss": 0.4265, "step": 4994 }, { "epoch": 0.3453162806774974, "grad_norm": 0.7743838245297573, "learning_rate": 9.476007239197557e-06, "loss": 0.3004, "step": 4995 }, { "epoch": 0.34538541306602144, "grad_norm": 0.7924039081044058, "learning_rate": 9.475757747891907e-06, "loss": 0.3537, "step": 4996 }, { "epoch": 0.34545454545454546, "grad_norm": 0.6933480334407025, "learning_rate": 9.47550820049062e-06, "loss": 0.4217, "step": 4997 }, { "epoch": 0.3455236778430695, "grad_norm": 0.792371269030063, "learning_rate": 9.475258596996825e-06, "loss": 0.3574, "step": 4998 }, { "epoch": 0.3455928102315935, "grad_norm": 0.7038264187047527, "learning_rate": 9.47500893741365e-06, "loss": 0.3174, "step": 4999 }, { "epoch": 0.3456619426201175, "grad_norm": 1.1250000593628153, "learning_rate": 9.474759221744222e-06, "loss": 0.3601, "step": 5000 }, { "epoch": 0.34573107500864153, "grad_norm": 0.8856746480949073, "learning_rate": 9.474509449991676e-06, "loss": 0.4635, "step": 5001 }, { "epoch": 0.34580020739716555, "grad_norm": 0.9979918156154136, "learning_rate": 9.474259622159137e-06, "loss": 0.323, "step": 5002 }, { "epoch": 0.3458693397856896, "grad_norm": 0.8082805267699309, "learning_rate": 9.474009738249738e-06, "loss": 0.3238, "step": 5003 }, { "epoch": 0.34593847217421364, "grad_norm": 0.7752640648780007, "learning_rate": 9.473759798266613e-06, "loss": 0.4406, "step": 5004 }, { "epoch": 0.34600760456273766, "grad_norm": 0.7677320929828367, "learning_rate": 9.473509802212891e-06, "loss": 0.3891, "step": 5005 }, { "epoch": 0.3460767369512617, "grad_norm": 0.9034460137761219, "learning_rate": 9.473259750091708e-06, "loss": 0.2979, "step": 5006 }, { "epoch": 0.3461458693397857, "grad_norm": 0.77715884591707, "learning_rate": 9.473009641906196e-06, "loss": 0.3823, "step": 5007 }, { "epoch": 0.3462150017283097, "grad_norm": 0.7789088206316562, "learning_rate": 9.472759477659492e-06, "loss": 0.4086, "step": 5008 }, { "epoch": 0.34628413411683373, "grad_norm": 0.7290231240250932, "learning_rate": 9.47250925735473e-06, "loss": 0.2793, "step": 5009 }, { "epoch": 0.34635326650535775, "grad_norm": 0.733276492336311, "learning_rate": 9.472258980995044e-06, "loss": 0.2825, "step": 5010 }, { "epoch": 0.34642239889388177, "grad_norm": 0.8091407703448943, "learning_rate": 9.472008648583575e-06, "loss": 0.4434, "step": 5011 }, { "epoch": 0.3464915312824058, "grad_norm": 0.7565304845689506, "learning_rate": 9.471758260123457e-06, "loss": 0.4093, "step": 5012 }, { "epoch": 0.34656066367092986, "grad_norm": 1.0555008137284019, "learning_rate": 9.47150781561783e-06, "loss": 0.334, "step": 5013 }, { "epoch": 0.3466297960594539, "grad_norm": 1.1048145865032273, "learning_rate": 9.47125731506983e-06, "loss": 0.3331, "step": 5014 }, { "epoch": 0.3466989284479779, "grad_norm": 0.7508998729479911, "learning_rate": 9.471006758482601e-06, "loss": 0.4327, "step": 5015 }, { "epoch": 0.3467680608365019, "grad_norm": 0.8972210594615778, "learning_rate": 9.470756145859282e-06, "loss": 0.4003, "step": 5016 }, { "epoch": 0.34683719322502593, "grad_norm": 0.8280334200752104, "learning_rate": 9.470505477203011e-06, "loss": 0.3038, "step": 5017 }, { "epoch": 0.34690632561354995, "grad_norm": 0.7894309998779773, "learning_rate": 9.470254752516934e-06, "loss": 0.3608, "step": 5018 }, { "epoch": 0.34697545800207397, "grad_norm": 0.8972013005014741, "learning_rate": 9.470003971804191e-06, "loss": 0.381, "step": 5019 }, { "epoch": 0.347044590390598, "grad_norm": 0.7162645439386968, "learning_rate": 9.469753135067923e-06, "loss": 0.3082, "step": 5020 }, { "epoch": 0.347113722779122, "grad_norm": 0.7125258467918697, "learning_rate": 9.469502242311278e-06, "loss": 0.3116, "step": 5021 }, { "epoch": 0.347182855167646, "grad_norm": 0.8095015013620253, "learning_rate": 9.469251293537398e-06, "loss": 0.4286, "step": 5022 }, { "epoch": 0.34725198755617004, "grad_norm": 0.8506331383929067, "learning_rate": 9.469000288749429e-06, "loss": 0.3227, "step": 5023 }, { "epoch": 0.3473211199446941, "grad_norm": 0.6763347929151867, "learning_rate": 9.468749227950515e-06, "loss": 0.2649, "step": 5024 }, { "epoch": 0.34739025233321813, "grad_norm": 0.8692463606191372, "learning_rate": 9.468498111143806e-06, "loss": 0.3602, "step": 5025 }, { "epoch": 0.34745938472174215, "grad_norm": 0.7084560462454406, "learning_rate": 9.468246938332448e-06, "loss": 0.4157, "step": 5026 }, { "epoch": 0.34752851711026617, "grad_norm": 0.6920276855183534, "learning_rate": 9.467995709519589e-06, "loss": 0.2765, "step": 5027 }, { "epoch": 0.3475976494987902, "grad_norm": 0.825403188957246, "learning_rate": 9.467744424708376e-06, "loss": 0.2706, "step": 5028 }, { "epoch": 0.3476667818873142, "grad_norm": 0.9054132381910379, "learning_rate": 9.46749308390196e-06, "loss": 0.4234, "step": 5029 }, { "epoch": 0.3477359142758382, "grad_norm": 1.2969268826331746, "learning_rate": 9.46724168710349e-06, "loss": 0.3815, "step": 5030 }, { "epoch": 0.34780504666436224, "grad_norm": 0.7095083985653311, "learning_rate": 9.466990234316119e-06, "loss": 0.2933, "step": 5031 }, { "epoch": 0.34787417905288626, "grad_norm": 0.7862675921601826, "learning_rate": 9.466738725542997e-06, "loss": 0.3417, "step": 5032 }, { "epoch": 0.3479433114414103, "grad_norm": 0.8031993769726203, "learning_rate": 9.466487160787275e-06, "loss": 0.4089, "step": 5033 }, { "epoch": 0.34801244382993435, "grad_norm": 0.774336766773565, "learning_rate": 9.466235540052107e-06, "loss": 0.317, "step": 5034 }, { "epoch": 0.34808157621845837, "grad_norm": 0.7178940709697372, "learning_rate": 9.465983863340648e-06, "loss": 0.29, "step": 5035 }, { "epoch": 0.3481507086069824, "grad_norm": 0.7386942652141321, "learning_rate": 9.465732130656051e-06, "loss": 0.3526, "step": 5036 }, { "epoch": 0.3482198409955064, "grad_norm": 0.7634452454975742, "learning_rate": 9.465480342001472e-06, "loss": 0.4666, "step": 5037 }, { "epoch": 0.3482889733840304, "grad_norm": 0.7214783448274825, "learning_rate": 9.465228497380064e-06, "loss": 0.3565, "step": 5038 }, { "epoch": 0.34835810577255444, "grad_norm": 0.7663142113934786, "learning_rate": 9.464976596794987e-06, "loss": 0.317, "step": 5039 }, { "epoch": 0.34842723816107846, "grad_norm": 0.8915498978810541, "learning_rate": 9.464724640249396e-06, "loss": 0.3847, "step": 5040 }, { "epoch": 0.3484963705496025, "grad_norm": 0.8838060149900363, "learning_rate": 9.46447262774645e-06, "loss": 0.387, "step": 5041 }, { "epoch": 0.3485655029381265, "grad_norm": 0.6326959745769363, "learning_rate": 9.464220559289305e-06, "loss": 0.3162, "step": 5042 }, { "epoch": 0.3486346353266505, "grad_norm": 0.7971693243083654, "learning_rate": 9.463968434881123e-06, "loss": 0.4001, "step": 5043 }, { "epoch": 0.3487037677151746, "grad_norm": 0.7750331405462179, "learning_rate": 9.463716254525063e-06, "loss": 0.4354, "step": 5044 }, { "epoch": 0.3487729001036986, "grad_norm": 0.7098162654624376, "learning_rate": 9.463464018224287e-06, "loss": 0.3459, "step": 5045 }, { "epoch": 0.3488420324922226, "grad_norm": 0.7401711149558393, "learning_rate": 9.463211725981954e-06, "loss": 0.3231, "step": 5046 }, { "epoch": 0.34891116488074664, "grad_norm": 0.8264122145801335, "learning_rate": 9.462959377801225e-06, "loss": 0.4343, "step": 5047 }, { "epoch": 0.34898029726927066, "grad_norm": 0.8060327433145362, "learning_rate": 9.462706973685267e-06, "loss": 0.3367, "step": 5048 }, { "epoch": 0.3490494296577947, "grad_norm": 1.1320830980541245, "learning_rate": 9.46245451363724e-06, "loss": 0.2857, "step": 5049 }, { "epoch": 0.3491185620463187, "grad_norm": 1.3923114201435551, "learning_rate": 9.46220199766031e-06, "loss": 0.3619, "step": 5050 }, { "epoch": 0.3491876944348427, "grad_norm": 0.8303819531870851, "learning_rate": 9.46194942575764e-06, "loss": 0.4672, "step": 5051 }, { "epoch": 0.34925682682336673, "grad_norm": 0.7199666988039974, "learning_rate": 9.461696797932397e-06, "loss": 0.3188, "step": 5052 }, { "epoch": 0.34932595921189075, "grad_norm": 0.7455093978355668, "learning_rate": 9.461444114187749e-06, "loss": 0.3008, "step": 5053 }, { "epoch": 0.3493950916004148, "grad_norm": 0.8671903347625958, "learning_rate": 9.461191374526858e-06, "loss": 0.4718, "step": 5054 }, { "epoch": 0.34946422398893884, "grad_norm": 0.7619438911706763, "learning_rate": 9.460938578952896e-06, "loss": 0.4016, "step": 5055 }, { "epoch": 0.34953335637746286, "grad_norm": 0.8028833103313148, "learning_rate": 9.46068572746903e-06, "loss": 0.3136, "step": 5056 }, { "epoch": 0.3496024887659869, "grad_norm": 0.8120900181514048, "learning_rate": 9.460432820078429e-06, "loss": 0.4172, "step": 5057 }, { "epoch": 0.3496716211545109, "grad_norm": 1.4061151335226445, "learning_rate": 9.460179856784262e-06, "loss": 0.429, "step": 5058 }, { "epoch": 0.3497407535430349, "grad_norm": 0.7685616158359263, "learning_rate": 9.459926837589699e-06, "loss": 0.3275, "step": 5059 }, { "epoch": 0.34980988593155893, "grad_norm": 0.8609540420214331, "learning_rate": 9.459673762497914e-06, "loss": 0.3121, "step": 5060 }, { "epoch": 0.34987901832008295, "grad_norm": 0.8102847340226237, "learning_rate": 9.459420631512076e-06, "loss": 0.4086, "step": 5061 }, { "epoch": 0.34994815070860696, "grad_norm": 0.8277815527617538, "learning_rate": 9.459167444635359e-06, "loss": 0.4005, "step": 5062 }, { "epoch": 0.350017283097131, "grad_norm": 0.6984456128326076, "learning_rate": 9.458914201870935e-06, "loss": 0.344, "step": 5063 }, { "epoch": 0.35008641548565506, "grad_norm": 0.6910312055248874, "learning_rate": 9.45866090322198e-06, "loss": 0.3088, "step": 5064 }, { "epoch": 0.3501555478741791, "grad_norm": 0.8446504086149553, "learning_rate": 9.458407548691667e-06, "loss": 0.3785, "step": 5065 }, { "epoch": 0.3502246802627031, "grad_norm": 0.799519302267221, "learning_rate": 9.458154138283171e-06, "loss": 0.4326, "step": 5066 }, { "epoch": 0.3502938126512271, "grad_norm": 0.7300104133727124, "learning_rate": 9.457900671999668e-06, "loss": 0.2787, "step": 5067 }, { "epoch": 0.35036294503975113, "grad_norm": 0.8906457617830665, "learning_rate": 9.457647149844337e-06, "loss": 0.4426, "step": 5068 }, { "epoch": 0.35043207742827515, "grad_norm": 1.6567706937307725, "learning_rate": 9.457393571820355e-06, "loss": 0.424, "step": 5069 }, { "epoch": 0.35050120981679916, "grad_norm": 0.762421115503675, "learning_rate": 9.457139937930899e-06, "loss": 0.3515, "step": 5070 }, { "epoch": 0.3505703422053232, "grad_norm": 0.995396337650368, "learning_rate": 9.456886248179147e-06, "loss": 0.3277, "step": 5071 }, { "epoch": 0.3506394745938472, "grad_norm": 0.7679387224331456, "learning_rate": 9.45663250256828e-06, "loss": 0.4228, "step": 5072 }, { "epoch": 0.3507086069823712, "grad_norm": 0.726442015883792, "learning_rate": 9.456378701101476e-06, "loss": 0.3501, "step": 5073 }, { "epoch": 0.3507777393708953, "grad_norm": 0.7156294017816137, "learning_rate": 9.45612484378192e-06, "loss": 0.3162, "step": 5074 }, { "epoch": 0.3508468717594193, "grad_norm": 0.7640077562113434, "learning_rate": 9.45587093061279e-06, "loss": 0.3454, "step": 5075 }, { "epoch": 0.35091600414794333, "grad_norm": 0.7667952119847784, "learning_rate": 9.455616961597272e-06, "loss": 0.4645, "step": 5076 }, { "epoch": 0.35098513653646735, "grad_norm": 0.79457687734073, "learning_rate": 9.455362936738543e-06, "loss": 0.3802, "step": 5077 }, { "epoch": 0.35105426892499136, "grad_norm": 0.8159399389467352, "learning_rate": 9.455108856039795e-06, "loss": 0.2773, "step": 5078 }, { "epoch": 0.3511234013135154, "grad_norm": 3.7822461586508345, "learning_rate": 9.454854719504204e-06, "loss": 0.4104, "step": 5079 }, { "epoch": 0.3511925337020394, "grad_norm": 0.8370903347678351, "learning_rate": 9.45460052713496e-06, "loss": 0.374, "step": 5080 }, { "epoch": 0.3512616660905634, "grad_norm": 0.7285686613706867, "learning_rate": 9.45434627893525e-06, "loss": 0.3056, "step": 5081 }, { "epoch": 0.35133079847908744, "grad_norm": 0.7540023877844891, "learning_rate": 9.454091974908258e-06, "loss": 0.3626, "step": 5082 }, { "epoch": 0.35139993086761145, "grad_norm": 0.7354565186367578, "learning_rate": 9.453837615057171e-06, "loss": 0.4114, "step": 5083 }, { "epoch": 0.35146906325613547, "grad_norm": 0.7261136505160113, "learning_rate": 9.453583199385177e-06, "loss": 0.3279, "step": 5084 }, { "epoch": 0.35153819564465955, "grad_norm": 0.7622034317550341, "learning_rate": 9.453328727895467e-06, "loss": 0.3375, "step": 5085 }, { "epoch": 0.35160732803318356, "grad_norm": 1.9317302565152195, "learning_rate": 9.453074200591229e-06, "loss": 0.3886, "step": 5086 }, { "epoch": 0.3516764604217076, "grad_norm": 0.7236069050564694, "learning_rate": 9.45281961747565e-06, "loss": 0.3827, "step": 5087 }, { "epoch": 0.3517455928102316, "grad_norm": 1.0552878804126522, "learning_rate": 9.452564978551926e-06, "loss": 0.3652, "step": 5088 }, { "epoch": 0.3518147251987556, "grad_norm": 1.2989957867524058, "learning_rate": 9.452310283823244e-06, "loss": 0.3245, "step": 5089 }, { "epoch": 0.35188385758727964, "grad_norm": 0.8466447555699156, "learning_rate": 9.452055533292798e-06, "loss": 0.5055, "step": 5090 }, { "epoch": 0.35195298997580365, "grad_norm": 0.7303634865632366, "learning_rate": 9.451800726963784e-06, "loss": 0.3739, "step": 5091 }, { "epoch": 0.35202212236432767, "grad_norm": 0.7228185461131941, "learning_rate": 9.451545864839391e-06, "loss": 0.2963, "step": 5092 }, { "epoch": 0.3520912547528517, "grad_norm": 0.7345692788548759, "learning_rate": 9.451290946922812e-06, "loss": 0.3731, "step": 5093 }, { "epoch": 0.3521603871413757, "grad_norm": 0.7849891509194573, "learning_rate": 9.451035973217248e-06, "loss": 0.4613, "step": 5094 }, { "epoch": 0.3522295195298998, "grad_norm": 0.7557085584698936, "learning_rate": 9.450780943725891e-06, "loss": 0.3283, "step": 5095 }, { "epoch": 0.3522986519184238, "grad_norm": 0.7461324256858624, "learning_rate": 9.450525858451938e-06, "loss": 0.2781, "step": 5096 }, { "epoch": 0.3523677843069478, "grad_norm": 1.0554983385157732, "learning_rate": 9.450270717398584e-06, "loss": 0.4166, "step": 5097 }, { "epoch": 0.35243691669547184, "grad_norm": 0.8005074142937281, "learning_rate": 9.45001552056903e-06, "loss": 0.3874, "step": 5098 }, { "epoch": 0.35250604908399585, "grad_norm": 0.9028999763536619, "learning_rate": 9.44976026796647e-06, "loss": 0.2628, "step": 5099 }, { "epoch": 0.35257518147251987, "grad_norm": 0.9919180186353348, "learning_rate": 9.44950495959411e-06, "loss": 0.3531, "step": 5100 }, { "epoch": 0.3526443138610439, "grad_norm": 0.7709789314958866, "learning_rate": 9.449249595455143e-06, "loss": 0.4471, "step": 5101 }, { "epoch": 0.3527134462495679, "grad_norm": 1.1673868614226253, "learning_rate": 9.448994175552773e-06, "loss": 0.3188, "step": 5102 }, { "epoch": 0.3527825786380919, "grad_norm": 0.7551099504528598, "learning_rate": 9.448738699890201e-06, "loss": 0.3082, "step": 5103 }, { "epoch": 0.35285171102661594, "grad_norm": 0.8097275689006168, "learning_rate": 9.448483168470628e-06, "loss": 0.408, "step": 5104 }, { "epoch": 0.35292084341514, "grad_norm": 0.7475827386216384, "learning_rate": 9.448227581297257e-06, "loss": 0.3471, "step": 5105 }, { "epoch": 0.35298997580366404, "grad_norm": 0.7251520662585846, "learning_rate": 9.447971938373293e-06, "loss": 0.3025, "step": 5106 }, { "epoch": 0.35305910819218805, "grad_norm": 0.7809658879939385, "learning_rate": 9.447716239701937e-06, "loss": 0.3046, "step": 5107 }, { "epoch": 0.35312824058071207, "grad_norm": 0.8210483391413201, "learning_rate": 9.447460485286395e-06, "loss": 0.4188, "step": 5108 }, { "epoch": 0.3531973729692361, "grad_norm": 0.7628800430338696, "learning_rate": 9.447204675129873e-06, "loss": 0.3869, "step": 5109 }, { "epoch": 0.3532665053577601, "grad_norm": 0.9247658586096792, "learning_rate": 9.446948809235578e-06, "loss": 0.2993, "step": 5110 }, { "epoch": 0.3533356377462841, "grad_norm": 0.9674140700679064, "learning_rate": 9.446692887606715e-06, "loss": 0.4102, "step": 5111 }, { "epoch": 0.35340477013480814, "grad_norm": 0.7131940074261578, "learning_rate": 9.446436910246491e-06, "loss": 0.4056, "step": 5112 }, { "epoch": 0.35347390252333216, "grad_norm": 0.9674167111855053, "learning_rate": 9.446180877158116e-06, "loss": 0.3229, "step": 5113 }, { "epoch": 0.3535430349118562, "grad_norm": 0.7967567671135585, "learning_rate": 9.445924788344799e-06, "loss": 0.3056, "step": 5114 }, { "epoch": 0.35361216730038025, "grad_norm": 0.8059108724888784, "learning_rate": 9.445668643809749e-06, "loss": 0.3963, "step": 5115 }, { "epoch": 0.35368129968890427, "grad_norm": 1.4420525447600472, "learning_rate": 9.445412443556174e-06, "loss": 0.3795, "step": 5116 }, { "epoch": 0.3537504320774283, "grad_norm": 0.7538991905992956, "learning_rate": 9.44515618758729e-06, "loss": 0.273, "step": 5117 }, { "epoch": 0.3538195644659523, "grad_norm": 0.943297280521891, "learning_rate": 9.444899875906305e-06, "loss": 0.3464, "step": 5118 }, { "epoch": 0.3538886968544763, "grad_norm": 0.7362235277141719, "learning_rate": 9.444643508516432e-06, "loss": 0.3785, "step": 5119 }, { "epoch": 0.35395782924300034, "grad_norm": 0.6849125684848424, "learning_rate": 9.444387085420884e-06, "loss": 0.2728, "step": 5120 }, { "epoch": 0.35402696163152436, "grad_norm": 0.778152866529319, "learning_rate": 9.444130606622876e-06, "loss": 0.3486, "step": 5121 }, { "epoch": 0.3540960940200484, "grad_norm": 0.8979554610425406, "learning_rate": 9.443874072125621e-06, "loss": 0.3803, "step": 5122 }, { "epoch": 0.3541652264085724, "grad_norm": 0.9111204551120708, "learning_rate": 9.443617481932337e-06, "loss": 0.3426, "step": 5123 }, { "epoch": 0.3542343587970964, "grad_norm": 0.6819857976079516, "learning_rate": 9.443360836046236e-06, "loss": 0.3053, "step": 5124 }, { "epoch": 0.3543034911856205, "grad_norm": 0.8296867219162537, "learning_rate": 9.443104134470535e-06, "loss": 0.4518, "step": 5125 }, { "epoch": 0.3543726235741445, "grad_norm": 0.8016951503052822, "learning_rate": 9.442847377208455e-06, "loss": 0.3621, "step": 5126 }, { "epoch": 0.3544417559626685, "grad_norm": 1.0235601045292197, "learning_rate": 9.442590564263211e-06, "loss": 0.3971, "step": 5127 }, { "epoch": 0.35451088835119254, "grad_norm": 0.7602860618892667, "learning_rate": 9.442333695638023e-06, "loss": 0.2643, "step": 5128 }, { "epoch": 0.35458002073971656, "grad_norm": 0.7509447011578286, "learning_rate": 9.44207677133611e-06, "loss": 0.4132, "step": 5129 }, { "epoch": 0.3546491531282406, "grad_norm": 0.8884959295187773, "learning_rate": 9.44181979136069e-06, "loss": 0.3656, "step": 5130 }, { "epoch": 0.3547182855167646, "grad_norm": 1.237272195826416, "learning_rate": 9.441562755714988e-06, "loss": 0.2984, "step": 5131 }, { "epoch": 0.3547874179052886, "grad_norm": 0.7911774317477787, "learning_rate": 9.441305664402223e-06, "loss": 0.4444, "step": 5132 }, { "epoch": 0.35485655029381263, "grad_norm": 0.9490760908507526, "learning_rate": 9.441048517425617e-06, "loss": 0.4538, "step": 5133 }, { "epoch": 0.35492568268233665, "grad_norm": 1.5110075290201237, "learning_rate": 9.440791314788394e-06, "loss": 0.3197, "step": 5134 }, { "epoch": 0.3549948150708607, "grad_norm": 0.6903542908204394, "learning_rate": 9.440534056493774e-06, "loss": 0.3026, "step": 5135 }, { "epoch": 0.35506394745938474, "grad_norm": 1.2092572850908272, "learning_rate": 9.440276742544987e-06, "loss": 0.4065, "step": 5136 }, { "epoch": 0.35513307984790876, "grad_norm": 1.5630427769178692, "learning_rate": 9.440019372945252e-06, "loss": 0.4439, "step": 5137 }, { "epoch": 0.3552022122364328, "grad_norm": 0.9398825326433823, "learning_rate": 9.439761947697801e-06, "loss": 0.2901, "step": 5138 }, { "epoch": 0.3552713446249568, "grad_norm": 0.8845044990098803, "learning_rate": 9.439504466805855e-06, "loss": 0.3573, "step": 5139 }, { "epoch": 0.3553404770134808, "grad_norm": 1.7256686225193083, "learning_rate": 9.439246930272644e-06, "loss": 0.4103, "step": 5140 }, { "epoch": 0.35540960940200483, "grad_norm": 0.8858885365254882, "learning_rate": 9.438989338101396e-06, "loss": 0.3425, "step": 5141 }, { "epoch": 0.35547874179052885, "grad_norm": 0.7986339738272006, "learning_rate": 9.438731690295336e-06, "loss": 0.324, "step": 5142 }, { "epoch": 0.35554787417905287, "grad_norm": 1.7795057111380237, "learning_rate": 9.438473986857698e-06, "loss": 0.3652, "step": 5143 }, { "epoch": 0.3556170065675769, "grad_norm": 0.8312062565715942, "learning_rate": 9.438216227791708e-06, "loss": 0.4041, "step": 5144 }, { "epoch": 0.3556861389561009, "grad_norm": 0.7900546810257327, "learning_rate": 9.437958413100598e-06, "loss": 0.3112, "step": 5145 }, { "epoch": 0.355755271344625, "grad_norm": 0.9318310293282573, "learning_rate": 9.437700542787599e-06, "loss": 0.3258, "step": 5146 }, { "epoch": 0.355824403733149, "grad_norm": 1.1961884281240256, "learning_rate": 9.437442616855944e-06, "loss": 0.5044, "step": 5147 }, { "epoch": 0.355893536121673, "grad_norm": 0.8795280369361262, "learning_rate": 9.437184635308862e-06, "loss": 0.4376, "step": 5148 }, { "epoch": 0.35596266851019703, "grad_norm": 0.7510520891718206, "learning_rate": 9.436926598149592e-06, "loss": 0.3094, "step": 5149 }, { "epoch": 0.35603180089872105, "grad_norm": 1.0023358937451816, "learning_rate": 9.436668505381364e-06, "loss": 0.3103, "step": 5150 }, { "epoch": 0.35610093328724507, "grad_norm": 0.8597160860026004, "learning_rate": 9.436410357007415e-06, "loss": 0.4389, "step": 5151 }, { "epoch": 0.3561700656757691, "grad_norm": 0.9965596810124617, "learning_rate": 9.43615215303098e-06, "loss": 0.3201, "step": 5152 }, { "epoch": 0.3562391980642931, "grad_norm": 0.8562668563854787, "learning_rate": 9.435893893455292e-06, "loss": 0.3462, "step": 5153 }, { "epoch": 0.3563083304528171, "grad_norm": 0.8202913403047765, "learning_rate": 9.43563557828359e-06, "loss": 0.3973, "step": 5154 }, { "epoch": 0.35637746284134114, "grad_norm": 0.7202790687509604, "learning_rate": 9.435377207519115e-06, "loss": 0.3725, "step": 5155 }, { "epoch": 0.3564465952298652, "grad_norm": 0.755024310721575, "learning_rate": 9.435118781165099e-06, "loss": 0.3602, "step": 5156 }, { "epoch": 0.35651572761838923, "grad_norm": 0.7777503175625641, "learning_rate": 9.434860299224784e-06, "loss": 0.4437, "step": 5157 }, { "epoch": 0.35658486000691325, "grad_norm": 0.9645524520216238, "learning_rate": 9.434601761701412e-06, "loss": 0.4558, "step": 5158 }, { "epoch": 0.35665399239543727, "grad_norm": 0.9892568869043546, "learning_rate": 9.434343168598218e-06, "loss": 0.3387, "step": 5159 }, { "epoch": 0.3567231247839613, "grad_norm": 0.8562312964977528, "learning_rate": 9.434084519918447e-06, "loss": 0.2921, "step": 5160 }, { "epoch": 0.3567922571724853, "grad_norm": 0.7980643230323031, "learning_rate": 9.433825815665341e-06, "loss": 0.3999, "step": 5161 }, { "epoch": 0.3568613895610093, "grad_norm": 0.7652596011753123, "learning_rate": 9.433567055842138e-06, "loss": 0.3877, "step": 5162 }, { "epoch": 0.35693052194953334, "grad_norm": 0.6918286613280449, "learning_rate": 9.433308240452086e-06, "loss": 0.3305, "step": 5163 }, { "epoch": 0.35699965433805736, "grad_norm": 0.892000146400956, "learning_rate": 9.433049369498426e-06, "loss": 0.3969, "step": 5164 }, { "epoch": 0.3570687867265814, "grad_norm": 0.7557373904240686, "learning_rate": 9.432790442984404e-06, "loss": 0.4464, "step": 5165 }, { "epoch": 0.35713791911510545, "grad_norm": 0.7051339742805501, "learning_rate": 9.432531460913264e-06, "loss": 0.2791, "step": 5166 }, { "epoch": 0.35720705150362947, "grad_norm": 0.780790855650142, "learning_rate": 9.432272423288252e-06, "loss": 0.3268, "step": 5167 }, { "epoch": 0.3572761838921535, "grad_norm": 0.7632796637179932, "learning_rate": 9.432013330112617e-06, "loss": 0.3417, "step": 5168 }, { "epoch": 0.3573453162806775, "grad_norm": 0.817753284509001, "learning_rate": 9.431754181389601e-06, "loss": 0.4351, "step": 5169 }, { "epoch": 0.3574144486692015, "grad_norm": 0.89447024977008, "learning_rate": 9.431494977122457e-06, "loss": 0.3269, "step": 5170 }, { "epoch": 0.35748358105772554, "grad_norm": 0.7042484745554792, "learning_rate": 9.431235717314432e-06, "loss": 0.3433, "step": 5171 }, { "epoch": 0.35755271344624956, "grad_norm": 0.7116242107954066, "learning_rate": 9.430976401968775e-06, "loss": 0.3814, "step": 5172 }, { "epoch": 0.3576218458347736, "grad_norm": 0.7707843707474106, "learning_rate": 9.430717031088736e-06, "loss": 0.3451, "step": 5173 }, { "epoch": 0.3576909782232976, "grad_norm": 0.7269814714565572, "learning_rate": 9.430457604677566e-06, "loss": 0.3659, "step": 5174 }, { "epoch": 0.3577601106118216, "grad_norm": 0.7531903242739428, "learning_rate": 9.430198122738516e-06, "loss": 0.324, "step": 5175 }, { "epoch": 0.3578292430003457, "grad_norm": 0.8169730983251163, "learning_rate": 9.429938585274838e-06, "loss": 0.3837, "step": 5176 }, { "epoch": 0.3578983753888697, "grad_norm": 0.8142113231488958, "learning_rate": 9.429678992289787e-06, "loss": 0.3215, "step": 5177 }, { "epoch": 0.3579675077773937, "grad_norm": 0.7231339616009681, "learning_rate": 9.429419343786614e-06, "loss": 0.2984, "step": 5178 }, { "epoch": 0.35803664016591774, "grad_norm": 0.858553801863839, "learning_rate": 9.429159639768575e-06, "loss": 0.4237, "step": 5179 }, { "epoch": 0.35810577255444176, "grad_norm": 0.7888778659653474, "learning_rate": 9.428899880238923e-06, "loss": 0.4044, "step": 5180 }, { "epoch": 0.3581749049429658, "grad_norm": 0.796884519652423, "learning_rate": 9.428640065200916e-06, "loss": 0.3373, "step": 5181 }, { "epoch": 0.3582440373314898, "grad_norm": 0.8892668091042614, "learning_rate": 9.428380194657807e-06, "loss": 0.3596, "step": 5182 }, { "epoch": 0.3583131697200138, "grad_norm": 0.703399663752259, "learning_rate": 9.428120268612857e-06, "loss": 0.4492, "step": 5183 }, { "epoch": 0.35838230210853783, "grad_norm": 0.7365359576572584, "learning_rate": 9.42786028706932e-06, "loss": 0.314, "step": 5184 }, { "epoch": 0.35845143449706185, "grad_norm": 0.7635696892480756, "learning_rate": 9.42760025003046e-06, "loss": 0.3504, "step": 5185 }, { "epoch": 0.3585205668855859, "grad_norm": 0.7468953957030868, "learning_rate": 9.427340157499529e-06, "loss": 0.4113, "step": 5186 }, { "epoch": 0.35858969927410994, "grad_norm": 0.7529681542649069, "learning_rate": 9.42708000947979e-06, "loss": 0.484, "step": 5187 }, { "epoch": 0.35865883166263396, "grad_norm": 0.6847073742165605, "learning_rate": 9.426819805974504e-06, "loss": 0.3111, "step": 5188 }, { "epoch": 0.358727964051158, "grad_norm": 1.3371308392859862, "learning_rate": 9.426559546986932e-06, "loss": 0.3317, "step": 5189 }, { "epoch": 0.358797096439682, "grad_norm": 0.7210732686885076, "learning_rate": 9.426299232520334e-06, "loss": 0.4115, "step": 5190 }, { "epoch": 0.358866228828206, "grad_norm": 0.693941138925969, "learning_rate": 9.426038862577977e-06, "loss": 0.4009, "step": 5191 }, { "epoch": 0.35893536121673003, "grad_norm": 0.7908285507308495, "learning_rate": 9.42577843716312e-06, "loss": 0.2937, "step": 5192 }, { "epoch": 0.35900449360525405, "grad_norm": 0.8006927102531223, "learning_rate": 9.425517956279028e-06, "loss": 0.4231, "step": 5193 }, { "epoch": 0.35907362599377807, "grad_norm": 0.8944778290042562, "learning_rate": 9.425257419928967e-06, "loss": 0.4411, "step": 5194 }, { "epoch": 0.3591427583823021, "grad_norm": 1.952250238670979, "learning_rate": 9.424996828116202e-06, "loss": 0.2827, "step": 5195 }, { "epoch": 0.35921189077082616, "grad_norm": 0.7629072050910537, "learning_rate": 9.424736180843999e-06, "loss": 0.3072, "step": 5196 }, { "epoch": 0.3592810231593502, "grad_norm": 0.758373998082916, "learning_rate": 9.424475478115622e-06, "loss": 0.4324, "step": 5197 }, { "epoch": 0.3593501555478742, "grad_norm": 1.128562848846035, "learning_rate": 9.424214719934341e-06, "loss": 0.335, "step": 5198 }, { "epoch": 0.3594192879363982, "grad_norm": 0.7539065788309643, "learning_rate": 9.423953906303425e-06, "loss": 0.336, "step": 5199 }, { "epoch": 0.35948842032492223, "grad_norm": 0.7588770117016899, "learning_rate": 9.423693037226143e-06, "loss": 0.3263, "step": 5200 }, { "epoch": 0.35955755271344625, "grad_norm": 0.7256430828057875, "learning_rate": 9.423432112705761e-06, "loss": 0.3561, "step": 5201 }, { "epoch": 0.35962668510197027, "grad_norm": 0.7581051662201787, "learning_rate": 9.423171132745552e-06, "loss": 0.3272, "step": 5202 }, { "epoch": 0.3596958174904943, "grad_norm": 0.7454792015564901, "learning_rate": 9.422910097348787e-06, "loss": 0.2966, "step": 5203 }, { "epoch": 0.3597649498790183, "grad_norm": 0.7654702268702888, "learning_rate": 9.422649006518736e-06, "loss": 0.423, "step": 5204 }, { "epoch": 0.3598340822675423, "grad_norm": 0.6798041513875434, "learning_rate": 9.422387860258673e-06, "loss": 0.3453, "step": 5205 }, { "epoch": 0.3599032146560664, "grad_norm": 0.70918263664983, "learning_rate": 9.422126658571869e-06, "loss": 0.331, "step": 5206 }, { "epoch": 0.3599723470445904, "grad_norm": 0.7405059406381406, "learning_rate": 9.4218654014616e-06, "loss": 0.3338, "step": 5207 }, { "epoch": 0.36004147943311443, "grad_norm": 0.735038964228842, "learning_rate": 9.42160408893114e-06, "loss": 0.4378, "step": 5208 }, { "epoch": 0.36011061182163845, "grad_norm": 0.6936465554332584, "learning_rate": 9.421342720983761e-06, "loss": 0.3858, "step": 5209 }, { "epoch": 0.36017974421016247, "grad_norm": 0.7041329979422005, "learning_rate": 9.421081297622744e-06, "loss": 0.2949, "step": 5210 }, { "epoch": 0.3602488765986865, "grad_norm": 0.8741494270142933, "learning_rate": 9.420819818851361e-06, "loss": 0.3658, "step": 5211 }, { "epoch": 0.3603180089872105, "grad_norm": 0.7695430553001377, "learning_rate": 9.420558284672892e-06, "loss": 0.4606, "step": 5212 }, { "epoch": 0.3603871413757345, "grad_norm": 1.0354590228004854, "learning_rate": 9.420296695090615e-06, "loss": 0.3401, "step": 5213 }, { "epoch": 0.36045627376425854, "grad_norm": 2.4563780866056057, "learning_rate": 9.420035050107803e-06, "loss": 0.3099, "step": 5214 }, { "epoch": 0.36052540615278256, "grad_norm": 0.9314771503589201, "learning_rate": 9.419773349727743e-06, "loss": 0.5103, "step": 5215 }, { "epoch": 0.3605945385413066, "grad_norm": 0.7186235412911777, "learning_rate": 9.41951159395371e-06, "loss": 0.3566, "step": 5216 }, { "epoch": 0.36066367092983065, "grad_norm": 0.7322964335434877, "learning_rate": 9.419249782788987e-06, "loss": 0.3166, "step": 5217 }, { "epoch": 0.36073280331835467, "grad_norm": 0.8156175074246161, "learning_rate": 9.418987916236856e-06, "loss": 0.3852, "step": 5218 }, { "epoch": 0.3608019357068787, "grad_norm": 0.735700475774543, "learning_rate": 9.418725994300594e-06, "loss": 0.4199, "step": 5219 }, { "epoch": 0.3608710680954027, "grad_norm": 0.8874197776477912, "learning_rate": 9.418464016983488e-06, "loss": 0.3047, "step": 5220 }, { "epoch": 0.3609402004839267, "grad_norm": 0.7906910229184023, "learning_rate": 9.418201984288822e-06, "loss": 0.2593, "step": 5221 }, { "epoch": 0.36100933287245074, "grad_norm": 0.7144623007135179, "learning_rate": 9.41793989621988e-06, "loss": 0.3852, "step": 5222 }, { "epoch": 0.36107846526097476, "grad_norm": 0.7270257793300333, "learning_rate": 9.417677752779944e-06, "loss": 0.3813, "step": 5223 }, { "epoch": 0.3611475976494988, "grad_norm": 0.7927883307223467, "learning_rate": 9.4174155539723e-06, "loss": 0.3615, "step": 5224 }, { "epoch": 0.3612167300380228, "grad_norm": 0.7408975367020164, "learning_rate": 9.417153299800237e-06, "loss": 0.3828, "step": 5225 }, { "epoch": 0.3612858624265468, "grad_norm": 0.8078532374736841, "learning_rate": 9.416890990267037e-06, "loss": 0.455, "step": 5226 }, { "epoch": 0.3613549948150709, "grad_norm": 0.7182812667399018, "learning_rate": 9.416628625375994e-06, "loss": 0.3785, "step": 5227 }, { "epoch": 0.3614241272035949, "grad_norm": 0.7115454281567138, "learning_rate": 9.416366205130393e-06, "loss": 0.2801, "step": 5228 }, { "epoch": 0.3614932595921189, "grad_norm": 0.6877125365939935, "learning_rate": 9.41610372953352e-06, "loss": 0.3654, "step": 5229 }, { "epoch": 0.36156239198064294, "grad_norm": 0.8261150813850439, "learning_rate": 9.415841198588671e-06, "loss": 0.357, "step": 5230 }, { "epoch": 0.36163152436916696, "grad_norm": 0.7736083148171602, "learning_rate": 9.415578612299131e-06, "loss": 0.297, "step": 5231 }, { "epoch": 0.361700656757691, "grad_norm": 0.7374190263662124, "learning_rate": 9.415315970668196e-06, "loss": 0.3939, "step": 5232 }, { "epoch": 0.361769789146215, "grad_norm": 0.8167982937821451, "learning_rate": 9.415053273699151e-06, "loss": 0.4802, "step": 5233 }, { "epoch": 0.361838921534739, "grad_norm": 1.7069574588809207, "learning_rate": 9.414790521395296e-06, "loss": 0.2402, "step": 5234 }, { "epoch": 0.36190805392326303, "grad_norm": 0.7331192899201113, "learning_rate": 9.414527713759918e-06, "loss": 0.3247, "step": 5235 }, { "epoch": 0.36197718631178705, "grad_norm": 0.8696062425162464, "learning_rate": 9.414264850796313e-06, "loss": 0.358, "step": 5236 }, { "epoch": 0.3620463187003111, "grad_norm": 0.7540137056132435, "learning_rate": 9.414001932507775e-06, "loss": 0.4793, "step": 5237 }, { "epoch": 0.36211545108883514, "grad_norm": 0.8082603637881394, "learning_rate": 9.413738958897602e-06, "loss": 0.2603, "step": 5238 }, { "epoch": 0.36218458347735916, "grad_norm": 0.742636652233084, "learning_rate": 9.413475929969086e-06, "loss": 0.2773, "step": 5239 }, { "epoch": 0.3622537158658832, "grad_norm": 0.7812721528768076, "learning_rate": 9.413212845725526e-06, "loss": 0.4109, "step": 5240 }, { "epoch": 0.3623228482544072, "grad_norm": 0.6918406856368278, "learning_rate": 9.41294970617022e-06, "loss": 0.3357, "step": 5241 }, { "epoch": 0.3623919806429312, "grad_norm": 0.7530279501436279, "learning_rate": 9.412686511306464e-06, "loss": 0.349, "step": 5242 }, { "epoch": 0.36246111303145523, "grad_norm": 0.7365327566671795, "learning_rate": 9.412423261137557e-06, "loss": 0.3904, "step": 5243 }, { "epoch": 0.36253024541997925, "grad_norm": 0.8293702308356893, "learning_rate": 9.412159955666797e-06, "loss": 0.4814, "step": 5244 }, { "epoch": 0.36259937780850326, "grad_norm": 0.9232817132676273, "learning_rate": 9.41189659489749e-06, "loss": 0.3382, "step": 5245 }, { "epoch": 0.3626685101970273, "grad_norm": 0.7725525587979868, "learning_rate": 9.411633178832928e-06, "loss": 0.2693, "step": 5246 }, { "epoch": 0.36273764258555136, "grad_norm": 0.7452415229888364, "learning_rate": 9.41136970747642e-06, "loss": 0.4259, "step": 5247 }, { "epoch": 0.3628067749740754, "grad_norm": 0.771205522483191, "learning_rate": 9.411106180831265e-06, "loss": 0.3838, "step": 5248 }, { "epoch": 0.3628759073625994, "grad_norm": 1.0827477121986582, "learning_rate": 9.410842598900764e-06, "loss": 0.2568, "step": 5249 }, { "epoch": 0.3629450397511234, "grad_norm": 0.7344401162858187, "learning_rate": 9.410578961688224e-06, "loss": 0.3688, "step": 5250 }, { "epoch": 0.36301417213964743, "grad_norm": 0.8068462835292847, "learning_rate": 9.410315269196948e-06, "loss": 0.5648, "step": 5251 }, { "epoch": 0.36308330452817145, "grad_norm": 0.7027990240387214, "learning_rate": 9.41005152143024e-06, "loss": 0.3437, "step": 5252 }, { "epoch": 0.36315243691669546, "grad_norm": 0.9594389451523617, "learning_rate": 9.409787718391408e-06, "loss": 0.3166, "step": 5253 }, { "epoch": 0.3632215693052195, "grad_norm": 0.7618713880786568, "learning_rate": 9.409523860083753e-06, "loss": 0.4398, "step": 5254 }, { "epoch": 0.3632907016937435, "grad_norm": 0.8263060090996974, "learning_rate": 9.40925994651059e-06, "loss": 0.3541, "step": 5255 }, { "epoch": 0.3633598340822675, "grad_norm": 0.7300995570046634, "learning_rate": 9.40899597767522e-06, "loss": 0.2949, "step": 5256 }, { "epoch": 0.3634289664707916, "grad_norm": 0.95274311316622, "learning_rate": 9.408731953580952e-06, "loss": 0.2999, "step": 5257 }, { "epoch": 0.3634980988593156, "grad_norm": 2.591652503133523, "learning_rate": 9.408467874231098e-06, "loss": 0.4429, "step": 5258 }, { "epoch": 0.36356723124783963, "grad_norm": 0.7438268797168084, "learning_rate": 9.408203739628967e-06, "loss": 0.3759, "step": 5259 }, { "epoch": 0.36363636363636365, "grad_norm": 0.694102589072205, "learning_rate": 9.40793954977787e-06, "loss": 0.33, "step": 5260 }, { "epoch": 0.36370549602488766, "grad_norm": 0.8268967201854777, "learning_rate": 9.407675304681114e-06, "loss": 0.3661, "step": 5261 }, { "epoch": 0.3637746284134117, "grad_norm": 0.9315383741224808, "learning_rate": 9.407411004342015e-06, "loss": 0.3676, "step": 5262 }, { "epoch": 0.3638437608019357, "grad_norm": 0.792828220066815, "learning_rate": 9.407146648763886e-06, "loss": 0.2999, "step": 5263 }, { "epoch": 0.3639128931904597, "grad_norm": 0.7273585073254388, "learning_rate": 9.406882237950038e-06, "loss": 0.3327, "step": 5264 }, { "epoch": 0.36398202557898374, "grad_norm": 0.8442280517268304, "learning_rate": 9.406617771903785e-06, "loss": 0.3986, "step": 5265 }, { "epoch": 0.36405115796750775, "grad_norm": 0.7821476323668427, "learning_rate": 9.406353250628442e-06, "loss": 0.3277, "step": 5266 }, { "epoch": 0.36412029035603183, "grad_norm": 0.667424997578523, "learning_rate": 9.406088674127324e-06, "loss": 0.2634, "step": 5267 }, { "epoch": 0.36418942274455585, "grad_norm": 0.7712812166517299, "learning_rate": 9.40582404240375e-06, "loss": 0.399, "step": 5268 }, { "epoch": 0.36425855513307986, "grad_norm": 0.8208605664489614, "learning_rate": 9.405559355461032e-06, "loss": 0.3988, "step": 5269 }, { "epoch": 0.3643276875216039, "grad_norm": 0.8344239128097048, "learning_rate": 9.405294613302491e-06, "loss": 0.3254, "step": 5270 }, { "epoch": 0.3643968199101279, "grad_norm": 0.7146656312978454, "learning_rate": 9.405029815931442e-06, "loss": 0.3144, "step": 5271 }, { "epoch": 0.3644659522986519, "grad_norm": 0.8158706512415329, "learning_rate": 9.404764963351208e-06, "loss": 0.4826, "step": 5272 }, { "epoch": 0.36453508468717594, "grad_norm": 0.7429441662724309, "learning_rate": 9.404500055565102e-06, "loss": 0.357, "step": 5273 }, { "epoch": 0.36460421707569995, "grad_norm": 0.8856415999087189, "learning_rate": 9.40423509257645e-06, "loss": 0.3406, "step": 5274 }, { "epoch": 0.364673349464224, "grad_norm": 0.8264156048674413, "learning_rate": 9.403970074388571e-06, "loss": 0.3822, "step": 5275 }, { "epoch": 0.364742481852748, "grad_norm": 0.7611668115119244, "learning_rate": 9.403705001004786e-06, "loss": 0.4061, "step": 5276 }, { "epoch": 0.364811614241272, "grad_norm": 0.7775247599026587, "learning_rate": 9.403439872428419e-06, "loss": 0.2975, "step": 5277 }, { "epoch": 0.3648807466297961, "grad_norm": 0.6940085775435586, "learning_rate": 9.40317468866279e-06, "loss": 0.3233, "step": 5278 }, { "epoch": 0.3649498790183201, "grad_norm": 0.8732055921479049, "learning_rate": 9.402909449711224e-06, "loss": 0.4275, "step": 5279 }, { "epoch": 0.3650190114068441, "grad_norm": 0.7328894327434883, "learning_rate": 9.402644155577046e-06, "loss": 0.4656, "step": 5280 }, { "epoch": 0.36508814379536814, "grad_norm": 0.6568674110398286, "learning_rate": 9.402378806263581e-06, "loss": 0.3146, "step": 5281 }, { "epoch": 0.36515727618389215, "grad_norm": 0.8146662209035467, "learning_rate": 9.402113401774152e-06, "loss": 0.3449, "step": 5282 }, { "epoch": 0.3652264085724162, "grad_norm": 0.7097246462324952, "learning_rate": 9.401847942112088e-06, "loss": 0.404, "step": 5283 }, { "epoch": 0.3652955409609402, "grad_norm": 0.7529259077808859, "learning_rate": 9.401582427280716e-06, "loss": 0.3149, "step": 5284 }, { "epoch": 0.3653646733494642, "grad_norm": 0.7624699776917555, "learning_rate": 9.401316857283365e-06, "loss": 0.3208, "step": 5285 }, { "epoch": 0.3654338057379882, "grad_norm": 0.8999927512271563, "learning_rate": 9.40105123212336e-06, "loss": 0.3563, "step": 5286 }, { "epoch": 0.36550293812651224, "grad_norm": 0.8487841035728735, "learning_rate": 9.40078555180403e-06, "loss": 0.4028, "step": 5287 }, { "epoch": 0.3655720705150363, "grad_norm": 0.7803384641000539, "learning_rate": 9.40051981632871e-06, "loss": 0.3459, "step": 5288 }, { "epoch": 0.36564120290356034, "grad_norm": 0.7743601281518061, "learning_rate": 9.400254025700726e-06, "loss": 0.3447, "step": 5289 }, { "epoch": 0.36571033529208435, "grad_norm": 0.7485804308251611, "learning_rate": 9.39998817992341e-06, "loss": 0.4305, "step": 5290 }, { "epoch": 0.36577946768060837, "grad_norm": 0.7140449194610529, "learning_rate": 9.399722279000094e-06, "loss": 0.3888, "step": 5291 }, { "epoch": 0.3658486000691324, "grad_norm": 0.8938520706738279, "learning_rate": 9.399456322934111e-06, "loss": 0.3261, "step": 5292 }, { "epoch": 0.3659177324576564, "grad_norm": 0.937123187689992, "learning_rate": 9.399190311728795e-06, "loss": 0.4894, "step": 5293 }, { "epoch": 0.3659868648461804, "grad_norm": 0.7684586987411564, "learning_rate": 9.398924245387479e-06, "loss": 0.4413, "step": 5294 }, { "epoch": 0.36605599723470444, "grad_norm": 0.9498272206819899, "learning_rate": 9.398658123913497e-06, "loss": 0.3427, "step": 5295 }, { "epoch": 0.36612512962322846, "grad_norm": 1.8514586989265929, "learning_rate": 9.398391947310186e-06, "loss": 0.2962, "step": 5296 }, { "epoch": 0.3661942620117525, "grad_norm": 0.7174109521922508, "learning_rate": 9.39812571558088e-06, "loss": 0.4373, "step": 5297 }, { "epoch": 0.36626339440027655, "grad_norm": 0.7906529987265983, "learning_rate": 9.397859428728918e-06, "loss": 0.4548, "step": 5298 }, { "epoch": 0.36633252678880057, "grad_norm": 0.7424284140037105, "learning_rate": 9.397593086757636e-06, "loss": 0.3031, "step": 5299 }, { "epoch": 0.3664016591773246, "grad_norm": 0.745275318240749, "learning_rate": 9.397326689670374e-06, "loss": 0.3652, "step": 5300 }, { "epoch": 0.3664707915658486, "grad_norm": 0.7866951804018836, "learning_rate": 9.397060237470468e-06, "loss": 0.4826, "step": 5301 }, { "epoch": 0.3665399239543726, "grad_norm": 0.8000415869139051, "learning_rate": 9.396793730161258e-06, "loss": 0.3388, "step": 5302 }, { "epoch": 0.36660905634289664, "grad_norm": 0.9934926504131845, "learning_rate": 9.396527167746086e-06, "loss": 0.2828, "step": 5303 }, { "epoch": 0.36667818873142066, "grad_norm": 0.7433700149068195, "learning_rate": 9.396260550228292e-06, "loss": 0.4546, "step": 5304 }, { "epoch": 0.3667473211199447, "grad_norm": 0.701365952382144, "learning_rate": 9.395993877611216e-06, "loss": 0.4182, "step": 5305 }, { "epoch": 0.3668164535084687, "grad_norm": 0.927634884326083, "learning_rate": 9.395727149898204e-06, "loss": 0.2527, "step": 5306 }, { "epoch": 0.3668855858969927, "grad_norm": 0.8234531724645208, "learning_rate": 9.395460367092596e-06, "loss": 0.3135, "step": 5307 }, { "epoch": 0.3669547182855168, "grad_norm": 0.9074764973502811, "learning_rate": 9.395193529197736e-06, "loss": 0.4693, "step": 5308 }, { "epoch": 0.3670238506740408, "grad_norm": 0.7308877526434175, "learning_rate": 9.394926636216968e-06, "loss": 0.3156, "step": 5309 }, { "epoch": 0.3670929830625648, "grad_norm": 3.2634802909468386, "learning_rate": 9.39465968815364e-06, "loss": 0.3262, "step": 5310 }, { "epoch": 0.36716211545108884, "grad_norm": 0.7985520311381991, "learning_rate": 9.394392685011093e-06, "loss": 0.4235, "step": 5311 }, { "epoch": 0.36723124783961286, "grad_norm": 0.7059029734245043, "learning_rate": 9.394125626792677e-06, "loss": 0.3722, "step": 5312 }, { "epoch": 0.3673003802281369, "grad_norm": 1.342594944024043, "learning_rate": 9.393858513501737e-06, "loss": 0.306, "step": 5313 }, { "epoch": 0.3673695126166609, "grad_norm": 1.0004402927444467, "learning_rate": 9.393591345141624e-06, "loss": 0.3378, "step": 5314 }, { "epoch": 0.3674386450051849, "grad_norm": 0.8509865699337494, "learning_rate": 9.393324121715681e-06, "loss": 0.4307, "step": 5315 }, { "epoch": 0.36750777739370893, "grad_norm": 0.8108887453783534, "learning_rate": 9.393056843227263e-06, "loss": 0.3313, "step": 5316 }, { "epoch": 0.36757690978223295, "grad_norm": 0.7560201339780102, "learning_rate": 9.392789509679717e-06, "loss": 0.333, "step": 5317 }, { "epoch": 0.367646042170757, "grad_norm": 0.7374244766350222, "learning_rate": 9.392522121076393e-06, "loss": 0.3697, "step": 5318 }, { "epoch": 0.36771517455928104, "grad_norm": 0.7842777741818544, "learning_rate": 9.392254677420643e-06, "loss": 0.5306, "step": 5319 }, { "epoch": 0.36778430694780506, "grad_norm": 0.7803798818279273, "learning_rate": 9.391987178715817e-06, "loss": 0.344, "step": 5320 }, { "epoch": 0.3678534393363291, "grad_norm": 1.8765244863891273, "learning_rate": 9.391719624965271e-06, "loss": 0.3054, "step": 5321 }, { "epoch": 0.3679225717248531, "grad_norm": 0.8146469529185845, "learning_rate": 9.39145201617236e-06, "loss": 0.4047, "step": 5322 }, { "epoch": 0.3679917041133771, "grad_norm": 0.8330544429368921, "learning_rate": 9.391184352340433e-06, "loss": 0.3811, "step": 5323 }, { "epoch": 0.36806083650190113, "grad_norm": 0.9150752892962589, "learning_rate": 9.390916633472843e-06, "loss": 0.3234, "step": 5324 }, { "epoch": 0.36812996889042515, "grad_norm": 0.7646751374842801, "learning_rate": 9.390648859572952e-06, "loss": 0.3144, "step": 5325 }, { "epoch": 0.36819910127894917, "grad_norm": 0.7538193039427274, "learning_rate": 9.390381030644114e-06, "loss": 0.4071, "step": 5326 }, { "epoch": 0.3682682336674732, "grad_norm": 1.0073668591735012, "learning_rate": 9.390113146689683e-06, "loss": 0.3, "step": 5327 }, { "epoch": 0.36833736605599726, "grad_norm": 0.8441716995272891, "learning_rate": 9.38984520771302e-06, "loss": 0.4085, "step": 5328 }, { "epoch": 0.3684064984445213, "grad_norm": 0.8657425398666646, "learning_rate": 9.38957721371748e-06, "loss": 0.37, "step": 5329 }, { "epoch": 0.3684756308330453, "grad_norm": 0.7989865403865737, "learning_rate": 9.389309164706422e-06, "loss": 0.3628, "step": 5330 }, { "epoch": 0.3685447632215693, "grad_norm": 1.0995959165333746, "learning_rate": 9.389041060683209e-06, "loss": 0.358, "step": 5331 }, { "epoch": 0.36861389561009333, "grad_norm": 1.0274755634920183, "learning_rate": 9.388772901651198e-06, "loss": 0.3039, "step": 5332 }, { "epoch": 0.36868302799861735, "grad_norm": 0.8056488283411094, "learning_rate": 9.388504687613751e-06, "loss": 0.4083, "step": 5333 }, { "epoch": 0.36875216038714137, "grad_norm": 0.7302204151214166, "learning_rate": 9.388236418574227e-06, "loss": 0.3176, "step": 5334 }, { "epoch": 0.3688212927756654, "grad_norm": 0.7932831367585536, "learning_rate": 9.387968094535995e-06, "loss": 0.3054, "step": 5335 }, { "epoch": 0.3688904251641894, "grad_norm": 0.8062955315552983, "learning_rate": 9.38769971550241e-06, "loss": 0.3906, "step": 5336 }, { "epoch": 0.3689595575527134, "grad_norm": 0.6943303567050758, "learning_rate": 9.387431281476842e-06, "loss": 0.3999, "step": 5337 }, { "epoch": 0.36902868994123744, "grad_norm": 0.8039650621584293, "learning_rate": 9.387162792462651e-06, "loss": 0.2763, "step": 5338 }, { "epoch": 0.3690978223297615, "grad_norm": 0.8994669334078851, "learning_rate": 9.386894248463204e-06, "loss": 0.3646, "step": 5339 }, { "epoch": 0.36916695471828553, "grad_norm": 0.7442394021625041, "learning_rate": 9.386625649481866e-06, "loss": 0.3915, "step": 5340 }, { "epoch": 0.36923608710680955, "grad_norm": 1.0363323218383473, "learning_rate": 9.386356995522004e-06, "loss": 0.3569, "step": 5341 }, { "epoch": 0.36930521949533357, "grad_norm": 0.7216108297111786, "learning_rate": 9.386088286586984e-06, "loss": 0.3354, "step": 5342 }, { "epoch": 0.3693743518838576, "grad_norm": 0.7437581361087444, "learning_rate": 9.385819522680177e-06, "loss": 0.3961, "step": 5343 }, { "epoch": 0.3694434842723816, "grad_norm": 0.7324895247648301, "learning_rate": 9.385550703804948e-06, "loss": 0.406, "step": 5344 }, { "epoch": 0.3695126166609056, "grad_norm": 0.7388764189924178, "learning_rate": 9.385281829964667e-06, "loss": 0.317, "step": 5345 }, { "epoch": 0.36958174904942964, "grad_norm": 0.7103493855494055, "learning_rate": 9.385012901162703e-06, "loss": 0.3414, "step": 5346 }, { "epoch": 0.36965088143795366, "grad_norm": 0.780895120234698, "learning_rate": 9.38474391740243e-06, "loss": 0.5072, "step": 5347 }, { "epoch": 0.3697200138264777, "grad_norm": 0.8438607097241538, "learning_rate": 9.384474878687216e-06, "loss": 0.3526, "step": 5348 }, { "epoch": 0.36978914621500175, "grad_norm": 0.8315061317618369, "learning_rate": 9.384205785020433e-06, "loss": 0.3226, "step": 5349 }, { "epoch": 0.36985827860352577, "grad_norm": 0.7340674170023712, "learning_rate": 9.383936636405456e-06, "loss": 0.3938, "step": 5350 }, { "epoch": 0.3699274109920498, "grad_norm": 0.7116538662742103, "learning_rate": 9.383667432845655e-06, "loss": 0.4515, "step": 5351 }, { "epoch": 0.3699965433805738, "grad_norm": 0.8806864182634397, "learning_rate": 9.383398174344405e-06, "loss": 0.33, "step": 5352 }, { "epoch": 0.3700656757690978, "grad_norm": 0.7411110643086064, "learning_rate": 9.383128860905084e-06, "loss": 0.2909, "step": 5353 }, { "epoch": 0.37013480815762184, "grad_norm": 0.7816888138686904, "learning_rate": 9.382859492531064e-06, "loss": 0.4118, "step": 5354 }, { "epoch": 0.37020394054614586, "grad_norm": 0.854640510767425, "learning_rate": 9.38259006922572e-06, "loss": 0.3741, "step": 5355 }, { "epoch": 0.3702730729346699, "grad_norm": 0.6699036496637032, "learning_rate": 9.382320590992432e-06, "loss": 0.2659, "step": 5356 }, { "epoch": 0.3703422053231939, "grad_norm": 1.402507867274114, "learning_rate": 9.382051057834577e-06, "loss": 0.3422, "step": 5357 }, { "epoch": 0.3704113377117179, "grad_norm": 0.747122350488144, "learning_rate": 9.381781469755531e-06, "loss": 0.4362, "step": 5358 }, { "epoch": 0.370480470100242, "grad_norm": 0.8791662002083807, "learning_rate": 9.381511826758673e-06, "loss": 0.3964, "step": 5359 }, { "epoch": 0.370549602488766, "grad_norm": 0.7098407224984424, "learning_rate": 9.381242128847385e-06, "loss": 0.3337, "step": 5360 }, { "epoch": 0.37061873487729, "grad_norm": 3.5939241245110214, "learning_rate": 9.380972376025046e-06, "loss": 0.4145, "step": 5361 }, { "epoch": 0.37068786726581404, "grad_norm": 0.7594761808501731, "learning_rate": 9.380702568295035e-06, "loss": 0.3768, "step": 5362 }, { "epoch": 0.37075699965433806, "grad_norm": 0.6949134720871982, "learning_rate": 9.380432705660736e-06, "loss": 0.3105, "step": 5363 }, { "epoch": 0.3708261320428621, "grad_norm": 0.8679266505341563, "learning_rate": 9.38016278812553e-06, "loss": 0.3208, "step": 5364 }, { "epoch": 0.3708952644313861, "grad_norm": 0.7967043314243616, "learning_rate": 9.379892815692799e-06, "loss": 0.4922, "step": 5365 }, { "epoch": 0.3709643968199101, "grad_norm": 0.8627013908135303, "learning_rate": 9.37962278836593e-06, "loss": 0.3673, "step": 5366 }, { "epoch": 0.37103352920843413, "grad_norm": 0.7843917472653112, "learning_rate": 9.379352706148304e-06, "loss": 0.3356, "step": 5367 }, { "epoch": 0.37110266159695815, "grad_norm": 0.9235781022298456, "learning_rate": 9.379082569043307e-06, "loss": 0.3634, "step": 5368 }, { "epoch": 0.3711717939854822, "grad_norm": 0.7566414610610992, "learning_rate": 9.378812377054325e-06, "loss": 0.4198, "step": 5369 }, { "epoch": 0.37124092637400624, "grad_norm": 0.7002707356598911, "learning_rate": 9.378542130184744e-06, "loss": 0.2795, "step": 5370 }, { "epoch": 0.37131005876253026, "grad_norm": 0.8354929684385749, "learning_rate": 9.378271828437952e-06, "loss": 0.32, "step": 5371 }, { "epoch": 0.3713791911510543, "grad_norm": 0.7447252598720162, "learning_rate": 9.378001471817335e-06, "loss": 0.3828, "step": 5372 }, { "epoch": 0.3714483235395783, "grad_norm": 0.8770706078934405, "learning_rate": 9.377731060326283e-06, "loss": 0.3817, "step": 5373 }, { "epoch": 0.3715174559281023, "grad_norm": 0.7033009665757487, "learning_rate": 9.377460593968183e-06, "loss": 0.33, "step": 5374 }, { "epoch": 0.37158658831662633, "grad_norm": 0.8215723385844699, "learning_rate": 9.37719007274643e-06, "loss": 0.3359, "step": 5375 }, { "epoch": 0.37165572070515035, "grad_norm": 0.7962825650179607, "learning_rate": 9.376919496664408e-06, "loss": 0.4517, "step": 5376 }, { "epoch": 0.37172485309367437, "grad_norm": 0.7567473987711821, "learning_rate": 9.376648865725513e-06, "loss": 0.3868, "step": 5377 }, { "epoch": 0.3717939854821984, "grad_norm": 0.7686308551012225, "learning_rate": 9.376378179933132e-06, "loss": 0.294, "step": 5378 }, { "epoch": 0.37186311787072246, "grad_norm": 0.8483497295715162, "learning_rate": 9.376107439290662e-06, "loss": 0.3808, "step": 5379 }, { "epoch": 0.3719322502592465, "grad_norm": 1.094936709639165, "learning_rate": 9.375836643801496e-06, "loss": 0.3771, "step": 5380 }, { "epoch": 0.3720013826477705, "grad_norm": 0.7429489103501196, "learning_rate": 9.375565793469026e-06, "loss": 0.317, "step": 5381 }, { "epoch": 0.3720705150362945, "grad_norm": 0.7936273042589506, "learning_rate": 9.375294888296647e-06, "loss": 0.4255, "step": 5382 }, { "epoch": 0.37213964742481853, "grad_norm": 0.8319059351575864, "learning_rate": 9.375023928287754e-06, "loss": 0.4271, "step": 5383 }, { "epoch": 0.37220877981334255, "grad_norm": 0.7981527289653637, "learning_rate": 9.374752913445745e-06, "loss": 0.3346, "step": 5384 }, { "epoch": 0.37227791220186657, "grad_norm": 0.7419804240134684, "learning_rate": 9.374481843774015e-06, "loss": 0.2939, "step": 5385 }, { "epoch": 0.3723470445903906, "grad_norm": 0.7949470093390589, "learning_rate": 9.37421071927596e-06, "loss": 0.418, "step": 5386 }, { "epoch": 0.3724161769789146, "grad_norm": 0.9012834626735746, "learning_rate": 9.373939539954981e-06, "loss": 0.3964, "step": 5387 }, { "epoch": 0.3724853093674386, "grad_norm": 0.8854964605630428, "learning_rate": 9.373668305814477e-06, "loss": 0.3901, "step": 5388 }, { "epoch": 0.3725544417559627, "grad_norm": 0.7316557983605787, "learning_rate": 9.373397016857844e-06, "loss": 0.2833, "step": 5389 }, { "epoch": 0.3726235741444867, "grad_norm": 0.8906209306755458, "learning_rate": 9.373125673088486e-06, "loss": 0.4136, "step": 5390 }, { "epoch": 0.37269270653301073, "grad_norm": 0.7397470703607805, "learning_rate": 9.372854274509799e-06, "loss": 0.3857, "step": 5391 }, { "epoch": 0.37276183892153475, "grad_norm": 0.739048882842144, "learning_rate": 9.372582821125189e-06, "loss": 0.3926, "step": 5392 }, { "epoch": 0.37283097131005877, "grad_norm": 1.3323012854874194, "learning_rate": 9.372311312938056e-06, "loss": 0.3458, "step": 5393 }, { "epoch": 0.3729001036985828, "grad_norm": 0.7303073091841844, "learning_rate": 9.372039749951805e-06, "loss": 0.4183, "step": 5394 }, { "epoch": 0.3729692360871068, "grad_norm": 0.7205478249612993, "learning_rate": 9.371768132169837e-06, "loss": 0.2873, "step": 5395 }, { "epoch": 0.3730383684756308, "grad_norm": 0.7085858717162545, "learning_rate": 9.371496459595556e-06, "loss": 0.3553, "step": 5396 }, { "epoch": 0.37310750086415484, "grad_norm": 0.8066655671810293, "learning_rate": 9.37122473223237e-06, "loss": 0.4067, "step": 5397 }, { "epoch": 0.37317663325267886, "grad_norm": 0.7438649003110214, "learning_rate": 9.370952950083683e-06, "loss": 0.4149, "step": 5398 }, { "epoch": 0.37324576564120293, "grad_norm": 0.7823846749918905, "learning_rate": 9.3706811131529e-06, "loss": 0.3563, "step": 5399 }, { "epoch": 0.37331489802972695, "grad_norm": 0.8682619419552442, "learning_rate": 9.370409221443429e-06, "loss": 0.3695, "step": 5400 }, { "epoch": 0.37338403041825097, "grad_norm": 0.86172781159521, "learning_rate": 9.370137274958677e-06, "loss": 0.4016, "step": 5401 }, { "epoch": 0.373453162806775, "grad_norm": 0.7263604303102881, "learning_rate": 9.369865273702055e-06, "loss": 0.3495, "step": 5402 }, { "epoch": 0.373522295195299, "grad_norm": 1.7587675996771797, "learning_rate": 9.36959321767697e-06, "loss": 0.2881, "step": 5403 }, { "epoch": 0.373591427583823, "grad_norm": 0.8023556004243274, "learning_rate": 9.369321106886831e-06, "loss": 0.4314, "step": 5404 }, { "epoch": 0.37366055997234704, "grad_norm": 1.1124328702666395, "learning_rate": 9.369048941335049e-06, "loss": 0.3762, "step": 5405 }, { "epoch": 0.37372969236087106, "grad_norm": 0.756447938509917, "learning_rate": 9.368776721025035e-06, "loss": 0.2967, "step": 5406 }, { "epoch": 0.3737988247493951, "grad_norm": 1.2888761745166426, "learning_rate": 9.368504445960203e-06, "loss": 0.3024, "step": 5407 }, { "epoch": 0.3738679571379191, "grad_norm": 0.9834179062099333, "learning_rate": 9.368232116143962e-06, "loss": 0.5295, "step": 5408 }, { "epoch": 0.3739370895264431, "grad_norm": 0.8647450290920933, "learning_rate": 9.367959731579728e-06, "loss": 0.3705, "step": 5409 }, { "epoch": 0.3740062219149672, "grad_norm": 0.8163278678200042, "learning_rate": 9.367687292270915e-06, "loss": 0.3936, "step": 5410 }, { "epoch": 0.3740753543034912, "grad_norm": 0.771917718462412, "learning_rate": 9.367414798220933e-06, "loss": 0.3492, "step": 5411 }, { "epoch": 0.3741444866920152, "grad_norm": 0.7640218047346414, "learning_rate": 9.367142249433204e-06, "loss": 0.3755, "step": 5412 }, { "epoch": 0.37421361908053924, "grad_norm": 0.9921526657852763, "learning_rate": 9.366869645911139e-06, "loss": 0.3859, "step": 5413 }, { "epoch": 0.37428275146906326, "grad_norm": 0.7956622267220368, "learning_rate": 9.366596987658153e-06, "loss": 0.388, "step": 5414 }, { "epoch": 0.3743518838575873, "grad_norm": 0.7856555885938262, "learning_rate": 9.36632427467767e-06, "loss": 0.4685, "step": 5415 }, { "epoch": 0.3744210162461113, "grad_norm": 0.8519112478085558, "learning_rate": 9.366051506973105e-06, "loss": 0.3264, "step": 5416 }, { "epoch": 0.3744901486346353, "grad_norm": 0.9038276910501997, "learning_rate": 9.365778684547873e-06, "loss": 0.2925, "step": 5417 }, { "epoch": 0.37455928102315933, "grad_norm": 0.7571317388102373, "learning_rate": 9.365505807405398e-06, "loss": 0.3398, "step": 5418 }, { "epoch": 0.37462841341168335, "grad_norm": 0.8288160369694738, "learning_rate": 9.365232875549098e-06, "loss": 0.4314, "step": 5419 }, { "epoch": 0.3746975458002074, "grad_norm": 0.6442163178910011, "learning_rate": 9.364959888982393e-06, "loss": 0.3081, "step": 5420 }, { "epoch": 0.37476667818873144, "grad_norm": 0.6419218866482828, "learning_rate": 9.364686847708705e-06, "loss": 0.2826, "step": 5421 }, { "epoch": 0.37483581057725546, "grad_norm": 0.7560616913131442, "learning_rate": 9.36441375173146e-06, "loss": 0.3966, "step": 5422 }, { "epoch": 0.3749049429657795, "grad_norm": 0.7620754537810035, "learning_rate": 9.364140601054075e-06, "loss": 0.4071, "step": 5423 }, { "epoch": 0.3749740753543035, "grad_norm": 0.9299162106998424, "learning_rate": 9.363867395679976e-06, "loss": 0.3408, "step": 5424 }, { "epoch": 0.3750432077428275, "grad_norm": 0.7844903694027425, "learning_rate": 9.363594135612585e-06, "loss": 0.4067, "step": 5425 }, { "epoch": 0.37511234013135153, "grad_norm": 0.9494243238068878, "learning_rate": 9.36332082085533e-06, "loss": 0.4414, "step": 5426 }, { "epoch": 0.37518147251987555, "grad_norm": 0.7809776131960591, "learning_rate": 9.363047451411637e-06, "loss": 0.3221, "step": 5427 }, { "epoch": 0.37525060490839957, "grad_norm": 0.7974022695299576, "learning_rate": 9.362774027284929e-06, "loss": 0.2869, "step": 5428 }, { "epoch": 0.3753197372969236, "grad_norm": 0.8470044573196237, "learning_rate": 9.362500548478633e-06, "loss": 0.4193, "step": 5429 }, { "epoch": 0.37538886968544766, "grad_norm": 0.8660451920301495, "learning_rate": 9.362227014996178e-06, "loss": 0.4096, "step": 5430 }, { "epoch": 0.3754580020739717, "grad_norm": 0.7053511054112648, "learning_rate": 9.361953426840996e-06, "loss": 0.3366, "step": 5431 }, { "epoch": 0.3755271344624957, "grad_norm": 0.8011624379763813, "learning_rate": 9.361679784016508e-06, "loss": 0.3776, "step": 5432 }, { "epoch": 0.3755962668510197, "grad_norm": 0.7614630089796651, "learning_rate": 9.36140608652615e-06, "loss": 0.4225, "step": 5433 }, { "epoch": 0.37566539923954373, "grad_norm": 0.7906107563590697, "learning_rate": 9.361132334373349e-06, "loss": 0.314, "step": 5434 }, { "epoch": 0.37573453162806775, "grad_norm": 0.8778819240849294, "learning_rate": 9.360858527561536e-06, "loss": 0.306, "step": 5435 }, { "epoch": 0.37580366401659177, "grad_norm": 0.8454873847219706, "learning_rate": 9.360584666094144e-06, "loss": 0.3656, "step": 5436 }, { "epoch": 0.3758727964051158, "grad_norm": 0.8374818983537985, "learning_rate": 9.360310749974606e-06, "loss": 0.3951, "step": 5437 }, { "epoch": 0.3759419287936398, "grad_norm": 0.879601329240751, "learning_rate": 9.360036779206354e-06, "loss": 0.2747, "step": 5438 }, { "epoch": 0.3760110611821638, "grad_norm": 0.7278569739294193, "learning_rate": 9.35976275379282e-06, "loss": 0.3488, "step": 5439 }, { "epoch": 0.3760801935706879, "grad_norm": 0.7606873734548697, "learning_rate": 9.359488673737443e-06, "loss": 0.4737, "step": 5440 }, { "epoch": 0.3761493259592119, "grad_norm": 0.7065017194272336, "learning_rate": 9.359214539043653e-06, "loss": 0.3289, "step": 5441 }, { "epoch": 0.37621845834773593, "grad_norm": 0.7656981090641043, "learning_rate": 9.35894034971489e-06, "loss": 0.3217, "step": 5442 }, { "epoch": 0.37628759073625995, "grad_norm": 0.8317788137225988, "learning_rate": 9.358666105754589e-06, "loss": 0.328, "step": 5443 }, { "epoch": 0.37635672312478397, "grad_norm": 0.6967849680336803, "learning_rate": 9.358391807166183e-06, "loss": 0.4649, "step": 5444 }, { "epoch": 0.376425855513308, "grad_norm": 0.8984464786483359, "learning_rate": 9.358117453953118e-06, "loss": 0.3438, "step": 5445 }, { "epoch": 0.376494987901832, "grad_norm": 0.6979574614521559, "learning_rate": 9.357843046118824e-06, "loss": 0.301, "step": 5446 }, { "epoch": 0.376564120290356, "grad_norm": 0.731040110516467, "learning_rate": 9.357568583666747e-06, "loss": 0.4849, "step": 5447 }, { "epoch": 0.37663325267888004, "grad_norm": 0.8317777902989375, "learning_rate": 9.357294066600325e-06, "loss": 0.396, "step": 5448 }, { "epoch": 0.37670238506740406, "grad_norm": 0.8521506674351278, "learning_rate": 9.357019494922997e-06, "loss": 0.3394, "step": 5449 }, { "epoch": 0.37677151745592813, "grad_norm": 0.8101200140426909, "learning_rate": 9.356744868638203e-06, "loss": 0.3617, "step": 5450 }, { "epoch": 0.37684064984445215, "grad_norm": 0.736029798731726, "learning_rate": 9.356470187749388e-06, "loss": 0.4639, "step": 5451 }, { "epoch": 0.37690978223297616, "grad_norm": 0.8603067982047762, "learning_rate": 9.356195452259995e-06, "loss": 0.3774, "step": 5452 }, { "epoch": 0.3769789146215002, "grad_norm": 0.7021136272733065, "learning_rate": 9.355920662173466e-06, "loss": 0.3089, "step": 5453 }, { "epoch": 0.3770480470100242, "grad_norm": 0.8669483048978855, "learning_rate": 9.355645817493245e-06, "loss": 0.4298, "step": 5454 }, { "epoch": 0.3771171793985482, "grad_norm": 0.8211144876497264, "learning_rate": 9.355370918222775e-06, "loss": 0.3707, "step": 5455 }, { "epoch": 0.37718631178707224, "grad_norm": 0.7512586921384247, "learning_rate": 9.355095964365504e-06, "loss": 0.3451, "step": 5456 }, { "epoch": 0.37725544417559626, "grad_norm": 0.6922179656822218, "learning_rate": 9.354820955924877e-06, "loss": 0.3075, "step": 5457 }, { "epoch": 0.3773245765641203, "grad_norm": 0.7404559073061155, "learning_rate": 9.354545892904342e-06, "loss": 0.3938, "step": 5458 }, { "epoch": 0.3773937089526443, "grad_norm": 0.7367523497658621, "learning_rate": 9.354270775307343e-06, "loss": 0.389, "step": 5459 }, { "epoch": 0.37746284134116836, "grad_norm": 0.7682987290670606, "learning_rate": 9.353995603137332e-06, "loss": 0.2837, "step": 5460 }, { "epoch": 0.3775319737296924, "grad_norm": 0.7850371060359977, "learning_rate": 9.353720376397755e-06, "loss": 0.3758, "step": 5461 }, { "epoch": 0.3776011061182164, "grad_norm": 0.7235526899755902, "learning_rate": 9.353445095092065e-06, "loss": 0.4227, "step": 5462 }, { "epoch": 0.3776702385067404, "grad_norm": 0.7456747089461875, "learning_rate": 9.353169759223709e-06, "loss": 0.3569, "step": 5463 }, { "epoch": 0.37773937089526444, "grad_norm": 0.8625575433566456, "learning_rate": 9.352894368796138e-06, "loss": 0.3743, "step": 5464 }, { "epoch": 0.37780850328378845, "grad_norm": 0.7014091126437116, "learning_rate": 9.352618923812803e-06, "loss": 0.3771, "step": 5465 }, { "epoch": 0.3778776356723125, "grad_norm": 0.8141409415776125, "learning_rate": 9.352343424277158e-06, "loss": 0.3723, "step": 5466 }, { "epoch": 0.3779467680608365, "grad_norm": 0.6738302391853642, "learning_rate": 9.352067870192655e-06, "loss": 0.3017, "step": 5467 }, { "epoch": 0.3780159004493605, "grad_norm": 0.9117772135631298, "learning_rate": 9.351792261562748e-06, "loss": 0.3155, "step": 5468 }, { "epoch": 0.3780850328378845, "grad_norm": 0.8771224844114747, "learning_rate": 9.351516598390893e-06, "loss": 0.3861, "step": 5469 }, { "epoch": 0.37815416522640855, "grad_norm": 0.7031424029988704, "learning_rate": 9.351240880680542e-06, "loss": 0.3119, "step": 5470 }, { "epoch": 0.3782232976149326, "grad_norm": 0.6794789180407454, "learning_rate": 9.35096510843515e-06, "loss": 0.3262, "step": 5471 }, { "epoch": 0.37829243000345664, "grad_norm": 0.8190818138205748, "learning_rate": 9.350689281658179e-06, "loss": 0.4334, "step": 5472 }, { "epoch": 0.37836156239198065, "grad_norm": 0.6737694914570564, "learning_rate": 9.350413400353077e-06, "loss": 0.3735, "step": 5473 }, { "epoch": 0.3784306947805047, "grad_norm": 0.7811406002144948, "learning_rate": 9.35013746452331e-06, "loss": 0.306, "step": 5474 }, { "epoch": 0.3784998271690287, "grad_norm": 0.6992871889033142, "learning_rate": 9.349861474172332e-06, "loss": 0.3685, "step": 5475 }, { "epoch": 0.3785689595575527, "grad_norm": 0.8012188698222837, "learning_rate": 9.349585429303604e-06, "loss": 0.4055, "step": 5476 }, { "epoch": 0.3786380919460767, "grad_norm": 0.734315112749426, "learning_rate": 9.349309329920583e-06, "loss": 0.319, "step": 5477 }, { "epoch": 0.37870722433460074, "grad_norm": 0.7368689098461072, "learning_rate": 9.349033176026733e-06, "loss": 0.3017, "step": 5478 }, { "epoch": 0.37877635672312476, "grad_norm": 0.7310730991013153, "learning_rate": 9.348756967625515e-06, "loss": 0.4466, "step": 5479 }, { "epoch": 0.3788454891116488, "grad_norm": 0.8548296227636102, "learning_rate": 9.348480704720386e-06, "loss": 0.3713, "step": 5480 }, { "epoch": 0.37891462150017285, "grad_norm": 1.2228576179240487, "learning_rate": 9.348204387314811e-06, "loss": 0.3432, "step": 5481 }, { "epoch": 0.3789837538886969, "grad_norm": 0.8465176507364137, "learning_rate": 9.347928015412256e-06, "loss": 0.3767, "step": 5482 }, { "epoch": 0.3790528862772209, "grad_norm": 0.7790437031970004, "learning_rate": 9.347651589016182e-06, "loss": 0.5166, "step": 5483 }, { "epoch": 0.3791220186657449, "grad_norm": 0.6931112136365866, "learning_rate": 9.347375108130054e-06, "loss": 0.3356, "step": 5484 }, { "epoch": 0.3791911510542689, "grad_norm": 0.740829654025946, "learning_rate": 9.347098572757337e-06, "loss": 0.3007, "step": 5485 }, { "epoch": 0.37926028344279294, "grad_norm": 0.7861165130820944, "learning_rate": 9.346821982901497e-06, "loss": 0.4684, "step": 5486 }, { "epoch": 0.37932941583131696, "grad_norm": 0.7966465122062292, "learning_rate": 9.346545338566e-06, "loss": 0.3928, "step": 5487 }, { "epoch": 0.379398548219841, "grad_norm": 0.9508581974553924, "learning_rate": 9.346268639754315e-06, "loss": 0.3172, "step": 5488 }, { "epoch": 0.379467680608365, "grad_norm": 0.8710625549648738, "learning_rate": 9.345991886469909e-06, "loss": 0.3268, "step": 5489 }, { "epoch": 0.379536812996889, "grad_norm": 0.7086245130595743, "learning_rate": 9.345715078716251e-06, "loss": 0.4004, "step": 5490 }, { "epoch": 0.3796059453854131, "grad_norm": 0.7211914223951962, "learning_rate": 9.345438216496808e-06, "loss": 0.3407, "step": 5491 }, { "epoch": 0.3796750777739371, "grad_norm": 0.7062970221586998, "learning_rate": 9.345161299815053e-06, "loss": 0.2687, "step": 5492 }, { "epoch": 0.3797442101624611, "grad_norm": 0.7723864698835605, "learning_rate": 9.344884328674454e-06, "loss": 0.3634, "step": 5493 }, { "epoch": 0.37981334255098514, "grad_norm": 0.766658487585225, "learning_rate": 9.344607303078486e-06, "loss": 0.4461, "step": 5494 }, { "epoch": 0.37988247493950916, "grad_norm": 0.8156817520964127, "learning_rate": 9.344330223030617e-06, "loss": 0.3819, "step": 5495 }, { "epoch": 0.3799516073280332, "grad_norm": 0.8940382789908703, "learning_rate": 9.344053088534322e-06, "loss": 0.3073, "step": 5496 }, { "epoch": 0.3800207397165572, "grad_norm": 0.8146532529415907, "learning_rate": 9.343775899593075e-06, "loss": 0.4422, "step": 5497 }, { "epoch": 0.3800898721050812, "grad_norm": 0.7036342270617767, "learning_rate": 9.343498656210347e-06, "loss": 0.3414, "step": 5498 }, { "epoch": 0.38015900449360523, "grad_norm": 0.795796689914394, "learning_rate": 9.343221358389616e-06, "loss": 0.2735, "step": 5499 }, { "epoch": 0.38022813688212925, "grad_norm": 0.9751358667566771, "learning_rate": 9.342944006134355e-06, "loss": 0.3102, "step": 5500 }, { "epoch": 0.3802972692706533, "grad_norm": 0.6975440479312658, "learning_rate": 9.342666599448042e-06, "loss": 0.4274, "step": 5501 }, { "epoch": 0.38036640165917734, "grad_norm": 0.6981542849691679, "learning_rate": 9.342389138334152e-06, "loss": 0.3134, "step": 5502 }, { "epoch": 0.38043553404770136, "grad_norm": 0.6911672481095312, "learning_rate": 9.342111622796165e-06, "loss": 0.3213, "step": 5503 }, { "epoch": 0.3805046664362254, "grad_norm": 0.7492922933736511, "learning_rate": 9.341834052837557e-06, "loss": 0.3833, "step": 5504 }, { "epoch": 0.3805737988247494, "grad_norm": 0.7609698692979322, "learning_rate": 9.341556428461806e-06, "loss": 0.4593, "step": 5505 }, { "epoch": 0.3806429312132734, "grad_norm": 0.9891931344184518, "learning_rate": 9.341278749672395e-06, "loss": 0.3059, "step": 5506 }, { "epoch": 0.38071206360179743, "grad_norm": 0.8661790417972096, "learning_rate": 9.341001016472802e-06, "loss": 0.377, "step": 5507 }, { "epoch": 0.38078119599032145, "grad_norm": 0.7638868476052401, "learning_rate": 9.340723228866508e-06, "loss": 0.4005, "step": 5508 }, { "epoch": 0.38085032837884547, "grad_norm": 0.835495900103658, "learning_rate": 9.340445386856995e-06, "loss": 0.3154, "step": 5509 }, { "epoch": 0.3809194607673695, "grad_norm": 0.6793967922907515, "learning_rate": 9.340167490447745e-06, "loss": 0.2694, "step": 5510 }, { "epoch": 0.38098859315589356, "grad_norm": 0.8433338472637033, "learning_rate": 9.33988953964224e-06, "loss": 0.3655, "step": 5511 }, { "epoch": 0.3810577255444176, "grad_norm": 0.7683576478163097, "learning_rate": 9.339611534443964e-06, "loss": 0.4332, "step": 5512 }, { "epoch": 0.3811268579329416, "grad_norm": 0.9940068464959075, "learning_rate": 9.339333474856402e-06, "loss": 0.2457, "step": 5513 }, { "epoch": 0.3811959903214656, "grad_norm": 0.8416901930439417, "learning_rate": 9.339055360883041e-06, "loss": 0.371, "step": 5514 }, { "epoch": 0.38126512270998963, "grad_norm": 0.7400499962839794, "learning_rate": 9.338777192527362e-06, "loss": 0.3927, "step": 5515 }, { "epoch": 0.38133425509851365, "grad_norm": 1.2585407399121673, "learning_rate": 9.338498969792855e-06, "loss": 0.369, "step": 5516 }, { "epoch": 0.38140338748703767, "grad_norm": 0.8631174808810037, "learning_rate": 9.338220692683005e-06, "loss": 0.3432, "step": 5517 }, { "epoch": 0.3814725198755617, "grad_norm": 0.7549927152905864, "learning_rate": 9.337942361201302e-06, "loss": 0.3697, "step": 5518 }, { "epoch": 0.3815416522640857, "grad_norm": 0.7395763818986694, "learning_rate": 9.337663975351231e-06, "loss": 0.4855, "step": 5519 }, { "epoch": 0.3816107846526097, "grad_norm": 0.7010910097873992, "learning_rate": 9.337385535136284e-06, "loss": 0.361, "step": 5520 }, { "epoch": 0.3816799170411338, "grad_norm": 0.7792562047089107, "learning_rate": 9.337107040559949e-06, "loss": 0.3042, "step": 5521 }, { "epoch": 0.3817490494296578, "grad_norm": 0.734959359760675, "learning_rate": 9.336828491625719e-06, "loss": 0.382, "step": 5522 }, { "epoch": 0.38181818181818183, "grad_norm": 0.9969884900074594, "learning_rate": 9.336549888337081e-06, "loss": 0.3323, "step": 5523 }, { "epoch": 0.38188731420670585, "grad_norm": 0.8048946351925947, "learning_rate": 9.33627123069753e-06, "loss": 0.3382, "step": 5524 }, { "epoch": 0.38195644659522987, "grad_norm": 1.036313695846114, "learning_rate": 9.335992518710558e-06, "loss": 0.3819, "step": 5525 }, { "epoch": 0.3820255789837539, "grad_norm": 0.9075382134642958, "learning_rate": 9.335713752379656e-06, "loss": 0.3551, "step": 5526 }, { "epoch": 0.3820947113722779, "grad_norm": 0.7621761693222566, "learning_rate": 9.335434931708322e-06, "loss": 0.3233, "step": 5527 }, { "epoch": 0.3821638437608019, "grad_norm": 0.7619741420940191, "learning_rate": 9.335156056700048e-06, "loss": 0.3215, "step": 5528 }, { "epoch": 0.38223297614932594, "grad_norm": 0.7745542931073102, "learning_rate": 9.334877127358327e-06, "loss": 0.3912, "step": 5529 }, { "epoch": 0.38230210853784996, "grad_norm": 0.8013143926907947, "learning_rate": 9.33459814368666e-06, "loss": 0.3741, "step": 5530 }, { "epoch": 0.382371240926374, "grad_norm": 0.8639321128165744, "learning_rate": 9.334319105688538e-06, "loss": 0.3548, "step": 5531 }, { "epoch": 0.38244037331489805, "grad_norm": 0.7595668719569085, "learning_rate": 9.334040013367462e-06, "loss": 0.3976, "step": 5532 }, { "epoch": 0.38250950570342207, "grad_norm": 0.8301189254087276, "learning_rate": 9.333760866726928e-06, "loss": 0.4435, "step": 5533 }, { "epoch": 0.3825786380919461, "grad_norm": 0.7692775975928237, "learning_rate": 9.333481665770437e-06, "loss": 0.3239, "step": 5534 }, { "epoch": 0.3826477704804701, "grad_norm": 0.7216523833453521, "learning_rate": 9.333202410501487e-06, "loss": 0.331, "step": 5535 }, { "epoch": 0.3827169028689941, "grad_norm": 0.7746690559212834, "learning_rate": 9.332923100923577e-06, "loss": 0.3661, "step": 5536 }, { "epoch": 0.38278603525751814, "grad_norm": 0.8304586076082211, "learning_rate": 9.332643737040209e-06, "loss": 0.4658, "step": 5537 }, { "epoch": 0.38285516764604216, "grad_norm": 0.8197350509101758, "learning_rate": 9.332364318854881e-06, "loss": 0.3387, "step": 5538 }, { "epoch": 0.3829243000345662, "grad_norm": 0.7750421811660386, "learning_rate": 9.332084846371099e-06, "loss": 0.3982, "step": 5539 }, { "epoch": 0.3829934324230902, "grad_norm": 0.7653284270437949, "learning_rate": 9.331805319592366e-06, "loss": 0.4806, "step": 5540 }, { "epoch": 0.3830625648116142, "grad_norm": 1.0264752533746808, "learning_rate": 9.331525738522183e-06, "loss": 0.3151, "step": 5541 }, { "epoch": 0.3831316972001383, "grad_norm": 0.6731034425821232, "learning_rate": 9.331246103164055e-06, "loss": 0.3136, "step": 5542 }, { "epoch": 0.3832008295886623, "grad_norm": 0.8148566916498952, "learning_rate": 9.330966413521484e-06, "loss": 0.3632, "step": 5543 }, { "epoch": 0.3832699619771863, "grad_norm": 0.8856747045269088, "learning_rate": 9.330686669597982e-06, "loss": 0.3964, "step": 5544 }, { "epoch": 0.38333909436571034, "grad_norm": 1.4260734303471498, "learning_rate": 9.330406871397048e-06, "loss": 0.3192, "step": 5545 }, { "epoch": 0.38340822675423436, "grad_norm": 0.7195776154897245, "learning_rate": 9.330127018922195e-06, "loss": 0.3106, "step": 5546 }, { "epoch": 0.3834773591427584, "grad_norm": 0.7914961314276971, "learning_rate": 9.329847112176924e-06, "loss": 0.4093, "step": 5547 }, { "epoch": 0.3835464915312824, "grad_norm": 0.7777003185778267, "learning_rate": 9.329567151164748e-06, "loss": 0.3827, "step": 5548 }, { "epoch": 0.3836156239198064, "grad_norm": 1.2138678391315176, "learning_rate": 9.329287135889174e-06, "loss": 0.3264, "step": 5549 }, { "epoch": 0.38368475630833043, "grad_norm": 0.7995245466986933, "learning_rate": 9.329007066353712e-06, "loss": 0.3787, "step": 5550 }, { "epoch": 0.38375388869685445, "grad_norm": 0.7274504660915462, "learning_rate": 9.328726942561872e-06, "loss": 0.3393, "step": 5551 }, { "epoch": 0.3838230210853785, "grad_norm": 0.7070685034148395, "learning_rate": 9.328446764517164e-06, "loss": 0.307, "step": 5552 }, { "epoch": 0.38389215347390254, "grad_norm": 0.8448817109213975, "learning_rate": 9.3281665322231e-06, "loss": 0.2957, "step": 5553 }, { "epoch": 0.38396128586242656, "grad_norm": 1.087563063980764, "learning_rate": 9.327886245683193e-06, "loss": 0.4221, "step": 5554 }, { "epoch": 0.3840304182509506, "grad_norm": 0.9142305092638077, "learning_rate": 9.327605904900956e-06, "loss": 0.3455, "step": 5555 }, { "epoch": 0.3840995506394746, "grad_norm": 0.6980880940711767, "learning_rate": 9.3273255098799e-06, "loss": 0.2858, "step": 5556 }, { "epoch": 0.3841686830279986, "grad_norm": 0.8612074563967138, "learning_rate": 9.327045060623543e-06, "loss": 0.3748, "step": 5557 }, { "epoch": 0.38423781541652263, "grad_norm": 0.7474024708659587, "learning_rate": 9.326764557135398e-06, "loss": 0.4136, "step": 5558 }, { "epoch": 0.38430694780504665, "grad_norm": 0.6774560385677807, "learning_rate": 9.32648399941898e-06, "loss": 0.3228, "step": 5559 }, { "epoch": 0.38437608019357067, "grad_norm": 1.1476731889951335, "learning_rate": 9.326203387477807e-06, "loss": 0.2622, "step": 5560 }, { "epoch": 0.3844452125820947, "grad_norm": 0.636669769730431, "learning_rate": 9.325922721315394e-06, "loss": 0.3755, "step": 5561 }, { "epoch": 0.38451434497061876, "grad_norm": 0.7440180401021126, "learning_rate": 9.32564200093526e-06, "loss": 0.3908, "step": 5562 }, { "epoch": 0.3845834773591428, "grad_norm": 0.9392132328223636, "learning_rate": 9.325361226340923e-06, "loss": 0.2757, "step": 5563 }, { "epoch": 0.3846526097476668, "grad_norm": 0.6901123689698307, "learning_rate": 9.325080397535901e-06, "loss": 0.2894, "step": 5564 }, { "epoch": 0.3847217421361908, "grad_norm": 0.7490069705948112, "learning_rate": 9.324799514523718e-06, "loss": 0.4255, "step": 5565 }, { "epoch": 0.38479087452471483, "grad_norm": 0.6967913635249495, "learning_rate": 9.324518577307888e-06, "loss": 0.3496, "step": 5566 }, { "epoch": 0.38486000691323885, "grad_norm": 0.8184725865079084, "learning_rate": 9.324237585891935e-06, "loss": 0.2641, "step": 5567 }, { "epoch": 0.38492913930176287, "grad_norm": 0.7173177115298992, "learning_rate": 9.323956540279381e-06, "loss": 0.3918, "step": 5568 }, { "epoch": 0.3849982716902869, "grad_norm": 0.7258973186219448, "learning_rate": 9.32367544047375e-06, "loss": 0.4174, "step": 5569 }, { "epoch": 0.3850674040788109, "grad_norm": 0.7470110483446574, "learning_rate": 9.323394286478562e-06, "loss": 0.4032, "step": 5570 }, { "epoch": 0.3851365364673349, "grad_norm": 0.9898634251844995, "learning_rate": 9.323113078297342e-06, "loss": 0.2947, "step": 5571 }, { "epoch": 0.385205668855859, "grad_norm": 0.8214217647795167, "learning_rate": 9.322831815933616e-06, "loss": 0.4358, "step": 5572 }, { "epoch": 0.385274801244383, "grad_norm": 0.8115808806308293, "learning_rate": 9.322550499390906e-06, "loss": 0.3306, "step": 5573 }, { "epoch": 0.38534393363290703, "grad_norm": 0.7595488369868654, "learning_rate": 9.322269128672738e-06, "loss": 0.3187, "step": 5574 }, { "epoch": 0.38541306602143105, "grad_norm": 0.7102645560943821, "learning_rate": 9.321987703782643e-06, "loss": 0.3001, "step": 5575 }, { "epoch": 0.38548219840995507, "grad_norm": 0.822074407339414, "learning_rate": 9.321706224724143e-06, "loss": 0.4441, "step": 5576 }, { "epoch": 0.3855513307984791, "grad_norm": 0.8984765159833723, "learning_rate": 9.321424691500769e-06, "loss": 0.3238, "step": 5577 }, { "epoch": 0.3856204631870031, "grad_norm": 0.8136249161628542, "learning_rate": 9.321143104116049e-06, "loss": 0.3248, "step": 5578 }, { "epoch": 0.3856895955755271, "grad_norm": 0.8490759626451327, "learning_rate": 9.32086146257351e-06, "loss": 0.4204, "step": 5579 }, { "epoch": 0.38575872796405114, "grad_norm": 0.7744017224503557, "learning_rate": 9.320579766876684e-06, "loss": 0.3835, "step": 5580 }, { "epoch": 0.38582786035257516, "grad_norm": 0.7583761127663352, "learning_rate": 9.320298017029102e-06, "loss": 0.3532, "step": 5581 }, { "epoch": 0.38589699274109923, "grad_norm": 0.7726691394407786, "learning_rate": 9.320016213034293e-06, "loss": 0.303, "step": 5582 }, { "epoch": 0.38596612512962325, "grad_norm": 0.7849983522734351, "learning_rate": 9.319734354895791e-06, "loss": 0.4179, "step": 5583 }, { "epoch": 0.38603525751814727, "grad_norm": 0.8703632354874501, "learning_rate": 9.319452442617127e-06, "loss": 0.3089, "step": 5584 }, { "epoch": 0.3861043899066713, "grad_norm": 0.8065216525831638, "learning_rate": 9.319170476201835e-06, "loss": 0.3648, "step": 5585 }, { "epoch": 0.3861735222951953, "grad_norm": 0.7925510083747632, "learning_rate": 9.31888845565345e-06, "loss": 0.4081, "step": 5586 }, { "epoch": 0.3862426546837193, "grad_norm": 0.7344728690964692, "learning_rate": 9.318606380975505e-06, "loss": 0.4151, "step": 5587 }, { "epoch": 0.38631178707224334, "grad_norm": 0.9392430631935254, "learning_rate": 9.318324252171535e-06, "loss": 0.2923, "step": 5588 }, { "epoch": 0.38638091946076736, "grad_norm": 1.4378426030394882, "learning_rate": 9.318042069245078e-06, "loss": 0.3732, "step": 5589 }, { "epoch": 0.3864500518492914, "grad_norm": 1.2951548916726963, "learning_rate": 9.31775983219967e-06, "loss": 0.4137, "step": 5590 }, { "epoch": 0.3865191842378154, "grad_norm": 0.7684045732840844, "learning_rate": 9.317477541038848e-06, "loss": 0.3812, "step": 5591 }, { "epoch": 0.38658831662633947, "grad_norm": 0.6605867582991259, "learning_rate": 9.317195195766149e-06, "loss": 0.2945, "step": 5592 }, { "epoch": 0.3866574490148635, "grad_norm": 0.7503565073674747, "learning_rate": 9.316912796385112e-06, "loss": 0.3559, "step": 5593 }, { "epoch": 0.3867265814033875, "grad_norm": 0.7964548566156533, "learning_rate": 9.316630342899278e-06, "loss": 0.4139, "step": 5594 }, { "epoch": 0.3867957137919115, "grad_norm": 1.2046149731631417, "learning_rate": 9.316347835312186e-06, "loss": 0.3394, "step": 5595 }, { "epoch": 0.38686484618043554, "grad_norm": 0.7084929430065334, "learning_rate": 9.316065273627376e-06, "loss": 0.2648, "step": 5596 }, { "epoch": 0.38693397856895956, "grad_norm": 0.8062881541981013, "learning_rate": 9.315782657848391e-06, "loss": 0.3782, "step": 5597 }, { "epoch": 0.3870031109574836, "grad_norm": 0.8533897236594338, "learning_rate": 9.315499987978772e-06, "loss": 0.39, "step": 5598 }, { "epoch": 0.3870722433460076, "grad_norm": 0.6762797964119385, "learning_rate": 9.315217264022064e-06, "loss": 0.2921, "step": 5599 }, { "epoch": 0.3871413757345316, "grad_norm": 0.8422652070644578, "learning_rate": 9.314934485981805e-06, "loss": 0.3362, "step": 5600 }, { "epoch": 0.38721050812305563, "grad_norm": 0.7598958259526435, "learning_rate": 9.314651653861544e-06, "loss": 0.462, "step": 5601 }, { "epoch": 0.38727964051157965, "grad_norm": 0.8225525634362401, "learning_rate": 9.314368767664824e-06, "loss": 0.2912, "step": 5602 }, { "epoch": 0.3873487729001037, "grad_norm": 0.8105485391575759, "learning_rate": 9.314085827395192e-06, "loss": 0.3132, "step": 5603 }, { "epoch": 0.38741790528862774, "grad_norm": 0.8049068992276434, "learning_rate": 9.313802833056191e-06, "loss": 0.3828, "step": 5604 }, { "epoch": 0.38748703767715176, "grad_norm": 0.7353044076801539, "learning_rate": 9.313519784651371e-06, "loss": 0.3784, "step": 5605 }, { "epoch": 0.3875561700656758, "grad_norm": 0.6654930594462191, "learning_rate": 9.313236682184279e-06, "loss": 0.3147, "step": 5606 }, { "epoch": 0.3876253024541998, "grad_norm": 0.8974767317292955, "learning_rate": 9.312953525658461e-06, "loss": 0.3477, "step": 5607 }, { "epoch": 0.3876944348427238, "grad_norm": 0.7244420096106257, "learning_rate": 9.312670315077469e-06, "loss": 0.4077, "step": 5608 }, { "epoch": 0.38776356723124783, "grad_norm": 0.7764859293220783, "learning_rate": 9.31238705044485e-06, "loss": 0.3286, "step": 5609 }, { "epoch": 0.38783269961977185, "grad_norm": 0.8190358525526689, "learning_rate": 9.312103731764155e-06, "loss": 0.3087, "step": 5610 }, { "epoch": 0.38790183200829587, "grad_norm": 0.7668944465864669, "learning_rate": 9.311820359038936e-06, "loss": 0.3977, "step": 5611 }, { "epoch": 0.3879709643968199, "grad_norm": 0.874848430183661, "learning_rate": 9.311536932272741e-06, "loss": 0.4535, "step": 5612 }, { "epoch": 0.38804009678534396, "grad_norm": 0.8447542887396411, "learning_rate": 9.311253451469128e-06, "loss": 0.3027, "step": 5613 }, { "epoch": 0.388109229173868, "grad_norm": 0.7416094091298107, "learning_rate": 9.310969916631645e-06, "loss": 0.3021, "step": 5614 }, { "epoch": 0.388178361562392, "grad_norm": 1.038155532960713, "learning_rate": 9.310686327763848e-06, "loss": 0.4485, "step": 5615 }, { "epoch": 0.388247493950916, "grad_norm": 0.9706910808996171, "learning_rate": 9.310402684869291e-06, "loss": 0.3487, "step": 5616 }, { "epoch": 0.38831662633944003, "grad_norm": 0.843485885667575, "learning_rate": 9.310118987951528e-06, "loss": 0.3568, "step": 5617 }, { "epoch": 0.38838575872796405, "grad_norm": 0.827881888836062, "learning_rate": 9.309835237014116e-06, "loss": 0.3929, "step": 5618 }, { "epoch": 0.38845489111648807, "grad_norm": 0.7322006685599313, "learning_rate": 9.30955143206061e-06, "loss": 0.467, "step": 5619 }, { "epoch": 0.3885240235050121, "grad_norm": 0.7714330213977539, "learning_rate": 9.309267573094568e-06, "loss": 0.3131, "step": 5620 }, { "epoch": 0.3885931558935361, "grad_norm": 0.8113851805395174, "learning_rate": 9.308983660119546e-06, "loss": 0.3946, "step": 5621 }, { "epoch": 0.3886622882820601, "grad_norm": 0.7886040369351567, "learning_rate": 9.308699693139102e-06, "loss": 0.3507, "step": 5622 }, { "epoch": 0.3887314206705842, "grad_norm": 0.9070490646670161, "learning_rate": 9.3084156721568e-06, "loss": 0.3777, "step": 5623 }, { "epoch": 0.3888005530591082, "grad_norm": 0.7477709493723498, "learning_rate": 9.308131597176195e-06, "loss": 0.2892, "step": 5624 }, { "epoch": 0.38886968544763223, "grad_norm": 0.7494681868995894, "learning_rate": 9.307847468200847e-06, "loss": 0.2895, "step": 5625 }, { "epoch": 0.38893881783615625, "grad_norm": 0.8931369836671266, "learning_rate": 9.307563285234323e-06, "loss": 0.4291, "step": 5626 }, { "epoch": 0.38900795022468027, "grad_norm": 0.8984869071478857, "learning_rate": 9.307279048280176e-06, "loss": 0.3257, "step": 5627 }, { "epoch": 0.3890770826132043, "grad_norm": 0.682012614874959, "learning_rate": 9.306994757341975e-06, "loss": 0.3105, "step": 5628 }, { "epoch": 0.3891462150017283, "grad_norm": 0.766080655035707, "learning_rate": 9.30671041242328e-06, "loss": 0.4541, "step": 5629 }, { "epoch": 0.3892153473902523, "grad_norm": 0.8224974960264609, "learning_rate": 9.306426013527656e-06, "loss": 0.3948, "step": 5630 }, { "epoch": 0.38928447977877634, "grad_norm": 0.7953475808842402, "learning_rate": 9.306141560658666e-06, "loss": 0.3193, "step": 5631 }, { "epoch": 0.38935361216730036, "grad_norm": 0.824900152354724, "learning_rate": 9.305857053819878e-06, "loss": 0.3726, "step": 5632 }, { "epoch": 0.38942274455582443, "grad_norm": 0.9442669754802299, "learning_rate": 9.305572493014856e-06, "loss": 0.4389, "step": 5633 }, { "epoch": 0.38949187694434845, "grad_norm": 0.786940556105852, "learning_rate": 9.305287878247165e-06, "loss": 0.3768, "step": 5634 }, { "epoch": 0.38956100933287247, "grad_norm": 0.8408082278719266, "learning_rate": 9.305003209520374e-06, "loss": 0.3661, "step": 5635 }, { "epoch": 0.3896301417213965, "grad_norm": 0.7925100143619043, "learning_rate": 9.304718486838051e-06, "loss": 0.3301, "step": 5636 }, { "epoch": 0.3896992741099205, "grad_norm": 0.8485144303003614, "learning_rate": 9.304433710203764e-06, "loss": 0.402, "step": 5637 }, { "epoch": 0.3897684064984445, "grad_norm": 0.6187973688892197, "learning_rate": 9.304148879621082e-06, "loss": 0.3267, "step": 5638 }, { "epoch": 0.38983753888696854, "grad_norm": 0.6625708453145085, "learning_rate": 9.303863995093574e-06, "loss": 0.2903, "step": 5639 }, { "epoch": 0.38990667127549256, "grad_norm": 0.7487379203190024, "learning_rate": 9.303579056624813e-06, "loss": 0.4326, "step": 5640 }, { "epoch": 0.3899758036640166, "grad_norm": 0.9134498458418431, "learning_rate": 9.303294064218367e-06, "loss": 0.3361, "step": 5641 }, { "epoch": 0.3900449360525406, "grad_norm": 0.7506531572929119, "learning_rate": 9.30300901787781e-06, "loss": 0.3265, "step": 5642 }, { "epoch": 0.39011406844106467, "grad_norm": 0.8035346943681374, "learning_rate": 9.302723917606714e-06, "loss": 0.3083, "step": 5643 }, { "epoch": 0.3901832008295887, "grad_norm": 0.9545276851894711, "learning_rate": 9.302438763408652e-06, "loss": 0.4265, "step": 5644 }, { "epoch": 0.3902523332181127, "grad_norm": 1.0389291522789668, "learning_rate": 9.302153555287199e-06, "loss": 0.2548, "step": 5645 }, { "epoch": 0.3903214656066367, "grad_norm": 0.7901662094413472, "learning_rate": 9.301868293245927e-06, "loss": 0.3416, "step": 5646 }, { "epoch": 0.39039059799516074, "grad_norm": 0.7802791432071516, "learning_rate": 9.301582977288415e-06, "loss": 0.4342, "step": 5647 }, { "epoch": 0.39045973038368476, "grad_norm": 0.7107957124211113, "learning_rate": 9.301297607418237e-06, "loss": 0.3757, "step": 5648 }, { "epoch": 0.3905288627722088, "grad_norm": 0.8217937993525619, "learning_rate": 9.301012183638968e-06, "loss": 0.3323, "step": 5649 }, { "epoch": 0.3905979951607328, "grad_norm": 1.082981441869016, "learning_rate": 9.300726705954186e-06, "loss": 0.4023, "step": 5650 }, { "epoch": 0.3906671275492568, "grad_norm": 0.8299895347660524, "learning_rate": 9.300441174367472e-06, "loss": 0.4167, "step": 5651 }, { "epoch": 0.3907362599377808, "grad_norm": 1.0524537412636232, "learning_rate": 9.3001555888824e-06, "loss": 0.3089, "step": 5652 }, { "epoch": 0.3908053923263049, "grad_norm": 0.7777951237133194, "learning_rate": 9.299869949502554e-06, "loss": 0.3022, "step": 5653 }, { "epoch": 0.3908745247148289, "grad_norm": 0.7864225613288977, "learning_rate": 9.299584256231511e-06, "loss": 0.4469, "step": 5654 }, { "epoch": 0.39094365710335294, "grad_norm": 0.7786610179703127, "learning_rate": 9.299298509072851e-06, "loss": 0.4503, "step": 5655 }, { "epoch": 0.39101278949187696, "grad_norm": 0.9290365869819865, "learning_rate": 9.29901270803016e-06, "loss": 0.302, "step": 5656 }, { "epoch": 0.391081921880401, "grad_norm": 0.9876820231116362, "learning_rate": 9.298726853107013e-06, "loss": 0.3313, "step": 5657 }, { "epoch": 0.391151054268925, "grad_norm": 0.7286702388508002, "learning_rate": 9.298440944306997e-06, "loss": 0.4753, "step": 5658 }, { "epoch": 0.391220186657449, "grad_norm": 0.7856886749420099, "learning_rate": 9.298154981633695e-06, "loss": 0.3387, "step": 5659 }, { "epoch": 0.391289319045973, "grad_norm": 0.797271192891857, "learning_rate": 9.297868965090692e-06, "loss": 0.3366, "step": 5660 }, { "epoch": 0.39135845143449705, "grad_norm": 0.7131541188644854, "learning_rate": 9.29758289468157e-06, "loss": 0.4498, "step": 5661 }, { "epoch": 0.39142758382302106, "grad_norm": 0.8606284293365956, "learning_rate": 9.297296770409917e-06, "loss": 0.4324, "step": 5662 }, { "epoch": 0.3914967162115451, "grad_norm": 0.6896701574477425, "learning_rate": 9.297010592279316e-06, "loss": 0.2834, "step": 5663 }, { "epoch": 0.39156584860006916, "grad_norm": 1.0782817090692867, "learning_rate": 9.296724360293358e-06, "loss": 0.3105, "step": 5664 }, { "epoch": 0.3916349809885932, "grad_norm": 1.146472733086988, "learning_rate": 9.296438074455626e-06, "loss": 0.4243, "step": 5665 }, { "epoch": 0.3917041133771172, "grad_norm": 0.7818652786696039, "learning_rate": 9.296151734769711e-06, "loss": 0.3354, "step": 5666 }, { "epoch": 0.3917732457656412, "grad_norm": 1.0037674471831164, "learning_rate": 9.295865341239201e-06, "loss": 0.3613, "step": 5667 }, { "epoch": 0.3918423781541652, "grad_norm": 0.7451909021061452, "learning_rate": 9.295578893867684e-06, "loss": 0.3557, "step": 5668 }, { "epoch": 0.39191151054268925, "grad_norm": 0.749811761887417, "learning_rate": 9.295292392658752e-06, "loss": 0.4564, "step": 5669 }, { "epoch": 0.39198064293121326, "grad_norm": 0.7008738090068146, "learning_rate": 9.295005837615996e-06, "loss": 0.3275, "step": 5670 }, { "epoch": 0.3920497753197373, "grad_norm": 0.9466045656105737, "learning_rate": 9.294719228743005e-06, "loss": 0.2817, "step": 5671 }, { "epoch": 0.3921189077082613, "grad_norm": 0.9012276605598167, "learning_rate": 9.294432566043375e-06, "loss": 0.446, "step": 5672 }, { "epoch": 0.3921880400967853, "grad_norm": 0.8160037969402255, "learning_rate": 9.294145849520694e-06, "loss": 0.3918, "step": 5673 }, { "epoch": 0.3922571724853094, "grad_norm": 0.6912150910729257, "learning_rate": 9.29385907917856e-06, "loss": 0.3252, "step": 5674 }, { "epoch": 0.3923263048738334, "grad_norm": 0.7937357832206106, "learning_rate": 9.293572255020564e-06, "loss": 0.3667, "step": 5675 }, { "epoch": 0.3923954372623574, "grad_norm": 0.7366020473479912, "learning_rate": 9.293285377050303e-06, "loss": 0.4216, "step": 5676 }, { "epoch": 0.39246456965088145, "grad_norm": 0.8957070033542112, "learning_rate": 9.29299844527137e-06, "loss": 0.3283, "step": 5677 }, { "epoch": 0.39253370203940546, "grad_norm": 0.9640935467779058, "learning_rate": 9.292711459687364e-06, "loss": 0.3538, "step": 5678 }, { "epoch": 0.3926028344279295, "grad_norm": 0.7200364179182446, "learning_rate": 9.29242442030188e-06, "loss": 0.4175, "step": 5679 }, { "epoch": 0.3926719668164535, "grad_norm": 0.7807278684904299, "learning_rate": 9.292137327118517e-06, "loss": 0.3344, "step": 5680 }, { "epoch": 0.3927410992049775, "grad_norm": 2.154128517054838, "learning_rate": 9.291850180140872e-06, "loss": 0.3005, "step": 5681 }, { "epoch": 0.39281023159350154, "grad_norm": 0.7945999985391663, "learning_rate": 9.291562979372542e-06, "loss": 0.3138, "step": 5682 }, { "epoch": 0.39287936398202555, "grad_norm": 0.8077020187965209, "learning_rate": 9.291275724817132e-06, "loss": 0.3884, "step": 5683 }, { "epoch": 0.3929484963705496, "grad_norm": 0.7050249811951277, "learning_rate": 9.290988416478236e-06, "loss": 0.3145, "step": 5684 }, { "epoch": 0.39301762875907365, "grad_norm": 0.7261765437294871, "learning_rate": 9.29070105435946e-06, "loss": 0.3166, "step": 5685 }, { "epoch": 0.39308676114759766, "grad_norm": 1.6871138186092212, "learning_rate": 9.290413638464402e-06, "loss": 0.3745, "step": 5686 }, { "epoch": 0.3931558935361217, "grad_norm": 1.0101834844898643, "learning_rate": 9.290126168796665e-06, "loss": 0.4348, "step": 5687 }, { "epoch": 0.3932250259246457, "grad_norm": 0.8837927327237333, "learning_rate": 9.289838645359855e-06, "loss": 0.3074, "step": 5688 }, { "epoch": 0.3932941583131697, "grad_norm": 0.8536677386104639, "learning_rate": 9.289551068157573e-06, "loss": 0.3145, "step": 5689 }, { "epoch": 0.39336329070169374, "grad_norm": 0.7514621518575553, "learning_rate": 9.289263437193421e-06, "loss": 0.4209, "step": 5690 }, { "epoch": 0.39343242309021775, "grad_norm": 0.7081201891440368, "learning_rate": 9.288975752471008e-06, "loss": 0.3414, "step": 5691 }, { "epoch": 0.39350155547874177, "grad_norm": 0.6800550722454943, "learning_rate": 9.288688013993937e-06, "loss": 0.3225, "step": 5692 }, { "epoch": 0.3935706878672658, "grad_norm": 0.8164485730463228, "learning_rate": 9.288400221765815e-06, "loss": 0.3771, "step": 5693 }, { "epoch": 0.39363982025578986, "grad_norm": 0.811913858722617, "learning_rate": 9.288112375790251e-06, "loss": 0.4441, "step": 5694 }, { "epoch": 0.3937089526443139, "grad_norm": 0.7600584962499611, "learning_rate": 9.28782447607085e-06, "loss": 0.3075, "step": 5695 }, { "epoch": 0.3937780850328379, "grad_norm": 0.7497575408162275, "learning_rate": 9.28753652261122e-06, "loss": 0.2823, "step": 5696 }, { "epoch": 0.3938472174213619, "grad_norm": 0.7215620949122237, "learning_rate": 9.287248515414972e-06, "loss": 0.4348, "step": 5697 }, { "epoch": 0.39391634980988594, "grad_norm": 0.7229426990069305, "learning_rate": 9.286960454485714e-06, "loss": 0.3883, "step": 5698 }, { "epoch": 0.39398548219840995, "grad_norm": 0.6773378054204687, "learning_rate": 9.28667233982706e-06, "loss": 0.2997, "step": 5699 }, { "epoch": 0.39405461458693397, "grad_norm": 0.8561351057149755, "learning_rate": 9.286384171442615e-06, "loss": 0.3137, "step": 5700 }, { "epoch": 0.394123746975458, "grad_norm": 0.7935395148489265, "learning_rate": 9.286095949335996e-06, "loss": 0.4482, "step": 5701 }, { "epoch": 0.394192879363982, "grad_norm": 0.8912573393662374, "learning_rate": 9.285807673510812e-06, "loss": 0.3818, "step": 5702 }, { "epoch": 0.394262011752506, "grad_norm": 0.8488969759053842, "learning_rate": 9.285519343970678e-06, "loss": 0.2748, "step": 5703 }, { "epoch": 0.3943311441410301, "grad_norm": 0.7462328436690806, "learning_rate": 9.285230960719205e-06, "loss": 0.3892, "step": 5704 }, { "epoch": 0.3944002765295541, "grad_norm": 0.8186774462441583, "learning_rate": 9.28494252376001e-06, "loss": 0.4106, "step": 5705 }, { "epoch": 0.39446940891807813, "grad_norm": 0.720723167748007, "learning_rate": 9.284654033096706e-06, "loss": 0.3533, "step": 5706 }, { "epoch": 0.39453854130660215, "grad_norm": 0.9461549371285717, "learning_rate": 9.284365488732914e-06, "loss": 0.3432, "step": 5707 }, { "epoch": 0.39460767369512617, "grad_norm": 0.8829721485791786, "learning_rate": 9.284076890672245e-06, "loss": 0.3878, "step": 5708 }, { "epoch": 0.3946768060836502, "grad_norm": 0.7292108762935947, "learning_rate": 9.283788238918316e-06, "loss": 0.3452, "step": 5709 }, { "epoch": 0.3947459384721742, "grad_norm": 4.055162564942597, "learning_rate": 9.283499533474748e-06, "loss": 0.289, "step": 5710 }, { "epoch": 0.3948150708606982, "grad_norm": 0.7525939542975159, "learning_rate": 9.283210774345156e-06, "loss": 0.3537, "step": 5711 }, { "epoch": 0.39488420324922224, "grad_norm": 0.7953599887204608, "learning_rate": 9.28292196153316e-06, "loss": 0.4095, "step": 5712 }, { "epoch": 0.39495333563774626, "grad_norm": 0.751938671370377, "learning_rate": 9.282633095042382e-06, "loss": 0.3552, "step": 5713 }, { "epoch": 0.39502246802627033, "grad_norm": 0.9551873783082179, "learning_rate": 9.282344174876438e-06, "loss": 0.344, "step": 5714 }, { "epoch": 0.39509160041479435, "grad_norm": 0.909360735965372, "learning_rate": 9.282055201038956e-06, "loss": 0.4385, "step": 5715 }, { "epoch": 0.39516073280331837, "grad_norm": 0.9428319622686024, "learning_rate": 9.281766173533553e-06, "loss": 0.3424, "step": 5716 }, { "epoch": 0.3952298651918424, "grad_norm": 0.7388980683580109, "learning_rate": 9.28147709236385e-06, "loss": 0.2875, "step": 5717 }, { "epoch": 0.3952989975803664, "grad_norm": 0.8038764166355122, "learning_rate": 9.281187957533474e-06, "loss": 0.3763, "step": 5718 }, { "epoch": 0.3953681299688904, "grad_norm": 0.730881965305495, "learning_rate": 9.280898769046045e-06, "loss": 0.4145, "step": 5719 }, { "epoch": 0.39543726235741444, "grad_norm": 0.8044415552140532, "learning_rate": 9.280609526905192e-06, "loss": 0.268, "step": 5720 }, { "epoch": 0.39550639474593846, "grad_norm": 0.6705812698045064, "learning_rate": 9.280320231114537e-06, "loss": 0.2998, "step": 5721 }, { "epoch": 0.3955755271344625, "grad_norm": 0.776657565302007, "learning_rate": 9.280030881677706e-06, "loss": 0.4768, "step": 5722 }, { "epoch": 0.3956446595229865, "grad_norm": 1.0663617897928388, "learning_rate": 9.279741478598326e-06, "loss": 0.4169, "step": 5723 }, { "epoch": 0.3957137919115105, "grad_norm": 0.7626512940540631, "learning_rate": 9.279452021880025e-06, "loss": 0.3209, "step": 5724 }, { "epoch": 0.3957829243000346, "grad_norm": 0.7600015733504007, "learning_rate": 9.279162511526428e-06, "loss": 0.3417, "step": 5725 }, { "epoch": 0.3958520566885586, "grad_norm": 0.7544092840186709, "learning_rate": 9.278872947541167e-06, "loss": 0.3829, "step": 5726 }, { "epoch": 0.3959211890770826, "grad_norm": 0.8508473377375662, "learning_rate": 9.278583329927869e-06, "loss": 0.3077, "step": 5727 }, { "epoch": 0.39599032146560664, "grad_norm": 0.7500274932627755, "learning_rate": 9.278293658690165e-06, "loss": 0.2841, "step": 5728 }, { "epoch": 0.39605945385413066, "grad_norm": 0.7851152917312857, "learning_rate": 9.278003933831684e-06, "loss": 0.4207, "step": 5729 }, { "epoch": 0.3961285862426547, "grad_norm": 0.8773532077100747, "learning_rate": 9.277714155356059e-06, "loss": 0.3546, "step": 5730 }, { "epoch": 0.3961977186311787, "grad_norm": 0.8562016143811715, "learning_rate": 9.277424323266922e-06, "loss": 0.3513, "step": 5731 }, { "epoch": 0.3962668510197027, "grad_norm": 0.8253309765658945, "learning_rate": 9.277134437567903e-06, "loss": 0.3086, "step": 5732 }, { "epoch": 0.39633598340822673, "grad_norm": 0.8188473502977557, "learning_rate": 9.276844498262636e-06, "loss": 0.3791, "step": 5733 }, { "epoch": 0.39640511579675075, "grad_norm": 0.6904660757671104, "learning_rate": 9.276554505354757e-06, "loss": 0.3336, "step": 5734 }, { "epoch": 0.3964742481852748, "grad_norm": 0.738420053708746, "learning_rate": 9.2762644588479e-06, "loss": 0.3566, "step": 5735 }, { "epoch": 0.39654338057379884, "grad_norm": 0.8048729019891641, "learning_rate": 9.275974358745697e-06, "loss": 0.4073, "step": 5736 }, { "epoch": 0.39661251296232286, "grad_norm": 1.170627947768466, "learning_rate": 9.275684205051787e-06, "loss": 0.401, "step": 5737 }, { "epoch": 0.3966816453508469, "grad_norm": 0.6868345083970933, "learning_rate": 9.275393997769806e-06, "loss": 0.3084, "step": 5738 }, { "epoch": 0.3967507777393709, "grad_norm": 0.8383602073858524, "learning_rate": 9.27510373690339e-06, "loss": 0.3682, "step": 5739 }, { "epoch": 0.3968199101278949, "grad_norm": 0.778160714091385, "learning_rate": 9.27481342245618e-06, "loss": 0.4311, "step": 5740 }, { "epoch": 0.39688904251641893, "grad_norm": 0.7418550805992923, "learning_rate": 9.274523054431811e-06, "loss": 0.3259, "step": 5741 }, { "epoch": 0.39695817490494295, "grad_norm": 0.6734426470096366, "learning_rate": 9.274232632833923e-06, "loss": 0.319, "step": 5742 }, { "epoch": 0.39702730729346697, "grad_norm": 0.8023277433701747, "learning_rate": 9.27394215766616e-06, "loss": 0.3929, "step": 5743 }, { "epoch": 0.397096439681991, "grad_norm": 0.6732175268001224, "learning_rate": 9.273651628932156e-06, "loss": 0.4003, "step": 5744 }, { "epoch": 0.39716557207051506, "grad_norm": 0.7692184503024039, "learning_rate": 9.273361046635557e-06, "loss": 0.3062, "step": 5745 }, { "epoch": 0.3972347044590391, "grad_norm": 0.735572076305346, "learning_rate": 9.273070410780002e-06, "loss": 0.2826, "step": 5746 }, { "epoch": 0.3973038368475631, "grad_norm": 0.7264559761253871, "learning_rate": 9.272779721369138e-06, "loss": 0.4386, "step": 5747 }, { "epoch": 0.3973729692360871, "grad_norm": 0.7509461870202998, "learning_rate": 9.272488978406602e-06, "loss": 0.3858, "step": 5748 }, { "epoch": 0.39744210162461113, "grad_norm": 0.714263110858345, "learning_rate": 9.272198181896043e-06, "loss": 0.321, "step": 5749 }, { "epoch": 0.39751123401313515, "grad_norm": 0.7643605581644427, "learning_rate": 9.271907331841104e-06, "loss": 0.3916, "step": 5750 }, { "epoch": 0.39758036640165917, "grad_norm": 0.7263980617032544, "learning_rate": 9.27161642824543e-06, "loss": 0.4483, "step": 5751 }, { "epoch": 0.3976494987901832, "grad_norm": 0.7978824573786498, "learning_rate": 9.271325471112667e-06, "loss": 0.312, "step": 5752 }, { "epoch": 0.3977186311787072, "grad_norm": 0.6280166280016684, "learning_rate": 9.27103446044646e-06, "loss": 0.3455, "step": 5753 }, { "epoch": 0.3977877635672312, "grad_norm": 0.7416940697968267, "learning_rate": 9.270743396250461e-06, "loss": 0.4369, "step": 5754 }, { "epoch": 0.3978568959557553, "grad_norm": 0.686346901670936, "learning_rate": 9.270452278528312e-06, "loss": 0.381, "step": 5755 }, { "epoch": 0.3979260283442793, "grad_norm": 0.7387297873796979, "learning_rate": 9.270161107283668e-06, "loss": 0.3225, "step": 5756 }, { "epoch": 0.39799516073280333, "grad_norm": 0.810254367225157, "learning_rate": 9.269869882520173e-06, "loss": 0.375, "step": 5757 }, { "epoch": 0.39806429312132735, "grad_norm": 0.8053964410441543, "learning_rate": 9.269578604241478e-06, "loss": 0.4588, "step": 5758 }, { "epoch": 0.39813342550985137, "grad_norm": 0.7967100900048755, "learning_rate": 9.269287272451238e-06, "loss": 0.3316, "step": 5759 }, { "epoch": 0.3982025578983754, "grad_norm": 1.0216257506833857, "learning_rate": 9.268995887153097e-06, "loss": 0.3161, "step": 5760 }, { "epoch": 0.3982716902868994, "grad_norm": 0.8005108699594793, "learning_rate": 9.268704448350713e-06, "loss": 0.4249, "step": 5761 }, { "epoch": 0.3983408226754234, "grad_norm": 0.8095019563139728, "learning_rate": 9.268412956047736e-06, "loss": 0.4314, "step": 5762 }, { "epoch": 0.39840995506394744, "grad_norm": 0.739728131751306, "learning_rate": 9.26812141024782e-06, "loss": 0.3324, "step": 5763 }, { "epoch": 0.39847908745247146, "grad_norm": 0.7812744062878475, "learning_rate": 9.26782981095462e-06, "loss": 0.3007, "step": 5764 }, { "epoch": 0.39854821984099553, "grad_norm": 0.851864145258182, "learning_rate": 9.267538158171788e-06, "loss": 0.4527, "step": 5765 }, { "epoch": 0.39861735222951955, "grad_norm": 0.7403043844887074, "learning_rate": 9.267246451902981e-06, "loss": 0.3603, "step": 5766 }, { "epoch": 0.39868648461804357, "grad_norm": 0.7348631099330103, "learning_rate": 9.266954692151857e-06, "loss": 0.36, "step": 5767 }, { "epoch": 0.3987556170065676, "grad_norm": 0.8752205369982152, "learning_rate": 9.266662878922067e-06, "loss": 0.3348, "step": 5768 }, { "epoch": 0.3988247493950916, "grad_norm": 0.8058235477018942, "learning_rate": 9.266371012217275e-06, "loss": 0.4329, "step": 5769 }, { "epoch": 0.3988938817836156, "grad_norm": 0.6666798247611122, "learning_rate": 9.266079092041136e-06, "loss": 0.3108, "step": 5770 }, { "epoch": 0.39896301417213964, "grad_norm": 0.8714096619592161, "learning_rate": 9.265787118397306e-06, "loss": 0.3473, "step": 5771 }, { "epoch": 0.39903214656066366, "grad_norm": 0.7729227540019615, "learning_rate": 9.265495091289451e-06, "loss": 0.4072, "step": 5772 }, { "epoch": 0.3991012789491877, "grad_norm": 0.7300362857048316, "learning_rate": 9.265203010721225e-06, "loss": 0.3587, "step": 5773 }, { "epoch": 0.3991704113377117, "grad_norm": 0.8172334013557433, "learning_rate": 9.26491087669629e-06, "loss": 0.301, "step": 5774 }, { "epoch": 0.39923954372623577, "grad_norm": 0.8082005594598805, "learning_rate": 9.26461868921831e-06, "loss": 0.3793, "step": 5775 }, { "epoch": 0.3993086761147598, "grad_norm": 0.7003957497736044, "learning_rate": 9.264326448290945e-06, "loss": 0.4354, "step": 5776 }, { "epoch": 0.3993778085032838, "grad_norm": 1.1906947727064099, "learning_rate": 9.264034153917858e-06, "loss": 0.3487, "step": 5777 }, { "epoch": 0.3994469408918078, "grad_norm": 0.6823447505116776, "learning_rate": 9.263741806102712e-06, "loss": 0.2731, "step": 5778 }, { "epoch": 0.39951607328033184, "grad_norm": 1.0145236878914132, "learning_rate": 9.263449404849171e-06, "loss": 0.4313, "step": 5779 }, { "epoch": 0.39958520566885586, "grad_norm": 0.9236455935241505, "learning_rate": 9.2631569501609e-06, "loss": 0.407, "step": 5780 }, { "epoch": 0.3996543380573799, "grad_norm": 0.7056221664918071, "learning_rate": 9.262864442041566e-06, "loss": 0.3571, "step": 5781 }, { "epoch": 0.3997234704459039, "grad_norm": 0.7493069816323925, "learning_rate": 9.262571880494834e-06, "loss": 0.4005, "step": 5782 }, { "epoch": 0.3997926028344279, "grad_norm": 0.7669271391067947, "learning_rate": 9.262279265524368e-06, "loss": 0.4642, "step": 5783 }, { "epoch": 0.39986173522295193, "grad_norm": 0.8444600260594454, "learning_rate": 9.26198659713384e-06, "loss": 0.3331, "step": 5784 }, { "epoch": 0.399930867611476, "grad_norm": 0.7746757945913868, "learning_rate": 9.261693875326915e-06, "loss": 0.3092, "step": 5785 }, { "epoch": 0.4, "grad_norm": 1.036472967422782, "learning_rate": 9.261401100107262e-06, "loss": 0.34, "step": 5786 }, { "epoch": 0.40006913238852404, "grad_norm": 0.7162611422060466, "learning_rate": 9.261108271478551e-06, "loss": 0.4067, "step": 5787 }, { "epoch": 0.40013826477704806, "grad_norm": 0.7970299701940126, "learning_rate": 9.260815389444454e-06, "loss": 0.3347, "step": 5788 }, { "epoch": 0.4002073971655721, "grad_norm": 0.749630900164082, "learning_rate": 9.260522454008637e-06, "loss": 0.3266, "step": 5789 }, { "epoch": 0.4002765295540961, "grad_norm": 0.8047564164800186, "learning_rate": 9.260229465174776e-06, "loss": 0.5051, "step": 5790 }, { "epoch": 0.4003456619426201, "grad_norm": 0.9104483882479429, "learning_rate": 9.259936422946543e-06, "loss": 0.3804, "step": 5791 }, { "epoch": 0.40041479433114413, "grad_norm": 0.7013466053682431, "learning_rate": 9.259643327327606e-06, "loss": 0.2737, "step": 5792 }, { "epoch": 0.40048392671966815, "grad_norm": 0.7132162913471415, "learning_rate": 9.259350178321643e-06, "loss": 0.3167, "step": 5793 }, { "epoch": 0.40055305910819217, "grad_norm": 0.7857078198927056, "learning_rate": 9.259056975932326e-06, "loss": 0.4215, "step": 5794 }, { "epoch": 0.4006221914967162, "grad_norm": 0.9075900262507887, "learning_rate": 9.258763720163332e-06, "loss": 0.3079, "step": 5795 }, { "epoch": 0.40069132388524026, "grad_norm": 0.7492023969411328, "learning_rate": 9.258470411018334e-06, "loss": 0.3213, "step": 5796 }, { "epoch": 0.4007604562737643, "grad_norm": 0.7448123387864498, "learning_rate": 9.25817704850101e-06, "loss": 0.4367, "step": 5797 }, { "epoch": 0.4008295886622883, "grad_norm": 0.6795740746105283, "learning_rate": 9.257883632615032e-06, "loss": 0.3478, "step": 5798 }, { "epoch": 0.4008987210508123, "grad_norm": 0.8497483600021978, "learning_rate": 9.257590163364085e-06, "loss": 0.2966, "step": 5799 }, { "epoch": 0.40096785343933633, "grad_norm": 0.7852892834426792, "learning_rate": 9.257296640751843e-06, "loss": 0.3615, "step": 5800 }, { "epoch": 0.40103698582786035, "grad_norm": 0.8807469236414917, "learning_rate": 9.257003064781982e-06, "loss": 0.3519, "step": 5801 }, { "epoch": 0.40110611821638437, "grad_norm": 1.0104660450930087, "learning_rate": 9.256709435458188e-06, "loss": 0.2424, "step": 5802 }, { "epoch": 0.4011752506049084, "grad_norm": 0.7477929284447523, "learning_rate": 9.256415752784136e-06, "loss": 0.3131, "step": 5803 }, { "epoch": 0.4012443829934324, "grad_norm": 0.8987757419188854, "learning_rate": 9.256122016763508e-06, "loss": 0.3709, "step": 5804 }, { "epoch": 0.4013135153819564, "grad_norm": 0.8283471324145874, "learning_rate": 9.255828227399988e-06, "loss": 0.4095, "step": 5805 }, { "epoch": 0.4013826477704805, "grad_norm": 0.7445985184494582, "learning_rate": 9.255534384697256e-06, "loss": 0.2612, "step": 5806 }, { "epoch": 0.4014517801590045, "grad_norm": 0.8879983015241721, "learning_rate": 9.255240488658993e-06, "loss": 0.2818, "step": 5807 }, { "epoch": 0.40152091254752853, "grad_norm": 0.8686708095809579, "learning_rate": 9.254946539288885e-06, "loss": 0.4273, "step": 5808 }, { "epoch": 0.40159004493605255, "grad_norm": 0.9356432261338746, "learning_rate": 9.254652536590615e-06, "loss": 0.3494, "step": 5809 }, { "epoch": 0.40165917732457657, "grad_norm": 0.777735003751609, "learning_rate": 9.254358480567869e-06, "loss": 0.3299, "step": 5810 }, { "epoch": 0.4017283097131006, "grad_norm": 0.8016540736603031, "learning_rate": 9.254064371224332e-06, "loss": 0.3399, "step": 5811 }, { "epoch": 0.4017974421016246, "grad_norm": 0.8016300886174291, "learning_rate": 9.253770208563688e-06, "loss": 0.4028, "step": 5812 }, { "epoch": 0.4018665744901486, "grad_norm": 0.6861961794785674, "learning_rate": 9.253475992589627e-06, "loss": 0.3242, "step": 5813 }, { "epoch": 0.40193570687867264, "grad_norm": 1.205694400503189, "learning_rate": 9.253181723305834e-06, "loss": 0.366, "step": 5814 }, { "epoch": 0.40200483926719666, "grad_norm": 0.8224051771535963, "learning_rate": 9.252887400716e-06, "loss": 0.372, "step": 5815 }, { "epoch": 0.40207397165572073, "grad_norm": 0.7767011131123505, "learning_rate": 9.252593024823811e-06, "loss": 0.3168, "step": 5816 }, { "epoch": 0.40214310404424475, "grad_norm": 0.8437265756237393, "learning_rate": 9.252298595632958e-06, "loss": 0.3001, "step": 5817 }, { "epoch": 0.40221223643276877, "grad_norm": 0.7967106789232642, "learning_rate": 9.25200411314713e-06, "loss": 0.4272, "step": 5818 }, { "epoch": 0.4022813688212928, "grad_norm": 0.7812567193243234, "learning_rate": 9.251709577370018e-06, "loss": 0.4175, "step": 5819 }, { "epoch": 0.4023505012098168, "grad_norm": 0.847885457097568, "learning_rate": 9.251414988305315e-06, "loss": 0.3297, "step": 5820 }, { "epoch": 0.4024196335983408, "grad_norm": 0.7328891152101211, "learning_rate": 9.251120345956713e-06, "loss": 0.2872, "step": 5821 }, { "epoch": 0.40248876598686484, "grad_norm": 0.9092324242306484, "learning_rate": 9.250825650327902e-06, "loss": 0.3553, "step": 5822 }, { "epoch": 0.40255789837538886, "grad_norm": 0.826021954973845, "learning_rate": 9.250530901422579e-06, "loss": 0.3806, "step": 5823 }, { "epoch": 0.4026270307639129, "grad_norm": 0.7959611270087659, "learning_rate": 9.250236099244435e-06, "loss": 0.3035, "step": 5824 }, { "epoch": 0.4026961631524369, "grad_norm": 0.7981796755076075, "learning_rate": 9.249941243797166e-06, "loss": 0.3174, "step": 5825 }, { "epoch": 0.40276529554096097, "grad_norm": 0.8221216293971804, "learning_rate": 9.24964633508447e-06, "loss": 0.4164, "step": 5826 }, { "epoch": 0.402834427929485, "grad_norm": 0.7603356577544589, "learning_rate": 9.24935137311004e-06, "loss": 0.2835, "step": 5827 }, { "epoch": 0.402903560318009, "grad_norm": 0.8115056851821646, "learning_rate": 9.249056357877574e-06, "loss": 0.2741, "step": 5828 }, { "epoch": 0.402972692706533, "grad_norm": 0.9221609352870214, "learning_rate": 9.248761289390768e-06, "loss": 0.3857, "step": 5829 }, { "epoch": 0.40304182509505704, "grad_norm": 0.9101784022211696, "learning_rate": 9.248466167653324e-06, "loss": 0.3583, "step": 5830 }, { "epoch": 0.40311095748358106, "grad_norm": 0.7930319761385244, "learning_rate": 9.248170992668936e-06, "loss": 0.3247, "step": 5831 }, { "epoch": 0.4031800898721051, "grad_norm": 0.8366573843310235, "learning_rate": 9.247875764441307e-06, "loss": 0.3134, "step": 5832 }, { "epoch": 0.4032492222606291, "grad_norm": 0.7638136685277351, "learning_rate": 9.247580482974137e-06, "loss": 0.3984, "step": 5833 }, { "epoch": 0.4033183546491531, "grad_norm": 0.7644409502300975, "learning_rate": 9.247285148271125e-06, "loss": 0.3283, "step": 5834 }, { "epoch": 0.40338748703767713, "grad_norm": 0.8399864503699008, "learning_rate": 9.246989760335972e-06, "loss": 0.3089, "step": 5835 }, { "epoch": 0.4034566194262012, "grad_norm": 0.8258015215168077, "learning_rate": 9.246694319172382e-06, "loss": 0.3874, "step": 5836 }, { "epoch": 0.4035257518147252, "grad_norm": 0.8226454069445841, "learning_rate": 9.246398824784058e-06, "loss": 0.4441, "step": 5837 }, { "epoch": 0.40359488420324924, "grad_norm": 0.6561901503276837, "learning_rate": 9.246103277174702e-06, "loss": 0.28, "step": 5838 }, { "epoch": 0.40366401659177326, "grad_norm": 0.7776405007830516, "learning_rate": 9.24580767634802e-06, "loss": 0.3345, "step": 5839 }, { "epoch": 0.4037331489802973, "grad_norm": 0.8680068569762469, "learning_rate": 9.245512022307715e-06, "loss": 0.4666, "step": 5840 }, { "epoch": 0.4038022813688213, "grad_norm": 0.7347430938449294, "learning_rate": 9.245216315057494e-06, "loss": 0.4112, "step": 5841 }, { "epoch": 0.4038714137573453, "grad_norm": 0.7246909098268464, "learning_rate": 9.244920554601061e-06, "loss": 0.2957, "step": 5842 }, { "epoch": 0.40394054614586933, "grad_norm": 1.0184679794531493, "learning_rate": 9.244624740942125e-06, "loss": 0.3379, "step": 5843 }, { "epoch": 0.40400967853439335, "grad_norm": 0.7107707501387259, "learning_rate": 9.244328874084393e-06, "loss": 0.4071, "step": 5844 }, { "epoch": 0.40407881092291736, "grad_norm": 0.6681282573311131, "learning_rate": 9.244032954031571e-06, "loss": 0.3075, "step": 5845 }, { "epoch": 0.40414794331144144, "grad_norm": 0.7315257644981619, "learning_rate": 9.243736980787374e-06, "loss": 0.3449, "step": 5846 }, { "epoch": 0.40421707569996546, "grad_norm": 0.7998304668550075, "learning_rate": 9.243440954355504e-06, "loss": 0.4211, "step": 5847 }, { "epoch": 0.4042862080884895, "grad_norm": 0.9508404821573978, "learning_rate": 9.243144874739676e-06, "loss": 0.3233, "step": 5848 }, { "epoch": 0.4043553404770135, "grad_norm": 0.8049738505584754, "learning_rate": 9.242848741943598e-06, "loss": 0.3414, "step": 5849 }, { "epoch": 0.4044244728655375, "grad_norm": 0.7616478337606721, "learning_rate": 9.242552555970982e-06, "loss": 0.3784, "step": 5850 }, { "epoch": 0.40449360525406153, "grad_norm": 0.760067138985286, "learning_rate": 9.242256316825541e-06, "loss": 0.3862, "step": 5851 }, { "epoch": 0.40456273764258555, "grad_norm": 0.8453069080314127, "learning_rate": 9.24196002451099e-06, "loss": 0.2701, "step": 5852 }, { "epoch": 0.40463187003110956, "grad_norm": 0.7704631567365703, "learning_rate": 9.241663679031039e-06, "loss": 0.2688, "step": 5853 }, { "epoch": 0.4047010024196336, "grad_norm": 0.8564896859918053, "learning_rate": 9.241367280389402e-06, "loss": 0.3996, "step": 5854 }, { "epoch": 0.4047701348081576, "grad_norm": 0.6700386969463986, "learning_rate": 9.241070828589797e-06, "loss": 0.3725, "step": 5855 }, { "epoch": 0.4048392671966816, "grad_norm": 0.8147556499934033, "learning_rate": 9.240774323635938e-06, "loss": 0.2888, "step": 5856 }, { "epoch": 0.4049083995852057, "grad_norm": 0.7836374794465598, "learning_rate": 9.24047776553154e-06, "loss": 0.3316, "step": 5857 }, { "epoch": 0.4049775319737297, "grad_norm": 0.9091585095183338, "learning_rate": 9.240181154280319e-06, "loss": 0.4842, "step": 5858 }, { "epoch": 0.4050466643622537, "grad_norm": 0.7269962407950995, "learning_rate": 9.239884489885997e-06, "loss": 0.345, "step": 5859 }, { "epoch": 0.40511579675077775, "grad_norm": 0.7073707752803938, "learning_rate": 9.239587772352288e-06, "loss": 0.3028, "step": 5860 }, { "epoch": 0.40518492913930176, "grad_norm": 0.7520822114032841, "learning_rate": 9.23929100168291e-06, "loss": 0.3745, "step": 5861 }, { "epoch": 0.4052540615278258, "grad_norm": 0.721003660467907, "learning_rate": 9.238994177881589e-06, "loss": 0.3572, "step": 5862 }, { "epoch": 0.4053231939163498, "grad_norm": 0.7128911199812495, "learning_rate": 9.238697300952038e-06, "loss": 0.3361, "step": 5863 }, { "epoch": 0.4053923263048738, "grad_norm": 0.711074897590764, "learning_rate": 9.238400370897982e-06, "loss": 0.3284, "step": 5864 }, { "epoch": 0.40546145869339784, "grad_norm": 0.8152246624402241, "learning_rate": 9.23810338772314e-06, "loss": 0.4097, "step": 5865 }, { "epoch": 0.40553059108192185, "grad_norm": 0.775418679161734, "learning_rate": 9.237806351431235e-06, "loss": 0.394, "step": 5866 }, { "epoch": 0.4055997234704459, "grad_norm": 0.7686621420190143, "learning_rate": 9.237509262025992e-06, "loss": 0.3165, "step": 5867 }, { "epoch": 0.40566885585896995, "grad_norm": 0.7841333548988759, "learning_rate": 9.237212119511133e-06, "loss": 0.386, "step": 5868 }, { "epoch": 0.40573798824749396, "grad_norm": 0.8803369716540396, "learning_rate": 9.236914923890379e-06, "loss": 0.4751, "step": 5869 }, { "epoch": 0.405807120636018, "grad_norm": 0.7148941124862501, "learning_rate": 9.236617675167459e-06, "loss": 0.3133, "step": 5870 }, { "epoch": 0.405876253024542, "grad_norm": 0.7852322574137554, "learning_rate": 9.236320373346095e-06, "loss": 0.3292, "step": 5871 }, { "epoch": 0.405945385413066, "grad_norm": 0.7948854063057113, "learning_rate": 9.236023018430018e-06, "loss": 0.4217, "step": 5872 }, { "epoch": 0.40601451780159004, "grad_norm": 0.7275715195475604, "learning_rate": 9.23572561042295e-06, "loss": 0.3491, "step": 5873 }, { "epoch": 0.40608365019011405, "grad_norm": 0.800776153888425, "learning_rate": 9.235428149328622e-06, "loss": 0.3351, "step": 5874 }, { "epoch": 0.40615278257863807, "grad_norm": 0.8265417945754902, "learning_rate": 9.23513063515076e-06, "loss": 0.3649, "step": 5875 }, { "epoch": 0.4062219149671621, "grad_norm": 0.6751500603513053, "learning_rate": 9.234833067893093e-06, "loss": 0.3812, "step": 5876 }, { "epoch": 0.40629104735568616, "grad_norm": 0.7382928080537494, "learning_rate": 9.234535447559352e-06, "loss": 0.2865, "step": 5877 }, { "epoch": 0.4063601797442102, "grad_norm": 0.7481680453389169, "learning_rate": 9.234237774153265e-06, "loss": 0.3145, "step": 5878 }, { "epoch": 0.4064293121327342, "grad_norm": 0.7491981747053487, "learning_rate": 9.233940047678564e-06, "loss": 0.3305, "step": 5879 }, { "epoch": 0.4064984445212582, "grad_norm": 0.741562433421216, "learning_rate": 9.23364226813898e-06, "loss": 0.3562, "step": 5880 }, { "epoch": 0.40656757690978224, "grad_norm": 0.8209337371067723, "learning_rate": 9.233344435538247e-06, "loss": 0.3538, "step": 5881 }, { "epoch": 0.40663670929830625, "grad_norm": 0.7027385573108703, "learning_rate": 9.233046549880093e-06, "loss": 0.3189, "step": 5882 }, { "epoch": 0.40670584168683027, "grad_norm": 0.7012186506538348, "learning_rate": 9.232748611168258e-06, "loss": 0.3501, "step": 5883 }, { "epoch": 0.4067749740753543, "grad_norm": 0.7798821076649116, "learning_rate": 9.232450619406471e-06, "loss": 0.2955, "step": 5884 }, { "epoch": 0.4068441064638783, "grad_norm": 0.7610897311796663, "learning_rate": 9.232152574598469e-06, "loss": 0.2863, "step": 5885 }, { "epoch": 0.4069132388524023, "grad_norm": 0.7607671284687785, "learning_rate": 9.231854476747988e-06, "loss": 0.353, "step": 5886 }, { "epoch": 0.4069823712409264, "grad_norm": 0.6935323605011007, "learning_rate": 9.231556325858763e-06, "loss": 0.4111, "step": 5887 }, { "epoch": 0.4070515036294504, "grad_norm": 0.679090848943119, "learning_rate": 9.23125812193453e-06, "loss": 0.2914, "step": 5888 }, { "epoch": 0.40712063601797444, "grad_norm": 0.8424670285311471, "learning_rate": 9.230959864979029e-06, "loss": 0.3522, "step": 5889 }, { "epoch": 0.40718976840649845, "grad_norm": 0.8189997436785734, "learning_rate": 9.230661554995995e-06, "loss": 0.4759, "step": 5890 }, { "epoch": 0.40725890079502247, "grad_norm": 0.8114250883926316, "learning_rate": 9.230363191989168e-06, "loss": 0.3055, "step": 5891 }, { "epoch": 0.4073280331835465, "grad_norm": 0.7349438348637949, "learning_rate": 9.23006477596229e-06, "loss": 0.2898, "step": 5892 }, { "epoch": 0.4073971655720705, "grad_norm": 0.741142122898886, "learning_rate": 9.229766306919098e-06, "loss": 0.3383, "step": 5893 }, { "epoch": 0.4074662979605945, "grad_norm": 0.7981689004896684, "learning_rate": 9.229467784863333e-06, "loss": 0.3793, "step": 5894 }, { "epoch": 0.40753543034911854, "grad_norm": 0.6942724955544741, "learning_rate": 9.229169209798738e-06, "loss": 0.294, "step": 5895 }, { "epoch": 0.40760456273764256, "grad_norm": 1.01120631784513, "learning_rate": 9.228870581729054e-06, "loss": 0.324, "step": 5896 }, { "epoch": 0.40767369512616664, "grad_norm": 0.8731934459252213, "learning_rate": 9.228571900658023e-06, "loss": 0.3959, "step": 5897 }, { "epoch": 0.40774282751469065, "grad_norm": 0.7021956873817896, "learning_rate": 9.22827316658939e-06, "loss": 0.3466, "step": 5898 }, { "epoch": 0.40781195990321467, "grad_norm": 0.6368232511560032, "learning_rate": 9.2279743795269e-06, "loss": 0.3147, "step": 5899 }, { "epoch": 0.4078810922917387, "grad_norm": 0.7016552937886236, "learning_rate": 9.227675539474294e-06, "loss": 0.402, "step": 5900 }, { "epoch": 0.4079502246802627, "grad_norm": 1.765226555236555, "learning_rate": 9.227376646435322e-06, "loss": 0.4395, "step": 5901 }, { "epoch": 0.4080193570687867, "grad_norm": 0.8047124257715865, "learning_rate": 9.227077700413727e-06, "loss": 0.3389, "step": 5902 }, { "epoch": 0.40808848945731074, "grad_norm": 0.6935563055356966, "learning_rate": 9.226778701413256e-06, "loss": 0.2557, "step": 5903 }, { "epoch": 0.40815762184583476, "grad_norm": 0.7758946585940284, "learning_rate": 9.226479649437657e-06, "loss": 0.3901, "step": 5904 }, { "epoch": 0.4082267542343588, "grad_norm": 0.7734646987114505, "learning_rate": 9.226180544490678e-06, "loss": 0.4735, "step": 5905 }, { "epoch": 0.4082958866228828, "grad_norm": 0.7115216320164838, "learning_rate": 9.225881386576068e-06, "loss": 0.2919, "step": 5906 }, { "epoch": 0.40836501901140687, "grad_norm": 1.4094035725373038, "learning_rate": 9.225582175697577e-06, "loss": 0.3352, "step": 5907 }, { "epoch": 0.4084341513999309, "grad_norm": 1.1396666998864409, "learning_rate": 9.225282911858954e-06, "loss": 0.3936, "step": 5908 }, { "epoch": 0.4085032837884549, "grad_norm": 0.7830347508958286, "learning_rate": 9.22498359506395e-06, "loss": 0.3406, "step": 5909 }, { "epoch": 0.4085724161769789, "grad_norm": 0.671591646718202, "learning_rate": 9.224684225316317e-06, "loss": 0.3241, "step": 5910 }, { "epoch": 0.40864154856550294, "grad_norm": 0.8271579371327962, "learning_rate": 9.224384802619805e-06, "loss": 0.4187, "step": 5911 }, { "epoch": 0.40871068095402696, "grad_norm": 0.7746060663210413, "learning_rate": 9.22408532697817e-06, "loss": 0.4509, "step": 5912 }, { "epoch": 0.408779813342551, "grad_norm": 0.7593933664026917, "learning_rate": 9.223785798395162e-06, "loss": 0.3407, "step": 5913 }, { "epoch": 0.408848945731075, "grad_norm": 0.8358613759738631, "learning_rate": 9.223486216874537e-06, "loss": 0.314, "step": 5914 }, { "epoch": 0.408918078119599, "grad_norm": 0.7576454565412083, "learning_rate": 9.22318658242005e-06, "loss": 0.4756, "step": 5915 }, { "epoch": 0.40898721050812303, "grad_norm": 0.7634494931575799, "learning_rate": 9.222886895035456e-06, "loss": 0.3495, "step": 5916 }, { "epoch": 0.40905634289664705, "grad_norm": 0.838726507108799, "learning_rate": 9.22258715472451e-06, "loss": 0.3826, "step": 5917 }, { "epoch": 0.4091254752851711, "grad_norm": 0.8185000217314439, "learning_rate": 9.22228736149097e-06, "loss": 0.3637, "step": 5918 }, { "epoch": 0.40919460767369514, "grad_norm": 0.7027133376549116, "learning_rate": 9.221987515338593e-06, "loss": 0.3935, "step": 5919 }, { "epoch": 0.40926374006221916, "grad_norm": 0.7055151717825422, "learning_rate": 9.221687616271138e-06, "loss": 0.3459, "step": 5920 }, { "epoch": 0.4093328724507432, "grad_norm": 0.7653022271647727, "learning_rate": 9.22138766429236e-06, "loss": 0.2878, "step": 5921 }, { "epoch": 0.4094020048392672, "grad_norm": 0.776717611760686, "learning_rate": 9.221087659406023e-06, "loss": 0.384, "step": 5922 }, { "epoch": 0.4094711372277912, "grad_norm": 0.7570849760655687, "learning_rate": 9.220787601615885e-06, "loss": 0.3621, "step": 5923 }, { "epoch": 0.40954026961631523, "grad_norm": 0.7177613130227402, "learning_rate": 9.220487490925707e-06, "loss": 0.3265, "step": 5924 }, { "epoch": 0.40960940200483925, "grad_norm": 0.888968252437308, "learning_rate": 9.22018732733925e-06, "loss": 0.3493, "step": 5925 }, { "epoch": 0.40967853439336327, "grad_norm": 0.8069064149686782, "learning_rate": 9.219887110860274e-06, "loss": 0.4118, "step": 5926 }, { "epoch": 0.4097476667818873, "grad_norm": 0.7299725732187953, "learning_rate": 9.219586841492548e-06, "loss": 0.3202, "step": 5927 }, { "epoch": 0.40981679917041136, "grad_norm": 0.7990065487870913, "learning_rate": 9.219286519239828e-06, "loss": 0.3194, "step": 5928 }, { "epoch": 0.4098859315589354, "grad_norm": 0.7363135332383467, "learning_rate": 9.21898614410588e-06, "loss": 0.4271, "step": 5929 }, { "epoch": 0.4099550639474594, "grad_norm": 1.73420523578672, "learning_rate": 9.218685716094474e-06, "loss": 0.4817, "step": 5930 }, { "epoch": 0.4100241963359834, "grad_norm": 0.8787450793670899, "learning_rate": 9.218385235209368e-06, "loss": 0.3369, "step": 5931 }, { "epoch": 0.41009332872450743, "grad_norm": 0.7765949099248929, "learning_rate": 9.21808470145433e-06, "loss": 0.3027, "step": 5932 }, { "epoch": 0.41016246111303145, "grad_norm": 0.7573085845032685, "learning_rate": 9.21778411483313e-06, "loss": 0.4404, "step": 5933 }, { "epoch": 0.41023159350155547, "grad_norm": 0.6946393706941737, "learning_rate": 9.217483475349533e-06, "loss": 0.3675, "step": 5934 }, { "epoch": 0.4103007258900795, "grad_norm": 0.797066696789385, "learning_rate": 9.217182783007309e-06, "loss": 0.2932, "step": 5935 }, { "epoch": 0.4103698582786035, "grad_norm": 1.241130759687672, "learning_rate": 9.21688203781022e-06, "loss": 0.4158, "step": 5936 }, { "epoch": 0.4104389906671275, "grad_norm": 0.6861418537029809, "learning_rate": 9.216581239762043e-06, "loss": 0.3467, "step": 5937 }, { "epoch": 0.4105081230556516, "grad_norm": 0.7893174867430917, "learning_rate": 9.216280388866545e-06, "loss": 0.3202, "step": 5938 }, { "epoch": 0.4105772554441756, "grad_norm": 0.7456266286498119, "learning_rate": 9.215979485127496e-06, "loss": 0.3392, "step": 5939 }, { "epoch": 0.41064638783269963, "grad_norm": 0.8716361752159602, "learning_rate": 9.215678528548668e-06, "loss": 0.4127, "step": 5940 }, { "epoch": 0.41071552022122365, "grad_norm": 0.8067830098505681, "learning_rate": 9.215377519133832e-06, "loss": 0.323, "step": 5941 }, { "epoch": 0.41078465260974767, "grad_norm": 0.7491419691375198, "learning_rate": 9.215076456886762e-06, "loss": 0.2826, "step": 5942 }, { "epoch": 0.4108537849982717, "grad_norm": 0.7284839399613317, "learning_rate": 9.214775341811234e-06, "loss": 0.4017, "step": 5943 }, { "epoch": 0.4109229173867957, "grad_norm": 0.7450836114817948, "learning_rate": 9.214474173911014e-06, "loss": 0.3947, "step": 5944 }, { "epoch": 0.4109920497753197, "grad_norm": 0.7271851395281749, "learning_rate": 9.214172953189883e-06, "loss": 0.3205, "step": 5945 }, { "epoch": 0.41106118216384374, "grad_norm": 0.7992917397697128, "learning_rate": 9.213871679651614e-06, "loss": 0.4075, "step": 5946 }, { "epoch": 0.41113031455236776, "grad_norm": 0.815659060358486, "learning_rate": 9.213570353299986e-06, "loss": 0.4364, "step": 5947 }, { "epoch": 0.41119944694089183, "grad_norm": 0.7168444637160755, "learning_rate": 9.21326897413877e-06, "loss": 0.3633, "step": 5948 }, { "epoch": 0.41126857932941585, "grad_norm": 0.7863296298107337, "learning_rate": 9.212967542171747e-06, "loss": 0.3103, "step": 5949 }, { "epoch": 0.41133771171793987, "grad_norm": 0.8670342084175976, "learning_rate": 9.212666057402695e-06, "loss": 0.3366, "step": 5950 }, { "epoch": 0.4114068441064639, "grad_norm": 0.6981201215140851, "learning_rate": 9.212364519835392e-06, "loss": 0.4141, "step": 5951 }, { "epoch": 0.4114759764949879, "grad_norm": 0.8714010602522894, "learning_rate": 9.212062929473616e-06, "loss": 0.33, "step": 5952 }, { "epoch": 0.4115451088835119, "grad_norm": 0.9597998180608975, "learning_rate": 9.21176128632115e-06, "loss": 0.281, "step": 5953 }, { "epoch": 0.41161424127203594, "grad_norm": 0.6739112185092503, "learning_rate": 9.21145959038177e-06, "loss": 0.3633, "step": 5954 }, { "epoch": 0.41168337366055996, "grad_norm": 0.7540027019716854, "learning_rate": 9.21115784165926e-06, "loss": 0.3805, "step": 5955 }, { "epoch": 0.411752506049084, "grad_norm": 0.7780349364380573, "learning_rate": 9.210856040157403e-06, "loss": 0.3164, "step": 5956 }, { "epoch": 0.411821638437608, "grad_norm": 0.824827866808208, "learning_rate": 9.21055418587998e-06, "loss": 0.3077, "step": 5957 }, { "epoch": 0.41189077082613207, "grad_norm": 0.7500753014758501, "learning_rate": 9.210252278830774e-06, "loss": 0.4492, "step": 5958 }, { "epoch": 0.4119599032146561, "grad_norm": 0.7287278908948636, "learning_rate": 9.209950319013568e-06, "loss": 0.3594, "step": 5959 }, { "epoch": 0.4120290356031801, "grad_norm": 0.6843157286361005, "learning_rate": 9.20964830643215e-06, "loss": 0.2561, "step": 5960 }, { "epoch": 0.4120981679917041, "grad_norm": 0.8789906912468474, "learning_rate": 9.209346241090303e-06, "loss": 0.3478, "step": 5961 }, { "epoch": 0.41216730038022814, "grad_norm": 0.7511435985705811, "learning_rate": 9.209044122991811e-06, "loss": 0.4099, "step": 5962 }, { "epoch": 0.41223643276875216, "grad_norm": 0.693910886324646, "learning_rate": 9.208741952140462e-06, "loss": 0.3352, "step": 5963 }, { "epoch": 0.4123055651572762, "grad_norm": 0.6957000857394369, "learning_rate": 9.208439728540044e-06, "loss": 0.3431, "step": 5964 }, { "epoch": 0.4123746975458002, "grad_norm": 0.7271045027277185, "learning_rate": 9.208137452194345e-06, "loss": 0.4678, "step": 5965 }, { "epoch": 0.4124438299343242, "grad_norm": 0.8132198447293908, "learning_rate": 9.207835123107151e-06, "loss": 0.2507, "step": 5966 }, { "epoch": 0.41251296232284823, "grad_norm": 0.7187629420856481, "learning_rate": 9.207532741282256e-06, "loss": 0.3121, "step": 5967 }, { "epoch": 0.4125820947113723, "grad_norm": 0.7497310400110726, "learning_rate": 9.207230306723445e-06, "loss": 0.386, "step": 5968 }, { "epoch": 0.4126512270998963, "grad_norm": 0.7500975595982919, "learning_rate": 9.20692781943451e-06, "loss": 0.4262, "step": 5969 }, { "epoch": 0.41272035948842034, "grad_norm": 0.928871431636108, "learning_rate": 9.206625279419243e-06, "loss": 0.3199, "step": 5970 }, { "epoch": 0.41278949187694436, "grad_norm": 0.7268043236558448, "learning_rate": 9.206322686681435e-06, "loss": 0.3266, "step": 5971 }, { "epoch": 0.4128586242654684, "grad_norm": 0.6798915865902558, "learning_rate": 9.20602004122488e-06, "loss": 0.3835, "step": 5972 }, { "epoch": 0.4129277566539924, "grad_norm": 0.7916894358129996, "learning_rate": 9.205717343053367e-06, "loss": 0.3798, "step": 5973 }, { "epoch": 0.4129968890425164, "grad_norm": 1.010333412872344, "learning_rate": 9.205414592170694e-06, "loss": 0.2977, "step": 5974 }, { "epoch": 0.41306602143104043, "grad_norm": 0.7752078352870968, "learning_rate": 9.205111788580655e-06, "loss": 0.327, "step": 5975 }, { "epoch": 0.41313515381956445, "grad_norm": 0.7335362512381832, "learning_rate": 9.204808932287041e-06, "loss": 0.4187, "step": 5976 }, { "epoch": 0.41320428620808847, "grad_norm": 0.7332422010849067, "learning_rate": 9.204506023293653e-06, "loss": 0.3498, "step": 5977 }, { "epoch": 0.41327341859661254, "grad_norm": 0.6632237883231663, "learning_rate": 9.204203061604287e-06, "loss": 0.2933, "step": 5978 }, { "epoch": 0.41334255098513656, "grad_norm": 0.8131771545602466, "learning_rate": 9.203900047222736e-06, "loss": 0.4274, "step": 5979 }, { "epoch": 0.4134116833736606, "grad_norm": 0.8944020197638657, "learning_rate": 9.2035969801528e-06, "loss": 0.3558, "step": 5980 }, { "epoch": 0.4134808157621846, "grad_norm": 0.6399366672639653, "learning_rate": 9.203293860398278e-06, "loss": 0.3644, "step": 5981 }, { "epoch": 0.4135499481507086, "grad_norm": 1.0517774321656004, "learning_rate": 9.20299068796297e-06, "loss": 0.3929, "step": 5982 }, { "epoch": 0.41361908053923263, "grad_norm": 0.7299527435636181, "learning_rate": 9.202687462850673e-06, "loss": 0.3956, "step": 5983 }, { "epoch": 0.41368821292775665, "grad_norm": 0.7828420441170814, "learning_rate": 9.202384185065188e-06, "loss": 0.2866, "step": 5984 }, { "epoch": 0.41375734531628067, "grad_norm": 0.6591752596930504, "learning_rate": 9.202080854610317e-06, "loss": 0.3333, "step": 5985 }, { "epoch": 0.4138264777048047, "grad_norm": 0.7234721889752416, "learning_rate": 9.201777471489861e-06, "loss": 0.4027, "step": 5986 }, { "epoch": 0.4138956100933287, "grad_norm": 0.7129644218901402, "learning_rate": 9.201474035707625e-06, "loss": 0.3914, "step": 5987 }, { "epoch": 0.4139647424818527, "grad_norm": 1.916528303732055, "learning_rate": 9.201170547267408e-06, "loss": 0.2855, "step": 5988 }, { "epoch": 0.4140338748703768, "grad_norm": 0.6951833272523722, "learning_rate": 9.200867006173015e-06, "loss": 0.2977, "step": 5989 }, { "epoch": 0.4141030072589008, "grad_norm": 1.0148335815128642, "learning_rate": 9.200563412428253e-06, "loss": 0.4446, "step": 5990 }, { "epoch": 0.41417213964742483, "grad_norm": 0.8301652132140145, "learning_rate": 9.200259766036923e-06, "loss": 0.302, "step": 5991 }, { "epoch": 0.41424127203594885, "grad_norm": 0.8205240991564526, "learning_rate": 9.199956067002833e-06, "loss": 0.2875, "step": 5992 }, { "epoch": 0.41431040442447287, "grad_norm": 0.830500630379696, "learning_rate": 9.19965231532979e-06, "loss": 0.3424, "step": 5993 }, { "epoch": 0.4143795368129969, "grad_norm": 0.8132545764654813, "learning_rate": 9.199348511021599e-06, "loss": 0.445, "step": 5994 }, { "epoch": 0.4144486692015209, "grad_norm": 0.6477327498084274, "learning_rate": 9.199044654082067e-06, "loss": 0.3572, "step": 5995 }, { "epoch": 0.4145178015900449, "grad_norm": 0.8181089576952414, "learning_rate": 9.198740744515006e-06, "loss": 0.3618, "step": 5996 }, { "epoch": 0.41458693397856894, "grad_norm": 0.6916168597520217, "learning_rate": 9.198436782324223e-06, "loss": 0.4118, "step": 5997 }, { "epoch": 0.41465606636709296, "grad_norm": 0.7336408866567418, "learning_rate": 9.198132767513526e-06, "loss": 0.3524, "step": 5998 }, { "epoch": 0.41472519875561703, "grad_norm": 0.7675807857369725, "learning_rate": 9.197828700086726e-06, "loss": 0.277, "step": 5999 }, { "epoch": 0.41479433114414105, "grad_norm": 0.8976450269599809, "learning_rate": 9.197524580047636e-06, "loss": 0.3921, "step": 6000 }, { "epoch": 0.41486346353266507, "grad_norm": 0.7512564048879008, "learning_rate": 9.197220407400066e-06, "loss": 0.4301, "step": 6001 }, { "epoch": 0.4149325959211891, "grad_norm": 0.7737046199656612, "learning_rate": 9.196916182147829e-06, "loss": 0.3649, "step": 6002 }, { "epoch": 0.4150017283097131, "grad_norm": 0.724447171329517, "learning_rate": 9.196611904294737e-06, "loss": 0.316, "step": 6003 }, { "epoch": 0.4150708606982371, "grad_norm": 0.7535103159287565, "learning_rate": 9.196307573844603e-06, "loss": 0.3353, "step": 6004 }, { "epoch": 0.41513999308676114, "grad_norm": 0.7060810146629686, "learning_rate": 9.196003190801242e-06, "loss": 0.4293, "step": 6005 }, { "epoch": 0.41520912547528516, "grad_norm": 0.7668351582810334, "learning_rate": 9.19569875516847e-06, "loss": 0.3206, "step": 6006 }, { "epoch": 0.4152782578638092, "grad_norm": 0.6856661296901649, "learning_rate": 9.195394266950101e-06, "loss": 0.3444, "step": 6007 }, { "epoch": 0.4153473902523332, "grad_norm": 0.8060881600214373, "learning_rate": 9.195089726149954e-06, "loss": 0.4896, "step": 6008 }, { "epoch": 0.41541652264085727, "grad_norm": 0.7717283159880561, "learning_rate": 9.194785132771841e-06, "loss": 0.2983, "step": 6009 }, { "epoch": 0.4154856550293813, "grad_norm": 0.6264327377363421, "learning_rate": 9.194480486819584e-06, "loss": 0.272, "step": 6010 }, { "epoch": 0.4155547874179053, "grad_norm": 1.0605908109795912, "learning_rate": 9.194175788297e-06, "loss": 0.41, "step": 6011 }, { "epoch": 0.4156239198064293, "grad_norm": 0.7260009840364785, "learning_rate": 9.193871037207907e-06, "loss": 0.3215, "step": 6012 }, { "epoch": 0.41569305219495334, "grad_norm": 0.7558460817121762, "learning_rate": 9.193566233556123e-06, "loss": 0.3706, "step": 6013 }, { "epoch": 0.41576218458347736, "grad_norm": 0.783280076757972, "learning_rate": 9.193261377345472e-06, "loss": 0.3139, "step": 6014 }, { "epoch": 0.4158313169720014, "grad_norm": 0.7500095578967066, "learning_rate": 9.192956468579773e-06, "loss": 0.4134, "step": 6015 }, { "epoch": 0.4159004493605254, "grad_norm": 0.7350694323342337, "learning_rate": 9.192651507262845e-06, "loss": 0.3498, "step": 6016 }, { "epoch": 0.4159695817490494, "grad_norm": 0.8113286581524547, "learning_rate": 9.192346493398514e-06, "loss": 0.3272, "step": 6017 }, { "epoch": 0.41603871413757343, "grad_norm": 0.9089931328310983, "learning_rate": 9.192041426990601e-06, "loss": 0.3674, "step": 6018 }, { "epoch": 0.4161078465260975, "grad_norm": 0.7896095172396312, "learning_rate": 9.19173630804293e-06, "loss": 0.4317, "step": 6019 }, { "epoch": 0.4161769789146215, "grad_norm": 0.8733151954951858, "learning_rate": 9.191431136559326e-06, "loss": 0.2876, "step": 6020 }, { "epoch": 0.41624611130314554, "grad_norm": 0.7641490317045173, "learning_rate": 9.19112591254361e-06, "loss": 0.2945, "step": 6021 }, { "epoch": 0.41631524369166956, "grad_norm": 0.7657748457259966, "learning_rate": 9.190820635999613e-06, "loss": 0.3936, "step": 6022 }, { "epoch": 0.4163843760801936, "grad_norm": 0.8536136177017826, "learning_rate": 9.190515306931156e-06, "loss": 0.3934, "step": 6023 }, { "epoch": 0.4164535084687176, "grad_norm": 0.9716187923253017, "learning_rate": 9.190209925342068e-06, "loss": 0.342, "step": 6024 }, { "epoch": 0.4165226408572416, "grad_norm": 0.9409014020977831, "learning_rate": 9.189904491236178e-06, "loss": 0.3831, "step": 6025 }, { "epoch": 0.41659177324576563, "grad_norm": 1.049232897666686, "learning_rate": 9.189599004617312e-06, "loss": 0.4092, "step": 6026 }, { "epoch": 0.41666090563428965, "grad_norm": 0.7825682480204623, "learning_rate": 9.189293465489297e-06, "loss": 0.3531, "step": 6027 }, { "epoch": 0.41673003802281366, "grad_norm": 0.780833998154156, "learning_rate": 9.188987873855966e-06, "loss": 0.2717, "step": 6028 }, { "epoch": 0.41679917041133774, "grad_norm": 0.811008776017749, "learning_rate": 9.188682229721149e-06, "loss": 0.4055, "step": 6029 }, { "epoch": 0.41686830279986176, "grad_norm": 0.9246843268117864, "learning_rate": 9.188376533088674e-06, "loss": 0.4107, "step": 6030 }, { "epoch": 0.4169374351883858, "grad_norm": 0.8564431790531729, "learning_rate": 9.188070783962373e-06, "loss": 0.345, "step": 6031 }, { "epoch": 0.4170065675769098, "grad_norm": 0.7752728447704261, "learning_rate": 9.18776498234608e-06, "loss": 0.3595, "step": 6032 }, { "epoch": 0.4170756999654338, "grad_norm": 0.8197453751834854, "learning_rate": 9.187459128243625e-06, "loss": 0.4396, "step": 6033 }, { "epoch": 0.41714483235395783, "grad_norm": 0.7376940500588499, "learning_rate": 9.187153221658843e-06, "loss": 0.3524, "step": 6034 }, { "epoch": 0.41721396474248185, "grad_norm": 0.6537963445497162, "learning_rate": 9.186847262595569e-06, "loss": 0.3196, "step": 6035 }, { "epoch": 0.41728309713100586, "grad_norm": 0.8298798374894748, "learning_rate": 9.186541251057635e-06, "loss": 0.3625, "step": 6036 }, { "epoch": 0.4173522295195299, "grad_norm": 0.7210716345924203, "learning_rate": 9.186235187048877e-06, "loss": 0.4937, "step": 6037 }, { "epoch": 0.4174213619080539, "grad_norm": 0.6750351631605693, "learning_rate": 9.185929070573134e-06, "loss": 0.316, "step": 6038 }, { "epoch": 0.417490494296578, "grad_norm": 0.8486216886297877, "learning_rate": 9.185622901634239e-06, "loss": 0.3704, "step": 6039 }, { "epoch": 0.417559626685102, "grad_norm": 0.7672593659163607, "learning_rate": 9.18531668023603e-06, "loss": 0.3773, "step": 6040 }, { "epoch": 0.417628759073626, "grad_norm": 0.8297812949991965, "learning_rate": 9.185010406382346e-06, "loss": 0.3513, "step": 6041 }, { "epoch": 0.41769789146215003, "grad_norm": 0.7187447217029841, "learning_rate": 9.184704080077026e-06, "loss": 0.3408, "step": 6042 }, { "epoch": 0.41776702385067405, "grad_norm": 1.153454923343402, "learning_rate": 9.184397701323907e-06, "loss": 0.3642, "step": 6043 }, { "epoch": 0.41783615623919806, "grad_norm": 0.8200506719885211, "learning_rate": 9.18409127012683e-06, "loss": 0.3534, "step": 6044 }, { "epoch": 0.4179052886277221, "grad_norm": 0.7536445943113325, "learning_rate": 9.183784786489636e-06, "loss": 0.3296, "step": 6045 }, { "epoch": 0.4179744210162461, "grad_norm": 0.7102976186179942, "learning_rate": 9.183478250416164e-06, "loss": 0.2864, "step": 6046 }, { "epoch": 0.4180435534047701, "grad_norm": 0.7965611318684419, "learning_rate": 9.183171661910262e-06, "loss": 0.4785, "step": 6047 }, { "epoch": 0.41811268579329414, "grad_norm": 0.762491571473106, "learning_rate": 9.182865020975765e-06, "loss": 0.403, "step": 6048 }, { "epoch": 0.41818181818181815, "grad_norm": 0.895166373660685, "learning_rate": 9.182558327616522e-06, "loss": 0.2982, "step": 6049 }, { "epoch": 0.41825095057034223, "grad_norm": 0.7346168874247799, "learning_rate": 9.182251581836372e-06, "loss": 0.3395, "step": 6050 }, { "epoch": 0.41832008295886625, "grad_norm": 0.719889935220398, "learning_rate": 9.181944783639165e-06, "loss": 0.4338, "step": 6051 }, { "epoch": 0.41838921534739026, "grad_norm": 0.694824739081107, "learning_rate": 9.181637933028741e-06, "loss": 0.29, "step": 6052 }, { "epoch": 0.4184583477359143, "grad_norm": 1.049475404009846, "learning_rate": 9.181331030008949e-06, "loss": 0.2902, "step": 6053 }, { "epoch": 0.4185274801244383, "grad_norm": 0.7636524191929616, "learning_rate": 9.181024074583633e-06, "loss": 0.4535, "step": 6054 }, { "epoch": 0.4185966125129623, "grad_norm": 0.7317604106806445, "learning_rate": 9.180717066756643e-06, "loss": 0.4285, "step": 6055 }, { "epoch": 0.41866574490148634, "grad_norm": 0.7399858303013385, "learning_rate": 9.180410006531825e-06, "loss": 0.3117, "step": 6056 }, { "epoch": 0.41873487729001035, "grad_norm": 0.7820277712450603, "learning_rate": 9.180102893913028e-06, "loss": 0.3653, "step": 6057 }, { "epoch": 0.4188040096785344, "grad_norm": 1.513541458788304, "learning_rate": 9.179795728904102e-06, "loss": 0.4031, "step": 6058 }, { "epoch": 0.4188731420670584, "grad_norm": 0.6902872460081335, "learning_rate": 9.179488511508893e-06, "loss": 0.3144, "step": 6059 }, { "epoch": 0.41894227445558246, "grad_norm": 0.9569454638535523, "learning_rate": 9.179181241731256e-06, "loss": 0.2495, "step": 6060 }, { "epoch": 0.4190114068441065, "grad_norm": 0.9798486034682581, "learning_rate": 9.178873919575042e-06, "loss": 0.3897, "step": 6061 }, { "epoch": 0.4190805392326305, "grad_norm": 0.8457983669840924, "learning_rate": 9.178566545044098e-06, "loss": 0.4529, "step": 6062 }, { "epoch": 0.4191496716211545, "grad_norm": 0.7642433181654098, "learning_rate": 9.17825911814228e-06, "loss": 0.2714, "step": 6063 }, { "epoch": 0.41921880400967854, "grad_norm": 0.9148917194756694, "learning_rate": 9.17795163887344e-06, "loss": 0.2654, "step": 6064 }, { "epoch": 0.41928793639820255, "grad_norm": 1.1033626705196413, "learning_rate": 9.177644107241434e-06, "loss": 0.4166, "step": 6065 }, { "epoch": 0.4193570687867266, "grad_norm": 1.1989601192369521, "learning_rate": 9.177336523250112e-06, "loss": 0.4125, "step": 6066 }, { "epoch": 0.4194262011752506, "grad_norm": 0.6649249187506453, "learning_rate": 9.177028886903332e-06, "loss": 0.2408, "step": 6067 }, { "epoch": 0.4194953335637746, "grad_norm": 0.8168951690861038, "learning_rate": 9.176721198204951e-06, "loss": 0.3856, "step": 6068 }, { "epoch": 0.4195644659522986, "grad_norm": 0.7946925881370518, "learning_rate": 9.176413457158821e-06, "loss": 0.4273, "step": 6069 }, { "epoch": 0.4196335983408227, "grad_norm": 0.7332679535152908, "learning_rate": 9.176105663768804e-06, "loss": 0.3088, "step": 6070 }, { "epoch": 0.4197027307293467, "grad_norm": 0.7758402722771407, "learning_rate": 9.175797818038755e-06, "loss": 0.3601, "step": 6071 }, { "epoch": 0.41977186311787074, "grad_norm": 0.7319196748357474, "learning_rate": 9.17548991997253e-06, "loss": 0.4348, "step": 6072 }, { "epoch": 0.41984099550639475, "grad_norm": 1.0273908076758658, "learning_rate": 9.175181969573993e-06, "loss": 0.3822, "step": 6073 }, { "epoch": 0.41991012789491877, "grad_norm": 0.7302793537009349, "learning_rate": 9.174873966846998e-06, "loss": 0.3423, "step": 6074 }, { "epoch": 0.4199792602834428, "grad_norm": 0.7457278223935945, "learning_rate": 9.17456591179541e-06, "loss": 0.3477, "step": 6075 }, { "epoch": 0.4200483926719668, "grad_norm": 0.702562217033282, "learning_rate": 9.174257804423087e-06, "loss": 0.4186, "step": 6076 }, { "epoch": 0.4201175250604908, "grad_norm": 0.7658198653251606, "learning_rate": 9.173949644733894e-06, "loss": 0.3194, "step": 6077 }, { "epoch": 0.42018665744901484, "grad_norm": 0.7777366275509462, "learning_rate": 9.17364143273169e-06, "loss": 0.3654, "step": 6078 }, { "epoch": 0.42025578983753886, "grad_norm": 0.7880789073273691, "learning_rate": 9.173333168420338e-06, "loss": 0.3999, "step": 6079 }, { "epoch": 0.42032492222606294, "grad_norm": 0.8204429119222207, "learning_rate": 9.173024851803702e-06, "loss": 0.4256, "step": 6080 }, { "epoch": 0.42039405461458695, "grad_norm": 1.0222996024290552, "learning_rate": 9.172716482885648e-06, "loss": 0.2915, "step": 6081 }, { "epoch": 0.42046318700311097, "grad_norm": 0.7696875470300519, "learning_rate": 9.17240806167004e-06, "loss": 0.3212, "step": 6082 }, { "epoch": 0.420532319391635, "grad_norm": 0.6980572832615036, "learning_rate": 9.172099588160741e-06, "loss": 0.435, "step": 6083 }, { "epoch": 0.420601451780159, "grad_norm": 0.8545763360726387, "learning_rate": 9.17179106236162e-06, "loss": 0.3318, "step": 6084 }, { "epoch": 0.420670584168683, "grad_norm": 0.6876273321739806, "learning_rate": 9.171482484276542e-06, "loss": 0.3363, "step": 6085 }, { "epoch": 0.42073971655720704, "grad_norm": 0.826820599515449, "learning_rate": 9.171173853909377e-06, "loss": 0.4319, "step": 6086 }, { "epoch": 0.42080884894573106, "grad_norm": 0.7555544081137043, "learning_rate": 9.17086517126399e-06, "loss": 0.4518, "step": 6087 }, { "epoch": 0.4208779813342551, "grad_norm": 0.6882283685851948, "learning_rate": 9.170556436344252e-06, "loss": 0.3357, "step": 6088 }, { "epoch": 0.4209471137227791, "grad_norm": 0.7172149048368137, "learning_rate": 9.170247649154032e-06, "loss": 0.3383, "step": 6089 }, { "epoch": 0.42101624611130317, "grad_norm": 0.7528217334460501, "learning_rate": 9.1699388096972e-06, "loss": 0.4599, "step": 6090 }, { "epoch": 0.4210853784998272, "grad_norm": 0.7897643880846663, "learning_rate": 9.169629917977625e-06, "loss": 0.3587, "step": 6091 }, { "epoch": 0.4211545108883512, "grad_norm": 0.6645279652869305, "learning_rate": 9.169320973999183e-06, "loss": 0.2561, "step": 6092 }, { "epoch": 0.4212236432768752, "grad_norm": 0.7539493704888661, "learning_rate": 9.16901197776574e-06, "loss": 0.4097, "step": 6093 }, { "epoch": 0.42129277566539924, "grad_norm": 0.7398293137095394, "learning_rate": 9.168702929281174e-06, "loss": 0.3638, "step": 6094 }, { "epoch": 0.42136190805392326, "grad_norm": 0.6875295674244876, "learning_rate": 9.168393828549356e-06, "loss": 0.3046, "step": 6095 }, { "epoch": 0.4214310404424473, "grad_norm": 0.8272337610890195, "learning_rate": 9.168084675574159e-06, "loss": 0.3024, "step": 6096 }, { "epoch": 0.4215001728309713, "grad_norm": 0.741388378011804, "learning_rate": 9.16777547035946e-06, "loss": 0.4159, "step": 6097 }, { "epoch": 0.4215693052194953, "grad_norm": 0.681252364607122, "learning_rate": 9.167466212909131e-06, "loss": 0.3157, "step": 6098 }, { "epoch": 0.42163843760801933, "grad_norm": 0.6588094654792963, "learning_rate": 9.167156903227053e-06, "loss": 0.2943, "step": 6099 }, { "epoch": 0.4217075699965434, "grad_norm": 0.8069531155132341, "learning_rate": 9.166847541317098e-06, "loss": 0.4295, "step": 6100 }, { "epoch": 0.4217767023850674, "grad_norm": 1.783622565157874, "learning_rate": 9.166538127183147e-06, "loss": 0.3953, "step": 6101 }, { "epoch": 0.42184583477359144, "grad_norm": 0.7664745235354665, "learning_rate": 9.166228660829075e-06, "loss": 0.3224, "step": 6102 }, { "epoch": 0.42191496716211546, "grad_norm": 0.670253011086567, "learning_rate": 9.165919142258762e-06, "loss": 0.2854, "step": 6103 }, { "epoch": 0.4219840995506395, "grad_norm": 0.727996913115067, "learning_rate": 9.165609571476086e-06, "loss": 0.4474, "step": 6104 }, { "epoch": 0.4220532319391635, "grad_norm": 0.949431253413613, "learning_rate": 9.165299948484927e-06, "loss": 0.4108, "step": 6105 }, { "epoch": 0.4221223643276875, "grad_norm": 0.7881133269501936, "learning_rate": 9.164990273289169e-06, "loss": 0.3593, "step": 6106 }, { "epoch": 0.42219149671621153, "grad_norm": 0.7263084688006254, "learning_rate": 9.16468054589269e-06, "loss": 0.2856, "step": 6107 }, { "epoch": 0.42226062910473555, "grad_norm": 1.727096193865661, "learning_rate": 9.16437076629937e-06, "loss": 0.4435, "step": 6108 }, { "epoch": 0.42232976149325957, "grad_norm": 0.8073429091951653, "learning_rate": 9.164060934513097e-06, "loss": 0.359, "step": 6109 }, { "epoch": 0.4223988938817836, "grad_norm": 0.7739094068279279, "learning_rate": 9.163751050537751e-06, "loss": 0.3112, "step": 6110 }, { "epoch": 0.42246802627030766, "grad_norm": 0.7408045015824977, "learning_rate": 9.163441114377217e-06, "loss": 0.3796, "step": 6111 }, { "epoch": 0.4225371586588317, "grad_norm": 0.7345449576023889, "learning_rate": 9.163131126035376e-06, "loss": 0.3271, "step": 6112 }, { "epoch": 0.4226062910473557, "grad_norm": 0.7745944456355331, "learning_rate": 9.162821085516117e-06, "loss": 0.3046, "step": 6113 }, { "epoch": 0.4226754234358797, "grad_norm": 0.7736722356110027, "learning_rate": 9.162510992823326e-06, "loss": 0.3232, "step": 6114 }, { "epoch": 0.42274455582440373, "grad_norm": 0.7526089775659079, "learning_rate": 9.162200847960886e-06, "loss": 0.4133, "step": 6115 }, { "epoch": 0.42281368821292775, "grad_norm": 0.7882670309318056, "learning_rate": 9.161890650932686e-06, "loss": 0.3873, "step": 6116 }, { "epoch": 0.42288282060145177, "grad_norm": 0.6968617947392298, "learning_rate": 9.161580401742615e-06, "loss": 0.2796, "step": 6117 }, { "epoch": 0.4229519529899758, "grad_norm": 0.7044717814803986, "learning_rate": 9.161270100394561e-06, "loss": 0.3499, "step": 6118 }, { "epoch": 0.4230210853784998, "grad_norm": 0.7604302463312409, "learning_rate": 9.16095974689241e-06, "loss": 0.4157, "step": 6119 }, { "epoch": 0.4230902177670238, "grad_norm": 0.8797988888728132, "learning_rate": 9.160649341240056e-06, "loss": 0.3311, "step": 6120 }, { "epoch": 0.4231593501555479, "grad_norm": 0.756893679351092, "learning_rate": 9.160338883441388e-06, "loss": 0.3608, "step": 6121 }, { "epoch": 0.4232284825440719, "grad_norm": 0.6907618748378681, "learning_rate": 9.160028373500296e-06, "loss": 0.437, "step": 6122 }, { "epoch": 0.42329761493259593, "grad_norm": 0.6494992189703707, "learning_rate": 9.159717811420671e-06, "loss": 0.3829, "step": 6123 }, { "epoch": 0.42336674732111995, "grad_norm": 0.6835487743268326, "learning_rate": 9.159407197206407e-06, "loss": 0.3318, "step": 6124 }, { "epoch": 0.42343587970964397, "grad_norm": 0.703356411098253, "learning_rate": 9.159096530861397e-06, "loss": 0.3189, "step": 6125 }, { "epoch": 0.423505012098168, "grad_norm": 0.7742320702022546, "learning_rate": 9.158785812389535e-06, "loss": 0.5122, "step": 6126 }, { "epoch": 0.423574144486692, "grad_norm": 0.8073496742586448, "learning_rate": 9.158475041794715e-06, "loss": 0.2827, "step": 6127 }, { "epoch": 0.423643276875216, "grad_norm": 0.6555639725911577, "learning_rate": 9.15816421908083e-06, "loss": 0.2876, "step": 6128 }, { "epoch": 0.42371240926374004, "grad_norm": 0.7949861914947146, "learning_rate": 9.157853344251776e-06, "loss": 0.3206, "step": 6129 }, { "epoch": 0.42378154165226406, "grad_norm": 0.8085152599300279, "learning_rate": 9.157542417311454e-06, "loss": 0.4485, "step": 6130 }, { "epoch": 0.42385067404078813, "grad_norm": 0.7432703082867805, "learning_rate": 9.157231438263753e-06, "loss": 0.2899, "step": 6131 }, { "epoch": 0.42391980642931215, "grad_norm": 0.7623146822869118, "learning_rate": 9.156920407112578e-06, "loss": 0.3438, "step": 6132 }, { "epoch": 0.42398893881783617, "grad_norm": 0.7562005375745081, "learning_rate": 9.156609323861822e-06, "loss": 0.4838, "step": 6133 }, { "epoch": 0.4240580712063602, "grad_norm": 0.6915312626714524, "learning_rate": 9.156298188515389e-06, "loss": 0.3488, "step": 6134 }, { "epoch": 0.4241272035948842, "grad_norm": 0.7066486258745016, "learning_rate": 9.155987001077173e-06, "loss": 0.3321, "step": 6135 }, { "epoch": 0.4241963359834082, "grad_norm": 1.0475875486072384, "learning_rate": 9.155675761551078e-06, "loss": 0.3907, "step": 6136 }, { "epoch": 0.42426546837193224, "grad_norm": 1.2420562470781915, "learning_rate": 9.155364469941001e-06, "loss": 0.4152, "step": 6137 }, { "epoch": 0.42433460076045626, "grad_norm": 0.7462933165951063, "learning_rate": 9.155053126250848e-06, "loss": 0.3442, "step": 6138 }, { "epoch": 0.4244037331489803, "grad_norm": 0.7186495407459711, "learning_rate": 9.154741730484518e-06, "loss": 0.2966, "step": 6139 }, { "epoch": 0.4244728655375043, "grad_norm": 0.744496285881556, "learning_rate": 9.154430282645915e-06, "loss": 0.4183, "step": 6140 }, { "epoch": 0.42454199792602837, "grad_norm": 0.9473787953178074, "learning_rate": 9.154118782738944e-06, "loss": 0.3523, "step": 6141 }, { "epoch": 0.4246111303145524, "grad_norm": 0.715577919254516, "learning_rate": 9.153807230767506e-06, "loss": 0.3276, "step": 6142 }, { "epoch": 0.4246802627030764, "grad_norm": 0.959061177732029, "learning_rate": 9.153495626735506e-06, "loss": 0.3903, "step": 6143 }, { "epoch": 0.4247493950916004, "grad_norm": 0.7372347888595517, "learning_rate": 9.153183970646853e-06, "loss": 0.4493, "step": 6144 }, { "epoch": 0.42481852748012444, "grad_norm": 0.965109725298633, "learning_rate": 9.15287226250545e-06, "loss": 0.3314, "step": 6145 }, { "epoch": 0.42488765986864846, "grad_norm": 0.761054710754326, "learning_rate": 9.152560502315202e-06, "loss": 0.3714, "step": 6146 }, { "epoch": 0.4249567922571725, "grad_norm": 0.7997600762796254, "learning_rate": 9.15224869008002e-06, "loss": 0.4288, "step": 6147 }, { "epoch": 0.4250259246456965, "grad_norm": 0.7680100703726123, "learning_rate": 9.151936825803813e-06, "loss": 0.3637, "step": 6148 }, { "epoch": 0.4250950570342205, "grad_norm": 0.9298725871481253, "learning_rate": 9.151624909490485e-06, "loss": 0.3013, "step": 6149 }, { "epoch": 0.42516418942274453, "grad_norm": 0.7076040415862941, "learning_rate": 9.151312941143948e-06, "loss": 0.3331, "step": 6150 }, { "epoch": 0.4252333218112686, "grad_norm": 0.7381881513067922, "learning_rate": 9.151000920768112e-06, "loss": 0.402, "step": 6151 }, { "epoch": 0.4253024541997926, "grad_norm": 0.6743568852220869, "learning_rate": 9.150688848366886e-06, "loss": 0.3034, "step": 6152 }, { "epoch": 0.42537158658831664, "grad_norm": 0.7595150438162623, "learning_rate": 9.150376723944185e-06, "loss": 0.3287, "step": 6153 }, { "epoch": 0.42544071897684066, "grad_norm": 0.7575190875269078, "learning_rate": 9.150064547503917e-06, "loss": 0.3818, "step": 6154 }, { "epoch": 0.4255098513653647, "grad_norm": 0.760867821490607, "learning_rate": 9.149752319049997e-06, "loss": 0.4247, "step": 6155 }, { "epoch": 0.4255789837538887, "grad_norm": 0.7232185172557513, "learning_rate": 9.149440038586335e-06, "loss": 0.3065, "step": 6156 }, { "epoch": 0.4256481161424127, "grad_norm": 0.7528466221502707, "learning_rate": 9.149127706116849e-06, "loss": 0.3546, "step": 6157 }, { "epoch": 0.42571724853093673, "grad_norm": 0.7279497053122541, "learning_rate": 9.14881532164545e-06, "loss": 0.4329, "step": 6158 }, { "epoch": 0.42578638091946075, "grad_norm": 0.856397557839375, "learning_rate": 9.148502885176057e-06, "loss": 0.2545, "step": 6159 }, { "epoch": 0.42585551330798477, "grad_norm": 0.6911888853694046, "learning_rate": 9.148190396712581e-06, "loss": 0.2811, "step": 6160 }, { "epoch": 0.42592464569650884, "grad_norm": 0.7412605881030363, "learning_rate": 9.147877856258943e-06, "loss": 0.3422, "step": 6161 }, { "epoch": 0.42599377808503286, "grad_norm": 0.6905439882071649, "learning_rate": 9.147565263819058e-06, "loss": 0.3642, "step": 6162 }, { "epoch": 0.4260629104735569, "grad_norm": 0.6676024055381361, "learning_rate": 9.147252619396844e-06, "loss": 0.2993, "step": 6163 }, { "epoch": 0.4261320428620809, "grad_norm": 0.873011610656225, "learning_rate": 9.146939922996221e-06, "loss": 0.3403, "step": 6164 }, { "epoch": 0.4262011752506049, "grad_norm": 0.7444252832717196, "learning_rate": 9.146627174621104e-06, "loss": 0.4407, "step": 6165 }, { "epoch": 0.42627030763912893, "grad_norm": 0.9855727356120234, "learning_rate": 9.146314374275417e-06, "loss": 0.3228, "step": 6166 }, { "epoch": 0.42633944002765295, "grad_norm": 0.8634852492186234, "learning_rate": 9.146001521963077e-06, "loss": 0.2476, "step": 6167 }, { "epoch": 0.42640857241617697, "grad_norm": 0.7568149791444683, "learning_rate": 9.14568861768801e-06, "loss": 0.3916, "step": 6168 }, { "epoch": 0.426477704804701, "grad_norm": 0.7701928521810557, "learning_rate": 9.145375661454133e-06, "loss": 0.3578, "step": 6169 }, { "epoch": 0.426546837193225, "grad_norm": 0.7501741533621974, "learning_rate": 9.14506265326537e-06, "loss": 0.3374, "step": 6170 }, { "epoch": 0.4266159695817491, "grad_norm": 0.8577982735382241, "learning_rate": 9.144749593125641e-06, "loss": 0.3995, "step": 6171 }, { "epoch": 0.4266851019702731, "grad_norm": 0.8729933719853372, "learning_rate": 9.144436481038876e-06, "loss": 0.448, "step": 6172 }, { "epoch": 0.4267542343587971, "grad_norm": 0.7220502376292278, "learning_rate": 9.144123317008995e-06, "loss": 0.3539, "step": 6173 }, { "epoch": 0.42682336674732113, "grad_norm": 0.7966558246240268, "learning_rate": 9.143810101039922e-06, "loss": 0.3409, "step": 6174 }, { "epoch": 0.42689249913584515, "grad_norm": 0.6943011112463545, "learning_rate": 9.143496833135585e-06, "loss": 0.3158, "step": 6175 }, { "epoch": 0.42696163152436917, "grad_norm": 0.7653653114364465, "learning_rate": 9.143183513299912e-06, "loss": 0.4272, "step": 6176 }, { "epoch": 0.4270307639128932, "grad_norm": 0.7319652280753038, "learning_rate": 9.142870141536827e-06, "loss": 0.2921, "step": 6177 }, { "epoch": 0.4270998963014172, "grad_norm": 0.8721781611493197, "learning_rate": 9.142556717850256e-06, "loss": 0.3236, "step": 6178 }, { "epoch": 0.4271690286899412, "grad_norm": 0.80395026622576, "learning_rate": 9.14224324224413e-06, "loss": 0.4172, "step": 6179 }, { "epoch": 0.42723816107846524, "grad_norm": 1.1070598999756254, "learning_rate": 9.141929714722378e-06, "loss": 0.3489, "step": 6180 }, { "epoch": 0.42730729346698926, "grad_norm": 0.7260250723480279, "learning_rate": 9.141616135288927e-06, "loss": 0.3351, "step": 6181 }, { "epoch": 0.42737642585551333, "grad_norm": 0.6931081283917674, "learning_rate": 9.141302503947709e-06, "loss": 0.3538, "step": 6182 }, { "epoch": 0.42744555824403735, "grad_norm": 2.715626756585954, "learning_rate": 9.140988820702655e-06, "loss": 0.3999, "step": 6183 }, { "epoch": 0.42751469063256137, "grad_norm": 0.7295236002767811, "learning_rate": 9.140675085557697e-06, "loss": 0.2966, "step": 6184 }, { "epoch": 0.4275838230210854, "grad_norm": 0.7637408626904454, "learning_rate": 9.140361298516765e-06, "loss": 0.4002, "step": 6185 }, { "epoch": 0.4276529554096094, "grad_norm": 0.7715580452455157, "learning_rate": 9.140047459583795e-06, "loss": 0.3775, "step": 6186 }, { "epoch": 0.4277220877981334, "grad_norm": 0.7545837175518462, "learning_rate": 9.139733568762716e-06, "loss": 0.4091, "step": 6187 }, { "epoch": 0.42779122018665744, "grad_norm": 0.6766192224856337, "learning_rate": 9.139419626057464e-06, "loss": 0.3231, "step": 6188 }, { "epoch": 0.42786035257518146, "grad_norm": 0.7626028920749361, "learning_rate": 9.139105631471977e-06, "loss": 0.3287, "step": 6189 }, { "epoch": 0.4279294849637055, "grad_norm": 0.7393316529987144, "learning_rate": 9.138791585010185e-06, "loss": 0.396, "step": 6190 }, { "epoch": 0.4279986173522295, "grad_norm": 0.6962971105499787, "learning_rate": 9.138477486676027e-06, "loss": 0.3079, "step": 6191 }, { "epoch": 0.42806774974075357, "grad_norm": 0.6653641392030981, "learning_rate": 9.138163336473439e-06, "loss": 0.2825, "step": 6192 }, { "epoch": 0.4281368821292776, "grad_norm": 0.7600057775147213, "learning_rate": 9.137849134406358e-06, "loss": 0.3839, "step": 6193 }, { "epoch": 0.4282060145178016, "grad_norm": 0.736336212791896, "learning_rate": 9.137534880478724e-06, "loss": 0.4046, "step": 6194 }, { "epoch": 0.4282751469063256, "grad_norm": 0.6735032966176084, "learning_rate": 9.137220574694473e-06, "loss": 0.2885, "step": 6195 }, { "epoch": 0.42834427929484964, "grad_norm": 0.6793023383777501, "learning_rate": 9.136906217057546e-06, "loss": 0.2759, "step": 6196 }, { "epoch": 0.42841341168337366, "grad_norm": 0.686535578566101, "learning_rate": 9.136591807571881e-06, "loss": 0.4122, "step": 6197 }, { "epoch": 0.4284825440718977, "grad_norm": 0.8354593840536687, "learning_rate": 9.13627734624142e-06, "loss": 0.3839, "step": 6198 }, { "epoch": 0.4285516764604217, "grad_norm": 0.694528461682357, "learning_rate": 9.135962833070105e-06, "loss": 0.2738, "step": 6199 }, { "epoch": 0.4286208088489457, "grad_norm": 0.762115697035187, "learning_rate": 9.135648268061876e-06, "loss": 0.3398, "step": 6200 }, { "epoch": 0.42868994123746973, "grad_norm": 1.0653776318999544, "learning_rate": 9.135333651220677e-06, "loss": 0.4075, "step": 6201 }, { "epoch": 0.4287590736259938, "grad_norm": 0.6714332857866735, "learning_rate": 9.13501898255045e-06, "loss": 0.3513, "step": 6202 }, { "epoch": 0.4288282060145178, "grad_norm": 0.7825262076449369, "learning_rate": 9.134704262055141e-06, "loss": 0.2845, "step": 6203 }, { "epoch": 0.42889733840304184, "grad_norm": 0.7387495243977558, "learning_rate": 9.13438948973869e-06, "loss": 0.4162, "step": 6204 }, { "epoch": 0.42896647079156586, "grad_norm": 0.8669601160930805, "learning_rate": 9.134074665605048e-06, "loss": 0.3713, "step": 6205 }, { "epoch": 0.4290356031800899, "grad_norm": 0.9435052234418196, "learning_rate": 9.133759789658156e-06, "loss": 0.3593, "step": 6206 }, { "epoch": 0.4291047355686139, "grad_norm": 0.6795478881965917, "learning_rate": 9.133444861901962e-06, "loss": 0.3148, "step": 6207 }, { "epoch": 0.4291738679571379, "grad_norm": 0.690921634838335, "learning_rate": 9.133129882340413e-06, "loss": 0.4666, "step": 6208 }, { "epoch": 0.42924300034566193, "grad_norm": 0.7382115597864023, "learning_rate": 9.132814850977456e-06, "loss": 0.3085, "step": 6209 }, { "epoch": 0.42931213273418595, "grad_norm": 0.707140704952794, "learning_rate": 9.132499767817042e-06, "loss": 0.352, "step": 6210 }, { "epoch": 0.42938126512270997, "grad_norm": 0.7375361635350082, "learning_rate": 9.132184632863116e-06, "loss": 0.3484, "step": 6211 }, { "epoch": 0.42945039751123404, "grad_norm": 0.7941030412949971, "learning_rate": 9.131869446119632e-06, "loss": 0.3747, "step": 6212 }, { "epoch": 0.42951952989975806, "grad_norm": 0.7773585355427799, "learning_rate": 9.131554207590537e-06, "loss": 0.2937, "step": 6213 }, { "epoch": 0.4295886622882821, "grad_norm": 0.7241948845964519, "learning_rate": 9.131238917279784e-06, "loss": 0.3413, "step": 6214 }, { "epoch": 0.4296577946768061, "grad_norm": 0.7373866452240305, "learning_rate": 9.130923575191323e-06, "loss": 0.4758, "step": 6215 }, { "epoch": 0.4297269270653301, "grad_norm": 1.2879014398399327, "learning_rate": 9.130608181329107e-06, "loss": 0.3645, "step": 6216 }, { "epoch": 0.42979605945385413, "grad_norm": 0.6777187443474942, "learning_rate": 9.130292735697088e-06, "loss": 0.3234, "step": 6217 }, { "epoch": 0.42986519184237815, "grad_norm": 0.7340212038522755, "learning_rate": 9.12997723829922e-06, "loss": 0.3697, "step": 6218 }, { "epoch": 0.42993432423090217, "grad_norm": 0.7558960816539587, "learning_rate": 9.129661689139462e-06, "loss": 0.4152, "step": 6219 }, { "epoch": 0.4300034566194262, "grad_norm": 0.6908235376200306, "learning_rate": 9.12934608822176e-06, "loss": 0.3118, "step": 6220 }, { "epoch": 0.4300725890079502, "grad_norm": 0.6224972570381457, "learning_rate": 9.129030435550076e-06, "loss": 0.2807, "step": 6221 }, { "epoch": 0.4301417213964743, "grad_norm": 0.8084310084572888, "learning_rate": 9.128714731128363e-06, "loss": 0.4161, "step": 6222 }, { "epoch": 0.4302108537849983, "grad_norm": 0.9037646620544182, "learning_rate": 9.128398974960579e-06, "loss": 0.3729, "step": 6223 }, { "epoch": 0.4302799861735223, "grad_norm": 0.6764839925055779, "learning_rate": 9.128083167050681e-06, "loss": 0.3301, "step": 6224 }, { "epoch": 0.43034911856204633, "grad_norm": 0.7787661292141405, "learning_rate": 9.127767307402628e-06, "loss": 0.3882, "step": 6225 }, { "epoch": 0.43041825095057035, "grad_norm": 0.9449788782427255, "learning_rate": 9.12745139602038e-06, "loss": 0.3611, "step": 6226 }, { "epoch": 0.43048738333909436, "grad_norm": 0.6982179369017916, "learning_rate": 9.12713543290789e-06, "loss": 0.3373, "step": 6227 }, { "epoch": 0.4305565157276184, "grad_norm": 0.6508632038341994, "learning_rate": 9.126819418069126e-06, "loss": 0.2856, "step": 6228 }, { "epoch": 0.4306256481161424, "grad_norm": 0.7657735267402946, "learning_rate": 9.126503351508044e-06, "loss": 0.4406, "step": 6229 }, { "epoch": 0.4306947805046664, "grad_norm": 0.7879856354650915, "learning_rate": 9.126187233228608e-06, "loss": 0.4193, "step": 6230 }, { "epoch": 0.43076391289319044, "grad_norm": 0.7209129519804071, "learning_rate": 9.125871063234776e-06, "loss": 0.3097, "step": 6231 }, { "epoch": 0.4308330452817145, "grad_norm": 0.812955588575853, "learning_rate": 9.125554841530515e-06, "loss": 0.3368, "step": 6232 }, { "epoch": 0.43090217767023853, "grad_norm": 0.7492342599400623, "learning_rate": 9.125238568119786e-06, "loss": 0.4828, "step": 6233 }, { "epoch": 0.43097131005876255, "grad_norm": 0.9167493136890961, "learning_rate": 9.124922243006553e-06, "loss": 0.3194, "step": 6234 }, { "epoch": 0.43104044244728656, "grad_norm": 0.7329736298126117, "learning_rate": 9.124605866194782e-06, "loss": 0.3348, "step": 6235 }, { "epoch": 0.4311095748358106, "grad_norm": 0.9831956134618856, "learning_rate": 9.124289437688436e-06, "loss": 0.3518, "step": 6236 }, { "epoch": 0.4311787072243346, "grad_norm": 0.753036076365689, "learning_rate": 9.123972957491484e-06, "loss": 0.4199, "step": 6237 }, { "epoch": 0.4312478396128586, "grad_norm": 0.7211188556194484, "learning_rate": 9.123656425607888e-06, "loss": 0.3752, "step": 6238 }, { "epoch": 0.43131697200138264, "grad_norm": 0.785547392591652, "learning_rate": 9.123339842041619e-06, "loss": 0.3546, "step": 6239 }, { "epoch": 0.43138610438990665, "grad_norm": 0.7916889598919076, "learning_rate": 9.123023206796643e-06, "loss": 0.4238, "step": 6240 }, { "epoch": 0.4314552367784307, "grad_norm": 0.8084585295652165, "learning_rate": 9.122706519876929e-06, "loss": 0.3633, "step": 6241 }, { "epoch": 0.4315243691669547, "grad_norm": 0.7351821655411858, "learning_rate": 9.122389781286447e-06, "loss": 0.3174, "step": 6242 }, { "epoch": 0.43159350155547876, "grad_norm": 0.7892668699634419, "learning_rate": 9.122072991029165e-06, "loss": 0.3758, "step": 6243 }, { "epoch": 0.4316626339440028, "grad_norm": 1.5644455035582099, "learning_rate": 9.121756149109055e-06, "loss": 0.3866, "step": 6244 }, { "epoch": 0.4317317663325268, "grad_norm": 1.024505616074995, "learning_rate": 9.121439255530086e-06, "loss": 0.3176, "step": 6245 }, { "epoch": 0.4318008987210508, "grad_norm": 0.9082056725898605, "learning_rate": 9.121122310296232e-06, "loss": 0.2855, "step": 6246 }, { "epoch": 0.43187003110957484, "grad_norm": 1.053355345039126, "learning_rate": 9.120805313411465e-06, "loss": 0.377, "step": 6247 }, { "epoch": 0.43193916349809885, "grad_norm": 0.7154273760610006, "learning_rate": 9.120488264879755e-06, "loss": 0.3381, "step": 6248 }, { "epoch": 0.4320082958866229, "grad_norm": 0.7047517705119725, "learning_rate": 9.120171164705079e-06, "loss": 0.3687, "step": 6249 }, { "epoch": 0.4320774282751469, "grad_norm": 0.6788688899675228, "learning_rate": 9.119854012891411e-06, "loss": 0.3171, "step": 6250 }, { "epoch": 0.4321465606636709, "grad_norm": 0.808972564453711, "learning_rate": 9.119536809442724e-06, "loss": 0.3819, "step": 6251 }, { "epoch": 0.4322156930521949, "grad_norm": 0.6740245307949667, "learning_rate": 9.119219554362995e-06, "loss": 0.3278, "step": 6252 }, { "epoch": 0.432284825440719, "grad_norm": 1.1658151821160514, "learning_rate": 9.1189022476562e-06, "loss": 0.2734, "step": 6253 }, { "epoch": 0.432353957829243, "grad_norm": 0.7256222021822732, "learning_rate": 9.118584889326318e-06, "loss": 0.3721, "step": 6254 }, { "epoch": 0.43242309021776704, "grad_norm": 0.7571190604659306, "learning_rate": 9.118267479377323e-06, "loss": 0.3436, "step": 6255 }, { "epoch": 0.43249222260629105, "grad_norm": 0.77309567358065, "learning_rate": 9.117950017813192e-06, "loss": 0.3369, "step": 6256 }, { "epoch": 0.4325613549948151, "grad_norm": 0.7894486510479555, "learning_rate": 9.117632504637908e-06, "loss": 0.3176, "step": 6257 }, { "epoch": 0.4326304873833391, "grad_norm": 0.8510682970129985, "learning_rate": 9.11731493985545e-06, "loss": 0.4209, "step": 6258 }, { "epoch": 0.4326996197718631, "grad_norm": 0.8500308675746993, "learning_rate": 9.116997323469796e-06, "loss": 0.347, "step": 6259 }, { "epoch": 0.4327687521603871, "grad_norm": 0.8167178601933145, "learning_rate": 9.116679655484928e-06, "loss": 0.317, "step": 6260 }, { "epoch": 0.43283788454891114, "grad_norm": 0.7732119845203255, "learning_rate": 9.116361935904828e-06, "loss": 0.4183, "step": 6261 }, { "epoch": 0.43290701693743516, "grad_norm": 0.7560444163923359, "learning_rate": 9.116044164733476e-06, "loss": 0.383, "step": 6262 }, { "epoch": 0.43297614932595924, "grad_norm": 0.8328421008559289, "learning_rate": 9.115726341974854e-06, "loss": 0.3301, "step": 6263 }, { "epoch": 0.43304528171448325, "grad_norm": 0.7640128768890181, "learning_rate": 9.115408467632949e-06, "loss": 0.2657, "step": 6264 }, { "epoch": 0.4331144141030073, "grad_norm": 0.7272900028924361, "learning_rate": 9.115090541711744e-06, "loss": 0.3936, "step": 6265 }, { "epoch": 0.4331835464915313, "grad_norm": 0.7722034357597208, "learning_rate": 9.11477256421522e-06, "loss": 0.3105, "step": 6266 }, { "epoch": 0.4332526788800553, "grad_norm": 0.6999348478076836, "learning_rate": 9.114454535147367e-06, "loss": 0.2513, "step": 6267 }, { "epoch": 0.4333218112685793, "grad_norm": 0.7601113017408531, "learning_rate": 9.114136454512169e-06, "loss": 0.3566, "step": 6268 }, { "epoch": 0.43339094365710334, "grad_norm": 0.8615551934864862, "learning_rate": 9.113818322313612e-06, "loss": 0.427, "step": 6269 }, { "epoch": 0.43346007604562736, "grad_norm": 0.6406830571467785, "learning_rate": 9.113500138555684e-06, "loss": 0.3247, "step": 6270 }, { "epoch": 0.4335292084341514, "grad_norm": 0.7372137090357258, "learning_rate": 9.113181903242375e-06, "loss": 0.3283, "step": 6271 }, { "epoch": 0.4335983408226754, "grad_norm": 0.9774760223691795, "learning_rate": 9.112863616377667e-06, "loss": 0.4332, "step": 6272 }, { "epoch": 0.4336674732111995, "grad_norm": 0.7289672619152411, "learning_rate": 9.112545277965555e-06, "loss": 0.4061, "step": 6273 }, { "epoch": 0.4337366055997235, "grad_norm": 0.897582315517491, "learning_rate": 9.112226888010028e-06, "loss": 0.2856, "step": 6274 }, { "epoch": 0.4338057379882475, "grad_norm": 0.6844074989798243, "learning_rate": 9.111908446515073e-06, "loss": 0.3195, "step": 6275 }, { "epoch": 0.4338748703767715, "grad_norm": 0.7883906460764908, "learning_rate": 9.111589953484688e-06, "loss": 0.4391, "step": 6276 }, { "epoch": 0.43394400276529554, "grad_norm": 0.8083237136655157, "learning_rate": 9.111271408922858e-06, "loss": 0.3746, "step": 6277 }, { "epoch": 0.43401313515381956, "grad_norm": 0.7757394228622982, "learning_rate": 9.110952812833577e-06, "loss": 0.339, "step": 6278 }, { "epoch": 0.4340822675423436, "grad_norm": 0.728829494598322, "learning_rate": 9.110634165220839e-06, "loss": 0.3379, "step": 6279 }, { "epoch": 0.4341513999308676, "grad_norm": 1.1497167670575381, "learning_rate": 9.110315466088638e-06, "loss": 0.3493, "step": 6280 }, { "epoch": 0.4342205323193916, "grad_norm": 0.8294764499411291, "learning_rate": 9.109996715440968e-06, "loss": 0.2966, "step": 6281 }, { "epoch": 0.43428966470791563, "grad_norm": 0.6871994149801903, "learning_rate": 9.109677913281824e-06, "loss": 0.3519, "step": 6282 }, { "epoch": 0.4343587970964397, "grad_norm": 0.7447181999815176, "learning_rate": 9.109359059615201e-06, "loss": 0.4325, "step": 6283 }, { "epoch": 0.4344279294849637, "grad_norm": 0.7820226959353582, "learning_rate": 9.109040154445094e-06, "loss": 0.2606, "step": 6284 }, { "epoch": 0.43449706187348774, "grad_norm": 1.017243881167651, "learning_rate": 9.108721197775504e-06, "loss": 0.3769, "step": 6285 }, { "epoch": 0.43456619426201176, "grad_norm": 0.7311864295577012, "learning_rate": 9.108402189610425e-06, "loss": 0.3558, "step": 6286 }, { "epoch": 0.4346353266505358, "grad_norm": 0.918595905297802, "learning_rate": 9.108083129953857e-06, "loss": 0.4737, "step": 6287 }, { "epoch": 0.4347044590390598, "grad_norm": 0.8151323393865477, "learning_rate": 9.107764018809799e-06, "loss": 0.2862, "step": 6288 }, { "epoch": 0.4347735914275838, "grad_norm": 1.1332166988610048, "learning_rate": 9.107444856182249e-06, "loss": 0.3271, "step": 6289 }, { "epoch": 0.43484272381610783, "grad_norm": 0.7683802648568232, "learning_rate": 9.107125642075206e-06, "loss": 0.4064, "step": 6290 }, { "epoch": 0.43491185620463185, "grad_norm": 0.6583524475071062, "learning_rate": 9.106806376492674e-06, "loss": 0.3573, "step": 6291 }, { "epoch": 0.43498098859315587, "grad_norm": 0.8233623911145699, "learning_rate": 9.106487059438653e-06, "loss": 0.3053, "step": 6292 }, { "epoch": 0.43505012098167994, "grad_norm": 0.770633677088767, "learning_rate": 9.106167690917144e-06, "loss": 0.4101, "step": 6293 }, { "epoch": 0.43511925337020396, "grad_norm": 1.0017523679913094, "learning_rate": 9.105848270932151e-06, "loss": 0.4489, "step": 6294 }, { "epoch": 0.435188385758728, "grad_norm": 0.7383900917681399, "learning_rate": 9.10552879948768e-06, "loss": 0.3443, "step": 6295 }, { "epoch": 0.435257518147252, "grad_norm": 0.6868612334012838, "learning_rate": 9.10520927658773e-06, "loss": 0.2785, "step": 6296 }, { "epoch": 0.435326650535776, "grad_norm": 0.742028098223047, "learning_rate": 9.104889702236307e-06, "loss": 0.4379, "step": 6297 }, { "epoch": 0.43539578292430003, "grad_norm": 0.706276558793016, "learning_rate": 9.104570076437417e-06, "loss": 0.3412, "step": 6298 }, { "epoch": 0.43546491531282405, "grad_norm": 0.6859746805728167, "learning_rate": 9.104250399195068e-06, "loss": 0.3003, "step": 6299 }, { "epoch": 0.43553404770134807, "grad_norm": 0.7366562920623303, "learning_rate": 9.103930670513263e-06, "loss": 0.3277, "step": 6300 }, { "epoch": 0.4356031800898721, "grad_norm": 0.7454223632051964, "learning_rate": 9.103610890396012e-06, "loss": 0.3995, "step": 6301 }, { "epoch": 0.4356723124783961, "grad_norm": 0.7170485505412527, "learning_rate": 9.103291058847321e-06, "loss": 0.3257, "step": 6302 }, { "epoch": 0.4357414448669201, "grad_norm": 0.6777109567612473, "learning_rate": 9.102971175871199e-06, "loss": 0.2897, "step": 6303 }, { "epoch": 0.4358105772554442, "grad_norm": 0.7178143561581346, "learning_rate": 9.102651241471656e-06, "loss": 0.4346, "step": 6304 }, { "epoch": 0.4358797096439682, "grad_norm": 0.7033418029190651, "learning_rate": 9.102331255652701e-06, "loss": 0.3639, "step": 6305 }, { "epoch": 0.43594884203249223, "grad_norm": 0.6828541329577417, "learning_rate": 9.102011218418345e-06, "loss": 0.2819, "step": 6306 }, { "epoch": 0.43601797442101625, "grad_norm": 0.9243918391078273, "learning_rate": 9.101691129772598e-06, "loss": 0.3398, "step": 6307 }, { "epoch": 0.43608710680954027, "grad_norm": 0.6819027269701925, "learning_rate": 9.101370989719473e-06, "loss": 0.4149, "step": 6308 }, { "epoch": 0.4361562391980643, "grad_norm": 0.7203042245657196, "learning_rate": 9.101050798262982e-06, "loss": 0.3112, "step": 6309 }, { "epoch": 0.4362253715865883, "grad_norm": 0.7443089574391629, "learning_rate": 9.100730555407136e-06, "loss": 0.2786, "step": 6310 }, { "epoch": 0.4362945039751123, "grad_norm": 0.7319550052829698, "learning_rate": 9.100410261155952e-06, "loss": 0.4261, "step": 6311 }, { "epoch": 0.43636363636363634, "grad_norm": 0.7443385142097992, "learning_rate": 9.100089915513443e-06, "loss": 0.3919, "step": 6312 }, { "epoch": 0.43643276875216036, "grad_norm": 0.7447579462625348, "learning_rate": 9.099769518483625e-06, "loss": 0.3326, "step": 6313 }, { "epoch": 0.43650190114068443, "grad_norm": 0.7811349607154083, "learning_rate": 9.099449070070511e-06, "loss": 0.3295, "step": 6314 }, { "epoch": 0.43657103352920845, "grad_norm": 0.7216509458121446, "learning_rate": 9.099128570278119e-06, "loss": 0.4656, "step": 6315 }, { "epoch": 0.43664016591773247, "grad_norm": 0.7232969057275282, "learning_rate": 9.098808019110466e-06, "loss": 0.3602, "step": 6316 }, { "epoch": 0.4367092983062565, "grad_norm": 0.7004722316679791, "learning_rate": 9.098487416571568e-06, "loss": 0.3366, "step": 6317 }, { "epoch": 0.4367784306947805, "grad_norm": 0.8392146258120159, "learning_rate": 9.098166762665446e-06, "loss": 0.4986, "step": 6318 }, { "epoch": 0.4368475630833045, "grad_norm": 0.7735470099753496, "learning_rate": 9.097846057396115e-06, "loss": 0.3716, "step": 6319 }, { "epoch": 0.43691669547182854, "grad_norm": 0.6776538487004155, "learning_rate": 9.097525300767599e-06, "loss": 0.3349, "step": 6320 }, { "epoch": 0.43698582786035256, "grad_norm": 0.6812636879674085, "learning_rate": 9.097204492783913e-06, "loss": 0.2813, "step": 6321 }, { "epoch": 0.4370549602488766, "grad_norm": 0.8048490671020712, "learning_rate": 9.096883633449083e-06, "loss": 0.3952, "step": 6322 }, { "epoch": 0.4371240926374006, "grad_norm": 0.6762657232271823, "learning_rate": 9.096562722767127e-06, "loss": 0.3428, "step": 6323 }, { "epoch": 0.43719322502592467, "grad_norm": 0.6843629830765933, "learning_rate": 9.096241760742067e-06, "loss": 0.287, "step": 6324 }, { "epoch": 0.4372623574144487, "grad_norm": 0.6975468238709432, "learning_rate": 9.095920747377927e-06, "loss": 0.3508, "step": 6325 }, { "epoch": 0.4373314898029727, "grad_norm": 0.7645563672797856, "learning_rate": 9.09559968267873e-06, "loss": 0.4184, "step": 6326 }, { "epoch": 0.4374006221914967, "grad_norm": 0.9844022188742496, "learning_rate": 9.095278566648501e-06, "loss": 0.4019, "step": 6327 }, { "epoch": 0.43746975458002074, "grad_norm": 0.7252839490061997, "learning_rate": 9.094957399291263e-06, "loss": 0.3413, "step": 6328 }, { "epoch": 0.43753888696854476, "grad_norm": 0.7534086311098024, "learning_rate": 9.09463618061104e-06, "loss": 0.4376, "step": 6329 }, { "epoch": 0.4376080193570688, "grad_norm": 0.7226140042147978, "learning_rate": 9.094314910611862e-06, "loss": 0.3797, "step": 6330 }, { "epoch": 0.4376771517455928, "grad_norm": 0.6930177262004226, "learning_rate": 9.093993589297752e-06, "loss": 0.2605, "step": 6331 }, { "epoch": 0.4377462841341168, "grad_norm": 0.7933772645427984, "learning_rate": 9.09367221667274e-06, "loss": 0.329, "step": 6332 }, { "epoch": 0.43781541652264083, "grad_norm": 1.150280258556767, "learning_rate": 9.093350792740851e-06, "loss": 0.463, "step": 6333 }, { "epoch": 0.4378845489111649, "grad_norm": 0.7783965294631281, "learning_rate": 9.093029317506114e-06, "loss": 0.3702, "step": 6334 }, { "epoch": 0.4379536812996889, "grad_norm": 0.8027594479933465, "learning_rate": 9.092707790972559e-06, "loss": 0.2944, "step": 6335 }, { "epoch": 0.43802281368821294, "grad_norm": 0.9954175639043655, "learning_rate": 9.092386213144218e-06, "loss": 0.3764, "step": 6336 }, { "epoch": 0.43809194607673696, "grad_norm": 0.7504244204275918, "learning_rate": 9.092064584025116e-06, "loss": 0.4149, "step": 6337 }, { "epoch": 0.438161078465261, "grad_norm": 0.6865421293525382, "learning_rate": 9.09174290361929e-06, "loss": 0.2848, "step": 6338 }, { "epoch": 0.438230210853785, "grad_norm": 0.8317441088777484, "learning_rate": 9.091421171930766e-06, "loss": 0.3315, "step": 6339 }, { "epoch": 0.438299343242309, "grad_norm": 1.1051592131651888, "learning_rate": 9.09109938896358e-06, "loss": 0.4582, "step": 6340 }, { "epoch": 0.43836847563083303, "grad_norm": 0.7522670399665192, "learning_rate": 9.090777554721764e-06, "loss": 0.3613, "step": 6341 }, { "epoch": 0.43843760801935705, "grad_norm": 0.9336556941942286, "learning_rate": 9.090455669209353e-06, "loss": 0.3507, "step": 6342 }, { "epoch": 0.43850674040788107, "grad_norm": 0.8981236351721957, "learning_rate": 9.090133732430377e-06, "loss": 0.3306, "step": 6343 }, { "epoch": 0.43857587279640514, "grad_norm": 0.7841453606617801, "learning_rate": 9.089811744388877e-06, "loss": 0.4932, "step": 6344 }, { "epoch": 0.43864500518492916, "grad_norm": 1.0145723522033432, "learning_rate": 9.089489705088884e-06, "loss": 0.2957, "step": 6345 }, { "epoch": 0.4387141375734532, "grad_norm": 0.6795331869713894, "learning_rate": 9.089167614534435e-06, "loss": 0.3076, "step": 6346 }, { "epoch": 0.4387832699619772, "grad_norm": 1.2973513857902303, "learning_rate": 9.088845472729569e-06, "loss": 0.3845, "step": 6347 }, { "epoch": 0.4388524023505012, "grad_norm": 0.8605020258784145, "learning_rate": 9.088523279678322e-06, "loss": 0.3406, "step": 6348 }, { "epoch": 0.43892153473902523, "grad_norm": 0.8680958599193472, "learning_rate": 9.08820103538473e-06, "loss": 0.3127, "step": 6349 }, { "epoch": 0.43899066712754925, "grad_norm": 0.6926300587218199, "learning_rate": 9.087878739852833e-06, "loss": 0.254, "step": 6350 }, { "epoch": 0.43905979951607327, "grad_norm": 0.7490047184628374, "learning_rate": 9.087556393086673e-06, "loss": 0.4774, "step": 6351 }, { "epoch": 0.4391289319045973, "grad_norm": 1.0462714205110721, "learning_rate": 9.087233995090288e-06, "loss": 0.2721, "step": 6352 }, { "epoch": 0.4391980642931213, "grad_norm": 0.738216686463042, "learning_rate": 9.086911545867718e-06, "loss": 0.3091, "step": 6353 }, { "epoch": 0.4392671966816454, "grad_norm": 0.7171576592094127, "learning_rate": 9.086589045423006e-06, "loss": 0.4219, "step": 6354 }, { "epoch": 0.4393363290701694, "grad_norm": 0.7497292557298714, "learning_rate": 9.086266493760194e-06, "loss": 0.3345, "step": 6355 }, { "epoch": 0.4394054614586934, "grad_norm": 1.0039504351827313, "learning_rate": 9.085943890883323e-06, "loss": 0.3145, "step": 6356 }, { "epoch": 0.43947459384721743, "grad_norm": 0.7373199428634287, "learning_rate": 9.085621236796437e-06, "loss": 0.3296, "step": 6357 }, { "epoch": 0.43954372623574145, "grad_norm": 0.7607741630428557, "learning_rate": 9.08529853150358e-06, "loss": 0.4092, "step": 6358 }, { "epoch": 0.43961285862426547, "grad_norm": 0.7237930223808717, "learning_rate": 9.084975775008796e-06, "loss": 0.3511, "step": 6359 }, { "epoch": 0.4396819910127895, "grad_norm": 0.7437123205968591, "learning_rate": 9.08465296731613e-06, "loss": 0.3184, "step": 6360 }, { "epoch": 0.4397511234013135, "grad_norm": 0.7291882528221826, "learning_rate": 9.08433010842963e-06, "loss": 0.3564, "step": 6361 }, { "epoch": 0.4398202557898375, "grad_norm": 0.7044062161833476, "learning_rate": 9.08400719835334e-06, "loss": 0.374, "step": 6362 }, { "epoch": 0.43988938817836154, "grad_norm": 0.759301730304605, "learning_rate": 9.083684237091309e-06, "loss": 0.2714, "step": 6363 }, { "epoch": 0.4399585205668856, "grad_norm": 0.7916612971347035, "learning_rate": 9.083361224647581e-06, "loss": 0.3062, "step": 6364 }, { "epoch": 0.44002765295540963, "grad_norm": 0.7510308734898524, "learning_rate": 9.08303816102621e-06, "loss": 0.4109, "step": 6365 }, { "epoch": 0.44009678534393365, "grad_norm": 0.6721772111879691, "learning_rate": 9.082715046231243e-06, "loss": 0.333, "step": 6366 }, { "epoch": 0.44016591773245767, "grad_norm": 0.691889493062469, "learning_rate": 9.082391880266725e-06, "loss": 0.3606, "step": 6367 }, { "epoch": 0.4402350501209817, "grad_norm": 1.0288452597210458, "learning_rate": 9.082068663136714e-06, "loss": 0.3146, "step": 6368 }, { "epoch": 0.4403041825095057, "grad_norm": 0.7423159196656436, "learning_rate": 9.081745394845253e-06, "loss": 0.4514, "step": 6369 }, { "epoch": 0.4403733148980297, "grad_norm": 0.7131115080533925, "learning_rate": 9.081422075396403e-06, "loss": 0.332, "step": 6370 }, { "epoch": 0.44044244728655374, "grad_norm": 0.8587651792841162, "learning_rate": 9.081098704794207e-06, "loss": 0.3587, "step": 6371 }, { "epoch": 0.44051157967507776, "grad_norm": 0.7232372564925491, "learning_rate": 9.080775283042722e-06, "loss": 0.4439, "step": 6372 }, { "epoch": 0.4405807120636018, "grad_norm": 0.8542365797584955, "learning_rate": 9.080451810146002e-06, "loss": 0.3428, "step": 6373 }, { "epoch": 0.4406498444521258, "grad_norm": 0.7871721565721626, "learning_rate": 9.0801282861081e-06, "loss": 0.3189, "step": 6374 }, { "epoch": 0.44071897684064987, "grad_norm": 0.7992232689716072, "learning_rate": 9.079804710933074e-06, "loss": 0.3358, "step": 6375 }, { "epoch": 0.4407881092291739, "grad_norm": 0.6490101102236772, "learning_rate": 9.079481084624973e-06, "loss": 0.4264, "step": 6376 }, { "epoch": 0.4408572416176979, "grad_norm": 0.7734647830829803, "learning_rate": 9.079157407187859e-06, "loss": 0.319, "step": 6377 }, { "epoch": 0.4409263740062219, "grad_norm": 0.6795353312709004, "learning_rate": 9.078833678625784e-06, "loss": 0.2779, "step": 6378 }, { "epoch": 0.44099550639474594, "grad_norm": 0.7444548861045297, "learning_rate": 9.078509898942812e-06, "loss": 0.3502, "step": 6379 }, { "epoch": 0.44106463878326996, "grad_norm": 0.7119834943843911, "learning_rate": 9.078186068142993e-06, "loss": 0.3903, "step": 6380 }, { "epoch": 0.441133771171794, "grad_norm": 0.7222098453236276, "learning_rate": 9.077862186230391e-06, "loss": 0.3119, "step": 6381 }, { "epoch": 0.441202903560318, "grad_norm": 0.7738078634387773, "learning_rate": 9.077538253209064e-06, "loss": 0.3813, "step": 6382 }, { "epoch": 0.441272035948842, "grad_norm": 0.8624588129906142, "learning_rate": 9.07721426908307e-06, "loss": 0.4624, "step": 6383 }, { "epoch": 0.44134116833736603, "grad_norm": 0.7598443115095123, "learning_rate": 9.076890233856473e-06, "loss": 0.3598, "step": 6384 }, { "epoch": 0.4414103007258901, "grad_norm": 0.7281082580106383, "learning_rate": 9.076566147533332e-06, "loss": 0.2748, "step": 6385 }, { "epoch": 0.4414794331144141, "grad_norm": 0.7281439937199069, "learning_rate": 9.07624201011771e-06, "loss": 0.3578, "step": 6386 }, { "epoch": 0.44154856550293814, "grad_norm": 0.7125093913204783, "learning_rate": 9.075917821613669e-06, "loss": 0.4013, "step": 6387 }, { "epoch": 0.44161769789146216, "grad_norm": 0.7052451940530139, "learning_rate": 9.07559358202527e-06, "loss": 0.3031, "step": 6388 }, { "epoch": 0.4416868302799862, "grad_norm": 0.7316874794536862, "learning_rate": 9.075269291356579e-06, "loss": 0.2902, "step": 6389 }, { "epoch": 0.4417559626685102, "grad_norm": 0.759629485240293, "learning_rate": 9.074944949611662e-06, "loss": 0.4204, "step": 6390 }, { "epoch": 0.4418250950570342, "grad_norm": 0.6748728806277109, "learning_rate": 9.07462055679458e-06, "loss": 0.3587, "step": 6391 }, { "epoch": 0.44189422744555823, "grad_norm": 0.6817088695820429, "learning_rate": 9.074296112909402e-06, "loss": 0.3118, "step": 6392 }, { "epoch": 0.44196335983408225, "grad_norm": 0.6788146008996275, "learning_rate": 9.073971617960194e-06, "loss": 0.3402, "step": 6393 }, { "epoch": 0.44203249222260627, "grad_norm": 0.8877565575264503, "learning_rate": 9.07364707195102e-06, "loss": 0.4393, "step": 6394 }, { "epoch": 0.44210162461113034, "grad_norm": 0.7706051854553607, "learning_rate": 9.07332247488595e-06, "loss": 0.3712, "step": 6395 }, { "epoch": 0.44217075699965436, "grad_norm": 0.6746767037061019, "learning_rate": 9.072997826769052e-06, "loss": 0.3128, "step": 6396 }, { "epoch": 0.4422398893881784, "grad_norm": 1.0532618288976165, "learning_rate": 9.072673127604395e-06, "loss": 0.4145, "step": 6397 }, { "epoch": 0.4423090217767024, "grad_norm": 0.6815122783651234, "learning_rate": 9.072348377396049e-06, "loss": 0.3489, "step": 6398 }, { "epoch": 0.4423781541652264, "grad_norm": 0.6630603941894065, "learning_rate": 9.072023576148084e-06, "loss": 0.335, "step": 6399 }, { "epoch": 0.44244728655375043, "grad_norm": 0.731221972568181, "learning_rate": 9.071698723864571e-06, "loss": 0.3535, "step": 6400 }, { "epoch": 0.44251641894227445, "grad_norm": 0.7219710060752114, "learning_rate": 9.071373820549578e-06, "loss": 0.4807, "step": 6401 }, { "epoch": 0.44258555133079847, "grad_norm": 0.7116043739664323, "learning_rate": 9.071048866207181e-06, "loss": 0.3117, "step": 6402 }, { "epoch": 0.4426546837193225, "grad_norm": 0.6811470922061815, "learning_rate": 9.070723860841452e-06, "loss": 0.2628, "step": 6403 }, { "epoch": 0.4427238161078465, "grad_norm": 0.7476578595783268, "learning_rate": 9.070398804456463e-06, "loss": 0.4367, "step": 6404 }, { "epoch": 0.4427929484963706, "grad_norm": 0.6742049002061088, "learning_rate": 9.07007369705629e-06, "loss": 0.3713, "step": 6405 }, { "epoch": 0.4428620808848946, "grad_norm": 0.6882774900199958, "learning_rate": 9.069748538645007e-06, "loss": 0.3404, "step": 6406 }, { "epoch": 0.4429312132734186, "grad_norm": 0.8091118393553589, "learning_rate": 9.069423329226688e-06, "loss": 0.334, "step": 6407 }, { "epoch": 0.44300034566194263, "grad_norm": 0.7581613472352232, "learning_rate": 9.06909806880541e-06, "loss": 0.4437, "step": 6408 }, { "epoch": 0.44306947805046665, "grad_norm": 0.7367929861978443, "learning_rate": 9.068772757385249e-06, "loss": 0.368, "step": 6409 }, { "epoch": 0.44313861043899067, "grad_norm": 1.4007885358435657, "learning_rate": 9.068447394970284e-06, "loss": 0.3091, "step": 6410 }, { "epoch": 0.4432077428275147, "grad_norm": 0.7358491081919103, "learning_rate": 9.068121981564588e-06, "loss": 0.38, "step": 6411 }, { "epoch": 0.4432768752160387, "grad_norm": 0.7016245766943666, "learning_rate": 9.067796517172246e-06, "loss": 0.3733, "step": 6412 }, { "epoch": 0.4433460076045627, "grad_norm": 0.8017265535918666, "learning_rate": 9.067471001797333e-06, "loss": 0.2826, "step": 6413 }, { "epoch": 0.44341513999308674, "grad_norm": 0.6760131773065283, "learning_rate": 9.067145435443929e-06, "loss": 0.269, "step": 6414 }, { "epoch": 0.4434842723816108, "grad_norm": 0.7786249807741323, "learning_rate": 9.066819818116116e-06, "loss": 0.3885, "step": 6415 }, { "epoch": 0.44355340477013483, "grad_norm": 0.6917370402303012, "learning_rate": 9.066494149817974e-06, "loss": 0.2607, "step": 6416 }, { "epoch": 0.44362253715865885, "grad_norm": 0.7123056846735374, "learning_rate": 9.066168430553587e-06, "loss": 0.3264, "step": 6417 }, { "epoch": 0.44369166954718287, "grad_norm": 0.6937922493830451, "learning_rate": 9.065842660327032e-06, "loss": 0.3544, "step": 6418 }, { "epoch": 0.4437608019357069, "grad_norm": 0.6995304622662076, "learning_rate": 9.065516839142393e-06, "loss": 0.3658, "step": 6419 }, { "epoch": 0.4438299343242309, "grad_norm": 0.98339354440512, "learning_rate": 9.06519096700376e-06, "loss": 0.296, "step": 6420 }, { "epoch": 0.4438990667127549, "grad_norm": 0.8388669400528206, "learning_rate": 9.064865043915212e-06, "loss": 0.2916, "step": 6421 }, { "epoch": 0.44396819910127894, "grad_norm": 0.7224371756958081, "learning_rate": 9.064539069880834e-06, "loss": 0.4696, "step": 6422 }, { "epoch": 0.44403733148980296, "grad_norm": 0.7749428846654827, "learning_rate": 9.064213044904712e-06, "loss": 0.439, "step": 6423 }, { "epoch": 0.444106463878327, "grad_norm": 0.6548725583392947, "learning_rate": 9.063886968990934e-06, "loss": 0.2796, "step": 6424 }, { "epoch": 0.44417559626685105, "grad_norm": 0.7552320368919115, "learning_rate": 9.063560842143583e-06, "loss": 0.3951, "step": 6425 }, { "epoch": 0.44424472865537507, "grad_norm": 0.7707029955826203, "learning_rate": 9.06323466436675e-06, "loss": 0.4234, "step": 6426 }, { "epoch": 0.4443138610438991, "grad_norm": 0.687426056502676, "learning_rate": 9.062908435664521e-06, "loss": 0.3185, "step": 6427 }, { "epoch": 0.4443829934324231, "grad_norm": 0.730196746310413, "learning_rate": 9.062582156040987e-06, "loss": 0.2611, "step": 6428 }, { "epoch": 0.4444521258209471, "grad_norm": 1.0923106678254404, "learning_rate": 9.062255825500233e-06, "loss": 0.4236, "step": 6429 }, { "epoch": 0.44452125820947114, "grad_norm": 0.6851689136222986, "learning_rate": 9.061929444046353e-06, "loss": 0.3231, "step": 6430 }, { "epoch": 0.44459039059799516, "grad_norm": 0.7040894475066763, "learning_rate": 9.061603011683437e-06, "loss": 0.3049, "step": 6431 }, { "epoch": 0.4446595229865192, "grad_norm": 0.7655758786412123, "learning_rate": 9.061276528415575e-06, "loss": 0.328, "step": 6432 }, { "epoch": 0.4447286553750432, "grad_norm": 0.7300735352712124, "learning_rate": 9.060949994246859e-06, "loss": 0.3917, "step": 6433 }, { "epoch": 0.4447977877635672, "grad_norm": 0.7372463463034158, "learning_rate": 9.060623409181382e-06, "loss": 0.3366, "step": 6434 }, { "epoch": 0.4448669201520912, "grad_norm": 0.6764797644228505, "learning_rate": 9.060296773223238e-06, "loss": 0.3137, "step": 6435 }, { "epoch": 0.4449360525406153, "grad_norm": 0.7414254000251076, "learning_rate": 9.059970086376518e-06, "loss": 0.3875, "step": 6436 }, { "epoch": 0.4450051849291393, "grad_norm": 0.8291299892883327, "learning_rate": 9.05964334864532e-06, "loss": 0.4708, "step": 6437 }, { "epoch": 0.44507431731766334, "grad_norm": 0.6886517730870103, "learning_rate": 9.059316560033735e-06, "loss": 0.3021, "step": 6438 }, { "epoch": 0.44514344970618736, "grad_norm": 0.7323577033300378, "learning_rate": 9.058989720545864e-06, "loss": 0.3432, "step": 6439 }, { "epoch": 0.4452125820947114, "grad_norm": 0.7738503114505737, "learning_rate": 9.058662830185799e-06, "loss": 0.4084, "step": 6440 }, { "epoch": 0.4452817144832354, "grad_norm": 0.7739917071703581, "learning_rate": 9.05833588895764e-06, "loss": 0.3291, "step": 6441 }, { "epoch": 0.4453508468717594, "grad_norm": 1.529501289447225, "learning_rate": 9.05800889686548e-06, "loss": 0.2735, "step": 6442 }, { "epoch": 0.4454199792602834, "grad_norm": 0.829821789746477, "learning_rate": 9.057681853913422e-06, "loss": 0.4304, "step": 6443 }, { "epoch": 0.44548911164880745, "grad_norm": 0.6945181541213088, "learning_rate": 9.057354760105564e-06, "loss": 0.4189, "step": 6444 }, { "epoch": 0.44555824403733146, "grad_norm": 0.755107134268622, "learning_rate": 9.057027615446003e-06, "loss": 0.2701, "step": 6445 }, { "epoch": 0.44562737642585554, "grad_norm": 0.7575685586603199, "learning_rate": 9.056700419938843e-06, "loss": 0.2985, "step": 6446 }, { "epoch": 0.44569650881437956, "grad_norm": 1.962279998563031, "learning_rate": 9.056373173588182e-06, "loss": 0.406, "step": 6447 }, { "epoch": 0.4457656412029036, "grad_norm": 0.755758146074145, "learning_rate": 9.056045876398122e-06, "loss": 0.3745, "step": 6448 }, { "epoch": 0.4458347735914276, "grad_norm": 4.5093048516256005, "learning_rate": 9.055718528372765e-06, "loss": 0.2685, "step": 6449 }, { "epoch": 0.4459039059799516, "grad_norm": 0.7654140945555884, "learning_rate": 9.055391129516213e-06, "loss": 0.3196, "step": 6450 }, { "epoch": 0.4459730383684756, "grad_norm": 0.7434091411412779, "learning_rate": 9.055063679832572e-06, "loss": 0.4107, "step": 6451 }, { "epoch": 0.44604217075699965, "grad_norm": 0.6908094209199536, "learning_rate": 9.054736179325945e-06, "loss": 0.307, "step": 6452 }, { "epoch": 0.44611130314552366, "grad_norm": 0.6622496708506481, "learning_rate": 9.054408628000435e-06, "loss": 0.279, "step": 6453 }, { "epoch": 0.4461804355340477, "grad_norm": 0.7437251324539156, "learning_rate": 9.054081025860148e-06, "loss": 0.3984, "step": 6454 }, { "epoch": 0.4462495679225717, "grad_norm": 0.8359668978911545, "learning_rate": 9.05375337290919e-06, "loss": 0.4199, "step": 6455 }, { "epoch": 0.4463187003110958, "grad_norm": 1.2229084149361036, "learning_rate": 9.053425669151668e-06, "loss": 0.3192, "step": 6456 }, { "epoch": 0.4463878326996198, "grad_norm": 0.8428660099376072, "learning_rate": 9.053097914591689e-06, "loss": 0.3549, "step": 6457 }, { "epoch": 0.4464569650881438, "grad_norm": 0.7266422641274135, "learning_rate": 9.052770109233359e-06, "loss": 0.4182, "step": 6458 }, { "epoch": 0.4465260974766678, "grad_norm": 0.7970495448839512, "learning_rate": 9.052442253080791e-06, "loss": 0.346, "step": 6459 }, { "epoch": 0.44659522986519185, "grad_norm": 0.6759571490075683, "learning_rate": 9.05211434613809e-06, "loss": 0.2765, "step": 6460 }, { "epoch": 0.44666436225371586, "grad_norm": 0.7082375515483428, "learning_rate": 9.051786388409367e-06, "loss": 0.3691, "step": 6461 }, { "epoch": 0.4467334946422399, "grad_norm": 0.7500741002227135, "learning_rate": 9.051458379898734e-06, "loss": 0.4591, "step": 6462 }, { "epoch": 0.4468026270307639, "grad_norm": 0.7671341481069345, "learning_rate": 9.051130320610298e-06, "loss": 0.3484, "step": 6463 }, { "epoch": 0.4468717594192879, "grad_norm": 0.8045753324405441, "learning_rate": 9.050802210548173e-06, "loss": 0.3171, "step": 6464 }, { "epoch": 0.44694089180781194, "grad_norm": 0.724728666799281, "learning_rate": 9.050474049716473e-06, "loss": 0.4546, "step": 6465 }, { "epoch": 0.447010024196336, "grad_norm": 0.8059548412358742, "learning_rate": 9.05014583811931e-06, "loss": 0.3562, "step": 6466 }, { "epoch": 0.44707915658486, "grad_norm": 0.8529411258682092, "learning_rate": 9.049817575760792e-06, "loss": 0.3496, "step": 6467 }, { "epoch": 0.44714828897338404, "grad_norm": 0.7149811293820009, "learning_rate": 9.04948926264504e-06, "loss": 0.3313, "step": 6468 }, { "epoch": 0.44721742136190806, "grad_norm": 0.8267871366712396, "learning_rate": 9.049160898776171e-06, "loss": 0.3708, "step": 6469 }, { "epoch": 0.4472865537504321, "grad_norm": 0.7179011173179057, "learning_rate": 9.048832484158291e-06, "loss": 0.3355, "step": 6470 }, { "epoch": 0.4473556861389561, "grad_norm": 0.9524875990959951, "learning_rate": 9.048504018795523e-06, "loss": 0.385, "step": 6471 }, { "epoch": 0.4474248185274801, "grad_norm": 0.7315266518956577, "learning_rate": 9.048175502691983e-06, "loss": 0.4248, "step": 6472 }, { "epoch": 0.44749395091600414, "grad_norm": 0.6904219123035161, "learning_rate": 9.047846935851786e-06, "loss": 0.3491, "step": 6473 }, { "epoch": 0.44756308330452815, "grad_norm": 0.8388436017750341, "learning_rate": 9.047518318279052e-06, "loss": 0.2865, "step": 6474 }, { "epoch": 0.44763221569305217, "grad_norm": 0.7571282977986252, "learning_rate": 9.047189649977898e-06, "loss": 0.3308, "step": 6475 }, { "epoch": 0.44770134808157624, "grad_norm": 0.7011753497824148, "learning_rate": 9.046860930952445e-06, "loss": 0.3599, "step": 6476 }, { "epoch": 0.44777048047010026, "grad_norm": 0.6446370322572555, "learning_rate": 9.046532161206814e-06, "loss": 0.2777, "step": 6477 }, { "epoch": 0.4478396128586243, "grad_norm": 0.6942423393413152, "learning_rate": 9.046203340745122e-06, "loss": 0.3107, "step": 6478 }, { "epoch": 0.4479087452471483, "grad_norm": 0.836022121067939, "learning_rate": 9.045874469571491e-06, "loss": 0.4812, "step": 6479 }, { "epoch": 0.4479778776356723, "grad_norm": 0.6798133308517963, "learning_rate": 9.045545547690045e-06, "loss": 0.3252, "step": 6480 }, { "epoch": 0.44804701002419633, "grad_norm": 0.701743501367178, "learning_rate": 9.045216575104905e-06, "loss": 0.32, "step": 6481 }, { "epoch": 0.44811614241272035, "grad_norm": 0.906960412300934, "learning_rate": 9.044887551820195e-06, "loss": 0.3349, "step": 6482 }, { "epoch": 0.44818527480124437, "grad_norm": 0.7414321461054829, "learning_rate": 9.044558477840035e-06, "loss": 0.4552, "step": 6483 }, { "epoch": 0.4482544071897684, "grad_norm": 0.8708427388673857, "learning_rate": 9.044229353168554e-06, "loss": 0.3146, "step": 6484 }, { "epoch": 0.4483235395782924, "grad_norm": 0.8154076760282252, "learning_rate": 9.043900177809876e-06, "loss": 0.2778, "step": 6485 }, { "epoch": 0.4483926719668165, "grad_norm": 0.6963956328365487, "learning_rate": 9.043570951768126e-06, "loss": 0.3713, "step": 6486 }, { "epoch": 0.4484618043553405, "grad_norm": 0.8042776444210465, "learning_rate": 9.04324167504743e-06, "loss": 0.4131, "step": 6487 }, { "epoch": 0.4485309367438645, "grad_norm": 0.7045906138213173, "learning_rate": 9.042912347651916e-06, "loss": 0.2815, "step": 6488 }, { "epoch": 0.44860006913238853, "grad_norm": 1.0850095087812104, "learning_rate": 9.042582969585709e-06, "loss": 0.3673, "step": 6489 }, { "epoch": 0.44866920152091255, "grad_norm": 0.7186879404659243, "learning_rate": 9.042253540852939e-06, "loss": 0.4773, "step": 6490 }, { "epoch": 0.44873833390943657, "grad_norm": 0.7303832959402395, "learning_rate": 9.041924061457734e-06, "loss": 0.3463, "step": 6491 }, { "epoch": 0.4488074662979606, "grad_norm": 0.7696454163918668, "learning_rate": 9.041594531404226e-06, "loss": 0.3384, "step": 6492 }, { "epoch": 0.4488765986864846, "grad_norm": 1.167604347717613, "learning_rate": 9.041264950696542e-06, "loss": 0.3651, "step": 6493 }, { "epoch": 0.4489457310750086, "grad_norm": 0.7826534294040194, "learning_rate": 9.040935319338813e-06, "loss": 0.3935, "step": 6494 }, { "epoch": 0.44901486346353264, "grad_norm": 0.9525097638534864, "learning_rate": 9.040605637335172e-06, "loss": 0.336, "step": 6495 }, { "epoch": 0.44908399585205666, "grad_norm": 0.841102629468297, "learning_rate": 9.04027590468975e-06, "loss": 0.375, "step": 6496 }, { "epoch": 0.44915312824058073, "grad_norm": 0.8227697736048281, "learning_rate": 9.03994612140668e-06, "loss": 0.4273, "step": 6497 }, { "epoch": 0.44922226062910475, "grad_norm": 1.1809729365011086, "learning_rate": 9.039616287490095e-06, "loss": 0.356, "step": 6498 }, { "epoch": 0.44929139301762877, "grad_norm": 0.7031911611857284, "learning_rate": 9.039286402944127e-06, "loss": 0.3189, "step": 6499 }, { "epoch": 0.4493605254061528, "grad_norm": 0.7937455464739462, "learning_rate": 9.038956467772914e-06, "loss": 0.4163, "step": 6500 }, { "epoch": 0.4494296577946768, "grad_norm": 0.7681742094023551, "learning_rate": 9.03862648198059e-06, "loss": 0.3998, "step": 6501 }, { "epoch": 0.4494987901832008, "grad_norm": 0.6793345034833391, "learning_rate": 9.03829644557129e-06, "loss": 0.3056, "step": 6502 }, { "epoch": 0.44956792257172484, "grad_norm": 0.8924610398854693, "learning_rate": 9.03796635854915e-06, "loss": 0.3669, "step": 6503 }, { "epoch": 0.44963705496024886, "grad_norm": 0.7422864445130399, "learning_rate": 9.037636220918308e-06, "loss": 0.3766, "step": 6504 }, { "epoch": 0.4497061873487729, "grad_norm": 0.9406611346003471, "learning_rate": 9.037306032682904e-06, "loss": 0.3653, "step": 6505 }, { "epoch": 0.4497753197372969, "grad_norm": 0.7177962474783395, "learning_rate": 9.036975793847071e-06, "loss": 0.2895, "step": 6506 }, { "epoch": 0.44984445212582097, "grad_norm": 0.8509926803824487, "learning_rate": 9.036645504414951e-06, "loss": 0.3757, "step": 6507 }, { "epoch": 0.449913584514345, "grad_norm": 0.7315707477617797, "learning_rate": 9.036315164390684e-06, "loss": 0.4613, "step": 6508 }, { "epoch": 0.449982716902869, "grad_norm": 0.7818969633106109, "learning_rate": 9.035984773778408e-06, "loss": 0.3826, "step": 6509 }, { "epoch": 0.450051849291393, "grad_norm": 0.6715780537857218, "learning_rate": 9.035654332582267e-06, "loss": 0.2765, "step": 6510 }, { "epoch": 0.45012098167991704, "grad_norm": 0.7099203287354776, "learning_rate": 9.0353238408064e-06, "loss": 0.3832, "step": 6511 }, { "epoch": 0.45019011406844106, "grad_norm": 0.7390826536706386, "learning_rate": 9.03499329845495e-06, "loss": 0.4489, "step": 6512 }, { "epoch": 0.4502592464569651, "grad_norm": 0.8613777085266563, "learning_rate": 9.034662705532061e-06, "loss": 0.3405, "step": 6513 }, { "epoch": 0.4503283788454891, "grad_norm": 0.7570642347102788, "learning_rate": 9.034332062041875e-06, "loss": 0.3665, "step": 6514 }, { "epoch": 0.4503975112340131, "grad_norm": 0.7390064645083844, "learning_rate": 9.034001367988536e-06, "loss": 0.4511, "step": 6515 }, { "epoch": 0.45046664362253713, "grad_norm": 0.7265170896208021, "learning_rate": 9.033670623376187e-06, "loss": 0.3059, "step": 6516 }, { "epoch": 0.4505357760110612, "grad_norm": 1.0081151586315156, "learning_rate": 9.033339828208979e-06, "loss": 0.3371, "step": 6517 }, { "epoch": 0.4506049083995852, "grad_norm": 1.1628346315375204, "learning_rate": 9.033008982491049e-06, "loss": 0.3782, "step": 6518 }, { "epoch": 0.45067404078810924, "grad_norm": 0.8185816848040828, "learning_rate": 9.032678086226552e-06, "loss": 0.4202, "step": 6519 }, { "epoch": 0.45074317317663326, "grad_norm": 0.7152312975081675, "learning_rate": 9.032347139419631e-06, "loss": 0.363, "step": 6520 }, { "epoch": 0.4508123055651573, "grad_norm": 0.8129962612775543, "learning_rate": 9.032016142074435e-06, "loss": 0.371, "step": 6521 }, { "epoch": 0.4508814379536813, "grad_norm": 1.0735305160294482, "learning_rate": 9.031685094195111e-06, "loss": 0.4854, "step": 6522 }, { "epoch": 0.4509505703422053, "grad_norm": 0.7021986522214357, "learning_rate": 9.03135399578581e-06, "loss": 0.3638, "step": 6523 }, { "epoch": 0.45101970273072933, "grad_norm": 0.7015946834994514, "learning_rate": 9.031022846850682e-06, "loss": 0.3286, "step": 6524 }, { "epoch": 0.45108883511925335, "grad_norm": 0.7405057273975622, "learning_rate": 9.030691647393875e-06, "loss": 0.3681, "step": 6525 }, { "epoch": 0.45115796750777737, "grad_norm": 1.3638021542717647, "learning_rate": 9.030360397419542e-06, "loss": 0.4217, "step": 6526 }, { "epoch": 0.45122709989630144, "grad_norm": 0.8316026674244503, "learning_rate": 9.030029096931832e-06, "loss": 0.3163, "step": 6527 }, { "epoch": 0.45129623228482546, "grad_norm": 0.7589859694481547, "learning_rate": 9.029697745934902e-06, "loss": 0.3162, "step": 6528 }, { "epoch": 0.4513653646733495, "grad_norm": 0.7256887897838246, "learning_rate": 9.029366344432901e-06, "loss": 0.3844, "step": 6529 }, { "epoch": 0.4514344970618735, "grad_norm": 0.7987168861656946, "learning_rate": 9.029034892429982e-06, "loss": 0.4315, "step": 6530 }, { "epoch": 0.4515036294503975, "grad_norm": 0.8187053533196872, "learning_rate": 9.028703389930303e-06, "loss": 0.3783, "step": 6531 }, { "epoch": 0.45157276183892153, "grad_norm": 0.8044869702120293, "learning_rate": 9.028371836938014e-06, "loss": 0.3069, "step": 6532 }, { "epoch": 0.45164189422744555, "grad_norm": 0.9141774665123166, "learning_rate": 9.028040233457276e-06, "loss": 0.4404, "step": 6533 }, { "epoch": 0.45171102661596957, "grad_norm": 0.7161166000401678, "learning_rate": 9.02770857949224e-06, "loss": 0.3515, "step": 6534 }, { "epoch": 0.4517801590044936, "grad_norm": 0.7267651777697216, "learning_rate": 9.027376875047066e-06, "loss": 0.3047, "step": 6535 }, { "epoch": 0.4518492913930176, "grad_norm": 0.7304465162969794, "learning_rate": 9.027045120125911e-06, "loss": 0.4051, "step": 6536 }, { "epoch": 0.4519184237815417, "grad_norm": 0.9387554716760563, "learning_rate": 9.026713314732928e-06, "loss": 0.4542, "step": 6537 }, { "epoch": 0.4519875561700657, "grad_norm": 0.746029334509571, "learning_rate": 9.026381458872283e-06, "loss": 0.3395, "step": 6538 }, { "epoch": 0.4520566885585897, "grad_norm": 0.7804333655453662, "learning_rate": 9.02604955254813e-06, "loss": 0.3414, "step": 6539 }, { "epoch": 0.45212582094711373, "grad_norm": 0.7449963825607343, "learning_rate": 9.025717595764632e-06, "loss": 0.4149, "step": 6540 }, { "epoch": 0.45219495333563775, "grad_norm": 1.2679487774178426, "learning_rate": 9.025385588525948e-06, "loss": 0.3468, "step": 6541 }, { "epoch": 0.45226408572416177, "grad_norm": 0.7358996174337405, "learning_rate": 9.025053530836238e-06, "loss": 0.3263, "step": 6542 }, { "epoch": 0.4523332181126858, "grad_norm": 0.7118978827978174, "learning_rate": 9.024721422699667e-06, "loss": 0.3617, "step": 6543 }, { "epoch": 0.4524023505012098, "grad_norm": 0.700852805319862, "learning_rate": 9.024389264120395e-06, "loss": 0.4661, "step": 6544 }, { "epoch": 0.4524714828897338, "grad_norm": 0.6044741851359254, "learning_rate": 9.024057055102584e-06, "loss": 0.2896, "step": 6545 }, { "epoch": 0.45254061527825784, "grad_norm": 0.6534142109184067, "learning_rate": 9.0237247956504e-06, "loss": 0.3046, "step": 6546 }, { "epoch": 0.4526097476667819, "grad_norm": 0.6722817323407146, "learning_rate": 9.023392485768007e-06, "loss": 0.3487, "step": 6547 }, { "epoch": 0.45267888005530593, "grad_norm": 0.7966658998024406, "learning_rate": 9.023060125459568e-06, "loss": 0.3467, "step": 6548 }, { "epoch": 0.45274801244382995, "grad_norm": 0.6768614222435757, "learning_rate": 9.02272771472925e-06, "loss": 0.3166, "step": 6549 }, { "epoch": 0.45281714483235397, "grad_norm": 0.724036787474197, "learning_rate": 9.02239525358122e-06, "loss": 0.3103, "step": 6550 }, { "epoch": 0.452886277220878, "grad_norm": 0.7319874238101803, "learning_rate": 9.022062742019643e-06, "loss": 0.4318, "step": 6551 }, { "epoch": 0.452955409609402, "grad_norm": 0.9144488631337301, "learning_rate": 9.021730180048687e-06, "loss": 0.4011, "step": 6552 }, { "epoch": 0.453024541997926, "grad_norm": 0.9334790219106278, "learning_rate": 9.021397567672521e-06, "loss": 0.2992, "step": 6553 }, { "epoch": 0.45309367438645004, "grad_norm": 0.8255781884321436, "learning_rate": 9.021064904895312e-06, "loss": 0.3732, "step": 6554 }, { "epoch": 0.45316280677497406, "grad_norm": 0.7125292430477935, "learning_rate": 9.020732191721231e-06, "loss": 0.3685, "step": 6555 }, { "epoch": 0.4532319391634981, "grad_norm": 0.8405561425392769, "learning_rate": 9.020399428154446e-06, "loss": 0.3066, "step": 6556 }, { "epoch": 0.45330107155202215, "grad_norm": 0.7635696603035704, "learning_rate": 9.02006661419913e-06, "loss": 0.2974, "step": 6557 }, { "epoch": 0.45337020394054617, "grad_norm": 0.7507613641124754, "learning_rate": 9.019733749859454e-06, "loss": 0.381, "step": 6558 }, { "epoch": 0.4534393363290702, "grad_norm": 1.0330258126197132, "learning_rate": 9.019400835139589e-06, "loss": 0.3164, "step": 6559 }, { "epoch": 0.4535084687175942, "grad_norm": 0.8259397638845455, "learning_rate": 9.019067870043705e-06, "loss": 0.2967, "step": 6560 }, { "epoch": 0.4535776011061182, "grad_norm": 0.7278840870493722, "learning_rate": 9.018734854575981e-06, "loss": 0.4027, "step": 6561 }, { "epoch": 0.45364673349464224, "grad_norm": 0.8467932894886115, "learning_rate": 9.018401788740585e-06, "loss": 0.3977, "step": 6562 }, { "epoch": 0.45371586588316626, "grad_norm": 0.7763271007881506, "learning_rate": 9.018068672541694e-06, "loss": 0.2742, "step": 6563 }, { "epoch": 0.4537849982716903, "grad_norm": 0.7386991304646083, "learning_rate": 9.017735505983482e-06, "loss": 0.3364, "step": 6564 }, { "epoch": 0.4538541306602143, "grad_norm": 0.7820688598212315, "learning_rate": 9.017402289070127e-06, "loss": 0.3928, "step": 6565 }, { "epoch": 0.4539232630487383, "grad_norm": 0.8989978394871179, "learning_rate": 9.017069021805803e-06, "loss": 0.3502, "step": 6566 }, { "epoch": 0.45399239543726233, "grad_norm": 0.7149390629350485, "learning_rate": 9.016735704194688e-06, "loss": 0.2989, "step": 6567 }, { "epoch": 0.4540615278257864, "grad_norm": 0.8041235560533649, "learning_rate": 9.01640233624096e-06, "loss": 0.372, "step": 6568 }, { "epoch": 0.4541306602143104, "grad_norm": 0.758367844156459, "learning_rate": 9.016068917948794e-06, "loss": 0.4143, "step": 6569 }, { "epoch": 0.45419979260283444, "grad_norm": 0.7247891967228476, "learning_rate": 9.015735449322372e-06, "loss": 0.2869, "step": 6570 }, { "epoch": 0.45426892499135846, "grad_norm": 1.374129933628954, "learning_rate": 9.015401930365873e-06, "loss": 0.3197, "step": 6571 }, { "epoch": 0.4543380573798825, "grad_norm": 0.752785313368202, "learning_rate": 9.015068361083476e-06, "loss": 0.4465, "step": 6572 }, { "epoch": 0.4544071897684065, "grad_norm": 0.6883622241382054, "learning_rate": 9.014734741479364e-06, "loss": 0.3733, "step": 6573 }, { "epoch": 0.4544763221569305, "grad_norm": 0.7046811289708211, "learning_rate": 9.014401071557715e-06, "loss": 0.3537, "step": 6574 }, { "epoch": 0.45454545454545453, "grad_norm": 0.7500046007248972, "learning_rate": 9.014067351322712e-06, "loss": 0.3821, "step": 6575 }, { "epoch": 0.45461458693397855, "grad_norm": 0.681533043519445, "learning_rate": 9.01373358077854e-06, "loss": 0.4474, "step": 6576 }, { "epoch": 0.45468371932250257, "grad_norm": 0.6982743233037316, "learning_rate": 9.01339975992938e-06, "loss": 0.3094, "step": 6577 }, { "epoch": 0.45475285171102664, "grad_norm": 0.7255205545882981, "learning_rate": 9.013065888779416e-06, "loss": 0.2651, "step": 6578 }, { "epoch": 0.45482198409955066, "grad_norm": 0.7333570743431391, "learning_rate": 9.012731967332833e-06, "loss": 0.377, "step": 6579 }, { "epoch": 0.4548911164880747, "grad_norm": 0.7479142232740069, "learning_rate": 9.012397995593814e-06, "loss": 0.4173, "step": 6580 }, { "epoch": 0.4549602488765987, "grad_norm": 0.8716036722407094, "learning_rate": 9.012063973566547e-06, "loss": 0.3403, "step": 6581 }, { "epoch": 0.4550293812651227, "grad_norm": 0.9693601641068784, "learning_rate": 9.011729901255219e-06, "loss": 0.3791, "step": 6582 }, { "epoch": 0.45509851365364673, "grad_norm": 0.743825090738039, "learning_rate": 9.011395778664017e-06, "loss": 0.4231, "step": 6583 }, { "epoch": 0.45516764604217075, "grad_norm": 0.6850129232299437, "learning_rate": 9.011061605797126e-06, "loss": 0.2583, "step": 6584 }, { "epoch": 0.45523677843069477, "grad_norm": 0.6330951061141296, "learning_rate": 9.010727382658737e-06, "loss": 0.2778, "step": 6585 }, { "epoch": 0.4553059108192188, "grad_norm": 1.6474734561770374, "learning_rate": 9.010393109253037e-06, "loss": 0.3665, "step": 6586 }, { "epoch": 0.4553750432077428, "grad_norm": 0.7107933876437771, "learning_rate": 9.010058785584216e-06, "loss": 0.3862, "step": 6587 }, { "epoch": 0.4554441755962669, "grad_norm": 0.6876010886150297, "learning_rate": 9.009724411656464e-06, "loss": 0.3118, "step": 6588 }, { "epoch": 0.4555133079847909, "grad_norm": 0.7518924345957031, "learning_rate": 9.009389987473973e-06, "loss": 0.3314, "step": 6589 }, { "epoch": 0.4555824403733149, "grad_norm": 0.7699727917445518, "learning_rate": 9.009055513040934e-06, "loss": 0.4531, "step": 6590 }, { "epoch": 0.45565157276183893, "grad_norm": 0.7753190723991434, "learning_rate": 9.008720988361537e-06, "loss": 0.3468, "step": 6591 }, { "epoch": 0.45572070515036295, "grad_norm": 0.6953975943310132, "learning_rate": 9.008386413439978e-06, "loss": 0.3315, "step": 6592 }, { "epoch": 0.45578983753888697, "grad_norm": 0.6827720897438949, "learning_rate": 9.008051788280447e-06, "loss": 0.3291, "step": 6593 }, { "epoch": 0.455858969927411, "grad_norm": 0.6931354080110093, "learning_rate": 9.00771711288714e-06, "loss": 0.4162, "step": 6594 }, { "epoch": 0.455928102315935, "grad_norm": 0.6598900842367952, "learning_rate": 9.007382387264252e-06, "loss": 0.2993, "step": 6595 }, { "epoch": 0.455997234704459, "grad_norm": 0.7476537806392741, "learning_rate": 9.007047611415977e-06, "loss": 0.297, "step": 6596 }, { "epoch": 0.45606636709298304, "grad_norm": 0.75480412937777, "learning_rate": 9.006712785346512e-06, "loss": 0.4314, "step": 6597 }, { "epoch": 0.4561354994815071, "grad_norm": 0.7648948310715844, "learning_rate": 9.00637790906005e-06, "loss": 0.4142, "step": 6598 }, { "epoch": 0.45620463187003113, "grad_norm": 0.7214999799495027, "learning_rate": 9.006042982560792e-06, "loss": 0.3357, "step": 6599 }, { "epoch": 0.45627376425855515, "grad_norm": 0.761870493711548, "learning_rate": 9.005708005852933e-06, "loss": 0.3761, "step": 6600 }, { "epoch": 0.45634289664707917, "grad_norm": 0.720630002291415, "learning_rate": 9.005372978940675e-06, "loss": 0.448, "step": 6601 }, { "epoch": 0.4564120290356032, "grad_norm": 0.7692969699282058, "learning_rate": 9.005037901828212e-06, "loss": 0.3517, "step": 6602 }, { "epoch": 0.4564811614241272, "grad_norm": 0.7124386172183975, "learning_rate": 9.004702774519748e-06, "loss": 0.3122, "step": 6603 }, { "epoch": 0.4565502938126512, "grad_norm": 0.7156906639922229, "learning_rate": 9.004367597019483e-06, "loss": 0.4028, "step": 6604 }, { "epoch": 0.45661942620117524, "grad_norm": 0.7382460402729085, "learning_rate": 9.004032369331615e-06, "loss": 0.3361, "step": 6605 }, { "epoch": 0.45668855858969926, "grad_norm": 0.9430143431277589, "learning_rate": 9.003697091460346e-06, "loss": 0.331, "step": 6606 }, { "epoch": 0.4567576909782233, "grad_norm": 0.6472880209321247, "learning_rate": 9.00336176340988e-06, "loss": 0.3231, "step": 6607 }, { "epoch": 0.45682682336674735, "grad_norm": 0.759209112490301, "learning_rate": 9.003026385184417e-06, "loss": 0.4421, "step": 6608 }, { "epoch": 0.45689595575527137, "grad_norm": 0.7394833601330852, "learning_rate": 9.002690956788163e-06, "loss": 0.3922, "step": 6609 }, { "epoch": 0.4569650881437954, "grad_norm": 0.6980749986477941, "learning_rate": 9.00235547822532e-06, "loss": 0.345, "step": 6610 }, { "epoch": 0.4570342205323194, "grad_norm": 0.7025980927287527, "learning_rate": 9.002019949500093e-06, "loss": 0.3751, "step": 6611 }, { "epoch": 0.4571033529208434, "grad_norm": 0.6964973920383373, "learning_rate": 9.00168437061669e-06, "loss": 0.3712, "step": 6612 }, { "epoch": 0.45717248530936744, "grad_norm": 0.6577450858204603, "learning_rate": 9.001348741579315e-06, "loss": 0.301, "step": 6613 }, { "epoch": 0.45724161769789146, "grad_norm": 0.7441134437081487, "learning_rate": 9.001013062392172e-06, "loss": 0.383, "step": 6614 }, { "epoch": 0.4573107500864155, "grad_norm": 0.7266116263088636, "learning_rate": 9.000677333059471e-06, "loss": 0.4063, "step": 6615 }, { "epoch": 0.4573798824749395, "grad_norm": 0.8481028268324342, "learning_rate": 9.000341553585421e-06, "loss": 0.3474, "step": 6616 }, { "epoch": 0.4574490148634635, "grad_norm": 0.8431618343285016, "learning_rate": 9.000005723974226e-06, "loss": 0.3463, "step": 6617 }, { "epoch": 0.4575181472519876, "grad_norm": 0.7392030186398644, "learning_rate": 8.9996698442301e-06, "loss": 0.3206, "step": 6618 }, { "epoch": 0.4575872796405116, "grad_norm": 0.809215503621876, "learning_rate": 8.99933391435725e-06, "loss": 0.4149, "step": 6619 }, { "epoch": 0.4576564120290356, "grad_norm": 1.2453485237832804, "learning_rate": 8.998997934359884e-06, "loss": 0.2974, "step": 6620 }, { "epoch": 0.45772554441755964, "grad_norm": 0.7267602594551715, "learning_rate": 8.998661904242217e-06, "loss": 0.2953, "step": 6621 }, { "epoch": 0.45779467680608366, "grad_norm": 0.697766094986507, "learning_rate": 8.998325824008459e-06, "loss": 0.4255, "step": 6622 }, { "epoch": 0.4578638091946077, "grad_norm": 0.6947526021970664, "learning_rate": 8.997989693662823e-06, "loss": 0.363, "step": 6623 }, { "epoch": 0.4579329415831317, "grad_norm": 0.5889864623159676, "learning_rate": 8.99765351320952e-06, "loss": 0.3098, "step": 6624 }, { "epoch": 0.4580020739716557, "grad_norm": 0.6892627061756599, "learning_rate": 8.997317282652764e-06, "loss": 0.3721, "step": 6625 }, { "epoch": 0.45807120636017973, "grad_norm": 0.7515647123128357, "learning_rate": 8.99698100199677e-06, "loss": 0.3896, "step": 6626 }, { "epoch": 0.45814033874870375, "grad_norm": 0.7756510693246812, "learning_rate": 8.996644671245752e-06, "loss": 0.3, "step": 6627 }, { "epoch": 0.45820947113722776, "grad_norm": 0.701876027561804, "learning_rate": 8.996308290403924e-06, "loss": 0.3096, "step": 6628 }, { "epoch": 0.45827860352575184, "grad_norm": 0.6741746878848859, "learning_rate": 8.995971859475504e-06, "loss": 0.3907, "step": 6629 }, { "epoch": 0.45834773591427586, "grad_norm": 0.9277818404705876, "learning_rate": 8.995635378464709e-06, "loss": 0.372, "step": 6630 }, { "epoch": 0.4584168683027999, "grad_norm": 0.6087733915194573, "learning_rate": 8.995298847375755e-06, "loss": 0.2561, "step": 6631 }, { "epoch": 0.4584860006913239, "grad_norm": 0.738359956150136, "learning_rate": 8.994962266212859e-06, "loss": 0.3606, "step": 6632 }, { "epoch": 0.4585551330798479, "grad_norm": 0.8829557924528134, "learning_rate": 8.994625634980241e-06, "loss": 0.4416, "step": 6633 }, { "epoch": 0.45862426546837193, "grad_norm": 0.7229973747455245, "learning_rate": 8.99428895368212e-06, "loss": 0.3628, "step": 6634 }, { "epoch": 0.45869339785689595, "grad_norm": 0.7725185240081482, "learning_rate": 8.993952222322712e-06, "loss": 0.2841, "step": 6635 }, { "epoch": 0.45876253024541996, "grad_norm": 0.7005586595238447, "learning_rate": 8.993615440906244e-06, "loss": 0.3837, "step": 6636 }, { "epoch": 0.458831662633944, "grad_norm": 0.7059988009482394, "learning_rate": 8.993278609436932e-06, "loss": 0.3304, "step": 6637 }, { "epoch": 0.458900795022468, "grad_norm": 0.652440556275487, "learning_rate": 8.992941727918999e-06, "loss": 0.2774, "step": 6638 }, { "epoch": 0.4589699274109921, "grad_norm": 0.6936773015563166, "learning_rate": 8.992604796356667e-06, "loss": 0.2986, "step": 6639 }, { "epoch": 0.4590390597995161, "grad_norm": 0.7144068468793797, "learning_rate": 8.992267814754158e-06, "loss": 0.421, "step": 6640 }, { "epoch": 0.4591081921880401, "grad_norm": 0.7448876628076626, "learning_rate": 8.991930783115699e-06, "loss": 0.3597, "step": 6641 }, { "epoch": 0.4591773245765641, "grad_norm": 0.7764618289683963, "learning_rate": 8.991593701445508e-06, "loss": 0.2321, "step": 6642 }, { "epoch": 0.45924645696508815, "grad_norm": 0.7356271640349155, "learning_rate": 8.991256569747818e-06, "loss": 0.3488, "step": 6643 }, { "epoch": 0.45931558935361216, "grad_norm": 0.7578074819087217, "learning_rate": 8.990919388026846e-06, "loss": 0.4062, "step": 6644 }, { "epoch": 0.4593847217421362, "grad_norm": 0.8044778872993527, "learning_rate": 8.990582156286822e-06, "loss": 0.3099, "step": 6645 }, { "epoch": 0.4594538541306602, "grad_norm": 0.8452559803942463, "learning_rate": 8.990244874531973e-06, "loss": 0.3293, "step": 6646 }, { "epoch": 0.4595229865191842, "grad_norm": 0.8596801975298284, "learning_rate": 8.989907542766525e-06, "loss": 0.3882, "step": 6647 }, { "epoch": 0.45959211890770824, "grad_norm": 0.9363546986872803, "learning_rate": 8.989570160994708e-06, "loss": 0.3557, "step": 6648 }, { "epoch": 0.4596612512962323, "grad_norm": 0.8058283454733567, "learning_rate": 8.989232729220745e-06, "loss": 0.3381, "step": 6649 }, { "epoch": 0.4597303836847563, "grad_norm": 0.798554796951841, "learning_rate": 8.988895247448872e-06, "loss": 0.3036, "step": 6650 }, { "epoch": 0.45979951607328035, "grad_norm": 0.7366298956358044, "learning_rate": 8.988557715683314e-06, "loss": 0.4412, "step": 6651 }, { "epoch": 0.45986864846180436, "grad_norm": 0.730084916621682, "learning_rate": 8.988220133928306e-06, "loss": 0.3227, "step": 6652 }, { "epoch": 0.4599377808503284, "grad_norm": 0.7670554523420908, "learning_rate": 8.987882502188072e-06, "loss": 0.2549, "step": 6653 }, { "epoch": 0.4600069132388524, "grad_norm": 0.7452271560460105, "learning_rate": 8.987544820466848e-06, "loss": 0.3556, "step": 6654 }, { "epoch": 0.4600760456273764, "grad_norm": 0.7271854373872205, "learning_rate": 8.987207088768868e-06, "loss": 0.3695, "step": 6655 }, { "epoch": 0.46014517801590044, "grad_norm": 0.6811239945670416, "learning_rate": 8.98686930709836e-06, "loss": 0.2944, "step": 6656 }, { "epoch": 0.46021431040442445, "grad_norm": 0.673135569579394, "learning_rate": 8.986531475459562e-06, "loss": 0.3028, "step": 6657 }, { "epoch": 0.46028344279294847, "grad_norm": 0.7426796124376639, "learning_rate": 8.986193593856706e-06, "loss": 0.4671, "step": 6658 }, { "epoch": 0.46035257518147255, "grad_norm": 0.7542860208198731, "learning_rate": 8.985855662294028e-06, "loss": 0.3338, "step": 6659 }, { "epoch": 0.46042170756999656, "grad_norm": 0.6267697810586069, "learning_rate": 8.985517680775759e-06, "loss": 0.2895, "step": 6660 }, { "epoch": 0.4604908399585206, "grad_norm": 1.6325545663099816, "learning_rate": 8.985179649306142e-06, "loss": 0.3491, "step": 6661 }, { "epoch": 0.4605599723470446, "grad_norm": 0.7330838204506273, "learning_rate": 8.984841567889407e-06, "loss": 0.3631, "step": 6662 }, { "epoch": 0.4606291047355686, "grad_norm": 0.886426771820518, "learning_rate": 8.984503436529797e-06, "loss": 0.2989, "step": 6663 }, { "epoch": 0.46069823712409264, "grad_norm": 0.7082727735474307, "learning_rate": 8.984165255231544e-06, "loss": 0.3161, "step": 6664 }, { "epoch": 0.46076736951261665, "grad_norm": 0.6767045678840995, "learning_rate": 8.983827023998892e-06, "loss": 0.4162, "step": 6665 }, { "epoch": 0.46083650190114067, "grad_norm": 0.7075735512315487, "learning_rate": 8.983488742836077e-06, "loss": 0.3285, "step": 6666 }, { "epoch": 0.4609056342896647, "grad_norm": 0.7056093833631067, "learning_rate": 8.98315041174734e-06, "loss": 0.3301, "step": 6667 }, { "epoch": 0.4609747666781887, "grad_norm": 0.784103898047662, "learning_rate": 8.98281203073692e-06, "loss": 0.3546, "step": 6668 }, { "epoch": 0.4610438990667128, "grad_norm": 0.7985120461535611, "learning_rate": 8.982473599809058e-06, "loss": 0.377, "step": 6669 }, { "epoch": 0.4611130314552368, "grad_norm": 0.6586566635484575, "learning_rate": 8.982135118967999e-06, "loss": 0.2697, "step": 6670 }, { "epoch": 0.4611821638437608, "grad_norm": 0.7040316609218624, "learning_rate": 8.981796588217981e-06, "loss": 0.3429, "step": 6671 }, { "epoch": 0.46125129623228484, "grad_norm": 0.7430460046531225, "learning_rate": 8.981458007563249e-06, "loss": 0.4105, "step": 6672 }, { "epoch": 0.46132042862080885, "grad_norm": 0.7637925075017574, "learning_rate": 8.981119377008046e-06, "loss": 0.3111, "step": 6673 }, { "epoch": 0.46138956100933287, "grad_norm": 0.6856918822417455, "learning_rate": 8.980780696556617e-06, "loss": 0.358, "step": 6674 }, { "epoch": 0.4614586933978569, "grad_norm": 0.7093411822522024, "learning_rate": 8.980441966213205e-06, "loss": 0.3778, "step": 6675 }, { "epoch": 0.4615278257863809, "grad_norm": 0.9842355839857353, "learning_rate": 8.980103185982056e-06, "loss": 0.4126, "step": 6676 }, { "epoch": 0.4615969581749049, "grad_norm": 0.8012898301517513, "learning_rate": 8.979764355867418e-06, "loss": 0.2946, "step": 6677 }, { "epoch": 0.46166609056342894, "grad_norm": 0.7776556673169575, "learning_rate": 8.979425475873534e-06, "loss": 0.3118, "step": 6678 }, { "epoch": 0.461735222951953, "grad_norm": 0.7608134653658124, "learning_rate": 8.979086546004655e-06, "loss": 0.4038, "step": 6679 }, { "epoch": 0.46180435534047704, "grad_norm": 0.7073812915247962, "learning_rate": 8.978747566265027e-06, "loss": 0.375, "step": 6680 }, { "epoch": 0.46187348772900105, "grad_norm": 0.6841457825967688, "learning_rate": 8.978408536658897e-06, "loss": 0.2457, "step": 6681 }, { "epoch": 0.46194262011752507, "grad_norm": 1.1280592284035658, "learning_rate": 8.978069457190517e-06, "loss": 0.3477, "step": 6682 }, { "epoch": 0.4620117525060491, "grad_norm": 0.6940711386824272, "learning_rate": 8.977730327864135e-06, "loss": 0.4615, "step": 6683 }, { "epoch": 0.4620808848945731, "grad_norm": 0.6387409988423195, "learning_rate": 8.977391148684003e-06, "loss": 0.328, "step": 6684 }, { "epoch": 0.4621500172830971, "grad_norm": 0.6598941296748123, "learning_rate": 8.977051919654369e-06, "loss": 0.2998, "step": 6685 }, { "epoch": 0.46221914967162114, "grad_norm": 1.5170917107104394, "learning_rate": 8.976712640779487e-06, "loss": 0.3647, "step": 6686 }, { "epoch": 0.46228828206014516, "grad_norm": 0.761508183438485, "learning_rate": 8.976373312063611e-06, "loss": 0.3976, "step": 6687 }, { "epoch": 0.4623574144486692, "grad_norm": 1.0789758730304357, "learning_rate": 8.97603393351099e-06, "loss": 0.2906, "step": 6688 }, { "epoch": 0.4624265468371932, "grad_norm": 0.7444091714328274, "learning_rate": 8.97569450512588e-06, "loss": 0.3488, "step": 6689 }, { "epoch": 0.46249567922571727, "grad_norm": 0.73985146344407, "learning_rate": 8.975355026912533e-06, "loss": 0.4187, "step": 6690 }, { "epoch": 0.4625648116142413, "grad_norm": 0.6782220363006144, "learning_rate": 8.975015498875207e-06, "loss": 0.3139, "step": 6691 }, { "epoch": 0.4626339440027653, "grad_norm": 0.6722348702381398, "learning_rate": 8.974675921018154e-06, "loss": 0.2642, "step": 6692 }, { "epoch": 0.4627030763912893, "grad_norm": 1.0249454658894073, "learning_rate": 8.974336293345631e-06, "loss": 0.3317, "step": 6693 }, { "epoch": 0.46277220877981334, "grad_norm": 0.7418535252870408, "learning_rate": 8.973996615861898e-06, "loss": 0.4581, "step": 6694 }, { "epoch": 0.46284134116833736, "grad_norm": 0.8618659895458864, "learning_rate": 8.973656888571206e-06, "loss": 0.255, "step": 6695 }, { "epoch": 0.4629104735568614, "grad_norm": 0.7531014842921753, "learning_rate": 8.973317111477817e-06, "loss": 0.3114, "step": 6696 }, { "epoch": 0.4629796059453854, "grad_norm": 0.6691454228392443, "learning_rate": 8.97297728458599e-06, "loss": 0.4258, "step": 6697 }, { "epoch": 0.4630487383339094, "grad_norm": 0.731035044422297, "learning_rate": 8.972637407899982e-06, "loss": 0.4016, "step": 6698 }, { "epoch": 0.46311787072243343, "grad_norm": 0.687808174156826, "learning_rate": 8.972297481424053e-06, "loss": 0.2751, "step": 6699 }, { "epoch": 0.4631870031109575, "grad_norm": 0.7529088683378496, "learning_rate": 8.971957505162464e-06, "loss": 0.3386, "step": 6700 }, { "epoch": 0.4632561354994815, "grad_norm": 0.644444839599815, "learning_rate": 8.971617479119476e-06, "loss": 0.3621, "step": 6701 }, { "epoch": 0.46332526788800554, "grad_norm": 0.7363602825594899, "learning_rate": 8.971277403299353e-06, "loss": 0.3463, "step": 6702 }, { "epoch": 0.46339440027652956, "grad_norm": 0.9314680250818421, "learning_rate": 8.970937277706352e-06, "loss": 0.3335, "step": 6703 }, { "epoch": 0.4634635326650536, "grad_norm": 0.8350325786591176, "learning_rate": 8.97059710234474e-06, "loss": 0.4075, "step": 6704 }, { "epoch": 0.4635326650535776, "grad_norm": 0.7456104413199737, "learning_rate": 8.97025687721878e-06, "loss": 0.4522, "step": 6705 }, { "epoch": 0.4636017974421016, "grad_norm": 0.6963868008509593, "learning_rate": 8.969916602332732e-06, "loss": 0.341, "step": 6706 }, { "epoch": 0.46367092983062563, "grad_norm": 0.7231273070617151, "learning_rate": 8.969576277690867e-06, "loss": 0.2896, "step": 6707 }, { "epoch": 0.46374006221914965, "grad_norm": 0.7300544445656797, "learning_rate": 8.969235903297446e-06, "loss": 0.4699, "step": 6708 }, { "epoch": 0.46380919460767367, "grad_norm": 0.7289469784360856, "learning_rate": 8.968895479156736e-06, "loss": 0.2972, "step": 6709 }, { "epoch": 0.46387832699619774, "grad_norm": 1.2173397391592025, "learning_rate": 8.968555005273005e-06, "loss": 0.2846, "step": 6710 }, { "epoch": 0.46394745938472176, "grad_norm": 0.9188896098129637, "learning_rate": 8.968214481650518e-06, "loss": 0.359, "step": 6711 }, { "epoch": 0.4640165917732458, "grad_norm": 0.7029157261863236, "learning_rate": 8.967873908293547e-06, "loss": 0.4035, "step": 6712 }, { "epoch": 0.4640857241617698, "grad_norm": 0.6984782992047277, "learning_rate": 8.967533285206356e-06, "loss": 0.3693, "step": 6713 }, { "epoch": 0.4641548565502938, "grad_norm": 0.8856215146615504, "learning_rate": 8.967192612393214e-06, "loss": 0.295, "step": 6714 }, { "epoch": 0.46422398893881783, "grad_norm": 0.8766231114992405, "learning_rate": 8.966851889858394e-06, "loss": 0.4356, "step": 6715 }, { "epoch": 0.46429312132734185, "grad_norm": 0.741973236858448, "learning_rate": 8.966511117606166e-06, "loss": 0.3919, "step": 6716 }, { "epoch": 0.46436225371586587, "grad_norm": 0.7093648187617734, "learning_rate": 8.966170295640796e-06, "loss": 0.3091, "step": 6717 }, { "epoch": 0.4644313861043899, "grad_norm": 0.6937569263558636, "learning_rate": 8.96582942396656e-06, "loss": 0.3793, "step": 6718 }, { "epoch": 0.4645005184929139, "grad_norm": 0.652094689729191, "learning_rate": 8.965488502587732e-06, "loss": 0.3773, "step": 6719 }, { "epoch": 0.464569650881438, "grad_norm": 0.7178731916813338, "learning_rate": 8.965147531508582e-06, "loss": 0.3208, "step": 6720 }, { "epoch": 0.464638783269962, "grad_norm": 0.775990632325407, "learning_rate": 8.964806510733383e-06, "loss": 0.3413, "step": 6721 }, { "epoch": 0.464707915658486, "grad_norm": 0.7301005038132143, "learning_rate": 8.96446544026641e-06, "loss": 0.4325, "step": 6722 }, { "epoch": 0.46477704804701003, "grad_norm": 0.7291772768855195, "learning_rate": 8.964124320111938e-06, "loss": 0.3711, "step": 6723 }, { "epoch": 0.46484618043553405, "grad_norm": 0.6976078806051411, "learning_rate": 8.963783150274242e-06, "loss": 0.2797, "step": 6724 }, { "epoch": 0.46491531282405807, "grad_norm": 0.7481757613010153, "learning_rate": 8.963441930757595e-06, "loss": 0.3609, "step": 6725 }, { "epoch": 0.4649844452125821, "grad_norm": 0.7280966167906295, "learning_rate": 8.96310066156628e-06, "loss": 0.4166, "step": 6726 }, { "epoch": 0.4650535776011061, "grad_norm": 0.7136409708271683, "learning_rate": 8.96275934270457e-06, "loss": 0.3821, "step": 6727 }, { "epoch": 0.4651227099896301, "grad_norm": 0.8725094820906332, "learning_rate": 8.962417974176742e-06, "loss": 0.2716, "step": 6728 }, { "epoch": 0.46519184237815414, "grad_norm": 0.7088990014339245, "learning_rate": 8.962076555987078e-06, "loss": 0.4028, "step": 6729 }, { "epoch": 0.4652609747666782, "grad_norm": 0.7792140907661189, "learning_rate": 8.961735088139853e-06, "loss": 0.4546, "step": 6730 }, { "epoch": 0.46533010715520223, "grad_norm": 0.6245297269181093, "learning_rate": 8.96139357063935e-06, "loss": 0.3082, "step": 6731 }, { "epoch": 0.46539923954372625, "grad_norm": 0.8609469213799837, "learning_rate": 8.961052003489847e-06, "loss": 0.3302, "step": 6732 }, { "epoch": 0.46546837193225027, "grad_norm": 0.651670423194108, "learning_rate": 8.960710386695627e-06, "loss": 0.4221, "step": 6733 }, { "epoch": 0.4655375043207743, "grad_norm": 0.6581261163310232, "learning_rate": 8.96036872026097e-06, "loss": 0.2972, "step": 6734 }, { "epoch": 0.4656066367092983, "grad_norm": 0.7978020582652008, "learning_rate": 8.96002700419016e-06, "loss": 0.2599, "step": 6735 }, { "epoch": 0.4656757690978223, "grad_norm": 0.7071695993654137, "learning_rate": 8.959685238487477e-06, "loss": 0.395, "step": 6736 }, { "epoch": 0.46574490148634634, "grad_norm": 0.6910608139673035, "learning_rate": 8.959343423157205e-06, "loss": 0.3935, "step": 6737 }, { "epoch": 0.46581403387487036, "grad_norm": 0.7236767437807252, "learning_rate": 8.959001558203631e-06, "loss": 0.2834, "step": 6738 }, { "epoch": 0.4658831662633944, "grad_norm": 0.7201228435518996, "learning_rate": 8.958659643631037e-06, "loss": 0.3381, "step": 6739 }, { "epoch": 0.46595229865191845, "grad_norm": 0.7109614973421885, "learning_rate": 8.95831767944371e-06, "loss": 0.4898, "step": 6740 }, { "epoch": 0.46602143104044247, "grad_norm": 0.7541588398606254, "learning_rate": 8.957975665645934e-06, "loss": 0.3311, "step": 6741 }, { "epoch": 0.4660905634289665, "grad_norm": 0.775896151647335, "learning_rate": 8.957633602241995e-06, "loss": 0.3195, "step": 6742 }, { "epoch": 0.4661596958174905, "grad_norm": 0.8012618917542798, "learning_rate": 8.957291489236181e-06, "loss": 0.3839, "step": 6743 }, { "epoch": 0.4662288282060145, "grad_norm": 0.8844143332262577, "learning_rate": 8.956949326632782e-06, "loss": 0.5074, "step": 6744 }, { "epoch": 0.46629796059453854, "grad_norm": 0.6326648977627545, "learning_rate": 8.956607114436085e-06, "loss": 0.3108, "step": 6745 }, { "epoch": 0.46636709298306256, "grad_norm": 0.8223125889324734, "learning_rate": 8.956264852650377e-06, "loss": 0.3109, "step": 6746 }, { "epoch": 0.4664362253715866, "grad_norm": 0.679559470034893, "learning_rate": 8.95592254127995e-06, "loss": 0.3868, "step": 6747 }, { "epoch": 0.4665053577601106, "grad_norm": 0.8380583405408063, "learning_rate": 8.955580180329096e-06, "loss": 0.3771, "step": 6748 }, { "epoch": 0.4665744901486346, "grad_norm": 0.7415777852979465, "learning_rate": 8.9552377698021e-06, "loss": 0.3186, "step": 6749 }, { "epoch": 0.4666436225371587, "grad_norm": 0.6744183786476449, "learning_rate": 8.95489530970326e-06, "loss": 0.3095, "step": 6750 }, { "epoch": 0.4667127549256827, "grad_norm": 0.6751551183266574, "learning_rate": 8.954552800036863e-06, "loss": 0.4053, "step": 6751 }, { "epoch": 0.4667818873142067, "grad_norm": 0.7061122119673886, "learning_rate": 8.954210240807203e-06, "loss": 0.3614, "step": 6752 }, { "epoch": 0.46685101970273074, "grad_norm": 0.8182561282371661, "learning_rate": 8.953867632018577e-06, "loss": 0.259, "step": 6753 }, { "epoch": 0.46692015209125476, "grad_norm": 0.7349456870478664, "learning_rate": 8.953524973675275e-06, "loss": 0.4132, "step": 6754 }, { "epoch": 0.4669892844797788, "grad_norm": 0.7384895347553962, "learning_rate": 8.953182265781593e-06, "loss": 0.3936, "step": 6755 }, { "epoch": 0.4670584168683028, "grad_norm": 0.6337370725500375, "learning_rate": 8.952839508341826e-06, "loss": 0.2585, "step": 6756 }, { "epoch": 0.4671275492568268, "grad_norm": 0.6956392650140901, "learning_rate": 8.952496701360269e-06, "loss": 0.2832, "step": 6757 }, { "epoch": 0.46719668164535083, "grad_norm": 0.8019479876102099, "learning_rate": 8.952153844841221e-06, "loss": 0.455, "step": 6758 }, { "epoch": 0.46726581403387485, "grad_norm": 0.7692290502099948, "learning_rate": 8.951810938788976e-06, "loss": 0.3322, "step": 6759 }, { "epoch": 0.46733494642239887, "grad_norm": 0.6553449402300942, "learning_rate": 8.951467983207834e-06, "loss": 0.2851, "step": 6760 }, { "epoch": 0.46740407881092294, "grad_norm": 0.7534483395778299, "learning_rate": 8.951124978102092e-06, "loss": 0.3161, "step": 6761 }, { "epoch": 0.46747321119944696, "grad_norm": 0.9260506856463369, "learning_rate": 8.95078192347605e-06, "loss": 0.4082, "step": 6762 }, { "epoch": 0.467542343587971, "grad_norm": 0.685618524546557, "learning_rate": 8.950438819334008e-06, "loss": 0.3471, "step": 6763 }, { "epoch": 0.467611475976495, "grad_norm": 0.6676561286903379, "learning_rate": 8.950095665680265e-06, "loss": 0.2834, "step": 6764 }, { "epoch": 0.467680608365019, "grad_norm": 0.7430902103577628, "learning_rate": 8.949752462519121e-06, "loss": 0.4701, "step": 6765 }, { "epoch": 0.46774974075354303, "grad_norm": 0.661373549485301, "learning_rate": 8.949409209854878e-06, "loss": 0.3282, "step": 6766 }, { "epoch": 0.46781887314206705, "grad_norm": 0.6436050510607105, "learning_rate": 8.949065907691841e-06, "loss": 0.3025, "step": 6767 }, { "epoch": 0.46788800553059107, "grad_norm": 0.7156309915092056, "learning_rate": 8.948722556034309e-06, "loss": 0.389, "step": 6768 }, { "epoch": 0.4679571379191151, "grad_norm": 0.8020236441574746, "learning_rate": 8.948379154886584e-06, "loss": 0.3969, "step": 6769 }, { "epoch": 0.4680262703076391, "grad_norm": 0.6886012559327223, "learning_rate": 8.948035704252974e-06, "loss": 0.3093, "step": 6770 }, { "epoch": 0.4680954026961632, "grad_norm": 0.6713656020626786, "learning_rate": 8.947692204137785e-06, "loss": 0.2903, "step": 6771 }, { "epoch": 0.4681645350846872, "grad_norm": 0.7020464284885354, "learning_rate": 8.947348654545318e-06, "loss": 0.4008, "step": 6772 }, { "epoch": 0.4682336674732112, "grad_norm": 0.7469915119839915, "learning_rate": 8.947005055479878e-06, "loss": 0.3821, "step": 6773 }, { "epoch": 0.46830279986173523, "grad_norm": 1.4738052403866935, "learning_rate": 8.946661406945775e-06, "loss": 0.3333, "step": 6774 }, { "epoch": 0.46837193225025925, "grad_norm": 0.8786645850579158, "learning_rate": 8.946317708947315e-06, "loss": 0.4195, "step": 6775 }, { "epoch": 0.46844106463878327, "grad_norm": 0.7141754031468658, "learning_rate": 8.945973961488803e-06, "loss": 0.4749, "step": 6776 }, { "epoch": 0.4685101970273073, "grad_norm": 0.7366811546387271, "learning_rate": 8.945630164574551e-06, "loss": 0.3267, "step": 6777 }, { "epoch": 0.4685793294158313, "grad_norm": 0.8540202666455795, "learning_rate": 8.945286318208864e-06, "loss": 0.3234, "step": 6778 }, { "epoch": 0.4686484618043553, "grad_norm": 0.6933055421840694, "learning_rate": 8.944942422396055e-06, "loss": 0.4297, "step": 6779 }, { "epoch": 0.46871759419287934, "grad_norm": 0.6849498339300345, "learning_rate": 8.944598477140432e-06, "loss": 0.3504, "step": 6780 }, { "epoch": 0.4687867265814034, "grad_norm": 0.8121389862036998, "learning_rate": 8.944254482446308e-06, "loss": 0.3829, "step": 6781 }, { "epoch": 0.46885585896992743, "grad_norm": 0.7086096869725237, "learning_rate": 8.943910438317993e-06, "loss": 0.3065, "step": 6782 }, { "epoch": 0.46892499135845145, "grad_norm": 0.7288306736787424, "learning_rate": 8.943566344759798e-06, "loss": 0.4288, "step": 6783 }, { "epoch": 0.46899412374697547, "grad_norm": 0.717014065323168, "learning_rate": 8.943222201776036e-06, "loss": 0.2676, "step": 6784 }, { "epoch": 0.4690632561354995, "grad_norm": 0.6884761861855603, "learning_rate": 8.94287800937102e-06, "loss": 0.2969, "step": 6785 }, { "epoch": 0.4691323885240235, "grad_norm": 0.7631135602935177, "learning_rate": 8.942533767549068e-06, "loss": 0.3487, "step": 6786 }, { "epoch": 0.4692015209125475, "grad_norm": 0.7343009474114665, "learning_rate": 8.942189476314489e-06, "loss": 0.3757, "step": 6787 }, { "epoch": 0.46927065330107154, "grad_norm": 0.6664445670930889, "learning_rate": 8.9418451356716e-06, "loss": 0.3429, "step": 6788 }, { "epoch": 0.46933978568959556, "grad_norm": 0.785355512536, "learning_rate": 8.941500745624717e-06, "loss": 0.3284, "step": 6789 }, { "epoch": 0.4694089180781196, "grad_norm": 0.8196599528249926, "learning_rate": 8.941156306178154e-06, "loss": 0.3552, "step": 6790 }, { "epoch": 0.46947805046664365, "grad_norm": 0.762548866366587, "learning_rate": 8.940811817336232e-06, "loss": 0.3373, "step": 6791 }, { "epoch": 0.46954718285516767, "grad_norm": 0.724164496084513, "learning_rate": 8.940467279103267e-06, "loss": 0.2962, "step": 6792 }, { "epoch": 0.4696163152436917, "grad_norm": 0.7304754450560395, "learning_rate": 8.940122691483576e-06, "loss": 0.3342, "step": 6793 }, { "epoch": 0.4696854476322157, "grad_norm": 0.78297684689258, "learning_rate": 8.939778054481478e-06, "loss": 0.4947, "step": 6794 }, { "epoch": 0.4697545800207397, "grad_norm": 0.7141972985950984, "learning_rate": 8.939433368101293e-06, "loss": 0.2747, "step": 6795 }, { "epoch": 0.46982371240926374, "grad_norm": 0.8203386815668797, "learning_rate": 8.939088632347342e-06, "loss": 0.3039, "step": 6796 }, { "epoch": 0.46989284479778776, "grad_norm": 0.8272960405244292, "learning_rate": 8.938743847223945e-06, "loss": 0.3973, "step": 6797 }, { "epoch": 0.4699619771863118, "grad_norm": 0.7588805233648909, "learning_rate": 8.938399012735421e-06, "loss": 0.4006, "step": 6798 }, { "epoch": 0.4700311095748358, "grad_norm": 0.7912590671363371, "learning_rate": 8.938054128886096e-06, "loss": 0.2802, "step": 6799 }, { "epoch": 0.4701002419633598, "grad_norm": 1.2353181544999425, "learning_rate": 8.937709195680288e-06, "loss": 0.3679, "step": 6800 }, { "epoch": 0.4701693743518839, "grad_norm": 0.8138079648712381, "learning_rate": 8.937364213122324e-06, "loss": 0.4668, "step": 6801 }, { "epoch": 0.4702385067404079, "grad_norm": 0.681772868847315, "learning_rate": 8.937019181216526e-06, "loss": 0.3033, "step": 6802 }, { "epoch": 0.4703076391289319, "grad_norm": 0.7821149152510298, "learning_rate": 8.936674099967217e-06, "loss": 0.322, "step": 6803 }, { "epoch": 0.47037677151745594, "grad_norm": 0.7224410310316377, "learning_rate": 8.936328969378724e-06, "loss": 0.4187, "step": 6804 }, { "epoch": 0.47044590390597996, "grad_norm": 0.7111291696882986, "learning_rate": 8.935983789455373e-06, "loss": 0.3993, "step": 6805 }, { "epoch": 0.470515036294504, "grad_norm": 0.6895319053637791, "learning_rate": 8.93563856020149e-06, "loss": 0.3122, "step": 6806 }, { "epoch": 0.470584168683028, "grad_norm": 0.7811644022565633, "learning_rate": 8.9352932816214e-06, "loss": 0.39, "step": 6807 }, { "epoch": 0.470653301071552, "grad_norm": 0.7080807791678851, "learning_rate": 8.93494795371943e-06, "loss": 0.4008, "step": 6808 }, { "epoch": 0.47072243346007603, "grad_norm": 0.8130912938014714, "learning_rate": 8.93460257649991e-06, "loss": 0.3025, "step": 6809 }, { "epoch": 0.47079156584860005, "grad_norm": 0.7577620455888178, "learning_rate": 8.934257149967172e-06, "loss": 0.3176, "step": 6810 }, { "epoch": 0.4708606982371241, "grad_norm": 1.2018789199223867, "learning_rate": 8.933911674125538e-06, "loss": 0.512, "step": 6811 }, { "epoch": 0.47092983062564814, "grad_norm": 0.7651209919281876, "learning_rate": 8.933566148979342e-06, "loss": 0.3682, "step": 6812 }, { "epoch": 0.47099896301417216, "grad_norm": 0.665660159706433, "learning_rate": 8.933220574532916e-06, "loss": 0.334, "step": 6813 }, { "epoch": 0.4710680954026962, "grad_norm": 0.6555352095432939, "learning_rate": 8.932874950790588e-06, "loss": 0.3352, "step": 6814 }, { "epoch": 0.4711372277912202, "grad_norm": 0.7195987689571974, "learning_rate": 8.93252927775669e-06, "loss": 0.4437, "step": 6815 }, { "epoch": 0.4712063601797442, "grad_norm": 1.204850135161217, "learning_rate": 8.932183555435556e-06, "loss": 0.3104, "step": 6816 }, { "epoch": 0.47127549256826823, "grad_norm": 0.7510430480766469, "learning_rate": 8.93183778383152e-06, "loss": 0.3161, "step": 6817 }, { "epoch": 0.47134462495679225, "grad_norm": 0.7874285863871933, "learning_rate": 8.931491962948913e-06, "loss": 0.3727, "step": 6818 }, { "epoch": 0.47141375734531626, "grad_norm": 0.7467965763879578, "learning_rate": 8.93114609279207e-06, "loss": 0.3272, "step": 6819 }, { "epoch": 0.4714828897338403, "grad_norm": 0.781649154095138, "learning_rate": 8.930800173365326e-06, "loss": 0.3, "step": 6820 }, { "epoch": 0.4715520221223643, "grad_norm": 0.7695979943145519, "learning_rate": 8.930454204673017e-06, "loss": 0.2864, "step": 6821 }, { "epoch": 0.4716211545108884, "grad_norm": 0.8087513516861927, "learning_rate": 8.93010818671948e-06, "loss": 0.4151, "step": 6822 }, { "epoch": 0.4716902868994124, "grad_norm": 0.6982491674666422, "learning_rate": 8.929762119509049e-06, "loss": 0.3019, "step": 6823 }, { "epoch": 0.4717594192879364, "grad_norm": 0.7861562532317348, "learning_rate": 8.929416003046063e-06, "loss": 0.3304, "step": 6824 }, { "epoch": 0.47182855167646043, "grad_norm": 0.7043538475277341, "learning_rate": 8.92906983733486e-06, "loss": 0.3605, "step": 6825 }, { "epoch": 0.47189768406498445, "grad_norm": 0.7406100548434752, "learning_rate": 8.928723622379778e-06, "loss": 0.4181, "step": 6826 }, { "epoch": 0.47196681645350846, "grad_norm": 1.2604410868306961, "learning_rate": 8.928377358185156e-06, "loss": 0.2943, "step": 6827 }, { "epoch": 0.4720359488420325, "grad_norm": 0.8395649812671842, "learning_rate": 8.928031044755335e-06, "loss": 0.3436, "step": 6828 }, { "epoch": 0.4721050812305565, "grad_norm": 0.8741557779461769, "learning_rate": 8.927684682094655e-06, "loss": 0.3839, "step": 6829 }, { "epoch": 0.4721742136190805, "grad_norm": 0.8793334111068359, "learning_rate": 8.927338270207456e-06, "loss": 0.3936, "step": 6830 }, { "epoch": 0.47224334600760454, "grad_norm": 0.8535328124253732, "learning_rate": 8.926991809098082e-06, "loss": 0.2818, "step": 6831 }, { "epoch": 0.4723124783961286, "grad_norm": 0.7812186047366628, "learning_rate": 8.926645298770872e-06, "loss": 0.4073, "step": 6832 }, { "epoch": 0.47238161078465263, "grad_norm": 0.7746396600745759, "learning_rate": 8.926298739230172e-06, "loss": 0.4308, "step": 6833 }, { "epoch": 0.47245074317317665, "grad_norm": 0.7749677151374729, "learning_rate": 8.925952130480322e-06, "loss": 0.2841, "step": 6834 }, { "epoch": 0.47251987556170066, "grad_norm": 0.6932263279346516, "learning_rate": 8.92560547252567e-06, "loss": 0.2881, "step": 6835 }, { "epoch": 0.4725890079502247, "grad_norm": 0.7451341148059608, "learning_rate": 8.92525876537056e-06, "loss": 0.3993, "step": 6836 }, { "epoch": 0.4726581403387487, "grad_norm": 0.7960287069329907, "learning_rate": 8.924912009019333e-06, "loss": 0.4279, "step": 6837 }, { "epoch": 0.4727272727272727, "grad_norm": 0.7950085953647287, "learning_rate": 8.92456520347634e-06, "loss": 0.3131, "step": 6838 }, { "epoch": 0.47279640511579674, "grad_norm": 0.7144032550765445, "learning_rate": 8.924218348745927e-06, "loss": 0.3313, "step": 6839 }, { "epoch": 0.47286553750432075, "grad_norm": 0.6877705472717613, "learning_rate": 8.923871444832439e-06, "loss": 0.4679, "step": 6840 }, { "epoch": 0.4729346698928448, "grad_norm": 0.7529263970356069, "learning_rate": 8.923524491740225e-06, "loss": 0.415, "step": 6841 }, { "epoch": 0.47300380228136885, "grad_norm": 0.7704410431242801, "learning_rate": 8.923177489473633e-06, "loss": 0.315, "step": 6842 }, { "epoch": 0.47307293466989286, "grad_norm": 0.9250288970785209, "learning_rate": 8.922830438037013e-06, "loss": 0.3463, "step": 6843 }, { "epoch": 0.4731420670584169, "grad_norm": 0.8372523010982791, "learning_rate": 8.922483337434712e-06, "loss": 0.3759, "step": 6844 }, { "epoch": 0.4732111994469409, "grad_norm": 0.7335176554795824, "learning_rate": 8.922136187671085e-06, "loss": 0.3436, "step": 6845 }, { "epoch": 0.4732803318354649, "grad_norm": 0.700936350801214, "learning_rate": 8.921788988750479e-06, "loss": 0.2769, "step": 6846 }, { "epoch": 0.47334946422398894, "grad_norm": 0.7327209039003465, "learning_rate": 8.921441740677249e-06, "loss": 0.3703, "step": 6847 }, { "epoch": 0.47341859661251295, "grad_norm": 0.9134086710535474, "learning_rate": 8.921094443455743e-06, "loss": 0.3773, "step": 6848 }, { "epoch": 0.47348772900103697, "grad_norm": 0.9549313210576131, "learning_rate": 8.920747097090314e-06, "loss": 0.3062, "step": 6849 }, { "epoch": 0.473556861389561, "grad_norm": 0.8036163685447131, "learning_rate": 8.920399701585319e-06, "loss": 0.3972, "step": 6850 }, { "epoch": 0.473625993778085, "grad_norm": 0.7434611305307159, "learning_rate": 8.920052256945112e-06, "loss": 0.4637, "step": 6851 }, { "epoch": 0.4736951261666091, "grad_norm": 1.5112172387053662, "learning_rate": 8.919704763174042e-06, "loss": 0.3294, "step": 6852 }, { "epoch": 0.4737642585551331, "grad_norm": 0.8866634928384292, "learning_rate": 8.919357220276468e-06, "loss": 0.362, "step": 6853 }, { "epoch": 0.4738333909436571, "grad_norm": 0.7586256574247047, "learning_rate": 8.919009628256748e-06, "loss": 0.3722, "step": 6854 }, { "epoch": 0.47390252333218114, "grad_norm": 0.6943182481553529, "learning_rate": 8.918661987119235e-06, "loss": 0.3535, "step": 6855 }, { "epoch": 0.47397165572070515, "grad_norm": 0.7235150367845113, "learning_rate": 8.918314296868288e-06, "loss": 0.3316, "step": 6856 }, { "epoch": 0.47404078810922917, "grad_norm": 0.6794795580829925, "learning_rate": 8.917966557508262e-06, "loss": 0.3119, "step": 6857 }, { "epoch": 0.4741099204977532, "grad_norm": 0.7345484099688735, "learning_rate": 8.917618769043519e-06, "loss": 0.3811, "step": 6858 }, { "epoch": 0.4741790528862772, "grad_norm": 0.7521841761927404, "learning_rate": 8.917270931478413e-06, "loss": 0.3363, "step": 6859 }, { "epoch": 0.4742481852748012, "grad_norm": 0.6434275306132561, "learning_rate": 8.91692304481731e-06, "loss": 0.2994, "step": 6860 }, { "epoch": 0.47431731766332524, "grad_norm": 0.6950339999933022, "learning_rate": 8.916575109064564e-06, "loss": 0.3861, "step": 6861 }, { "epoch": 0.4743864500518493, "grad_norm": 0.7109971841940177, "learning_rate": 8.91622712422454e-06, "loss": 0.4598, "step": 6862 }, { "epoch": 0.47445558244037334, "grad_norm": 0.6543988054600413, "learning_rate": 8.915879090301598e-06, "loss": 0.2837, "step": 6863 }, { "epoch": 0.47452471482889735, "grad_norm": 0.7862227422807403, "learning_rate": 8.9155310073001e-06, "loss": 0.3404, "step": 6864 }, { "epoch": 0.47459384721742137, "grad_norm": 0.7063185845533221, "learning_rate": 8.915182875224407e-06, "loss": 0.3799, "step": 6865 }, { "epoch": 0.4746629796059454, "grad_norm": 0.663773929592175, "learning_rate": 8.914834694078885e-06, "loss": 0.311, "step": 6866 }, { "epoch": 0.4747321119944694, "grad_norm": 0.6980335302106118, "learning_rate": 8.914486463867896e-06, "loss": 0.3303, "step": 6867 }, { "epoch": 0.4748012443829934, "grad_norm": 0.7124964979426074, "learning_rate": 8.914138184595803e-06, "loss": 0.4051, "step": 6868 }, { "epoch": 0.47487037677151744, "grad_norm": 0.7283190007959723, "learning_rate": 8.913789856266976e-06, "loss": 0.4638, "step": 6869 }, { "epoch": 0.47493950916004146, "grad_norm": 0.6461387157883247, "learning_rate": 8.913441478885776e-06, "loss": 0.3062, "step": 6870 }, { "epoch": 0.4750086415485655, "grad_norm": 0.6216151106181023, "learning_rate": 8.913093052456571e-06, "loss": 0.2787, "step": 6871 }, { "epoch": 0.47507777393708955, "grad_norm": 0.8351253794578911, "learning_rate": 8.91274457698373e-06, "loss": 0.4974, "step": 6872 }, { "epoch": 0.47514690632561357, "grad_norm": 0.6826326749376646, "learning_rate": 8.912396052471614e-06, "loss": 0.391, "step": 6873 }, { "epoch": 0.4752160387141376, "grad_norm": 0.7131068318839321, "learning_rate": 8.9120474789246e-06, "loss": 0.2866, "step": 6874 }, { "epoch": 0.4752851711026616, "grad_norm": 0.7382845559246698, "learning_rate": 8.911698856347048e-06, "loss": 0.3224, "step": 6875 }, { "epoch": 0.4753543034911856, "grad_norm": 0.6812780017554183, "learning_rate": 8.911350184743335e-06, "loss": 0.4278, "step": 6876 }, { "epoch": 0.47542343587970964, "grad_norm": 0.683983784798323, "learning_rate": 8.911001464117825e-06, "loss": 0.3139, "step": 6877 }, { "epoch": 0.47549256826823366, "grad_norm": 0.6438802531244567, "learning_rate": 8.910652694474893e-06, "loss": 0.2948, "step": 6878 }, { "epoch": 0.4755617006567577, "grad_norm": 0.717398821473885, "learning_rate": 8.910303875818908e-06, "loss": 0.3926, "step": 6879 }, { "epoch": 0.4756308330452817, "grad_norm": 0.7272970430626221, "learning_rate": 8.909955008154242e-06, "loss": 0.3453, "step": 6880 }, { "epoch": 0.4756999654338057, "grad_norm": 0.6076586790830285, "learning_rate": 8.909606091485267e-06, "loss": 0.3193, "step": 6881 }, { "epoch": 0.47576909782232973, "grad_norm": 0.665820252709973, "learning_rate": 8.909257125816357e-06, "loss": 0.3457, "step": 6882 }, { "epoch": 0.4758382302108538, "grad_norm": 0.6788715474760204, "learning_rate": 8.908908111151886e-06, "loss": 0.4377, "step": 6883 }, { "epoch": 0.4759073625993778, "grad_norm": 0.7005185560454621, "learning_rate": 8.908559047496227e-06, "loss": 0.3403, "step": 6884 }, { "epoch": 0.47597649498790184, "grad_norm": 0.6412853287667463, "learning_rate": 8.908209934853755e-06, "loss": 0.2861, "step": 6885 }, { "epoch": 0.47604562737642586, "grad_norm": 0.7011998470226933, "learning_rate": 8.907860773228848e-06, "loss": 0.4233, "step": 6886 }, { "epoch": 0.4761147597649499, "grad_norm": 0.7196793901760212, "learning_rate": 8.907511562625876e-06, "loss": 0.3713, "step": 6887 }, { "epoch": 0.4761838921534739, "grad_norm": 1.0078998032048225, "learning_rate": 8.907162303049223e-06, "loss": 0.3437, "step": 6888 }, { "epoch": 0.4762530245419979, "grad_norm": 0.8142028579714489, "learning_rate": 8.906812994503264e-06, "loss": 0.3092, "step": 6889 }, { "epoch": 0.47632215693052193, "grad_norm": 0.6977582040419451, "learning_rate": 8.906463636992374e-06, "loss": 0.445, "step": 6890 }, { "epoch": 0.47639128931904595, "grad_norm": 0.9263580128128482, "learning_rate": 8.906114230520934e-06, "loss": 0.4046, "step": 6891 }, { "epoch": 0.47646042170756997, "grad_norm": 0.6540940328176935, "learning_rate": 8.905764775093325e-06, "loss": 0.3304, "step": 6892 }, { "epoch": 0.47652955409609404, "grad_norm": 2.4326319562719467, "learning_rate": 8.905415270713923e-06, "loss": 0.3094, "step": 6893 }, { "epoch": 0.47659868648461806, "grad_norm": 0.7550382621176468, "learning_rate": 8.905065717387109e-06, "loss": 0.4344, "step": 6894 }, { "epoch": 0.4766678188731421, "grad_norm": 0.8711119341348961, "learning_rate": 8.904716115117266e-06, "loss": 0.3384, "step": 6895 }, { "epoch": 0.4767369512616661, "grad_norm": 0.7097361034521676, "learning_rate": 8.904366463908776e-06, "loss": 0.3238, "step": 6896 }, { "epoch": 0.4768060836501901, "grad_norm": 0.6728260593447957, "learning_rate": 8.904016763766017e-06, "loss": 0.3714, "step": 6897 }, { "epoch": 0.47687521603871413, "grad_norm": 0.6620955370351121, "learning_rate": 8.903667014693378e-06, "loss": 0.3602, "step": 6898 }, { "epoch": 0.47694434842723815, "grad_norm": 0.765974286634655, "learning_rate": 8.90331721669524e-06, "loss": 0.3243, "step": 6899 }, { "epoch": 0.47701348081576217, "grad_norm": 0.7486559264281113, "learning_rate": 8.902967369775983e-06, "loss": 0.3321, "step": 6900 }, { "epoch": 0.4770826132042862, "grad_norm": 0.7927790251150808, "learning_rate": 8.902617473939996e-06, "loss": 0.4214, "step": 6901 }, { "epoch": 0.4771517455928102, "grad_norm": 0.6571197939636172, "learning_rate": 8.902267529191666e-06, "loss": 0.2716, "step": 6902 }, { "epoch": 0.4772208779813343, "grad_norm": 0.7528469898673655, "learning_rate": 8.901917535535373e-06, "loss": 0.2811, "step": 6903 }, { "epoch": 0.4772900103698583, "grad_norm": 0.7148843826250723, "learning_rate": 8.90156749297551e-06, "loss": 0.4075, "step": 6904 }, { "epoch": 0.4773591427583823, "grad_norm": 0.7457689314777555, "learning_rate": 8.90121740151646e-06, "loss": 0.3787, "step": 6905 }, { "epoch": 0.47742827514690633, "grad_norm": 0.7216499346908607, "learning_rate": 8.900867261162612e-06, "loss": 0.3344, "step": 6906 }, { "epoch": 0.47749740753543035, "grad_norm": 0.757474881281227, "learning_rate": 8.900517071918356e-06, "loss": 0.4055, "step": 6907 }, { "epoch": 0.47756653992395437, "grad_norm": 0.8412702781462996, "learning_rate": 8.900166833788075e-06, "loss": 0.4277, "step": 6908 }, { "epoch": 0.4776356723124784, "grad_norm": 0.6909583745801379, "learning_rate": 8.899816546776167e-06, "loss": 0.3188, "step": 6909 }, { "epoch": 0.4777048047010024, "grad_norm": 0.6953073636633657, "learning_rate": 8.899466210887015e-06, "loss": 0.2927, "step": 6910 }, { "epoch": 0.4777739370895264, "grad_norm": 0.7530158550611533, "learning_rate": 8.899115826125015e-06, "loss": 0.3969, "step": 6911 }, { "epoch": 0.47784306947805044, "grad_norm": 0.8588502322626896, "learning_rate": 8.898765392494557e-06, "loss": 0.4205, "step": 6912 }, { "epoch": 0.4779122018665745, "grad_norm": 0.6884904017026343, "learning_rate": 8.89841491000003e-06, "loss": 0.2646, "step": 6913 }, { "epoch": 0.47798133425509853, "grad_norm": 1.0165060589976624, "learning_rate": 8.89806437864583e-06, "loss": 0.3597, "step": 6914 }, { "epoch": 0.47805046664362255, "grad_norm": 0.7272560281276461, "learning_rate": 8.897713798436351e-06, "loss": 0.4099, "step": 6915 }, { "epoch": 0.47811959903214657, "grad_norm": 0.7803739969652161, "learning_rate": 8.897363169375984e-06, "loss": 0.3043, "step": 6916 }, { "epoch": 0.4781887314206706, "grad_norm": 0.6636845673923643, "learning_rate": 8.897012491469125e-06, "loss": 0.2944, "step": 6917 }, { "epoch": 0.4782578638091946, "grad_norm": 0.6974497859115154, "learning_rate": 8.89666176472017e-06, "loss": 0.3441, "step": 6918 }, { "epoch": 0.4783269961977186, "grad_norm": 0.7143700851689524, "learning_rate": 8.896310989133513e-06, "loss": 0.3651, "step": 6919 }, { "epoch": 0.47839612858624264, "grad_norm": 0.7470750608301486, "learning_rate": 8.895960164713549e-06, "loss": 0.2671, "step": 6920 }, { "epoch": 0.47846526097476666, "grad_norm": 0.6952935159807753, "learning_rate": 8.895609291464679e-06, "loss": 0.2844, "step": 6921 }, { "epoch": 0.4785343933632907, "grad_norm": 0.737007047379456, "learning_rate": 8.895258369391298e-06, "loss": 0.4387, "step": 6922 }, { "epoch": 0.47860352575181475, "grad_norm": 0.8327127907448741, "learning_rate": 8.894907398497805e-06, "loss": 0.3594, "step": 6923 }, { "epoch": 0.47867265814033877, "grad_norm": 0.8004633391011231, "learning_rate": 8.8945563787886e-06, "loss": 0.3055, "step": 6924 }, { "epoch": 0.4787417905288628, "grad_norm": 0.690004651475706, "learning_rate": 8.894205310268078e-06, "loss": 0.3215, "step": 6925 }, { "epoch": 0.4788109229173868, "grad_norm": 0.8162224479893608, "learning_rate": 8.893854192940645e-06, "loss": 0.3771, "step": 6926 }, { "epoch": 0.4788800553059108, "grad_norm": 0.8341458704919263, "learning_rate": 8.893503026810697e-06, "loss": 0.3138, "step": 6927 }, { "epoch": 0.47894918769443484, "grad_norm": 0.6458111226230919, "learning_rate": 8.893151811882636e-06, "loss": 0.3452, "step": 6928 }, { "epoch": 0.47901832008295886, "grad_norm": 0.9672193797266225, "learning_rate": 8.892800548160866e-06, "loss": 0.3985, "step": 6929 }, { "epoch": 0.4790874524714829, "grad_norm": 0.8599039592363427, "learning_rate": 8.892449235649788e-06, "loss": 0.4085, "step": 6930 }, { "epoch": 0.4791565848600069, "grad_norm": 0.8019802009266814, "learning_rate": 8.892097874353804e-06, "loss": 0.283, "step": 6931 }, { "epoch": 0.4792257172485309, "grad_norm": 0.8464204277101596, "learning_rate": 8.89174646427732e-06, "loss": 0.3651, "step": 6932 }, { "epoch": 0.479294849637055, "grad_norm": 0.742890011395056, "learning_rate": 8.891395005424738e-06, "loss": 0.435, "step": 6933 }, { "epoch": 0.479363982025579, "grad_norm": 0.6694611389119555, "learning_rate": 8.891043497800466e-06, "loss": 0.3197, "step": 6934 }, { "epoch": 0.479433114414103, "grad_norm": 0.7419740769388314, "learning_rate": 8.890691941408906e-06, "loss": 0.2987, "step": 6935 }, { "epoch": 0.47950224680262704, "grad_norm": 0.6643553213048435, "learning_rate": 8.890340336254467e-06, "loss": 0.3655, "step": 6936 }, { "epoch": 0.47957137919115106, "grad_norm": 0.8257984802305969, "learning_rate": 8.889988682341555e-06, "loss": 0.4177, "step": 6937 }, { "epoch": 0.4796405115796751, "grad_norm": 1.1371231563240587, "learning_rate": 8.889636979674575e-06, "loss": 0.318, "step": 6938 }, { "epoch": 0.4797096439681991, "grad_norm": 0.7246595027843474, "learning_rate": 8.889285228257938e-06, "loss": 0.2896, "step": 6939 }, { "epoch": 0.4797787763567231, "grad_norm": 0.7345202860250378, "learning_rate": 8.888933428096051e-06, "loss": 0.4068, "step": 6940 }, { "epoch": 0.47984790874524713, "grad_norm": 0.6675263119883342, "learning_rate": 8.888581579193324e-06, "loss": 0.2765, "step": 6941 }, { "epoch": 0.47991704113377115, "grad_norm": 0.6870626233350384, "learning_rate": 8.888229681554167e-06, "loss": 0.3285, "step": 6942 }, { "epoch": 0.4799861735222952, "grad_norm": 0.7705693996253776, "learning_rate": 8.887877735182987e-06, "loss": 0.4177, "step": 6943 }, { "epoch": 0.48005530591081924, "grad_norm": 0.6861605165294891, "learning_rate": 8.8875257400842e-06, "loss": 0.4297, "step": 6944 }, { "epoch": 0.48012443829934326, "grad_norm": 0.7675793767370501, "learning_rate": 8.887173696262216e-06, "loss": 0.3598, "step": 6945 }, { "epoch": 0.4801935706878673, "grad_norm": 0.7063473700144475, "learning_rate": 8.886821603721446e-06, "loss": 0.3169, "step": 6946 }, { "epoch": 0.4802627030763913, "grad_norm": 0.7500862786300833, "learning_rate": 8.886469462466303e-06, "loss": 0.401, "step": 6947 }, { "epoch": 0.4803318354649153, "grad_norm": 0.6774129208836336, "learning_rate": 8.886117272501202e-06, "loss": 0.3591, "step": 6948 }, { "epoch": 0.48040096785343933, "grad_norm": 0.6609226555726041, "learning_rate": 8.885765033830555e-06, "loss": 0.3438, "step": 6949 }, { "epoch": 0.48047010024196335, "grad_norm": 0.7260607890204057, "learning_rate": 8.885412746458778e-06, "loss": 0.3548, "step": 6950 }, { "epoch": 0.48053923263048737, "grad_norm": 0.7837347946594212, "learning_rate": 8.885060410390287e-06, "loss": 0.36, "step": 6951 }, { "epoch": 0.4806083650190114, "grad_norm": 0.7532789704691527, "learning_rate": 8.884708025629496e-06, "loss": 0.3572, "step": 6952 }, { "epoch": 0.4806774974075354, "grad_norm": 0.6559719745525177, "learning_rate": 8.884355592180823e-06, "loss": 0.2903, "step": 6953 }, { "epoch": 0.4807466297960595, "grad_norm": 0.689355434070117, "learning_rate": 8.884003110048684e-06, "loss": 0.3907, "step": 6954 }, { "epoch": 0.4808157621845835, "grad_norm": 0.7163092281039298, "learning_rate": 8.883650579237497e-06, "loss": 0.3237, "step": 6955 }, { "epoch": 0.4808848945731075, "grad_norm": 0.6824212324967285, "learning_rate": 8.883297999751682e-06, "loss": 0.2517, "step": 6956 }, { "epoch": 0.48095402696163153, "grad_norm": 0.7094204845983315, "learning_rate": 8.882945371595655e-06, "loss": 0.3145, "step": 6957 }, { "epoch": 0.48102315935015555, "grad_norm": 0.7595336237888266, "learning_rate": 8.88259269477384e-06, "loss": 0.4003, "step": 6958 }, { "epoch": 0.48109229173867957, "grad_norm": 0.9145089160814953, "learning_rate": 8.882239969290652e-06, "loss": 0.2825, "step": 6959 }, { "epoch": 0.4811614241272036, "grad_norm": 0.7653196758211986, "learning_rate": 8.881887195150514e-06, "loss": 0.2919, "step": 6960 }, { "epoch": 0.4812305565157276, "grad_norm": 0.8327655529168762, "learning_rate": 8.881534372357848e-06, "loss": 0.363, "step": 6961 }, { "epoch": 0.4812996889042516, "grad_norm": 0.7130410532848431, "learning_rate": 8.881181500917076e-06, "loss": 0.3588, "step": 6962 }, { "epoch": 0.48136882129277564, "grad_norm": 0.815735454906776, "learning_rate": 8.88082858083262e-06, "loss": 0.4056, "step": 6963 }, { "epoch": 0.4814379536812997, "grad_norm": 0.7027261301870297, "learning_rate": 8.880475612108904e-06, "loss": 0.3694, "step": 6964 }, { "epoch": 0.48150708606982373, "grad_norm": 0.8158742316599211, "learning_rate": 8.880122594750349e-06, "loss": 0.3953, "step": 6965 }, { "epoch": 0.48157621845834775, "grad_norm": 0.7575985920997999, "learning_rate": 8.879769528761384e-06, "loss": 0.3495, "step": 6966 }, { "epoch": 0.48164535084687177, "grad_norm": 0.6909927222512665, "learning_rate": 8.879416414146431e-06, "loss": 0.3065, "step": 6967 }, { "epoch": 0.4817144832353958, "grad_norm": 0.7718427611116035, "learning_rate": 8.879063250909916e-06, "loss": 0.4142, "step": 6968 }, { "epoch": 0.4817836156239198, "grad_norm": 0.6526644054842434, "learning_rate": 8.878710039056267e-06, "loss": 0.3665, "step": 6969 }, { "epoch": 0.4818527480124438, "grad_norm": 0.7137330951970087, "learning_rate": 8.878356778589908e-06, "loss": 0.2896, "step": 6970 }, { "epoch": 0.48192188040096784, "grad_norm": 0.638824399616581, "learning_rate": 8.878003469515269e-06, "loss": 0.2895, "step": 6971 }, { "epoch": 0.48199101278949186, "grad_norm": 0.8450679793202559, "learning_rate": 8.877650111836775e-06, "loss": 0.4233, "step": 6972 }, { "epoch": 0.4820601451780159, "grad_norm": 0.757707393353937, "learning_rate": 8.87729670555886e-06, "loss": 0.3795, "step": 6973 }, { "epoch": 0.48212927756653995, "grad_norm": 0.6939114228004383, "learning_rate": 8.876943250685947e-06, "loss": 0.3579, "step": 6974 }, { "epoch": 0.48219840995506397, "grad_norm": 0.7762023758388598, "learning_rate": 8.876589747222473e-06, "loss": 0.3279, "step": 6975 }, { "epoch": 0.482267542343588, "grad_norm": 0.8270133455301114, "learning_rate": 8.87623619517286e-06, "loss": 0.4602, "step": 6976 }, { "epoch": 0.482336674732112, "grad_norm": 0.6972015725961015, "learning_rate": 8.875882594541546e-06, "loss": 0.3281, "step": 6977 }, { "epoch": 0.482405807120636, "grad_norm": 0.6844765902211587, "learning_rate": 8.87552894533296e-06, "loss": 0.3384, "step": 6978 }, { "epoch": 0.48247493950916004, "grad_norm": 0.7883844528237881, "learning_rate": 8.875175247551538e-06, "loss": 0.4499, "step": 6979 }, { "epoch": 0.48254407189768406, "grad_norm": 0.7989189539335617, "learning_rate": 8.874821501201707e-06, "loss": 0.3981, "step": 6980 }, { "epoch": 0.4826132042862081, "grad_norm": 0.7130936558431297, "learning_rate": 8.874467706287902e-06, "loss": 0.2788, "step": 6981 }, { "epoch": 0.4826823366747321, "grad_norm": 0.8312866819558596, "learning_rate": 8.874113862814561e-06, "loss": 0.3501, "step": 6982 }, { "epoch": 0.4827514690632561, "grad_norm": 0.8058026955891664, "learning_rate": 8.873759970786115e-06, "loss": 0.4324, "step": 6983 }, { "epoch": 0.4828206014517802, "grad_norm": 0.8003471091035679, "learning_rate": 8.873406030207001e-06, "loss": 0.3, "step": 6984 }, { "epoch": 0.4828897338403042, "grad_norm": 0.7298892601976752, "learning_rate": 8.873052041081655e-06, "loss": 0.3227, "step": 6985 }, { "epoch": 0.4829588662288282, "grad_norm": 0.757330208632104, "learning_rate": 8.872698003414515e-06, "loss": 0.4042, "step": 6986 }, { "epoch": 0.48302799861735224, "grad_norm": 0.728048724704002, "learning_rate": 8.872343917210016e-06, "loss": 0.4433, "step": 6987 }, { "epoch": 0.48309713100587626, "grad_norm": 0.7742429028038883, "learning_rate": 8.871989782472595e-06, "loss": 0.3234, "step": 6988 }, { "epoch": 0.4831662633944003, "grad_norm": 0.7336822820343846, "learning_rate": 8.871635599206693e-06, "loss": 0.3386, "step": 6989 }, { "epoch": 0.4832353957829243, "grad_norm": 0.8280406150640405, "learning_rate": 8.871281367416747e-06, "loss": 0.3672, "step": 6990 }, { "epoch": 0.4833045281714483, "grad_norm": 1.880824658235605, "learning_rate": 8.8709270871072e-06, "loss": 0.4083, "step": 6991 }, { "epoch": 0.48337366055997233, "grad_norm": 0.7757172337379798, "learning_rate": 8.870572758282488e-06, "loss": 0.2661, "step": 6992 }, { "epoch": 0.48344279294849635, "grad_norm": 0.7095420609967059, "learning_rate": 8.870218380947054e-06, "loss": 0.4075, "step": 6993 }, { "epoch": 0.4835119253370204, "grad_norm": 0.8354572052779855, "learning_rate": 8.869863955105339e-06, "loss": 0.4104, "step": 6994 }, { "epoch": 0.48358105772554444, "grad_norm": 0.7105560314515823, "learning_rate": 8.869509480761786e-06, "loss": 0.3962, "step": 6995 }, { "epoch": 0.48365019011406846, "grad_norm": 0.6703263781745621, "learning_rate": 8.869154957920838e-06, "loss": 0.2699, "step": 6996 }, { "epoch": 0.4837193225025925, "grad_norm": 0.7324233716806422, "learning_rate": 8.868800386586935e-06, "loss": 0.4523, "step": 6997 }, { "epoch": 0.4837884548911165, "grad_norm": 0.8844461316768151, "learning_rate": 8.868445766764527e-06, "loss": 0.3795, "step": 6998 }, { "epoch": 0.4838575872796405, "grad_norm": 0.7296836828683839, "learning_rate": 8.868091098458052e-06, "loss": 0.2988, "step": 6999 }, { "epoch": 0.48392671966816453, "grad_norm": 0.7930985919361508, "learning_rate": 8.867736381671958e-06, "loss": 0.3685, "step": 7000 }, { "epoch": 0.48399585205668855, "grad_norm": 0.6849731596043526, "learning_rate": 8.867381616410691e-06, "loss": 0.4179, "step": 7001 }, { "epoch": 0.48406498444521256, "grad_norm": 1.0176978387012496, "learning_rate": 8.867026802678698e-06, "loss": 0.3035, "step": 7002 }, { "epoch": 0.4841341168337366, "grad_norm": 0.7247211121757883, "learning_rate": 8.866671940480424e-06, "loss": 0.2512, "step": 7003 }, { "epoch": 0.48420324922226066, "grad_norm": 0.6908615847337172, "learning_rate": 8.86631702982032e-06, "loss": 0.3454, "step": 7004 }, { "epoch": 0.4842723816107847, "grad_norm": 0.8360814444518837, "learning_rate": 8.86596207070283e-06, "loss": 0.3824, "step": 7005 }, { "epoch": 0.4843415139993087, "grad_norm": 0.8336343764261283, "learning_rate": 8.865607063132402e-06, "loss": 0.3001, "step": 7006 }, { "epoch": 0.4844106463878327, "grad_norm": 0.8922437920831642, "learning_rate": 8.865252007113491e-06, "loss": 0.3281, "step": 7007 }, { "epoch": 0.48447977877635673, "grad_norm": 1.1542618973572258, "learning_rate": 8.864896902650545e-06, "loss": 0.4631, "step": 7008 }, { "epoch": 0.48454891116488075, "grad_norm": 0.8384810519524789, "learning_rate": 8.86454174974801e-06, "loss": 0.3369, "step": 7009 }, { "epoch": 0.48461804355340476, "grad_norm": 0.6433528827719818, "learning_rate": 8.864186548410344e-06, "loss": 0.3043, "step": 7010 }, { "epoch": 0.4846871759419288, "grad_norm": 0.7555412643955319, "learning_rate": 8.863831298641994e-06, "loss": 0.3587, "step": 7011 }, { "epoch": 0.4847563083304528, "grad_norm": 0.6873264270007714, "learning_rate": 8.863476000447415e-06, "loss": 0.4118, "step": 7012 }, { "epoch": 0.4848254407189768, "grad_norm": 0.6465561683927754, "learning_rate": 8.863120653831057e-06, "loss": 0.3134, "step": 7013 }, { "epoch": 0.48489457310750084, "grad_norm": 0.7541449601052285, "learning_rate": 8.862765258797378e-06, "loss": 0.342, "step": 7014 }, { "epoch": 0.4849637054960249, "grad_norm": 0.7114049509331285, "learning_rate": 8.862409815350828e-06, "loss": 0.3875, "step": 7015 }, { "epoch": 0.48503283788454893, "grad_norm": 0.7361853898472397, "learning_rate": 8.862054323495864e-06, "loss": 0.3159, "step": 7016 }, { "epoch": 0.48510197027307295, "grad_norm": 0.6432728762946074, "learning_rate": 8.861698783236943e-06, "loss": 0.3246, "step": 7017 }, { "epoch": 0.48517110266159696, "grad_norm": 0.6374958141182988, "learning_rate": 8.861343194578518e-06, "loss": 0.3178, "step": 7018 }, { "epoch": 0.485240235050121, "grad_norm": 0.7622286099395658, "learning_rate": 8.860987557525048e-06, "loss": 0.4666, "step": 7019 }, { "epoch": 0.485309367438645, "grad_norm": 0.7193505637167229, "learning_rate": 8.860631872080986e-06, "loss": 0.3476, "step": 7020 }, { "epoch": 0.485378499827169, "grad_norm": 0.9257514208916444, "learning_rate": 8.860276138250796e-06, "loss": 0.3504, "step": 7021 }, { "epoch": 0.48544763221569304, "grad_norm": 0.7689987579931505, "learning_rate": 8.859920356038933e-06, "loss": 0.4676, "step": 7022 }, { "epoch": 0.48551676460421705, "grad_norm": 0.7804100980607068, "learning_rate": 8.859564525449856e-06, "loss": 0.3248, "step": 7023 }, { "epoch": 0.4855858969927411, "grad_norm": 0.6757077970754448, "learning_rate": 8.859208646488027e-06, "loss": 0.3622, "step": 7024 }, { "epoch": 0.48565502938126515, "grad_norm": 0.7410949473428525, "learning_rate": 8.858852719157902e-06, "loss": 0.3325, "step": 7025 }, { "epoch": 0.48572416176978916, "grad_norm": 0.7380300741814796, "learning_rate": 8.858496743463948e-06, "loss": 0.4711, "step": 7026 }, { "epoch": 0.4857932941583132, "grad_norm": 0.6888437498170399, "learning_rate": 8.858140719410619e-06, "loss": 0.317, "step": 7027 }, { "epoch": 0.4858624265468372, "grad_norm": 0.754642832336563, "learning_rate": 8.857784647002386e-06, "loss": 0.3304, "step": 7028 }, { "epoch": 0.4859315589353612, "grad_norm": 0.6952104676661234, "learning_rate": 8.857428526243704e-06, "loss": 0.3767, "step": 7029 }, { "epoch": 0.48600069132388524, "grad_norm": 0.7227586815450877, "learning_rate": 8.857072357139039e-06, "loss": 0.3885, "step": 7030 }, { "epoch": 0.48606982371240925, "grad_norm": 0.6614543443615791, "learning_rate": 8.856716139692856e-06, "loss": 0.3008, "step": 7031 }, { "epoch": 0.4861389561009333, "grad_norm": 0.7249879303497995, "learning_rate": 8.85635987390962e-06, "loss": 0.2727, "step": 7032 }, { "epoch": 0.4862080884894573, "grad_norm": 0.7065973343452007, "learning_rate": 8.856003559793793e-06, "loss": 0.3823, "step": 7033 }, { "epoch": 0.4862772208779813, "grad_norm": 0.6912487168815866, "learning_rate": 8.855647197349844e-06, "loss": 0.3538, "step": 7034 }, { "epoch": 0.4863463532665054, "grad_norm": 0.7498959360606604, "learning_rate": 8.855290786582238e-06, "loss": 0.2937, "step": 7035 }, { "epoch": 0.4864154856550294, "grad_norm": 0.8098903912193655, "learning_rate": 8.854934327495442e-06, "loss": 0.3783, "step": 7036 }, { "epoch": 0.4864846180435534, "grad_norm": 0.7582307012193266, "learning_rate": 8.854577820093924e-06, "loss": 0.3765, "step": 7037 }, { "epoch": 0.48655375043207744, "grad_norm": 0.6848929835207546, "learning_rate": 8.85422126438215e-06, "loss": 0.2707, "step": 7038 }, { "epoch": 0.48662288282060145, "grad_norm": 0.999478252669891, "learning_rate": 8.853864660364592e-06, "loss": 0.2982, "step": 7039 }, { "epoch": 0.4866920152091255, "grad_norm": 0.822448675290931, "learning_rate": 8.853508008045716e-06, "loss": 0.4391, "step": 7040 }, { "epoch": 0.4867611475976495, "grad_norm": 0.7495693427643344, "learning_rate": 8.853151307429996e-06, "loss": 0.3412, "step": 7041 }, { "epoch": 0.4868302799861735, "grad_norm": 0.7512229437732677, "learning_rate": 8.852794558521902e-06, "loss": 0.2666, "step": 7042 }, { "epoch": 0.4868994123746975, "grad_norm": 0.7193090321199068, "learning_rate": 8.852437761325903e-06, "loss": 0.3887, "step": 7043 }, { "epoch": 0.48696854476322154, "grad_norm": 0.84791215462381, "learning_rate": 8.852080915846471e-06, "loss": 0.4724, "step": 7044 }, { "epoch": 0.4870376771517456, "grad_norm": 0.8495913936658281, "learning_rate": 8.851724022088077e-06, "loss": 0.354, "step": 7045 }, { "epoch": 0.48710680954026964, "grad_norm": 0.7001854807303473, "learning_rate": 8.851367080055199e-06, "loss": 0.2899, "step": 7046 }, { "epoch": 0.48717594192879365, "grad_norm": 0.7435472680593994, "learning_rate": 8.851010089752309e-06, "loss": 0.4119, "step": 7047 }, { "epoch": 0.4872450743173177, "grad_norm": 0.7715437707841705, "learning_rate": 8.850653051183875e-06, "loss": 0.3553, "step": 7048 }, { "epoch": 0.4873142067058417, "grad_norm": 0.8207019382092866, "learning_rate": 8.85029596435438e-06, "loss": 0.278, "step": 7049 }, { "epoch": 0.4873833390943657, "grad_norm": 0.7730108576951442, "learning_rate": 8.849938829268297e-06, "loss": 0.4082, "step": 7050 }, { "epoch": 0.4874524714828897, "grad_norm": 1.6015155173924578, "learning_rate": 8.849581645930099e-06, "loss": 0.3683, "step": 7051 }, { "epoch": 0.48752160387141374, "grad_norm": 0.8488534929382404, "learning_rate": 8.849224414344266e-06, "loss": 0.3659, "step": 7052 }, { "epoch": 0.48759073625993776, "grad_norm": 0.8660685659956375, "learning_rate": 8.848867134515275e-06, "loss": 0.306, "step": 7053 }, { "epoch": 0.4876598686484618, "grad_norm": 0.723687282648183, "learning_rate": 8.848509806447602e-06, "loss": 0.4311, "step": 7054 }, { "epoch": 0.48772900103698585, "grad_norm": 0.6720686275466142, "learning_rate": 8.848152430145726e-06, "loss": 0.3296, "step": 7055 }, { "epoch": 0.48779813342550987, "grad_norm": 0.6152261155672386, "learning_rate": 8.847795005614127e-06, "loss": 0.291, "step": 7056 }, { "epoch": 0.4878672658140339, "grad_norm": 0.7194977548870197, "learning_rate": 8.847437532857286e-06, "loss": 0.3188, "step": 7057 }, { "epoch": 0.4879363982025579, "grad_norm": 0.7135701891028312, "learning_rate": 8.84708001187968e-06, "loss": 0.4336, "step": 7058 }, { "epoch": 0.4880055305910819, "grad_norm": 0.7204866512109219, "learning_rate": 8.846722442685793e-06, "loss": 0.2902, "step": 7059 }, { "epoch": 0.48807466297960594, "grad_norm": 0.7105437154839878, "learning_rate": 8.846364825280105e-06, "loss": 0.2498, "step": 7060 }, { "epoch": 0.48814379536812996, "grad_norm": 0.7704374657000923, "learning_rate": 8.846007159667097e-06, "loss": 0.3522, "step": 7061 }, { "epoch": 0.488212927756654, "grad_norm": 0.7020779373876747, "learning_rate": 8.845649445851251e-06, "loss": 0.4475, "step": 7062 }, { "epoch": 0.488282060145178, "grad_norm": 0.6836123007770404, "learning_rate": 8.845291683837054e-06, "loss": 0.3224, "step": 7063 }, { "epoch": 0.488351192533702, "grad_norm": 0.8310098532905429, "learning_rate": 8.84493387362899e-06, "loss": 0.3636, "step": 7064 }, { "epoch": 0.4884203249222261, "grad_norm": 0.7846827741323289, "learning_rate": 8.844576015231537e-06, "loss": 0.3846, "step": 7065 }, { "epoch": 0.4884894573107501, "grad_norm": 0.6718478175562049, "learning_rate": 8.844218108649187e-06, "loss": 0.3317, "step": 7066 }, { "epoch": 0.4885585896992741, "grad_norm": 0.6299717272069162, "learning_rate": 8.843860153886422e-06, "loss": 0.2767, "step": 7067 }, { "epoch": 0.48862772208779814, "grad_norm": 0.7314035846833425, "learning_rate": 8.843502150947732e-06, "loss": 0.3401, "step": 7068 }, { "epoch": 0.48869685447632216, "grad_norm": 0.7283683872374346, "learning_rate": 8.8431440998376e-06, "loss": 0.4621, "step": 7069 }, { "epoch": 0.4887659868648462, "grad_norm": 0.6546541947872407, "learning_rate": 8.842786000560517e-06, "loss": 0.2812, "step": 7070 }, { "epoch": 0.4888351192533702, "grad_norm": 0.6850389672004652, "learning_rate": 8.842427853120967e-06, "loss": 0.3092, "step": 7071 }, { "epoch": 0.4889042516418942, "grad_norm": 0.6992962991372292, "learning_rate": 8.842069657523442e-06, "loss": 0.3922, "step": 7072 }, { "epoch": 0.48897338403041823, "grad_norm": 0.7559967571479209, "learning_rate": 8.841711413772429e-06, "loss": 0.4357, "step": 7073 }, { "epoch": 0.48904251641894225, "grad_norm": 0.6752618408316962, "learning_rate": 8.841353121872421e-06, "loss": 0.3758, "step": 7074 }, { "epoch": 0.48911164880746627, "grad_norm": 0.6948988200647794, "learning_rate": 8.840994781827906e-06, "loss": 0.3441, "step": 7075 }, { "epoch": 0.48918078119599034, "grad_norm": 0.7099150224009126, "learning_rate": 8.840636393643376e-06, "loss": 0.439, "step": 7076 }, { "epoch": 0.48924991358451436, "grad_norm": 0.6897718062430273, "learning_rate": 8.840277957323323e-06, "loss": 0.2993, "step": 7077 }, { "epoch": 0.4893190459730384, "grad_norm": 0.7038744023799944, "learning_rate": 8.83991947287224e-06, "loss": 0.3293, "step": 7078 }, { "epoch": 0.4893881783615624, "grad_norm": 0.732925770531463, "learning_rate": 8.839560940294619e-06, "loss": 0.4191, "step": 7079 }, { "epoch": 0.4894573107500864, "grad_norm": 0.7374136603459099, "learning_rate": 8.83920235959495e-06, "loss": 0.4003, "step": 7080 }, { "epoch": 0.48952644313861043, "grad_norm": 0.6412934127798451, "learning_rate": 8.838843730777735e-06, "loss": 0.2694, "step": 7081 }, { "epoch": 0.48959557552713445, "grad_norm": 0.7995228227132278, "learning_rate": 8.838485053847462e-06, "loss": 0.3862, "step": 7082 }, { "epoch": 0.48966470791565847, "grad_norm": 0.7563087518062762, "learning_rate": 8.838126328808631e-06, "loss": 0.3843, "step": 7083 }, { "epoch": 0.4897338403041825, "grad_norm": 0.6972204054675978, "learning_rate": 8.837767555665734e-06, "loss": 0.3079, "step": 7084 }, { "epoch": 0.4898029726927065, "grad_norm": 0.7171234485633534, "learning_rate": 8.83740873442327e-06, "loss": 0.3288, "step": 7085 }, { "epoch": 0.4898721050812306, "grad_norm": 0.6864665572105496, "learning_rate": 8.837049865085736e-06, "loss": 0.3519, "step": 7086 }, { "epoch": 0.4899412374697546, "grad_norm": 0.720818155559665, "learning_rate": 8.83669094765763e-06, "loss": 0.4185, "step": 7087 }, { "epoch": 0.4900103698582786, "grad_norm": 1.5092601291932863, "learning_rate": 8.836331982143449e-06, "loss": 0.3099, "step": 7088 }, { "epoch": 0.49007950224680263, "grad_norm": 0.7175803759271784, "learning_rate": 8.835972968547692e-06, "loss": 0.3563, "step": 7089 }, { "epoch": 0.49014863463532665, "grad_norm": 0.7284733430809027, "learning_rate": 8.835613906874862e-06, "loss": 0.4038, "step": 7090 }, { "epoch": 0.49021776702385067, "grad_norm": 0.7106534410298614, "learning_rate": 8.835254797129455e-06, "loss": 0.311, "step": 7091 }, { "epoch": 0.4902868994123747, "grad_norm": 0.7223231380446498, "learning_rate": 8.834895639315973e-06, "loss": 0.3311, "step": 7092 }, { "epoch": 0.4903560318008987, "grad_norm": 0.666536724841127, "learning_rate": 8.834536433438919e-06, "loss": 0.3259, "step": 7093 }, { "epoch": 0.4904251641894227, "grad_norm": 0.7575675676355376, "learning_rate": 8.834177179502795e-06, "loss": 0.4864, "step": 7094 }, { "epoch": 0.49049429657794674, "grad_norm": 0.8413928972519437, "learning_rate": 8.833817877512098e-06, "loss": 0.3351, "step": 7095 }, { "epoch": 0.4905634289664708, "grad_norm": 0.7329940560401061, "learning_rate": 8.83345852747134e-06, "loss": 0.2981, "step": 7096 }, { "epoch": 0.49063256135499483, "grad_norm": 0.7473399894033388, "learning_rate": 8.833099129385019e-06, "loss": 0.4837, "step": 7097 }, { "epoch": 0.49070169374351885, "grad_norm": 0.6570583588792714, "learning_rate": 8.83273968325764e-06, "loss": 0.2905, "step": 7098 }, { "epoch": 0.49077082613204287, "grad_norm": 0.6639190729282125, "learning_rate": 8.83238018909371e-06, "loss": 0.3001, "step": 7099 }, { "epoch": 0.4908399585205669, "grad_norm": 0.7875811640316887, "learning_rate": 8.832020646897733e-06, "loss": 0.376, "step": 7100 }, { "epoch": 0.4909090909090909, "grad_norm": 0.7005310130670539, "learning_rate": 8.831661056674216e-06, "loss": 0.4142, "step": 7101 }, { "epoch": 0.4909782232976149, "grad_norm": 0.6473478705385707, "learning_rate": 8.831301418427666e-06, "loss": 0.3356, "step": 7102 }, { "epoch": 0.49104735568613894, "grad_norm": 0.6787901219482579, "learning_rate": 8.83094173216259e-06, "loss": 0.2969, "step": 7103 }, { "epoch": 0.49111648807466296, "grad_norm": 0.723372295655662, "learning_rate": 8.830581997883496e-06, "loss": 0.3971, "step": 7104 }, { "epoch": 0.491185620463187, "grad_norm": 0.7263979812182614, "learning_rate": 8.83022221559489e-06, "loss": 0.3745, "step": 7105 }, { "epoch": 0.49125475285171105, "grad_norm": 0.6991207314180884, "learning_rate": 8.829862385301287e-06, "loss": 0.3594, "step": 7106 }, { "epoch": 0.49132388524023507, "grad_norm": 1.2287707136284367, "learning_rate": 8.829502507007193e-06, "loss": 0.403, "step": 7107 }, { "epoch": 0.4913930176287591, "grad_norm": 0.7100025935893856, "learning_rate": 8.829142580717118e-06, "loss": 0.4331, "step": 7108 }, { "epoch": 0.4914621500172831, "grad_norm": 2.6670819103127537, "learning_rate": 8.828782606435575e-06, "loss": 0.3062, "step": 7109 }, { "epoch": 0.4915312824058071, "grad_norm": 0.6897285458027286, "learning_rate": 8.828422584167075e-06, "loss": 0.2756, "step": 7110 }, { "epoch": 0.49160041479433114, "grad_norm": 0.810551458095822, "learning_rate": 8.82806251391613e-06, "loss": 0.364, "step": 7111 }, { "epoch": 0.49166954718285516, "grad_norm": 0.7327064887132347, "learning_rate": 8.827702395687253e-06, "loss": 0.4185, "step": 7112 }, { "epoch": 0.4917386795713792, "grad_norm": 0.7922662139812967, "learning_rate": 8.827342229484957e-06, "loss": 0.3551, "step": 7113 }, { "epoch": 0.4918078119599032, "grad_norm": 0.8974682964659177, "learning_rate": 8.826982015313757e-06, "loss": 0.319, "step": 7114 }, { "epoch": 0.4918769443484272, "grad_norm": 0.7562950888912281, "learning_rate": 8.826621753178167e-06, "loss": 0.3997, "step": 7115 }, { "epoch": 0.4919460767369513, "grad_norm": 0.7212452146955767, "learning_rate": 8.8262614430827e-06, "loss": 0.3444, "step": 7116 }, { "epoch": 0.4920152091254753, "grad_norm": 0.8875358914265895, "learning_rate": 8.825901085031876e-06, "loss": 0.2848, "step": 7117 }, { "epoch": 0.4920843415139993, "grad_norm": 0.7433918946805139, "learning_rate": 8.82554067903021e-06, "loss": 0.3743, "step": 7118 }, { "epoch": 0.49215347390252334, "grad_norm": 0.6859802805862434, "learning_rate": 8.825180225082218e-06, "loss": 0.4244, "step": 7119 }, { "epoch": 0.49222260629104736, "grad_norm": 0.7218621309032026, "learning_rate": 8.824819723192417e-06, "loss": 0.2835, "step": 7120 }, { "epoch": 0.4922917386795714, "grad_norm": 0.6183404664221662, "learning_rate": 8.824459173365328e-06, "loss": 0.3131, "step": 7121 }, { "epoch": 0.4923608710680954, "grad_norm": 0.6909881673036463, "learning_rate": 8.824098575605468e-06, "loss": 0.4538, "step": 7122 }, { "epoch": 0.4924300034566194, "grad_norm": 0.8079055899217253, "learning_rate": 8.823737929917355e-06, "loss": 0.3772, "step": 7123 }, { "epoch": 0.49249913584514343, "grad_norm": 0.750910804547806, "learning_rate": 8.823377236305514e-06, "loss": 0.3301, "step": 7124 }, { "epoch": 0.49256826823366745, "grad_norm": 0.7229745798513201, "learning_rate": 8.82301649477446e-06, "loss": 0.2926, "step": 7125 }, { "epoch": 0.4926374006221915, "grad_norm": 0.6872851344386369, "learning_rate": 8.822655705328717e-06, "loss": 0.3806, "step": 7126 }, { "epoch": 0.49270653301071554, "grad_norm": 0.7403458496958606, "learning_rate": 8.822294867972807e-06, "loss": 0.333, "step": 7127 }, { "epoch": 0.49277566539923956, "grad_norm": 0.6266937874629585, "learning_rate": 8.821933982711252e-06, "loss": 0.302, "step": 7128 }, { "epoch": 0.4928447977877636, "grad_norm": 0.735149781015028, "learning_rate": 8.821573049548576e-06, "loss": 0.4333, "step": 7129 }, { "epoch": 0.4929139301762876, "grad_norm": 0.6674194117015939, "learning_rate": 8.821212068489301e-06, "loss": 0.3604, "step": 7130 }, { "epoch": 0.4929830625648116, "grad_norm": 0.6818896669343472, "learning_rate": 8.820851039537953e-06, "loss": 0.2565, "step": 7131 }, { "epoch": 0.49305219495333563, "grad_norm": 0.9600175308410597, "learning_rate": 8.820489962699053e-06, "loss": 0.3859, "step": 7132 }, { "epoch": 0.49312132734185965, "grad_norm": 0.7847887482469247, "learning_rate": 8.820128837977132e-06, "loss": 0.4199, "step": 7133 }, { "epoch": 0.49319045973038367, "grad_norm": 2.707515430703905, "learning_rate": 8.819767665376713e-06, "loss": 0.359, "step": 7134 }, { "epoch": 0.4932595921189077, "grad_norm": 0.6608110410253474, "learning_rate": 8.819406444902321e-06, "loss": 0.282, "step": 7135 }, { "epoch": 0.49332872450743176, "grad_norm": 0.741710448225196, "learning_rate": 8.819045176558488e-06, "loss": 0.3911, "step": 7136 }, { "epoch": 0.4933978568959558, "grad_norm": 0.7632942807076096, "learning_rate": 8.818683860349738e-06, "loss": 0.4223, "step": 7137 }, { "epoch": 0.4934669892844798, "grad_norm": 0.7034956210411893, "learning_rate": 8.8183224962806e-06, "loss": 0.3977, "step": 7138 }, { "epoch": 0.4935361216730038, "grad_norm": 0.6863272007112592, "learning_rate": 8.817961084355604e-06, "loss": 0.3509, "step": 7139 }, { "epoch": 0.49360525406152783, "grad_norm": 0.7737395310552606, "learning_rate": 8.817599624579279e-06, "loss": 0.3662, "step": 7140 }, { "epoch": 0.49367438645005185, "grad_norm": 0.6969772895152875, "learning_rate": 8.817238116956154e-06, "loss": 0.3492, "step": 7141 }, { "epoch": 0.49374351883857587, "grad_norm": 0.6733361456319299, "learning_rate": 8.816876561490764e-06, "loss": 0.2701, "step": 7142 }, { "epoch": 0.4938126512270999, "grad_norm": 0.6931910443283273, "learning_rate": 8.816514958187637e-06, "loss": 0.3743, "step": 7143 }, { "epoch": 0.4938817836156239, "grad_norm": 0.7243595811649918, "learning_rate": 8.816153307051305e-06, "loss": 0.3958, "step": 7144 }, { "epoch": 0.4939509160041479, "grad_norm": 0.8257562109174447, "learning_rate": 8.815791608086303e-06, "loss": 0.3087, "step": 7145 }, { "epoch": 0.49402004839267194, "grad_norm": 0.6795185426828068, "learning_rate": 8.815429861297163e-06, "loss": 0.3286, "step": 7146 }, { "epoch": 0.494089180781196, "grad_norm": 0.6685075003794637, "learning_rate": 8.815068066688417e-06, "loss": 0.3698, "step": 7147 }, { "epoch": 0.49415831316972003, "grad_norm": 3.295883571209638, "learning_rate": 8.814706224264602e-06, "loss": 0.3526, "step": 7148 }, { "epoch": 0.49422744555824405, "grad_norm": 0.5916722587060558, "learning_rate": 8.814344334030251e-06, "loss": 0.2682, "step": 7149 }, { "epoch": 0.49429657794676807, "grad_norm": 0.8829293628723415, "learning_rate": 8.813982395989902e-06, "loss": 0.328, "step": 7150 }, { "epoch": 0.4943657103352921, "grad_norm": 0.7125488247340717, "learning_rate": 8.81362041014809e-06, "loss": 0.4072, "step": 7151 }, { "epoch": 0.4944348427238161, "grad_norm": 0.8287135830746595, "learning_rate": 8.813258376509353e-06, "loss": 0.3011, "step": 7152 }, { "epoch": 0.4945039751123401, "grad_norm": 0.9649647204584445, "learning_rate": 8.812896295078224e-06, "loss": 0.3113, "step": 7153 }, { "epoch": 0.49457310750086414, "grad_norm": 0.7702000790974532, "learning_rate": 8.812534165859248e-06, "loss": 0.4261, "step": 7154 }, { "epoch": 0.49464223988938816, "grad_norm": 0.7204298887068984, "learning_rate": 8.812171988856956e-06, "loss": 0.4218, "step": 7155 }, { "epoch": 0.4947113722779122, "grad_norm": 0.6700966777590373, "learning_rate": 8.811809764075894e-06, "loss": 0.2897, "step": 7156 }, { "epoch": 0.49478050466643625, "grad_norm": 0.7311620640447385, "learning_rate": 8.811447491520599e-06, "loss": 0.3137, "step": 7157 }, { "epoch": 0.49484963705496027, "grad_norm": 0.7393167885794478, "learning_rate": 8.811085171195611e-06, "loss": 0.4409, "step": 7158 }, { "epoch": 0.4949187694434843, "grad_norm": 0.9393034259528968, "learning_rate": 8.810722803105472e-06, "loss": 0.311, "step": 7159 }, { "epoch": 0.4949879018320083, "grad_norm": 0.8549571635390476, "learning_rate": 8.810360387254722e-06, "loss": 0.3358, "step": 7160 }, { "epoch": 0.4950570342205323, "grad_norm": 1.4333103898135986, "learning_rate": 8.809997923647906e-06, "loss": 0.3267, "step": 7161 }, { "epoch": 0.49512616660905634, "grad_norm": 0.7726722357452447, "learning_rate": 8.809635412289565e-06, "loss": 0.4039, "step": 7162 }, { "epoch": 0.49519529899758036, "grad_norm": 0.6363704690753401, "learning_rate": 8.809272853184243e-06, "loss": 0.2892, "step": 7163 }, { "epoch": 0.4952644313861044, "grad_norm": 0.7705965047580364, "learning_rate": 8.808910246336482e-06, "loss": 0.3057, "step": 7164 }, { "epoch": 0.4953335637746284, "grad_norm": 0.7640362908530474, "learning_rate": 8.80854759175083e-06, "loss": 0.4487, "step": 7165 }, { "epoch": 0.4954026961631524, "grad_norm": 0.7182065569434497, "learning_rate": 8.80818488943183e-06, "loss": 0.3324, "step": 7166 }, { "epoch": 0.4954718285516765, "grad_norm": 0.7915862856675194, "learning_rate": 8.807822139384028e-06, "loss": 0.3092, "step": 7167 }, { "epoch": 0.4955409609402005, "grad_norm": 0.8706887342533348, "learning_rate": 8.807459341611972e-06, "loss": 0.3373, "step": 7168 }, { "epoch": 0.4956100933287245, "grad_norm": 0.716055512658913, "learning_rate": 8.807096496120206e-06, "loss": 0.4171, "step": 7169 }, { "epoch": 0.49567922571724854, "grad_norm": 0.7351983990741874, "learning_rate": 8.806733602913282e-06, "loss": 0.2998, "step": 7170 }, { "epoch": 0.49574835810577256, "grad_norm": 0.893426146186369, "learning_rate": 8.806370661995744e-06, "loss": 0.3991, "step": 7171 }, { "epoch": 0.4958174904942966, "grad_norm": 0.7460516024370328, "learning_rate": 8.806007673372142e-06, "loss": 0.4105, "step": 7172 }, { "epoch": 0.4958866228828206, "grad_norm": 0.6841136278918841, "learning_rate": 8.805644637047026e-06, "loss": 0.3827, "step": 7173 }, { "epoch": 0.4959557552713446, "grad_norm": 0.8199077652496649, "learning_rate": 8.805281553024948e-06, "loss": 0.3522, "step": 7174 }, { "epoch": 0.49602488765986863, "grad_norm": 0.8501307544061234, "learning_rate": 8.804918421310456e-06, "loss": 0.3392, "step": 7175 }, { "epoch": 0.49609402004839265, "grad_norm": 0.6920204055527338, "learning_rate": 8.8045552419081e-06, "loss": 0.4093, "step": 7176 }, { "epoch": 0.4961631524369167, "grad_norm": 0.965548729778015, "learning_rate": 8.804192014822435e-06, "loss": 0.2661, "step": 7177 }, { "epoch": 0.49623228482544074, "grad_norm": 0.7189878788792604, "learning_rate": 8.80382874005801e-06, "loss": 0.2432, "step": 7178 }, { "epoch": 0.49630141721396476, "grad_norm": 0.7908541505765023, "learning_rate": 8.803465417619383e-06, "loss": 0.3923, "step": 7179 }, { "epoch": 0.4963705496024888, "grad_norm": 0.733999552052557, "learning_rate": 8.803102047511104e-06, "loss": 0.4381, "step": 7180 }, { "epoch": 0.4964396819910128, "grad_norm": 0.9361661733797244, "learning_rate": 8.802738629737726e-06, "loss": 0.2925, "step": 7181 }, { "epoch": 0.4965088143795368, "grad_norm": 0.9347521001050568, "learning_rate": 8.802375164303806e-06, "loss": 0.3202, "step": 7182 }, { "epoch": 0.49657794676806083, "grad_norm": 0.6693981064263086, "learning_rate": 8.8020116512139e-06, "loss": 0.4508, "step": 7183 }, { "epoch": 0.49664707915658485, "grad_norm": 0.7026595339223397, "learning_rate": 8.801648090472563e-06, "loss": 0.315, "step": 7184 }, { "epoch": 0.49671621154510887, "grad_norm": 0.699040361741338, "learning_rate": 8.801284482084351e-06, "loss": 0.2795, "step": 7185 }, { "epoch": 0.4967853439336329, "grad_norm": 1.0442365894833048, "learning_rate": 8.800920826053822e-06, "loss": 0.4, "step": 7186 }, { "epoch": 0.49685447632215696, "grad_norm": 0.7296613015520171, "learning_rate": 8.800557122385533e-06, "loss": 0.3847, "step": 7187 }, { "epoch": 0.496923608710681, "grad_norm": 1.0114630479395545, "learning_rate": 8.800193371084043e-06, "loss": 0.3971, "step": 7188 }, { "epoch": 0.496992741099205, "grad_norm": 0.7975693620437648, "learning_rate": 8.799829572153911e-06, "loss": 0.3206, "step": 7189 }, { "epoch": 0.497061873487729, "grad_norm": 0.8033657317396065, "learning_rate": 8.799465725599697e-06, "loss": 0.4439, "step": 7190 }, { "epoch": 0.49713100587625303, "grad_norm": 0.7382819559158703, "learning_rate": 8.79910183142596e-06, "loss": 0.3436, "step": 7191 }, { "epoch": 0.49720013826477705, "grad_norm": 0.833271207245566, "learning_rate": 8.798737889637262e-06, "loss": 0.325, "step": 7192 }, { "epoch": 0.49726927065330107, "grad_norm": 0.7435079296535043, "learning_rate": 8.798373900238161e-06, "loss": 0.3278, "step": 7193 }, { "epoch": 0.4973384030418251, "grad_norm": 0.7756779252632101, "learning_rate": 8.798009863233224e-06, "loss": 0.4029, "step": 7194 }, { "epoch": 0.4974075354303491, "grad_norm": 0.8010587853038862, "learning_rate": 8.79764577862701e-06, "loss": 0.3083, "step": 7195 }, { "epoch": 0.4974766678188731, "grad_norm": 0.677821419939719, "learning_rate": 8.797281646424086e-06, "loss": 0.3413, "step": 7196 }, { "epoch": 0.4975458002073972, "grad_norm": 0.7151440359481972, "learning_rate": 8.796917466629011e-06, "loss": 0.424, "step": 7197 }, { "epoch": 0.4976149325959212, "grad_norm": 0.7427660229824969, "learning_rate": 8.796553239246351e-06, "loss": 0.4074, "step": 7198 }, { "epoch": 0.49768406498444523, "grad_norm": 0.7685626602908708, "learning_rate": 8.796188964280672e-06, "loss": 0.3396, "step": 7199 }, { "epoch": 0.49775319737296925, "grad_norm": 0.8223133623686608, "learning_rate": 8.795824641736539e-06, "loss": 0.373, "step": 7200 }, { "epoch": 0.49782232976149327, "grad_norm": 0.7160914062168652, "learning_rate": 8.795460271618519e-06, "loss": 0.4144, "step": 7201 }, { "epoch": 0.4978914621500173, "grad_norm": 0.7336776026169307, "learning_rate": 8.795095853931176e-06, "loss": 0.3099, "step": 7202 }, { "epoch": 0.4979605945385413, "grad_norm": 0.6545042664204672, "learning_rate": 8.794731388679081e-06, "loss": 0.3289, "step": 7203 }, { "epoch": 0.4980297269270653, "grad_norm": 0.7198968954134549, "learning_rate": 8.7943668758668e-06, "loss": 0.3876, "step": 7204 }, { "epoch": 0.49809885931558934, "grad_norm": 0.6998824389464534, "learning_rate": 8.7940023154989e-06, "loss": 0.3806, "step": 7205 }, { "epoch": 0.49816799170411336, "grad_norm": 0.7986064506393385, "learning_rate": 8.793637707579951e-06, "loss": 0.2942, "step": 7206 }, { "epoch": 0.4982371240926374, "grad_norm": 0.715600320420797, "learning_rate": 8.793273052114526e-06, "loss": 0.3424, "step": 7207 }, { "epoch": 0.49830625648116145, "grad_norm": 0.6960231987246661, "learning_rate": 8.792908349107193e-06, "loss": 0.4588, "step": 7208 }, { "epoch": 0.49837538886968547, "grad_norm": 0.8286630992750051, "learning_rate": 8.792543598562519e-06, "loss": 0.2741, "step": 7209 }, { "epoch": 0.4984445212582095, "grad_norm": 0.7398966673179956, "learning_rate": 8.792178800485082e-06, "loss": 0.2846, "step": 7210 }, { "epoch": 0.4985136536467335, "grad_norm": 0.6814587400220827, "learning_rate": 8.79181395487945e-06, "loss": 0.356, "step": 7211 }, { "epoch": 0.4985827860352575, "grad_norm": 0.9054928897710399, "learning_rate": 8.791449061750197e-06, "loss": 0.398, "step": 7212 }, { "epoch": 0.49865191842378154, "grad_norm": 0.840819081355012, "learning_rate": 8.791084121101895e-06, "loss": 0.3233, "step": 7213 }, { "epoch": 0.49872105081230556, "grad_norm": 0.73868336188029, "learning_rate": 8.790719132939123e-06, "loss": 0.3411, "step": 7214 }, { "epoch": 0.4987901832008296, "grad_norm": 0.7439224137673585, "learning_rate": 8.790354097266447e-06, "loss": 0.4243, "step": 7215 }, { "epoch": 0.4988593155893536, "grad_norm": 0.7441286056767734, "learning_rate": 8.789989014088449e-06, "loss": 0.3124, "step": 7216 }, { "epoch": 0.4989284479778776, "grad_norm": 0.8116394422310349, "learning_rate": 8.789623883409703e-06, "loss": 0.2692, "step": 7217 }, { "epoch": 0.4989975803664017, "grad_norm": 0.8153995236929303, "learning_rate": 8.789258705234783e-06, "loss": 0.4158, "step": 7218 }, { "epoch": 0.4990667127549257, "grad_norm": 0.7401025028227954, "learning_rate": 8.788893479568268e-06, "loss": 0.4505, "step": 7219 }, { "epoch": 0.4991358451434497, "grad_norm": 0.6632680137465278, "learning_rate": 8.788528206414733e-06, "loss": 0.255, "step": 7220 }, { "epoch": 0.49920497753197374, "grad_norm": 0.6540570877455489, "learning_rate": 8.78816288577876e-06, "loss": 0.2823, "step": 7221 }, { "epoch": 0.49927410992049776, "grad_norm": 0.6806868574096684, "learning_rate": 8.787797517664924e-06, "loss": 0.3819, "step": 7222 }, { "epoch": 0.4993432423090218, "grad_norm": 0.7917038292355716, "learning_rate": 8.787432102077808e-06, "loss": 0.3297, "step": 7223 }, { "epoch": 0.4994123746975458, "grad_norm": 0.7093885140682704, "learning_rate": 8.787066639021987e-06, "loss": 0.3247, "step": 7224 }, { "epoch": 0.4994815070860698, "grad_norm": 1.0040998274112145, "learning_rate": 8.786701128502045e-06, "loss": 0.3347, "step": 7225 }, { "epoch": 0.4995506394745938, "grad_norm": 0.7068745180469804, "learning_rate": 8.786335570522564e-06, "loss": 0.4778, "step": 7226 }, { "epoch": 0.49961977186311785, "grad_norm": 0.7008164654881484, "learning_rate": 8.78596996508812e-06, "loss": 0.3204, "step": 7227 }, { "epoch": 0.4996889042516419, "grad_norm": 0.7381029687187919, "learning_rate": 8.785604312203302e-06, "loss": 0.3022, "step": 7228 }, { "epoch": 0.49975803664016594, "grad_norm": 0.7493231394241221, "learning_rate": 8.785238611872687e-06, "loss": 0.4181, "step": 7229 }, { "epoch": 0.49982716902868995, "grad_norm": 0.701340128767454, "learning_rate": 8.784872864100862e-06, "loss": 0.3538, "step": 7230 }, { "epoch": 0.499896301417214, "grad_norm": 0.6868314909345957, "learning_rate": 8.784507068892411e-06, "loss": 0.3193, "step": 7231 }, { "epoch": 0.499965433805738, "grad_norm": 0.733386361739376, "learning_rate": 8.784141226251916e-06, "loss": 0.3326, "step": 7232 }, { "epoch": 0.500034566194262, "grad_norm": 0.8329335858409125, "learning_rate": 8.783775336183965e-06, "loss": 0.414, "step": 7233 }, { "epoch": 0.500103698582786, "grad_norm": 0.7780847197008526, "learning_rate": 8.783409398693144e-06, "loss": 0.3022, "step": 7234 }, { "epoch": 0.50017283097131, "grad_norm": 0.7393196456360291, "learning_rate": 8.783043413784036e-06, "loss": 0.2808, "step": 7235 }, { "epoch": 0.5002419633598341, "grad_norm": 0.684943393109104, "learning_rate": 8.78267738146123e-06, "loss": 0.3556, "step": 7236 }, { "epoch": 0.5003110957483581, "grad_norm": 0.7059934723304699, "learning_rate": 8.782311301729315e-06, "loss": 0.3871, "step": 7237 }, { "epoch": 0.5003802281368821, "grad_norm": 0.7372527928622089, "learning_rate": 8.781945174592877e-06, "loss": 0.3218, "step": 7238 }, { "epoch": 0.5004493605254061, "grad_norm": 0.7722314320475572, "learning_rate": 8.781579000056506e-06, "loss": 0.3382, "step": 7239 }, { "epoch": 0.5005184929139301, "grad_norm": 0.7762424765131486, "learning_rate": 8.781212778124789e-06, "loss": 0.4308, "step": 7240 }, { "epoch": 0.5005876253024542, "grad_norm": 1.1640703129338927, "learning_rate": 8.780846508802319e-06, "loss": 0.3171, "step": 7241 }, { "epoch": 0.5006567576909782, "grad_norm": 0.6746314294640773, "learning_rate": 8.780480192093683e-06, "loss": 0.3155, "step": 7242 }, { "epoch": 0.5007258900795023, "grad_norm": 0.6931014539870505, "learning_rate": 8.780113828003476e-06, "loss": 0.4411, "step": 7243 }, { "epoch": 0.5007950224680263, "grad_norm": 0.8136752503532187, "learning_rate": 8.779747416536289e-06, "loss": 0.4265, "step": 7244 }, { "epoch": 0.5008641548565503, "grad_norm": 0.7093373803517885, "learning_rate": 8.779380957696712e-06, "loss": 0.3314, "step": 7245 }, { "epoch": 0.5009332872450744, "grad_norm": 0.7598363123192301, "learning_rate": 8.77901445148934e-06, "loss": 0.3518, "step": 7246 }, { "epoch": 0.5010024196335984, "grad_norm": 0.8012146649512576, "learning_rate": 8.778647897918766e-06, "loss": 0.4202, "step": 7247 }, { "epoch": 0.5010715520221224, "grad_norm": 0.6656500052086619, "learning_rate": 8.778281296989584e-06, "loss": 0.3539, "step": 7248 }, { "epoch": 0.5011406844106464, "grad_norm": 0.7108203117895703, "learning_rate": 8.777914648706387e-06, "loss": 0.3274, "step": 7249 }, { "epoch": 0.5012098167991704, "grad_norm": 0.8215272259684662, "learning_rate": 8.777547953073775e-06, "loss": 0.3742, "step": 7250 }, { "epoch": 0.5012789491876944, "grad_norm": 0.6665590072083067, "learning_rate": 8.777181210096339e-06, "loss": 0.3957, "step": 7251 }, { "epoch": 0.5013480815762185, "grad_norm": 0.672054184406966, "learning_rate": 8.776814419778677e-06, "loss": 0.2972, "step": 7252 }, { "epoch": 0.5014172139647425, "grad_norm": 0.6061553830913171, "learning_rate": 8.776447582125385e-06, "loss": 0.2879, "step": 7253 }, { "epoch": 0.5014863463532665, "grad_norm": 0.6678748740864803, "learning_rate": 8.776080697141063e-06, "loss": 0.4049, "step": 7254 }, { "epoch": 0.5015554787417905, "grad_norm": 0.8339897005191959, "learning_rate": 8.77571376483031e-06, "loss": 0.4232, "step": 7255 }, { "epoch": 0.5016246111303145, "grad_norm": 0.6825507566577897, "learning_rate": 8.77534678519772e-06, "loss": 0.2938, "step": 7256 }, { "epoch": 0.5016937435188386, "grad_norm": 0.739491761201439, "learning_rate": 8.774979758247896e-06, "loss": 0.3631, "step": 7257 }, { "epoch": 0.5017628759073626, "grad_norm": 0.693043971628971, "learning_rate": 8.77461268398544e-06, "loss": 0.3677, "step": 7258 }, { "epoch": 0.5018320082958866, "grad_norm": 0.7692232829560351, "learning_rate": 8.774245562414948e-06, "loss": 0.3171, "step": 7259 }, { "epoch": 0.5019011406844106, "grad_norm": 0.9766332331072791, "learning_rate": 8.773878393541023e-06, "loss": 0.2603, "step": 7260 }, { "epoch": 0.5019702730729346, "grad_norm": 0.6990958656191679, "learning_rate": 8.773511177368269e-06, "loss": 0.3478, "step": 7261 }, { "epoch": 0.5020394054614586, "grad_norm": 0.6956717527671069, "learning_rate": 8.773143913901286e-06, "loss": 0.3932, "step": 7262 }, { "epoch": 0.5021085378499828, "grad_norm": 0.7478889405691924, "learning_rate": 8.772776603144677e-06, "loss": 0.3205, "step": 7263 }, { "epoch": 0.5021776702385068, "grad_norm": 0.6902830000333977, "learning_rate": 8.772409245103046e-06, "loss": 0.3415, "step": 7264 }, { "epoch": 0.5022468026270308, "grad_norm": 0.965190461220723, "learning_rate": 8.772041839780997e-06, "loss": 0.4446, "step": 7265 }, { "epoch": 0.5023159350155548, "grad_norm": 0.6782695509994595, "learning_rate": 8.771674387183136e-06, "loss": 0.3412, "step": 7266 }, { "epoch": 0.5023850674040788, "grad_norm": 0.6549934366540211, "learning_rate": 8.771306887314067e-06, "loss": 0.313, "step": 7267 }, { "epoch": 0.5024541997926029, "grad_norm": 1.2240945935070682, "learning_rate": 8.770939340178395e-06, "loss": 0.3756, "step": 7268 }, { "epoch": 0.5025233321811269, "grad_norm": 0.7318901128254321, "learning_rate": 8.770571745780729e-06, "loss": 0.4363, "step": 7269 }, { "epoch": 0.5025924645696509, "grad_norm": 0.6618603239447004, "learning_rate": 8.770204104125676e-06, "loss": 0.3022, "step": 7270 }, { "epoch": 0.5026615969581749, "grad_norm": 0.7974141546200131, "learning_rate": 8.769836415217842e-06, "loss": 0.3801, "step": 7271 }, { "epoch": 0.5027307293466989, "grad_norm": 0.7790158012986538, "learning_rate": 8.769468679061835e-06, "loss": 0.4112, "step": 7272 }, { "epoch": 0.502799861735223, "grad_norm": 0.6634404467821335, "learning_rate": 8.769100895662266e-06, "loss": 0.3551, "step": 7273 }, { "epoch": 0.502868994123747, "grad_norm": 0.659290178616603, "learning_rate": 8.768733065023744e-06, "loss": 0.2983, "step": 7274 }, { "epoch": 0.502938126512271, "grad_norm": 0.6993408954844083, "learning_rate": 8.768365187150878e-06, "loss": 0.3272, "step": 7275 }, { "epoch": 0.503007258900795, "grad_norm": 0.8327316668705996, "learning_rate": 8.76799726204828e-06, "loss": 0.4536, "step": 7276 }, { "epoch": 0.503076391289319, "grad_norm": 0.6643114801463357, "learning_rate": 8.76762928972056e-06, "loss": 0.2756, "step": 7277 }, { "epoch": 0.503145523677843, "grad_norm": 0.7270196383876775, "learning_rate": 8.76726127017233e-06, "loss": 0.2804, "step": 7278 }, { "epoch": 0.5032146560663671, "grad_norm": 0.6668187705767218, "learning_rate": 8.766893203408205e-06, "loss": 0.3582, "step": 7279 }, { "epoch": 0.5032837884548911, "grad_norm": 0.8273251483851083, "learning_rate": 8.766525089432795e-06, "loss": 0.3625, "step": 7280 }, { "epoch": 0.5033529208434151, "grad_norm": 0.7588155611827125, "learning_rate": 8.766156928250714e-06, "loss": 0.2953, "step": 7281 }, { "epoch": 0.5034220532319391, "grad_norm": 0.8874114363909147, "learning_rate": 8.765788719866577e-06, "loss": 0.3525, "step": 7282 }, { "epoch": 0.5034911856204631, "grad_norm": 0.7154583755272775, "learning_rate": 8.765420464285e-06, "loss": 0.4363, "step": 7283 }, { "epoch": 0.5035603180089873, "grad_norm": 0.8324307286976007, "learning_rate": 8.765052161510595e-06, "loss": 0.2996, "step": 7284 }, { "epoch": 0.5036294503975113, "grad_norm": 1.0686243608537715, "learning_rate": 8.764683811547982e-06, "loss": 0.2776, "step": 7285 }, { "epoch": 0.5036985827860353, "grad_norm": 0.892036724957653, "learning_rate": 8.764315414401773e-06, "loss": 0.4061, "step": 7286 }, { "epoch": 0.5037677151745593, "grad_norm": 0.6650436156070698, "learning_rate": 8.76394697007659e-06, "loss": 0.4192, "step": 7287 }, { "epoch": 0.5038368475630833, "grad_norm": 0.6492793837808013, "learning_rate": 8.76357847857705e-06, "loss": 0.2914, "step": 7288 }, { "epoch": 0.5039059799516074, "grad_norm": 0.7356652453893201, "learning_rate": 8.763209939907771e-06, "loss": 0.3106, "step": 7289 }, { "epoch": 0.5039751123401314, "grad_norm": 0.7213031149167579, "learning_rate": 8.762841354073368e-06, "loss": 0.4074, "step": 7290 }, { "epoch": 0.5040442447286554, "grad_norm": 0.9690170526829384, "learning_rate": 8.762472721078465e-06, "loss": 0.2743, "step": 7291 }, { "epoch": 0.5041133771171794, "grad_norm": 0.6978709413620858, "learning_rate": 8.762104040927681e-06, "loss": 0.3331, "step": 7292 }, { "epoch": 0.5041825095057034, "grad_norm": 0.8023871770110412, "learning_rate": 8.761735313625636e-06, "loss": 0.4102, "step": 7293 }, { "epoch": 0.5042516418942274, "grad_norm": 0.9312667981737194, "learning_rate": 8.761366539176953e-06, "loss": 0.4301, "step": 7294 }, { "epoch": 0.5043207742827515, "grad_norm": 0.7484662997672593, "learning_rate": 8.760997717586253e-06, "loss": 0.2957, "step": 7295 }, { "epoch": 0.5043899066712755, "grad_norm": 0.7376508372484077, "learning_rate": 8.760628848858156e-06, "loss": 0.3768, "step": 7296 }, { "epoch": 0.5044590390597995, "grad_norm": 0.7134613644371611, "learning_rate": 8.76025993299729e-06, "loss": 0.3725, "step": 7297 }, { "epoch": 0.5045281714483235, "grad_norm": 0.6855105370511153, "learning_rate": 8.759890970008275e-06, "loss": 0.305, "step": 7298 }, { "epoch": 0.5045973038368475, "grad_norm": 0.7075680666426897, "learning_rate": 8.759521959895738e-06, "loss": 0.2863, "step": 7299 }, { "epoch": 0.5046664362253716, "grad_norm": 0.7642799688528372, "learning_rate": 8.759152902664302e-06, "loss": 0.3325, "step": 7300 }, { "epoch": 0.5047355686138956, "grad_norm": 0.8309207148960447, "learning_rate": 8.758783798318592e-06, "loss": 0.4356, "step": 7301 }, { "epoch": 0.5048047010024196, "grad_norm": 0.7021187285421511, "learning_rate": 8.758414646863235e-06, "loss": 0.3374, "step": 7302 }, { "epoch": 0.5048738333909436, "grad_norm": 0.8585895812275387, "learning_rate": 8.75804544830286e-06, "loss": 0.3177, "step": 7303 }, { "epoch": 0.5049429657794677, "grad_norm": 0.7346040708550626, "learning_rate": 8.757676202642087e-06, "loss": 0.4224, "step": 7304 }, { "epoch": 0.5050120981679918, "grad_norm": 0.6933910281611378, "learning_rate": 8.757306909885553e-06, "loss": 0.3834, "step": 7305 }, { "epoch": 0.5050812305565158, "grad_norm": 0.713289381052098, "learning_rate": 8.75693757003788e-06, "loss": 0.3044, "step": 7306 }, { "epoch": 0.5051503629450398, "grad_norm": 0.7197363759795828, "learning_rate": 8.7565681831037e-06, "loss": 0.2714, "step": 7307 }, { "epoch": 0.5052194953335638, "grad_norm": 0.7903263286624624, "learning_rate": 8.756198749087641e-06, "loss": 0.4734, "step": 7308 }, { "epoch": 0.5052886277220878, "grad_norm": 0.7255260292442629, "learning_rate": 8.755829267994334e-06, "loss": 0.3071, "step": 7309 }, { "epoch": 0.5053577601106118, "grad_norm": 0.7055619277532404, "learning_rate": 8.75545973982841e-06, "loss": 0.2973, "step": 7310 }, { "epoch": 0.5054268924991359, "grad_norm": 2.5663616452066536, "learning_rate": 8.7550901645945e-06, "loss": 0.3529, "step": 7311 }, { "epoch": 0.5054960248876599, "grad_norm": 0.7401036159946018, "learning_rate": 8.754720542297236e-06, "loss": 0.4278, "step": 7312 }, { "epoch": 0.5055651572761839, "grad_norm": 0.6745141391181038, "learning_rate": 8.754350872941251e-06, "loss": 0.3237, "step": 7313 }, { "epoch": 0.5056342896647079, "grad_norm": 0.673238211997469, "learning_rate": 8.753981156531178e-06, "loss": 0.3636, "step": 7314 }, { "epoch": 0.5057034220532319, "grad_norm": 0.7076394099908316, "learning_rate": 8.753611393071648e-06, "loss": 0.4402, "step": 7315 }, { "epoch": 0.505772554441756, "grad_norm": 0.7345425367900571, "learning_rate": 8.7532415825673e-06, "loss": 0.3176, "step": 7316 }, { "epoch": 0.50584168683028, "grad_norm": 0.6611990462019194, "learning_rate": 8.752871725022767e-06, "loss": 0.2818, "step": 7317 }, { "epoch": 0.505910819218804, "grad_norm": 0.7764299565041518, "learning_rate": 8.752501820442683e-06, "loss": 0.315, "step": 7318 }, { "epoch": 0.505979951607328, "grad_norm": 0.7365058019540414, "learning_rate": 8.752131868831686e-06, "loss": 0.4069, "step": 7319 }, { "epoch": 0.506049083995852, "grad_norm": 0.7223406871526461, "learning_rate": 8.751761870194412e-06, "loss": 0.3486, "step": 7320 }, { "epoch": 0.506118216384376, "grad_norm": 0.6675985877571308, "learning_rate": 8.751391824535497e-06, "loss": 0.372, "step": 7321 }, { "epoch": 0.5061873487729001, "grad_norm": 0.7025113974596033, "learning_rate": 8.751021731859582e-06, "loss": 0.4387, "step": 7322 }, { "epoch": 0.5062564811614241, "grad_norm": 1.3569694269897965, "learning_rate": 8.750651592171304e-06, "loss": 0.3855, "step": 7323 }, { "epoch": 0.5063256135499482, "grad_norm": 0.6278129702855986, "learning_rate": 8.7502814054753e-06, "loss": 0.2524, "step": 7324 }, { "epoch": 0.5063947459384722, "grad_norm": 0.880263805886742, "learning_rate": 8.749911171776212e-06, "loss": 0.3894, "step": 7325 }, { "epoch": 0.5064638783269962, "grad_norm": 0.9343495858253191, "learning_rate": 8.74954089107868e-06, "loss": 0.4019, "step": 7326 }, { "epoch": 0.5065330107155203, "grad_norm": 0.656149695066918, "learning_rate": 8.749170563387343e-06, "loss": 0.3046, "step": 7327 }, { "epoch": 0.5066021431040443, "grad_norm": 0.704059534765481, "learning_rate": 8.748800188706845e-06, "loss": 0.3551, "step": 7328 }, { "epoch": 0.5066712754925683, "grad_norm": 0.6976034951136455, "learning_rate": 8.748429767041826e-06, "loss": 0.4304, "step": 7329 }, { "epoch": 0.5067404078810923, "grad_norm": 0.7854181391415355, "learning_rate": 8.748059298396929e-06, "loss": 0.3928, "step": 7330 }, { "epoch": 0.5068095402696163, "grad_norm": 0.837571950718197, "learning_rate": 8.747688782776797e-06, "loss": 0.3264, "step": 7331 }, { "epoch": 0.5068786726581404, "grad_norm": 0.7397529182453393, "learning_rate": 8.747318220186073e-06, "loss": 0.287, "step": 7332 }, { "epoch": 0.5069478050466644, "grad_norm": 0.674485428031739, "learning_rate": 8.746947610629405e-06, "loss": 0.3683, "step": 7333 }, { "epoch": 0.5070169374351884, "grad_norm": 0.6683942640696658, "learning_rate": 8.746576954111437e-06, "loss": 0.3384, "step": 7334 }, { "epoch": 0.5070860698237124, "grad_norm": 0.7064046538112779, "learning_rate": 8.74620625063681e-06, "loss": 0.3192, "step": 7335 }, { "epoch": 0.5071552022122364, "grad_norm": 0.7000531246131894, "learning_rate": 8.745835500210173e-06, "loss": 0.3167, "step": 7336 }, { "epoch": 0.5072243346007604, "grad_norm": 0.8411081680716064, "learning_rate": 8.745464702836174e-06, "loss": 0.4079, "step": 7337 }, { "epoch": 0.5072934669892845, "grad_norm": 0.6148337566032159, "learning_rate": 8.745093858519459e-06, "loss": 0.2502, "step": 7338 }, { "epoch": 0.5073625993778085, "grad_norm": 0.7743077534624794, "learning_rate": 8.744722967264674e-06, "loss": 0.3196, "step": 7339 }, { "epoch": 0.5074317317663325, "grad_norm": 0.7296926495273792, "learning_rate": 8.744352029076472e-06, "loss": 0.4108, "step": 7340 }, { "epoch": 0.5075008641548565, "grad_norm": 0.7203129986026066, "learning_rate": 8.743981043959498e-06, "loss": 0.2907, "step": 7341 }, { "epoch": 0.5075699965433805, "grad_norm": 0.7101674576494691, "learning_rate": 8.743610011918404e-06, "loss": 0.3524, "step": 7342 }, { "epoch": 0.5076391289319045, "grad_norm": 0.7086920482183842, "learning_rate": 8.743238932957838e-06, "loss": 0.3748, "step": 7343 }, { "epoch": 0.5077082613204286, "grad_norm": 0.9087229462232599, "learning_rate": 8.742867807082452e-06, "loss": 0.3754, "step": 7344 }, { "epoch": 0.5077773937089527, "grad_norm": 0.7151239481646388, "learning_rate": 8.742496634296898e-06, "loss": 0.2865, "step": 7345 }, { "epoch": 0.5078465260974767, "grad_norm": 0.6862039760793262, "learning_rate": 8.742125414605828e-06, "loss": 0.2569, "step": 7346 }, { "epoch": 0.5079156584860007, "grad_norm": 0.6916296370377301, "learning_rate": 8.741754148013894e-06, "loss": 0.4653, "step": 7347 }, { "epoch": 0.5079847908745247, "grad_norm": 0.7135793802657767, "learning_rate": 8.741382834525748e-06, "loss": 0.3427, "step": 7348 }, { "epoch": 0.5080539232630488, "grad_norm": 0.6676158120855353, "learning_rate": 8.741011474146047e-06, "loss": 0.2979, "step": 7349 }, { "epoch": 0.5081230556515728, "grad_norm": 0.8170501427276428, "learning_rate": 8.740640066879441e-06, "loss": 0.3385, "step": 7350 }, { "epoch": 0.5081921880400968, "grad_norm": 0.786445861172437, "learning_rate": 8.740268612730589e-06, "loss": 0.3731, "step": 7351 }, { "epoch": 0.5082613204286208, "grad_norm": 0.7882477990084653, "learning_rate": 8.739897111704143e-06, "loss": 0.3173, "step": 7352 }, { "epoch": 0.5083304528171448, "grad_norm": 0.7064560422968591, "learning_rate": 8.739525563804762e-06, "loss": 0.2991, "step": 7353 }, { "epoch": 0.5083995852056689, "grad_norm": 0.7343542908212918, "learning_rate": 8.739153969037102e-06, "loss": 0.3903, "step": 7354 }, { "epoch": 0.5084687175941929, "grad_norm": 0.6798307140952107, "learning_rate": 8.73878232740582e-06, "loss": 0.3275, "step": 7355 }, { "epoch": 0.5085378499827169, "grad_norm": 0.8162172382586231, "learning_rate": 8.738410638915571e-06, "loss": 0.3298, "step": 7356 }, { "epoch": 0.5086069823712409, "grad_norm": 0.7040527875633217, "learning_rate": 8.738038903571018e-06, "loss": 0.3884, "step": 7357 }, { "epoch": 0.5086761147597649, "grad_norm": 0.6947916786464337, "learning_rate": 8.737667121376819e-06, "loss": 0.4529, "step": 7358 }, { "epoch": 0.508745247148289, "grad_norm": 0.7057342135782583, "learning_rate": 8.737295292337633e-06, "loss": 0.3421, "step": 7359 }, { "epoch": 0.508814379536813, "grad_norm": 0.7136654210283907, "learning_rate": 8.736923416458118e-06, "loss": 0.2587, "step": 7360 }, { "epoch": 0.508883511925337, "grad_norm": 0.7108826451713263, "learning_rate": 8.736551493742938e-06, "loss": 0.3891, "step": 7361 }, { "epoch": 0.508952644313861, "grad_norm": 0.701569612812752, "learning_rate": 8.736179524196753e-06, "loss": 0.4253, "step": 7362 }, { "epoch": 0.509021776702385, "grad_norm": 0.7535032032948316, "learning_rate": 8.735807507824225e-06, "loss": 0.3206, "step": 7363 }, { "epoch": 0.509090909090909, "grad_norm": 0.6939992026661617, "learning_rate": 8.735435444630019e-06, "loss": 0.2885, "step": 7364 }, { "epoch": 0.5091600414794332, "grad_norm": 0.677075011295498, "learning_rate": 8.735063334618791e-06, "loss": 0.3791, "step": 7365 }, { "epoch": 0.5092291738679572, "grad_norm": 0.6586765176613218, "learning_rate": 8.734691177795214e-06, "loss": 0.3983, "step": 7366 }, { "epoch": 0.5092983062564812, "grad_norm": 0.6706999978584203, "learning_rate": 8.734318974163947e-06, "loss": 0.2653, "step": 7367 }, { "epoch": 0.5093674386450052, "grad_norm": 0.8127511888745323, "learning_rate": 8.733946723729655e-06, "loss": 0.3076, "step": 7368 }, { "epoch": 0.5094365710335292, "grad_norm": 0.7280100100074057, "learning_rate": 8.733574426497004e-06, "loss": 0.4612, "step": 7369 }, { "epoch": 0.5095057034220533, "grad_norm": 0.6251131546734379, "learning_rate": 8.733202082470662e-06, "loss": 0.2903, "step": 7370 }, { "epoch": 0.5095748358105773, "grad_norm": 0.67208222551118, "learning_rate": 8.732829691655293e-06, "loss": 0.2798, "step": 7371 }, { "epoch": 0.5096439681991013, "grad_norm": 0.6718328488370959, "learning_rate": 8.732457254055566e-06, "loss": 0.4117, "step": 7372 }, { "epoch": 0.5097131005876253, "grad_norm": 0.730722387999817, "learning_rate": 8.732084769676147e-06, "loss": 0.4087, "step": 7373 }, { "epoch": 0.5097822329761493, "grad_norm": 0.6838785938981862, "learning_rate": 8.731712238521707e-06, "loss": 0.2677, "step": 7374 }, { "epoch": 0.5098513653646733, "grad_norm": 0.7389380497925441, "learning_rate": 8.731339660596912e-06, "loss": 0.3034, "step": 7375 }, { "epoch": 0.5099204977531974, "grad_norm": 0.7684964295282148, "learning_rate": 8.730967035906434e-06, "loss": 0.3878, "step": 7376 }, { "epoch": 0.5099896301417214, "grad_norm": 0.6812488143733847, "learning_rate": 8.730594364454942e-06, "loss": 0.3041, "step": 7377 }, { "epoch": 0.5100587625302454, "grad_norm": 0.7364785812834722, "learning_rate": 8.73022164624711e-06, "loss": 0.2739, "step": 7378 }, { "epoch": 0.5101278949187694, "grad_norm": 0.7122098868639782, "learning_rate": 8.729848881287604e-06, "loss": 0.4131, "step": 7379 }, { "epoch": 0.5101970273072934, "grad_norm": 0.7604634884702852, "learning_rate": 8.729476069581098e-06, "loss": 0.484, "step": 7380 }, { "epoch": 0.5102661596958175, "grad_norm": 0.6967553596598028, "learning_rate": 8.729103211132267e-06, "loss": 0.325, "step": 7381 }, { "epoch": 0.5103352920843415, "grad_norm": 0.7427044935346359, "learning_rate": 8.72873030594578e-06, "loss": 0.331, "step": 7382 }, { "epoch": 0.5104044244728655, "grad_norm": 0.6771006245627624, "learning_rate": 8.728357354026314e-06, "loss": 0.4262, "step": 7383 }, { "epoch": 0.5104735568613895, "grad_norm": 0.7183694416166454, "learning_rate": 8.727984355378541e-06, "loss": 0.4273, "step": 7384 }, { "epoch": 0.5105426892499136, "grad_norm": 0.8109821203693076, "learning_rate": 8.727611310007138e-06, "loss": 0.3219, "step": 7385 }, { "epoch": 0.5106118216384377, "grad_norm": 0.6890988974728921, "learning_rate": 8.72723821791678e-06, "loss": 0.3681, "step": 7386 }, { "epoch": 0.5106809540269617, "grad_norm": 0.9775428541611, "learning_rate": 8.72686507911214e-06, "loss": 0.3987, "step": 7387 }, { "epoch": 0.5107500864154857, "grad_norm": 1.2740728061357758, "learning_rate": 8.726491893597901e-06, "loss": 0.2725, "step": 7388 }, { "epoch": 0.5108192188040097, "grad_norm": 0.6634358904948547, "learning_rate": 8.726118661378735e-06, "loss": 0.3281, "step": 7389 }, { "epoch": 0.5108883511925337, "grad_norm": 0.7710635195698069, "learning_rate": 8.72574538245932e-06, "loss": 0.4545, "step": 7390 }, { "epoch": 0.5109574835810577, "grad_norm": 0.7108083104409514, "learning_rate": 8.725372056844335e-06, "loss": 0.3181, "step": 7391 }, { "epoch": 0.5110266159695818, "grad_norm": 0.7527686186647572, "learning_rate": 8.724998684538462e-06, "loss": 0.3317, "step": 7392 }, { "epoch": 0.5110957483581058, "grad_norm": 0.7707996523408019, "learning_rate": 8.724625265546375e-06, "loss": 0.3828, "step": 7393 }, { "epoch": 0.5111648807466298, "grad_norm": 0.7812737622288851, "learning_rate": 8.72425179987276e-06, "loss": 0.4415, "step": 7394 }, { "epoch": 0.5112340131351538, "grad_norm": 0.7015285630936912, "learning_rate": 8.723878287522293e-06, "loss": 0.2884, "step": 7395 }, { "epoch": 0.5113031455236778, "grad_norm": 0.8358747413966614, "learning_rate": 8.723504728499658e-06, "loss": 0.262, "step": 7396 }, { "epoch": 0.5113722779122019, "grad_norm": 0.729589606587566, "learning_rate": 8.723131122809536e-06, "loss": 0.3931, "step": 7397 }, { "epoch": 0.5114414103007259, "grad_norm": 0.6536724455098987, "learning_rate": 8.722757470456608e-06, "loss": 0.3328, "step": 7398 }, { "epoch": 0.5115105426892499, "grad_norm": 0.6572362776476199, "learning_rate": 8.722383771445562e-06, "loss": 0.3256, "step": 7399 }, { "epoch": 0.5115796750777739, "grad_norm": 0.6982900639895172, "learning_rate": 8.722010025781075e-06, "loss": 0.3217, "step": 7400 }, { "epoch": 0.5116488074662979, "grad_norm": 1.17420977474046, "learning_rate": 8.721636233467837e-06, "loss": 0.3891, "step": 7401 }, { "epoch": 0.511717939854822, "grad_norm": 1.3299347575957448, "learning_rate": 8.72126239451053e-06, "loss": 0.2769, "step": 7402 }, { "epoch": 0.511787072243346, "grad_norm": 0.679614177405733, "learning_rate": 8.720888508913838e-06, "loss": 0.2653, "step": 7403 }, { "epoch": 0.51185620463187, "grad_norm": 2.505762395250688, "learning_rate": 8.72051457668245e-06, "loss": 0.3662, "step": 7404 }, { "epoch": 0.511925337020394, "grad_norm": 0.6627263401610342, "learning_rate": 8.720140597821052e-06, "loss": 0.3792, "step": 7405 }, { "epoch": 0.5119944694089181, "grad_norm": 0.7395329291978675, "learning_rate": 8.719766572334331e-06, "loss": 0.3242, "step": 7406 }, { "epoch": 0.5120636017974421, "grad_norm": 0.7389979719242369, "learning_rate": 8.719392500226972e-06, "loss": 0.2962, "step": 7407 }, { "epoch": 0.5121327341859662, "grad_norm": 0.6153164125998015, "learning_rate": 8.719018381503666e-06, "loss": 0.3369, "step": 7408 }, { "epoch": 0.5122018665744902, "grad_norm": 0.7208767199752787, "learning_rate": 8.718644216169103e-06, "loss": 0.2892, "step": 7409 }, { "epoch": 0.5122709989630142, "grad_norm": 0.658699847015141, "learning_rate": 8.71827000422797e-06, "loss": 0.3021, "step": 7410 }, { "epoch": 0.5123401313515382, "grad_norm": 0.8942993996035165, "learning_rate": 8.717895745684958e-06, "loss": 0.3895, "step": 7411 }, { "epoch": 0.5124092637400622, "grad_norm": 0.7455390806354042, "learning_rate": 8.717521440544758e-06, "loss": 0.3943, "step": 7412 }, { "epoch": 0.5124783961285863, "grad_norm": 1.7798227023913424, "learning_rate": 8.717147088812062e-06, "loss": 0.2788, "step": 7413 }, { "epoch": 0.5125475285171103, "grad_norm": 0.6862396421523795, "learning_rate": 8.716772690491558e-06, "loss": 0.3275, "step": 7414 }, { "epoch": 0.5126166609056343, "grad_norm": 0.7503003009152537, "learning_rate": 8.716398245587943e-06, "loss": 0.4093, "step": 7415 }, { "epoch": 0.5126857932941583, "grad_norm": 0.8261406399760485, "learning_rate": 8.716023754105908e-06, "loss": 0.3306, "step": 7416 }, { "epoch": 0.5127549256826823, "grad_norm": 0.6784600683622829, "learning_rate": 8.715649216050147e-06, "loss": 0.3318, "step": 7417 }, { "epoch": 0.5128240580712063, "grad_norm": 0.8860754177640343, "learning_rate": 8.715274631425354e-06, "loss": 0.4236, "step": 7418 }, { "epoch": 0.5128931904597304, "grad_norm": 0.8611072154180766, "learning_rate": 8.714900000236222e-06, "loss": 0.4022, "step": 7419 }, { "epoch": 0.5129623228482544, "grad_norm": 0.6628274739938594, "learning_rate": 8.71452532248745e-06, "loss": 0.2943, "step": 7420 }, { "epoch": 0.5130314552367784, "grad_norm": 0.7214629400633594, "learning_rate": 8.714150598183733e-06, "loss": 0.2522, "step": 7421 }, { "epoch": 0.5131005876253024, "grad_norm": 0.7323482415327505, "learning_rate": 8.713775827329764e-06, "loss": 0.3667, "step": 7422 }, { "epoch": 0.5131697200138264, "grad_norm": 0.714258689307448, "learning_rate": 8.713401009930243e-06, "loss": 0.4115, "step": 7423 }, { "epoch": 0.5132388524023505, "grad_norm": 0.6963218124143015, "learning_rate": 8.713026145989868e-06, "loss": 0.2765, "step": 7424 }, { "epoch": 0.5133079847908745, "grad_norm": 0.7040501766426582, "learning_rate": 8.712651235513337e-06, "loss": 0.3162, "step": 7425 }, { "epoch": 0.5133771171793986, "grad_norm": 0.7148509215762878, "learning_rate": 8.712276278505347e-06, "loss": 0.3871, "step": 7426 }, { "epoch": 0.5134462495679226, "grad_norm": 0.6860106599119347, "learning_rate": 8.711901274970598e-06, "loss": 0.273, "step": 7427 }, { "epoch": 0.5135153819564466, "grad_norm": 0.8451571646225972, "learning_rate": 8.711526224913792e-06, "loss": 0.3453, "step": 7428 }, { "epoch": 0.5135845143449707, "grad_norm": 0.7456612615386154, "learning_rate": 8.711151128339628e-06, "loss": 0.4101, "step": 7429 }, { "epoch": 0.5136536467334947, "grad_norm": 0.6964732177740882, "learning_rate": 8.710775985252808e-06, "loss": 0.3451, "step": 7430 }, { "epoch": 0.5137227791220187, "grad_norm": 0.6446551399349296, "learning_rate": 8.710400795658031e-06, "loss": 0.2689, "step": 7431 }, { "epoch": 0.5137919115105427, "grad_norm": 0.6944764407947039, "learning_rate": 8.710025559560005e-06, "loss": 0.3301, "step": 7432 }, { "epoch": 0.5138610438990667, "grad_norm": 0.9764455125212754, "learning_rate": 8.709650276963427e-06, "loss": 0.4173, "step": 7433 }, { "epoch": 0.5139301762875907, "grad_norm": 0.735750778083511, "learning_rate": 8.709274947873003e-06, "loss": 0.3666, "step": 7434 }, { "epoch": 0.5139993086761148, "grad_norm": 0.6895376265950872, "learning_rate": 8.70889957229344e-06, "loss": 0.258, "step": 7435 }, { "epoch": 0.5140684410646388, "grad_norm": 0.7313318575786133, "learning_rate": 8.708524150229435e-06, "loss": 0.4025, "step": 7436 }, { "epoch": 0.5141375734531628, "grad_norm": 0.8712864409621011, "learning_rate": 8.708148681685702e-06, "loss": 0.3479, "step": 7437 }, { "epoch": 0.5142067058416868, "grad_norm": 0.7268057761297395, "learning_rate": 8.70777316666694e-06, "loss": 0.3508, "step": 7438 }, { "epoch": 0.5142758382302108, "grad_norm": 0.7912405446676626, "learning_rate": 8.70739760517786e-06, "loss": 0.259, "step": 7439 }, { "epoch": 0.5143449706187349, "grad_norm": 0.8512461545322558, "learning_rate": 8.707021997223167e-06, "loss": 0.4838, "step": 7440 }, { "epoch": 0.5144141030072589, "grad_norm": 0.6578956299748839, "learning_rate": 8.706646342807568e-06, "loss": 0.3018, "step": 7441 }, { "epoch": 0.5144832353957829, "grad_norm": 0.7482147638186369, "learning_rate": 8.706270641935772e-06, "loss": 0.2814, "step": 7442 }, { "epoch": 0.5145523677843069, "grad_norm": 0.7982333161868432, "learning_rate": 8.705894894612486e-06, "loss": 0.387, "step": 7443 }, { "epoch": 0.5146215001728309, "grad_norm": 0.6924795690655297, "learning_rate": 8.705519100842424e-06, "loss": 0.3667, "step": 7444 }, { "epoch": 0.5146906325613549, "grad_norm": 0.7165478664499761, "learning_rate": 8.705143260630291e-06, "loss": 0.2778, "step": 7445 }, { "epoch": 0.5147597649498791, "grad_norm": 0.9328703458758509, "learning_rate": 8.7047673739808e-06, "loss": 0.3149, "step": 7446 }, { "epoch": 0.5148288973384031, "grad_norm": 0.7298553230269462, "learning_rate": 8.70439144089866e-06, "loss": 0.43, "step": 7447 }, { "epoch": 0.5148980297269271, "grad_norm": 0.6872661799323456, "learning_rate": 8.704015461388587e-06, "loss": 0.3702, "step": 7448 }, { "epoch": 0.5149671621154511, "grad_norm": 0.6971282035883618, "learning_rate": 8.703639435455287e-06, "loss": 0.3166, "step": 7449 }, { "epoch": 0.5150362945039751, "grad_norm": 0.8145706560312902, "learning_rate": 8.703263363103477e-06, "loss": 0.398, "step": 7450 }, { "epoch": 0.5151054268924992, "grad_norm": 0.7353203437298227, "learning_rate": 8.702887244337872e-06, "loss": 0.4241, "step": 7451 }, { "epoch": 0.5151745592810232, "grad_norm": 0.7283774549407462, "learning_rate": 8.702511079163181e-06, "loss": 0.2964, "step": 7452 }, { "epoch": 0.5152436916695472, "grad_norm": 0.7611758389956685, "learning_rate": 8.702134867584122e-06, "loss": 0.304, "step": 7453 }, { "epoch": 0.5153128240580712, "grad_norm": 1.1197911420099838, "learning_rate": 8.70175860960541e-06, "loss": 0.395, "step": 7454 }, { "epoch": 0.5153819564465952, "grad_norm": 0.8480723965156164, "learning_rate": 8.701382305231758e-06, "loss": 0.3481, "step": 7455 }, { "epoch": 0.5154510888351193, "grad_norm": 0.7609638504778117, "learning_rate": 8.701005954467886e-06, "loss": 0.3319, "step": 7456 }, { "epoch": 0.5155202212236433, "grad_norm": 0.6341785814149375, "learning_rate": 8.700629557318508e-06, "loss": 0.2879, "step": 7457 }, { "epoch": 0.5155893536121673, "grad_norm": 0.7251927210282613, "learning_rate": 8.700253113788343e-06, "loss": 0.3833, "step": 7458 }, { "epoch": 0.5156584860006913, "grad_norm": 0.7312966873059658, "learning_rate": 8.69987662388211e-06, "loss": 0.3631, "step": 7459 }, { "epoch": 0.5157276183892153, "grad_norm": 0.7527632121897345, "learning_rate": 8.699500087604522e-06, "loss": 0.3104, "step": 7460 }, { "epoch": 0.5157967507777393, "grad_norm": 0.9829267650832858, "learning_rate": 8.699123504960307e-06, "loss": 0.3474, "step": 7461 }, { "epoch": 0.5158658831662634, "grad_norm": 0.8786154069867105, "learning_rate": 8.698746875954177e-06, "loss": 0.4142, "step": 7462 }, { "epoch": 0.5159350155547874, "grad_norm": 2.9752744953837835, "learning_rate": 8.698370200590857e-06, "loss": 0.3859, "step": 7463 }, { "epoch": 0.5160041479433114, "grad_norm": 0.7503005822700219, "learning_rate": 8.697993478875067e-06, "loss": 0.311, "step": 7464 }, { "epoch": 0.5160732803318354, "grad_norm": 0.7910068395288812, "learning_rate": 8.697616710811528e-06, "loss": 0.4161, "step": 7465 }, { "epoch": 0.5161424127203594, "grad_norm": 0.709530219338052, "learning_rate": 8.697239896404962e-06, "loss": 0.3344, "step": 7466 }, { "epoch": 0.5162115451088836, "grad_norm": 0.726936250654519, "learning_rate": 8.69686303566009e-06, "loss": 0.3686, "step": 7467 }, { "epoch": 0.5162806774974076, "grad_norm": 0.8648157134117841, "learning_rate": 8.69648612858164e-06, "loss": 0.3094, "step": 7468 }, { "epoch": 0.5163498098859316, "grad_norm": 0.9278721634903947, "learning_rate": 8.696109175174332e-06, "loss": 0.4247, "step": 7469 }, { "epoch": 0.5164189422744556, "grad_norm": 0.779025715574508, "learning_rate": 8.695732175442893e-06, "loss": 0.3622, "step": 7470 }, { "epoch": 0.5164880746629796, "grad_norm": 0.7449054478791796, "learning_rate": 8.695355129392045e-06, "loss": 0.2859, "step": 7471 }, { "epoch": 0.5165572070515037, "grad_norm": 0.7339839983739457, "learning_rate": 8.694978037026514e-06, "loss": 0.4257, "step": 7472 }, { "epoch": 0.5166263394400277, "grad_norm": 1.2705925320674412, "learning_rate": 8.69460089835103e-06, "loss": 0.3948, "step": 7473 }, { "epoch": 0.5166954718285517, "grad_norm": 2.333872962178124, "learning_rate": 8.694223713370315e-06, "loss": 0.2647, "step": 7474 }, { "epoch": 0.5167646042170757, "grad_norm": 0.7932093863549823, "learning_rate": 8.693846482089099e-06, "loss": 0.3636, "step": 7475 }, { "epoch": 0.5168337366055997, "grad_norm": 0.670264840642015, "learning_rate": 8.693469204512108e-06, "loss": 0.4251, "step": 7476 }, { "epoch": 0.5169028689941237, "grad_norm": 0.8140843288060962, "learning_rate": 8.693091880644074e-06, "loss": 0.2862, "step": 7477 }, { "epoch": 0.5169720013826478, "grad_norm": 0.7895702021583553, "learning_rate": 8.692714510489722e-06, "loss": 0.3387, "step": 7478 }, { "epoch": 0.5170411337711718, "grad_norm": 1.10164079622572, "learning_rate": 8.692337094053786e-06, "loss": 0.3132, "step": 7479 }, { "epoch": 0.5171102661596958, "grad_norm": 0.7989460004604606, "learning_rate": 8.691959631340991e-06, "loss": 0.3899, "step": 7480 }, { "epoch": 0.5171793985482198, "grad_norm": 0.8040913346189155, "learning_rate": 8.691582122356073e-06, "loss": 0.2929, "step": 7481 }, { "epoch": 0.5172485309367438, "grad_norm": 3.445592786767106, "learning_rate": 8.691204567103759e-06, "loss": 0.3646, "step": 7482 }, { "epoch": 0.5173176633252679, "grad_norm": 0.6997286059744776, "learning_rate": 8.690826965588784e-06, "loss": 0.4225, "step": 7483 }, { "epoch": 0.5173867957137919, "grad_norm": 0.7132602037814207, "learning_rate": 8.69044931781588e-06, "loss": 0.3073, "step": 7484 }, { "epoch": 0.5174559281023159, "grad_norm": 0.9751148736090374, "learning_rate": 8.690071623789778e-06, "loss": 0.3332, "step": 7485 }, { "epoch": 0.5175250604908399, "grad_norm": 0.8476557581425384, "learning_rate": 8.689693883515215e-06, "loss": 0.4537, "step": 7486 }, { "epoch": 0.517594192879364, "grad_norm": 1.235625648164384, "learning_rate": 8.689316096996923e-06, "loss": 0.3679, "step": 7487 }, { "epoch": 0.517663325267888, "grad_norm": 0.8097205903819287, "learning_rate": 8.688938264239639e-06, "loss": 0.322, "step": 7488 }, { "epoch": 0.5177324576564121, "grad_norm": 0.8649223123473355, "learning_rate": 8.688560385248097e-06, "loss": 0.3286, "step": 7489 }, { "epoch": 0.5178015900449361, "grad_norm": 0.70358300412953, "learning_rate": 8.688182460027031e-06, "loss": 0.4331, "step": 7490 }, { "epoch": 0.5178707224334601, "grad_norm": 0.7565425105645979, "learning_rate": 8.68780448858118e-06, "loss": 0.3686, "step": 7491 }, { "epoch": 0.5179398548219841, "grad_norm": 0.6874627740242792, "learning_rate": 8.687426470915282e-06, "loss": 0.3423, "step": 7492 }, { "epoch": 0.5180089872105081, "grad_norm": 0.757592167949352, "learning_rate": 8.687048407034073e-06, "loss": 0.3712, "step": 7493 }, { "epoch": 0.5180781195990322, "grad_norm": 1.0750041034440874, "learning_rate": 8.686670296942293e-06, "loss": 0.4824, "step": 7494 }, { "epoch": 0.5181472519875562, "grad_norm": 0.7545630589653597, "learning_rate": 8.68629214064468e-06, "loss": 0.2819, "step": 7495 }, { "epoch": 0.5182163843760802, "grad_norm": 0.8607588197926476, "learning_rate": 8.685913938145974e-06, "loss": 0.2697, "step": 7496 }, { "epoch": 0.5182855167646042, "grad_norm": 0.8258070235023925, "learning_rate": 8.685535689450913e-06, "loss": 0.432, "step": 7497 }, { "epoch": 0.5183546491531282, "grad_norm": 0.7253821590260887, "learning_rate": 8.685157394564239e-06, "loss": 0.3315, "step": 7498 }, { "epoch": 0.5184237815416523, "grad_norm": 0.7931948863280662, "learning_rate": 8.684779053490694e-06, "loss": 0.3088, "step": 7499 }, { "epoch": 0.5184929139301763, "grad_norm": 0.9604228074500285, "learning_rate": 8.68440066623502e-06, "loss": 0.3654, "step": 7500 }, { "epoch": 0.5185620463187003, "grad_norm": 0.7588637231130274, "learning_rate": 8.68402223280196e-06, "loss": 0.4454, "step": 7501 }, { "epoch": 0.5186311787072243, "grad_norm": 0.7166366120536786, "learning_rate": 8.683643753196254e-06, "loss": 0.333, "step": 7502 }, { "epoch": 0.5187003110957483, "grad_norm": 0.7761418624067448, "learning_rate": 8.683265227422647e-06, "loss": 0.3442, "step": 7503 }, { "epoch": 0.5187694434842723, "grad_norm": 0.7059145716118089, "learning_rate": 8.682886655485883e-06, "loss": 0.3908, "step": 7504 }, { "epoch": 0.5188385758727964, "grad_norm": 1.3010899457045977, "learning_rate": 8.68250803739071e-06, "loss": 0.3608, "step": 7505 }, { "epoch": 0.5189077082613204, "grad_norm": 0.7722961624298342, "learning_rate": 8.682129373141868e-06, "loss": 0.2836, "step": 7506 }, { "epoch": 0.5189768406498445, "grad_norm": 0.7398649290219372, "learning_rate": 8.681750662744106e-06, "loss": 0.3421, "step": 7507 }, { "epoch": 0.5190459730383685, "grad_norm": 0.7560694557363906, "learning_rate": 8.68137190620217e-06, "loss": 0.4311, "step": 7508 }, { "epoch": 0.5191151054268925, "grad_norm": 0.741229096389807, "learning_rate": 8.680993103520805e-06, "loss": 0.2869, "step": 7509 }, { "epoch": 0.5191842378154166, "grad_norm": 0.7376841443733468, "learning_rate": 8.680614254704763e-06, "loss": 0.3215, "step": 7510 }, { "epoch": 0.5192533702039406, "grad_norm": 0.6860562441319955, "learning_rate": 8.680235359758789e-06, "loss": 0.3205, "step": 7511 }, { "epoch": 0.5193225025924646, "grad_norm": 0.7744781490099042, "learning_rate": 8.679856418687633e-06, "loss": 0.3608, "step": 7512 }, { "epoch": 0.5193916349809886, "grad_norm": 0.7077348643003722, "learning_rate": 8.679477431496042e-06, "loss": 0.3721, "step": 7513 }, { "epoch": 0.5194607673695126, "grad_norm": 0.7902483250197928, "learning_rate": 8.67909839818877e-06, "loss": 0.2984, "step": 7514 }, { "epoch": 0.5195298997580367, "grad_norm": 0.7378838944744361, "learning_rate": 8.678719318770562e-06, "loss": 0.419, "step": 7515 }, { "epoch": 0.5195990321465607, "grad_norm": 0.6419847039381806, "learning_rate": 8.678340193246175e-06, "loss": 0.3264, "step": 7516 }, { "epoch": 0.5196681645350847, "grad_norm": 0.6662524419946408, "learning_rate": 8.677961021620357e-06, "loss": 0.3132, "step": 7517 }, { "epoch": 0.5197372969236087, "grad_norm": 0.7177291967873639, "learning_rate": 8.677581803897863e-06, "loss": 0.3621, "step": 7518 }, { "epoch": 0.5198064293121327, "grad_norm": 0.6911186383299761, "learning_rate": 8.677202540083443e-06, "loss": 0.4095, "step": 7519 }, { "epoch": 0.5198755617006567, "grad_norm": 0.760687204648128, "learning_rate": 8.67682323018185e-06, "loss": 0.2462, "step": 7520 }, { "epoch": 0.5199446940891808, "grad_norm": 0.7589750065427064, "learning_rate": 8.676443874197839e-06, "loss": 0.324, "step": 7521 }, { "epoch": 0.5200138264777048, "grad_norm": 0.7485505595892048, "learning_rate": 8.676064472136168e-06, "loss": 0.4359, "step": 7522 }, { "epoch": 0.5200829588662288, "grad_norm": 0.6514390680750547, "learning_rate": 8.675685024001588e-06, "loss": 0.3519, "step": 7523 }, { "epoch": 0.5201520912547528, "grad_norm": 0.7201014153151631, "learning_rate": 8.675305529798854e-06, "loss": 0.242, "step": 7524 }, { "epoch": 0.5202212236432768, "grad_norm": 0.7457675585562403, "learning_rate": 8.674925989532726e-06, "loss": 0.336, "step": 7525 }, { "epoch": 0.5202903560318008, "grad_norm": 0.7517490248200134, "learning_rate": 8.674546403207958e-06, "loss": 0.3652, "step": 7526 }, { "epoch": 0.5203594884203249, "grad_norm": 0.7535856772068635, "learning_rate": 8.674166770829309e-06, "loss": 0.3395, "step": 7527 }, { "epoch": 0.520428620808849, "grad_norm": 0.8746436634041019, "learning_rate": 8.673787092401536e-06, "loss": 0.3128, "step": 7528 }, { "epoch": 0.520497753197373, "grad_norm": 0.7596681830473455, "learning_rate": 8.673407367929396e-06, "loss": 0.4051, "step": 7529 }, { "epoch": 0.520566885585897, "grad_norm": 0.7213026107299915, "learning_rate": 8.673027597417652e-06, "loss": 0.3821, "step": 7530 }, { "epoch": 0.520636017974421, "grad_norm": 0.7913527353208554, "learning_rate": 8.672647780871062e-06, "loss": 0.3254, "step": 7531 }, { "epoch": 0.5207051503629451, "grad_norm": 0.7053923046746128, "learning_rate": 8.672267918294386e-06, "loss": 0.3098, "step": 7532 }, { "epoch": 0.5207742827514691, "grad_norm": 0.7593618265820998, "learning_rate": 8.671888009692386e-06, "loss": 0.3964, "step": 7533 }, { "epoch": 0.5208434151399931, "grad_norm": 0.8077341927929506, "learning_rate": 8.671508055069822e-06, "loss": 0.3029, "step": 7534 }, { "epoch": 0.5209125475285171, "grad_norm": 0.937969897019191, "learning_rate": 8.671128054431456e-06, "loss": 0.2491, "step": 7535 }, { "epoch": 0.5209816799170411, "grad_norm": 0.7810165447489295, "learning_rate": 8.67074800778205e-06, "loss": 0.3503, "step": 7536 }, { "epoch": 0.5210508123055652, "grad_norm": 0.6721153498161423, "learning_rate": 8.670367915126372e-06, "loss": 0.3279, "step": 7537 }, { "epoch": 0.5211199446940892, "grad_norm": 0.590937066627104, "learning_rate": 8.66998777646918e-06, "loss": 0.2832, "step": 7538 }, { "epoch": 0.5211890770826132, "grad_norm": 0.7462287170522002, "learning_rate": 8.66960759181524e-06, "loss": 0.3072, "step": 7539 }, { "epoch": 0.5212582094711372, "grad_norm": 0.74788992870387, "learning_rate": 8.66922736116932e-06, "loss": 0.4678, "step": 7540 }, { "epoch": 0.5213273418596612, "grad_norm": 0.7111641605892757, "learning_rate": 8.668847084536181e-06, "loss": 0.309, "step": 7541 }, { "epoch": 0.5213964742481852, "grad_norm": 0.6759035565739582, "learning_rate": 8.66846676192059e-06, "loss": 0.2901, "step": 7542 }, { "epoch": 0.5214656066367093, "grad_norm": 0.6608202715994492, "learning_rate": 8.668086393327317e-06, "loss": 0.3586, "step": 7543 }, { "epoch": 0.5215347390252333, "grad_norm": 0.7574437563898221, "learning_rate": 8.667705978761128e-06, "loss": 0.5323, "step": 7544 }, { "epoch": 0.5216038714137573, "grad_norm": 0.8421569874765122, "learning_rate": 8.667325518226788e-06, "loss": 0.3588, "step": 7545 }, { "epoch": 0.5216730038022813, "grad_norm": 0.8255478042068064, "learning_rate": 8.666945011729068e-06, "loss": 0.2651, "step": 7546 }, { "epoch": 0.5217421361908053, "grad_norm": 0.8462061130300863, "learning_rate": 8.666564459272735e-06, "loss": 0.4421, "step": 7547 }, { "epoch": 0.5218112685793295, "grad_norm": 0.7849392326322184, "learning_rate": 8.666183860862561e-06, "loss": 0.3957, "step": 7548 }, { "epoch": 0.5218804009678535, "grad_norm": 0.7849893348023099, "learning_rate": 8.665803216503315e-06, "loss": 0.3442, "step": 7549 }, { "epoch": 0.5219495333563775, "grad_norm": 0.7269424004464325, "learning_rate": 8.665422526199767e-06, "loss": 0.3932, "step": 7550 }, { "epoch": 0.5220186657449015, "grad_norm": 0.7542858731913662, "learning_rate": 8.665041789956689e-06, "loss": 0.4432, "step": 7551 }, { "epoch": 0.5220877981334255, "grad_norm": 1.1933639231783277, "learning_rate": 8.664661007778853e-06, "loss": 0.3199, "step": 7552 }, { "epoch": 0.5221569305219496, "grad_norm": 0.6786155555073191, "learning_rate": 8.664280179671031e-06, "loss": 0.2991, "step": 7553 }, { "epoch": 0.5222260629104736, "grad_norm": 0.6931719227146265, "learning_rate": 8.663899305637994e-06, "loss": 0.3514, "step": 7554 }, { "epoch": 0.5222951952989976, "grad_norm": 0.9969730853123104, "learning_rate": 8.663518385684521e-06, "loss": 0.3923, "step": 7555 }, { "epoch": 0.5223643276875216, "grad_norm": 0.7567772176460273, "learning_rate": 8.663137419815381e-06, "loss": 0.3332, "step": 7556 }, { "epoch": 0.5224334600760456, "grad_norm": 0.9120408150282744, "learning_rate": 8.662756408035353e-06, "loss": 0.3521, "step": 7557 }, { "epoch": 0.5225025924645696, "grad_norm": 0.6664559781170573, "learning_rate": 8.662375350349209e-06, "loss": 0.3628, "step": 7558 }, { "epoch": 0.5225717248530937, "grad_norm": 0.7291884658337887, "learning_rate": 8.661994246761724e-06, "loss": 0.3013, "step": 7559 }, { "epoch": 0.5226408572416177, "grad_norm": 0.6694872509860389, "learning_rate": 8.661613097277677e-06, "loss": 0.3162, "step": 7560 }, { "epoch": 0.5227099896301417, "grad_norm": 0.6733824243831875, "learning_rate": 8.661231901901845e-06, "loss": 0.4154, "step": 7561 }, { "epoch": 0.5227791220186657, "grad_norm": 0.6978574273733181, "learning_rate": 8.660850660639004e-06, "loss": 0.3795, "step": 7562 }, { "epoch": 0.5228482544071897, "grad_norm": 0.6716650592162299, "learning_rate": 8.660469373493933e-06, "loss": 0.3438, "step": 7563 }, { "epoch": 0.5229173867957138, "grad_norm": 0.7328434100086498, "learning_rate": 8.660088040471411e-06, "loss": 0.3797, "step": 7564 }, { "epoch": 0.5229865191842378, "grad_norm": 0.7168567235331962, "learning_rate": 8.659706661576218e-06, "loss": 0.41, "step": 7565 }, { "epoch": 0.5230556515727618, "grad_norm": 0.6986633366659073, "learning_rate": 8.659325236813131e-06, "loss": 0.3034, "step": 7566 }, { "epoch": 0.5231247839612858, "grad_norm": 0.8008419316278813, "learning_rate": 8.658943766186936e-06, "loss": 0.3117, "step": 7567 }, { "epoch": 0.5231939163498099, "grad_norm": 0.726992996064352, "learning_rate": 8.658562249702407e-06, "loss": 0.3279, "step": 7568 }, { "epoch": 0.523263048738334, "grad_norm": 0.7082770880862489, "learning_rate": 8.658180687364333e-06, "loss": 0.438, "step": 7569 }, { "epoch": 0.523332181126858, "grad_norm": 0.6977999595896553, "learning_rate": 8.65779907917749e-06, "loss": 0.3276, "step": 7570 }, { "epoch": 0.523401313515382, "grad_norm": 0.7458181570195497, "learning_rate": 8.657417425146662e-06, "loss": 0.2947, "step": 7571 }, { "epoch": 0.523470445903906, "grad_norm": 0.9848353985054392, "learning_rate": 8.657035725276634e-06, "loss": 0.4778, "step": 7572 }, { "epoch": 0.52353957829243, "grad_norm": 0.9112039986792949, "learning_rate": 8.656653979572192e-06, "loss": 0.3388, "step": 7573 }, { "epoch": 0.523608710680954, "grad_norm": 0.6388357832113191, "learning_rate": 8.656272188038115e-06, "loss": 0.2705, "step": 7574 }, { "epoch": 0.5236778430694781, "grad_norm": 0.8765464515555971, "learning_rate": 8.655890350679193e-06, "loss": 0.3313, "step": 7575 }, { "epoch": 0.5237469754580021, "grad_norm": 0.6852178028389763, "learning_rate": 8.655508467500208e-06, "loss": 0.4087, "step": 7576 }, { "epoch": 0.5238161078465261, "grad_norm": 0.7858554008731906, "learning_rate": 8.655126538505949e-06, "loss": 0.3496, "step": 7577 }, { "epoch": 0.5238852402350501, "grad_norm": 0.7755538440037971, "learning_rate": 8.654744563701201e-06, "loss": 0.2827, "step": 7578 }, { "epoch": 0.5239543726235741, "grad_norm": 0.7494823859120393, "learning_rate": 8.654362543090754e-06, "loss": 0.4233, "step": 7579 }, { "epoch": 0.5240235050120982, "grad_norm": 0.7138148812628082, "learning_rate": 8.653980476679393e-06, "loss": 0.3533, "step": 7580 }, { "epoch": 0.5240926374006222, "grad_norm": 0.8337755924341236, "learning_rate": 8.653598364471907e-06, "loss": 0.2421, "step": 7581 }, { "epoch": 0.5241617697891462, "grad_norm": 0.7107372118873863, "learning_rate": 8.653216206473086e-06, "loss": 0.3176, "step": 7582 }, { "epoch": 0.5242309021776702, "grad_norm": 0.6981410595031913, "learning_rate": 8.652834002687718e-06, "loss": 0.421, "step": 7583 }, { "epoch": 0.5243000345661942, "grad_norm": 0.7052642023688538, "learning_rate": 8.652451753120597e-06, "loss": 0.3195, "step": 7584 }, { "epoch": 0.5243691669547182, "grad_norm": 0.593863616150374, "learning_rate": 8.652069457776508e-06, "loss": 0.306, "step": 7585 }, { "epoch": 0.5244382993432423, "grad_norm": 0.7228954331066543, "learning_rate": 8.651687116660249e-06, "loss": 0.3577, "step": 7586 }, { "epoch": 0.5245074317317663, "grad_norm": 0.7248889518447128, "learning_rate": 8.651304729776607e-06, "loss": 0.38, "step": 7587 }, { "epoch": 0.5245765641202903, "grad_norm": 0.7143675525174957, "learning_rate": 8.650922297130378e-06, "loss": 0.3068, "step": 7588 }, { "epoch": 0.5246456965088144, "grad_norm": 0.6458121728183354, "learning_rate": 8.650539818726351e-06, "loss": 0.2515, "step": 7589 }, { "epoch": 0.5247148288973384, "grad_norm": 0.7708555720545899, "learning_rate": 8.650157294569324e-06, "loss": 0.4204, "step": 7590 }, { "epoch": 0.5247839612858625, "grad_norm": 0.7165429981655944, "learning_rate": 8.649774724664088e-06, "loss": 0.3352, "step": 7591 }, { "epoch": 0.5248530936743865, "grad_norm": 0.7529554575993228, "learning_rate": 8.64939210901544e-06, "loss": 0.2815, "step": 7592 }, { "epoch": 0.5249222260629105, "grad_norm": 0.7372070758849387, "learning_rate": 8.649009447628174e-06, "loss": 0.3549, "step": 7593 }, { "epoch": 0.5249913584514345, "grad_norm": 0.6942362013513762, "learning_rate": 8.648626740507087e-06, "loss": 0.3901, "step": 7594 }, { "epoch": 0.5250604908399585, "grad_norm": 0.6971373675124944, "learning_rate": 8.648243987656975e-06, "loss": 0.3304, "step": 7595 }, { "epoch": 0.5251296232284826, "grad_norm": 1.1237689697252733, "learning_rate": 8.647861189082635e-06, "loss": 0.2687, "step": 7596 }, { "epoch": 0.5251987556170066, "grad_norm": 0.7040374827234753, "learning_rate": 8.647478344788864e-06, "loss": 0.3945, "step": 7597 }, { "epoch": 0.5252678880055306, "grad_norm": 1.004866940903212, "learning_rate": 8.647095454780463e-06, "loss": 0.42, "step": 7598 }, { "epoch": 0.5253370203940546, "grad_norm": 0.6840216481138913, "learning_rate": 8.646712519062228e-06, "loss": 0.2988, "step": 7599 }, { "epoch": 0.5254061527825786, "grad_norm": 0.7325837558577382, "learning_rate": 8.646329537638959e-06, "loss": 0.3918, "step": 7600 }, { "epoch": 0.5254752851711026, "grad_norm": 0.7537292880905994, "learning_rate": 8.645946510515457e-06, "loss": 0.4371, "step": 7601 }, { "epoch": 0.5255444175596267, "grad_norm": 0.9781196783464451, "learning_rate": 8.645563437696523e-06, "loss": 0.3357, "step": 7602 }, { "epoch": 0.5256135499481507, "grad_norm": 0.9389530263790121, "learning_rate": 8.645180319186955e-06, "loss": 0.3763, "step": 7603 }, { "epoch": 0.5256826823366747, "grad_norm": 0.7868089071892341, "learning_rate": 8.644797154991558e-06, "loss": 0.4324, "step": 7604 }, { "epoch": 0.5257518147251987, "grad_norm": 0.7457582432380825, "learning_rate": 8.644413945115133e-06, "loss": 0.3899, "step": 7605 }, { "epoch": 0.5258209471137227, "grad_norm": 0.6826456806261508, "learning_rate": 8.644030689562483e-06, "loss": 0.364, "step": 7606 }, { "epoch": 0.5258900795022468, "grad_norm": 0.7384018293872416, "learning_rate": 8.643647388338411e-06, "loss": 0.3753, "step": 7607 }, { "epoch": 0.5259592118907708, "grad_norm": 0.6963048335010675, "learning_rate": 8.643264041447722e-06, "loss": 0.4249, "step": 7608 }, { "epoch": 0.5260283442792949, "grad_norm": 0.7951118072146034, "learning_rate": 8.642880648895219e-06, "loss": 0.2875, "step": 7609 }, { "epoch": 0.5260974766678189, "grad_norm": 0.8339883552837828, "learning_rate": 8.64249721068571e-06, "loss": 0.2616, "step": 7610 }, { "epoch": 0.5261666090563429, "grad_norm": 0.7121069507230068, "learning_rate": 8.642113726823997e-06, "loss": 0.3771, "step": 7611 }, { "epoch": 0.526235741444867, "grad_norm": 0.8635144314843931, "learning_rate": 8.641730197314888e-06, "loss": 0.3633, "step": 7612 }, { "epoch": 0.526304873833391, "grad_norm": 0.693154050474944, "learning_rate": 8.64134662216319e-06, "loss": 0.3309, "step": 7613 }, { "epoch": 0.526374006221915, "grad_norm": 0.7913331967238918, "learning_rate": 8.64096300137371e-06, "loss": 0.3591, "step": 7614 }, { "epoch": 0.526443138610439, "grad_norm": 0.7198393600502643, "learning_rate": 8.640579334951258e-06, "loss": 0.3942, "step": 7615 }, { "epoch": 0.526512270998963, "grad_norm": 0.7370681715639599, "learning_rate": 8.64019562290064e-06, "loss": 0.3009, "step": 7616 }, { "epoch": 0.526581403387487, "grad_norm": 0.6843128548535979, "learning_rate": 8.639811865226666e-06, "loss": 0.3033, "step": 7617 }, { "epoch": 0.5266505357760111, "grad_norm": 1.290773915401141, "learning_rate": 8.639428061934145e-06, "loss": 0.3967, "step": 7618 }, { "epoch": 0.5267196681645351, "grad_norm": 0.7292461800715415, "learning_rate": 8.63904421302789e-06, "loss": 0.3855, "step": 7619 }, { "epoch": 0.5267888005530591, "grad_norm": 0.6463221699443699, "learning_rate": 8.63866031851271e-06, "loss": 0.3106, "step": 7620 }, { "epoch": 0.5268579329415831, "grad_norm": 0.7413966805832901, "learning_rate": 8.638276378393415e-06, "loss": 0.3156, "step": 7621 }, { "epoch": 0.5269270653301071, "grad_norm": 0.7681092720440541, "learning_rate": 8.637892392674819e-06, "loss": 0.4078, "step": 7622 }, { "epoch": 0.5269961977186312, "grad_norm": 0.7000171844079864, "learning_rate": 8.637508361361733e-06, "loss": 0.3871, "step": 7623 }, { "epoch": 0.5270653301071552, "grad_norm": 0.6383937565778535, "learning_rate": 8.637124284458973e-06, "loss": 0.3323, "step": 7624 }, { "epoch": 0.5271344624956792, "grad_norm": 0.7054697132478852, "learning_rate": 8.636740161971349e-06, "loss": 0.3382, "step": 7625 }, { "epoch": 0.5272035948842032, "grad_norm": 0.705939479632571, "learning_rate": 8.636355993903677e-06, "loss": 0.4281, "step": 7626 }, { "epoch": 0.5272727272727272, "grad_norm": 0.8742059169662298, "learning_rate": 8.635971780260773e-06, "loss": 0.3021, "step": 7627 }, { "epoch": 0.5273418596612512, "grad_norm": 0.7864015803232663, "learning_rate": 8.635587521047452e-06, "loss": 0.3403, "step": 7628 }, { "epoch": 0.5274109920497754, "grad_norm": 0.7686166911823586, "learning_rate": 8.635203216268528e-06, "loss": 0.4629, "step": 7629 }, { "epoch": 0.5274801244382994, "grad_norm": 0.7632010413043447, "learning_rate": 8.63481886592882e-06, "loss": 0.391, "step": 7630 }, { "epoch": 0.5275492568268234, "grad_norm": 0.6757431472943846, "learning_rate": 8.634434470033144e-06, "loss": 0.2884, "step": 7631 }, { "epoch": 0.5276183892153474, "grad_norm": 0.7048340391256808, "learning_rate": 8.634050028586318e-06, "loss": 0.2825, "step": 7632 }, { "epoch": 0.5276875216038714, "grad_norm": 0.7896946668121501, "learning_rate": 8.633665541593159e-06, "loss": 0.4092, "step": 7633 }, { "epoch": 0.5277566539923955, "grad_norm": 0.6414866777808638, "learning_rate": 8.633281009058488e-06, "loss": 0.3419, "step": 7634 }, { "epoch": 0.5278257863809195, "grad_norm": 0.6808661574882717, "learning_rate": 8.632896430987124e-06, "loss": 0.3186, "step": 7635 }, { "epoch": 0.5278949187694435, "grad_norm": 0.6910350764205739, "learning_rate": 8.632511807383885e-06, "loss": 0.3383, "step": 7636 }, { "epoch": 0.5279640511579675, "grad_norm": 0.8279231856836863, "learning_rate": 8.632127138253594e-06, "loss": 0.3905, "step": 7637 }, { "epoch": 0.5280331835464915, "grad_norm": 1.4026260717167405, "learning_rate": 8.63174242360107e-06, "loss": 0.2877, "step": 7638 }, { "epoch": 0.5281023159350156, "grad_norm": 2.2884216154877075, "learning_rate": 8.631357663431138e-06, "loss": 0.3515, "step": 7639 }, { "epoch": 0.5281714483235396, "grad_norm": 0.8490866279464522, "learning_rate": 8.630972857748616e-06, "loss": 0.4584, "step": 7640 }, { "epoch": 0.5282405807120636, "grad_norm": 1.8660115994199702, "learning_rate": 8.630588006558331e-06, "loss": 0.3167, "step": 7641 }, { "epoch": 0.5283097131005876, "grad_norm": 0.7825851273773415, "learning_rate": 8.630203109865104e-06, "loss": 0.2922, "step": 7642 }, { "epoch": 0.5283788454891116, "grad_norm": 0.697018570854484, "learning_rate": 8.629818167673757e-06, "loss": 0.3401, "step": 7643 }, { "epoch": 0.5284479778776356, "grad_norm": 0.9779707415581214, "learning_rate": 8.62943317998912e-06, "loss": 0.4321, "step": 7644 }, { "epoch": 0.5285171102661597, "grad_norm": 0.6340804577208701, "learning_rate": 8.629048146816012e-06, "loss": 0.3256, "step": 7645 }, { "epoch": 0.5285862426546837, "grad_norm": 0.6861949650325202, "learning_rate": 8.628663068159264e-06, "loss": 0.3141, "step": 7646 }, { "epoch": 0.5286553750432077, "grad_norm": 0.6717976375300925, "learning_rate": 8.6282779440237e-06, "loss": 0.4272, "step": 7647 }, { "epoch": 0.5287245074317317, "grad_norm": 0.8842058761727984, "learning_rate": 8.627892774414147e-06, "loss": 0.292, "step": 7648 }, { "epoch": 0.5287936398202558, "grad_norm": 0.708221069717197, "learning_rate": 8.627507559335431e-06, "loss": 0.322, "step": 7649 }, { "epoch": 0.5288627722087799, "grad_norm": 0.7670872476890052, "learning_rate": 8.627122298792383e-06, "loss": 0.3533, "step": 7650 }, { "epoch": 0.5289319045973039, "grad_norm": 0.7386822863030075, "learning_rate": 8.626736992789828e-06, "loss": 0.4125, "step": 7651 }, { "epoch": 0.5290010369858279, "grad_norm": 0.722559825850742, "learning_rate": 8.626351641332599e-06, "loss": 0.372, "step": 7652 }, { "epoch": 0.5290701693743519, "grad_norm": 0.7928386961569086, "learning_rate": 8.625966244425521e-06, "loss": 0.2906, "step": 7653 }, { "epoch": 0.5291393017628759, "grad_norm": 0.7337084202221581, "learning_rate": 8.62558080207343e-06, "loss": 0.4206, "step": 7654 }, { "epoch": 0.5292084341514, "grad_norm": 0.7413810466252406, "learning_rate": 8.625195314281151e-06, "loss": 0.3867, "step": 7655 }, { "epoch": 0.529277566539924, "grad_norm": 0.7113787615295937, "learning_rate": 8.624809781053521e-06, "loss": 0.2742, "step": 7656 }, { "epoch": 0.529346698928448, "grad_norm": 0.7744649408231934, "learning_rate": 8.624424202395367e-06, "loss": 0.3045, "step": 7657 }, { "epoch": 0.529415831316972, "grad_norm": 0.6778974118154469, "learning_rate": 8.624038578311526e-06, "loss": 0.4577, "step": 7658 }, { "epoch": 0.529484963705496, "grad_norm": 1.0085309656359163, "learning_rate": 8.623652908806827e-06, "loss": 0.2818, "step": 7659 }, { "epoch": 0.52955409609402, "grad_norm": 0.8560144400897357, "learning_rate": 8.623267193886105e-06, "loss": 0.2874, "step": 7660 }, { "epoch": 0.5296232284825441, "grad_norm": 0.7017376294813894, "learning_rate": 8.622881433554195e-06, "loss": 0.4409, "step": 7661 }, { "epoch": 0.5296923608710681, "grad_norm": 0.6364432813528945, "learning_rate": 8.622495627815934e-06, "loss": 0.4122, "step": 7662 }, { "epoch": 0.5297614932595921, "grad_norm": 0.7092301460140253, "learning_rate": 8.622109776676152e-06, "loss": 0.2827, "step": 7663 }, { "epoch": 0.5298306256481161, "grad_norm": 0.7395962246256195, "learning_rate": 8.62172388013969e-06, "loss": 0.3199, "step": 7664 }, { "epoch": 0.5298997580366401, "grad_norm": 0.719389050240962, "learning_rate": 8.62133793821138e-06, "loss": 0.3947, "step": 7665 }, { "epoch": 0.5299688904251642, "grad_norm": 0.6578655364443787, "learning_rate": 8.620951950896063e-06, "loss": 0.2863, "step": 7666 }, { "epoch": 0.5300380228136882, "grad_norm": 0.6647877623008287, "learning_rate": 8.620565918198576e-06, "loss": 0.3098, "step": 7667 }, { "epoch": 0.5301071552022122, "grad_norm": 1.1523041407413313, "learning_rate": 8.620179840123754e-06, "loss": 0.3355, "step": 7668 }, { "epoch": 0.5301762875907362, "grad_norm": 0.7526413391393942, "learning_rate": 8.619793716676441e-06, "loss": 0.3779, "step": 7669 }, { "epoch": 0.5302454199792603, "grad_norm": 0.6899689943230893, "learning_rate": 8.619407547861473e-06, "loss": 0.3066, "step": 7670 }, { "epoch": 0.5303145523677844, "grad_norm": 0.7014660659835166, "learning_rate": 8.619021333683689e-06, "loss": 0.3112, "step": 7671 }, { "epoch": 0.5303836847563084, "grad_norm": 0.8004992694343569, "learning_rate": 8.618635074147933e-06, "loss": 0.4299, "step": 7672 }, { "epoch": 0.5304528171448324, "grad_norm": 0.73179515608075, "learning_rate": 8.61824876925904e-06, "loss": 0.3509, "step": 7673 }, { "epoch": 0.5305219495333564, "grad_norm": 0.6831244997775465, "learning_rate": 8.617862419021861e-06, "loss": 0.3718, "step": 7674 }, { "epoch": 0.5305910819218804, "grad_norm": 0.6781707459736555, "learning_rate": 8.61747602344123e-06, "loss": 0.3281, "step": 7675 }, { "epoch": 0.5306602143104044, "grad_norm": 0.6894160867863921, "learning_rate": 8.617089582521994e-06, "loss": 0.3915, "step": 7676 }, { "epoch": 0.5307293466989285, "grad_norm": 0.7061556252751965, "learning_rate": 8.616703096268994e-06, "loss": 0.3638, "step": 7677 }, { "epoch": 0.5307984790874525, "grad_norm": 0.7357522593459455, "learning_rate": 8.616316564687075e-06, "loss": 0.3247, "step": 7678 }, { "epoch": 0.5308676114759765, "grad_norm": 0.834330946291809, "learning_rate": 8.615929987781083e-06, "loss": 0.3943, "step": 7679 }, { "epoch": 0.5309367438645005, "grad_norm": 0.6660874980695387, "learning_rate": 8.615543365555859e-06, "loss": 0.3506, "step": 7680 }, { "epoch": 0.5310058762530245, "grad_norm": 0.6558433590542483, "learning_rate": 8.615156698016253e-06, "loss": 0.3295, "step": 7681 }, { "epoch": 0.5310750086415486, "grad_norm": 0.7301174649354604, "learning_rate": 8.614769985167109e-06, "loss": 0.2986, "step": 7682 }, { "epoch": 0.5311441410300726, "grad_norm": 0.7156910803257679, "learning_rate": 8.614383227013273e-06, "loss": 0.4222, "step": 7683 }, { "epoch": 0.5312132734185966, "grad_norm": 0.9236096668984758, "learning_rate": 8.613996423559594e-06, "loss": 0.3061, "step": 7684 }, { "epoch": 0.5312824058071206, "grad_norm": 0.6713160104876802, "learning_rate": 8.61360957481092e-06, "loss": 0.2893, "step": 7685 }, { "epoch": 0.5313515381956446, "grad_norm": 0.6705196059658621, "learning_rate": 8.613222680772097e-06, "loss": 0.3363, "step": 7686 }, { "epoch": 0.5314206705841686, "grad_norm": 0.6854566296983627, "learning_rate": 8.612835741447979e-06, "loss": 0.3559, "step": 7687 }, { "epoch": 0.5314898029726927, "grad_norm": 0.718238479355648, "learning_rate": 8.61244875684341e-06, "loss": 0.3306, "step": 7688 }, { "epoch": 0.5315589353612167, "grad_norm": 0.7165194967407049, "learning_rate": 8.612061726963242e-06, "loss": 0.268, "step": 7689 }, { "epoch": 0.5316280677497408, "grad_norm": 0.7406120573955306, "learning_rate": 8.611674651812327e-06, "loss": 0.4073, "step": 7690 }, { "epoch": 0.5316972001382648, "grad_norm": 0.7048852211137704, "learning_rate": 8.611287531395518e-06, "loss": 0.3424, "step": 7691 }, { "epoch": 0.5317663325267888, "grad_norm": 0.8346623606724397, "learning_rate": 8.61090036571766e-06, "loss": 0.2577, "step": 7692 }, { "epoch": 0.5318354649153129, "grad_norm": 0.7805002727683642, "learning_rate": 8.610513154783613e-06, "loss": 0.3538, "step": 7693 }, { "epoch": 0.5319045973038369, "grad_norm": 0.7540329442482315, "learning_rate": 8.610125898598225e-06, "loss": 0.4549, "step": 7694 }, { "epoch": 0.5319737296923609, "grad_norm": 0.6718496086663993, "learning_rate": 8.609738597166353e-06, "loss": 0.2985, "step": 7695 }, { "epoch": 0.5320428620808849, "grad_norm": 0.7101513090477213, "learning_rate": 8.609351250492849e-06, "loss": 0.3123, "step": 7696 }, { "epoch": 0.5321119944694089, "grad_norm": 1.073283047319174, "learning_rate": 8.608963858582569e-06, "loss": 0.5082, "step": 7697 }, { "epoch": 0.532181126857933, "grad_norm": 0.7072581209894768, "learning_rate": 8.608576421440367e-06, "loss": 0.3148, "step": 7698 }, { "epoch": 0.532250259246457, "grad_norm": 0.6271239033778339, "learning_rate": 8.608188939071099e-06, "loss": 0.2997, "step": 7699 }, { "epoch": 0.532319391634981, "grad_norm": 0.7130703858207906, "learning_rate": 8.607801411479622e-06, "loss": 0.3358, "step": 7700 }, { "epoch": 0.532388524023505, "grad_norm": 0.813859352247736, "learning_rate": 8.607413838670791e-06, "loss": 0.4409, "step": 7701 }, { "epoch": 0.532457656412029, "grad_norm": 0.8141668380675433, "learning_rate": 8.607026220649467e-06, "loss": 0.3042, "step": 7702 }, { "epoch": 0.532526788800553, "grad_norm": 0.9825931844508112, "learning_rate": 8.606638557420508e-06, "loss": 0.3222, "step": 7703 }, { "epoch": 0.5325959211890771, "grad_norm": 0.6821712825314971, "learning_rate": 8.60625084898877e-06, "loss": 0.3914, "step": 7704 }, { "epoch": 0.5326650535776011, "grad_norm": 0.6445862751028177, "learning_rate": 8.60586309535911e-06, "loss": 0.3485, "step": 7705 }, { "epoch": 0.5327341859661251, "grad_norm": 0.6257323242605765, "learning_rate": 8.605475296536395e-06, "loss": 0.2768, "step": 7706 }, { "epoch": 0.5328033183546491, "grad_norm": 0.9318683759282687, "learning_rate": 8.605087452525478e-06, "loss": 0.3404, "step": 7707 }, { "epoch": 0.5328724507431731, "grad_norm": 0.6538116304247075, "learning_rate": 8.604699563331226e-06, "loss": 0.374, "step": 7708 }, { "epoch": 0.5329415831316971, "grad_norm": 0.809754654173368, "learning_rate": 8.604311628958498e-06, "loss": 0.3036, "step": 7709 }, { "epoch": 0.5330107155202213, "grad_norm": 0.6755753967108241, "learning_rate": 8.603923649412153e-06, "loss": 0.3257, "step": 7710 }, { "epoch": 0.5330798479087453, "grad_norm": 0.757716277169688, "learning_rate": 8.603535624697058e-06, "loss": 0.3396, "step": 7711 }, { "epoch": 0.5331489802972693, "grad_norm": 0.663588716096797, "learning_rate": 8.603147554818077e-06, "loss": 0.3885, "step": 7712 }, { "epoch": 0.5332181126857933, "grad_norm": 0.6368802398633411, "learning_rate": 8.60275943978007e-06, "loss": 0.3, "step": 7713 }, { "epoch": 0.5332872450743174, "grad_norm": 1.155310520167106, "learning_rate": 8.602371279587902e-06, "loss": 0.2917, "step": 7714 }, { "epoch": 0.5333563774628414, "grad_norm": 0.7277447465723658, "learning_rate": 8.60198307424644e-06, "loss": 0.3546, "step": 7715 }, { "epoch": 0.5334255098513654, "grad_norm": 0.6768088405177033, "learning_rate": 8.601594823760547e-06, "loss": 0.2722, "step": 7716 }, { "epoch": 0.5334946422398894, "grad_norm": 0.7013496038204251, "learning_rate": 8.601206528135091e-06, "loss": 0.3244, "step": 7717 }, { "epoch": 0.5335637746284134, "grad_norm": 0.7519684927442722, "learning_rate": 8.600818187374938e-06, "loss": 0.3474, "step": 7718 }, { "epoch": 0.5336329070169374, "grad_norm": 0.7067556006041905, "learning_rate": 8.600429801484952e-06, "loss": 0.3989, "step": 7719 }, { "epoch": 0.5337020394054615, "grad_norm": 0.6382345689056911, "learning_rate": 8.600041370470008e-06, "loss": 0.2781, "step": 7720 }, { "epoch": 0.5337711717939855, "grad_norm": 0.6560221726783856, "learning_rate": 8.599652894334969e-06, "loss": 0.2926, "step": 7721 }, { "epoch": 0.5338403041825095, "grad_norm": 0.7433783716598233, "learning_rate": 8.599264373084704e-06, "loss": 0.4216, "step": 7722 }, { "epoch": 0.5339094365710335, "grad_norm": 0.6629741770101338, "learning_rate": 8.598875806724082e-06, "loss": 0.3228, "step": 7723 }, { "epoch": 0.5339785689595575, "grad_norm": 0.7181414847397704, "learning_rate": 8.598487195257976e-06, "loss": 0.2789, "step": 7724 }, { "epoch": 0.5340477013480815, "grad_norm": 0.7250969127614888, "learning_rate": 8.598098538691255e-06, "loss": 0.3254, "step": 7725 }, { "epoch": 0.5341168337366056, "grad_norm": 0.6646208845354065, "learning_rate": 8.597709837028789e-06, "loss": 0.4154, "step": 7726 }, { "epoch": 0.5341859661251296, "grad_norm": 1.0237641253021348, "learning_rate": 8.59732109027545e-06, "loss": 0.3049, "step": 7727 }, { "epoch": 0.5342550985136536, "grad_norm": 0.8104811769401393, "learning_rate": 8.596932298436113e-06, "loss": 0.3, "step": 7728 }, { "epoch": 0.5343242309021776, "grad_norm": 0.8472857192192451, "learning_rate": 8.596543461515648e-06, "loss": 0.5155, "step": 7729 }, { "epoch": 0.5343933632907016, "grad_norm": 0.6487388544328351, "learning_rate": 8.596154579518928e-06, "loss": 0.3259, "step": 7730 }, { "epoch": 0.5344624956792258, "grad_norm": 0.6990612278553852, "learning_rate": 8.595765652450827e-06, "loss": 0.2956, "step": 7731 }, { "epoch": 0.5345316280677498, "grad_norm": 0.7221211301332964, "learning_rate": 8.595376680316223e-06, "loss": 0.3133, "step": 7732 }, { "epoch": 0.5346007604562738, "grad_norm": 0.7064928439936186, "learning_rate": 8.594987663119986e-06, "loss": 0.4659, "step": 7733 }, { "epoch": 0.5346698928447978, "grad_norm": 0.8082326636845208, "learning_rate": 8.594598600866997e-06, "loss": 0.3098, "step": 7734 }, { "epoch": 0.5347390252333218, "grad_norm": 0.6885869686259142, "learning_rate": 8.594209493562127e-06, "loss": 0.353, "step": 7735 }, { "epoch": 0.5348081576218459, "grad_norm": 0.7416222724727434, "learning_rate": 8.593820341210257e-06, "loss": 0.4101, "step": 7736 }, { "epoch": 0.5348772900103699, "grad_norm": 0.7011385843275155, "learning_rate": 8.593431143816262e-06, "loss": 0.3359, "step": 7737 }, { "epoch": 0.5349464223988939, "grad_norm": 0.6898826746303591, "learning_rate": 8.59304190138502e-06, "loss": 0.3085, "step": 7738 }, { "epoch": 0.5350155547874179, "grad_norm": 0.7249698992670431, "learning_rate": 8.592652613921408e-06, "loss": 0.321, "step": 7739 }, { "epoch": 0.5350846871759419, "grad_norm": 0.7234405270595325, "learning_rate": 8.59226328143031e-06, "loss": 0.4584, "step": 7740 }, { "epoch": 0.535153819564466, "grad_norm": 0.7792548533938681, "learning_rate": 8.591873903916601e-06, "loss": 0.3608, "step": 7741 }, { "epoch": 0.53522295195299, "grad_norm": 0.6208815792424374, "learning_rate": 8.591484481385162e-06, "loss": 0.2575, "step": 7742 }, { "epoch": 0.535292084341514, "grad_norm": 0.7732273802525161, "learning_rate": 8.591095013840875e-06, "loss": 0.3356, "step": 7743 }, { "epoch": 0.535361216730038, "grad_norm": 0.820879833222494, "learning_rate": 8.59070550128862e-06, "loss": 0.4509, "step": 7744 }, { "epoch": 0.535430349118562, "grad_norm": 0.8628091686351206, "learning_rate": 8.590315943733282e-06, "loss": 0.3041, "step": 7745 }, { "epoch": 0.535499481507086, "grad_norm": 0.7121195531950782, "learning_rate": 8.589926341179738e-06, "loss": 0.2759, "step": 7746 }, { "epoch": 0.5355686138956101, "grad_norm": 0.7213236646128736, "learning_rate": 8.589536693632875e-06, "loss": 0.4703, "step": 7747 }, { "epoch": 0.5356377462841341, "grad_norm": 0.9207094755615584, "learning_rate": 8.589147001097576e-06, "loss": 0.289, "step": 7748 }, { "epoch": 0.5357068786726581, "grad_norm": 0.7025128380870046, "learning_rate": 8.588757263578722e-06, "loss": 0.2875, "step": 7749 }, { "epoch": 0.5357760110611821, "grad_norm": 0.8574438720865188, "learning_rate": 8.588367481081202e-06, "loss": 0.3336, "step": 7750 }, { "epoch": 0.5358451434497062, "grad_norm": 0.6772278130186239, "learning_rate": 8.5879776536099e-06, "loss": 0.3952, "step": 7751 }, { "epoch": 0.5359142758382303, "grad_norm": 0.755444138710279, "learning_rate": 8.5875877811697e-06, "loss": 0.2837, "step": 7752 }, { "epoch": 0.5359834082267543, "grad_norm": 1.1058376833865828, "learning_rate": 8.587197863765488e-06, "loss": 0.3338, "step": 7753 }, { "epoch": 0.5360525406152783, "grad_norm": 0.8828390455855477, "learning_rate": 8.586807901402155e-06, "loss": 0.4514, "step": 7754 }, { "epoch": 0.5361216730038023, "grad_norm": 0.8395634073975504, "learning_rate": 8.586417894084585e-06, "loss": 0.3739, "step": 7755 }, { "epoch": 0.5361908053923263, "grad_norm": 0.8141923958647914, "learning_rate": 8.586027841817668e-06, "loss": 0.3171, "step": 7756 }, { "epoch": 0.5362599377808503, "grad_norm": 0.7452766514372648, "learning_rate": 8.58563774460629e-06, "loss": 0.3439, "step": 7757 }, { "epoch": 0.5363290701693744, "grad_norm": 0.6736057168167016, "learning_rate": 8.585247602455342e-06, "loss": 0.4018, "step": 7758 }, { "epoch": 0.5363982025578984, "grad_norm": 0.6914100760724559, "learning_rate": 8.584857415369715e-06, "loss": 0.2919, "step": 7759 }, { "epoch": 0.5364673349464224, "grad_norm": 0.840460741838169, "learning_rate": 8.584467183354295e-06, "loss": 0.368, "step": 7760 }, { "epoch": 0.5365364673349464, "grad_norm": 0.6871774609769576, "learning_rate": 8.584076906413978e-06, "loss": 0.3641, "step": 7761 }, { "epoch": 0.5366055997234704, "grad_norm": 0.648891186882484, "learning_rate": 8.583686584553651e-06, "loss": 0.3826, "step": 7762 }, { "epoch": 0.5366747321119945, "grad_norm": 0.7063486006331637, "learning_rate": 8.58329621777821e-06, "loss": 0.3184, "step": 7763 }, { "epoch": 0.5367438645005185, "grad_norm": 0.7309914342291735, "learning_rate": 8.582905806092546e-06, "loss": 0.3734, "step": 7764 }, { "epoch": 0.5368129968890425, "grad_norm": 0.7002935888785912, "learning_rate": 8.58251534950155e-06, "loss": 0.44, "step": 7765 }, { "epoch": 0.5368821292775665, "grad_norm": 0.7052774725991694, "learning_rate": 8.582124848010118e-06, "loss": 0.3556, "step": 7766 }, { "epoch": 0.5369512616660905, "grad_norm": 1.027599902164386, "learning_rate": 8.581734301623143e-06, "loss": 0.3271, "step": 7767 }, { "epoch": 0.5370203940546145, "grad_norm": 0.7260626534313204, "learning_rate": 8.58134371034552e-06, "loss": 0.337, "step": 7768 }, { "epoch": 0.5370895264431386, "grad_norm": 0.9198653719443481, "learning_rate": 8.580953074182147e-06, "loss": 0.3831, "step": 7769 }, { "epoch": 0.5371586588316626, "grad_norm": 0.6348239599548662, "learning_rate": 8.580562393137916e-06, "loss": 0.2768, "step": 7770 }, { "epoch": 0.5372277912201867, "grad_norm": 0.7568375580887114, "learning_rate": 8.580171667217724e-06, "loss": 0.3133, "step": 7771 }, { "epoch": 0.5372969236087107, "grad_norm": 0.8062372969292357, "learning_rate": 8.579780896426473e-06, "loss": 0.3964, "step": 7772 }, { "epoch": 0.5373660559972347, "grad_norm": 0.7300528935177849, "learning_rate": 8.579390080769054e-06, "loss": 0.3578, "step": 7773 }, { "epoch": 0.5374351883857588, "grad_norm": 0.7081954533851608, "learning_rate": 8.57899922025037e-06, "loss": 0.3035, "step": 7774 }, { "epoch": 0.5375043207742828, "grad_norm": 0.7652652328522448, "learning_rate": 8.578608314875316e-06, "loss": 0.326, "step": 7775 }, { "epoch": 0.5375734531628068, "grad_norm": 0.7091558835486581, "learning_rate": 8.578217364648794e-06, "loss": 0.3771, "step": 7776 }, { "epoch": 0.5376425855513308, "grad_norm": 0.6387304878187697, "learning_rate": 8.577826369575702e-06, "loss": 0.3196, "step": 7777 }, { "epoch": 0.5377117179398548, "grad_norm": 0.6265014253914921, "learning_rate": 8.577435329660941e-06, "loss": 0.3378, "step": 7778 }, { "epoch": 0.5377808503283789, "grad_norm": 0.831365192231375, "learning_rate": 8.577044244909414e-06, "loss": 0.395, "step": 7779 }, { "epoch": 0.5378499827169029, "grad_norm": 0.7483879889326948, "learning_rate": 8.576653115326021e-06, "loss": 0.3984, "step": 7780 }, { "epoch": 0.5379191151054269, "grad_norm": 0.7111287381488457, "learning_rate": 8.576261940915663e-06, "loss": 0.3406, "step": 7781 }, { "epoch": 0.5379882474939509, "grad_norm": 0.753372769238642, "learning_rate": 8.575870721683243e-06, "loss": 0.2702, "step": 7782 }, { "epoch": 0.5380573798824749, "grad_norm": 0.7213534890548936, "learning_rate": 8.575479457633667e-06, "loss": 0.4496, "step": 7783 }, { "epoch": 0.538126512270999, "grad_norm": 0.9451293946205607, "learning_rate": 8.575088148771836e-06, "loss": 0.3618, "step": 7784 }, { "epoch": 0.538195644659523, "grad_norm": 0.6385697971775048, "learning_rate": 8.574696795102653e-06, "loss": 0.3019, "step": 7785 }, { "epoch": 0.538264777048047, "grad_norm": 0.7897376678295627, "learning_rate": 8.574305396631028e-06, "loss": 0.3843, "step": 7786 }, { "epoch": 0.538333909436571, "grad_norm": 0.6489261362129208, "learning_rate": 8.573913953361862e-06, "loss": 0.4319, "step": 7787 }, { "epoch": 0.538403041825095, "grad_norm": 0.6226102404946713, "learning_rate": 8.573522465300062e-06, "loss": 0.2933, "step": 7788 }, { "epoch": 0.538472174213619, "grad_norm": 0.8449632275370795, "learning_rate": 8.573130932450538e-06, "loss": 0.3156, "step": 7789 }, { "epoch": 0.538541306602143, "grad_norm": 0.9128986448258776, "learning_rate": 8.572739354818192e-06, "loss": 0.4071, "step": 7790 }, { "epoch": 0.5386104389906671, "grad_norm": 0.6708760830965391, "learning_rate": 8.572347732407934e-06, "loss": 0.3325, "step": 7791 }, { "epoch": 0.5386795713791912, "grad_norm": 0.6480865756042182, "learning_rate": 8.571956065224671e-06, "loss": 0.3252, "step": 7792 }, { "epoch": 0.5387487037677152, "grad_norm": 0.8178291432826195, "learning_rate": 8.571564353273315e-06, "loss": 0.4399, "step": 7793 }, { "epoch": 0.5388178361562392, "grad_norm": 0.7729347238337737, "learning_rate": 8.571172596558773e-06, "loss": 0.4198, "step": 7794 }, { "epoch": 0.5388869685447633, "grad_norm": 3.337193174140823, "learning_rate": 8.570780795085956e-06, "loss": 0.3063, "step": 7795 }, { "epoch": 0.5389561009332873, "grad_norm": 0.7379182669759452, "learning_rate": 8.570388948859774e-06, "loss": 0.2699, "step": 7796 }, { "epoch": 0.5390252333218113, "grad_norm": 0.6385828395530819, "learning_rate": 8.569997057885138e-06, "loss": 0.3923, "step": 7797 }, { "epoch": 0.5390943657103353, "grad_norm": 0.7910312802472687, "learning_rate": 8.56960512216696e-06, "loss": 0.3044, "step": 7798 }, { "epoch": 0.5391634980988593, "grad_norm": 0.8397517472500474, "learning_rate": 8.569213141710152e-06, "loss": 0.2777, "step": 7799 }, { "epoch": 0.5392326304873833, "grad_norm": 0.9082652231889331, "learning_rate": 8.568821116519627e-06, "loss": 0.3998, "step": 7800 }, { "epoch": 0.5393017628759074, "grad_norm": 0.7225398779469069, "learning_rate": 8.568429046600297e-06, "loss": 0.426, "step": 7801 }, { "epoch": 0.5393708952644314, "grad_norm": 0.7130074963705485, "learning_rate": 8.568036931957079e-06, "loss": 0.3284, "step": 7802 }, { "epoch": 0.5394400276529554, "grad_norm": 0.7122632912888623, "learning_rate": 8.567644772594885e-06, "loss": 0.3272, "step": 7803 }, { "epoch": 0.5395091600414794, "grad_norm": 0.6965091714690889, "learning_rate": 8.56725256851863e-06, "loss": 0.4035, "step": 7804 }, { "epoch": 0.5395782924300034, "grad_norm": 1.2403403853065107, "learning_rate": 8.56686031973323e-06, "loss": 0.3882, "step": 7805 }, { "epoch": 0.5396474248185275, "grad_norm": 0.7716889268066006, "learning_rate": 8.5664680262436e-06, "loss": 0.3348, "step": 7806 }, { "epoch": 0.5397165572070515, "grad_norm": 0.7481918984163154, "learning_rate": 8.56607568805466e-06, "loss": 0.3545, "step": 7807 }, { "epoch": 0.5397856895955755, "grad_norm": 0.7418019211959547, "learning_rate": 8.565683305171324e-06, "loss": 0.3667, "step": 7808 }, { "epoch": 0.5398548219840995, "grad_norm": 0.7506696442718837, "learning_rate": 8.56529087759851e-06, "loss": 0.3482, "step": 7809 }, { "epoch": 0.5399239543726235, "grad_norm": 0.7738393553165188, "learning_rate": 8.564898405341142e-06, "loss": 0.3016, "step": 7810 }, { "epoch": 0.5399930867611475, "grad_norm": 0.7043157286042319, "learning_rate": 8.56450588840413e-06, "loss": 0.3634, "step": 7811 }, { "epoch": 0.5400622191496717, "grad_norm": 0.6473519361925242, "learning_rate": 8.5641133267924e-06, "loss": 0.3933, "step": 7812 }, { "epoch": 0.5401313515381957, "grad_norm": 0.7688457023623138, "learning_rate": 8.563720720510869e-06, "loss": 0.2851, "step": 7813 }, { "epoch": 0.5402004839267197, "grad_norm": 0.8014328133957687, "learning_rate": 8.563328069564458e-06, "loss": 0.3556, "step": 7814 }, { "epoch": 0.5402696163152437, "grad_norm": 0.7181758638151653, "learning_rate": 8.562935373958088e-06, "loss": 0.4437, "step": 7815 }, { "epoch": 0.5403387487037677, "grad_norm": 0.692849378066667, "learning_rate": 8.562542633696683e-06, "loss": 0.3367, "step": 7816 }, { "epoch": 0.5404078810922918, "grad_norm": 0.6633463467738352, "learning_rate": 8.562149848785163e-06, "loss": 0.3088, "step": 7817 }, { "epoch": 0.5404770134808158, "grad_norm": 0.7590491112636292, "learning_rate": 8.561757019228452e-06, "loss": 0.3307, "step": 7818 }, { "epoch": 0.5405461458693398, "grad_norm": 0.6728011116173911, "learning_rate": 8.561364145031474e-06, "loss": 0.3834, "step": 7819 }, { "epoch": 0.5406152782578638, "grad_norm": 0.8097880481319198, "learning_rate": 8.560971226199151e-06, "loss": 0.2991, "step": 7820 }, { "epoch": 0.5406844106463878, "grad_norm": 0.7289930938559579, "learning_rate": 8.560578262736407e-06, "loss": 0.3379, "step": 7821 }, { "epoch": 0.5407535430349119, "grad_norm": 0.7441531216158865, "learning_rate": 8.56018525464817e-06, "loss": 0.3888, "step": 7822 }, { "epoch": 0.5408226754234359, "grad_norm": 0.8598374363199609, "learning_rate": 8.559792201939365e-06, "loss": 0.3592, "step": 7823 }, { "epoch": 0.5408918078119599, "grad_norm": 0.6437300326394445, "learning_rate": 8.559399104614915e-06, "loss": 0.3183, "step": 7824 }, { "epoch": 0.5409609402004839, "grad_norm": 0.7224551349858354, "learning_rate": 8.559005962679752e-06, "loss": 0.3691, "step": 7825 }, { "epoch": 0.5410300725890079, "grad_norm": 0.8505071015897682, "learning_rate": 8.5586127761388e-06, "loss": 0.4288, "step": 7826 }, { "epoch": 0.541099204977532, "grad_norm": 0.7381928637989621, "learning_rate": 8.558219544996986e-06, "loss": 0.3157, "step": 7827 }, { "epoch": 0.541168337366056, "grad_norm": 0.6813716354683569, "learning_rate": 8.557826269259242e-06, "loss": 0.3454, "step": 7828 }, { "epoch": 0.54123746975458, "grad_norm": 0.6673909013125208, "learning_rate": 8.557432948930494e-06, "loss": 0.3935, "step": 7829 }, { "epoch": 0.541306602143104, "grad_norm": 0.8604187738074474, "learning_rate": 8.557039584015673e-06, "loss": 0.3344, "step": 7830 }, { "epoch": 0.541375734531628, "grad_norm": 0.7782373591549796, "learning_rate": 8.556646174519707e-06, "loss": 0.3546, "step": 7831 }, { "epoch": 0.5414448669201521, "grad_norm": 0.7056952898969516, "learning_rate": 8.556252720447531e-06, "loss": 0.2755, "step": 7832 }, { "epoch": 0.5415139993086762, "grad_norm": 0.7081899562103438, "learning_rate": 8.55585922180407e-06, "loss": 0.4756, "step": 7833 }, { "epoch": 0.5415831316972002, "grad_norm": 0.6729664905353879, "learning_rate": 8.555465678594263e-06, "loss": 0.3434, "step": 7834 }, { "epoch": 0.5416522640857242, "grad_norm": 0.7826549708147953, "learning_rate": 8.555072090823037e-06, "loss": 0.3282, "step": 7835 }, { "epoch": 0.5417213964742482, "grad_norm": 0.7678899520618335, "learning_rate": 8.554678458495327e-06, "loss": 0.4535, "step": 7836 }, { "epoch": 0.5417905288627722, "grad_norm": 0.6442318651340861, "learning_rate": 8.554284781616066e-06, "loss": 0.3784, "step": 7837 }, { "epoch": 0.5418596612512963, "grad_norm": 0.652234981125277, "learning_rate": 8.553891060190187e-06, "loss": 0.3365, "step": 7838 }, { "epoch": 0.5419287936398203, "grad_norm": 0.7806046659320509, "learning_rate": 8.553497294222627e-06, "loss": 0.3149, "step": 7839 }, { "epoch": 0.5419979260283443, "grad_norm": 0.7294764629976096, "learning_rate": 8.55310348371832e-06, "loss": 0.443, "step": 7840 }, { "epoch": 0.5420670584168683, "grad_norm": 0.7071497478132337, "learning_rate": 8.5527096286822e-06, "loss": 0.3301, "step": 7841 }, { "epoch": 0.5421361908053923, "grad_norm": 0.9113705157953849, "learning_rate": 8.552315729119207e-06, "loss": 0.3186, "step": 7842 }, { "epoch": 0.5422053231939163, "grad_norm": 0.7408397251799641, "learning_rate": 8.551921785034277e-06, "loss": 0.3764, "step": 7843 }, { "epoch": 0.5422744555824404, "grad_norm": 0.7394446371195824, "learning_rate": 8.551527796432342e-06, "loss": 0.3786, "step": 7844 }, { "epoch": 0.5423435879709644, "grad_norm": 0.8798830638356673, "learning_rate": 8.551133763318347e-06, "loss": 0.402, "step": 7845 }, { "epoch": 0.5424127203594884, "grad_norm": 0.746745909878584, "learning_rate": 8.550739685697228e-06, "loss": 0.253, "step": 7846 }, { "epoch": 0.5424818527480124, "grad_norm": 0.8254035143404087, "learning_rate": 8.55034556357392e-06, "loss": 0.406, "step": 7847 }, { "epoch": 0.5425509851365364, "grad_norm": 0.7455106482158916, "learning_rate": 8.54995139695337e-06, "loss": 0.3795, "step": 7848 }, { "epoch": 0.5426201175250605, "grad_norm": 0.6875028829404589, "learning_rate": 8.549557185840512e-06, "loss": 0.2765, "step": 7849 }, { "epoch": 0.5426892499135845, "grad_norm": 0.9058310999653347, "learning_rate": 8.54916293024029e-06, "loss": 0.3351, "step": 7850 }, { "epoch": 0.5427583823021085, "grad_norm": 0.7504028342661823, "learning_rate": 8.548768630157645e-06, "loss": 0.4012, "step": 7851 }, { "epoch": 0.5428275146906325, "grad_norm": 0.7142422522456767, "learning_rate": 8.548374285597519e-06, "loss": 0.2954, "step": 7852 }, { "epoch": 0.5428966470791566, "grad_norm": 0.9252422413448195, "learning_rate": 8.547979896564852e-06, "loss": 0.3279, "step": 7853 }, { "epoch": 0.5429657794676807, "grad_norm": 0.8048151755312773, "learning_rate": 8.54758546306459e-06, "loss": 0.4101, "step": 7854 }, { "epoch": 0.5430349118562047, "grad_norm": 0.7286656070401458, "learning_rate": 8.547190985101673e-06, "loss": 0.3573, "step": 7855 }, { "epoch": 0.5431040442447287, "grad_norm": 0.674729145486168, "learning_rate": 8.546796462681048e-06, "loss": 0.2502, "step": 7856 }, { "epoch": 0.5431731766332527, "grad_norm": 0.7424621369510143, "learning_rate": 8.546401895807662e-06, "loss": 0.3395, "step": 7857 }, { "epoch": 0.5432423090217767, "grad_norm": 0.7311215860615293, "learning_rate": 8.546007284486452e-06, "loss": 0.4228, "step": 7858 }, { "epoch": 0.5433114414103007, "grad_norm": 0.6680194126917167, "learning_rate": 8.545612628722373e-06, "loss": 0.33, "step": 7859 }, { "epoch": 0.5433805737988248, "grad_norm": 0.6741662009598802, "learning_rate": 8.545217928520363e-06, "loss": 0.3363, "step": 7860 }, { "epoch": 0.5434497061873488, "grad_norm": 0.6716706586069405, "learning_rate": 8.544823183885377e-06, "loss": 0.4054, "step": 7861 }, { "epoch": 0.5435188385758728, "grad_norm": 0.7029744350525929, "learning_rate": 8.544428394822357e-06, "loss": 0.4012, "step": 7862 }, { "epoch": 0.5435879709643968, "grad_norm": 0.641196266171862, "learning_rate": 8.544033561336253e-06, "loss": 0.3164, "step": 7863 }, { "epoch": 0.5436571033529208, "grad_norm": 0.7619576895392031, "learning_rate": 8.543638683432011e-06, "loss": 0.357, "step": 7864 }, { "epoch": 0.5437262357414449, "grad_norm": 0.7087204995889512, "learning_rate": 8.543243761114583e-06, "loss": 0.4185, "step": 7865 }, { "epoch": 0.5437953681299689, "grad_norm": 0.7164034145885424, "learning_rate": 8.542848794388918e-06, "loss": 0.284, "step": 7866 }, { "epoch": 0.5438645005184929, "grad_norm": 0.846569507896518, "learning_rate": 8.542453783259967e-06, "loss": 0.3188, "step": 7867 }, { "epoch": 0.5439336329070169, "grad_norm": 0.7169676440482492, "learning_rate": 8.542058727732679e-06, "loss": 0.3624, "step": 7868 }, { "epoch": 0.5440027652955409, "grad_norm": 0.6896164905757156, "learning_rate": 8.541663627812005e-06, "loss": 0.3869, "step": 7869 }, { "epoch": 0.5440718976840649, "grad_norm": 0.8739664784870801, "learning_rate": 8.541268483502898e-06, "loss": 0.3277, "step": 7870 }, { "epoch": 0.544141030072589, "grad_norm": 0.7181012382111841, "learning_rate": 8.54087329481031e-06, "loss": 0.3226, "step": 7871 }, { "epoch": 0.544210162461113, "grad_norm": 0.8146538322424581, "learning_rate": 8.540478061739195e-06, "loss": 0.4653, "step": 7872 }, { "epoch": 0.5442792948496371, "grad_norm": 0.6502044895853705, "learning_rate": 8.540082784294505e-06, "loss": 0.3457, "step": 7873 }, { "epoch": 0.5443484272381611, "grad_norm": 0.6719664618712549, "learning_rate": 8.539687462481195e-06, "loss": 0.2968, "step": 7874 }, { "epoch": 0.5444175596266851, "grad_norm": 0.7349299787309798, "learning_rate": 8.539292096304218e-06, "loss": 0.4017, "step": 7875 }, { "epoch": 0.5444866920152092, "grad_norm": 0.8490820180669851, "learning_rate": 8.538896685768534e-06, "loss": 0.4506, "step": 7876 }, { "epoch": 0.5445558244037332, "grad_norm": 0.6425518182428853, "learning_rate": 8.538501230879095e-06, "loss": 0.2733, "step": 7877 }, { "epoch": 0.5446249567922572, "grad_norm": 0.6143773100921365, "learning_rate": 8.538105731640855e-06, "loss": 0.2924, "step": 7878 }, { "epoch": 0.5446940891807812, "grad_norm": 0.7982902905092533, "learning_rate": 8.537710188058777e-06, "loss": 0.3775, "step": 7879 }, { "epoch": 0.5447632215693052, "grad_norm": 0.675474795843583, "learning_rate": 8.537314600137814e-06, "loss": 0.4193, "step": 7880 }, { "epoch": 0.5448323539578293, "grad_norm": 0.656585196836492, "learning_rate": 8.536918967882923e-06, "loss": 0.3559, "step": 7881 }, { "epoch": 0.5449014863463533, "grad_norm": 0.7339796577580815, "learning_rate": 8.536523291299066e-06, "loss": 0.3246, "step": 7882 }, { "epoch": 0.5449706187348773, "grad_norm": 0.7534028925014938, "learning_rate": 8.536127570391202e-06, "loss": 0.4874, "step": 7883 }, { "epoch": 0.5450397511234013, "grad_norm": 0.6888900141363982, "learning_rate": 8.535731805164288e-06, "loss": 0.349, "step": 7884 }, { "epoch": 0.5451088835119253, "grad_norm": 0.6524794265984144, "learning_rate": 8.535335995623285e-06, "loss": 0.297, "step": 7885 }, { "epoch": 0.5451780159004493, "grad_norm": 0.6814350842864132, "learning_rate": 8.534940141773157e-06, "loss": 0.357, "step": 7886 }, { "epoch": 0.5452471482889734, "grad_norm": 0.7530130118554842, "learning_rate": 8.53454424361886e-06, "loss": 0.4744, "step": 7887 }, { "epoch": 0.5453162806774974, "grad_norm": 0.6438631975969447, "learning_rate": 8.534148301165361e-06, "loss": 0.2734, "step": 7888 }, { "epoch": 0.5453854130660214, "grad_norm": 0.7676216202251058, "learning_rate": 8.533752314417619e-06, "loss": 0.2855, "step": 7889 }, { "epoch": 0.5454545454545454, "grad_norm": 0.7007473856796159, "learning_rate": 8.533356283380598e-06, "loss": 0.3863, "step": 7890 }, { "epoch": 0.5455236778430694, "grad_norm": 0.6736427803400745, "learning_rate": 8.532960208059261e-06, "loss": 0.3278, "step": 7891 }, { "epoch": 0.5455928102315935, "grad_norm": 0.844440475007637, "learning_rate": 8.532564088458573e-06, "loss": 0.3207, "step": 7892 }, { "epoch": 0.5456619426201176, "grad_norm": 0.8916466426448553, "learning_rate": 8.532167924583498e-06, "loss": 0.3515, "step": 7893 }, { "epoch": 0.5457310750086416, "grad_norm": 0.7699588288856335, "learning_rate": 8.531771716439e-06, "loss": 0.3575, "step": 7894 }, { "epoch": 0.5458002073971656, "grad_norm": 0.7897957270430901, "learning_rate": 8.53137546403005e-06, "loss": 0.3523, "step": 7895 }, { "epoch": 0.5458693397856896, "grad_norm": 0.7563317965623018, "learning_rate": 8.530979167361609e-06, "loss": 0.3594, "step": 7896 }, { "epoch": 0.5459384721742137, "grad_norm": 0.725423899252532, "learning_rate": 8.530582826438646e-06, "loss": 0.4448, "step": 7897 }, { "epoch": 0.5460076045627377, "grad_norm": 0.6958796555328497, "learning_rate": 8.530186441266125e-06, "loss": 0.3502, "step": 7898 }, { "epoch": 0.5460767369512617, "grad_norm": 0.6827760947157621, "learning_rate": 8.529790011849019e-06, "loss": 0.302, "step": 7899 }, { "epoch": 0.5461458693397857, "grad_norm": 0.683215897720537, "learning_rate": 8.529393538192294e-06, "loss": 0.3509, "step": 7900 }, { "epoch": 0.5462150017283097, "grad_norm": 0.717754880947297, "learning_rate": 8.52899702030092e-06, "loss": 0.4643, "step": 7901 }, { "epoch": 0.5462841341168337, "grad_norm": 0.7470394623540357, "learning_rate": 8.528600458179865e-06, "loss": 0.3276, "step": 7902 }, { "epoch": 0.5463532665053578, "grad_norm": 0.6905945034535304, "learning_rate": 8.528203851834102e-06, "loss": 0.2897, "step": 7903 }, { "epoch": 0.5464223988938818, "grad_norm": 0.7289392162981326, "learning_rate": 8.5278072012686e-06, "loss": 0.3387, "step": 7904 }, { "epoch": 0.5464915312824058, "grad_norm": 0.6892507643673852, "learning_rate": 8.52741050648833e-06, "loss": 0.3251, "step": 7905 }, { "epoch": 0.5465606636709298, "grad_norm": 0.7979133413984274, "learning_rate": 8.527013767498262e-06, "loss": 0.3112, "step": 7906 }, { "epoch": 0.5466297960594538, "grad_norm": 0.7045200685627759, "learning_rate": 8.526616984303374e-06, "loss": 0.3205, "step": 7907 }, { "epoch": 0.5466989284479778, "grad_norm": 0.6813922111791473, "learning_rate": 8.526220156908633e-06, "loss": 0.3703, "step": 7908 }, { "epoch": 0.5467680608365019, "grad_norm": 0.7594653914003451, "learning_rate": 8.525823285319017e-06, "loss": 0.3334, "step": 7909 }, { "epoch": 0.5468371932250259, "grad_norm": 0.7114775803745574, "learning_rate": 8.525426369539498e-06, "loss": 0.3532, "step": 7910 }, { "epoch": 0.5469063256135499, "grad_norm": 0.7095060256162466, "learning_rate": 8.52502940957505e-06, "loss": 0.3625, "step": 7911 }, { "epoch": 0.5469754580020739, "grad_norm": 0.7535088526916838, "learning_rate": 8.524632405430649e-06, "loss": 0.4286, "step": 7912 }, { "epoch": 0.5470445903905979, "grad_norm": 0.6663826128315928, "learning_rate": 8.524235357111272e-06, "loss": 0.2879, "step": 7913 }, { "epoch": 0.5471137227791221, "grad_norm": 0.6676537857176755, "learning_rate": 8.523838264621894e-06, "loss": 0.3349, "step": 7914 }, { "epoch": 0.5471828551676461, "grad_norm": 0.7127913009774681, "learning_rate": 8.52344112796749e-06, "loss": 0.4519, "step": 7915 }, { "epoch": 0.5472519875561701, "grad_norm": 0.6633658221785019, "learning_rate": 8.52304394715304e-06, "loss": 0.3447, "step": 7916 }, { "epoch": 0.5473211199446941, "grad_norm": 0.6906742490467553, "learning_rate": 8.522646722183523e-06, "loss": 0.2807, "step": 7917 }, { "epoch": 0.5473902523332181, "grad_norm": 0.7178522303577852, "learning_rate": 8.522249453063914e-06, "loss": 0.3511, "step": 7918 }, { "epoch": 0.5474593847217422, "grad_norm": 0.7269428938543038, "learning_rate": 8.521852139799196e-06, "loss": 0.4334, "step": 7919 }, { "epoch": 0.5475285171102662, "grad_norm": 0.7634260251949061, "learning_rate": 8.521454782394342e-06, "loss": 0.3014, "step": 7920 }, { "epoch": 0.5475976494987902, "grad_norm": 0.6457210811292101, "learning_rate": 8.52105738085434e-06, "loss": 0.2847, "step": 7921 }, { "epoch": 0.5476667818873142, "grad_norm": 0.6818341089988518, "learning_rate": 8.520659935184168e-06, "loss": 0.3685, "step": 7922 }, { "epoch": 0.5477359142758382, "grad_norm": 0.6625052341460893, "learning_rate": 8.520262445388805e-06, "loss": 0.3676, "step": 7923 }, { "epoch": 0.5478050466643622, "grad_norm": 0.6814150716050065, "learning_rate": 8.519864911473235e-06, "loss": 0.3307, "step": 7924 }, { "epoch": 0.5478741790528863, "grad_norm": 0.7302262820706444, "learning_rate": 8.51946733344244e-06, "loss": 0.3709, "step": 7925 }, { "epoch": 0.5479433114414103, "grad_norm": 0.679715438558678, "learning_rate": 8.519069711301403e-06, "loss": 0.4317, "step": 7926 }, { "epoch": 0.5480124438299343, "grad_norm": 0.6353068161674847, "learning_rate": 8.518672045055106e-06, "loss": 0.3451, "step": 7927 }, { "epoch": 0.5480815762184583, "grad_norm": 0.704473538576459, "learning_rate": 8.518274334708537e-06, "loss": 0.2953, "step": 7928 }, { "epoch": 0.5481507086069823, "grad_norm": 0.7756111382943982, "learning_rate": 8.517876580266674e-06, "loss": 0.4513, "step": 7929 }, { "epoch": 0.5482198409955064, "grad_norm": 0.6945967942619298, "learning_rate": 8.517478781734507e-06, "loss": 0.3296, "step": 7930 }, { "epoch": 0.5482889733840304, "grad_norm": 0.6725410062689673, "learning_rate": 8.517080939117022e-06, "loss": 0.3644, "step": 7931 }, { "epoch": 0.5483581057725544, "grad_norm": 0.7195351530250159, "learning_rate": 8.516683052419204e-06, "loss": 0.3562, "step": 7932 }, { "epoch": 0.5484272381610784, "grad_norm": 0.8524470426231502, "learning_rate": 8.51628512164604e-06, "loss": 0.4164, "step": 7933 }, { "epoch": 0.5484963705496025, "grad_norm": 0.6315762803200823, "learning_rate": 8.515887146802514e-06, "loss": 0.2522, "step": 7934 }, { "epoch": 0.5485655029381266, "grad_norm": 0.6684414938153137, "learning_rate": 8.515489127893622e-06, "loss": 0.267, "step": 7935 }, { "epoch": 0.5486346353266506, "grad_norm": 0.7769575931316853, "learning_rate": 8.515091064924343e-06, "loss": 0.3752, "step": 7936 }, { "epoch": 0.5487037677151746, "grad_norm": 0.7082364089640263, "learning_rate": 8.514692957899673e-06, "loss": 0.3959, "step": 7937 }, { "epoch": 0.5487729001036986, "grad_norm": 0.7203192842291777, "learning_rate": 8.514294806824598e-06, "loss": 0.3109, "step": 7938 }, { "epoch": 0.5488420324922226, "grad_norm": 0.6423503471618106, "learning_rate": 8.513896611704108e-06, "loss": 0.348, "step": 7939 }, { "epoch": 0.5489111648807466, "grad_norm": 0.6177969406373922, "learning_rate": 8.513498372543197e-06, "loss": 0.3529, "step": 7940 }, { "epoch": 0.5489802972692707, "grad_norm": 0.7314666004949346, "learning_rate": 8.513100089346853e-06, "loss": 0.3529, "step": 7941 }, { "epoch": 0.5490494296577947, "grad_norm": 0.7411111426078094, "learning_rate": 8.51270176212007e-06, "loss": 0.279, "step": 7942 }, { "epoch": 0.5491185620463187, "grad_norm": 0.737668184529939, "learning_rate": 8.512303390867837e-06, "loss": 0.4028, "step": 7943 }, { "epoch": 0.5491876944348427, "grad_norm": 0.7105603138148818, "learning_rate": 8.51190497559515e-06, "loss": 0.4303, "step": 7944 }, { "epoch": 0.5492568268233667, "grad_norm": 0.6992849816038218, "learning_rate": 8.511506516307001e-06, "loss": 0.3461, "step": 7945 }, { "epoch": 0.5493259592118908, "grad_norm": 0.9298787545480309, "learning_rate": 8.511108013008387e-06, "loss": 0.2514, "step": 7946 }, { "epoch": 0.5493950916004148, "grad_norm": 0.7125618684714818, "learning_rate": 8.510709465704297e-06, "loss": 0.446, "step": 7947 }, { "epoch": 0.5494642239889388, "grad_norm": 0.6726373824573987, "learning_rate": 8.51031087439973e-06, "loss": 0.3045, "step": 7948 }, { "epoch": 0.5495333563774628, "grad_norm": 0.6623372769140344, "learning_rate": 8.50991223909968e-06, "loss": 0.3112, "step": 7949 }, { "epoch": 0.5496024887659868, "grad_norm": 0.7032778477416527, "learning_rate": 8.509513559809144e-06, "loss": 0.3331, "step": 7950 }, { "epoch": 0.5496716211545108, "grad_norm": 0.6831876123946428, "learning_rate": 8.50911483653312e-06, "loss": 0.4297, "step": 7951 }, { "epoch": 0.5497407535430349, "grad_norm": 0.7050839504081, "learning_rate": 8.508716069276603e-06, "loss": 0.3044, "step": 7952 }, { "epoch": 0.5498098859315589, "grad_norm": 0.695772363718387, "learning_rate": 8.50831725804459e-06, "loss": 0.2588, "step": 7953 }, { "epoch": 0.549879018320083, "grad_norm": 0.776867649173738, "learning_rate": 8.507918402842083e-06, "loss": 0.3623, "step": 7954 }, { "epoch": 0.549948150708607, "grad_norm": 0.69286547493345, "learning_rate": 8.507519503674078e-06, "loss": 0.3508, "step": 7955 }, { "epoch": 0.550017283097131, "grad_norm": 0.6344651349810909, "learning_rate": 8.507120560545576e-06, "loss": 0.2648, "step": 7956 }, { "epoch": 0.5500864154856551, "grad_norm": 0.6875380928312224, "learning_rate": 8.506721573461577e-06, "loss": 0.3159, "step": 7957 }, { "epoch": 0.5501555478741791, "grad_norm": 0.8620110738588455, "learning_rate": 8.506322542427079e-06, "loss": 0.42, "step": 7958 }, { "epoch": 0.5502246802627031, "grad_norm": 0.8126471938146211, "learning_rate": 8.505923467447088e-06, "loss": 0.3199, "step": 7959 }, { "epoch": 0.5502938126512271, "grad_norm": 0.7731749738987975, "learning_rate": 8.505524348526601e-06, "loss": 0.2679, "step": 7960 }, { "epoch": 0.5503629450397511, "grad_norm": 0.6697365971824352, "learning_rate": 8.505125185670623e-06, "loss": 0.359, "step": 7961 }, { "epoch": 0.5504320774282752, "grad_norm": 0.6899861424510744, "learning_rate": 8.504725978884154e-06, "loss": 0.4093, "step": 7962 }, { "epoch": 0.5505012098167992, "grad_norm": 0.7155699532525831, "learning_rate": 8.504326728172202e-06, "loss": 0.3137, "step": 7963 }, { "epoch": 0.5505703422053232, "grad_norm": 0.78596695382335, "learning_rate": 8.503927433539767e-06, "loss": 0.3313, "step": 7964 }, { "epoch": 0.5506394745938472, "grad_norm": 0.6802640581269906, "learning_rate": 8.503528094991856e-06, "loss": 0.3996, "step": 7965 }, { "epoch": 0.5507086069823712, "grad_norm": 0.7580911571602281, "learning_rate": 8.50312871253347e-06, "loss": 0.3543, "step": 7966 }, { "epoch": 0.5507777393708952, "grad_norm": 0.9122038832639876, "learning_rate": 8.502729286169619e-06, "loss": 0.316, "step": 7967 }, { "epoch": 0.5508468717594193, "grad_norm": 0.7414354349051043, "learning_rate": 8.502329815905307e-06, "loss": 0.3601, "step": 7968 }, { "epoch": 0.5509160041479433, "grad_norm": 0.6966831231409074, "learning_rate": 8.50193030174554e-06, "loss": 0.4154, "step": 7969 }, { "epoch": 0.5509851365364673, "grad_norm": 0.8367352881531545, "learning_rate": 8.501530743695328e-06, "loss": 0.2942, "step": 7970 }, { "epoch": 0.5510542689249913, "grad_norm": 0.6776672019973303, "learning_rate": 8.501131141759675e-06, "loss": 0.3546, "step": 7971 }, { "epoch": 0.5511234013135153, "grad_norm": 0.6635063537942042, "learning_rate": 8.500731495943592e-06, "loss": 0.3755, "step": 7972 }, { "epoch": 0.5511925337020394, "grad_norm": 0.9916574084922388, "learning_rate": 8.500331806252089e-06, "loss": 0.3182, "step": 7973 }, { "epoch": 0.5512616660905634, "grad_norm": 0.7340370644995856, "learning_rate": 8.499932072690171e-06, "loss": 0.3069, "step": 7974 }, { "epoch": 0.5513307984790875, "grad_norm": 0.7818118580969101, "learning_rate": 8.499532295262851e-06, "loss": 0.3432, "step": 7975 }, { "epoch": 0.5513999308676115, "grad_norm": 0.6913568886439453, "learning_rate": 8.499132473975138e-06, "loss": 0.4076, "step": 7976 }, { "epoch": 0.5514690632561355, "grad_norm": 0.6410473744006263, "learning_rate": 8.498732608832046e-06, "loss": 0.3153, "step": 7977 }, { "epoch": 0.5515381956446596, "grad_norm": 0.5843345482023914, "learning_rate": 8.498332699838583e-06, "loss": 0.211, "step": 7978 }, { "epoch": 0.5516073280331836, "grad_norm": 0.758075562246195, "learning_rate": 8.497932746999764e-06, "loss": 0.3443, "step": 7979 }, { "epoch": 0.5516764604217076, "grad_norm": 0.7122457657203013, "learning_rate": 8.497532750320598e-06, "loss": 0.445, "step": 7980 }, { "epoch": 0.5517455928102316, "grad_norm": 0.6297445997724647, "learning_rate": 8.497132709806103e-06, "loss": 0.2494, "step": 7981 }, { "epoch": 0.5518147251987556, "grad_norm": 0.8112621944689509, "learning_rate": 8.496732625461289e-06, "loss": 0.2804, "step": 7982 }, { "epoch": 0.5518838575872796, "grad_norm": 0.667043307028663, "learning_rate": 8.496332497291173e-06, "loss": 0.4793, "step": 7983 }, { "epoch": 0.5519529899758037, "grad_norm": 0.7113766096766865, "learning_rate": 8.495932325300766e-06, "loss": 0.3114, "step": 7984 }, { "epoch": 0.5520221223643277, "grad_norm": 0.8852948676684742, "learning_rate": 8.495532109495088e-06, "loss": 0.3619, "step": 7985 }, { "epoch": 0.5520912547528517, "grad_norm": 0.6348311545840808, "learning_rate": 8.495131849879153e-06, "loss": 0.3505, "step": 7986 }, { "epoch": 0.5521603871413757, "grad_norm": 0.780031496734543, "learning_rate": 8.494731546457976e-06, "loss": 0.3432, "step": 7987 }, { "epoch": 0.5522295195298997, "grad_norm": 0.819544589792929, "learning_rate": 8.494331199236578e-06, "loss": 0.2679, "step": 7988 }, { "epoch": 0.5522986519184238, "grad_norm": 0.7340126708413396, "learning_rate": 8.493930808219974e-06, "loss": 0.3568, "step": 7989 }, { "epoch": 0.5523677843069478, "grad_norm": 0.765481068995109, "learning_rate": 8.493530373413182e-06, "loss": 0.3946, "step": 7990 }, { "epoch": 0.5524369166954718, "grad_norm": 0.644837305851408, "learning_rate": 8.49312989482122e-06, "loss": 0.2537, "step": 7991 }, { "epoch": 0.5525060490839958, "grad_norm": 0.9376798149890616, "learning_rate": 8.492729372449112e-06, "loss": 0.3554, "step": 7992 }, { "epoch": 0.5525751814725198, "grad_norm": 0.8427250117140501, "learning_rate": 8.492328806301872e-06, "loss": 0.3908, "step": 7993 }, { "epoch": 0.5526443138610438, "grad_norm": 0.7605577584155352, "learning_rate": 8.491928196384522e-06, "loss": 0.416, "step": 7994 }, { "epoch": 0.552713446249568, "grad_norm": 0.6251764890541215, "learning_rate": 8.491527542702085e-06, "loss": 0.3094, "step": 7995 }, { "epoch": 0.552782578638092, "grad_norm": 0.7235378272919372, "learning_rate": 8.49112684525958e-06, "loss": 0.2581, "step": 7996 }, { "epoch": 0.552851711026616, "grad_norm": 0.7135662865355409, "learning_rate": 8.490726104062032e-06, "loss": 0.4153, "step": 7997 }, { "epoch": 0.55292084341514, "grad_norm": 0.6618851816102213, "learning_rate": 8.490325319114461e-06, "loss": 0.3597, "step": 7998 }, { "epoch": 0.552989975803664, "grad_norm": 0.6563299830778855, "learning_rate": 8.48992449042189e-06, "loss": 0.2877, "step": 7999 }, { "epoch": 0.5530591081921881, "grad_norm": 1.0030272610381998, "learning_rate": 8.489523617989345e-06, "loss": 0.4033, "step": 8000 }, { "epoch": 0.5531282405807121, "grad_norm": 0.758423705934095, "learning_rate": 8.489122701821849e-06, "loss": 0.4883, "step": 8001 }, { "epoch": 0.5531973729692361, "grad_norm": 0.6906227768479357, "learning_rate": 8.488721741924424e-06, "loss": 0.2685, "step": 8002 }, { "epoch": 0.5532665053577601, "grad_norm": 0.7507888156049546, "learning_rate": 8.488320738302099e-06, "loss": 0.3162, "step": 8003 }, { "epoch": 0.5533356377462841, "grad_norm": 0.7078242503850886, "learning_rate": 8.4879196909599e-06, "loss": 0.4223, "step": 8004 }, { "epoch": 0.5534047701348082, "grad_norm": 0.6353355545420842, "learning_rate": 8.48751859990285e-06, "loss": 0.3323, "step": 8005 }, { "epoch": 0.5534739025233322, "grad_norm": 0.6145640983230728, "learning_rate": 8.487117465135982e-06, "loss": 0.2939, "step": 8006 }, { "epoch": 0.5535430349118562, "grad_norm": 0.691440715128347, "learning_rate": 8.486716286664316e-06, "loss": 0.3169, "step": 8007 }, { "epoch": 0.5536121673003802, "grad_norm": 0.7181445684342321, "learning_rate": 8.486315064492885e-06, "loss": 0.3379, "step": 8008 }, { "epoch": 0.5536812996889042, "grad_norm": 0.7822722868146943, "learning_rate": 8.485913798626715e-06, "loss": 0.282, "step": 8009 }, { "epoch": 0.5537504320774282, "grad_norm": 0.690721275773608, "learning_rate": 8.485512489070837e-06, "loss": 0.2776, "step": 8010 }, { "epoch": 0.5538195644659523, "grad_norm": 0.7100520924048458, "learning_rate": 8.485111135830281e-06, "loss": 0.3282, "step": 8011 }, { "epoch": 0.5538886968544763, "grad_norm": 0.7112986225946251, "learning_rate": 8.484709738910076e-06, "loss": 0.3896, "step": 8012 }, { "epoch": 0.5539578292430003, "grad_norm": 0.7107379921725074, "learning_rate": 8.484308298315253e-06, "loss": 0.3543, "step": 8013 }, { "epoch": 0.5540269616315243, "grad_norm": 0.7435114414043997, "learning_rate": 8.483906814050843e-06, "loss": 0.3019, "step": 8014 }, { "epoch": 0.5540960940200484, "grad_norm": 0.7095381101983962, "learning_rate": 8.48350528612188e-06, "loss": 0.3946, "step": 8015 }, { "epoch": 0.5541652264085725, "grad_norm": 0.6781307578965516, "learning_rate": 8.483103714533393e-06, "loss": 0.3053, "step": 8016 }, { "epoch": 0.5542343587970965, "grad_norm": 0.7280224023135061, "learning_rate": 8.482702099290418e-06, "loss": 0.2832, "step": 8017 }, { "epoch": 0.5543034911856205, "grad_norm": 0.7188083933176016, "learning_rate": 8.482300440397988e-06, "loss": 0.3573, "step": 8018 }, { "epoch": 0.5543726235741445, "grad_norm": 0.7281991375226479, "learning_rate": 8.481898737861134e-06, "loss": 0.3748, "step": 8019 }, { "epoch": 0.5544417559626685, "grad_norm": 0.755899320016248, "learning_rate": 8.481496991684895e-06, "loss": 0.2995, "step": 8020 }, { "epoch": 0.5545108883511926, "grad_norm": 0.6390573270200601, "learning_rate": 8.481095201874303e-06, "loss": 0.3132, "step": 8021 }, { "epoch": 0.5545800207397166, "grad_norm": 0.7048838106696218, "learning_rate": 8.480693368434396e-06, "loss": 0.3996, "step": 8022 }, { "epoch": 0.5546491531282406, "grad_norm": 0.8758076485450457, "learning_rate": 8.480291491370209e-06, "loss": 0.3974, "step": 8023 }, { "epoch": 0.5547182855167646, "grad_norm": 0.8194136197212538, "learning_rate": 8.47988957068678e-06, "loss": 0.3026, "step": 8024 }, { "epoch": 0.5547874179052886, "grad_norm": 0.8176671479468188, "learning_rate": 8.479487606389142e-06, "loss": 0.3458, "step": 8025 }, { "epoch": 0.5548565502938126, "grad_norm": 0.6473045418545729, "learning_rate": 8.479085598482339e-06, "loss": 0.3774, "step": 8026 }, { "epoch": 0.5549256826823367, "grad_norm": 1.0002414692371013, "learning_rate": 8.478683546971406e-06, "loss": 0.2898, "step": 8027 }, { "epoch": 0.5549948150708607, "grad_norm": 0.6677454146657543, "learning_rate": 8.478281451861384e-06, "loss": 0.309, "step": 8028 }, { "epoch": 0.5550639474593847, "grad_norm": 1.3990822874225584, "learning_rate": 8.47787931315731e-06, "loss": 0.3743, "step": 8029 }, { "epoch": 0.5551330798479087, "grad_norm": 0.7197330519642283, "learning_rate": 8.477477130864224e-06, "loss": 0.3851, "step": 8030 }, { "epoch": 0.5552022122364327, "grad_norm": 0.6530838775746867, "learning_rate": 8.47707490498717e-06, "loss": 0.2836, "step": 8031 }, { "epoch": 0.5552713446249568, "grad_norm": 0.6887755323569134, "learning_rate": 8.476672635531187e-06, "loss": 0.2993, "step": 8032 }, { "epoch": 0.5553404770134808, "grad_norm": 0.7055763305030986, "learning_rate": 8.476270322501317e-06, "loss": 0.4438, "step": 8033 }, { "epoch": 0.5554096094020048, "grad_norm": 0.6391339730091401, "learning_rate": 8.475867965902603e-06, "loss": 0.3007, "step": 8034 }, { "epoch": 0.5554787417905289, "grad_norm": 1.1767939383676866, "learning_rate": 8.475465565740085e-06, "loss": 0.3154, "step": 8035 }, { "epoch": 0.5555478741790529, "grad_norm": 0.6887498883357822, "learning_rate": 8.475063122018808e-06, "loss": 0.3479, "step": 8036 }, { "epoch": 0.555617006567577, "grad_norm": 0.8463974359463085, "learning_rate": 8.474660634743818e-06, "loss": 0.4256, "step": 8037 }, { "epoch": 0.555686138956101, "grad_norm": 0.6611253482753043, "learning_rate": 8.474258103920156e-06, "loss": 0.318, "step": 8038 }, { "epoch": 0.555755271344625, "grad_norm": 0.7179847600666198, "learning_rate": 8.473855529552871e-06, "loss": 0.3189, "step": 8039 }, { "epoch": 0.555824403733149, "grad_norm": 0.7545694874542783, "learning_rate": 8.473452911647006e-06, "loss": 0.4294, "step": 8040 }, { "epoch": 0.555893536121673, "grad_norm": 0.7138375796618325, "learning_rate": 8.473050250207607e-06, "loss": 0.3827, "step": 8041 }, { "epoch": 0.555962668510197, "grad_norm": 0.6548298780480486, "learning_rate": 8.47264754523972e-06, "loss": 0.3424, "step": 8042 }, { "epoch": 0.5560318008987211, "grad_norm": 0.6908681021258547, "learning_rate": 8.472244796748395e-06, "loss": 0.3171, "step": 8043 }, { "epoch": 0.5561009332872451, "grad_norm": 0.722387139388581, "learning_rate": 8.471842004738676e-06, "loss": 0.4149, "step": 8044 }, { "epoch": 0.5561700656757691, "grad_norm": 0.7413579952254364, "learning_rate": 8.471439169215614e-06, "loss": 0.3379, "step": 8045 }, { "epoch": 0.5562391980642931, "grad_norm": 0.6966255316707237, "learning_rate": 8.471036290184258e-06, "loss": 0.3157, "step": 8046 }, { "epoch": 0.5563083304528171, "grad_norm": 0.6716363664012963, "learning_rate": 8.470633367649655e-06, "loss": 0.4193, "step": 8047 }, { "epoch": 0.5563774628413412, "grad_norm": 0.6587004914400411, "learning_rate": 8.470230401616858e-06, "loss": 0.388, "step": 8048 }, { "epoch": 0.5564465952298652, "grad_norm": 0.8574801549827202, "learning_rate": 8.469827392090915e-06, "loss": 0.2754, "step": 8049 }, { "epoch": 0.5565157276183892, "grad_norm": 0.9939890895349822, "learning_rate": 8.469424339076877e-06, "loss": 0.4123, "step": 8050 }, { "epoch": 0.5565848600069132, "grad_norm": 0.7620892534663383, "learning_rate": 8.469021242579798e-06, "loss": 0.4152, "step": 8051 }, { "epoch": 0.5566539923954372, "grad_norm": 0.6353393128194618, "learning_rate": 8.468618102604726e-06, "loss": 0.3333, "step": 8052 }, { "epoch": 0.5567231247839612, "grad_norm": 0.939219649146775, "learning_rate": 8.468214919156718e-06, "loss": 0.2794, "step": 8053 }, { "epoch": 0.5567922571724853, "grad_norm": 0.7749588641853064, "learning_rate": 8.467811692240826e-06, "loss": 0.3988, "step": 8054 }, { "epoch": 0.5568613895610093, "grad_norm": 0.7164273842475092, "learning_rate": 8.467408421862103e-06, "loss": 0.3292, "step": 8055 }, { "epoch": 0.5569305219495334, "grad_norm": 0.7232897210962076, "learning_rate": 8.4670051080256e-06, "loss": 0.3076, "step": 8056 }, { "epoch": 0.5569996543380574, "grad_norm": 0.7128666335126039, "learning_rate": 8.46660175073638e-06, "loss": 0.3337, "step": 8057 }, { "epoch": 0.5570687867265814, "grad_norm": 0.7063906311840226, "learning_rate": 8.466198349999489e-06, "loss": 0.3858, "step": 8058 }, { "epoch": 0.5571379191151055, "grad_norm": 0.7411243178735146, "learning_rate": 8.465794905819991e-06, "loss": 0.3606, "step": 8059 }, { "epoch": 0.5572070515036295, "grad_norm": 0.6988555700402398, "learning_rate": 8.465391418202937e-06, "loss": 0.3334, "step": 8060 }, { "epoch": 0.5572761838921535, "grad_norm": 0.6795014335674049, "learning_rate": 8.464987887153385e-06, "loss": 0.378, "step": 8061 }, { "epoch": 0.5573453162806775, "grad_norm": 0.7746989878285991, "learning_rate": 8.464584312676393e-06, "loss": 0.4431, "step": 8062 }, { "epoch": 0.5574144486692015, "grad_norm": 0.8256240366216051, "learning_rate": 8.464180694777023e-06, "loss": 0.3071, "step": 8063 }, { "epoch": 0.5574835810577256, "grad_norm": 0.7479618860030428, "learning_rate": 8.463777033460328e-06, "loss": 0.3127, "step": 8064 }, { "epoch": 0.5575527134462496, "grad_norm": 0.7492943819032698, "learning_rate": 8.463373328731368e-06, "loss": 0.3673, "step": 8065 }, { "epoch": 0.5576218458347736, "grad_norm": 0.6793350050172637, "learning_rate": 8.462969580595207e-06, "loss": 0.3596, "step": 8066 }, { "epoch": 0.5576909782232976, "grad_norm": 0.6860390178545296, "learning_rate": 8.4625657890569e-06, "loss": 0.298, "step": 8067 }, { "epoch": 0.5577601106118216, "grad_norm": 0.7276754488325788, "learning_rate": 8.46216195412151e-06, "loss": 0.2969, "step": 8068 }, { "epoch": 0.5578292430003456, "grad_norm": 0.6645130295570117, "learning_rate": 8.4617580757941e-06, "loss": 0.4449, "step": 8069 }, { "epoch": 0.5578983753888697, "grad_norm": 0.7238646942670229, "learning_rate": 8.461354154079729e-06, "loss": 0.2738, "step": 8070 }, { "epoch": 0.5579675077773937, "grad_norm": 0.6928585941350376, "learning_rate": 8.460950188983461e-06, "loss": 0.337, "step": 8071 }, { "epoch": 0.5580366401659177, "grad_norm": 0.7132997610713754, "learning_rate": 8.460546180510358e-06, "loss": 0.416, "step": 8072 }, { "epoch": 0.5581057725544417, "grad_norm": 0.8267392361678729, "learning_rate": 8.460142128665484e-06, "loss": 0.3684, "step": 8073 }, { "epoch": 0.5581749049429657, "grad_norm": 0.6518042798720094, "learning_rate": 8.459738033453905e-06, "loss": 0.2762, "step": 8074 }, { "epoch": 0.5582440373314898, "grad_norm": 0.7625644984497463, "learning_rate": 8.459333894880684e-06, "loss": 0.3465, "step": 8075 }, { "epoch": 0.5583131697200139, "grad_norm": 0.7367975583766324, "learning_rate": 8.458929712950885e-06, "loss": 0.3888, "step": 8076 }, { "epoch": 0.5583823021085379, "grad_norm": 0.6622876689534971, "learning_rate": 8.458525487669576e-06, "loss": 0.2647, "step": 8077 }, { "epoch": 0.5584514344970619, "grad_norm": 0.7208752376255321, "learning_rate": 8.45812121904182e-06, "loss": 0.3315, "step": 8078 }, { "epoch": 0.5585205668855859, "grad_norm": 0.6855354424485235, "learning_rate": 8.457716907072686e-06, "loss": 0.358, "step": 8079 }, { "epoch": 0.55858969927411, "grad_norm": 0.6804829511379206, "learning_rate": 8.457312551767242e-06, "loss": 0.3931, "step": 8080 }, { "epoch": 0.558658831662634, "grad_norm": 0.6610972267546353, "learning_rate": 8.456908153130557e-06, "loss": 0.2931, "step": 8081 }, { "epoch": 0.558727964051158, "grad_norm": 0.8361542616018683, "learning_rate": 8.456503711167695e-06, "loss": 0.4151, "step": 8082 }, { "epoch": 0.558797096439682, "grad_norm": 0.7548979869697044, "learning_rate": 8.456099225883727e-06, "loss": 0.4048, "step": 8083 }, { "epoch": 0.558866228828206, "grad_norm": 0.7698841536180896, "learning_rate": 8.455694697283725e-06, "loss": 0.2927, "step": 8084 }, { "epoch": 0.55893536121673, "grad_norm": 0.7126681185977977, "learning_rate": 8.455290125372756e-06, "loss": 0.3199, "step": 8085 }, { "epoch": 0.5590044936052541, "grad_norm": 0.7603859134108752, "learning_rate": 8.45488551015589e-06, "loss": 0.3463, "step": 8086 }, { "epoch": 0.5590736259937781, "grad_norm": 0.8181002294450814, "learning_rate": 8.454480851638203e-06, "loss": 0.4188, "step": 8087 }, { "epoch": 0.5591427583823021, "grad_norm": 0.9448849183204427, "learning_rate": 8.45407614982476e-06, "loss": 0.2673, "step": 8088 }, { "epoch": 0.5592118907708261, "grad_norm": 0.7285587700142578, "learning_rate": 8.453671404720639e-06, "loss": 0.2725, "step": 8089 }, { "epoch": 0.5592810231593501, "grad_norm": 0.7456006694903358, "learning_rate": 8.453266616330908e-06, "loss": 0.4294, "step": 8090 }, { "epoch": 0.5593501555478742, "grad_norm": 0.7827444533578679, "learning_rate": 8.452861784660644e-06, "loss": 0.3412, "step": 8091 }, { "epoch": 0.5594192879363982, "grad_norm": 0.7694879842415145, "learning_rate": 8.452456909714919e-06, "loss": 0.2915, "step": 8092 }, { "epoch": 0.5594884203249222, "grad_norm": 0.7201128906105569, "learning_rate": 8.452051991498808e-06, "loss": 0.3445, "step": 8093 }, { "epoch": 0.5595575527134462, "grad_norm": 0.7521044319249073, "learning_rate": 8.451647030017386e-06, "loss": 0.4092, "step": 8094 }, { "epoch": 0.5596266851019702, "grad_norm": 0.6382305244777701, "learning_rate": 8.451242025275727e-06, "loss": 0.2939, "step": 8095 }, { "epoch": 0.5596958174904944, "grad_norm": 0.6738397496414581, "learning_rate": 8.45083697727891e-06, "loss": 0.3409, "step": 8096 }, { "epoch": 0.5597649498790184, "grad_norm": 0.8039149001794165, "learning_rate": 8.450431886032009e-06, "loss": 0.4101, "step": 8097 }, { "epoch": 0.5598340822675424, "grad_norm": 0.6498011441038182, "learning_rate": 8.450026751540101e-06, "loss": 0.372, "step": 8098 }, { "epoch": 0.5599032146560664, "grad_norm": 0.7371922975141345, "learning_rate": 8.449621573808266e-06, "loss": 0.3198, "step": 8099 }, { "epoch": 0.5599723470445904, "grad_norm": 0.7106946861135799, "learning_rate": 8.44921635284158e-06, "loss": 0.3113, "step": 8100 }, { "epoch": 0.5600414794331144, "grad_norm": 0.740942532598145, "learning_rate": 8.448811088645123e-06, "loss": 0.4009, "step": 8101 }, { "epoch": 0.5601106118216385, "grad_norm": 1.0468707655660905, "learning_rate": 8.448405781223972e-06, "loss": 0.3014, "step": 8102 }, { "epoch": 0.5601797442101625, "grad_norm": 0.66672392589239, "learning_rate": 8.44800043058321e-06, "loss": 0.2885, "step": 8103 }, { "epoch": 0.5602488765986865, "grad_norm": 0.7148937219679338, "learning_rate": 8.447595036727916e-06, "loss": 0.3127, "step": 8104 }, { "epoch": 0.5603180089872105, "grad_norm": 0.8026639774183305, "learning_rate": 8.44718959966317e-06, "loss": 0.3994, "step": 8105 }, { "epoch": 0.5603871413757345, "grad_norm": 0.716246818102857, "learning_rate": 8.446784119394054e-06, "loss": 0.2754, "step": 8106 }, { "epoch": 0.5604562737642586, "grad_norm": 0.8094439288600819, "learning_rate": 8.44637859592565e-06, "loss": 0.2849, "step": 8107 }, { "epoch": 0.5605254061527826, "grad_norm": 0.7276266615117587, "learning_rate": 8.445973029263041e-06, "loss": 0.496, "step": 8108 }, { "epoch": 0.5605945385413066, "grad_norm": 0.7400264721003604, "learning_rate": 8.44556741941131e-06, "loss": 0.3225, "step": 8109 }, { "epoch": 0.5606636709298306, "grad_norm": 0.7563850897114954, "learning_rate": 8.445161766375541e-06, "loss": 0.2745, "step": 8110 }, { "epoch": 0.5607328033183546, "grad_norm": 0.6790662962306641, "learning_rate": 8.444756070160815e-06, "loss": 0.3396, "step": 8111 }, { "epoch": 0.5608019357068786, "grad_norm": 0.7196326506022295, "learning_rate": 8.444350330772222e-06, "loss": 0.3703, "step": 8112 }, { "epoch": 0.5608710680954027, "grad_norm": 0.7233020609332101, "learning_rate": 8.443944548214842e-06, "loss": 0.2808, "step": 8113 }, { "epoch": 0.5609402004839267, "grad_norm": 0.7143273822903243, "learning_rate": 8.443538722493765e-06, "loss": 0.3638, "step": 8114 }, { "epoch": 0.5610093328724507, "grad_norm": 0.7481190997635913, "learning_rate": 8.443132853614073e-06, "loss": 0.4208, "step": 8115 }, { "epoch": 0.5610784652609747, "grad_norm": 0.6924081405144761, "learning_rate": 8.442726941580858e-06, "loss": 0.3467, "step": 8116 }, { "epoch": 0.5611475976494988, "grad_norm": 0.6793449895227509, "learning_rate": 8.442320986399203e-06, "loss": 0.3149, "step": 8117 }, { "epoch": 0.5612167300380229, "grad_norm": 0.7137902873055774, "learning_rate": 8.441914988074199e-06, "loss": 0.3498, "step": 8118 }, { "epoch": 0.5612858624265469, "grad_norm": 1.0447108806168393, "learning_rate": 8.44150894661093e-06, "loss": 0.4406, "step": 8119 }, { "epoch": 0.5613549948150709, "grad_norm": 0.6655347361676495, "learning_rate": 8.44110286201449e-06, "loss": 0.2944, "step": 8120 }, { "epoch": 0.5614241272035949, "grad_norm": 1.007693827461194, "learning_rate": 8.440696734289966e-06, "loss": 0.3095, "step": 8121 }, { "epoch": 0.5614932595921189, "grad_norm": 0.7425262385276217, "learning_rate": 8.44029056344245e-06, "loss": 0.4611, "step": 8122 }, { "epoch": 0.561562391980643, "grad_norm": 0.7282031093250061, "learning_rate": 8.43988434947703e-06, "loss": 0.312, "step": 8123 }, { "epoch": 0.561631524369167, "grad_norm": 0.7150298750779346, "learning_rate": 8.439478092398798e-06, "loss": 0.3517, "step": 8124 }, { "epoch": 0.561700656757691, "grad_norm": 0.6687389512912341, "learning_rate": 8.439071792212847e-06, "loss": 0.2619, "step": 8125 }, { "epoch": 0.561769789146215, "grad_norm": 0.7893187660506714, "learning_rate": 8.438665448924267e-06, "loss": 0.4152, "step": 8126 }, { "epoch": 0.561838921534739, "grad_norm": 0.7158161599424264, "learning_rate": 8.438259062538155e-06, "loss": 0.3417, "step": 8127 }, { "epoch": 0.561908053923263, "grad_norm": 0.7368879459121885, "learning_rate": 8.4378526330596e-06, "loss": 0.3078, "step": 8128 }, { "epoch": 0.5619771863117871, "grad_norm": 1.184569805595319, "learning_rate": 8.437446160493696e-06, "loss": 0.4014, "step": 8129 }, { "epoch": 0.5620463187003111, "grad_norm": 0.6766882260402299, "learning_rate": 8.43703964484554e-06, "loss": 0.3757, "step": 8130 }, { "epoch": 0.5621154510888351, "grad_norm": 0.7503523368772824, "learning_rate": 8.436633086120227e-06, "loss": 0.2797, "step": 8131 }, { "epoch": 0.5621845834773591, "grad_norm": 0.7898740960682117, "learning_rate": 8.43622648432285e-06, "loss": 0.3193, "step": 8132 }, { "epoch": 0.5622537158658831, "grad_norm": 0.7076815905580636, "learning_rate": 8.435819839458506e-06, "loss": 0.4259, "step": 8133 }, { "epoch": 0.5623228482544071, "grad_norm": 0.6545703937879256, "learning_rate": 8.43541315153229e-06, "loss": 0.2816, "step": 8134 }, { "epoch": 0.5623919806429312, "grad_norm": 1.799487094538748, "learning_rate": 8.435006420549303e-06, "loss": 0.332, "step": 8135 }, { "epoch": 0.5624611130314552, "grad_norm": 0.6306319018976759, "learning_rate": 8.43459964651464e-06, "loss": 0.3387, "step": 8136 }, { "epoch": 0.5625302454199793, "grad_norm": 0.6754875498756786, "learning_rate": 8.434192829433401e-06, "loss": 0.3562, "step": 8137 }, { "epoch": 0.5625993778085033, "grad_norm": 1.00980959832455, "learning_rate": 8.433785969310684e-06, "loss": 0.307, "step": 8138 }, { "epoch": 0.5626685101970273, "grad_norm": 0.867429858909522, "learning_rate": 8.433379066151586e-06, "loss": 0.2788, "step": 8139 }, { "epoch": 0.5627376425855514, "grad_norm": 0.6886710760277236, "learning_rate": 8.43297211996121e-06, "loss": 0.3695, "step": 8140 }, { "epoch": 0.5628067749740754, "grad_norm": 0.6452469504108452, "learning_rate": 8.432565130744652e-06, "loss": 0.2869, "step": 8141 }, { "epoch": 0.5628759073625994, "grad_norm": 0.9823854955994452, "learning_rate": 8.432158098507017e-06, "loss": 0.2546, "step": 8142 }, { "epoch": 0.5629450397511234, "grad_norm": 0.8242446334544118, "learning_rate": 8.431751023253408e-06, "loss": 0.3616, "step": 8143 }, { "epoch": 0.5630141721396474, "grad_norm": 0.7418118569480476, "learning_rate": 8.43134390498892e-06, "loss": 0.3871, "step": 8144 }, { "epoch": 0.5630833045281715, "grad_norm": 0.7028422297775659, "learning_rate": 8.430936743718663e-06, "loss": 0.2987, "step": 8145 }, { "epoch": 0.5631524369166955, "grad_norm": 0.7890032944679576, "learning_rate": 8.430529539447735e-06, "loss": 0.2987, "step": 8146 }, { "epoch": 0.5632215693052195, "grad_norm": 0.8216211352682662, "learning_rate": 8.430122292181241e-06, "loss": 0.453, "step": 8147 }, { "epoch": 0.5632907016937435, "grad_norm": 0.7261593537545904, "learning_rate": 8.429715001924285e-06, "loss": 0.3074, "step": 8148 }, { "epoch": 0.5633598340822675, "grad_norm": 0.7868929577587673, "learning_rate": 8.429307668681974e-06, "loss": 0.2877, "step": 8149 }, { "epoch": 0.5634289664707915, "grad_norm": 0.9285543681268654, "learning_rate": 8.428900292459411e-06, "loss": 0.4331, "step": 8150 }, { "epoch": 0.5634980988593156, "grad_norm": 0.8821509231833735, "learning_rate": 8.428492873261702e-06, "loss": 0.3675, "step": 8151 }, { "epoch": 0.5635672312478396, "grad_norm": 0.6884515890564386, "learning_rate": 8.428085411093952e-06, "loss": 0.3286, "step": 8152 }, { "epoch": 0.5636363636363636, "grad_norm": 0.8020997462838955, "learning_rate": 8.42767790596127e-06, "loss": 0.2751, "step": 8153 }, { "epoch": 0.5637054960248876, "grad_norm": 1.0453054869623069, "learning_rate": 8.427270357868762e-06, "loss": 0.3803, "step": 8154 }, { "epoch": 0.5637746284134116, "grad_norm": 0.722602296585563, "learning_rate": 8.426862766821536e-06, "loss": 0.3307, "step": 8155 }, { "epoch": 0.5638437608019357, "grad_norm": 0.6306336206361877, "learning_rate": 8.426455132824702e-06, "loss": 0.2453, "step": 8156 }, { "epoch": 0.5639128931904598, "grad_norm": 0.659178457413867, "learning_rate": 8.426047455883367e-06, "loss": 0.3395, "step": 8157 }, { "epoch": 0.5639820255789838, "grad_norm": 0.7966910827803271, "learning_rate": 8.42563973600264e-06, "loss": 0.3922, "step": 8158 }, { "epoch": 0.5640511579675078, "grad_norm": 0.7070771211506005, "learning_rate": 8.425231973187632e-06, "loss": 0.4225, "step": 8159 }, { "epoch": 0.5641202903560318, "grad_norm": 0.6040541649402782, "learning_rate": 8.424824167443457e-06, "loss": 0.2318, "step": 8160 }, { "epoch": 0.5641894227445559, "grad_norm": 0.687327866919724, "learning_rate": 8.42441631877522e-06, "loss": 0.3651, "step": 8161 }, { "epoch": 0.5642585551330799, "grad_norm": 0.7014493698870103, "learning_rate": 8.424008427188036e-06, "loss": 0.437, "step": 8162 }, { "epoch": 0.5643276875216039, "grad_norm": 0.7262190564699678, "learning_rate": 8.423600492687019e-06, "loss": 0.3396, "step": 8163 }, { "epoch": 0.5643968199101279, "grad_norm": 0.7379597465400669, "learning_rate": 8.423192515277276e-06, "loss": 0.2846, "step": 8164 }, { "epoch": 0.5644659522986519, "grad_norm": 0.7002169841899943, "learning_rate": 8.422784494963926e-06, "loss": 0.418, "step": 8165 }, { "epoch": 0.564535084687176, "grad_norm": 0.6742918865088743, "learning_rate": 8.42237643175208e-06, "loss": 0.3736, "step": 8166 }, { "epoch": 0.5646042170757, "grad_norm": 0.7050905338503541, "learning_rate": 8.421968325646854e-06, "loss": 0.2757, "step": 8167 }, { "epoch": 0.564673349464224, "grad_norm": 0.7194340283625711, "learning_rate": 8.421560176653361e-06, "loss": 0.3432, "step": 8168 }, { "epoch": 0.564742481852748, "grad_norm": 0.8684970478993131, "learning_rate": 8.421151984776716e-06, "loss": 0.4262, "step": 8169 }, { "epoch": 0.564811614241272, "grad_norm": 0.749941088159382, "learning_rate": 8.420743750022038e-06, "loss": 0.2892, "step": 8170 }, { "epoch": 0.564880746629796, "grad_norm": 0.7379436795416674, "learning_rate": 8.42033547239444e-06, "loss": 0.3217, "step": 8171 }, { "epoch": 0.5649498790183201, "grad_norm": 0.6648672430526966, "learning_rate": 8.419927151899043e-06, "loss": 0.3821, "step": 8172 }, { "epoch": 0.5650190114068441, "grad_norm": 0.7575862411741562, "learning_rate": 8.41951878854096e-06, "loss": 0.4074, "step": 8173 }, { "epoch": 0.5650881437953681, "grad_norm": 0.6543115617347611, "learning_rate": 8.419110382325313e-06, "loss": 0.2568, "step": 8174 }, { "epoch": 0.5651572761838921, "grad_norm": 0.7104927144615433, "learning_rate": 8.418701933257219e-06, "loss": 0.3307, "step": 8175 }, { "epoch": 0.5652264085724161, "grad_norm": 0.7052764894491736, "learning_rate": 8.418293441341795e-06, "loss": 0.4094, "step": 8176 }, { "epoch": 0.5652955409609401, "grad_norm": 0.7680467742306604, "learning_rate": 8.417884906584167e-06, "loss": 0.2758, "step": 8177 }, { "epoch": 0.5653646733494643, "grad_norm": 0.7013799493213129, "learning_rate": 8.417476328989448e-06, "loss": 0.343, "step": 8178 }, { "epoch": 0.5654338057379883, "grad_norm": 0.7455514568497376, "learning_rate": 8.417067708562763e-06, "loss": 0.3615, "step": 8179 }, { "epoch": 0.5655029381265123, "grad_norm": 0.7098990312762129, "learning_rate": 8.41665904530923e-06, "loss": 0.3982, "step": 8180 }, { "epoch": 0.5655720705150363, "grad_norm": 0.6856965652832046, "learning_rate": 8.416250339233977e-06, "loss": 0.316, "step": 8181 }, { "epoch": 0.5656412029035603, "grad_norm": 0.751219096434979, "learning_rate": 8.41584159034212e-06, "loss": 0.342, "step": 8182 }, { "epoch": 0.5657103352920844, "grad_norm": 0.7098218296554091, "learning_rate": 8.415432798638787e-06, "loss": 0.431, "step": 8183 }, { "epoch": 0.5657794676806084, "grad_norm": 0.7366564264835354, "learning_rate": 8.415023964129097e-06, "loss": 0.326, "step": 8184 }, { "epoch": 0.5658486000691324, "grad_norm": 0.6395931836921851, "learning_rate": 8.414615086818177e-06, "loss": 0.2933, "step": 8185 }, { "epoch": 0.5659177324576564, "grad_norm": 0.8839430665444967, "learning_rate": 8.414206166711149e-06, "loss": 0.3631, "step": 8186 }, { "epoch": 0.5659868648461804, "grad_norm": 0.679243372866465, "learning_rate": 8.41379720381314e-06, "loss": 0.4195, "step": 8187 }, { "epoch": 0.5660559972347045, "grad_norm": 0.7125198816304411, "learning_rate": 8.413388198129275e-06, "loss": 0.3169, "step": 8188 }, { "epoch": 0.5661251296232285, "grad_norm": 0.6875240127532223, "learning_rate": 8.412979149664682e-06, "loss": 0.3008, "step": 8189 }, { "epoch": 0.5661942620117525, "grad_norm": 0.7973236866702861, "learning_rate": 8.412570058424485e-06, "loss": 0.4049, "step": 8190 }, { "epoch": 0.5662633944002765, "grad_norm": 0.6888847657029781, "learning_rate": 8.412160924413813e-06, "loss": 0.3278, "step": 8191 }, { "epoch": 0.5663325267888005, "grad_norm": 0.6605757633425093, "learning_rate": 8.411751747637792e-06, "loss": 0.3284, "step": 8192 }, { "epoch": 0.5664016591773245, "grad_norm": 0.7086332686979916, "learning_rate": 8.411342528101551e-06, "loss": 0.4078, "step": 8193 }, { "epoch": 0.5664707915658486, "grad_norm": 0.7884440228318856, "learning_rate": 8.410933265810221e-06, "loss": 0.4145, "step": 8194 }, { "epoch": 0.5665399239543726, "grad_norm": 0.6042212381109021, "learning_rate": 8.41052396076893e-06, "loss": 0.2991, "step": 8195 }, { "epoch": 0.5666090563428966, "grad_norm": 0.7171788261462894, "learning_rate": 8.410114612982805e-06, "loss": 0.2455, "step": 8196 }, { "epoch": 0.5666781887314206, "grad_norm": 0.7133608740154158, "learning_rate": 8.40970522245698e-06, "loss": 0.4296, "step": 8197 }, { "epoch": 0.5667473211199447, "grad_norm": 0.8048570197470296, "learning_rate": 8.409295789196586e-06, "loss": 0.3098, "step": 8198 }, { "epoch": 0.5668164535084688, "grad_norm": 0.9210812116625418, "learning_rate": 8.408886313206753e-06, "loss": 0.3235, "step": 8199 }, { "epoch": 0.5668855858969928, "grad_norm": 0.7212200131126534, "learning_rate": 8.408476794492612e-06, "loss": 0.3579, "step": 8200 }, { "epoch": 0.5669547182855168, "grad_norm": 0.65234840049405, "learning_rate": 8.408067233059297e-06, "loss": 0.3805, "step": 8201 }, { "epoch": 0.5670238506740408, "grad_norm": 0.7087610299867276, "learning_rate": 8.407657628911944e-06, "loss": 0.2499, "step": 8202 }, { "epoch": 0.5670929830625648, "grad_norm": 0.6865653129131277, "learning_rate": 8.407247982055683e-06, "loss": 0.3188, "step": 8203 }, { "epoch": 0.5671621154510889, "grad_norm": 0.6891396913590859, "learning_rate": 8.406838292495648e-06, "loss": 0.3607, "step": 8204 }, { "epoch": 0.5672312478396129, "grad_norm": 1.62642256575488, "learning_rate": 8.406428560236976e-06, "loss": 0.458, "step": 8205 }, { "epoch": 0.5673003802281369, "grad_norm": 0.6980884347381672, "learning_rate": 8.406018785284799e-06, "loss": 0.3388, "step": 8206 }, { "epoch": 0.5673695126166609, "grad_norm": 0.7441730243323756, "learning_rate": 8.405608967644257e-06, "loss": 0.344, "step": 8207 }, { "epoch": 0.5674386450051849, "grad_norm": 0.7061236945156514, "learning_rate": 8.405199107320483e-06, "loss": 0.384, "step": 8208 }, { "epoch": 0.567507777393709, "grad_norm": 0.6239568923268433, "learning_rate": 8.404789204318615e-06, "loss": 0.3214, "step": 8209 }, { "epoch": 0.567576909782233, "grad_norm": 0.7773095796527517, "learning_rate": 8.404379258643792e-06, "loss": 0.2945, "step": 8210 }, { "epoch": 0.567646042170757, "grad_norm": 0.69693831506534, "learning_rate": 8.403969270301147e-06, "loss": 0.3772, "step": 8211 }, { "epoch": 0.567715174559281, "grad_norm": 0.7471457186507144, "learning_rate": 8.403559239295825e-06, "loss": 0.4022, "step": 8212 }, { "epoch": 0.567784306947805, "grad_norm": 0.6552365013909581, "learning_rate": 8.40314916563296e-06, "loss": 0.2866, "step": 8213 }, { "epoch": 0.567853439336329, "grad_norm": 0.6867368505527427, "learning_rate": 8.402739049317694e-06, "loss": 0.3273, "step": 8214 }, { "epoch": 0.567922571724853, "grad_norm": 0.6588687371864601, "learning_rate": 8.402328890355166e-06, "loss": 0.4036, "step": 8215 }, { "epoch": 0.5679917041133771, "grad_norm": 1.0399333546186524, "learning_rate": 8.401918688750519e-06, "loss": 0.3884, "step": 8216 }, { "epoch": 0.5680608365019011, "grad_norm": 0.6100703906169883, "learning_rate": 8.401508444508892e-06, "loss": 0.3025, "step": 8217 }, { "epoch": 0.5681299688904252, "grad_norm": 0.7093144985974339, "learning_rate": 8.401098157635426e-06, "loss": 0.3756, "step": 8218 }, { "epoch": 0.5681991012789492, "grad_norm": 0.660494730337541, "learning_rate": 8.400687828135263e-06, "loss": 0.4243, "step": 8219 }, { "epoch": 0.5682682336674733, "grad_norm": 0.6839285374445375, "learning_rate": 8.400277456013549e-06, "loss": 0.322, "step": 8220 }, { "epoch": 0.5683373660559973, "grad_norm": 0.6831841259444483, "learning_rate": 8.399867041275423e-06, "loss": 0.3041, "step": 8221 }, { "epoch": 0.5684064984445213, "grad_norm": 0.8373679221778689, "learning_rate": 8.399456583926033e-06, "loss": 0.4647, "step": 8222 }, { "epoch": 0.5684756308330453, "grad_norm": 0.6734122538479854, "learning_rate": 8.39904608397052e-06, "loss": 0.3522, "step": 8223 }, { "epoch": 0.5685447632215693, "grad_norm": 0.6483874057188067, "learning_rate": 8.39863554141403e-06, "loss": 0.3112, "step": 8224 }, { "epoch": 0.5686138956100933, "grad_norm": 0.7113263384046319, "learning_rate": 8.39822495626171e-06, "loss": 0.3047, "step": 8225 }, { "epoch": 0.5686830279986174, "grad_norm": 0.7586457472588127, "learning_rate": 8.397814328518702e-06, "loss": 0.4696, "step": 8226 }, { "epoch": 0.5687521603871414, "grad_norm": 0.5851518210062117, "learning_rate": 8.397403658190157e-06, "loss": 0.2722, "step": 8227 }, { "epoch": 0.5688212927756654, "grad_norm": 0.5909985097571777, "learning_rate": 8.396992945281221e-06, "loss": 0.2945, "step": 8228 }, { "epoch": 0.5688904251641894, "grad_norm": 0.7426300731376776, "learning_rate": 8.396582189797038e-06, "loss": 0.4841, "step": 8229 }, { "epoch": 0.5689595575527134, "grad_norm": 0.6849111460953421, "learning_rate": 8.396171391742762e-06, "loss": 0.3236, "step": 8230 }, { "epoch": 0.5690286899412375, "grad_norm": 0.7623518419329421, "learning_rate": 8.395760551123535e-06, "loss": 0.3207, "step": 8231 }, { "epoch": 0.5690978223297615, "grad_norm": 1.2661881941755386, "learning_rate": 8.395349667944512e-06, "loss": 0.3833, "step": 8232 }, { "epoch": 0.5691669547182855, "grad_norm": 0.6665862473004454, "learning_rate": 8.394938742210838e-06, "loss": 0.4109, "step": 8233 }, { "epoch": 0.5692360871068095, "grad_norm": 0.6740538495024934, "learning_rate": 8.394527773927668e-06, "loss": 0.3546, "step": 8234 }, { "epoch": 0.5693052194953335, "grad_norm": 0.6033925701507998, "learning_rate": 8.394116763100148e-06, "loss": 0.2927, "step": 8235 }, { "epoch": 0.5693743518838575, "grad_norm": 0.7102175294779122, "learning_rate": 8.393705709733432e-06, "loss": 0.4222, "step": 8236 }, { "epoch": 0.5694434842723816, "grad_norm": 0.6256005317883793, "learning_rate": 8.393294613832673e-06, "loss": 0.3588, "step": 8237 }, { "epoch": 0.5695126166609056, "grad_norm": 0.6698575505638148, "learning_rate": 8.392883475403019e-06, "loss": 0.29, "step": 8238 }, { "epoch": 0.5695817490494297, "grad_norm": 0.7410140037878858, "learning_rate": 8.392472294449626e-06, "loss": 0.3723, "step": 8239 }, { "epoch": 0.5696508814379537, "grad_norm": 0.6456789116170438, "learning_rate": 8.392061070977648e-06, "loss": 0.3485, "step": 8240 }, { "epoch": 0.5697200138264777, "grad_norm": 0.640344286664746, "learning_rate": 8.391649804992239e-06, "loss": 0.3612, "step": 8241 }, { "epoch": 0.5697891462150018, "grad_norm": 0.6765347488434271, "learning_rate": 8.391238496498551e-06, "loss": 0.2891, "step": 8242 }, { "epoch": 0.5698582786035258, "grad_norm": 0.7204427009457831, "learning_rate": 8.390827145501739e-06, "loss": 0.3645, "step": 8243 }, { "epoch": 0.5699274109920498, "grad_norm": 0.7644472967322544, "learning_rate": 8.390415752006961e-06, "loss": 0.4029, "step": 8244 }, { "epoch": 0.5699965433805738, "grad_norm": 0.6617864029576205, "learning_rate": 8.390004316019373e-06, "loss": 0.309, "step": 8245 }, { "epoch": 0.5700656757690978, "grad_norm": 0.6837957189753778, "learning_rate": 8.38959283754413e-06, "loss": 0.3567, "step": 8246 }, { "epoch": 0.5701348081576219, "grad_norm": 0.6936026578548542, "learning_rate": 8.38918131658639e-06, "loss": 0.3993, "step": 8247 }, { "epoch": 0.5702039405461459, "grad_norm": 0.7207496452311587, "learning_rate": 8.388769753151309e-06, "loss": 0.3626, "step": 8248 }, { "epoch": 0.5702730729346699, "grad_norm": 0.6985338310216725, "learning_rate": 8.38835814724405e-06, "loss": 0.3523, "step": 8249 }, { "epoch": 0.5703422053231939, "grad_norm": 0.693971942076092, "learning_rate": 8.387946498869765e-06, "loss": 0.3191, "step": 8250 }, { "epoch": 0.5704113377117179, "grad_norm": 0.6745173799246258, "learning_rate": 8.387534808033618e-06, "loss": 0.3994, "step": 8251 }, { "epoch": 0.5704804701002419, "grad_norm": 0.674010655021362, "learning_rate": 8.387123074740767e-06, "loss": 0.3052, "step": 8252 }, { "epoch": 0.570549602488766, "grad_norm": 0.9574831649249985, "learning_rate": 8.386711298996373e-06, "loss": 0.3228, "step": 8253 }, { "epoch": 0.57061873487729, "grad_norm": 0.7335600233242179, "learning_rate": 8.386299480805598e-06, "loss": 0.3772, "step": 8254 }, { "epoch": 0.570687867265814, "grad_norm": 0.7904274500687334, "learning_rate": 8.3858876201736e-06, "loss": 0.3945, "step": 8255 }, { "epoch": 0.570756999654338, "grad_norm": 0.717051403741529, "learning_rate": 8.385475717105545e-06, "loss": 0.312, "step": 8256 }, { "epoch": 0.570826132042862, "grad_norm": 0.6964845468945532, "learning_rate": 8.385063771606592e-06, "loss": 0.2956, "step": 8257 }, { "epoch": 0.570895264431386, "grad_norm": 0.7908876683134533, "learning_rate": 8.384651783681905e-06, "loss": 0.4788, "step": 8258 }, { "epoch": 0.5709643968199102, "grad_norm": 0.9893069524579037, "learning_rate": 8.38423975333665e-06, "loss": 0.3442, "step": 8259 }, { "epoch": 0.5710335292084342, "grad_norm": 0.765954402166931, "learning_rate": 8.383827680575989e-06, "loss": 0.2968, "step": 8260 }, { "epoch": 0.5711026615969582, "grad_norm": 0.6909713223682137, "learning_rate": 8.383415565405086e-06, "loss": 0.3226, "step": 8261 }, { "epoch": 0.5711717939854822, "grad_norm": 0.714493618459702, "learning_rate": 8.383003407829107e-06, "loss": 0.4034, "step": 8262 }, { "epoch": 0.5712409263740063, "grad_norm": 0.6872573526761929, "learning_rate": 8.382591207853217e-06, "loss": 0.3277, "step": 8263 }, { "epoch": 0.5713100587625303, "grad_norm": 0.6900723121459874, "learning_rate": 8.382178965482584e-06, "loss": 0.3198, "step": 8264 }, { "epoch": 0.5713791911510543, "grad_norm": 0.6739251120412973, "learning_rate": 8.381766680722372e-06, "loss": 0.4148, "step": 8265 }, { "epoch": 0.5714483235395783, "grad_norm": 0.6893482642958385, "learning_rate": 8.38135435357775e-06, "loss": 0.3793, "step": 8266 }, { "epoch": 0.5715174559281023, "grad_norm": 0.6819491114557195, "learning_rate": 8.380941984053888e-06, "loss": 0.3046, "step": 8267 }, { "epoch": 0.5715865883166263, "grad_norm": 0.6737863250922238, "learning_rate": 8.38052957215595e-06, "loss": 0.3377, "step": 8268 }, { "epoch": 0.5716557207051504, "grad_norm": 0.6807591701105777, "learning_rate": 8.380117117889103e-06, "loss": 0.3756, "step": 8269 }, { "epoch": 0.5717248530936744, "grad_norm": 0.7911775433908769, "learning_rate": 8.379704621258524e-06, "loss": 0.3269, "step": 8270 }, { "epoch": 0.5717939854821984, "grad_norm": 0.7580063947632348, "learning_rate": 8.379292082269379e-06, "loss": 0.3894, "step": 8271 }, { "epoch": 0.5718631178707224, "grad_norm": 0.7254305957318016, "learning_rate": 8.378879500926838e-06, "loss": 0.3974, "step": 8272 }, { "epoch": 0.5719322502592464, "grad_norm": 0.7181227108980511, "learning_rate": 8.378466877236072e-06, "loss": 0.3165, "step": 8273 }, { "epoch": 0.5720013826477705, "grad_norm": 0.7313199145970175, "learning_rate": 8.378054211202254e-06, "loss": 0.3026, "step": 8274 }, { "epoch": 0.5720705150362945, "grad_norm": 0.7990551910988458, "learning_rate": 8.377641502830553e-06, "loss": 0.35, "step": 8275 }, { "epoch": 0.5721396474248185, "grad_norm": 0.7084431910534921, "learning_rate": 8.377228752126143e-06, "loss": 0.4141, "step": 8276 }, { "epoch": 0.5722087798133425, "grad_norm": 0.6640858110246561, "learning_rate": 8.376815959094199e-06, "loss": 0.3514, "step": 8277 }, { "epoch": 0.5722779122018665, "grad_norm": 0.6992729714256947, "learning_rate": 8.376403123739892e-06, "loss": 0.2845, "step": 8278 }, { "epoch": 0.5723470445903907, "grad_norm": 0.7032153495556754, "learning_rate": 8.375990246068399e-06, "loss": 0.4076, "step": 8279 }, { "epoch": 0.5724161769789147, "grad_norm": 0.7861260628843506, "learning_rate": 8.375577326084891e-06, "loss": 0.375, "step": 8280 }, { "epoch": 0.5724853093674387, "grad_norm": 0.6755894135073237, "learning_rate": 8.375164363794548e-06, "loss": 0.3078, "step": 8281 }, { "epoch": 0.5725544417559627, "grad_norm": 0.6977061374063745, "learning_rate": 8.37475135920254e-06, "loss": 0.3053, "step": 8282 }, { "epoch": 0.5726235741444867, "grad_norm": 0.7121961331450508, "learning_rate": 8.374338312314047e-06, "loss": 0.4443, "step": 8283 }, { "epoch": 0.5726927065330107, "grad_norm": 1.0133456963212433, "learning_rate": 8.373925223134246e-06, "loss": 0.2596, "step": 8284 }, { "epoch": 0.5727618389215348, "grad_norm": 0.6859257014641823, "learning_rate": 8.373512091668314e-06, "loss": 0.3279, "step": 8285 }, { "epoch": 0.5728309713100588, "grad_norm": 0.6594613942347972, "learning_rate": 8.373098917921427e-06, "loss": 0.4212, "step": 8286 }, { "epoch": 0.5729001036985828, "grad_norm": 0.6451330735577059, "learning_rate": 8.372685701898764e-06, "loss": 0.3343, "step": 8287 }, { "epoch": 0.5729692360871068, "grad_norm": 1.0659230858808506, "learning_rate": 8.372272443605506e-06, "loss": 0.2602, "step": 8288 }, { "epoch": 0.5730383684756308, "grad_norm": 0.9039960959742147, "learning_rate": 8.37185914304683e-06, "loss": 0.3135, "step": 8289 }, { "epoch": 0.5731075008641549, "grad_norm": 0.7176875377325842, "learning_rate": 8.371445800227918e-06, "loss": 0.3688, "step": 8290 }, { "epoch": 0.5731766332526789, "grad_norm": 0.6772650558841328, "learning_rate": 8.371032415153949e-06, "loss": 0.3518, "step": 8291 }, { "epoch": 0.5732457656412029, "grad_norm": 0.7030011915342709, "learning_rate": 8.370618987830104e-06, "loss": 0.2916, "step": 8292 }, { "epoch": 0.5733148980297269, "grad_norm": 0.7202569962904896, "learning_rate": 8.370205518261564e-06, "loss": 0.3391, "step": 8293 }, { "epoch": 0.5733840304182509, "grad_norm": 0.6738183839574622, "learning_rate": 8.369792006453514e-06, "loss": 0.3903, "step": 8294 }, { "epoch": 0.5734531628067749, "grad_norm": 0.6123426457217018, "learning_rate": 8.369378452411135e-06, "loss": 0.3302, "step": 8295 }, { "epoch": 0.573522295195299, "grad_norm": 0.7507025588060974, "learning_rate": 8.368964856139609e-06, "loss": 0.2581, "step": 8296 }, { "epoch": 0.573591427583823, "grad_norm": 0.7577806637612411, "learning_rate": 8.368551217644124e-06, "loss": 0.4232, "step": 8297 }, { "epoch": 0.573660559972347, "grad_norm": 0.6648898670842697, "learning_rate": 8.368137536929857e-06, "loss": 0.3188, "step": 8298 }, { "epoch": 0.573729692360871, "grad_norm": 0.6859560421219377, "learning_rate": 8.367723814001997e-06, "loss": 0.2578, "step": 8299 }, { "epoch": 0.5737988247493951, "grad_norm": 0.6891209011842849, "learning_rate": 8.367310048865732e-06, "loss": 0.3062, "step": 8300 }, { "epoch": 0.5738679571379192, "grad_norm": 0.7215634489452588, "learning_rate": 8.366896241526241e-06, "loss": 0.4736, "step": 8301 }, { "epoch": 0.5739370895264432, "grad_norm": 0.7561595956876697, "learning_rate": 8.366482391988716e-06, "loss": 0.4018, "step": 8302 }, { "epoch": 0.5740062219149672, "grad_norm": 0.9498976712135223, "learning_rate": 8.366068500258342e-06, "loss": 0.3197, "step": 8303 }, { "epoch": 0.5740753543034912, "grad_norm": 0.800687485081932, "learning_rate": 8.365654566340306e-06, "loss": 0.3859, "step": 8304 }, { "epoch": 0.5741444866920152, "grad_norm": 0.7036075169192375, "learning_rate": 8.365240590239797e-06, "loss": 0.3333, "step": 8305 }, { "epoch": 0.5742136190805393, "grad_norm": 0.6945999409036046, "learning_rate": 8.364826571962e-06, "loss": 0.268, "step": 8306 }, { "epoch": 0.5742827514690633, "grad_norm": 0.7636934634725957, "learning_rate": 8.36441251151211e-06, "loss": 0.3089, "step": 8307 }, { "epoch": 0.5743518838575873, "grad_norm": 1.125510680713598, "learning_rate": 8.363998408895312e-06, "loss": 0.3909, "step": 8308 }, { "epoch": 0.5744210162461113, "grad_norm": 0.7694988276555783, "learning_rate": 8.363584264116798e-06, "loss": 0.3086, "step": 8309 }, { "epoch": 0.5744901486346353, "grad_norm": 0.6767501034893546, "learning_rate": 8.363170077181757e-06, "loss": 0.2625, "step": 8310 }, { "epoch": 0.5745592810231593, "grad_norm": 0.7415299550383226, "learning_rate": 8.362755848095382e-06, "loss": 0.3547, "step": 8311 }, { "epoch": 0.5746284134116834, "grad_norm": 1.0958589236437712, "learning_rate": 8.362341576862863e-06, "loss": 0.4221, "step": 8312 }, { "epoch": 0.5746975458002074, "grad_norm": 0.7393522661722727, "learning_rate": 8.361927263489391e-06, "loss": 0.2744, "step": 8313 }, { "epoch": 0.5747666781887314, "grad_norm": 1.2189762157701314, "learning_rate": 8.361512907980163e-06, "loss": 0.2863, "step": 8314 }, { "epoch": 0.5748358105772554, "grad_norm": 1.3997520829251004, "learning_rate": 8.361098510340366e-06, "loss": 0.3808, "step": 8315 }, { "epoch": 0.5749049429657794, "grad_norm": 0.7196520263710671, "learning_rate": 8.360684070575201e-06, "loss": 0.3735, "step": 8316 }, { "epoch": 0.5749740753543034, "grad_norm": 0.6047741114398113, "learning_rate": 8.360269588689857e-06, "loss": 0.2734, "step": 8317 }, { "epoch": 0.5750432077428275, "grad_norm": 0.6634595949114526, "learning_rate": 8.35985506468953e-06, "loss": 0.3383, "step": 8318 }, { "epoch": 0.5751123401313515, "grad_norm": 1.787898404579677, "learning_rate": 8.359440498579416e-06, "loss": 0.4206, "step": 8319 }, { "epoch": 0.5751814725198756, "grad_norm": 0.664816947717923, "learning_rate": 8.359025890364713e-06, "loss": 0.3048, "step": 8320 }, { "epoch": 0.5752506049083996, "grad_norm": 0.8005973561401677, "learning_rate": 8.358611240050611e-06, "loss": 0.2836, "step": 8321 }, { "epoch": 0.5753197372969237, "grad_norm": 0.7393911783249267, "learning_rate": 8.358196547642313e-06, "loss": 0.4416, "step": 8322 }, { "epoch": 0.5753888696854477, "grad_norm": 1.066159256536026, "learning_rate": 8.357781813145014e-06, "loss": 0.3856, "step": 8323 }, { "epoch": 0.5754580020739717, "grad_norm": 0.6898085503946025, "learning_rate": 8.357367036563912e-06, "loss": 0.3044, "step": 8324 }, { "epoch": 0.5755271344624957, "grad_norm": 0.6888687966756142, "learning_rate": 8.356952217904206e-06, "loss": 0.3046, "step": 8325 }, { "epoch": 0.5755962668510197, "grad_norm": 0.6953591523591462, "learning_rate": 8.356537357171095e-06, "loss": 0.4154, "step": 8326 }, { "epoch": 0.5756653992395437, "grad_norm": 0.6643251254588661, "learning_rate": 8.356122454369777e-06, "loss": 0.3215, "step": 8327 }, { "epoch": 0.5757345316280678, "grad_norm": 0.6945174211242745, "learning_rate": 8.355707509505454e-06, "loss": 0.2786, "step": 8328 }, { "epoch": 0.5758036640165918, "grad_norm": 0.7887165128335345, "learning_rate": 8.355292522583327e-06, "loss": 0.3643, "step": 8329 }, { "epoch": 0.5758727964051158, "grad_norm": 0.6906450649550466, "learning_rate": 8.354877493608595e-06, "loss": 0.3978, "step": 8330 }, { "epoch": 0.5759419287936398, "grad_norm": 0.7638956631326601, "learning_rate": 8.35446242258646e-06, "loss": 0.2574, "step": 8331 }, { "epoch": 0.5760110611821638, "grad_norm": 0.6832972156140341, "learning_rate": 8.354047309522126e-06, "loss": 0.3607, "step": 8332 }, { "epoch": 0.5760801935706878, "grad_norm": 0.7220441501884701, "learning_rate": 8.353632154420793e-06, "loss": 0.4145, "step": 8333 }, { "epoch": 0.5761493259592119, "grad_norm": 0.7627426016134451, "learning_rate": 8.353216957287667e-06, "loss": 0.2917, "step": 8334 }, { "epoch": 0.5762184583477359, "grad_norm": 0.7688129373603307, "learning_rate": 8.35280171812795e-06, "loss": 0.2931, "step": 8335 }, { "epoch": 0.5762875907362599, "grad_norm": 0.7870569142464833, "learning_rate": 8.352386436946847e-06, "loss": 0.3499, "step": 8336 }, { "epoch": 0.5763567231247839, "grad_norm": 0.7649872433368551, "learning_rate": 8.351971113749564e-06, "loss": 0.4327, "step": 8337 }, { "epoch": 0.5764258555133079, "grad_norm": 0.633429901328398, "learning_rate": 8.351555748541303e-06, "loss": 0.2986, "step": 8338 }, { "epoch": 0.576494987901832, "grad_norm": 0.759214498998838, "learning_rate": 8.351140341327272e-06, "loss": 0.373, "step": 8339 }, { "epoch": 0.5765641202903561, "grad_norm": 0.7055426873813224, "learning_rate": 8.350724892112678e-06, "loss": 0.4248, "step": 8340 }, { "epoch": 0.5766332526788801, "grad_norm": 0.9084144315754238, "learning_rate": 8.350309400902727e-06, "loss": 0.3595, "step": 8341 }, { "epoch": 0.5767023850674041, "grad_norm": 0.7028379543611136, "learning_rate": 8.349893867702625e-06, "loss": 0.277, "step": 8342 }, { "epoch": 0.5767715174559281, "grad_norm": 0.6791430517824191, "learning_rate": 8.349478292517584e-06, "loss": 0.4148, "step": 8343 }, { "epoch": 0.5768406498444522, "grad_norm": 0.7226572136042868, "learning_rate": 8.349062675352808e-06, "loss": 0.419, "step": 8344 }, { "epoch": 0.5769097822329762, "grad_norm": 0.7041462191903396, "learning_rate": 8.348647016213508e-06, "loss": 0.3474, "step": 8345 }, { "epoch": 0.5769789146215002, "grad_norm": 0.6751761843404827, "learning_rate": 8.348231315104895e-06, "loss": 0.2855, "step": 8346 }, { "epoch": 0.5770480470100242, "grad_norm": 0.7083994592409394, "learning_rate": 8.347815572032179e-06, "loss": 0.4083, "step": 8347 }, { "epoch": 0.5771171793985482, "grad_norm": 0.7466666654235039, "learning_rate": 8.347399787000566e-06, "loss": 0.3531, "step": 8348 }, { "epoch": 0.5771863117870722, "grad_norm": 0.6383442178354286, "learning_rate": 8.346983960015272e-06, "loss": 0.3224, "step": 8349 }, { "epoch": 0.5772554441755963, "grad_norm": 0.690193729206602, "learning_rate": 8.346568091081507e-06, "loss": 0.4133, "step": 8350 }, { "epoch": 0.5773245765641203, "grad_norm": 0.6464574310925385, "learning_rate": 8.346152180204483e-06, "loss": 0.3935, "step": 8351 }, { "epoch": 0.5773937089526443, "grad_norm": 0.6389377939269365, "learning_rate": 8.345736227389412e-06, "loss": 0.3056, "step": 8352 }, { "epoch": 0.5774628413411683, "grad_norm": 0.6794597814358859, "learning_rate": 8.34532023264151e-06, "loss": 0.3926, "step": 8353 }, { "epoch": 0.5775319737296923, "grad_norm": 0.7691657807650485, "learning_rate": 8.344904195965988e-06, "loss": 0.3613, "step": 8354 }, { "epoch": 0.5776011061182164, "grad_norm": 0.6977564178616741, "learning_rate": 8.344488117368062e-06, "loss": 0.4046, "step": 8355 }, { "epoch": 0.5776702385067404, "grad_norm": 0.7572044346992869, "learning_rate": 8.344071996852944e-06, "loss": 0.328, "step": 8356 }, { "epoch": 0.5777393708952644, "grad_norm": 0.7577052829549517, "learning_rate": 8.343655834425853e-06, "loss": 0.3492, "step": 8357 }, { "epoch": 0.5778085032837884, "grad_norm": 0.7093150479136192, "learning_rate": 8.343239630092002e-06, "loss": 0.4232, "step": 8358 }, { "epoch": 0.5778776356723124, "grad_norm": 0.6781594827924485, "learning_rate": 8.34282338385661e-06, "loss": 0.3165, "step": 8359 }, { "epoch": 0.5779467680608364, "grad_norm": 0.795202994802811, "learning_rate": 8.34240709572489e-06, "loss": 0.3152, "step": 8360 }, { "epoch": 0.5780159004493606, "grad_norm": 0.6801360743824016, "learning_rate": 8.341990765702063e-06, "loss": 0.372, "step": 8361 }, { "epoch": 0.5780850328378846, "grad_norm": 0.6307270620155788, "learning_rate": 8.341574393793346e-06, "loss": 0.3512, "step": 8362 }, { "epoch": 0.5781541652264086, "grad_norm": 1.0827468157496698, "learning_rate": 8.341157980003959e-06, "loss": 0.3328, "step": 8363 }, { "epoch": 0.5782232976149326, "grad_norm": 0.7699145932483727, "learning_rate": 8.340741524339115e-06, "loss": 0.2748, "step": 8364 }, { "epoch": 0.5782924300034566, "grad_norm": 0.8425230383030174, "learning_rate": 8.340325026804041e-06, "loss": 0.4582, "step": 8365 }, { "epoch": 0.5783615623919807, "grad_norm": 0.7368363338324047, "learning_rate": 8.339908487403954e-06, "loss": 0.3356, "step": 8366 }, { "epoch": 0.5784306947805047, "grad_norm": 0.8130193434823256, "learning_rate": 8.339491906144072e-06, "loss": 0.3252, "step": 8367 }, { "epoch": 0.5784998271690287, "grad_norm": 0.709061559374945, "learning_rate": 8.339075283029621e-06, "loss": 0.3093, "step": 8368 }, { "epoch": 0.5785689595575527, "grad_norm": 0.7883817099106215, "learning_rate": 8.33865861806582e-06, "loss": 0.3725, "step": 8369 }, { "epoch": 0.5786380919460767, "grad_norm": 0.6428633240846053, "learning_rate": 8.338241911257889e-06, "loss": 0.3285, "step": 8370 }, { "epoch": 0.5787072243346008, "grad_norm": 0.6880093978491155, "learning_rate": 8.337825162611056e-06, "loss": 0.2917, "step": 8371 }, { "epoch": 0.5787763567231248, "grad_norm": 0.7071836009338603, "learning_rate": 8.33740837213054e-06, "loss": 0.3992, "step": 8372 }, { "epoch": 0.5788454891116488, "grad_norm": 0.6855526697790235, "learning_rate": 8.336991539821567e-06, "loss": 0.2829, "step": 8373 }, { "epoch": 0.5789146215001728, "grad_norm": 0.691829647639169, "learning_rate": 8.336574665689358e-06, "loss": 0.3432, "step": 8374 }, { "epoch": 0.5789837538886968, "grad_norm": 0.7502771865563048, "learning_rate": 8.33615774973914e-06, "loss": 0.327, "step": 8375 }, { "epoch": 0.5790528862772208, "grad_norm": 0.7368756661472934, "learning_rate": 8.33574079197614e-06, "loss": 0.4596, "step": 8376 }, { "epoch": 0.5791220186657449, "grad_norm": 0.8221209287799193, "learning_rate": 8.335323792405581e-06, "loss": 0.3337, "step": 8377 }, { "epoch": 0.5791911510542689, "grad_norm": 0.6651697863628174, "learning_rate": 8.334906751032691e-06, "loss": 0.2936, "step": 8378 }, { "epoch": 0.5792602834427929, "grad_norm": 0.7071280848627223, "learning_rate": 8.334489667862695e-06, "loss": 0.3917, "step": 8379 }, { "epoch": 0.5793294158313169, "grad_norm": 0.7668931705471903, "learning_rate": 8.334072542900825e-06, "loss": 0.4015, "step": 8380 }, { "epoch": 0.579398548219841, "grad_norm": 0.7024433502460877, "learning_rate": 8.3336553761523e-06, "loss": 0.3725, "step": 8381 }, { "epoch": 0.5794676806083651, "grad_norm": 0.6327412457994529, "learning_rate": 8.33323816762236e-06, "loss": 0.2953, "step": 8382 }, { "epoch": 0.5795368129968891, "grad_norm": 0.6878974035514096, "learning_rate": 8.332820917316225e-06, "loss": 0.4197, "step": 8383 }, { "epoch": 0.5796059453854131, "grad_norm": 0.6787768468500301, "learning_rate": 8.332403625239127e-06, "loss": 0.2974, "step": 8384 }, { "epoch": 0.5796750777739371, "grad_norm": 0.6190818785969876, "learning_rate": 8.331986291396297e-06, "loss": 0.231, "step": 8385 }, { "epoch": 0.5797442101624611, "grad_norm": 0.6894501199025113, "learning_rate": 8.331568915792965e-06, "loss": 0.381, "step": 8386 }, { "epoch": 0.5798133425509852, "grad_norm": 0.7215975583964228, "learning_rate": 8.331151498434363e-06, "loss": 0.3896, "step": 8387 }, { "epoch": 0.5798824749395092, "grad_norm": 1.1924669033231314, "learning_rate": 8.330734039325721e-06, "loss": 0.357, "step": 8388 }, { "epoch": 0.5799516073280332, "grad_norm": 1.211216565616607, "learning_rate": 8.330316538472271e-06, "loss": 0.4057, "step": 8389 }, { "epoch": 0.5800207397165572, "grad_norm": 0.6738740043914359, "learning_rate": 8.32989899587925e-06, "loss": 0.3769, "step": 8390 }, { "epoch": 0.5800898721050812, "grad_norm": 0.8329297692435811, "learning_rate": 8.329481411551883e-06, "loss": 0.3514, "step": 8391 }, { "epoch": 0.5801590044936052, "grad_norm": 0.7791647791724966, "learning_rate": 8.32906378549541e-06, "loss": 0.3388, "step": 8392 }, { "epoch": 0.5802281368821293, "grad_norm": 0.6969878371520776, "learning_rate": 8.328646117715063e-06, "loss": 0.3394, "step": 8393 }, { "epoch": 0.5802972692706533, "grad_norm": 0.6513491624793722, "learning_rate": 8.328228408216078e-06, "loss": 0.3936, "step": 8394 }, { "epoch": 0.5803664016591773, "grad_norm": 0.9099167951947522, "learning_rate": 8.32781065700369e-06, "loss": 0.3405, "step": 8395 }, { "epoch": 0.5804355340477013, "grad_norm": 0.8599181328751613, "learning_rate": 8.327392864083133e-06, "loss": 0.2757, "step": 8396 }, { "epoch": 0.5805046664362253, "grad_norm": 0.7448031546318178, "learning_rate": 8.326975029459643e-06, "loss": 0.4002, "step": 8397 }, { "epoch": 0.5805737988247494, "grad_norm": 0.7186071447528187, "learning_rate": 8.326557153138461e-06, "loss": 0.3543, "step": 8398 }, { "epoch": 0.5806429312132734, "grad_norm": 0.692730666292589, "learning_rate": 8.326139235124821e-06, "loss": 0.2631, "step": 8399 }, { "epoch": 0.5807120636017974, "grad_norm": 0.745476872809441, "learning_rate": 8.32572127542396e-06, "loss": 0.304, "step": 8400 }, { "epoch": 0.5807811959903215, "grad_norm": 0.8391178817226218, "learning_rate": 8.32530327404112e-06, "loss": 0.4317, "step": 8401 }, { "epoch": 0.5808503283788455, "grad_norm": 0.7917461733453751, "learning_rate": 8.324885230981536e-06, "loss": 0.3266, "step": 8402 }, { "epoch": 0.5809194607673696, "grad_norm": 0.6776098456327065, "learning_rate": 8.32446714625045e-06, "loss": 0.2564, "step": 8403 }, { "epoch": 0.5809885931558936, "grad_norm": 0.7442238676473981, "learning_rate": 8.3240490198531e-06, "loss": 0.425, "step": 8404 }, { "epoch": 0.5810577255444176, "grad_norm": 0.730377130036394, "learning_rate": 8.323630851794729e-06, "loss": 0.3724, "step": 8405 }, { "epoch": 0.5811268579329416, "grad_norm": 0.7248207085731005, "learning_rate": 8.323212642080576e-06, "loss": 0.2884, "step": 8406 }, { "epoch": 0.5811959903214656, "grad_norm": 3.5363585442583516, "learning_rate": 8.322794390715883e-06, "loss": 0.3297, "step": 8407 }, { "epoch": 0.5812651227099896, "grad_norm": 0.6903255272242808, "learning_rate": 8.32237609770589e-06, "loss": 0.449, "step": 8408 }, { "epoch": 0.5813342550985137, "grad_norm": 0.625789952891584, "learning_rate": 8.321957763055846e-06, "loss": 0.254, "step": 8409 }, { "epoch": 0.5814033874870377, "grad_norm": 0.8227686750627292, "learning_rate": 8.321539386770987e-06, "loss": 0.3433, "step": 8410 }, { "epoch": 0.5814725198755617, "grad_norm": 0.7023902833561364, "learning_rate": 8.32112096885656e-06, "loss": 0.3567, "step": 8411 }, { "epoch": 0.5815416522640857, "grad_norm": 0.790088453395632, "learning_rate": 8.320702509317808e-06, "loss": 0.4173, "step": 8412 }, { "epoch": 0.5816107846526097, "grad_norm": 0.7911904077404345, "learning_rate": 8.320284008159975e-06, "loss": 0.294, "step": 8413 }, { "epoch": 0.5816799170411338, "grad_norm": 0.7085893085206366, "learning_rate": 8.31986546538831e-06, "loss": 0.2799, "step": 8414 }, { "epoch": 0.5817490494296578, "grad_norm": 0.7114833301595411, "learning_rate": 8.319446881008053e-06, "loss": 0.3916, "step": 8415 }, { "epoch": 0.5818181818181818, "grad_norm": 1.0840695930554238, "learning_rate": 8.319028255024456e-06, "loss": 0.3782, "step": 8416 }, { "epoch": 0.5818873142067058, "grad_norm": 0.9072312057516108, "learning_rate": 8.318609587442761e-06, "loss": 0.2881, "step": 8417 }, { "epoch": 0.5819564465952298, "grad_norm": 0.6989154800211871, "learning_rate": 8.31819087826822e-06, "loss": 0.3186, "step": 8418 }, { "epoch": 0.5820255789837538, "grad_norm": 0.7367930438380246, "learning_rate": 8.317772127506074e-06, "loss": 0.4476, "step": 8419 }, { "epoch": 0.5820947113722779, "grad_norm": 0.9090317186755212, "learning_rate": 8.317353335161578e-06, "loss": 0.2954, "step": 8420 }, { "epoch": 0.582163843760802, "grad_norm": 0.7676642466082482, "learning_rate": 8.316934501239976e-06, "loss": 0.3349, "step": 8421 }, { "epoch": 0.582232976149326, "grad_norm": 0.7465260654342403, "learning_rate": 8.31651562574652e-06, "loss": 0.4382, "step": 8422 }, { "epoch": 0.58230210853785, "grad_norm": 0.6859970350564417, "learning_rate": 8.31609670868646e-06, "loss": 0.3061, "step": 8423 }, { "epoch": 0.582371240926374, "grad_norm": 0.86722131187771, "learning_rate": 8.315677750065045e-06, "loss": 0.3073, "step": 8424 }, { "epoch": 0.5824403733148981, "grad_norm": 0.7905521686228968, "learning_rate": 8.315258749887526e-06, "loss": 0.3528, "step": 8425 }, { "epoch": 0.5825095057034221, "grad_norm": 0.7072745711345643, "learning_rate": 8.314839708159157e-06, "loss": 0.432, "step": 8426 }, { "epoch": 0.5825786380919461, "grad_norm": 0.7234674953603031, "learning_rate": 8.314420624885187e-06, "loss": 0.2615, "step": 8427 }, { "epoch": 0.5826477704804701, "grad_norm": 0.8856049299918753, "learning_rate": 8.314001500070867e-06, "loss": 0.3116, "step": 8428 }, { "epoch": 0.5827169028689941, "grad_norm": 0.7757741096189997, "learning_rate": 8.313582333721456e-06, "loss": 0.3777, "step": 8429 }, { "epoch": 0.5827860352575182, "grad_norm": 0.6698147005949862, "learning_rate": 8.3131631258422e-06, "loss": 0.3713, "step": 8430 }, { "epoch": 0.5828551676460422, "grad_norm": 0.8136928791846841, "learning_rate": 8.31274387643836e-06, "loss": 0.2918, "step": 8431 }, { "epoch": 0.5829243000345662, "grad_norm": 0.8628251084320406, "learning_rate": 8.312324585515186e-06, "loss": 0.3002, "step": 8432 }, { "epoch": 0.5829934324230902, "grad_norm": 0.7573114742497122, "learning_rate": 8.311905253077935e-06, "loss": 0.4658, "step": 8433 }, { "epoch": 0.5830625648116142, "grad_norm": 0.9088615287469756, "learning_rate": 8.31148587913186e-06, "loss": 0.3271, "step": 8434 }, { "epoch": 0.5831316972001382, "grad_norm": 0.6788385929102323, "learning_rate": 8.31106646368222e-06, "loss": 0.2692, "step": 8435 }, { "epoch": 0.5832008295886623, "grad_norm": 0.7163892686338936, "learning_rate": 8.310647006734271e-06, "loss": 0.3669, "step": 8436 }, { "epoch": 0.5832699619771863, "grad_norm": 0.8317059574333291, "learning_rate": 8.310227508293272e-06, "loss": 0.4516, "step": 8437 }, { "epoch": 0.5833390943657103, "grad_norm": 0.9402158010811088, "learning_rate": 8.309807968364476e-06, "loss": 0.3367, "step": 8438 }, { "epoch": 0.5834082267542343, "grad_norm": 0.7355998147364563, "learning_rate": 8.309388386953144e-06, "loss": 0.3317, "step": 8439 }, { "epoch": 0.5834773591427583, "grad_norm": 0.7055655515738538, "learning_rate": 8.308968764064536e-06, "loss": 0.3752, "step": 8440 }, { "epoch": 0.5835464915312824, "grad_norm": 1.0271811899331849, "learning_rate": 8.308549099703907e-06, "loss": 0.2889, "step": 8441 }, { "epoch": 0.5836156239198065, "grad_norm": 0.8010098675169097, "learning_rate": 8.308129393876521e-06, "loss": 0.297, "step": 8442 }, { "epoch": 0.5836847563083305, "grad_norm": 0.7105022236381674, "learning_rate": 8.307709646587636e-06, "loss": 0.3017, "step": 8443 }, { "epoch": 0.5837538886968545, "grad_norm": 0.7675434847549664, "learning_rate": 8.307289857842515e-06, "loss": 0.4259, "step": 8444 }, { "epoch": 0.5838230210853785, "grad_norm": 0.6612584224990771, "learning_rate": 8.306870027646418e-06, "loss": 0.2729, "step": 8445 }, { "epoch": 0.5838921534739026, "grad_norm": 0.6943206059110673, "learning_rate": 8.306450156004605e-06, "loss": 0.2864, "step": 8446 }, { "epoch": 0.5839612858624266, "grad_norm": 0.9201015914537995, "learning_rate": 8.306030242922341e-06, "loss": 0.3642, "step": 8447 }, { "epoch": 0.5840304182509506, "grad_norm": 0.8241382944153652, "learning_rate": 8.305610288404886e-06, "loss": 0.4043, "step": 8448 }, { "epoch": 0.5840995506394746, "grad_norm": 0.702414849991717, "learning_rate": 8.305190292457508e-06, "loss": 0.3259, "step": 8449 }, { "epoch": 0.5841686830279986, "grad_norm": 0.7412613518560606, "learning_rate": 8.304770255085465e-06, "loss": 0.3275, "step": 8450 }, { "epoch": 0.5842378154165226, "grad_norm": 0.7344996859618985, "learning_rate": 8.304350176294026e-06, "loss": 0.3862, "step": 8451 }, { "epoch": 0.5843069478050467, "grad_norm": 0.7582764881778856, "learning_rate": 8.303930056088454e-06, "loss": 0.345, "step": 8452 }, { "epoch": 0.5843760801935707, "grad_norm": 0.6989061010013169, "learning_rate": 8.303509894474015e-06, "loss": 0.3035, "step": 8453 }, { "epoch": 0.5844452125820947, "grad_norm": 0.7216266260141695, "learning_rate": 8.303089691455975e-06, "loss": 0.377, "step": 8454 }, { "epoch": 0.5845143449706187, "grad_norm": 0.630030645120634, "learning_rate": 8.3026694470396e-06, "loss": 0.3303, "step": 8455 }, { "epoch": 0.5845834773591427, "grad_norm": 0.6990783694484134, "learning_rate": 8.302249161230157e-06, "loss": 0.3077, "step": 8456 }, { "epoch": 0.5846526097476668, "grad_norm": 0.9003507263129951, "learning_rate": 8.301828834032915e-06, "loss": 0.3492, "step": 8457 }, { "epoch": 0.5847217421361908, "grad_norm": 0.7194089797184833, "learning_rate": 8.301408465453141e-06, "loss": 0.4178, "step": 8458 }, { "epoch": 0.5847908745247148, "grad_norm": 0.6965852243908129, "learning_rate": 8.300988055496102e-06, "loss": 0.3053, "step": 8459 }, { "epoch": 0.5848600069132388, "grad_norm": 0.7505793428674565, "learning_rate": 8.300567604167071e-06, "loss": 0.2349, "step": 8460 }, { "epoch": 0.5849291393017628, "grad_norm": 0.7320287236748448, "learning_rate": 8.300147111471314e-06, "loss": 0.3598, "step": 8461 }, { "epoch": 0.584998271690287, "grad_norm": 0.8360777428560331, "learning_rate": 8.299726577414102e-06, "loss": 0.3866, "step": 8462 }, { "epoch": 0.585067404078811, "grad_norm": 0.8077918715895572, "learning_rate": 8.299306002000707e-06, "loss": 0.247, "step": 8463 }, { "epoch": 0.585136536467335, "grad_norm": 0.7654165679106615, "learning_rate": 8.298885385236397e-06, "loss": 0.351, "step": 8464 }, { "epoch": 0.585205668855859, "grad_norm": 0.7244572426998456, "learning_rate": 8.298464727126447e-06, "loss": 0.4854, "step": 8465 }, { "epoch": 0.585274801244383, "grad_norm": 0.8149113620057159, "learning_rate": 8.298044027676129e-06, "loss": 0.3472, "step": 8466 }, { "epoch": 0.585343933632907, "grad_norm": 0.6860737400273063, "learning_rate": 8.297623286890714e-06, "loss": 0.2556, "step": 8467 }, { "epoch": 0.5854130660214311, "grad_norm": 0.7794412525060556, "learning_rate": 8.297202504775476e-06, "loss": 0.3656, "step": 8468 }, { "epoch": 0.5854821984099551, "grad_norm": 0.8355161948080958, "learning_rate": 8.296781681335687e-06, "loss": 0.393, "step": 8469 }, { "epoch": 0.5855513307984791, "grad_norm": 0.6304070783158983, "learning_rate": 8.296360816576625e-06, "loss": 0.3118, "step": 8470 }, { "epoch": 0.5856204631870031, "grad_norm": 0.7599099110550958, "learning_rate": 8.295939910503563e-06, "loss": 0.3163, "step": 8471 }, { "epoch": 0.5856895955755271, "grad_norm": 0.7270026913033861, "learning_rate": 8.295518963121775e-06, "loss": 0.4179, "step": 8472 }, { "epoch": 0.5857587279640512, "grad_norm": 0.6964090275915666, "learning_rate": 8.29509797443654e-06, "loss": 0.3931, "step": 8473 }, { "epoch": 0.5858278603525752, "grad_norm": 0.6405256390016271, "learning_rate": 8.29467694445313e-06, "loss": 0.255, "step": 8474 }, { "epoch": 0.5858969927410992, "grad_norm": 0.7729259057964729, "learning_rate": 8.294255873176824e-06, "loss": 0.333, "step": 8475 }, { "epoch": 0.5859661251296232, "grad_norm": 0.6850978815006152, "learning_rate": 8.2938347606129e-06, "loss": 0.4664, "step": 8476 }, { "epoch": 0.5860352575181472, "grad_norm": 0.6105070848688126, "learning_rate": 8.293413606766636e-06, "loss": 0.3029, "step": 8477 }, { "epoch": 0.5861043899066712, "grad_norm": 0.633150651928455, "learning_rate": 8.292992411643311e-06, "loss": 0.2623, "step": 8478 }, { "epoch": 0.5861735222951953, "grad_norm": 0.682473627219383, "learning_rate": 8.2925711752482e-06, "loss": 0.4681, "step": 8479 }, { "epoch": 0.5862426546837193, "grad_norm": 0.821181993549544, "learning_rate": 8.292149897586586e-06, "loss": 0.4297, "step": 8480 }, { "epoch": 0.5863117870722433, "grad_norm": 0.7496118739287537, "learning_rate": 8.291728578663748e-06, "loss": 0.3058, "step": 8481 }, { "epoch": 0.5863809194607674, "grad_norm": 0.7836953668756956, "learning_rate": 8.291307218484968e-06, "loss": 0.3752, "step": 8482 }, { "epoch": 0.5864500518492914, "grad_norm": 0.6706081169468132, "learning_rate": 8.290885817055526e-06, "loss": 0.3443, "step": 8483 }, { "epoch": 0.5865191842378155, "grad_norm": 0.7026227649899649, "learning_rate": 8.2904643743807e-06, "loss": 0.2668, "step": 8484 }, { "epoch": 0.5865883166263395, "grad_norm": 0.6574240504614763, "learning_rate": 8.290042890465776e-06, "loss": 0.2625, "step": 8485 }, { "epoch": 0.5866574490148635, "grad_norm": 0.7002389336345428, "learning_rate": 8.289621365316039e-06, "loss": 0.4083, "step": 8486 }, { "epoch": 0.5867265814033875, "grad_norm": 0.6628434834344712, "learning_rate": 8.289199798936766e-06, "loss": 0.3815, "step": 8487 }, { "epoch": 0.5867957137919115, "grad_norm": 0.6840429641487152, "learning_rate": 8.288778191333244e-06, "loss": 0.298, "step": 8488 }, { "epoch": 0.5868648461804356, "grad_norm": 0.6709222994064499, "learning_rate": 8.288356542510755e-06, "loss": 0.2914, "step": 8489 }, { "epoch": 0.5869339785689596, "grad_norm": 0.7306776670506279, "learning_rate": 8.287934852474587e-06, "loss": 0.4073, "step": 8490 }, { "epoch": 0.5870031109574836, "grad_norm": 0.6947503362830878, "learning_rate": 8.287513121230022e-06, "loss": 0.3354, "step": 8491 }, { "epoch": 0.5870722433460076, "grad_norm": 0.6525843718666436, "learning_rate": 8.287091348782348e-06, "loss": 0.2776, "step": 8492 }, { "epoch": 0.5871413757345316, "grad_norm": 0.7318117625541777, "learning_rate": 8.286669535136849e-06, "loss": 0.3338, "step": 8493 }, { "epoch": 0.5872105081230556, "grad_norm": 0.8108023961463301, "learning_rate": 8.286247680298813e-06, "loss": 0.3774, "step": 8494 }, { "epoch": 0.5872796405115797, "grad_norm": 0.6204025852581677, "learning_rate": 8.28582578427353e-06, "loss": 0.2871, "step": 8495 }, { "epoch": 0.5873487729001037, "grad_norm": 0.6273472313089268, "learning_rate": 8.28540384706628e-06, "loss": 0.2545, "step": 8496 }, { "epoch": 0.5874179052886277, "grad_norm": 0.7838493377860669, "learning_rate": 8.284981868682358e-06, "loss": 0.4006, "step": 8497 }, { "epoch": 0.5874870376771517, "grad_norm": 0.8566091145648589, "learning_rate": 8.28455984912705e-06, "loss": 0.4141, "step": 8498 }, { "epoch": 0.5875561700656757, "grad_norm": 0.6018590483929257, "learning_rate": 8.284137788405647e-06, "loss": 0.2774, "step": 8499 }, { "epoch": 0.5876253024541997, "grad_norm": 0.6610373468951601, "learning_rate": 8.283715686523435e-06, "loss": 0.3202, "step": 8500 }, { "epoch": 0.5876944348427238, "grad_norm": 0.6957036885832131, "learning_rate": 8.283293543485709e-06, "loss": 0.4179, "step": 8501 }, { "epoch": 0.5877635672312478, "grad_norm": 0.6773734413024611, "learning_rate": 8.282871359297757e-06, "loss": 0.2729, "step": 8502 }, { "epoch": 0.5878326996197719, "grad_norm": 0.7707197934441845, "learning_rate": 8.282449133964872e-06, "loss": 0.3577, "step": 8503 }, { "epoch": 0.5879018320082959, "grad_norm": 0.686852121442987, "learning_rate": 8.282026867492343e-06, "loss": 0.3804, "step": 8504 }, { "epoch": 0.58797096439682, "grad_norm": 0.720165308540582, "learning_rate": 8.281604559885466e-06, "loss": 0.2938, "step": 8505 }, { "epoch": 0.588040096785344, "grad_norm": 0.7013435018691917, "learning_rate": 8.281182211149532e-06, "loss": 0.3076, "step": 8506 }, { "epoch": 0.588109229173868, "grad_norm": 0.7169797700534984, "learning_rate": 8.280759821289834e-06, "loss": 0.3459, "step": 8507 }, { "epoch": 0.588178361562392, "grad_norm": 0.7775487132420938, "learning_rate": 8.280337390311666e-06, "loss": 0.3838, "step": 8508 }, { "epoch": 0.588247493950916, "grad_norm": 0.6747273959520327, "learning_rate": 8.279914918220324e-06, "loss": 0.2483, "step": 8509 }, { "epoch": 0.58831662633944, "grad_norm": 0.6916308771316594, "learning_rate": 8.279492405021102e-06, "loss": 0.3161, "step": 8510 }, { "epoch": 0.5883857587279641, "grad_norm": 0.6608194457489284, "learning_rate": 8.279069850719293e-06, "loss": 0.3438, "step": 8511 }, { "epoch": 0.5884548911164881, "grad_norm": 0.7092998410809565, "learning_rate": 8.278647255320195e-06, "loss": 0.4286, "step": 8512 }, { "epoch": 0.5885240235050121, "grad_norm": 0.7388818740528021, "learning_rate": 8.278224618829107e-06, "loss": 0.3529, "step": 8513 }, { "epoch": 0.5885931558935361, "grad_norm": 0.6345035896535678, "learning_rate": 8.277801941251324e-06, "loss": 0.3651, "step": 8514 }, { "epoch": 0.5886622882820601, "grad_norm": 0.6795473624222435, "learning_rate": 8.277379222592141e-06, "loss": 0.4265, "step": 8515 }, { "epoch": 0.5887314206705841, "grad_norm": 0.6405440062744312, "learning_rate": 8.27695646285686e-06, "loss": 0.2969, "step": 8516 }, { "epoch": 0.5888005530591082, "grad_norm": 0.6280546990873539, "learning_rate": 8.276533662050777e-06, "loss": 0.2798, "step": 8517 }, { "epoch": 0.5888696854476322, "grad_norm": 0.8193572591233699, "learning_rate": 8.276110820179192e-06, "loss": 0.3599, "step": 8518 }, { "epoch": 0.5889388178361562, "grad_norm": 0.6746245064829288, "learning_rate": 8.275687937247403e-06, "loss": 0.3774, "step": 8519 }, { "epoch": 0.5890079502246802, "grad_norm": 0.6794138218211284, "learning_rate": 8.275265013260714e-06, "loss": 0.337, "step": 8520 }, { "epoch": 0.5890770826132042, "grad_norm": 0.6446823377681921, "learning_rate": 8.27484204822442e-06, "loss": 0.269, "step": 8521 }, { "epoch": 0.5891462150017283, "grad_norm": 0.6953220258651261, "learning_rate": 8.274419042143828e-06, "loss": 0.3808, "step": 8522 }, { "epoch": 0.5892153473902524, "grad_norm": 0.7188699775341645, "learning_rate": 8.273995995024235e-06, "loss": 0.3929, "step": 8523 }, { "epoch": 0.5892844797787764, "grad_norm": 0.8507280583395354, "learning_rate": 8.273572906870946e-06, "loss": 0.2591, "step": 8524 }, { "epoch": 0.5893536121673004, "grad_norm": 0.791678019541469, "learning_rate": 8.273149777689261e-06, "loss": 0.3112, "step": 8525 }, { "epoch": 0.5894227445558244, "grad_norm": 0.756833499824908, "learning_rate": 8.272726607484485e-06, "loss": 0.4185, "step": 8526 }, { "epoch": 0.5894918769443485, "grad_norm": 0.71286064587668, "learning_rate": 8.272303396261923e-06, "loss": 0.2977, "step": 8527 }, { "epoch": 0.5895610093328725, "grad_norm": 0.6898961844770508, "learning_rate": 8.271880144026877e-06, "loss": 0.2792, "step": 8528 }, { "epoch": 0.5896301417213965, "grad_norm": 0.77016877347308, "learning_rate": 8.271456850784651e-06, "loss": 0.4567, "step": 8529 }, { "epoch": 0.5896992741099205, "grad_norm": 0.7048380600463298, "learning_rate": 8.271033516540554e-06, "loss": 0.3768, "step": 8530 }, { "epoch": 0.5897684064984445, "grad_norm": 0.6535025947653964, "learning_rate": 8.270610141299887e-06, "loss": 0.3047, "step": 8531 }, { "epoch": 0.5898375388869685, "grad_norm": 0.7540822442022835, "learning_rate": 8.27018672506796e-06, "loss": 0.3467, "step": 8532 }, { "epoch": 0.5899066712754926, "grad_norm": 0.7019926923702854, "learning_rate": 8.269763267850077e-06, "loss": 0.4353, "step": 8533 }, { "epoch": 0.5899758036640166, "grad_norm": 0.8000505248816425, "learning_rate": 8.269339769651548e-06, "loss": 0.3808, "step": 8534 }, { "epoch": 0.5900449360525406, "grad_norm": 0.8157425503509991, "learning_rate": 8.268916230477679e-06, "loss": 0.2955, "step": 8535 }, { "epoch": 0.5901140684410646, "grad_norm": 0.670143237712882, "learning_rate": 8.268492650333777e-06, "loss": 0.3453, "step": 8536 }, { "epoch": 0.5901832008295886, "grad_norm": 0.708892104071164, "learning_rate": 8.268069029225154e-06, "loss": 0.3815, "step": 8537 }, { "epoch": 0.5902523332181127, "grad_norm": 0.707681012227319, "learning_rate": 8.267645367157117e-06, "loss": 0.2964, "step": 8538 }, { "epoch": 0.5903214656066367, "grad_norm": 0.7668405044733559, "learning_rate": 8.267221664134978e-06, "loss": 0.3411, "step": 8539 }, { "epoch": 0.5903905979951607, "grad_norm": 0.6592351210883769, "learning_rate": 8.266797920164048e-06, "loss": 0.4235, "step": 8540 }, { "epoch": 0.5904597303836847, "grad_norm": 0.7329540928748391, "learning_rate": 8.266374135249632e-06, "loss": 0.3975, "step": 8541 }, { "epoch": 0.5905288627722087, "grad_norm": 0.6904659593313573, "learning_rate": 8.265950309397046e-06, "loss": 0.2756, "step": 8542 }, { "epoch": 0.5905979951607329, "grad_norm": 0.6540316499614736, "learning_rate": 8.265526442611602e-06, "loss": 0.3381, "step": 8543 }, { "epoch": 0.5906671275492569, "grad_norm": 1.1556560747238098, "learning_rate": 8.265102534898613e-06, "loss": 0.354, "step": 8544 }, { "epoch": 0.5907362599377809, "grad_norm": 1.5166743896617227, "learning_rate": 8.264678586263389e-06, "loss": 0.3118, "step": 8545 }, { "epoch": 0.5908053923263049, "grad_norm": 1.8951518219592847, "learning_rate": 8.264254596711246e-06, "loss": 0.3199, "step": 8546 }, { "epoch": 0.5908745247148289, "grad_norm": 0.7248548354543988, "learning_rate": 8.263830566247494e-06, "loss": 0.4789, "step": 8547 }, { "epoch": 0.590943657103353, "grad_norm": 0.7026451887764538, "learning_rate": 8.263406494877452e-06, "loss": 0.3569, "step": 8548 }, { "epoch": 0.591012789491877, "grad_norm": 0.6809431063047855, "learning_rate": 8.262982382606435e-06, "loss": 0.2887, "step": 8549 }, { "epoch": 0.591081921880401, "grad_norm": 0.6526523740942624, "learning_rate": 8.262558229439756e-06, "loss": 0.2676, "step": 8550 }, { "epoch": 0.591151054268925, "grad_norm": 0.677503245329679, "learning_rate": 8.262134035382733e-06, "loss": 0.3827, "step": 8551 }, { "epoch": 0.591220186657449, "grad_norm": 0.8162639321722268, "learning_rate": 8.26170980044068e-06, "loss": 0.2902, "step": 8552 }, { "epoch": 0.591289319045973, "grad_norm": 0.6082387460184094, "learning_rate": 8.261285524618916e-06, "loss": 0.3061, "step": 8553 }, { "epoch": 0.5913584514344971, "grad_norm": 0.7265549289777348, "learning_rate": 8.260861207922757e-06, "loss": 0.4509, "step": 8554 }, { "epoch": 0.5914275838230211, "grad_norm": 0.8024229969828862, "learning_rate": 8.260436850357523e-06, "loss": 0.3753, "step": 8555 }, { "epoch": 0.5914967162115451, "grad_norm": 0.6476208583550815, "learning_rate": 8.26001245192853e-06, "loss": 0.2626, "step": 8556 }, { "epoch": 0.5915658486000691, "grad_norm": 0.7431252282288471, "learning_rate": 8.259588012641101e-06, "loss": 0.395, "step": 8557 }, { "epoch": 0.5916349809885931, "grad_norm": 0.6678152214049522, "learning_rate": 8.259163532500551e-06, "loss": 0.4033, "step": 8558 }, { "epoch": 0.5917041133771171, "grad_norm": 0.7340468589653127, "learning_rate": 8.258739011512204e-06, "loss": 0.289, "step": 8559 }, { "epoch": 0.5917732457656412, "grad_norm": 0.6942766881677516, "learning_rate": 8.258314449681379e-06, "loss": 0.3214, "step": 8560 }, { "epoch": 0.5918423781541652, "grad_norm": 0.8072315076472603, "learning_rate": 8.257889847013397e-06, "loss": 0.3841, "step": 8561 }, { "epoch": 0.5919115105426892, "grad_norm": 0.6638097516893553, "learning_rate": 8.25746520351358e-06, "loss": 0.4232, "step": 8562 }, { "epoch": 0.5919806429312132, "grad_norm": 0.6436652944875217, "learning_rate": 8.257040519187248e-06, "loss": 0.3517, "step": 8563 }, { "epoch": 0.5920497753197373, "grad_norm": 0.6991123598952343, "learning_rate": 8.256615794039726e-06, "loss": 0.2928, "step": 8564 }, { "epoch": 0.5921189077082614, "grad_norm": 0.9644303846990642, "learning_rate": 8.256191028076338e-06, "loss": 0.426, "step": 8565 }, { "epoch": 0.5921880400967854, "grad_norm": 0.5782323454657297, "learning_rate": 8.255766221302406e-06, "loss": 0.2734, "step": 8566 }, { "epoch": 0.5922571724853094, "grad_norm": 0.7747374022213656, "learning_rate": 8.255341373723253e-06, "loss": 0.35, "step": 8567 }, { "epoch": 0.5923263048738334, "grad_norm": 0.7024015298195901, "learning_rate": 8.254916485344209e-06, "loss": 0.3932, "step": 8568 }, { "epoch": 0.5923954372623574, "grad_norm": 0.720726389898214, "learning_rate": 8.254491556170592e-06, "loss": 0.3761, "step": 8569 }, { "epoch": 0.5924645696508815, "grad_norm": 0.710531095834933, "learning_rate": 8.254066586207733e-06, "loss": 0.3586, "step": 8570 }, { "epoch": 0.5925337020394055, "grad_norm": 0.6179679579986253, "learning_rate": 8.253641575460956e-06, "loss": 0.2874, "step": 8571 }, { "epoch": 0.5926028344279295, "grad_norm": 0.8861657551663459, "learning_rate": 8.253216523935588e-06, "loss": 0.386, "step": 8572 }, { "epoch": 0.5926719668164535, "grad_norm": 0.6549037216670953, "learning_rate": 8.252791431636957e-06, "loss": 0.3124, "step": 8573 }, { "epoch": 0.5927410992049775, "grad_norm": 0.6100181306894348, "learning_rate": 8.252366298570389e-06, "loss": 0.2892, "step": 8574 }, { "epoch": 0.5928102315935015, "grad_norm": 0.6856347014804747, "learning_rate": 8.251941124741214e-06, "loss": 0.2919, "step": 8575 }, { "epoch": 0.5928793639820256, "grad_norm": 0.7226520441670817, "learning_rate": 8.251515910154762e-06, "loss": 0.4093, "step": 8576 }, { "epoch": 0.5929484963705496, "grad_norm": 0.7368614332892439, "learning_rate": 8.25109065481636e-06, "loss": 0.3499, "step": 8577 }, { "epoch": 0.5930176287590736, "grad_norm": 0.6255716674394455, "learning_rate": 8.250665358731336e-06, "loss": 0.3337, "step": 8578 }, { "epoch": 0.5930867611475976, "grad_norm": 0.7202063108557115, "learning_rate": 8.250240021905026e-06, "loss": 0.3801, "step": 8579 }, { "epoch": 0.5931558935361216, "grad_norm": 0.788415225740051, "learning_rate": 8.249814644342756e-06, "loss": 0.3503, "step": 8580 }, { "epoch": 0.5932250259246457, "grad_norm": 0.7345292994763856, "learning_rate": 8.24938922604986e-06, "loss": 0.3498, "step": 8581 }, { "epoch": 0.5932941583131697, "grad_norm": 0.7406271446091836, "learning_rate": 8.248963767031669e-06, "loss": 0.3131, "step": 8582 }, { "epoch": 0.5933632907016937, "grad_norm": 0.834031439946859, "learning_rate": 8.248538267293513e-06, "loss": 0.4395, "step": 8583 }, { "epoch": 0.5934324230902178, "grad_norm": 0.6935836489155328, "learning_rate": 8.24811272684073e-06, "loss": 0.3189, "step": 8584 }, { "epoch": 0.5935015554787418, "grad_norm": 0.6711776225610142, "learning_rate": 8.247687145678647e-06, "loss": 0.2638, "step": 8585 }, { "epoch": 0.5935706878672659, "grad_norm": 0.7073360311882907, "learning_rate": 8.247261523812605e-06, "loss": 0.4312, "step": 8586 }, { "epoch": 0.5936398202557899, "grad_norm": 0.7289319361723662, "learning_rate": 8.246835861247933e-06, "loss": 0.4256, "step": 8587 }, { "epoch": 0.5937089526443139, "grad_norm": 0.7633871305864274, "learning_rate": 8.246410157989968e-06, "loss": 0.2954, "step": 8588 }, { "epoch": 0.5937780850328379, "grad_norm": 0.7293156832716751, "learning_rate": 8.245984414044045e-06, "loss": 0.3279, "step": 8589 }, { "epoch": 0.5938472174213619, "grad_norm": 1.5193629780415299, "learning_rate": 8.2455586294155e-06, "loss": 0.3943, "step": 8590 }, { "epoch": 0.593916349809886, "grad_norm": 0.6900744454942893, "learning_rate": 8.24513280410967e-06, "loss": 0.3869, "step": 8591 }, { "epoch": 0.59398548219841, "grad_norm": 0.7603745319077492, "learning_rate": 8.244706938131892e-06, "loss": 0.2839, "step": 8592 }, { "epoch": 0.594054614586934, "grad_norm": 0.7481225326641393, "learning_rate": 8.244281031487503e-06, "loss": 0.3587, "step": 8593 }, { "epoch": 0.594123746975458, "grad_norm": 0.6945109134598918, "learning_rate": 8.243855084181837e-06, "loss": 0.4596, "step": 8594 }, { "epoch": 0.594192879363982, "grad_norm": 0.6987727113652122, "learning_rate": 8.243429096220241e-06, "loss": 0.3377, "step": 8595 }, { "epoch": 0.594262011752506, "grad_norm": 0.9012631656795508, "learning_rate": 8.24300306760805e-06, "loss": 0.3244, "step": 8596 }, { "epoch": 0.59433114414103, "grad_norm": 0.848177669641396, "learning_rate": 8.2425769983506e-06, "loss": 0.4127, "step": 8597 }, { "epoch": 0.5944002765295541, "grad_norm": 0.7015144105176457, "learning_rate": 8.242150888453235e-06, "loss": 0.4594, "step": 8598 }, { "epoch": 0.5944694089180781, "grad_norm": 0.7079221440757828, "learning_rate": 8.241724737921292e-06, "loss": 0.2969, "step": 8599 }, { "epoch": 0.5945385413066021, "grad_norm": 0.7323860121344775, "learning_rate": 8.24129854676012e-06, "loss": 0.3522, "step": 8600 }, { "epoch": 0.5946076736951261, "grad_norm": 0.736730333733974, "learning_rate": 8.24087231497505e-06, "loss": 0.444, "step": 8601 }, { "epoch": 0.5946768060836501, "grad_norm": 0.706761320768767, "learning_rate": 8.240446042571431e-06, "loss": 0.2995, "step": 8602 }, { "epoch": 0.5947459384721742, "grad_norm": 0.7089709818223014, "learning_rate": 8.240019729554605e-06, "loss": 0.3263, "step": 8603 }, { "epoch": 0.5948150708606983, "grad_norm": 0.7636097414354708, "learning_rate": 8.239593375929913e-06, "loss": 0.3944, "step": 8604 }, { "epoch": 0.5948842032492223, "grad_norm": 0.78344974875255, "learning_rate": 8.239166981702698e-06, "loss": 0.386, "step": 8605 }, { "epoch": 0.5949533356377463, "grad_norm": 0.6095393056513773, "learning_rate": 8.238740546878307e-06, "loss": 0.2886, "step": 8606 }, { "epoch": 0.5950224680262703, "grad_norm": 0.649243895565633, "learning_rate": 8.238314071462081e-06, "loss": 0.3256, "step": 8607 }, { "epoch": 0.5950916004147944, "grad_norm": 0.7500965713520421, "learning_rate": 8.237887555459369e-06, "loss": 0.4111, "step": 8608 }, { "epoch": 0.5951607328033184, "grad_norm": 0.7373859811106876, "learning_rate": 8.237460998875515e-06, "loss": 0.2902, "step": 8609 }, { "epoch": 0.5952298651918424, "grad_norm": 0.6596354078475319, "learning_rate": 8.237034401715864e-06, "loss": 0.3321, "step": 8610 }, { "epoch": 0.5952989975803664, "grad_norm": 0.8388050172802135, "learning_rate": 8.236607763985764e-06, "loss": 0.4205, "step": 8611 }, { "epoch": 0.5953681299688904, "grad_norm": 0.7409340744190664, "learning_rate": 8.236181085690561e-06, "loss": 0.4169, "step": 8612 }, { "epoch": 0.5954372623574145, "grad_norm": 0.6722332704390811, "learning_rate": 8.235754366835605e-06, "loss": 0.2828, "step": 8613 }, { "epoch": 0.5955063947459385, "grad_norm": 0.7630555799653429, "learning_rate": 8.23532760742624e-06, "loss": 0.3132, "step": 8614 }, { "epoch": 0.5955755271344625, "grad_norm": 0.7651142857478161, "learning_rate": 8.234900807467818e-06, "loss": 0.4092, "step": 8615 }, { "epoch": 0.5956446595229865, "grad_norm": 0.6346400796044186, "learning_rate": 8.234473966965689e-06, "loss": 0.2368, "step": 8616 }, { "epoch": 0.5957137919115105, "grad_norm": 0.6172641476965398, "learning_rate": 8.2340470859252e-06, "loss": 0.2787, "step": 8617 }, { "epoch": 0.5957829243000345, "grad_norm": 0.7261881274195058, "learning_rate": 8.233620164351702e-06, "loss": 0.3531, "step": 8618 }, { "epoch": 0.5958520566885586, "grad_norm": 1.4945188658228137, "learning_rate": 8.233193202250546e-06, "loss": 0.4155, "step": 8619 }, { "epoch": 0.5959211890770826, "grad_norm": 0.6824406441659179, "learning_rate": 8.232766199627082e-06, "loss": 0.2593, "step": 8620 }, { "epoch": 0.5959903214656066, "grad_norm": 0.7213552092949274, "learning_rate": 8.232339156486665e-06, "loss": 0.2568, "step": 8621 }, { "epoch": 0.5960594538541306, "grad_norm": 0.7406806800000496, "learning_rate": 8.231912072834643e-06, "loss": 0.3593, "step": 8622 }, { "epoch": 0.5961285862426546, "grad_norm": 0.6206016865702951, "learning_rate": 8.231484948676371e-06, "loss": 0.2611, "step": 8623 }, { "epoch": 0.5961977186311787, "grad_norm": 0.6838827637813484, "learning_rate": 8.231057784017202e-06, "loss": 0.2503, "step": 8624 }, { "epoch": 0.5962668510197028, "grad_norm": 0.7522510472342904, "learning_rate": 8.230630578862491e-06, "loss": 0.2634, "step": 8625 }, { "epoch": 0.5963359834082268, "grad_norm": 0.7650147241218875, "learning_rate": 8.23020333321759e-06, "loss": 0.4711, "step": 8626 }, { "epoch": 0.5964051157967508, "grad_norm": 0.6651725431330762, "learning_rate": 8.229776047087856e-06, "loss": 0.2609, "step": 8627 }, { "epoch": 0.5964742481852748, "grad_norm": 0.6788369584989525, "learning_rate": 8.229348720478641e-06, "loss": 0.3018, "step": 8628 }, { "epoch": 0.5965433805737989, "grad_norm": 0.7776124580531444, "learning_rate": 8.228921353395304e-06, "loss": 0.4012, "step": 8629 }, { "epoch": 0.5966125129623229, "grad_norm": 0.7042794777959517, "learning_rate": 8.228493945843201e-06, "loss": 0.3813, "step": 8630 }, { "epoch": 0.5966816453508469, "grad_norm": 0.6984345607779402, "learning_rate": 8.228066497827688e-06, "loss": 0.317, "step": 8631 }, { "epoch": 0.5967507777393709, "grad_norm": 0.6928088303549811, "learning_rate": 8.227639009354121e-06, "loss": 0.2724, "step": 8632 }, { "epoch": 0.5968199101278949, "grad_norm": 0.738070934934987, "learning_rate": 8.227211480427857e-06, "loss": 0.4021, "step": 8633 }, { "epoch": 0.5968890425164189, "grad_norm": 0.7031051304601471, "learning_rate": 8.22678391105426e-06, "loss": 0.2771, "step": 8634 }, { "epoch": 0.596958174904943, "grad_norm": 0.6845508691264887, "learning_rate": 8.226356301238683e-06, "loss": 0.3125, "step": 8635 }, { "epoch": 0.597027307293467, "grad_norm": 0.6774827571596194, "learning_rate": 8.225928650986487e-06, "loss": 0.395, "step": 8636 }, { "epoch": 0.597096439681991, "grad_norm": 0.6986194809964948, "learning_rate": 8.225500960303034e-06, "loss": 0.4546, "step": 8637 }, { "epoch": 0.597165572070515, "grad_norm": 0.5983109511279313, "learning_rate": 8.225073229193681e-06, "loss": 0.2898, "step": 8638 }, { "epoch": 0.597234704459039, "grad_norm": 0.6448673673449149, "learning_rate": 8.224645457663793e-06, "loss": 0.3544, "step": 8639 }, { "epoch": 0.597303836847563, "grad_norm": 0.7567948691896828, "learning_rate": 8.224217645718726e-06, "loss": 0.4747, "step": 8640 }, { "epoch": 0.5973729692360871, "grad_norm": 0.7181642277787494, "learning_rate": 8.223789793363846e-06, "loss": 0.3105, "step": 8641 }, { "epoch": 0.5974421016246111, "grad_norm": 0.8713835449997153, "learning_rate": 8.223361900604513e-06, "loss": 0.285, "step": 8642 }, { "epoch": 0.5975112340131351, "grad_norm": 1.0842276300010107, "learning_rate": 8.222933967446093e-06, "loss": 0.3269, "step": 8643 }, { "epoch": 0.5975803664016591, "grad_norm": 0.7630280164943263, "learning_rate": 8.222505993893947e-06, "loss": 0.3974, "step": 8644 }, { "epoch": 0.5976494987901833, "grad_norm": 0.7933974750380246, "learning_rate": 8.222077979953438e-06, "loss": 0.3562, "step": 8645 }, { "epoch": 0.5977186311787073, "grad_norm": 0.8108176787318585, "learning_rate": 8.221649925629932e-06, "loss": 0.2797, "step": 8646 }, { "epoch": 0.5977877635672313, "grad_norm": 0.7967773757469974, "learning_rate": 8.221221830928794e-06, "loss": 0.465, "step": 8647 }, { "epoch": 0.5978568959557553, "grad_norm": 0.7428875220009115, "learning_rate": 8.220793695855386e-06, "loss": 0.3191, "step": 8648 }, { "epoch": 0.5979260283442793, "grad_norm": 0.6841368426219766, "learning_rate": 8.22036552041508e-06, "loss": 0.3037, "step": 8649 }, { "epoch": 0.5979951607328033, "grad_norm": 0.9692045589249244, "learning_rate": 8.21993730461324e-06, "loss": 0.3012, "step": 8650 }, { "epoch": 0.5980642931213274, "grad_norm": 0.6719699232483709, "learning_rate": 8.219509048455229e-06, "loss": 0.3537, "step": 8651 }, { "epoch": 0.5981334255098514, "grad_norm": 0.8577097434947011, "learning_rate": 8.219080751946419e-06, "loss": 0.2877, "step": 8652 }, { "epoch": 0.5982025578983754, "grad_norm": 0.7204906287844548, "learning_rate": 8.218652415092177e-06, "loss": 0.3367, "step": 8653 }, { "epoch": 0.5982716902868994, "grad_norm": 0.7485273619503919, "learning_rate": 8.218224037897871e-06, "loss": 0.405, "step": 8654 }, { "epoch": 0.5983408226754234, "grad_norm": 0.6798867975439085, "learning_rate": 8.21779562036887e-06, "loss": 0.3352, "step": 8655 }, { "epoch": 0.5984099550639475, "grad_norm": 0.689836352255314, "learning_rate": 8.217367162510542e-06, "loss": 0.2704, "step": 8656 }, { "epoch": 0.5984790874524715, "grad_norm": 0.7012124503535387, "learning_rate": 8.216938664328257e-06, "loss": 0.2944, "step": 8657 }, { "epoch": 0.5985482198409955, "grad_norm": 0.7706179876529368, "learning_rate": 8.216510125827389e-06, "loss": 0.446, "step": 8658 }, { "epoch": 0.5986173522295195, "grad_norm": 0.6840222634915362, "learning_rate": 8.216081547013305e-06, "loss": 0.3722, "step": 8659 }, { "epoch": 0.5986864846180435, "grad_norm": 0.7310320306596778, "learning_rate": 8.21565292789138e-06, "loss": 0.3074, "step": 8660 }, { "epoch": 0.5987556170065675, "grad_norm": 0.6664513381180898, "learning_rate": 8.215224268466984e-06, "loss": 0.3565, "step": 8661 }, { "epoch": 0.5988247493950916, "grad_norm": 0.7036004519929735, "learning_rate": 8.21479556874549e-06, "loss": 0.3831, "step": 8662 }, { "epoch": 0.5988938817836156, "grad_norm": 0.6788679367139743, "learning_rate": 8.21436682873227e-06, "loss": 0.3636, "step": 8663 }, { "epoch": 0.5989630141721396, "grad_norm": 0.7827743438885543, "learning_rate": 8.213938048432697e-06, "loss": 0.2985, "step": 8664 }, { "epoch": 0.5990321465606637, "grad_norm": 0.7409188935305451, "learning_rate": 8.213509227852147e-06, "loss": 0.4416, "step": 8665 }, { "epoch": 0.5991012789491877, "grad_norm": 0.7794339208096125, "learning_rate": 8.213080366995994e-06, "loss": 0.3207, "step": 8666 }, { "epoch": 0.5991704113377118, "grad_norm": 0.806808742915029, "learning_rate": 8.212651465869613e-06, "loss": 0.2835, "step": 8667 }, { "epoch": 0.5992395437262358, "grad_norm": 0.8689782420669453, "learning_rate": 8.212222524478377e-06, "loss": 0.3557, "step": 8668 }, { "epoch": 0.5993086761147598, "grad_norm": 0.6863765319488694, "learning_rate": 8.211793542827667e-06, "loss": 0.397, "step": 8669 }, { "epoch": 0.5993778085032838, "grad_norm": 1.5404916700266338, "learning_rate": 8.211364520922852e-06, "loss": 0.283, "step": 8670 }, { "epoch": 0.5994469408918078, "grad_norm": 0.6138309898959364, "learning_rate": 8.210935458769318e-06, "loss": 0.2719, "step": 8671 }, { "epoch": 0.5995160732803319, "grad_norm": 0.6588475470948432, "learning_rate": 8.210506356372436e-06, "loss": 0.4094, "step": 8672 }, { "epoch": 0.5995852056688559, "grad_norm": 0.6877814232586825, "learning_rate": 8.210077213737588e-06, "loss": 0.3118, "step": 8673 }, { "epoch": 0.5996543380573799, "grad_norm": 0.667429116715519, "learning_rate": 8.20964803087015e-06, "loss": 0.2594, "step": 8674 }, { "epoch": 0.5997234704459039, "grad_norm": 0.6695441832805268, "learning_rate": 8.209218807775501e-06, "loss": 0.3492, "step": 8675 }, { "epoch": 0.5997926028344279, "grad_norm": 0.7094657690409513, "learning_rate": 8.208789544459023e-06, "loss": 0.412, "step": 8676 }, { "epoch": 0.5998617352229519, "grad_norm": 0.702948299605332, "learning_rate": 8.208360240926094e-06, "loss": 0.3093, "step": 8677 }, { "epoch": 0.599930867611476, "grad_norm": 0.665808708289263, "learning_rate": 8.207930897182094e-06, "loss": 0.3252, "step": 8678 }, { "epoch": 0.6, "grad_norm": 0.7355513149480215, "learning_rate": 8.207501513232406e-06, "loss": 0.3948, "step": 8679 }, { "epoch": 0.600069132388524, "grad_norm": 0.6568835771701333, "learning_rate": 8.207072089082409e-06, "loss": 0.3345, "step": 8680 }, { "epoch": 0.600138264777048, "grad_norm": 0.6949930076114175, "learning_rate": 8.206642624737487e-06, "loss": 0.3326, "step": 8681 }, { "epoch": 0.600207397165572, "grad_norm": 1.0244254969743358, "learning_rate": 8.206213120203023e-06, "loss": 0.368, "step": 8682 }, { "epoch": 0.600276529554096, "grad_norm": 0.7541564445989815, "learning_rate": 8.2057835754844e-06, "loss": 0.4849, "step": 8683 }, { "epoch": 0.6003456619426201, "grad_norm": 0.6245920052998574, "learning_rate": 8.205353990587001e-06, "loss": 0.3265, "step": 8684 }, { "epoch": 0.6004147943311441, "grad_norm": 0.673022131018649, "learning_rate": 8.20492436551621e-06, "loss": 0.2976, "step": 8685 }, { "epoch": 0.6004839267196682, "grad_norm": 0.6916599383157922, "learning_rate": 8.204494700277409e-06, "loss": 0.3487, "step": 8686 }, { "epoch": 0.6005530591081922, "grad_norm": 1.0333344829262436, "learning_rate": 8.204064994875987e-06, "loss": 0.4148, "step": 8687 }, { "epoch": 0.6006221914967163, "grad_norm": 0.6765226503421975, "learning_rate": 8.203635249317327e-06, "loss": 0.3129, "step": 8688 }, { "epoch": 0.6006913238852403, "grad_norm": 0.7138122031015566, "learning_rate": 8.203205463606817e-06, "loss": 0.3653, "step": 8689 }, { "epoch": 0.6007604562737643, "grad_norm": 0.6777325847068254, "learning_rate": 8.202775637749844e-06, "loss": 0.3631, "step": 8690 }, { "epoch": 0.6008295886622883, "grad_norm": 0.7917569634288791, "learning_rate": 8.202345771751794e-06, "loss": 0.3658, "step": 8691 }, { "epoch": 0.6008987210508123, "grad_norm": 0.675232515032452, "learning_rate": 8.201915865618056e-06, "loss": 0.3162, "step": 8692 }, { "epoch": 0.6009678534393363, "grad_norm": 0.6789052724789333, "learning_rate": 8.201485919354013e-06, "loss": 0.3046, "step": 8693 }, { "epoch": 0.6010369858278604, "grad_norm": 0.7223610998853913, "learning_rate": 8.20105593296506e-06, "loss": 0.4169, "step": 8694 }, { "epoch": 0.6011061182163844, "grad_norm": 0.6832677728396357, "learning_rate": 8.200625906456582e-06, "loss": 0.3361, "step": 8695 }, { "epoch": 0.6011752506049084, "grad_norm": 0.7865556417015716, "learning_rate": 8.200195839833971e-06, "loss": 0.3113, "step": 8696 }, { "epoch": 0.6012443829934324, "grad_norm": 0.6938427123714841, "learning_rate": 8.199765733102615e-06, "loss": 0.3792, "step": 8697 }, { "epoch": 0.6013135153819564, "grad_norm": 0.6842277988627217, "learning_rate": 8.199335586267907e-06, "loss": 0.3287, "step": 8698 }, { "epoch": 0.6013826477704804, "grad_norm": 0.667918320659143, "learning_rate": 8.198905399335235e-06, "loss": 0.3157, "step": 8699 }, { "epoch": 0.6014517801590045, "grad_norm": 0.7103815239288784, "learning_rate": 8.198475172309994e-06, "loss": 0.3633, "step": 8700 }, { "epoch": 0.6015209125475285, "grad_norm": 0.7155985981836547, "learning_rate": 8.198044905197575e-06, "loss": 0.4157, "step": 8701 }, { "epoch": 0.6015900449360525, "grad_norm": 0.7356068097318732, "learning_rate": 8.19761459800337e-06, "loss": 0.2629, "step": 8702 }, { "epoch": 0.6016591773245765, "grad_norm": 0.8543074164761387, "learning_rate": 8.197184250732772e-06, "loss": 0.3678, "step": 8703 }, { "epoch": 0.6017283097131005, "grad_norm": 0.7031868143628609, "learning_rate": 8.196753863391175e-06, "loss": 0.369, "step": 8704 }, { "epoch": 0.6017974421016246, "grad_norm": 0.7300603026659721, "learning_rate": 8.196323435983973e-06, "loss": 0.4104, "step": 8705 }, { "epoch": 0.6018665744901487, "grad_norm": 0.7121194679532032, "learning_rate": 8.195892968516561e-06, "loss": 0.3957, "step": 8706 }, { "epoch": 0.6019357068786727, "grad_norm": 0.7072740646718643, "learning_rate": 8.195462460994335e-06, "loss": 0.3544, "step": 8707 }, { "epoch": 0.6020048392671967, "grad_norm": 0.7354780306805633, "learning_rate": 8.195031913422688e-06, "loss": 0.4091, "step": 8708 }, { "epoch": 0.6020739716557207, "grad_norm": 0.7086956825381164, "learning_rate": 8.194601325807019e-06, "loss": 0.3461, "step": 8709 }, { "epoch": 0.6021431040442448, "grad_norm": 0.6475328422310964, "learning_rate": 8.194170698152724e-06, "loss": 0.3418, "step": 8710 }, { "epoch": 0.6022122364327688, "grad_norm": 0.6762138601442726, "learning_rate": 8.193740030465199e-06, "loss": 0.3437, "step": 8711 }, { "epoch": 0.6022813688212928, "grad_norm": 0.7887379460801809, "learning_rate": 8.19330932274984e-06, "loss": 0.4379, "step": 8712 }, { "epoch": 0.6023505012098168, "grad_norm": 0.6124629642790544, "learning_rate": 8.19287857501205e-06, "loss": 0.2854, "step": 8713 }, { "epoch": 0.6024196335983408, "grad_norm": 0.7046902950706047, "learning_rate": 8.192447787257224e-06, "loss": 0.3296, "step": 8714 }, { "epoch": 0.6024887659868648, "grad_norm": 0.660890692477307, "learning_rate": 8.192016959490763e-06, "loss": 0.4516, "step": 8715 }, { "epoch": 0.6025578983753889, "grad_norm": 0.7812644705983885, "learning_rate": 8.191586091718064e-06, "loss": 0.3294, "step": 8716 }, { "epoch": 0.6026270307639129, "grad_norm": 0.6726840802252729, "learning_rate": 8.19115518394453e-06, "loss": 0.248, "step": 8717 }, { "epoch": 0.6026961631524369, "grad_norm": 0.666478146627543, "learning_rate": 8.19072423617556e-06, "loss": 0.3074, "step": 8718 }, { "epoch": 0.6027652955409609, "grad_norm": 0.7152510614243066, "learning_rate": 8.190293248416558e-06, "loss": 0.4311, "step": 8719 }, { "epoch": 0.6028344279294849, "grad_norm": 0.7632484387435082, "learning_rate": 8.189862220672922e-06, "loss": 0.3212, "step": 8720 }, { "epoch": 0.602903560318009, "grad_norm": 0.7098470243928693, "learning_rate": 8.189431152950055e-06, "loss": 0.2629, "step": 8721 }, { "epoch": 0.602972692706533, "grad_norm": 0.8204635181748714, "learning_rate": 8.189000045253361e-06, "loss": 0.3916, "step": 8722 }, { "epoch": 0.603041825095057, "grad_norm": 0.952392849687124, "learning_rate": 8.188568897588243e-06, "loss": 0.3558, "step": 8723 }, { "epoch": 0.603110957483581, "grad_norm": 0.7229904505151614, "learning_rate": 8.188137709960102e-06, "loss": 0.2624, "step": 8724 }, { "epoch": 0.603180089872105, "grad_norm": 0.7149958638767117, "learning_rate": 8.187706482374346e-06, "loss": 0.3292, "step": 8725 }, { "epoch": 0.6032492222606292, "grad_norm": 0.9165305905258713, "learning_rate": 8.187275214836379e-06, "loss": 0.4263, "step": 8726 }, { "epoch": 0.6033183546491532, "grad_norm": 0.9194297598806886, "learning_rate": 8.186843907351602e-06, "loss": 0.3148, "step": 8727 }, { "epoch": 0.6033874870376772, "grad_norm": 0.7794972060096961, "learning_rate": 8.186412559925427e-06, "loss": 0.3539, "step": 8728 }, { "epoch": 0.6034566194262012, "grad_norm": 0.9273880811005579, "learning_rate": 8.185981172563255e-06, "loss": 0.3976, "step": 8729 }, { "epoch": 0.6035257518147252, "grad_norm": 0.7239561099936905, "learning_rate": 8.185549745270496e-06, "loss": 0.3856, "step": 8730 }, { "epoch": 0.6035948842032492, "grad_norm": 1.0937715477690122, "learning_rate": 8.185118278052554e-06, "loss": 0.3103, "step": 8731 }, { "epoch": 0.6036640165917733, "grad_norm": 0.7306715737134885, "learning_rate": 8.184686770914841e-06, "loss": 0.3651, "step": 8732 }, { "epoch": 0.6037331489802973, "grad_norm": 0.7424289770491297, "learning_rate": 8.18425522386276e-06, "loss": 0.3957, "step": 8733 }, { "epoch": 0.6038022813688213, "grad_norm": 1.0022972275894655, "learning_rate": 8.183823636901724e-06, "loss": 0.3014, "step": 8734 }, { "epoch": 0.6038714137573453, "grad_norm": 0.7649045614705714, "learning_rate": 8.18339201003714e-06, "loss": 0.2546, "step": 8735 }, { "epoch": 0.6039405461458693, "grad_norm": 0.8316562063524053, "learning_rate": 8.182960343274418e-06, "loss": 0.3989, "step": 8736 }, { "epoch": 0.6040096785343934, "grad_norm": 0.6741571419919908, "learning_rate": 8.182528636618969e-06, "loss": 0.372, "step": 8737 }, { "epoch": 0.6040788109229174, "grad_norm": 0.6126095836232635, "learning_rate": 8.182096890076204e-06, "loss": 0.3257, "step": 8738 }, { "epoch": 0.6041479433114414, "grad_norm": 0.7330443054281327, "learning_rate": 8.181665103651531e-06, "loss": 0.3824, "step": 8739 }, { "epoch": 0.6042170756999654, "grad_norm": 0.8880163068581667, "learning_rate": 8.181233277350365e-06, "loss": 0.4225, "step": 8740 }, { "epoch": 0.6042862080884894, "grad_norm": 0.7237200157461979, "learning_rate": 8.180801411178116e-06, "loss": 0.3829, "step": 8741 }, { "epoch": 0.6043553404770134, "grad_norm": 0.7680118343216874, "learning_rate": 8.180369505140198e-06, "loss": 0.3636, "step": 8742 }, { "epoch": 0.6044244728655375, "grad_norm": 0.8758707373954182, "learning_rate": 8.179937559242023e-06, "loss": 0.3824, "step": 8743 }, { "epoch": 0.6044936052540615, "grad_norm": 0.7188837324076568, "learning_rate": 8.179505573489009e-06, "loss": 0.4468, "step": 8744 }, { "epoch": 0.6045627376425855, "grad_norm": 0.7896087293581981, "learning_rate": 8.179073547886564e-06, "loss": 0.3284, "step": 8745 }, { "epoch": 0.6046318700311095, "grad_norm": 4.892314315651547, "learning_rate": 8.178641482440106e-06, "loss": 0.298, "step": 8746 }, { "epoch": 0.6047010024196336, "grad_norm": 0.706251204738906, "learning_rate": 8.178209377155049e-06, "loss": 0.3915, "step": 8747 }, { "epoch": 0.6047701348081577, "grad_norm": 0.7243411796506078, "learning_rate": 8.17777723203681e-06, "loss": 0.3297, "step": 8748 }, { "epoch": 0.6048392671966817, "grad_norm": 0.8784358446062016, "learning_rate": 8.177345047090805e-06, "loss": 0.3044, "step": 8749 }, { "epoch": 0.6049083995852057, "grad_norm": 0.7504498135778039, "learning_rate": 8.176912822322449e-06, "loss": 0.2978, "step": 8750 }, { "epoch": 0.6049775319737297, "grad_norm": 0.7387686582720442, "learning_rate": 8.176480557737161e-06, "loss": 0.4102, "step": 8751 }, { "epoch": 0.6050466643622537, "grad_norm": 0.8697275627602573, "learning_rate": 8.176048253340357e-06, "loss": 0.3168, "step": 8752 }, { "epoch": 0.6051157967507778, "grad_norm": 0.6769789687160572, "learning_rate": 8.175615909137455e-06, "loss": 0.3323, "step": 8753 }, { "epoch": 0.6051849291393018, "grad_norm": 0.6599828667650288, "learning_rate": 8.175183525133877e-06, "loss": 0.4074, "step": 8754 }, { "epoch": 0.6052540615278258, "grad_norm": 0.8242681680560778, "learning_rate": 8.174751101335038e-06, "loss": 0.3706, "step": 8755 }, { "epoch": 0.6053231939163498, "grad_norm": 0.9904230184910505, "learning_rate": 8.174318637746361e-06, "loss": 0.3571, "step": 8756 }, { "epoch": 0.6053923263048738, "grad_norm": 0.812263862672458, "learning_rate": 8.173886134373264e-06, "loss": 0.3233, "step": 8757 }, { "epoch": 0.6054614586933978, "grad_norm": 0.8807970146898589, "learning_rate": 8.17345359122117e-06, "loss": 0.3971, "step": 8758 }, { "epoch": 0.6055305910819219, "grad_norm": 0.7387535573342165, "learning_rate": 8.173021008295497e-06, "loss": 0.2826, "step": 8759 }, { "epoch": 0.6055997234704459, "grad_norm": 0.7717357756837685, "learning_rate": 8.172588385601667e-06, "loss": 0.3444, "step": 8760 }, { "epoch": 0.6056688558589699, "grad_norm": 0.7588073346885913, "learning_rate": 8.172155723145105e-06, "loss": 0.3249, "step": 8761 }, { "epoch": 0.6057379882474939, "grad_norm": 0.7069735600529865, "learning_rate": 8.17172302093123e-06, "loss": 0.3549, "step": 8762 }, { "epoch": 0.6058071206360179, "grad_norm": 0.8871961579954066, "learning_rate": 8.17129027896547e-06, "loss": 0.3054, "step": 8763 }, { "epoch": 0.605876253024542, "grad_norm": 0.759869328606415, "learning_rate": 8.170857497253244e-06, "loss": 0.3962, "step": 8764 }, { "epoch": 0.605945385413066, "grad_norm": 0.6855068029186234, "learning_rate": 8.17042467579998e-06, "loss": 0.3581, "step": 8765 }, { "epoch": 0.60601451780159, "grad_norm": 0.6597074303721425, "learning_rate": 8.169991814611098e-06, "loss": 0.3487, "step": 8766 }, { "epoch": 0.6060836501901141, "grad_norm": 0.6462510321513383, "learning_rate": 8.169558913692027e-06, "loss": 0.2666, "step": 8767 }, { "epoch": 0.6061527825786381, "grad_norm": 0.7086135331162224, "learning_rate": 8.169125973048191e-06, "loss": 0.3172, "step": 8768 }, { "epoch": 0.6062219149671622, "grad_norm": 0.8210683980564917, "learning_rate": 8.168692992685016e-06, "loss": 0.4861, "step": 8769 }, { "epoch": 0.6062910473556862, "grad_norm": 0.6963245877424192, "learning_rate": 8.16825997260793e-06, "loss": 0.2888, "step": 8770 }, { "epoch": 0.6063601797442102, "grad_norm": 0.7119989649863895, "learning_rate": 8.167826912822358e-06, "loss": 0.2676, "step": 8771 }, { "epoch": 0.6064293121327342, "grad_norm": 0.8254982033387495, "learning_rate": 8.167393813333731e-06, "loss": 0.4598, "step": 8772 }, { "epoch": 0.6064984445212582, "grad_norm": 0.7510214727308049, "learning_rate": 8.166960674147472e-06, "loss": 0.3934, "step": 8773 }, { "epoch": 0.6065675769097822, "grad_norm": 0.9696278330183137, "learning_rate": 8.166527495269015e-06, "loss": 0.301, "step": 8774 }, { "epoch": 0.6066367092983063, "grad_norm": 0.7614114993344195, "learning_rate": 8.166094276703787e-06, "loss": 0.4069, "step": 8775 }, { "epoch": 0.6067058416868303, "grad_norm": 0.7344839513123345, "learning_rate": 8.165661018457216e-06, "loss": 0.4097, "step": 8776 }, { "epoch": 0.6067749740753543, "grad_norm": 0.7095252841673729, "learning_rate": 8.165227720534734e-06, "loss": 0.2859, "step": 8777 }, { "epoch": 0.6068441064638783, "grad_norm": 0.7623843523344775, "learning_rate": 8.164794382941771e-06, "loss": 0.293, "step": 8778 }, { "epoch": 0.6069132388524023, "grad_norm": 0.7215076442004739, "learning_rate": 8.164361005683758e-06, "loss": 0.4601, "step": 8779 }, { "epoch": 0.6069823712409264, "grad_norm": 0.7398877111150371, "learning_rate": 8.163927588766127e-06, "loss": 0.3077, "step": 8780 }, { "epoch": 0.6070515036294504, "grad_norm": 0.6458933205192195, "learning_rate": 8.163494132194312e-06, "loss": 0.2869, "step": 8781 }, { "epoch": 0.6071206360179744, "grad_norm": 0.917674614620778, "learning_rate": 8.16306063597374e-06, "loss": 0.331, "step": 8782 }, { "epoch": 0.6071897684064984, "grad_norm": 0.765536974268752, "learning_rate": 8.16262710010985e-06, "loss": 0.4454, "step": 8783 }, { "epoch": 0.6072589007950224, "grad_norm": 0.8169393966455947, "learning_rate": 8.162193524608072e-06, "loss": 0.3347, "step": 8784 }, { "epoch": 0.6073280331835464, "grad_norm": 0.8319982350208199, "learning_rate": 8.161759909473843e-06, "loss": 0.3182, "step": 8785 }, { "epoch": 0.6073971655720705, "grad_norm": 1.2603058671704297, "learning_rate": 8.161326254712596e-06, "loss": 0.4458, "step": 8786 }, { "epoch": 0.6074662979605946, "grad_norm": 0.7274050165438317, "learning_rate": 8.160892560329764e-06, "loss": 0.3676, "step": 8787 }, { "epoch": 0.6075354303491186, "grad_norm": 0.7523559755942408, "learning_rate": 8.160458826330785e-06, "loss": 0.3354, "step": 8788 }, { "epoch": 0.6076045627376426, "grad_norm": 0.7777350232549654, "learning_rate": 8.160025052721097e-06, "loss": 0.3671, "step": 8789 }, { "epoch": 0.6076736951261666, "grad_norm": 0.7216601741118706, "learning_rate": 8.159591239506133e-06, "loss": 0.4247, "step": 8790 }, { "epoch": 0.6077428275146907, "grad_norm": 0.6301856536042729, "learning_rate": 8.15915738669133e-06, "loss": 0.3082, "step": 8791 }, { "epoch": 0.6078119599032147, "grad_norm": 0.8459226915133615, "learning_rate": 8.158723494282127e-06, "loss": 0.3053, "step": 8792 }, { "epoch": 0.6078810922917387, "grad_norm": 0.7364971797882076, "learning_rate": 8.158289562283965e-06, "loss": 0.3575, "step": 8793 }, { "epoch": 0.6079502246802627, "grad_norm": 1.0847279464351798, "learning_rate": 8.157855590702277e-06, "loss": 0.4007, "step": 8794 }, { "epoch": 0.6080193570687867, "grad_norm": 0.6136202992528658, "learning_rate": 8.157421579542506e-06, "loss": 0.3107, "step": 8795 }, { "epoch": 0.6080884894573108, "grad_norm": 1.1750708746424352, "learning_rate": 8.156987528810087e-06, "loss": 0.2904, "step": 8796 }, { "epoch": 0.6081576218458348, "grad_norm": 0.8109783442906024, "learning_rate": 8.156553438510467e-06, "loss": 0.4173, "step": 8797 }, { "epoch": 0.6082267542343588, "grad_norm": 1.0627522283859119, "learning_rate": 8.15611930864908e-06, "loss": 0.3999, "step": 8798 }, { "epoch": 0.6082958866228828, "grad_norm": 0.7360381057763394, "learning_rate": 8.155685139231372e-06, "loss": 0.3378, "step": 8799 }, { "epoch": 0.6083650190114068, "grad_norm": 0.7301275454871046, "learning_rate": 8.155250930262782e-06, "loss": 0.2931, "step": 8800 }, { "epoch": 0.6084341513999308, "grad_norm": 0.7273775250456885, "learning_rate": 8.154816681748751e-06, "loss": 0.4329, "step": 8801 }, { "epoch": 0.6085032837884549, "grad_norm": 0.7209958665286906, "learning_rate": 8.154382393694724e-06, "loss": 0.3298, "step": 8802 }, { "epoch": 0.6085724161769789, "grad_norm": 0.65086659927895, "learning_rate": 8.153948066106144e-06, "loss": 0.27, "step": 8803 }, { "epoch": 0.6086415485655029, "grad_norm": 0.8379293629344575, "learning_rate": 8.15351369898845e-06, "loss": 0.4129, "step": 8804 }, { "epoch": 0.6087106809540269, "grad_norm": 0.7271294390025184, "learning_rate": 8.153079292347092e-06, "loss": 0.3691, "step": 8805 }, { "epoch": 0.6087798133425509, "grad_norm": 0.6192433877188818, "learning_rate": 8.152644846187511e-06, "loss": 0.271, "step": 8806 }, { "epoch": 0.6088489457310751, "grad_norm": 0.7572805473951021, "learning_rate": 8.152210360515154e-06, "loss": 0.3661, "step": 8807 }, { "epoch": 0.6089180781195991, "grad_norm": 0.881729914909347, "learning_rate": 8.151775835335467e-06, "loss": 0.3793, "step": 8808 }, { "epoch": 0.6089872105081231, "grad_norm": 0.6855707945959831, "learning_rate": 8.151341270653893e-06, "loss": 0.3124, "step": 8809 }, { "epoch": 0.6090563428966471, "grad_norm": 0.8475994804720625, "learning_rate": 8.150906666475879e-06, "loss": 0.3634, "step": 8810 }, { "epoch": 0.6091254752851711, "grad_norm": 0.746856495493701, "learning_rate": 8.150472022806873e-06, "loss": 0.3869, "step": 8811 }, { "epoch": 0.6091946076736952, "grad_norm": 0.8713990887032098, "learning_rate": 8.150037339652324e-06, "loss": 0.3574, "step": 8812 }, { "epoch": 0.6092637400622192, "grad_norm": 0.9750441251420302, "learning_rate": 8.149602617017677e-06, "loss": 0.3007, "step": 8813 }, { "epoch": 0.6093328724507432, "grad_norm": 0.7689585355611621, "learning_rate": 8.149167854908381e-06, "loss": 0.419, "step": 8814 }, { "epoch": 0.6094020048392672, "grad_norm": 0.7369037806257648, "learning_rate": 8.14873305332989e-06, "loss": 0.4162, "step": 8815 }, { "epoch": 0.6094711372277912, "grad_norm": 0.744815842948354, "learning_rate": 8.148298212287647e-06, "loss": 0.392, "step": 8816 }, { "epoch": 0.6095402696163152, "grad_norm": 0.6627261956368794, "learning_rate": 8.147863331787102e-06, "loss": 0.2834, "step": 8817 }, { "epoch": 0.6096094020048393, "grad_norm": 0.7018455081819979, "learning_rate": 8.147428411833712e-06, "loss": 0.3279, "step": 8818 }, { "epoch": 0.6096785343933633, "grad_norm": 0.8217019241662443, "learning_rate": 8.146993452432919e-06, "loss": 0.5141, "step": 8819 }, { "epoch": 0.6097476667818873, "grad_norm": 0.6530993758197184, "learning_rate": 8.146558453590183e-06, "loss": 0.3064, "step": 8820 }, { "epoch": 0.6098167991704113, "grad_norm": 0.699613021298422, "learning_rate": 8.14612341531095e-06, "loss": 0.2948, "step": 8821 }, { "epoch": 0.6098859315589353, "grad_norm": 0.6553175227497611, "learning_rate": 8.145688337600674e-06, "loss": 0.3823, "step": 8822 }, { "epoch": 0.6099550639474594, "grad_norm": 0.7527309794463135, "learning_rate": 8.14525322046481e-06, "loss": 0.3623, "step": 8823 }, { "epoch": 0.6100241963359834, "grad_norm": 0.8800256403564083, "learning_rate": 8.14481806390881e-06, "loss": 0.307, "step": 8824 }, { "epoch": 0.6100933287245074, "grad_norm": 0.7408452610106568, "learning_rate": 8.144382867938127e-06, "loss": 0.3068, "step": 8825 }, { "epoch": 0.6101624611130314, "grad_norm": 0.8097412365052119, "learning_rate": 8.143947632558215e-06, "loss": 0.4386, "step": 8826 }, { "epoch": 0.6102315935015554, "grad_norm": 0.6558531937419907, "learning_rate": 8.143512357774532e-06, "loss": 0.3121, "step": 8827 }, { "epoch": 0.6103007258900796, "grad_norm": 0.6134734150602129, "learning_rate": 8.14307704359253e-06, "loss": 0.2808, "step": 8828 }, { "epoch": 0.6103698582786036, "grad_norm": 0.7581625398037579, "learning_rate": 8.14264169001767e-06, "loss": 0.3898, "step": 8829 }, { "epoch": 0.6104389906671276, "grad_norm": 0.6858080716539202, "learning_rate": 8.142206297055401e-06, "loss": 0.3155, "step": 8830 }, { "epoch": 0.6105081230556516, "grad_norm": 0.8186139098366241, "learning_rate": 8.141770864711184e-06, "loss": 0.2836, "step": 8831 }, { "epoch": 0.6105772554441756, "grad_norm": 0.6712679100576436, "learning_rate": 8.141335392990477e-06, "loss": 0.3178, "step": 8832 }, { "epoch": 0.6106463878326996, "grad_norm": 0.6852187768645334, "learning_rate": 8.140899881898737e-06, "loss": 0.3505, "step": 8833 }, { "epoch": 0.6107155202212237, "grad_norm": 0.7358214344416389, "learning_rate": 8.14046433144142e-06, "loss": 0.2832, "step": 8834 }, { "epoch": 0.6107846526097477, "grad_norm": 0.6526047984880803, "learning_rate": 8.140028741623988e-06, "loss": 0.2628, "step": 8835 }, { "epoch": 0.6108537849982717, "grad_norm": 0.7575500808145376, "learning_rate": 8.139593112451901e-06, "loss": 0.3379, "step": 8836 }, { "epoch": 0.6109229173867957, "grad_norm": 0.8053614594191895, "learning_rate": 8.139157443930614e-06, "loss": 0.3464, "step": 8837 }, { "epoch": 0.6109920497753197, "grad_norm": 0.7918673405473675, "learning_rate": 8.138721736065593e-06, "loss": 0.3101, "step": 8838 }, { "epoch": 0.6110611821638438, "grad_norm": 0.8354615551272873, "learning_rate": 8.138285988862296e-06, "loss": 0.2535, "step": 8839 }, { "epoch": 0.6111303145523678, "grad_norm": 0.6539304443433402, "learning_rate": 8.137850202326185e-06, "loss": 0.387, "step": 8840 }, { "epoch": 0.6111994469408918, "grad_norm": 0.6554685813148244, "learning_rate": 8.13741437646272e-06, "loss": 0.3547, "step": 8841 }, { "epoch": 0.6112685793294158, "grad_norm": 0.675084786447232, "learning_rate": 8.136978511277366e-06, "loss": 0.3353, "step": 8842 }, { "epoch": 0.6113377117179398, "grad_norm": 0.741044008022738, "learning_rate": 8.136542606775583e-06, "loss": 0.3695, "step": 8843 }, { "epoch": 0.6114068441064638, "grad_norm": 0.794402314985284, "learning_rate": 8.136106662962838e-06, "loss": 0.4715, "step": 8844 }, { "epoch": 0.6114759764949879, "grad_norm": 0.6114747502848129, "learning_rate": 8.135670679844592e-06, "loss": 0.2945, "step": 8845 }, { "epoch": 0.6115451088835119, "grad_norm": 0.6911463027300012, "learning_rate": 8.13523465742631e-06, "loss": 0.263, "step": 8846 }, { "epoch": 0.6116142412720359, "grad_norm": 0.6954022402601668, "learning_rate": 8.134798595713456e-06, "loss": 0.3804, "step": 8847 }, { "epoch": 0.61168337366056, "grad_norm": 0.7502866108442456, "learning_rate": 8.134362494711496e-06, "loss": 0.2935, "step": 8848 }, { "epoch": 0.611752506049084, "grad_norm": 0.6948088623625075, "learning_rate": 8.133926354425895e-06, "loss": 0.3029, "step": 8849 }, { "epoch": 0.6118216384376081, "grad_norm": 0.670567525336453, "learning_rate": 8.13349017486212e-06, "loss": 0.3034, "step": 8850 }, { "epoch": 0.6118907708261321, "grad_norm": 0.920464620013753, "learning_rate": 8.133053956025638e-06, "loss": 0.4097, "step": 8851 }, { "epoch": 0.6119599032146561, "grad_norm": 0.612251853479613, "learning_rate": 8.132617697921917e-06, "loss": 0.275, "step": 8852 }, { "epoch": 0.6120290356031801, "grad_norm": 0.7031255007190471, "learning_rate": 8.132181400556422e-06, "loss": 0.2755, "step": 8853 }, { "epoch": 0.6120981679917041, "grad_norm": 0.6930356809463122, "learning_rate": 8.131745063934625e-06, "loss": 0.3631, "step": 8854 }, { "epoch": 0.6121673003802282, "grad_norm": 0.6688132630241057, "learning_rate": 8.13130868806199e-06, "loss": 0.309, "step": 8855 }, { "epoch": 0.6122364327687522, "grad_norm": 0.7398312009281285, "learning_rate": 8.130872272943991e-06, "loss": 0.3349, "step": 8856 }, { "epoch": 0.6123055651572762, "grad_norm": 0.666211209217936, "learning_rate": 8.130435818586093e-06, "loss": 0.2963, "step": 8857 }, { "epoch": 0.6123746975458002, "grad_norm": 0.7012161833385513, "learning_rate": 8.12999932499377e-06, "loss": 0.4204, "step": 8858 }, { "epoch": 0.6124438299343242, "grad_norm": 0.6335545744741653, "learning_rate": 8.12956279217249e-06, "loss": 0.3342, "step": 8859 }, { "epoch": 0.6125129623228482, "grad_norm": 0.8820570402621106, "learning_rate": 8.129126220127726e-06, "loss": 0.2722, "step": 8860 }, { "epoch": 0.6125820947113723, "grad_norm": 1.4655742684392168, "learning_rate": 8.128689608864949e-06, "loss": 0.395, "step": 8861 }, { "epoch": 0.6126512270998963, "grad_norm": 0.733510300101043, "learning_rate": 8.128252958389631e-06, "loss": 0.3887, "step": 8862 }, { "epoch": 0.6127203594884203, "grad_norm": 4.2899364336409, "learning_rate": 8.127816268707244e-06, "loss": 0.2571, "step": 8863 }, { "epoch": 0.6127894918769443, "grad_norm": 0.6831199468425679, "learning_rate": 8.127379539823264e-06, "loss": 0.2918, "step": 8864 }, { "epoch": 0.6128586242654683, "grad_norm": 0.784549483948934, "learning_rate": 8.12694277174316e-06, "loss": 0.4399, "step": 8865 }, { "epoch": 0.6129277566539924, "grad_norm": 0.6681803507605741, "learning_rate": 8.126505964472412e-06, "loss": 0.3013, "step": 8866 }, { "epoch": 0.6129968890425164, "grad_norm": 0.8375269243196998, "learning_rate": 8.12606911801649e-06, "loss": 0.3029, "step": 8867 }, { "epoch": 0.6130660214310405, "grad_norm": 0.6886651086673545, "learning_rate": 8.125632232380868e-06, "loss": 0.4076, "step": 8868 }, { "epoch": 0.6131351538195645, "grad_norm": 0.6744003525325675, "learning_rate": 8.125195307571025e-06, "loss": 0.4067, "step": 8869 }, { "epoch": 0.6132042862080885, "grad_norm": 0.9219338215442672, "learning_rate": 8.124758343592436e-06, "loss": 0.3228, "step": 8870 }, { "epoch": 0.6132734185966126, "grad_norm": 1.6848743065152125, "learning_rate": 8.124321340450579e-06, "loss": 0.3114, "step": 8871 }, { "epoch": 0.6133425509851366, "grad_norm": 0.81048298862883, "learning_rate": 8.123884298150928e-06, "loss": 0.3874, "step": 8872 }, { "epoch": 0.6134116833736606, "grad_norm": 0.80731164183716, "learning_rate": 8.123447216698962e-06, "loss": 0.3526, "step": 8873 }, { "epoch": 0.6134808157621846, "grad_norm": 0.7812359922642252, "learning_rate": 8.123010096100161e-06, "loss": 0.2919, "step": 8874 }, { "epoch": 0.6135499481507086, "grad_norm": 0.7083687896042519, "learning_rate": 8.122572936360001e-06, "loss": 0.2966, "step": 8875 }, { "epoch": 0.6136190805392326, "grad_norm": 0.7341954948976629, "learning_rate": 8.12213573748396e-06, "loss": 0.4075, "step": 8876 }, { "epoch": 0.6136882129277567, "grad_norm": 0.7531104718300331, "learning_rate": 8.121698499477521e-06, "loss": 0.3384, "step": 8877 }, { "epoch": 0.6137573453162807, "grad_norm": 0.6474034757636329, "learning_rate": 8.121261222346162e-06, "loss": 0.3083, "step": 8878 }, { "epoch": 0.6138264777048047, "grad_norm": 3.153296561452977, "learning_rate": 8.120823906095364e-06, "loss": 0.4262, "step": 8879 }, { "epoch": 0.6138956100933287, "grad_norm": 0.7610127672169602, "learning_rate": 8.120386550730606e-06, "loss": 0.3844, "step": 8880 }, { "epoch": 0.6139647424818527, "grad_norm": 0.7465650253355295, "learning_rate": 8.119949156257374e-06, "loss": 0.3396, "step": 8881 }, { "epoch": 0.6140338748703768, "grad_norm": 0.7547350843170222, "learning_rate": 8.119511722681146e-06, "loss": 0.3013, "step": 8882 }, { "epoch": 0.6141030072589008, "grad_norm": 0.7664737789964804, "learning_rate": 8.119074250007404e-06, "loss": 0.4417, "step": 8883 }, { "epoch": 0.6141721396474248, "grad_norm": 0.7450096119495824, "learning_rate": 8.118636738241634e-06, "loss": 0.3101, "step": 8884 }, { "epoch": 0.6142412720359488, "grad_norm": 0.6773562488612993, "learning_rate": 8.118199187389318e-06, "loss": 0.3002, "step": 8885 }, { "epoch": 0.6143104044244728, "grad_norm": 0.7324003615799864, "learning_rate": 8.11776159745594e-06, "loss": 0.3777, "step": 8886 }, { "epoch": 0.6143795368129968, "grad_norm": 0.8253128078075296, "learning_rate": 8.117323968446983e-06, "loss": 0.4398, "step": 8887 }, { "epoch": 0.6144486692015209, "grad_norm": 0.6514010242937124, "learning_rate": 8.116886300367934e-06, "loss": 0.2892, "step": 8888 }, { "epoch": 0.614517801590045, "grad_norm": 0.8029426954780089, "learning_rate": 8.116448593224277e-06, "loss": 0.3306, "step": 8889 }, { "epoch": 0.614586933978569, "grad_norm": 0.717320069275005, "learning_rate": 8.116010847021498e-06, "loss": 0.4676, "step": 8890 }, { "epoch": 0.614656066367093, "grad_norm": 0.6885761469638786, "learning_rate": 8.115573061765085e-06, "loss": 0.4021, "step": 8891 }, { "epoch": 0.614725198755617, "grad_norm": 0.6076441993361804, "learning_rate": 8.115135237460524e-06, "loss": 0.2588, "step": 8892 }, { "epoch": 0.6147943311441411, "grad_norm": 0.7639180626847065, "learning_rate": 8.114697374113301e-06, "loss": 0.331, "step": 8893 }, { "epoch": 0.6148634635326651, "grad_norm": 0.7504619589689462, "learning_rate": 8.114259471728905e-06, "loss": 0.4329, "step": 8894 }, { "epoch": 0.6149325959211891, "grad_norm": 0.6899542500149475, "learning_rate": 8.113821530312825e-06, "loss": 0.3636, "step": 8895 }, { "epoch": 0.6150017283097131, "grad_norm": 0.7337717542264947, "learning_rate": 8.113383549870548e-06, "loss": 0.3929, "step": 8896 }, { "epoch": 0.6150708606982371, "grad_norm": 0.7223817203741287, "learning_rate": 8.112945530407564e-06, "loss": 0.4432, "step": 8897 }, { "epoch": 0.6151399930867611, "grad_norm": 0.6945100721711076, "learning_rate": 8.112507471929365e-06, "loss": 0.3517, "step": 8898 }, { "epoch": 0.6152091254752852, "grad_norm": 0.7660380505278183, "learning_rate": 8.112069374441438e-06, "loss": 0.3089, "step": 8899 }, { "epoch": 0.6152782578638092, "grad_norm": 0.7294712496603442, "learning_rate": 8.111631237949275e-06, "loss": 0.3587, "step": 8900 }, { "epoch": 0.6153473902523332, "grad_norm": 0.721878611100321, "learning_rate": 8.111193062458367e-06, "loss": 0.3871, "step": 8901 }, { "epoch": 0.6154165226408572, "grad_norm": 0.6129589889195297, "learning_rate": 8.110754847974207e-06, "loss": 0.3017, "step": 8902 }, { "epoch": 0.6154856550293812, "grad_norm": 0.7085414360909978, "learning_rate": 8.110316594502286e-06, "loss": 0.2578, "step": 8903 }, { "epoch": 0.6155547874179053, "grad_norm": 0.6887309521295236, "learning_rate": 8.109878302048098e-06, "loss": 0.3938, "step": 8904 }, { "epoch": 0.6156239198064293, "grad_norm": 0.7153277735451141, "learning_rate": 8.109439970617134e-06, "loss": 0.3668, "step": 8905 }, { "epoch": 0.6156930521949533, "grad_norm": 0.6555015862802461, "learning_rate": 8.10900160021489e-06, "loss": 0.3285, "step": 8906 }, { "epoch": 0.6157621845834773, "grad_norm": 1.2351349126964883, "learning_rate": 8.108563190846858e-06, "loss": 0.3102, "step": 8907 }, { "epoch": 0.6158313169720013, "grad_norm": 0.7309332675808781, "learning_rate": 8.108124742518535e-06, "loss": 0.3735, "step": 8908 }, { "epoch": 0.6159004493605255, "grad_norm": 1.3282844383407213, "learning_rate": 8.107686255235413e-06, "loss": 0.3109, "step": 8909 }, { "epoch": 0.6159695817490495, "grad_norm": 0.6336109598826096, "learning_rate": 8.107247729002992e-06, "loss": 0.3446, "step": 8910 }, { "epoch": 0.6160387141375735, "grad_norm": 0.6681943842032548, "learning_rate": 8.106809163826765e-06, "loss": 0.2855, "step": 8911 }, { "epoch": 0.6161078465260975, "grad_norm": 0.7536436103640344, "learning_rate": 8.10637055971223e-06, "loss": 0.4352, "step": 8912 }, { "epoch": 0.6161769789146215, "grad_norm": 1.9520603515566373, "learning_rate": 8.105931916664883e-06, "loss": 0.2729, "step": 8913 }, { "epoch": 0.6162461113031455, "grad_norm": 0.7894410118359094, "learning_rate": 8.105493234690222e-06, "loss": 0.3152, "step": 8914 }, { "epoch": 0.6163152436916696, "grad_norm": 0.8013216859402137, "learning_rate": 8.105054513793747e-06, "loss": 0.412, "step": 8915 }, { "epoch": 0.6163843760801936, "grad_norm": 0.7402852705135259, "learning_rate": 8.104615753980953e-06, "loss": 0.3865, "step": 8916 }, { "epoch": 0.6164535084687176, "grad_norm": 0.6879106741963787, "learning_rate": 8.104176955257341e-06, "loss": 0.3527, "step": 8917 }, { "epoch": 0.6165226408572416, "grad_norm": 0.733222343841822, "learning_rate": 8.103738117628413e-06, "loss": 0.3284, "step": 8918 }, { "epoch": 0.6165917732457656, "grad_norm": 0.701760782093491, "learning_rate": 8.103299241099663e-06, "loss": 0.4269, "step": 8919 }, { "epoch": 0.6166609056342897, "grad_norm": 0.8202453881354106, "learning_rate": 8.102860325676598e-06, "loss": 0.3066, "step": 8920 }, { "epoch": 0.6167300380228137, "grad_norm": 0.7639393571768948, "learning_rate": 8.102421371364714e-06, "loss": 0.3171, "step": 8921 }, { "epoch": 0.6167991704113377, "grad_norm": 0.7556877372475392, "learning_rate": 8.101982378169515e-06, "loss": 0.4204, "step": 8922 }, { "epoch": 0.6168683027998617, "grad_norm": 1.6827091523970694, "learning_rate": 8.101543346096505e-06, "loss": 0.2902, "step": 8923 }, { "epoch": 0.6169374351883857, "grad_norm": 0.6893820155717539, "learning_rate": 8.10110427515118e-06, "loss": 0.2958, "step": 8924 }, { "epoch": 0.6170065675769097, "grad_norm": 0.7419244184558977, "learning_rate": 8.10066516533905e-06, "loss": 0.3458, "step": 8925 }, { "epoch": 0.6170756999654338, "grad_norm": 0.6781141551418345, "learning_rate": 8.100226016665616e-06, "loss": 0.388, "step": 8926 }, { "epoch": 0.6171448323539578, "grad_norm": 0.7822607263772048, "learning_rate": 8.09978682913638e-06, "loss": 0.3429, "step": 8927 }, { "epoch": 0.6172139647424818, "grad_norm": 0.7498366627806706, "learning_rate": 8.099347602756849e-06, "loss": 0.2601, "step": 8928 }, { "epoch": 0.6172830971310059, "grad_norm": 0.7536755727513066, "learning_rate": 8.098908337532524e-06, "loss": 0.3867, "step": 8929 }, { "epoch": 0.61735222951953, "grad_norm": 0.6557926496498574, "learning_rate": 8.098469033468917e-06, "loss": 0.366, "step": 8930 }, { "epoch": 0.617421361908054, "grad_norm": 0.6744954046863677, "learning_rate": 8.098029690571529e-06, "loss": 0.2687, "step": 8931 }, { "epoch": 0.617490494296578, "grad_norm": 0.7109430161729021, "learning_rate": 8.097590308845866e-06, "loss": 0.2719, "step": 8932 }, { "epoch": 0.617559626685102, "grad_norm": 0.7281680309964836, "learning_rate": 8.097150888297439e-06, "loss": 0.3705, "step": 8933 }, { "epoch": 0.617628759073626, "grad_norm": 0.7997754336624875, "learning_rate": 8.09671142893175e-06, "loss": 0.3493, "step": 8934 }, { "epoch": 0.61769789146215, "grad_norm": 0.7433006717812944, "learning_rate": 8.096271930754312e-06, "loss": 0.3574, "step": 8935 }, { "epoch": 0.6177670238506741, "grad_norm": 0.8546014218563185, "learning_rate": 8.095832393770629e-06, "loss": 0.4206, "step": 8936 }, { "epoch": 0.6178361562391981, "grad_norm": 0.6597806633036907, "learning_rate": 8.095392817986212e-06, "loss": 0.3709, "step": 8937 }, { "epoch": 0.6179052886277221, "grad_norm": 1.5889008378121192, "learning_rate": 8.094953203406571e-06, "loss": 0.2829, "step": 8938 }, { "epoch": 0.6179744210162461, "grad_norm": 0.9062039458496636, "learning_rate": 8.094513550037213e-06, "loss": 0.3399, "step": 8939 }, { "epoch": 0.6180435534047701, "grad_norm": 0.8074114148725854, "learning_rate": 8.094073857883652e-06, "loss": 0.414, "step": 8940 }, { "epoch": 0.6181126857932941, "grad_norm": 0.863404234786824, "learning_rate": 8.093634126951395e-06, "loss": 0.3441, "step": 8941 }, { "epoch": 0.6181818181818182, "grad_norm": 0.7357670727611505, "learning_rate": 8.093194357245955e-06, "loss": 0.3085, "step": 8942 }, { "epoch": 0.6182509505703422, "grad_norm": 1.5641706141125888, "learning_rate": 8.092754548772845e-06, "loss": 0.3547, "step": 8943 }, { "epoch": 0.6183200829588662, "grad_norm": 0.6932131863495402, "learning_rate": 8.092314701537577e-06, "loss": 0.409, "step": 8944 }, { "epoch": 0.6183892153473902, "grad_norm": 0.7745771056667561, "learning_rate": 8.091874815545661e-06, "loss": 0.3157, "step": 8945 }, { "epoch": 0.6184583477359142, "grad_norm": 0.8913579858039109, "learning_rate": 8.091434890802612e-06, "loss": 0.3488, "step": 8946 }, { "epoch": 0.6185274801244383, "grad_norm": 1.112729705016528, "learning_rate": 8.090994927313943e-06, "loss": 0.4089, "step": 8947 }, { "epoch": 0.6185966125129623, "grad_norm": 0.7319273079173926, "learning_rate": 8.09055492508517e-06, "loss": 0.387, "step": 8948 }, { "epoch": 0.6186657449014863, "grad_norm": 0.7354294384169564, "learning_rate": 8.090114884121808e-06, "loss": 0.3499, "step": 8949 }, { "epoch": 0.6187348772900104, "grad_norm": 0.9059939507971331, "learning_rate": 8.089674804429369e-06, "loss": 0.2763, "step": 8950 }, { "epoch": 0.6188040096785344, "grad_norm": 0.6828856349741974, "learning_rate": 8.08923468601337e-06, "loss": 0.3632, "step": 8951 }, { "epoch": 0.6188731420670585, "grad_norm": 0.786411876142158, "learning_rate": 8.088794528879325e-06, "loss": 0.343, "step": 8952 }, { "epoch": 0.6189422744555825, "grad_norm": 0.7259018840332337, "learning_rate": 8.088354333032755e-06, "loss": 0.2877, "step": 8953 }, { "epoch": 0.6190114068441065, "grad_norm": 0.6709135744799302, "learning_rate": 8.087914098479175e-06, "loss": 0.3645, "step": 8954 }, { "epoch": 0.6190805392326305, "grad_norm": 0.7174828109414169, "learning_rate": 8.087473825224102e-06, "loss": 0.3712, "step": 8955 }, { "epoch": 0.6191496716211545, "grad_norm": 1.1374354306618009, "learning_rate": 8.087033513273054e-06, "loss": 0.2653, "step": 8956 }, { "epoch": 0.6192188040096785, "grad_norm": 0.7967863975245181, "learning_rate": 8.08659316263155e-06, "loss": 0.357, "step": 8957 }, { "epoch": 0.6192879363982026, "grad_norm": 0.6500503144815729, "learning_rate": 8.08615277330511e-06, "loss": 0.4843, "step": 8958 }, { "epoch": 0.6193570687867266, "grad_norm": 0.7132574859913589, "learning_rate": 8.08571234529925e-06, "loss": 0.3252, "step": 8959 }, { "epoch": 0.6194262011752506, "grad_norm": 0.7910448351555418, "learning_rate": 8.085271878619495e-06, "loss": 0.3288, "step": 8960 }, { "epoch": 0.6194953335637746, "grad_norm": 0.6506475414448043, "learning_rate": 8.084831373271363e-06, "loss": 0.3118, "step": 8961 }, { "epoch": 0.6195644659522986, "grad_norm": 0.7345254725392774, "learning_rate": 8.084390829260374e-06, "loss": 0.4328, "step": 8962 }, { "epoch": 0.6196335983408227, "grad_norm": 0.654219685390054, "learning_rate": 8.083950246592049e-06, "loss": 0.2824, "step": 8963 }, { "epoch": 0.6197027307293467, "grad_norm": 0.6891805887263973, "learning_rate": 8.083509625271913e-06, "loss": 0.2548, "step": 8964 }, { "epoch": 0.6197718631178707, "grad_norm": 0.7844033643297421, "learning_rate": 8.083068965305486e-06, "loss": 0.4588, "step": 8965 }, { "epoch": 0.6198409955063947, "grad_norm": 0.7135662439501714, "learning_rate": 8.08262826669829e-06, "loss": 0.3319, "step": 8966 }, { "epoch": 0.6199101278949187, "grad_norm": 0.6920180809844706, "learning_rate": 8.082187529455853e-06, "loss": 0.3412, "step": 8967 }, { "epoch": 0.6199792602834427, "grad_norm": 0.6991821375494716, "learning_rate": 8.081746753583694e-06, "loss": 0.34, "step": 8968 }, { "epoch": 0.6200483926719668, "grad_norm": 0.7968480780919053, "learning_rate": 8.08130593908734e-06, "loss": 0.4254, "step": 8969 }, { "epoch": 0.6201175250604909, "grad_norm": 0.678901098110144, "learning_rate": 8.080865085972314e-06, "loss": 0.3015, "step": 8970 }, { "epoch": 0.6201866574490149, "grad_norm": 0.7427658238672956, "learning_rate": 8.080424194244142e-06, "loss": 0.2877, "step": 8971 }, { "epoch": 0.6202557898375389, "grad_norm": 0.6977722564706741, "learning_rate": 8.07998326390835e-06, "loss": 0.3812, "step": 8972 }, { "epoch": 0.620324922226063, "grad_norm": 1.088909823281779, "learning_rate": 8.079542294970463e-06, "loss": 0.3773, "step": 8973 }, { "epoch": 0.620394054614587, "grad_norm": 0.6435790178039327, "learning_rate": 8.079101287436012e-06, "loss": 0.3027, "step": 8974 }, { "epoch": 0.620463187003111, "grad_norm": 0.733345108315155, "learning_rate": 8.07866024131052e-06, "loss": 0.4045, "step": 8975 }, { "epoch": 0.620532319391635, "grad_norm": 0.7406848526440202, "learning_rate": 8.078219156599516e-06, "loss": 0.43, "step": 8976 }, { "epoch": 0.620601451780159, "grad_norm": 0.9896602506729488, "learning_rate": 8.077778033308528e-06, "loss": 0.3067, "step": 8977 }, { "epoch": 0.620670584168683, "grad_norm": 0.7524046728051097, "learning_rate": 8.077336871443085e-06, "loss": 0.319, "step": 8978 }, { "epoch": 0.620739716557207, "grad_norm": 0.6894499784247443, "learning_rate": 8.076895671008715e-06, "loss": 0.4024, "step": 8979 }, { "epoch": 0.6208088489457311, "grad_norm": 0.7629861032716764, "learning_rate": 8.07645443201095e-06, "loss": 0.3522, "step": 8980 }, { "epoch": 0.6208779813342551, "grad_norm": 0.6926752737528489, "learning_rate": 8.07601315445532e-06, "loss": 0.2973, "step": 8981 }, { "epoch": 0.6209471137227791, "grad_norm": 0.7982476952888273, "learning_rate": 8.075571838347351e-06, "loss": 0.3955, "step": 8982 }, { "epoch": 0.6210162461113031, "grad_norm": 0.7279347612101759, "learning_rate": 8.075130483692581e-06, "loss": 0.3907, "step": 8983 }, { "epoch": 0.6210853784998271, "grad_norm": 0.929302444356559, "learning_rate": 8.074689090496537e-06, "loss": 0.2764, "step": 8984 }, { "epoch": 0.6211545108883512, "grad_norm": 0.6941039994469166, "learning_rate": 8.074247658764752e-06, "loss": 0.2682, "step": 8985 }, { "epoch": 0.6212236432768752, "grad_norm": 0.716749715418506, "learning_rate": 8.07380618850276e-06, "loss": 0.37, "step": 8986 }, { "epoch": 0.6212927756653992, "grad_norm": 0.6372235023846582, "learning_rate": 8.073364679716091e-06, "loss": 0.3688, "step": 8987 }, { "epoch": 0.6213619080539232, "grad_norm": 0.6790250361966644, "learning_rate": 8.072923132410282e-06, "loss": 0.2968, "step": 8988 }, { "epoch": 0.6214310404424472, "grad_norm": 0.7459904461713474, "learning_rate": 8.072481546590865e-06, "loss": 0.3035, "step": 8989 }, { "epoch": 0.6215001728309714, "grad_norm": 0.776076933235328, "learning_rate": 8.072039922263374e-06, "loss": 0.4019, "step": 8990 }, { "epoch": 0.6215693052194954, "grad_norm": 0.7597649544440925, "learning_rate": 8.071598259433346e-06, "loss": 0.3407, "step": 8991 }, { "epoch": 0.6216384376080194, "grad_norm": 0.7426047319512442, "learning_rate": 8.071156558106314e-06, "loss": 0.2838, "step": 8992 }, { "epoch": 0.6217075699965434, "grad_norm": 1.4801826688808766, "learning_rate": 8.070714818287817e-06, "loss": 0.429, "step": 8993 }, { "epoch": 0.6217767023850674, "grad_norm": 0.66381062805938, "learning_rate": 8.070273039983388e-06, "loss": 0.3327, "step": 8994 }, { "epoch": 0.6218458347735915, "grad_norm": 0.6613666129599216, "learning_rate": 8.069831223198567e-06, "loss": 0.3407, "step": 8995 }, { "epoch": 0.6219149671621155, "grad_norm": 0.7940985105509494, "learning_rate": 8.069389367938889e-06, "loss": 0.3209, "step": 8996 }, { "epoch": 0.6219840995506395, "grad_norm": 1.384656753193058, "learning_rate": 8.068947474209893e-06, "loss": 0.3509, "step": 8997 }, { "epoch": 0.6220532319391635, "grad_norm": 0.6689723922740816, "learning_rate": 8.068505542017118e-06, "loss": 0.3377, "step": 8998 }, { "epoch": 0.6221223643276875, "grad_norm": 0.7022092354123981, "learning_rate": 8.068063571366102e-06, "loss": 0.3115, "step": 8999 }, { "epoch": 0.6221914967162115, "grad_norm": 0.7003884322969324, "learning_rate": 8.067621562262383e-06, "loss": 0.3334, "step": 9000 }, { "epoch": 0.6222606291047356, "grad_norm": 0.7348736278215839, "learning_rate": 8.067179514711502e-06, "loss": 0.4132, "step": 9001 }, { "epoch": 0.6223297614932596, "grad_norm": 0.7419191944467518, "learning_rate": 8.066737428719002e-06, "loss": 0.3481, "step": 9002 }, { "epoch": 0.6223988938817836, "grad_norm": 0.6447233691695906, "learning_rate": 8.066295304290419e-06, "loss": 0.2574, "step": 9003 }, { "epoch": 0.6224680262703076, "grad_norm": 0.7776468880172082, "learning_rate": 8.065853141431297e-06, "loss": 0.4426, "step": 9004 }, { "epoch": 0.6225371586588316, "grad_norm": 0.7778753325997729, "learning_rate": 8.065410940147177e-06, "loss": 0.33, "step": 9005 }, { "epoch": 0.6226062910473557, "grad_norm": 1.0423800265163725, "learning_rate": 8.0649687004436e-06, "loss": 0.3254, "step": 9006 }, { "epoch": 0.6226754234358797, "grad_norm": 0.6990578134635802, "learning_rate": 8.064526422326113e-06, "loss": 0.3208, "step": 9007 }, { "epoch": 0.6227445558244037, "grad_norm": 0.7034719155715231, "learning_rate": 8.064084105800254e-06, "loss": 0.4901, "step": 9008 }, { "epoch": 0.6228136882129277, "grad_norm": 0.7133463483700725, "learning_rate": 8.06364175087157e-06, "loss": 0.3123, "step": 9009 }, { "epoch": 0.6228828206014517, "grad_norm": 0.6778167760842916, "learning_rate": 8.063199357545604e-06, "loss": 0.3072, "step": 9010 }, { "epoch": 0.6229519529899759, "grad_norm": 0.6891392274414481, "learning_rate": 8.062756925827899e-06, "loss": 0.3246, "step": 9011 }, { "epoch": 0.6230210853784999, "grad_norm": 0.7273279976085697, "learning_rate": 8.062314455724003e-06, "loss": 0.4046, "step": 9012 }, { "epoch": 0.6230902177670239, "grad_norm": 0.9946480376198692, "learning_rate": 8.06187194723946e-06, "loss": 0.2895, "step": 9013 }, { "epoch": 0.6231593501555479, "grad_norm": 0.7149630413658415, "learning_rate": 8.061429400379816e-06, "loss": 0.2705, "step": 9014 }, { "epoch": 0.6232284825440719, "grad_norm": 0.715035050435029, "learning_rate": 8.060986815150618e-06, "loss": 0.3916, "step": 9015 }, { "epoch": 0.623297614932596, "grad_norm": 0.6910714288676929, "learning_rate": 8.060544191557413e-06, "loss": 0.3351, "step": 9016 }, { "epoch": 0.62336674732112, "grad_norm": 0.6801389263567832, "learning_rate": 8.060101529605748e-06, "loss": 0.3572, "step": 9017 }, { "epoch": 0.623435879709644, "grad_norm": 0.7843052341656507, "learning_rate": 8.059658829301172e-06, "loss": 0.2896, "step": 9018 }, { "epoch": 0.623505012098168, "grad_norm": 0.8157807908265599, "learning_rate": 8.059216090649232e-06, "loss": 0.4184, "step": 9019 }, { "epoch": 0.623574144486692, "grad_norm": 0.6389236732686853, "learning_rate": 8.05877331365548e-06, "loss": 0.292, "step": 9020 }, { "epoch": 0.623643276875216, "grad_norm": 0.6730071330579862, "learning_rate": 8.058330498325459e-06, "loss": 0.2934, "step": 9021 }, { "epoch": 0.62371240926374, "grad_norm": 0.7047098528387249, "learning_rate": 8.057887644664725e-06, "loss": 0.3975, "step": 9022 }, { "epoch": 0.6237815416522641, "grad_norm": 0.6819639759191204, "learning_rate": 8.057444752678827e-06, "loss": 0.2986, "step": 9023 }, { "epoch": 0.6238506740407881, "grad_norm": 1.458641645699343, "learning_rate": 8.057001822373316e-06, "loss": 0.3502, "step": 9024 }, { "epoch": 0.6239198064293121, "grad_norm": 0.7143287224913552, "learning_rate": 8.05655885375374e-06, "loss": 0.3592, "step": 9025 }, { "epoch": 0.6239889388178361, "grad_norm": 0.6646194919754611, "learning_rate": 8.056115846825656e-06, "loss": 0.4154, "step": 9026 }, { "epoch": 0.6240580712063601, "grad_norm": 0.9668469772996161, "learning_rate": 8.055672801594613e-06, "loss": 0.3179, "step": 9027 }, { "epoch": 0.6241272035948842, "grad_norm": 0.6496001103403517, "learning_rate": 8.055229718066163e-06, "loss": 0.3147, "step": 9028 }, { "epoch": 0.6241963359834082, "grad_norm": 0.7271210148934184, "learning_rate": 8.054786596245863e-06, "loss": 0.3859, "step": 9029 }, { "epoch": 0.6242654683719322, "grad_norm": 0.679244359063165, "learning_rate": 8.054343436139263e-06, "loss": 0.3675, "step": 9030 }, { "epoch": 0.6243346007604563, "grad_norm": 0.6864887519890236, "learning_rate": 8.05390023775192e-06, "loss": 0.3028, "step": 9031 }, { "epoch": 0.6244037331489803, "grad_norm": 0.6153030734225728, "learning_rate": 8.053457001089387e-06, "loss": 0.3047, "step": 9032 }, { "epoch": 0.6244728655375044, "grad_norm": 0.6679502100103932, "learning_rate": 8.05301372615722e-06, "loss": 0.371, "step": 9033 }, { "epoch": 0.6245419979260284, "grad_norm": 0.6997956951947742, "learning_rate": 8.052570412960972e-06, "loss": 0.3522, "step": 9034 }, { "epoch": 0.6246111303145524, "grad_norm": 0.6763128808802444, "learning_rate": 8.052127061506204e-06, "loss": 0.3014, "step": 9035 }, { "epoch": 0.6246802627030764, "grad_norm": 0.6698601746151784, "learning_rate": 8.05168367179847e-06, "loss": 0.3541, "step": 9036 }, { "epoch": 0.6247493950916004, "grad_norm": 0.6219483135444571, "learning_rate": 8.051240243843326e-06, "loss": 0.3236, "step": 9037 }, { "epoch": 0.6248185274801245, "grad_norm": 0.6646811783884826, "learning_rate": 8.05079677764633e-06, "loss": 0.2706, "step": 9038 }, { "epoch": 0.6248876598686485, "grad_norm": 0.7263356256096875, "learning_rate": 8.050353273213043e-06, "loss": 0.2722, "step": 9039 }, { "epoch": 0.6249567922571725, "grad_norm": 0.7225926094121468, "learning_rate": 8.049909730549019e-06, "loss": 0.3941, "step": 9040 }, { "epoch": 0.6250259246456965, "grad_norm": 0.7109173211457326, "learning_rate": 8.049466149659822e-06, "loss": 0.3018, "step": 9041 }, { "epoch": 0.6250950570342205, "grad_norm": 0.6677362935512704, "learning_rate": 8.049022530551006e-06, "loss": 0.2699, "step": 9042 }, { "epoch": 0.6251641894227445, "grad_norm": 0.7966406977113133, "learning_rate": 8.048578873228135e-06, "loss": 0.317, "step": 9043 }, { "epoch": 0.6252333218112686, "grad_norm": 0.6559469127432895, "learning_rate": 8.04813517769677e-06, "loss": 0.3351, "step": 9044 }, { "epoch": 0.6253024541997926, "grad_norm": 0.697158978217304, "learning_rate": 8.047691443962468e-06, "loss": 0.3935, "step": 9045 }, { "epoch": 0.6253715865883166, "grad_norm": 0.6396708180994112, "learning_rate": 8.047247672030792e-06, "loss": 0.2714, "step": 9046 }, { "epoch": 0.6254407189768406, "grad_norm": 1.109627831322626, "learning_rate": 8.046803861907307e-06, "loss": 0.4343, "step": 9047 }, { "epoch": 0.6255098513653646, "grad_norm": 0.7294708941462895, "learning_rate": 8.046360013597571e-06, "loss": 0.3512, "step": 9048 }, { "epoch": 0.6255789837538887, "grad_norm": 0.7609091236738695, "learning_rate": 8.045916127107149e-06, "loss": 0.247, "step": 9049 }, { "epoch": 0.6256481161424127, "grad_norm": 0.7645916500932881, "learning_rate": 8.045472202441604e-06, "loss": 0.3253, "step": 9050 }, { "epoch": 0.6257172485309368, "grad_norm": 0.702724831813369, "learning_rate": 8.0450282396065e-06, "loss": 0.3981, "step": 9051 }, { "epoch": 0.6257863809194608, "grad_norm": 0.6605394908052794, "learning_rate": 8.044584238607401e-06, "loss": 0.336, "step": 9052 }, { "epoch": 0.6258555133079848, "grad_norm": 0.6624619304161117, "learning_rate": 8.044140199449873e-06, "loss": 0.2846, "step": 9053 }, { "epoch": 0.6259246456965089, "grad_norm": 0.7378973473952641, "learning_rate": 8.043696122139479e-06, "loss": 0.4134, "step": 9054 }, { "epoch": 0.6259937780850329, "grad_norm": 0.6939266082049103, "learning_rate": 8.043252006681786e-06, "loss": 0.2988, "step": 9055 }, { "epoch": 0.6260629104735569, "grad_norm": 0.6668530980147583, "learning_rate": 8.04280785308236e-06, "loss": 0.3138, "step": 9056 }, { "epoch": 0.6261320428620809, "grad_norm": 0.648706862249787, "learning_rate": 8.042363661346767e-06, "loss": 0.3297, "step": 9057 }, { "epoch": 0.6262011752506049, "grad_norm": 0.7141675538628083, "learning_rate": 8.041919431480576e-06, "loss": 0.4315, "step": 9058 }, { "epoch": 0.6262703076391289, "grad_norm": 0.8096697464906458, "learning_rate": 8.041475163489354e-06, "loss": 0.343, "step": 9059 }, { "epoch": 0.626339440027653, "grad_norm": 0.6546940801894832, "learning_rate": 8.041030857378667e-06, "loss": 0.3056, "step": 9060 }, { "epoch": 0.626408572416177, "grad_norm": 0.8041508826353748, "learning_rate": 8.040586513154085e-06, "loss": 0.3356, "step": 9061 }, { "epoch": 0.626477704804701, "grad_norm": 0.8046667285032014, "learning_rate": 8.040142130821179e-06, "loss": 0.4428, "step": 9062 }, { "epoch": 0.626546837193225, "grad_norm": 0.6933725491998982, "learning_rate": 8.039697710385515e-06, "loss": 0.2597, "step": 9063 }, { "epoch": 0.626615969581749, "grad_norm": 0.7186714102933005, "learning_rate": 8.039253251852666e-06, "loss": 0.3041, "step": 9064 }, { "epoch": 0.626685101970273, "grad_norm": 0.653605343490922, "learning_rate": 8.0388087552282e-06, "loss": 0.3789, "step": 9065 }, { "epoch": 0.6267542343587971, "grad_norm": 0.670489788334486, "learning_rate": 8.038364220517692e-06, "loss": 0.3396, "step": 9066 }, { "epoch": 0.6268233667473211, "grad_norm": 0.8946998401307289, "learning_rate": 8.037919647726708e-06, "loss": 0.2918, "step": 9067 }, { "epoch": 0.6268924991358451, "grad_norm": 0.9191222354253358, "learning_rate": 8.037475036860823e-06, "loss": 0.3828, "step": 9068 }, { "epoch": 0.6269616315243691, "grad_norm": 1.0069158606827922, "learning_rate": 8.03703038792561e-06, "loss": 0.4028, "step": 9069 }, { "epoch": 0.6270307639128931, "grad_norm": 0.6575329535199396, "learning_rate": 8.036585700926641e-06, "loss": 0.2761, "step": 9070 }, { "epoch": 0.6270998963014172, "grad_norm": 0.7597519158152876, "learning_rate": 8.036140975869488e-06, "loss": 0.2782, "step": 9071 }, { "epoch": 0.6271690286899413, "grad_norm": 0.9619733537812277, "learning_rate": 8.035696212759727e-06, "loss": 0.4326, "step": 9072 }, { "epoch": 0.6272381610784653, "grad_norm": 0.7688167782585681, "learning_rate": 8.03525141160293e-06, "loss": 0.4088, "step": 9073 }, { "epoch": 0.6273072934669893, "grad_norm": 0.6485334447064518, "learning_rate": 8.034806572404674e-06, "loss": 0.2557, "step": 9074 }, { "epoch": 0.6273764258555133, "grad_norm": 0.680776543514422, "learning_rate": 8.034361695170533e-06, "loss": 0.3022, "step": 9075 }, { "epoch": 0.6274455582440374, "grad_norm": 0.7069598209494734, "learning_rate": 8.033916779906086e-06, "loss": 0.4129, "step": 9076 }, { "epoch": 0.6275146906325614, "grad_norm": 0.7625389700979446, "learning_rate": 8.033471826616905e-06, "loss": 0.2775, "step": 9077 }, { "epoch": 0.6275838230210854, "grad_norm": 0.7570560138169887, "learning_rate": 8.033026835308567e-06, "loss": 0.2826, "step": 9078 }, { "epoch": 0.6276529554096094, "grad_norm": 0.7509256997398043, "learning_rate": 8.03258180598665e-06, "loss": 0.3459, "step": 9079 }, { "epoch": 0.6277220877981334, "grad_norm": 0.9606750137325494, "learning_rate": 8.032136738656732e-06, "loss": 0.3466, "step": 9080 }, { "epoch": 0.6277912201866575, "grad_norm": 0.757848182619748, "learning_rate": 8.031691633324395e-06, "loss": 0.3146, "step": 9081 }, { "epoch": 0.6278603525751815, "grad_norm": 0.7181379894455852, "learning_rate": 8.03124648999521e-06, "loss": 0.3299, "step": 9082 }, { "epoch": 0.6279294849637055, "grad_norm": 0.7739720707649147, "learning_rate": 8.030801308674761e-06, "loss": 0.4232, "step": 9083 }, { "epoch": 0.6279986173522295, "grad_norm": 0.9847427761300344, "learning_rate": 8.030356089368625e-06, "loss": 0.3334, "step": 9084 }, { "epoch": 0.6280677497407535, "grad_norm": 0.6841753802030967, "learning_rate": 8.029910832082384e-06, "loss": 0.3306, "step": 9085 }, { "epoch": 0.6281368821292775, "grad_norm": 0.7171705643118765, "learning_rate": 8.029465536821617e-06, "loss": 0.3608, "step": 9086 }, { "epoch": 0.6282060145178016, "grad_norm": 0.7464352458292767, "learning_rate": 8.029020203591908e-06, "loss": 0.4012, "step": 9087 }, { "epoch": 0.6282751469063256, "grad_norm": 0.8110380906378999, "learning_rate": 8.028574832398834e-06, "loss": 0.2982, "step": 9088 }, { "epoch": 0.6283442792948496, "grad_norm": 0.7662449690064896, "learning_rate": 8.02812942324798e-06, "loss": 0.3501, "step": 9089 }, { "epoch": 0.6284134116833736, "grad_norm": 0.7586097273851357, "learning_rate": 8.027683976144927e-06, "loss": 0.3912, "step": 9090 }, { "epoch": 0.6284825440718976, "grad_norm": 0.7944438161684807, "learning_rate": 8.02723849109526e-06, "loss": 0.3402, "step": 9091 }, { "epoch": 0.6285516764604218, "grad_norm": 0.6609471235345059, "learning_rate": 8.02679296810456e-06, "loss": 0.3314, "step": 9092 }, { "epoch": 0.6286208088489458, "grad_norm": 0.7307408835152481, "learning_rate": 8.026347407178413e-06, "loss": 0.3364, "step": 9093 }, { "epoch": 0.6286899412374698, "grad_norm": 0.7058553494011682, "learning_rate": 8.0259018083224e-06, "loss": 0.4351, "step": 9094 }, { "epoch": 0.6287590736259938, "grad_norm": 0.6485450480059745, "learning_rate": 8.025456171542108e-06, "loss": 0.2952, "step": 9095 }, { "epoch": 0.6288282060145178, "grad_norm": 0.6912935518747759, "learning_rate": 8.025010496843124e-06, "loss": 0.2925, "step": 9096 }, { "epoch": 0.6288973384030419, "grad_norm": 0.9503545286604907, "learning_rate": 8.02456478423103e-06, "loss": 0.4636, "step": 9097 }, { "epoch": 0.6289664707915659, "grad_norm": 0.6897389445593844, "learning_rate": 8.024119033711415e-06, "loss": 0.3722, "step": 9098 }, { "epoch": 0.6290356031800899, "grad_norm": 0.6918188284060117, "learning_rate": 8.023673245289864e-06, "loss": 0.2406, "step": 9099 }, { "epoch": 0.6291047355686139, "grad_norm": 0.7989874662777243, "learning_rate": 8.023227418971964e-06, "loss": 0.3333, "step": 9100 }, { "epoch": 0.6291738679571379, "grad_norm": 0.758959265126224, "learning_rate": 8.022781554763304e-06, "loss": 0.3843, "step": 9101 }, { "epoch": 0.6292430003456619, "grad_norm": 0.7286868673192326, "learning_rate": 8.022335652669472e-06, "loss": 0.3179, "step": 9102 }, { "epoch": 0.629312132734186, "grad_norm": 0.7393465665231014, "learning_rate": 8.021889712696058e-06, "loss": 0.3116, "step": 9103 }, { "epoch": 0.62938126512271, "grad_norm": 0.7112868333082436, "learning_rate": 8.021443734848648e-06, "loss": 0.3649, "step": 9104 }, { "epoch": 0.629450397511234, "grad_norm": 0.7690272187059197, "learning_rate": 8.02099771913283e-06, "loss": 0.322, "step": 9105 }, { "epoch": 0.629519529899758, "grad_norm": 0.7088799373517529, "learning_rate": 8.0205516655542e-06, "loss": 0.3346, "step": 9106 }, { "epoch": 0.629588662288282, "grad_norm": 0.7229626744653389, "learning_rate": 8.020105574118346e-06, "loss": 0.3352, "step": 9107 }, { "epoch": 0.629657794676806, "grad_norm": 0.8031417962466373, "learning_rate": 8.019659444830857e-06, "loss": 0.4112, "step": 9108 }, { "epoch": 0.6297269270653301, "grad_norm": 0.6627970457260874, "learning_rate": 8.019213277697324e-06, "loss": 0.2963, "step": 9109 }, { "epoch": 0.6297960594538541, "grad_norm": 0.6842740477175934, "learning_rate": 8.018767072723342e-06, "loss": 0.3032, "step": 9110 }, { "epoch": 0.6298651918423781, "grad_norm": 0.7027791780651919, "learning_rate": 8.018320829914503e-06, "loss": 0.3378, "step": 9111 }, { "epoch": 0.6299343242309022, "grad_norm": 0.7669037093839346, "learning_rate": 8.017874549276396e-06, "loss": 0.3686, "step": 9112 }, { "epoch": 0.6300034566194262, "grad_norm": 0.6344785440922489, "learning_rate": 8.01742823081462e-06, "loss": 0.269, "step": 9113 }, { "epoch": 0.6300725890079503, "grad_norm": 0.6596652426834696, "learning_rate": 8.016981874534765e-06, "loss": 0.2442, "step": 9114 }, { "epoch": 0.6301417213964743, "grad_norm": 0.7005008042284332, "learning_rate": 8.016535480442427e-06, "loss": 0.415, "step": 9115 }, { "epoch": 0.6302108537849983, "grad_norm": 0.7501277002239214, "learning_rate": 8.016089048543198e-06, "loss": 0.4085, "step": 9116 }, { "epoch": 0.6302799861735223, "grad_norm": 0.8341856136535718, "learning_rate": 8.015642578842677e-06, "loss": 0.3669, "step": 9117 }, { "epoch": 0.6303491185620463, "grad_norm": 0.7131846179327386, "learning_rate": 8.015196071346457e-06, "loss": 0.3882, "step": 9118 }, { "epoch": 0.6304182509505704, "grad_norm": 0.6660203498558995, "learning_rate": 8.014749526060134e-06, "loss": 0.3669, "step": 9119 }, { "epoch": 0.6304873833390944, "grad_norm": 1.131053982464013, "learning_rate": 8.01430294298931e-06, "loss": 0.2938, "step": 9120 }, { "epoch": 0.6305565157276184, "grad_norm": 0.845161269188566, "learning_rate": 8.013856322139573e-06, "loss": 0.254, "step": 9121 }, { "epoch": 0.6306256481161424, "grad_norm": 0.6753666176548138, "learning_rate": 8.013409663516529e-06, "loss": 0.425, "step": 9122 }, { "epoch": 0.6306947805046664, "grad_norm": 0.7064316172223334, "learning_rate": 8.01296296712577e-06, "loss": 0.428, "step": 9123 }, { "epoch": 0.6307639128931904, "grad_norm": 0.665313543644682, "learning_rate": 8.012516232972897e-06, "loss": 0.3287, "step": 9124 }, { "epoch": 0.6308330452817145, "grad_norm": 0.7825662478497888, "learning_rate": 8.012069461063512e-06, "loss": 0.3557, "step": 9125 }, { "epoch": 0.6309021776702385, "grad_norm": 0.658923248821014, "learning_rate": 8.01162265140321e-06, "loss": 0.4081, "step": 9126 }, { "epoch": 0.6309713100587625, "grad_norm": 0.7832424603963016, "learning_rate": 8.011175803997592e-06, "loss": 0.3076, "step": 9127 }, { "epoch": 0.6310404424472865, "grad_norm": 0.7310273190148231, "learning_rate": 8.01072891885226e-06, "loss": 0.3043, "step": 9128 }, { "epoch": 0.6311095748358105, "grad_norm": 0.8690031630098266, "learning_rate": 8.010281995972813e-06, "loss": 0.3779, "step": 9129 }, { "epoch": 0.6311787072243346, "grad_norm": 0.7884040188493213, "learning_rate": 8.009835035364854e-06, "loss": 0.3526, "step": 9130 }, { "epoch": 0.6312478396128586, "grad_norm": 0.6845204520088486, "learning_rate": 8.009388037033983e-06, "loss": 0.275, "step": 9131 }, { "epoch": 0.6313169720013826, "grad_norm": 0.7653138074633492, "learning_rate": 8.008941000985803e-06, "loss": 0.3421, "step": 9132 }, { "epoch": 0.6313861043899067, "grad_norm": 0.8682222452977949, "learning_rate": 8.008493927225919e-06, "loss": 0.4315, "step": 9133 }, { "epoch": 0.6314552367784307, "grad_norm": 0.9852021329696354, "learning_rate": 8.008046815759933e-06, "loss": 0.3109, "step": 9134 }, { "epoch": 0.6315243691669548, "grad_norm": 0.7403641454677283, "learning_rate": 8.007599666593447e-06, "loss": 0.267, "step": 9135 }, { "epoch": 0.6315935015554788, "grad_norm": 1.1149457622361647, "learning_rate": 8.007152479732067e-06, "loss": 0.3924, "step": 9136 }, { "epoch": 0.6316626339440028, "grad_norm": 0.7529298587515918, "learning_rate": 8.006705255181396e-06, "loss": 0.404, "step": 9137 }, { "epoch": 0.6317317663325268, "grad_norm": 0.8511642484023321, "learning_rate": 8.006257992947041e-06, "loss": 0.274, "step": 9138 }, { "epoch": 0.6318008987210508, "grad_norm": 0.8330480237399555, "learning_rate": 8.005810693034608e-06, "loss": 0.3313, "step": 9139 }, { "epoch": 0.6318700311095748, "grad_norm": 0.7046733552699552, "learning_rate": 8.0053633554497e-06, "loss": 0.4228, "step": 9140 }, { "epoch": 0.6319391634980989, "grad_norm": 0.7878630493392677, "learning_rate": 8.004915980197927e-06, "loss": 0.3428, "step": 9141 }, { "epoch": 0.6320082958866229, "grad_norm": 0.7758161810994516, "learning_rate": 8.004468567284895e-06, "loss": 0.2643, "step": 9142 }, { "epoch": 0.6320774282751469, "grad_norm": 0.8405486015830054, "learning_rate": 8.004021116716211e-06, "loss": 0.346, "step": 9143 }, { "epoch": 0.6321465606636709, "grad_norm": 3.118952506952597, "learning_rate": 8.003573628497483e-06, "loss": 0.3895, "step": 9144 }, { "epoch": 0.6322156930521949, "grad_norm": 1.0028706338482583, "learning_rate": 8.00312610263432e-06, "loss": 0.3078, "step": 9145 }, { "epoch": 0.632284825440719, "grad_norm": 0.7619100748963425, "learning_rate": 8.00267853913233e-06, "loss": 0.2622, "step": 9146 }, { "epoch": 0.632353957829243, "grad_norm": 0.9091566013107413, "learning_rate": 8.002230937997123e-06, "loss": 0.3953, "step": 9147 }, { "epoch": 0.632423090217767, "grad_norm": 0.8069633768055462, "learning_rate": 8.00178329923431e-06, "loss": 0.3417, "step": 9148 }, { "epoch": 0.632492222606291, "grad_norm": 0.9176316387694494, "learning_rate": 8.0013356228495e-06, "loss": 0.3041, "step": 9149 }, { "epoch": 0.632561354994815, "grad_norm": 0.7062920969086798, "learning_rate": 8.000887908848303e-06, "loss": 0.3014, "step": 9150 }, { "epoch": 0.632630487383339, "grad_norm": 0.7718722299268134, "learning_rate": 8.00044015723633e-06, "loss": 0.4085, "step": 9151 }, { "epoch": 0.6326996197718631, "grad_norm": 1.3614787149657304, "learning_rate": 7.999992368019198e-06, "loss": 0.2872, "step": 9152 }, { "epoch": 0.6327687521603872, "grad_norm": 0.6332403057782166, "learning_rate": 7.999544541202514e-06, "loss": 0.2376, "step": 9153 }, { "epoch": 0.6328378845489112, "grad_norm": 0.7255101916341746, "learning_rate": 7.999096676791892e-06, "loss": 0.3707, "step": 9154 }, { "epoch": 0.6329070169374352, "grad_norm": 0.7197344431351709, "learning_rate": 7.998648774792944e-06, "loss": 0.386, "step": 9155 }, { "epoch": 0.6329761493259592, "grad_norm": 0.7574616009839249, "learning_rate": 7.998200835211285e-06, "loss": 0.2525, "step": 9156 }, { "epoch": 0.6330452817144833, "grad_norm": 0.7931541239091062, "learning_rate": 7.997752858052531e-06, "loss": 0.3741, "step": 9157 }, { "epoch": 0.6331144141030073, "grad_norm": 0.6967416189662772, "learning_rate": 7.997304843322294e-06, "loss": 0.4254, "step": 9158 }, { "epoch": 0.6331835464915313, "grad_norm": 1.3420977207685387, "learning_rate": 7.996856791026188e-06, "loss": 0.2845, "step": 9159 }, { "epoch": 0.6332526788800553, "grad_norm": 0.6294722647182627, "learning_rate": 7.996408701169832e-06, "loss": 0.3142, "step": 9160 }, { "epoch": 0.6333218112685793, "grad_norm": 0.665788657304387, "learning_rate": 7.99596057375884e-06, "loss": 0.2999, "step": 9161 }, { "epoch": 0.6333909436571034, "grad_norm": 0.7228801601921986, "learning_rate": 7.995512408798828e-06, "loss": 0.4091, "step": 9162 }, { "epoch": 0.6334600760456274, "grad_norm": 0.9513776593341178, "learning_rate": 7.995064206295415e-06, "loss": 0.2991, "step": 9163 }, { "epoch": 0.6335292084341514, "grad_norm": 0.7242973568228263, "learning_rate": 7.994615966254216e-06, "loss": 0.314, "step": 9164 }, { "epoch": 0.6335983408226754, "grad_norm": 0.8667349659813421, "learning_rate": 7.994167688680851e-06, "loss": 0.4085, "step": 9165 }, { "epoch": 0.6336674732111994, "grad_norm": 0.878965018697641, "learning_rate": 7.993719373580937e-06, "loss": 0.3016, "step": 9166 }, { "epoch": 0.6337366055997234, "grad_norm": 0.7336330292863106, "learning_rate": 7.993271020960092e-06, "loss": 0.3106, "step": 9167 }, { "epoch": 0.6338057379882475, "grad_norm": 0.6698178039130005, "learning_rate": 7.992822630823938e-06, "loss": 0.409, "step": 9168 }, { "epoch": 0.6338748703767715, "grad_norm": 0.6900433184690207, "learning_rate": 7.992374203178093e-06, "loss": 0.4433, "step": 9169 }, { "epoch": 0.6339440027652955, "grad_norm": 0.805681468999864, "learning_rate": 7.991925738028176e-06, "loss": 0.2978, "step": 9170 }, { "epoch": 0.6340131351538195, "grad_norm": 0.7763424068635502, "learning_rate": 7.99147723537981e-06, "loss": 0.3632, "step": 9171 }, { "epoch": 0.6340822675423435, "grad_norm": 0.8062887259086593, "learning_rate": 7.991028695238616e-06, "loss": 0.4099, "step": 9172 }, { "epoch": 0.6341513999308677, "grad_norm": 0.6906861508461262, "learning_rate": 7.990580117610214e-06, "loss": 0.3804, "step": 9173 }, { "epoch": 0.6342205323193917, "grad_norm": 0.7273326957379477, "learning_rate": 7.990131502500229e-06, "loss": 0.2801, "step": 9174 }, { "epoch": 0.6342896647079157, "grad_norm": 0.7669896627977217, "learning_rate": 7.98968284991428e-06, "loss": 0.3005, "step": 9175 }, { "epoch": 0.6343587970964397, "grad_norm": 0.8009610652684649, "learning_rate": 7.989234159857993e-06, "loss": 0.4141, "step": 9176 }, { "epoch": 0.6344279294849637, "grad_norm": 1.5449895074296687, "learning_rate": 7.98878543233699e-06, "loss": 0.2861, "step": 9177 }, { "epoch": 0.6344970618734878, "grad_norm": 0.677793832666429, "learning_rate": 7.988336667356896e-06, "loss": 0.2909, "step": 9178 }, { "epoch": 0.6345661942620118, "grad_norm": 0.7676293915365985, "learning_rate": 7.987887864923331e-06, "loss": 0.4496, "step": 9179 }, { "epoch": 0.6346353266505358, "grad_norm": 0.8925139448880234, "learning_rate": 7.987439025041927e-06, "loss": 0.4433, "step": 9180 }, { "epoch": 0.6347044590390598, "grad_norm": 0.7390613070093695, "learning_rate": 7.986990147718307e-06, "loss": 0.2869, "step": 9181 }, { "epoch": 0.6347735914275838, "grad_norm": 0.7790113386194666, "learning_rate": 7.986541232958095e-06, "loss": 0.3483, "step": 9182 }, { "epoch": 0.6348427238161078, "grad_norm": 0.7085830268922585, "learning_rate": 7.986092280766917e-06, "loss": 0.4994, "step": 9183 }, { "epoch": 0.6349118562046319, "grad_norm": 0.734161573304561, "learning_rate": 7.985643291150404e-06, "loss": 0.2989, "step": 9184 }, { "epoch": 0.6349809885931559, "grad_norm": 0.616763504524844, "learning_rate": 7.985194264114178e-06, "loss": 0.2878, "step": 9185 }, { "epoch": 0.6350501209816799, "grad_norm": 0.7557107216124909, "learning_rate": 7.98474519966387e-06, "loss": 0.3638, "step": 9186 }, { "epoch": 0.6351192533702039, "grad_norm": 0.6437135213612839, "learning_rate": 7.984296097805107e-06, "loss": 0.3863, "step": 9187 }, { "epoch": 0.6351883857587279, "grad_norm": 0.7159768974277952, "learning_rate": 7.983846958543519e-06, "loss": 0.3271, "step": 9188 }, { "epoch": 0.635257518147252, "grad_norm": 0.64678565397762, "learning_rate": 7.983397781884734e-06, "loss": 0.3389, "step": 9189 }, { "epoch": 0.635326650535776, "grad_norm": 0.7254354258972515, "learning_rate": 7.982948567834382e-06, "loss": 0.434, "step": 9190 }, { "epoch": 0.6353957829243, "grad_norm": 0.7189080767804572, "learning_rate": 7.982499316398094e-06, "loss": 0.3704, "step": 9191 }, { "epoch": 0.635464915312824, "grad_norm": 0.6654143220424615, "learning_rate": 7.982050027581498e-06, "loss": 0.3332, "step": 9192 }, { "epoch": 0.6355340477013481, "grad_norm": 0.7577474273974095, "learning_rate": 7.981600701390226e-06, "loss": 0.3565, "step": 9193 }, { "epoch": 0.6356031800898722, "grad_norm": 0.8019829651994869, "learning_rate": 7.981151337829911e-06, "loss": 0.4908, "step": 9194 }, { "epoch": 0.6356723124783962, "grad_norm": 0.6536532600264815, "learning_rate": 7.980701936906185e-06, "loss": 0.3504, "step": 9195 }, { "epoch": 0.6357414448669202, "grad_norm": 0.6487643609492237, "learning_rate": 7.980252498624678e-06, "loss": 0.3357, "step": 9196 }, { "epoch": 0.6358105772554442, "grad_norm": 0.6492299332723238, "learning_rate": 7.979803022991025e-06, "loss": 0.356, "step": 9197 }, { "epoch": 0.6358797096439682, "grad_norm": 1.1570204042564969, "learning_rate": 7.979353510010858e-06, "loss": 0.3693, "step": 9198 }, { "epoch": 0.6359488420324922, "grad_norm": 0.672496256129305, "learning_rate": 7.978903959689814e-06, "loss": 0.2736, "step": 9199 }, { "epoch": 0.6360179744210163, "grad_norm": 0.7157230191279577, "learning_rate": 7.978454372033524e-06, "loss": 0.3745, "step": 9200 }, { "epoch": 0.6360871068095403, "grad_norm": 0.7670826188254217, "learning_rate": 7.978004747047623e-06, "loss": 0.4125, "step": 9201 }, { "epoch": 0.6361562391980643, "grad_norm": 0.7814076376327244, "learning_rate": 7.977555084737747e-06, "loss": 0.2796, "step": 9202 }, { "epoch": 0.6362253715865883, "grad_norm": 1.331123020140714, "learning_rate": 7.977105385109532e-06, "loss": 0.2984, "step": 9203 }, { "epoch": 0.6362945039751123, "grad_norm": 0.6982729279819161, "learning_rate": 7.976655648168615e-06, "loss": 0.3373, "step": 9204 }, { "epoch": 0.6363636363636364, "grad_norm": 0.7694248390462894, "learning_rate": 7.97620587392063e-06, "loss": 0.4157, "step": 9205 }, { "epoch": 0.6364327687521604, "grad_norm": 0.6787625855182323, "learning_rate": 7.975756062371216e-06, "loss": 0.33, "step": 9206 }, { "epoch": 0.6365019011406844, "grad_norm": 0.7574591784035137, "learning_rate": 7.97530621352601e-06, "loss": 0.3484, "step": 9207 }, { "epoch": 0.6365710335292084, "grad_norm": 1.0219173979148632, "learning_rate": 7.97485632739065e-06, "loss": 0.4677, "step": 9208 }, { "epoch": 0.6366401659177324, "grad_norm": 0.7604384097607063, "learning_rate": 7.974406403970776e-06, "loss": 0.3273, "step": 9209 }, { "epoch": 0.6367092983062564, "grad_norm": 0.7599067850686446, "learning_rate": 7.973956443272025e-06, "loss": 0.2844, "step": 9210 }, { "epoch": 0.6367784306947805, "grad_norm": 0.7414473882716212, "learning_rate": 7.973506445300037e-06, "loss": 0.3966, "step": 9211 }, { "epoch": 0.6368475630833045, "grad_norm": 0.6686790056904878, "learning_rate": 7.973056410060454e-06, "loss": 0.4228, "step": 9212 }, { "epoch": 0.6369166954718285, "grad_norm": 0.6570953058282893, "learning_rate": 7.972606337558913e-06, "loss": 0.3217, "step": 9213 }, { "epoch": 0.6369858278603526, "grad_norm": 0.7337336019653232, "learning_rate": 7.972156227801055e-06, "loss": 0.3001, "step": 9214 }, { "epoch": 0.6370549602488766, "grad_norm": 0.7463545480448582, "learning_rate": 7.971706080792525e-06, "loss": 0.472, "step": 9215 }, { "epoch": 0.6371240926374007, "grad_norm": 0.7216301167316778, "learning_rate": 7.971255896538961e-06, "loss": 0.3069, "step": 9216 }, { "epoch": 0.6371932250259247, "grad_norm": 0.7821684335508757, "learning_rate": 7.97080567504601e-06, "loss": 0.3253, "step": 9217 }, { "epoch": 0.6372623574144487, "grad_norm": 0.7234048389656471, "learning_rate": 7.970355416319309e-06, "loss": 0.3667, "step": 9218 }, { "epoch": 0.6373314898029727, "grad_norm": 1.0140831255365141, "learning_rate": 7.969905120364503e-06, "loss": 0.4536, "step": 9219 }, { "epoch": 0.6374006221914967, "grad_norm": 0.6754734090304899, "learning_rate": 7.969454787187237e-06, "loss": 0.2822, "step": 9220 }, { "epoch": 0.6374697545800208, "grad_norm": 0.6365040331998775, "learning_rate": 7.969004416793155e-06, "loss": 0.3203, "step": 9221 }, { "epoch": 0.6375388869685448, "grad_norm": 0.8864799936811479, "learning_rate": 7.968554009187901e-06, "loss": 0.4058, "step": 9222 }, { "epoch": 0.6376080193570688, "grad_norm": 0.6708686884342251, "learning_rate": 7.968103564377118e-06, "loss": 0.3072, "step": 9223 }, { "epoch": 0.6376771517455928, "grad_norm": 0.7229240297026813, "learning_rate": 7.967653082366456e-06, "loss": 0.2989, "step": 9224 }, { "epoch": 0.6377462841341168, "grad_norm": 0.6797771321307609, "learning_rate": 7.967202563161559e-06, "loss": 0.3404, "step": 9225 }, { "epoch": 0.6378154165226408, "grad_norm": 0.7308107605191451, "learning_rate": 7.966752006768073e-06, "loss": 0.4473, "step": 9226 }, { "epoch": 0.6378845489111649, "grad_norm": 0.6605795891031137, "learning_rate": 7.966301413191643e-06, "loss": 0.3231, "step": 9227 }, { "epoch": 0.6379536812996889, "grad_norm": 0.5885697229627921, "learning_rate": 7.96585078243792e-06, "loss": 0.2911, "step": 9228 }, { "epoch": 0.6380228136882129, "grad_norm": 0.7014593444015567, "learning_rate": 7.96540011451255e-06, "loss": 0.3855, "step": 9229 }, { "epoch": 0.6380919460767369, "grad_norm": 0.6908572077984035, "learning_rate": 7.964949409421181e-06, "loss": 0.3564, "step": 9230 }, { "epoch": 0.6381610784652609, "grad_norm": 0.663841733110517, "learning_rate": 7.964498667169463e-06, "loss": 0.2917, "step": 9231 }, { "epoch": 0.638230210853785, "grad_norm": 0.7156628449047381, "learning_rate": 7.964047887763044e-06, "loss": 0.3607, "step": 9232 }, { "epoch": 0.638299343242309, "grad_norm": 0.6963346193214478, "learning_rate": 7.963597071207574e-06, "loss": 0.4046, "step": 9233 }, { "epoch": 0.6383684756308331, "grad_norm": 0.6680111149886884, "learning_rate": 7.963146217508704e-06, "loss": 0.3306, "step": 9234 }, { "epoch": 0.6384376080193571, "grad_norm": 1.0209470523254014, "learning_rate": 7.962695326672084e-06, "loss": 0.2949, "step": 9235 }, { "epoch": 0.6385067404078811, "grad_norm": 0.6712398003632105, "learning_rate": 7.962244398703365e-06, "loss": 0.3365, "step": 9236 }, { "epoch": 0.6385758727964052, "grad_norm": 0.7262371034695411, "learning_rate": 7.9617934336082e-06, "loss": 0.3408, "step": 9237 }, { "epoch": 0.6386450051849292, "grad_norm": 0.6851016940744991, "learning_rate": 7.961342431392238e-06, "loss": 0.2613, "step": 9238 }, { "epoch": 0.6387141375734532, "grad_norm": 0.727629607982611, "learning_rate": 7.960891392061137e-06, "loss": 0.3394, "step": 9239 }, { "epoch": 0.6387832699619772, "grad_norm": 0.8258658622257545, "learning_rate": 7.960440315620543e-06, "loss": 0.3412, "step": 9240 }, { "epoch": 0.6388524023505012, "grad_norm": 0.6615824113283557, "learning_rate": 7.959989202076113e-06, "loss": 0.3931, "step": 9241 }, { "epoch": 0.6389215347390252, "grad_norm": 0.6970635759026312, "learning_rate": 7.959538051433502e-06, "loss": 0.3759, "step": 9242 }, { "epoch": 0.6389906671275493, "grad_norm": 0.8127416338961575, "learning_rate": 7.959086863698363e-06, "loss": 0.3606, "step": 9243 }, { "epoch": 0.6390597995160733, "grad_norm": 0.6695606659049586, "learning_rate": 7.958635638876348e-06, "loss": 0.4472, "step": 9244 }, { "epoch": 0.6391289319045973, "grad_norm": 0.7205099128968269, "learning_rate": 7.95818437697312e-06, "loss": 0.3062, "step": 9245 }, { "epoch": 0.6391980642931213, "grad_norm": 0.6550453381745577, "learning_rate": 7.957733077994327e-06, "loss": 0.287, "step": 9246 }, { "epoch": 0.6392671966816453, "grad_norm": 0.7974846334678877, "learning_rate": 7.957281741945629e-06, "loss": 0.4134, "step": 9247 }, { "epoch": 0.6393363290701694, "grad_norm": 0.7291525409707043, "learning_rate": 7.956830368832679e-06, "loss": 0.3715, "step": 9248 }, { "epoch": 0.6394054614586934, "grad_norm": 0.6676007894929306, "learning_rate": 7.956378958661141e-06, "loss": 0.3017, "step": 9249 }, { "epoch": 0.6394745938472174, "grad_norm": 0.8166846856735762, "learning_rate": 7.955927511436667e-06, "loss": 0.3297, "step": 9250 }, { "epoch": 0.6395437262357414, "grad_norm": 0.6965658565943897, "learning_rate": 7.955476027164917e-06, "loss": 0.4341, "step": 9251 }, { "epoch": 0.6396128586242654, "grad_norm": 0.7043495756345257, "learning_rate": 7.955024505851549e-06, "loss": 0.3556, "step": 9252 }, { "epoch": 0.6396819910127894, "grad_norm": 0.7376997722738676, "learning_rate": 7.954572947502222e-06, "loss": 0.3091, "step": 9253 }, { "epoch": 0.6397511234013136, "grad_norm": 0.6491709723917186, "learning_rate": 7.954121352122597e-06, "loss": 0.392, "step": 9254 }, { "epoch": 0.6398202557898376, "grad_norm": 1.1871516761182719, "learning_rate": 7.95366971971833e-06, "loss": 0.3243, "step": 9255 }, { "epoch": 0.6398893881783616, "grad_norm": 0.6050166153884351, "learning_rate": 7.953218050295088e-06, "loss": 0.2479, "step": 9256 }, { "epoch": 0.6399585205668856, "grad_norm": 0.7649441307355224, "learning_rate": 7.952766343858527e-06, "loss": 0.3516, "step": 9257 }, { "epoch": 0.6400276529554096, "grad_norm": 1.1188981281429236, "learning_rate": 7.952314600414307e-06, "loss": 0.4469, "step": 9258 }, { "epoch": 0.6400967853439337, "grad_norm": 0.6788647051877863, "learning_rate": 7.951862819968093e-06, "loss": 0.3069, "step": 9259 }, { "epoch": 0.6401659177324577, "grad_norm": 0.6617353876048606, "learning_rate": 7.951411002525547e-06, "loss": 0.3186, "step": 9260 }, { "epoch": 0.6402350501209817, "grad_norm": 0.7482251390643777, "learning_rate": 7.950959148092332e-06, "loss": 0.3774, "step": 9261 }, { "epoch": 0.6403041825095057, "grad_norm": 0.692434491168211, "learning_rate": 7.950507256674109e-06, "loss": 0.3594, "step": 9262 }, { "epoch": 0.6403733148980297, "grad_norm": 0.6822003354883348, "learning_rate": 7.950055328276542e-06, "loss": 0.2398, "step": 9263 }, { "epoch": 0.6404424472865538, "grad_norm": 0.7340666897270517, "learning_rate": 7.949603362905299e-06, "loss": 0.3457, "step": 9264 }, { "epoch": 0.6405115796750778, "grad_norm": 0.7751337477325487, "learning_rate": 7.94915136056604e-06, "loss": 0.3737, "step": 9265 }, { "epoch": 0.6405807120636018, "grad_norm": 0.6267338280232948, "learning_rate": 7.948699321264431e-06, "loss": 0.3571, "step": 9266 }, { "epoch": 0.6406498444521258, "grad_norm": 0.6566956560620913, "learning_rate": 7.94824724500614e-06, "loss": 0.2905, "step": 9267 }, { "epoch": 0.6407189768406498, "grad_norm": 0.7743374723107873, "learning_rate": 7.947795131796829e-06, "loss": 0.3588, "step": 9268 }, { "epoch": 0.6407881092291738, "grad_norm": 0.7533848944050193, "learning_rate": 7.947342981642167e-06, "loss": 0.4002, "step": 9269 }, { "epoch": 0.6408572416176979, "grad_norm": 0.7089633474638277, "learning_rate": 7.946890794547822e-06, "loss": 0.258, "step": 9270 }, { "epoch": 0.6409263740062219, "grad_norm": 0.6563317654081017, "learning_rate": 7.94643857051946e-06, "loss": 0.2593, "step": 9271 }, { "epoch": 0.6409955063947459, "grad_norm": 0.7343172536995118, "learning_rate": 7.945986309562748e-06, "loss": 0.4115, "step": 9272 }, { "epoch": 0.6410646387832699, "grad_norm": 0.6844215469823126, "learning_rate": 7.945534011683354e-06, "loss": 0.3849, "step": 9273 }, { "epoch": 0.6411337711717939, "grad_norm": 0.7022675821786603, "learning_rate": 7.945081676886948e-06, "loss": 0.2761, "step": 9274 }, { "epoch": 0.6412029035603181, "grad_norm": 0.7414801095393563, "learning_rate": 7.9446293051792e-06, "loss": 0.3612, "step": 9275 }, { "epoch": 0.6412720359488421, "grad_norm": 0.6653369273077412, "learning_rate": 7.944176896565778e-06, "loss": 0.3577, "step": 9276 }, { "epoch": 0.6413411683373661, "grad_norm": 0.6430824593857271, "learning_rate": 7.943724451052351e-06, "loss": 0.3392, "step": 9277 }, { "epoch": 0.6414103007258901, "grad_norm": 0.6350588038097499, "learning_rate": 7.943271968644594e-06, "loss": 0.2506, "step": 9278 }, { "epoch": 0.6414794331144141, "grad_norm": 0.7616555025992781, "learning_rate": 7.942819449348173e-06, "loss": 0.4197, "step": 9279 }, { "epoch": 0.6415485655029382, "grad_norm": 0.6896082430910251, "learning_rate": 7.942366893168764e-06, "loss": 0.3724, "step": 9280 }, { "epoch": 0.6416176978914622, "grad_norm": 0.7161527398164863, "learning_rate": 7.941914300112036e-06, "loss": 0.2802, "step": 9281 }, { "epoch": 0.6416868302799862, "grad_norm": 0.7516725639555109, "learning_rate": 7.941461670183664e-06, "loss": 0.3096, "step": 9282 }, { "epoch": 0.6417559626685102, "grad_norm": 0.6758727088728468, "learning_rate": 7.941009003389318e-06, "loss": 0.38, "step": 9283 }, { "epoch": 0.6418250950570342, "grad_norm": 0.7223490507399006, "learning_rate": 7.940556299734671e-06, "loss": 0.2917, "step": 9284 }, { "epoch": 0.6418942274455582, "grad_norm": 0.707320130876344, "learning_rate": 7.9401035592254e-06, "loss": 0.2791, "step": 9285 }, { "epoch": 0.6419633598340823, "grad_norm": 0.6733469162193426, "learning_rate": 7.939650781867177e-06, "loss": 0.3715, "step": 9286 }, { "epoch": 0.6420324922226063, "grad_norm": 0.6472926239094923, "learning_rate": 7.939197967665677e-06, "loss": 0.3459, "step": 9287 }, { "epoch": 0.6421016246111303, "grad_norm": 0.691601670537219, "learning_rate": 7.938745116626578e-06, "loss": 0.3343, "step": 9288 }, { "epoch": 0.6421707569996543, "grad_norm": 0.7100766919511724, "learning_rate": 7.93829222875555e-06, "loss": 0.3008, "step": 9289 }, { "epoch": 0.6422398893881783, "grad_norm": 0.6624031821530265, "learning_rate": 7.937839304058277e-06, "loss": 0.3805, "step": 9290 }, { "epoch": 0.6423090217767023, "grad_norm": 0.6553672262512307, "learning_rate": 7.937386342540427e-06, "loss": 0.3439, "step": 9291 }, { "epoch": 0.6423781541652264, "grad_norm": 0.66259899185102, "learning_rate": 7.936933344207685e-06, "loss": 0.2989, "step": 9292 }, { "epoch": 0.6424472865537504, "grad_norm": 0.699691918474284, "learning_rate": 7.936480309065722e-06, "loss": 0.3177, "step": 9293 }, { "epoch": 0.6425164189422744, "grad_norm": 0.7299844385779042, "learning_rate": 7.936027237120218e-06, "loss": 0.4233, "step": 9294 }, { "epoch": 0.6425855513307985, "grad_norm": 0.7415239478169143, "learning_rate": 7.935574128376853e-06, "loss": 0.2749, "step": 9295 }, { "epoch": 0.6426546837193226, "grad_norm": 0.6770691596816604, "learning_rate": 7.935120982841307e-06, "loss": 0.2928, "step": 9296 }, { "epoch": 0.6427238161078466, "grad_norm": 0.7200353525164204, "learning_rate": 7.934667800519256e-06, "loss": 0.3776, "step": 9297 }, { "epoch": 0.6427929484963706, "grad_norm": 0.7817604355945658, "learning_rate": 7.93421458141638e-06, "loss": 0.3094, "step": 9298 }, { "epoch": 0.6428620808848946, "grad_norm": 0.690128964422097, "learning_rate": 7.933761325538364e-06, "loss": 0.3318, "step": 9299 }, { "epoch": 0.6429312132734186, "grad_norm": 0.7183670033337595, "learning_rate": 7.933308032890882e-06, "loss": 0.3554, "step": 9300 }, { "epoch": 0.6430003456619426, "grad_norm": 0.6994444643090972, "learning_rate": 7.93285470347962e-06, "loss": 0.3952, "step": 9301 }, { "epoch": 0.6430694780504667, "grad_norm": 0.7127805838229996, "learning_rate": 7.932401337310258e-06, "loss": 0.3108, "step": 9302 }, { "epoch": 0.6431386104389907, "grad_norm": 0.7308750966723395, "learning_rate": 7.931947934388478e-06, "loss": 0.3254, "step": 9303 }, { "epoch": 0.6432077428275147, "grad_norm": 0.6628024478946419, "learning_rate": 7.931494494719963e-06, "loss": 0.3515, "step": 9304 }, { "epoch": 0.6432768752160387, "grad_norm": 0.7060963758179766, "learning_rate": 7.931041018310396e-06, "loss": 0.3865, "step": 9305 }, { "epoch": 0.6433460076045627, "grad_norm": 0.6560291292154481, "learning_rate": 7.93058750516546e-06, "loss": 0.2452, "step": 9306 }, { "epoch": 0.6434151399930867, "grad_norm": 0.6599759299986768, "learning_rate": 7.93013395529084e-06, "loss": 0.3141, "step": 9307 }, { "epoch": 0.6434842723816108, "grad_norm": 0.8738408173770438, "learning_rate": 7.92968036869222e-06, "loss": 0.4548, "step": 9308 }, { "epoch": 0.6435534047701348, "grad_norm": 0.6804866873457573, "learning_rate": 7.929226745375285e-06, "loss": 0.3285, "step": 9309 }, { "epoch": 0.6436225371586588, "grad_norm": 0.585118024059706, "learning_rate": 7.92877308534572e-06, "loss": 0.2663, "step": 9310 }, { "epoch": 0.6436916695471828, "grad_norm": 0.7056695699245851, "learning_rate": 7.928319388609211e-06, "loss": 0.329, "step": 9311 }, { "epoch": 0.6437608019357068, "grad_norm": 0.7356405802810144, "learning_rate": 7.927865655171445e-06, "loss": 0.3974, "step": 9312 }, { "epoch": 0.6438299343242309, "grad_norm": 0.652522331905263, "learning_rate": 7.927411885038108e-06, "loss": 0.3093, "step": 9313 }, { "epoch": 0.6438990667127549, "grad_norm": 0.8434910046862981, "learning_rate": 7.926958078214886e-06, "loss": 0.2907, "step": 9314 }, { "epoch": 0.643968199101279, "grad_norm": 0.6963428318320208, "learning_rate": 7.926504234707468e-06, "loss": 0.3957, "step": 9315 }, { "epoch": 0.644037331489803, "grad_norm": 0.6788708106244785, "learning_rate": 7.926050354521542e-06, "loss": 0.3169, "step": 9316 }, { "epoch": 0.644106463878327, "grad_norm": 0.8538271521207365, "learning_rate": 7.925596437662798e-06, "loss": 0.2799, "step": 9317 }, { "epoch": 0.6441755962668511, "grad_norm": 0.5919447482868533, "learning_rate": 7.925142484136922e-06, "loss": 0.3083, "step": 9318 }, { "epoch": 0.6442447286553751, "grad_norm": 0.6928070343930528, "learning_rate": 7.924688493949605e-06, "loss": 0.3793, "step": 9319 }, { "epoch": 0.6443138610438991, "grad_norm": 0.6580372287418929, "learning_rate": 7.92423446710654e-06, "loss": 0.3708, "step": 9320 }, { "epoch": 0.6443829934324231, "grad_norm": 0.6721975465850156, "learning_rate": 7.92378040361341e-06, "loss": 0.2437, "step": 9321 }, { "epoch": 0.6444521258209471, "grad_norm": 0.6868799647614571, "learning_rate": 7.923326303475915e-06, "loss": 0.4332, "step": 9322 }, { "epoch": 0.6445212582094711, "grad_norm": 0.6734054340452983, "learning_rate": 7.922872166699738e-06, "loss": 0.2926, "step": 9323 }, { "epoch": 0.6445903905979952, "grad_norm": 0.6952392235713019, "learning_rate": 7.922417993290577e-06, "loss": 0.3697, "step": 9324 }, { "epoch": 0.6446595229865192, "grad_norm": 0.7659006039520142, "learning_rate": 7.921963783254121e-06, "loss": 0.3046, "step": 9325 }, { "epoch": 0.6447286553750432, "grad_norm": 1.1783144795314844, "learning_rate": 7.921509536596063e-06, "loss": 0.442, "step": 9326 }, { "epoch": 0.6447977877635672, "grad_norm": 0.6757663436563127, "learning_rate": 7.921055253322096e-06, "loss": 0.3319, "step": 9327 }, { "epoch": 0.6448669201520912, "grad_norm": 0.6610385463131235, "learning_rate": 7.920600933437916e-06, "loss": 0.3201, "step": 9328 }, { "epoch": 0.6449360525406153, "grad_norm": 0.669912636439883, "learning_rate": 7.920146576949214e-06, "loss": 0.3653, "step": 9329 }, { "epoch": 0.6450051849291393, "grad_norm": 0.7436772508245004, "learning_rate": 7.919692183861684e-06, "loss": 0.3793, "step": 9330 }, { "epoch": 0.6450743173176633, "grad_norm": 0.7650818316050811, "learning_rate": 7.919237754181026e-06, "loss": 0.3146, "step": 9331 }, { "epoch": 0.6451434497061873, "grad_norm": 0.84022398027903, "learning_rate": 7.918783287912932e-06, "loss": 0.3694, "step": 9332 }, { "epoch": 0.6452125820947113, "grad_norm": 0.6908104758680444, "learning_rate": 7.918328785063099e-06, "loss": 0.4159, "step": 9333 }, { "epoch": 0.6452817144832353, "grad_norm": 0.9646714995085361, "learning_rate": 7.91787424563722e-06, "loss": 0.3034, "step": 9334 }, { "epoch": 0.6453508468717594, "grad_norm": 0.7259464617886339, "learning_rate": 7.917419669640997e-06, "loss": 0.3303, "step": 9335 }, { "epoch": 0.6454199792602835, "grad_norm": 0.7098233396181228, "learning_rate": 7.916965057080124e-06, "loss": 0.3693, "step": 9336 }, { "epoch": 0.6454891116488075, "grad_norm": 0.6735305256864402, "learning_rate": 7.916510407960299e-06, "loss": 0.4265, "step": 9337 }, { "epoch": 0.6455582440373315, "grad_norm": 0.7575917710783083, "learning_rate": 7.91605572228722e-06, "loss": 0.256, "step": 9338 }, { "epoch": 0.6456273764258555, "grad_norm": 0.6752898442256935, "learning_rate": 7.915601000066589e-06, "loss": 0.3894, "step": 9339 }, { "epoch": 0.6456965088143796, "grad_norm": 0.6480280017362746, "learning_rate": 7.9151462413041e-06, "loss": 0.4233, "step": 9340 }, { "epoch": 0.6457656412029036, "grad_norm": 0.6845858851644762, "learning_rate": 7.914691446005455e-06, "loss": 0.3327, "step": 9341 }, { "epoch": 0.6458347735914276, "grad_norm": 0.663310643390757, "learning_rate": 7.914236614176358e-06, "loss": 0.3076, "step": 9342 }, { "epoch": 0.6459039059799516, "grad_norm": 0.6896583857036404, "learning_rate": 7.913781745822502e-06, "loss": 0.3484, "step": 9343 }, { "epoch": 0.6459730383684756, "grad_norm": 0.6394678845330802, "learning_rate": 7.913326840949593e-06, "loss": 0.3349, "step": 9344 }, { "epoch": 0.6460421707569997, "grad_norm": 0.7329305822956392, "learning_rate": 7.912871899563331e-06, "loss": 0.2922, "step": 9345 }, { "epoch": 0.6461113031455237, "grad_norm": 0.9935888588016734, "learning_rate": 7.91241692166942e-06, "loss": 0.3466, "step": 9346 }, { "epoch": 0.6461804355340477, "grad_norm": 0.7735302477733804, "learning_rate": 7.911961907273558e-06, "loss": 0.4021, "step": 9347 }, { "epoch": 0.6462495679225717, "grad_norm": 0.7089712526628075, "learning_rate": 7.91150685638145e-06, "loss": 0.3021, "step": 9348 }, { "epoch": 0.6463187003110957, "grad_norm": 0.6184799428738548, "learning_rate": 7.9110517689988e-06, "loss": 0.299, "step": 9349 }, { "epoch": 0.6463878326996197, "grad_norm": 0.6847402452266017, "learning_rate": 7.910596645131312e-06, "loss": 0.3264, "step": 9350 }, { "epoch": 0.6464569650881438, "grad_norm": 0.6432626870113912, "learning_rate": 7.910141484784689e-06, "loss": 0.3842, "step": 9351 }, { "epoch": 0.6465260974766678, "grad_norm": 0.6912944599662868, "learning_rate": 7.909686287964632e-06, "loss": 0.3148, "step": 9352 }, { "epoch": 0.6465952298651918, "grad_norm": 0.7160767859173124, "learning_rate": 7.909231054676855e-06, "loss": 0.2929, "step": 9353 }, { "epoch": 0.6466643622537158, "grad_norm": 0.9001569950996863, "learning_rate": 7.908775784927056e-06, "loss": 0.4363, "step": 9354 }, { "epoch": 0.6467334946422398, "grad_norm": 0.7093780231676805, "learning_rate": 7.908320478720945e-06, "loss": 0.364, "step": 9355 }, { "epoch": 0.646802627030764, "grad_norm": 0.6907647773332144, "learning_rate": 7.907865136064225e-06, "loss": 0.2853, "step": 9356 }, { "epoch": 0.646871759419288, "grad_norm": 0.8480617869091021, "learning_rate": 7.907409756962606e-06, "loss": 0.266, "step": 9357 }, { "epoch": 0.646940891807812, "grad_norm": 0.6601103443624516, "learning_rate": 7.906954341421793e-06, "loss": 0.3927, "step": 9358 }, { "epoch": 0.647010024196336, "grad_norm": 0.6968712284754982, "learning_rate": 7.906498889447498e-06, "loss": 0.3333, "step": 9359 }, { "epoch": 0.64707915658486, "grad_norm": 0.6104827328896667, "learning_rate": 7.906043401045424e-06, "loss": 0.222, "step": 9360 }, { "epoch": 0.6471482889733841, "grad_norm": 0.7087994477949116, "learning_rate": 7.905587876221281e-06, "loss": 0.3797, "step": 9361 }, { "epoch": 0.6472174213619081, "grad_norm": 0.7139758678924135, "learning_rate": 7.905132314980781e-06, "loss": 0.41, "step": 9362 }, { "epoch": 0.6472865537504321, "grad_norm": 0.7127861766469351, "learning_rate": 7.904676717329631e-06, "loss": 0.2692, "step": 9363 }, { "epoch": 0.6473556861389561, "grad_norm": 0.6714158359547604, "learning_rate": 7.904221083273543e-06, "loss": 0.2503, "step": 9364 }, { "epoch": 0.6474248185274801, "grad_norm": 0.6889214547605637, "learning_rate": 7.903765412818225e-06, "loss": 0.4878, "step": 9365 }, { "epoch": 0.6474939509160041, "grad_norm": 0.7258918994157657, "learning_rate": 7.90330970596939e-06, "loss": 0.2911, "step": 9366 }, { "epoch": 0.6475630833045282, "grad_norm": 0.6778196246558965, "learning_rate": 7.90285396273275e-06, "loss": 0.2804, "step": 9367 }, { "epoch": 0.6476322156930522, "grad_norm": 0.7509665280938694, "learning_rate": 7.902398183114017e-06, "loss": 0.3696, "step": 9368 }, { "epoch": 0.6477013480815762, "grad_norm": 0.7385295658935204, "learning_rate": 7.9019423671189e-06, "loss": 0.411, "step": 9369 }, { "epoch": 0.6477704804701002, "grad_norm": 0.7096053817011205, "learning_rate": 7.901486514753117e-06, "loss": 0.2496, "step": 9370 }, { "epoch": 0.6478396128586242, "grad_norm": 0.7074345076908107, "learning_rate": 7.901030626022375e-06, "loss": 0.2973, "step": 9371 }, { "epoch": 0.6479087452471483, "grad_norm": 0.6909569156347046, "learning_rate": 7.900574700932392e-06, "loss": 0.4523, "step": 9372 }, { "epoch": 0.6479778776356723, "grad_norm": 0.6236500716793038, "learning_rate": 7.900118739488882e-06, "loss": 0.3012, "step": 9373 }, { "epoch": 0.6480470100241963, "grad_norm": 0.6705685292983652, "learning_rate": 7.89966274169756e-06, "loss": 0.2676, "step": 9374 }, { "epoch": 0.6481161424127203, "grad_norm": 0.7411501474445318, "learning_rate": 7.89920670756414e-06, "loss": 0.3789, "step": 9375 }, { "epoch": 0.6481852748012444, "grad_norm": 0.7615601597974576, "learning_rate": 7.898750637094337e-06, "loss": 0.4221, "step": 9376 }, { "epoch": 0.6482544071897685, "grad_norm": 0.7229446026324613, "learning_rate": 7.89829453029387e-06, "loss": 0.2978, "step": 9377 }, { "epoch": 0.6483235395782925, "grad_norm": 0.6351165047304889, "learning_rate": 7.897838387168448e-06, "loss": 0.3215, "step": 9378 }, { "epoch": 0.6483926719668165, "grad_norm": 0.695213463657991, "learning_rate": 7.897382207723798e-06, "loss": 0.3852, "step": 9379 }, { "epoch": 0.6484618043553405, "grad_norm": 0.7055966104947382, "learning_rate": 7.896925991965632e-06, "loss": 0.3822, "step": 9380 }, { "epoch": 0.6485309367438645, "grad_norm": 0.7261171109683066, "learning_rate": 7.896469739899668e-06, "loss": 0.2445, "step": 9381 }, { "epoch": 0.6486000691323885, "grad_norm": 1.9564814564400013, "learning_rate": 7.896013451531623e-06, "loss": 0.304, "step": 9382 }, { "epoch": 0.6486692015209126, "grad_norm": 0.6733705933706399, "learning_rate": 7.89555712686722e-06, "loss": 0.3896, "step": 9383 }, { "epoch": 0.6487383339094366, "grad_norm": 0.7212652165706832, "learning_rate": 7.895100765912173e-06, "loss": 0.2981, "step": 9384 }, { "epoch": 0.6488074662979606, "grad_norm": 0.6351869409893978, "learning_rate": 7.894644368672208e-06, "loss": 0.2578, "step": 9385 }, { "epoch": 0.6488765986864846, "grad_norm": 0.7447220431770599, "learning_rate": 7.89418793515304e-06, "loss": 0.3969, "step": 9386 }, { "epoch": 0.6489457310750086, "grad_norm": 0.7134570772934485, "learning_rate": 7.893731465360391e-06, "loss": 0.3836, "step": 9387 }, { "epoch": 0.6490148634635327, "grad_norm": 0.6431276585931479, "learning_rate": 7.893274959299982e-06, "loss": 0.2789, "step": 9388 }, { "epoch": 0.6490839958520567, "grad_norm": 0.6818681748985133, "learning_rate": 7.892818416977534e-06, "loss": 0.3038, "step": 9389 }, { "epoch": 0.6491531282405807, "grad_norm": 0.7351886216027158, "learning_rate": 7.89236183839877e-06, "loss": 0.3791, "step": 9390 }, { "epoch": 0.6492222606291047, "grad_norm": 0.698807315514907, "learning_rate": 7.891905223569413e-06, "loss": 0.36, "step": 9391 }, { "epoch": 0.6492913930176287, "grad_norm": 0.7011603003772215, "learning_rate": 7.891448572495185e-06, "loss": 0.3007, "step": 9392 }, { "epoch": 0.6493605254061527, "grad_norm": 0.7012751708747821, "learning_rate": 7.890991885181808e-06, "loss": 0.3298, "step": 9393 }, { "epoch": 0.6494296577946768, "grad_norm": 0.6710088106879464, "learning_rate": 7.890535161635007e-06, "loss": 0.4009, "step": 9394 }, { "epoch": 0.6494987901832008, "grad_norm": 0.6883656286110449, "learning_rate": 7.890078401860507e-06, "loss": 0.2722, "step": 9395 }, { "epoch": 0.6495679225717248, "grad_norm": 0.6716764843525099, "learning_rate": 7.889621605864031e-06, "loss": 0.2685, "step": 9396 }, { "epoch": 0.6496370549602489, "grad_norm": 0.6970728660058426, "learning_rate": 7.889164773651305e-06, "loss": 0.3906, "step": 9397 }, { "epoch": 0.649706187348773, "grad_norm": 0.7179915844689602, "learning_rate": 7.888707905228055e-06, "loss": 0.2996, "step": 9398 }, { "epoch": 0.649775319737297, "grad_norm": 0.7243585078085288, "learning_rate": 7.888251000600007e-06, "loss": 0.3032, "step": 9399 }, { "epoch": 0.649844452125821, "grad_norm": 0.7452010780380552, "learning_rate": 7.887794059772887e-06, "loss": 0.3022, "step": 9400 }, { "epoch": 0.649913584514345, "grad_norm": 0.7854507069412942, "learning_rate": 7.88733708275242e-06, "loss": 0.3602, "step": 9401 }, { "epoch": 0.649982716902869, "grad_norm": 0.7030478799022977, "learning_rate": 7.886880069544337e-06, "loss": 0.2826, "step": 9402 }, { "epoch": 0.650051849291393, "grad_norm": 0.6401283252126088, "learning_rate": 7.886423020154364e-06, "loss": 0.2502, "step": 9403 }, { "epoch": 0.650120981679917, "grad_norm": 0.7132595007598056, "learning_rate": 7.885965934588229e-06, "loss": 0.3644, "step": 9404 }, { "epoch": 0.6501901140684411, "grad_norm": 0.9808169376227274, "learning_rate": 7.88550881285166e-06, "loss": 0.3603, "step": 9405 }, { "epoch": 0.6502592464569651, "grad_norm": 0.7859416024222199, "learning_rate": 7.88505165495039e-06, "loss": 0.2766, "step": 9406 }, { "epoch": 0.6503283788454891, "grad_norm": 0.7300187483774286, "learning_rate": 7.884594460890147e-06, "loss": 0.3401, "step": 9407 }, { "epoch": 0.6503975112340131, "grad_norm": 0.6522374763526416, "learning_rate": 7.884137230676658e-06, "loss": 0.37, "step": 9408 }, { "epoch": 0.6504666436225371, "grad_norm": 0.7699952966759738, "learning_rate": 7.883679964315656e-06, "loss": 0.3246, "step": 9409 }, { "epoch": 0.6505357760110612, "grad_norm": 0.6696367451090578, "learning_rate": 7.88322266181287e-06, "loss": 0.286, "step": 9410 }, { "epoch": 0.6506049083995852, "grad_norm": 0.6751028304850534, "learning_rate": 7.882765323174036e-06, "loss": 0.357, "step": 9411 }, { "epoch": 0.6506740407881092, "grad_norm": 0.6620930435448742, "learning_rate": 7.882307948404883e-06, "loss": 0.4092, "step": 9412 }, { "epoch": 0.6507431731766332, "grad_norm": 0.6975225311020273, "learning_rate": 7.881850537511143e-06, "loss": 0.3517, "step": 9413 }, { "epoch": 0.6508123055651572, "grad_norm": 0.69539224932436, "learning_rate": 7.88139309049855e-06, "loss": 0.2847, "step": 9414 }, { "epoch": 0.6508814379536813, "grad_norm": 0.6770351398236345, "learning_rate": 7.880935607372835e-06, "loss": 0.3665, "step": 9415 }, { "epoch": 0.6509505703422053, "grad_norm": 1.3880770844193582, "learning_rate": 7.880478088139736e-06, "loss": 0.3463, "step": 9416 }, { "epoch": 0.6510197027307294, "grad_norm": 0.6680768286815207, "learning_rate": 7.880020532804983e-06, "loss": 0.2752, "step": 9417 }, { "epoch": 0.6510888351192534, "grad_norm": 0.7769383013200557, "learning_rate": 7.879562941374312e-06, "loss": 0.3706, "step": 9418 }, { "epoch": 0.6511579675077774, "grad_norm": 0.7838195940375632, "learning_rate": 7.879105313853458e-06, "loss": 0.3775, "step": 9419 }, { "epoch": 0.6512270998963015, "grad_norm": 0.6481521359399449, "learning_rate": 7.878647650248159e-06, "loss": 0.3238, "step": 9420 }, { "epoch": 0.6512962322848255, "grad_norm": 0.6606355483524129, "learning_rate": 7.878189950564148e-06, "loss": 0.283, "step": 9421 }, { "epoch": 0.6513653646733495, "grad_norm": 0.6882134696280785, "learning_rate": 7.87773221480716e-06, "loss": 0.3927, "step": 9422 }, { "epoch": 0.6514344970618735, "grad_norm": 0.6758424189655813, "learning_rate": 7.877274442982938e-06, "loss": 0.3547, "step": 9423 }, { "epoch": 0.6515036294503975, "grad_norm": 0.6644533412296356, "learning_rate": 7.876816635097212e-06, "loss": 0.2585, "step": 9424 }, { "epoch": 0.6515727618389215, "grad_norm": 0.6551568885018979, "learning_rate": 7.876358791155727e-06, "loss": 0.2955, "step": 9425 }, { "epoch": 0.6516418942274456, "grad_norm": 0.6473863288238256, "learning_rate": 7.875900911164214e-06, "loss": 0.4093, "step": 9426 }, { "epoch": 0.6517110266159696, "grad_norm": 0.604895355509768, "learning_rate": 7.875442995128417e-06, "loss": 0.2732, "step": 9427 }, { "epoch": 0.6517801590044936, "grad_norm": 0.7196329944153391, "learning_rate": 7.874985043054074e-06, "loss": 0.2894, "step": 9428 }, { "epoch": 0.6518492913930176, "grad_norm": 0.7354892675312411, "learning_rate": 7.874527054946923e-06, "loss": 0.396, "step": 9429 }, { "epoch": 0.6519184237815416, "grad_norm": 0.5657118427286807, "learning_rate": 7.874069030812707e-06, "loss": 0.2909, "step": 9430 }, { "epoch": 0.6519875561700657, "grad_norm": 0.6502723798427923, "learning_rate": 7.873610970657163e-06, "loss": 0.3299, "step": 9431 }, { "epoch": 0.6520566885585897, "grad_norm": 0.6482881281367289, "learning_rate": 7.873152874486034e-06, "loss": 0.3231, "step": 9432 }, { "epoch": 0.6521258209471137, "grad_norm": 0.6068210536059977, "learning_rate": 7.872694742305062e-06, "loss": 0.3895, "step": 9433 }, { "epoch": 0.6521949533356377, "grad_norm": 0.5725437137954906, "learning_rate": 7.872236574119986e-06, "loss": 0.2609, "step": 9434 }, { "epoch": 0.6522640857241617, "grad_norm": 0.6599626660365017, "learning_rate": 7.871778369936552e-06, "loss": 0.2749, "step": 9435 }, { "epoch": 0.6523332181126857, "grad_norm": 0.6828892056587073, "learning_rate": 7.871320129760501e-06, "loss": 0.2919, "step": 9436 }, { "epoch": 0.6524023505012099, "grad_norm": 3.4807784876834775, "learning_rate": 7.870861853597576e-06, "loss": 0.4081, "step": 9437 }, { "epoch": 0.6524714828897339, "grad_norm": 0.638147901144887, "learning_rate": 7.870403541453521e-06, "loss": 0.3063, "step": 9438 }, { "epoch": 0.6525406152782579, "grad_norm": 0.6567055200009186, "learning_rate": 7.86994519333408e-06, "loss": 0.2844, "step": 9439 }, { "epoch": 0.6526097476667819, "grad_norm": 0.6883206089714317, "learning_rate": 7.869486809244998e-06, "loss": 0.3815, "step": 9440 }, { "epoch": 0.6526788800553059, "grad_norm": 0.722936296605328, "learning_rate": 7.869028389192019e-06, "loss": 0.3117, "step": 9441 }, { "epoch": 0.65274801244383, "grad_norm": 0.6332796244290908, "learning_rate": 7.86856993318089e-06, "loss": 0.3, "step": 9442 }, { "epoch": 0.652817144832354, "grad_norm": 0.6622428050690417, "learning_rate": 7.868111441217355e-06, "loss": 0.3461, "step": 9443 }, { "epoch": 0.652886277220878, "grad_norm": 0.7358209170290795, "learning_rate": 7.867652913307163e-06, "loss": 0.4297, "step": 9444 }, { "epoch": 0.652955409609402, "grad_norm": 0.8142956300459844, "learning_rate": 7.867194349456056e-06, "loss": 0.3184, "step": 9445 }, { "epoch": 0.653024541997926, "grad_norm": 0.6109406941288668, "learning_rate": 7.866735749669787e-06, "loss": 0.2817, "step": 9446 }, { "epoch": 0.65309367438645, "grad_norm": 0.7214072258075674, "learning_rate": 7.866277113954102e-06, "loss": 0.4345, "step": 9447 }, { "epoch": 0.6531628067749741, "grad_norm": 0.6476321026339424, "learning_rate": 7.865818442314748e-06, "loss": 0.2986, "step": 9448 }, { "epoch": 0.6532319391634981, "grad_norm": 0.5954805906441637, "learning_rate": 7.865359734757474e-06, "loss": 0.308, "step": 9449 }, { "epoch": 0.6533010715520221, "grad_norm": 0.6684939064709879, "learning_rate": 7.864900991288028e-06, "loss": 0.2866, "step": 9450 }, { "epoch": 0.6533702039405461, "grad_norm": 0.7225272355934739, "learning_rate": 7.864442211912162e-06, "loss": 0.4432, "step": 9451 }, { "epoch": 0.6534393363290701, "grad_norm": 0.8029296196351815, "learning_rate": 7.863983396635624e-06, "loss": 0.32, "step": 9452 }, { "epoch": 0.6535084687175942, "grad_norm": 0.7349768329289247, "learning_rate": 7.863524545464165e-06, "loss": 0.3005, "step": 9453 }, { "epoch": 0.6535776011061182, "grad_norm": 0.6813050528304234, "learning_rate": 7.863065658403537e-06, "loss": 0.3811, "step": 9454 }, { "epoch": 0.6536467334946422, "grad_norm": 0.7303036379804861, "learning_rate": 7.862606735459489e-06, "loss": 0.4212, "step": 9455 }, { "epoch": 0.6537158658831662, "grad_norm": 0.6347174135447871, "learning_rate": 7.862147776637774e-06, "loss": 0.2785, "step": 9456 }, { "epoch": 0.6537849982716902, "grad_norm": 0.6394750122294109, "learning_rate": 7.861688781944147e-06, "loss": 0.3068, "step": 9457 }, { "epoch": 0.6538541306602144, "grad_norm": 0.7998686602460908, "learning_rate": 7.861229751384357e-06, "loss": 0.4466, "step": 9458 }, { "epoch": 0.6539232630487384, "grad_norm": 0.8028030888344739, "learning_rate": 7.860770684964156e-06, "loss": 0.3432, "step": 9459 }, { "epoch": 0.6539923954372624, "grad_norm": 0.8354031566646629, "learning_rate": 7.860311582689303e-06, "loss": 0.2506, "step": 9460 }, { "epoch": 0.6540615278257864, "grad_norm": 0.7386864182998905, "learning_rate": 7.859852444565545e-06, "loss": 0.357, "step": 9461 }, { "epoch": 0.6541306602143104, "grad_norm": 0.7380807505820867, "learning_rate": 7.859393270598643e-06, "loss": 0.3478, "step": 9462 }, { "epoch": 0.6541997926028345, "grad_norm": 1.3183582662788635, "learning_rate": 7.858934060794348e-06, "loss": 0.2989, "step": 9463 }, { "epoch": 0.6542689249913585, "grad_norm": 0.8248537464883762, "learning_rate": 7.858474815158416e-06, "loss": 0.2864, "step": 9464 }, { "epoch": 0.6543380573798825, "grad_norm": 0.7010807870593624, "learning_rate": 7.858015533696606e-06, "loss": 0.3711, "step": 9465 }, { "epoch": 0.6544071897684065, "grad_norm": 0.6757681030982179, "learning_rate": 7.857556216414669e-06, "loss": 0.3052, "step": 9466 }, { "epoch": 0.6544763221569305, "grad_norm": 0.6547467603563054, "learning_rate": 7.857096863318366e-06, "loss": 0.3, "step": 9467 }, { "epoch": 0.6545454545454545, "grad_norm": 0.7732243490676804, "learning_rate": 7.856637474413452e-06, "loss": 0.3978, "step": 9468 }, { "epoch": 0.6546145869339786, "grad_norm": 0.6634955363415747, "learning_rate": 7.856178049705683e-06, "loss": 0.3325, "step": 9469 }, { "epoch": 0.6546837193225026, "grad_norm": 0.7299414888026118, "learning_rate": 7.855718589200822e-06, "loss": 0.3476, "step": 9470 }, { "epoch": 0.6547528517110266, "grad_norm": 0.6697132646698377, "learning_rate": 7.855259092904624e-06, "loss": 0.3197, "step": 9471 }, { "epoch": 0.6548219840995506, "grad_norm": 0.7024211126793174, "learning_rate": 7.854799560822849e-06, "loss": 0.4208, "step": 9472 }, { "epoch": 0.6548911164880746, "grad_norm": 0.7459331121954325, "learning_rate": 7.854339992961256e-06, "loss": 0.3951, "step": 9473 }, { "epoch": 0.6549602488765986, "grad_norm": 0.9476782122732964, "learning_rate": 7.853880389325604e-06, "loss": 0.2934, "step": 9474 }, { "epoch": 0.6550293812651227, "grad_norm": 0.9589855770458608, "learning_rate": 7.853420749921656e-06, "loss": 0.3676, "step": 9475 }, { "epoch": 0.6550985136536467, "grad_norm": 1.2168619310996691, "learning_rate": 7.85296107475517e-06, "loss": 0.3462, "step": 9476 }, { "epoch": 0.6551676460421707, "grad_norm": 0.6364821828144017, "learning_rate": 7.852501363831909e-06, "loss": 0.3011, "step": 9477 }, { "epoch": 0.6552367784306948, "grad_norm": 0.7031438947681222, "learning_rate": 7.852041617157634e-06, "loss": 0.3058, "step": 9478 }, { "epoch": 0.6553059108192189, "grad_norm": 0.6443902877931686, "learning_rate": 7.851581834738104e-06, "loss": 0.3158, "step": 9479 }, { "epoch": 0.6553750432077429, "grad_norm": 0.8977499872058007, "learning_rate": 7.851122016579086e-06, "loss": 0.3215, "step": 9480 }, { "epoch": 0.6554441755962669, "grad_norm": 0.7635647738570644, "learning_rate": 7.850662162686344e-06, "loss": 0.2793, "step": 9481 }, { "epoch": 0.6555133079847909, "grad_norm": 0.6872389414874097, "learning_rate": 7.850202273065638e-06, "loss": 0.3257, "step": 9482 }, { "epoch": 0.6555824403733149, "grad_norm": 0.6994843283841045, "learning_rate": 7.849742347722733e-06, "loss": 0.4409, "step": 9483 }, { "epoch": 0.6556515727618389, "grad_norm": 0.6173407377728977, "learning_rate": 7.849282386663393e-06, "loss": 0.3028, "step": 9484 }, { "epoch": 0.655720705150363, "grad_norm": 0.7078011117477743, "learning_rate": 7.848822389893383e-06, "loss": 0.2908, "step": 9485 }, { "epoch": 0.655789837538887, "grad_norm": 0.6953257379559904, "learning_rate": 7.848362357418466e-06, "loss": 0.3347, "step": 9486 }, { "epoch": 0.655858969927411, "grad_norm": 0.6643482879874151, "learning_rate": 7.847902289244413e-06, "loss": 0.356, "step": 9487 }, { "epoch": 0.655928102315935, "grad_norm": 0.7062805248443023, "learning_rate": 7.847442185376986e-06, "loss": 0.2432, "step": 9488 }, { "epoch": 0.655997234704459, "grad_norm": 0.687596741450974, "learning_rate": 7.846982045821954e-06, "loss": 0.2841, "step": 9489 }, { "epoch": 0.656066367092983, "grad_norm": 0.6881432414770656, "learning_rate": 7.84652187058508e-06, "loss": 0.4397, "step": 9490 }, { "epoch": 0.6561354994815071, "grad_norm": 0.9424200131820059, "learning_rate": 7.846061659672137e-06, "loss": 0.2658, "step": 9491 }, { "epoch": 0.6562046318700311, "grad_norm": 0.6725448863887799, "learning_rate": 7.845601413088887e-06, "loss": 0.2725, "step": 9492 }, { "epoch": 0.6562737642585551, "grad_norm": 0.7365098701828439, "learning_rate": 7.845141130841106e-06, "loss": 0.3317, "step": 9493 }, { "epoch": 0.6563428966470791, "grad_norm": 0.6972029144191373, "learning_rate": 7.844680812934555e-06, "loss": 0.3831, "step": 9494 }, { "epoch": 0.6564120290356031, "grad_norm": 0.6756055619323855, "learning_rate": 7.844220459375007e-06, "loss": 0.2746, "step": 9495 }, { "epoch": 0.6564811614241272, "grad_norm": 0.782547029733164, "learning_rate": 7.843760070168231e-06, "loss": 0.3088, "step": 9496 }, { "epoch": 0.6565502938126512, "grad_norm": 0.6700614587330229, "learning_rate": 7.843299645319998e-06, "loss": 0.4007, "step": 9497 }, { "epoch": 0.6566194262011753, "grad_norm": 0.6544406939236204, "learning_rate": 7.842839184836078e-06, "loss": 0.3586, "step": 9498 }, { "epoch": 0.6566885585896993, "grad_norm": 0.663874408837816, "learning_rate": 7.842378688722243e-06, "loss": 0.3689, "step": 9499 }, { "epoch": 0.6567576909782233, "grad_norm": 0.7014369637472883, "learning_rate": 7.841918156984261e-06, "loss": 0.3074, "step": 9500 }, { "epoch": 0.6568268233667474, "grad_norm": 0.8468317732769152, "learning_rate": 7.84145758962791e-06, "loss": 0.4702, "step": 9501 }, { "epoch": 0.6568959557552714, "grad_norm": 0.6757104955553216, "learning_rate": 7.840996986658957e-06, "loss": 0.2746, "step": 9502 }, { "epoch": 0.6569650881437954, "grad_norm": 0.7774071827707697, "learning_rate": 7.840536348083176e-06, "loss": 0.3447, "step": 9503 }, { "epoch": 0.6570342205323194, "grad_norm": 0.7299718595646026, "learning_rate": 7.840075673906343e-06, "loss": 0.3941, "step": 9504 }, { "epoch": 0.6571033529208434, "grad_norm": 0.7394545915488848, "learning_rate": 7.839614964134228e-06, "loss": 0.3303, "step": 9505 }, { "epoch": 0.6571724853093674, "grad_norm": 0.8214525870719526, "learning_rate": 7.839154218772608e-06, "loss": 0.327, "step": 9506 }, { "epoch": 0.6572416176978915, "grad_norm": 1.1492195598488832, "learning_rate": 7.838693437827255e-06, "loss": 0.3355, "step": 9507 }, { "epoch": 0.6573107500864155, "grad_norm": 0.7692367883291898, "learning_rate": 7.838232621303947e-06, "loss": 0.4309, "step": 9508 }, { "epoch": 0.6573798824749395, "grad_norm": 0.7197090200036745, "learning_rate": 7.837771769208458e-06, "loss": 0.2782, "step": 9509 }, { "epoch": 0.6574490148634635, "grad_norm": 0.6738823322693894, "learning_rate": 7.837310881546564e-06, "loss": 0.2921, "step": 9510 }, { "epoch": 0.6575181472519875, "grad_norm": 0.7993628737095506, "learning_rate": 7.836849958324042e-06, "loss": 0.3083, "step": 9511 }, { "epoch": 0.6575872796405116, "grad_norm": 0.7520610736591983, "learning_rate": 7.83638899954667e-06, "loss": 0.373, "step": 9512 }, { "epoch": 0.6576564120290356, "grad_norm": 0.7678842692243002, "learning_rate": 7.83592800522022e-06, "loss": 0.3107, "step": 9513 }, { "epoch": 0.6577255444175596, "grad_norm": 0.6434563287139701, "learning_rate": 7.835466975350475e-06, "loss": 0.3201, "step": 9514 }, { "epoch": 0.6577946768060836, "grad_norm": 0.7301399179277317, "learning_rate": 7.835005909943212e-06, "loss": 0.4019, "step": 9515 }, { "epoch": 0.6578638091946076, "grad_norm": 0.7140183171819734, "learning_rate": 7.834544809004207e-06, "loss": 0.3277, "step": 9516 }, { "epoch": 0.6579329415831316, "grad_norm": 0.6952643349247174, "learning_rate": 7.834083672539246e-06, "loss": 0.3194, "step": 9517 }, { "epoch": 0.6580020739716557, "grad_norm": 0.6547526502207175, "learning_rate": 7.833622500554099e-06, "loss": 0.3786, "step": 9518 }, { "epoch": 0.6580712063601798, "grad_norm": 0.7064811514327397, "learning_rate": 7.833161293054555e-06, "loss": 0.3864, "step": 9519 }, { "epoch": 0.6581403387487038, "grad_norm": 0.6592201797240936, "learning_rate": 7.832700050046386e-06, "loss": 0.2885, "step": 9520 }, { "epoch": 0.6582094711372278, "grad_norm": 0.7368476924068363, "learning_rate": 7.83223877153538e-06, "loss": 0.2782, "step": 9521 }, { "epoch": 0.6582786035257518, "grad_norm": 0.7393284467853445, "learning_rate": 7.831777457527314e-06, "loss": 0.3992, "step": 9522 }, { "epoch": 0.6583477359142759, "grad_norm": 0.7864959249179406, "learning_rate": 7.831316108027971e-06, "loss": 0.413, "step": 9523 }, { "epoch": 0.6584168683027999, "grad_norm": 0.6504387245523814, "learning_rate": 7.830854723043133e-06, "loss": 0.3177, "step": 9524 }, { "epoch": 0.6584860006913239, "grad_norm": 0.6965227414733174, "learning_rate": 7.830393302578585e-06, "loss": 0.3643, "step": 9525 }, { "epoch": 0.6585551330798479, "grad_norm": 0.699254633960867, "learning_rate": 7.829931846640106e-06, "loss": 0.3931, "step": 9526 }, { "epoch": 0.6586242654683719, "grad_norm": 0.7165015259121592, "learning_rate": 7.829470355233482e-06, "loss": 0.3142, "step": 9527 }, { "epoch": 0.658693397856896, "grad_norm": 0.6487147366971463, "learning_rate": 7.829008828364498e-06, "loss": 0.3033, "step": 9528 }, { "epoch": 0.65876253024542, "grad_norm": 0.7147512349126965, "learning_rate": 7.828547266038935e-06, "loss": 0.4056, "step": 9529 }, { "epoch": 0.658831662633944, "grad_norm": 0.7125128623536083, "learning_rate": 7.828085668262579e-06, "loss": 0.3913, "step": 9530 }, { "epoch": 0.658900795022468, "grad_norm": 0.6357458634086457, "learning_rate": 7.82762403504122e-06, "loss": 0.2999, "step": 9531 }, { "epoch": 0.658969927410992, "grad_norm": 1.1022356801317144, "learning_rate": 7.827162366380637e-06, "loss": 0.3354, "step": 9532 }, { "epoch": 0.659039059799516, "grad_norm": 0.7051146650806698, "learning_rate": 7.826700662286617e-06, "loss": 0.4214, "step": 9533 }, { "epoch": 0.6591081921880401, "grad_norm": 0.6935917991527231, "learning_rate": 7.826238922764952e-06, "loss": 0.3038, "step": 9534 }, { "epoch": 0.6591773245765641, "grad_norm": 0.6167372074868905, "learning_rate": 7.825777147821424e-06, "loss": 0.3066, "step": 9535 }, { "epoch": 0.6592464569650881, "grad_norm": 0.6715786460842303, "learning_rate": 7.825315337461825e-06, "loss": 0.374, "step": 9536 }, { "epoch": 0.6593155893536121, "grad_norm": 0.7084823406451566, "learning_rate": 7.824853491691938e-06, "loss": 0.4168, "step": 9537 }, { "epoch": 0.6593847217421361, "grad_norm": 0.6683245146827668, "learning_rate": 7.824391610517555e-06, "loss": 0.2697, "step": 9538 }, { "epoch": 0.6594538541306603, "grad_norm": 0.6333361213572024, "learning_rate": 7.82392969394446e-06, "loss": 0.3256, "step": 9539 }, { "epoch": 0.6595229865191843, "grad_norm": 0.6774602594992266, "learning_rate": 7.82346774197845e-06, "loss": 0.407, "step": 9540 }, { "epoch": 0.6595921189077083, "grad_norm": 0.6545600751405579, "learning_rate": 7.823005754625308e-06, "loss": 0.3225, "step": 9541 }, { "epoch": 0.6596612512962323, "grad_norm": 0.7065525993765618, "learning_rate": 7.822543731890828e-06, "loss": 0.2488, "step": 9542 }, { "epoch": 0.6597303836847563, "grad_norm": 0.6898648720898439, "learning_rate": 7.8220816737808e-06, "loss": 0.3444, "step": 9543 }, { "epoch": 0.6597995160732804, "grad_norm": 0.8303304332593955, "learning_rate": 7.821619580301014e-06, "loss": 0.3869, "step": 9544 }, { "epoch": 0.6598686484618044, "grad_norm": 0.7950554077883031, "learning_rate": 7.82115745145726e-06, "loss": 0.3743, "step": 9545 }, { "epoch": 0.6599377808503284, "grad_norm": 0.9016341965551642, "learning_rate": 7.820695287255333e-06, "loss": 0.3057, "step": 9546 }, { "epoch": 0.6600069132388524, "grad_norm": 0.7539540112086396, "learning_rate": 7.820233087701026e-06, "loss": 0.3678, "step": 9547 }, { "epoch": 0.6600760456273764, "grad_norm": 0.7112084625660439, "learning_rate": 7.819770852800127e-06, "loss": 0.3604, "step": 9548 }, { "epoch": 0.6601451780159004, "grad_norm": 0.7092228788709269, "learning_rate": 7.819308582558436e-06, "loss": 0.285, "step": 9549 }, { "epoch": 0.6602143104044245, "grad_norm": 0.8111559112705639, "learning_rate": 7.818846276981742e-06, "loss": 0.3181, "step": 9550 }, { "epoch": 0.6602834427929485, "grad_norm": 0.7498303117905373, "learning_rate": 7.818383936075839e-06, "loss": 0.4574, "step": 9551 }, { "epoch": 0.6603525751814725, "grad_norm": 0.7373236091278397, "learning_rate": 7.817921559846524e-06, "loss": 0.3118, "step": 9552 }, { "epoch": 0.6604217075699965, "grad_norm": 0.9470877296084417, "learning_rate": 7.817459148299591e-06, "loss": 0.263, "step": 9553 }, { "epoch": 0.6604908399585205, "grad_norm": 0.8668309799293681, "learning_rate": 7.816996701440837e-06, "loss": 0.3661, "step": 9554 }, { "epoch": 0.6605599723470446, "grad_norm": 0.7450555496940335, "learning_rate": 7.816534219276054e-06, "loss": 0.3924, "step": 9555 }, { "epoch": 0.6606291047355686, "grad_norm": 0.6347340917718309, "learning_rate": 7.816071701811043e-06, "loss": 0.2886, "step": 9556 }, { "epoch": 0.6606982371240926, "grad_norm": 0.6715020680180374, "learning_rate": 7.815609149051598e-06, "loss": 0.2995, "step": 9557 }, { "epoch": 0.6607673695126166, "grad_norm": 0.6954569497655385, "learning_rate": 7.815146561003514e-06, "loss": 0.3889, "step": 9558 }, { "epoch": 0.6608365019011407, "grad_norm": 0.6890518223960308, "learning_rate": 7.814683937672596e-06, "loss": 0.3635, "step": 9559 }, { "epoch": 0.6609056342896648, "grad_norm": 0.7947380131240662, "learning_rate": 7.814221279064637e-06, "loss": 0.2875, "step": 9560 }, { "epoch": 0.6609747666781888, "grad_norm": 0.7986247836494941, "learning_rate": 7.813758585185435e-06, "loss": 0.4138, "step": 9561 }, { "epoch": 0.6610438990667128, "grad_norm": 0.6842936475687341, "learning_rate": 7.81329585604079e-06, "loss": 0.3766, "step": 9562 }, { "epoch": 0.6611130314552368, "grad_norm": 0.5706750466730346, "learning_rate": 7.812833091636504e-06, "loss": 0.27, "step": 9563 }, { "epoch": 0.6611821638437608, "grad_norm": 0.7141041793156775, "learning_rate": 7.812370291978375e-06, "loss": 0.2957, "step": 9564 }, { "epoch": 0.6612512962322848, "grad_norm": 0.6762378032106428, "learning_rate": 7.811907457072202e-06, "loss": 0.4052, "step": 9565 }, { "epoch": 0.6613204286208089, "grad_norm": 0.7064017090802617, "learning_rate": 7.811444586923786e-06, "loss": 0.421, "step": 9566 }, { "epoch": 0.6613895610093329, "grad_norm": 0.6609348787603258, "learning_rate": 7.810981681538932e-06, "loss": 0.2665, "step": 9567 }, { "epoch": 0.6614586933978569, "grad_norm": 0.7346987783950126, "learning_rate": 7.810518740923438e-06, "loss": 0.3332, "step": 9568 }, { "epoch": 0.6615278257863809, "grad_norm": 2.772601546008353, "learning_rate": 7.810055765083108e-06, "loss": 0.415, "step": 9569 }, { "epoch": 0.6615969581749049, "grad_norm": 0.9923391495061868, "learning_rate": 7.809592754023743e-06, "loss": 0.3201, "step": 9570 }, { "epoch": 0.661666090563429, "grad_norm": 0.6518358942525033, "learning_rate": 7.809129707751146e-06, "loss": 0.2697, "step": 9571 }, { "epoch": 0.661735222951953, "grad_norm": 0.6925894381005137, "learning_rate": 7.808666626271122e-06, "loss": 0.4476, "step": 9572 }, { "epoch": 0.661804355340477, "grad_norm": 0.692592123738794, "learning_rate": 7.808203509589474e-06, "loss": 0.3815, "step": 9573 }, { "epoch": 0.661873487729001, "grad_norm": 0.631988296083081, "learning_rate": 7.807740357712006e-06, "loss": 0.2925, "step": 9574 }, { "epoch": 0.661942620117525, "grad_norm": 0.672741712160274, "learning_rate": 7.807277170644524e-06, "loss": 0.3003, "step": 9575 }, { "epoch": 0.662011752506049, "grad_norm": 0.8933049901266553, "learning_rate": 7.806813948392833e-06, "loss": 0.4408, "step": 9576 }, { "epoch": 0.6620808848945731, "grad_norm": 0.6833258151557327, "learning_rate": 7.806350690962736e-06, "loss": 0.341, "step": 9577 }, { "epoch": 0.6621500172830971, "grad_norm": 0.6924260711755734, "learning_rate": 7.805887398360044e-06, "loss": 0.2902, "step": 9578 }, { "epoch": 0.6622191496716212, "grad_norm": 0.6993709702636399, "learning_rate": 7.805424070590557e-06, "loss": 0.4252, "step": 9579 }, { "epoch": 0.6622882820601452, "grad_norm": 0.7376069141367644, "learning_rate": 7.804960707660089e-06, "loss": 0.3708, "step": 9580 }, { "epoch": 0.6623574144486692, "grad_norm": 1.408787360122909, "learning_rate": 7.804497309574443e-06, "loss": 0.2652, "step": 9581 }, { "epoch": 0.6624265468371933, "grad_norm": 1.481580976702888, "learning_rate": 7.80403387633943e-06, "loss": 0.3064, "step": 9582 }, { "epoch": 0.6624956792257173, "grad_norm": 0.6594479307323259, "learning_rate": 7.803570407960855e-06, "loss": 0.3784, "step": 9583 }, { "epoch": 0.6625648116142413, "grad_norm": 0.6742030316409728, "learning_rate": 7.803106904444527e-06, "loss": 0.3166, "step": 9584 }, { "epoch": 0.6626339440027653, "grad_norm": 0.6671091369113528, "learning_rate": 7.802643365796257e-06, "loss": 0.2905, "step": 9585 }, { "epoch": 0.6627030763912893, "grad_norm": 0.6674526615730036, "learning_rate": 7.802179792021854e-06, "loss": 0.3575, "step": 9586 }, { "epoch": 0.6627722087798134, "grad_norm": 0.6752621070731949, "learning_rate": 7.801716183127128e-06, "loss": 0.3791, "step": 9587 }, { "epoch": 0.6628413411683374, "grad_norm": 0.5925561192527972, "learning_rate": 7.80125253911789e-06, "loss": 0.3105, "step": 9588 }, { "epoch": 0.6629104735568614, "grad_norm": 0.6395734262419428, "learning_rate": 7.80078885999995e-06, "loss": 0.2571, "step": 9589 }, { "epoch": 0.6629796059453854, "grad_norm": 0.7691813680750379, "learning_rate": 7.800325145779121e-06, "loss": 0.4409, "step": 9590 }, { "epoch": 0.6630487383339094, "grad_norm": 0.5976706041096174, "learning_rate": 7.799861396461213e-06, "loss": 0.3237, "step": 9591 }, { "epoch": 0.6631178707224334, "grad_norm": 0.6745769653225412, "learning_rate": 7.799397612052039e-06, "loss": 0.2694, "step": 9592 }, { "epoch": 0.6631870031109575, "grad_norm": 0.7541980271935089, "learning_rate": 7.79893379255741e-06, "loss": 0.3097, "step": 9593 }, { "epoch": 0.6632561354994815, "grad_norm": 0.8536924936013217, "learning_rate": 7.798469937983144e-06, "loss": 0.3599, "step": 9594 }, { "epoch": 0.6633252678880055, "grad_norm": 0.6293463350432886, "learning_rate": 7.79800604833505e-06, "loss": 0.2998, "step": 9595 }, { "epoch": 0.6633944002765295, "grad_norm": 0.7118786069132311, "learning_rate": 7.797542123618942e-06, "loss": 0.292, "step": 9596 }, { "epoch": 0.6634635326650535, "grad_norm": 0.7515465030185245, "learning_rate": 7.797078163840636e-06, "loss": 0.4265, "step": 9597 }, { "epoch": 0.6635326650535776, "grad_norm": 0.725578324312504, "learning_rate": 7.796614169005949e-06, "loss": 0.3727, "step": 9598 }, { "epoch": 0.6636017974421016, "grad_norm": 0.6938643542902279, "learning_rate": 7.796150139120693e-06, "loss": 0.2708, "step": 9599 }, { "epoch": 0.6636709298306257, "grad_norm": 1.0368050883075601, "learning_rate": 7.795686074190685e-06, "loss": 0.3249, "step": 9600 }, { "epoch": 0.6637400622191497, "grad_norm": 0.7438962447325213, "learning_rate": 7.795221974221741e-06, "loss": 0.4473, "step": 9601 }, { "epoch": 0.6638091946076737, "grad_norm": 2.4296633965954313, "learning_rate": 7.79475783921968e-06, "loss": 0.3072, "step": 9602 }, { "epoch": 0.6638783269961978, "grad_norm": 0.674789902037814, "learning_rate": 7.794293669190313e-06, "loss": 0.2766, "step": 9603 }, { "epoch": 0.6639474593847218, "grad_norm": 0.6922993775898855, "learning_rate": 7.793829464139465e-06, "loss": 0.3893, "step": 9604 }, { "epoch": 0.6640165917732458, "grad_norm": 1.326224377272258, "learning_rate": 7.793365224072949e-06, "loss": 0.3951, "step": 9605 }, { "epoch": 0.6640857241617698, "grad_norm": 0.6381640630482894, "learning_rate": 7.792900948996584e-06, "loss": 0.2846, "step": 9606 }, { "epoch": 0.6641548565502938, "grad_norm": 1.020783946308574, "learning_rate": 7.792436638916192e-06, "loss": 0.3288, "step": 9607 }, { "epoch": 0.6642239889388178, "grad_norm": 0.7587845538032538, "learning_rate": 7.791972293837589e-06, "loss": 0.3987, "step": 9608 }, { "epoch": 0.6642931213273419, "grad_norm": 0.7910304705391894, "learning_rate": 7.791507913766595e-06, "loss": 0.3316, "step": 9609 }, { "epoch": 0.6643622537158659, "grad_norm": 0.7320221901292743, "learning_rate": 7.791043498709031e-06, "loss": 0.2585, "step": 9610 }, { "epoch": 0.6644313861043899, "grad_norm": 0.6859592284099657, "learning_rate": 7.790579048670719e-06, "loss": 0.3234, "step": 9611 }, { "epoch": 0.6645005184929139, "grad_norm": 0.6480176936361052, "learning_rate": 7.790114563657477e-06, "loss": 0.3826, "step": 9612 }, { "epoch": 0.6645696508814379, "grad_norm": 0.6447263637598515, "learning_rate": 7.789650043675129e-06, "loss": 0.3048, "step": 9613 }, { "epoch": 0.664638783269962, "grad_norm": 0.7598299896242177, "learning_rate": 7.789185488729497e-06, "loss": 0.296, "step": 9614 }, { "epoch": 0.664707915658486, "grad_norm": 0.689806120763988, "learning_rate": 7.7887208988264e-06, "loss": 0.3868, "step": 9615 }, { "epoch": 0.66477704804701, "grad_norm": 0.793723617681514, "learning_rate": 7.788256273971666e-06, "loss": 0.2756, "step": 9616 }, { "epoch": 0.664846180435534, "grad_norm": 0.663966540927037, "learning_rate": 7.787791614171112e-06, "loss": 0.2909, "step": 9617 }, { "epoch": 0.664915312824058, "grad_norm": 0.7821066841081084, "learning_rate": 7.78732691943057e-06, "loss": 0.2871, "step": 9618 }, { "epoch": 0.664984445212582, "grad_norm": 0.6508618727886756, "learning_rate": 7.786862189755856e-06, "loss": 0.3729, "step": 9619 }, { "epoch": 0.6650535776011062, "grad_norm": 0.7284044087043405, "learning_rate": 7.786397425152798e-06, "loss": 0.3015, "step": 9620 }, { "epoch": 0.6651227099896302, "grad_norm": 0.8662615225274054, "learning_rate": 7.78593262562722e-06, "loss": 0.356, "step": 9621 }, { "epoch": 0.6651918423781542, "grad_norm": 0.6916608116141729, "learning_rate": 7.785467791184951e-06, "loss": 0.3943, "step": 9622 }, { "epoch": 0.6652609747666782, "grad_norm": 0.6486076753735478, "learning_rate": 7.78500292183181e-06, "loss": 0.3435, "step": 9623 }, { "epoch": 0.6653301071552022, "grad_norm": 0.7204805947200107, "learning_rate": 7.78453801757363e-06, "loss": 0.2836, "step": 9624 }, { "epoch": 0.6653992395437263, "grad_norm": 0.6966695857366433, "learning_rate": 7.784073078416237e-06, "loss": 0.3468, "step": 9625 }, { "epoch": 0.6654683719322503, "grad_norm": 0.7683268105520412, "learning_rate": 7.783608104365455e-06, "loss": 0.4256, "step": 9626 }, { "epoch": 0.6655375043207743, "grad_norm": 0.7098698112069403, "learning_rate": 7.78314309542711e-06, "loss": 0.3134, "step": 9627 }, { "epoch": 0.6656066367092983, "grad_norm": 0.5906888385980978, "learning_rate": 7.782678051607036e-06, "loss": 0.2488, "step": 9628 }, { "epoch": 0.6656757690978223, "grad_norm": 0.6992882435125447, "learning_rate": 7.782212972911058e-06, "loss": 0.474, "step": 9629 }, { "epoch": 0.6657449014863464, "grad_norm": 0.6394485094208575, "learning_rate": 7.781747859345005e-06, "loss": 0.3971, "step": 9630 }, { "epoch": 0.6658140338748704, "grad_norm": 0.9350977013709585, "learning_rate": 7.781282710914709e-06, "loss": 0.3342, "step": 9631 }, { "epoch": 0.6658831662633944, "grad_norm": 0.8096310228554284, "learning_rate": 7.780817527625996e-06, "loss": 0.292, "step": 9632 }, { "epoch": 0.6659522986519184, "grad_norm": 0.7295541977288512, "learning_rate": 7.780352309484698e-06, "loss": 0.4735, "step": 9633 }, { "epoch": 0.6660214310404424, "grad_norm": 0.67340661776295, "learning_rate": 7.779887056496645e-06, "loss": 0.3865, "step": 9634 }, { "epoch": 0.6660905634289664, "grad_norm": 0.6447801484686796, "learning_rate": 7.77942176866767e-06, "loss": 0.2479, "step": 9635 }, { "epoch": 0.6661596958174905, "grad_norm": 0.7584604988639143, "learning_rate": 7.778956446003602e-06, "loss": 0.3494, "step": 9636 }, { "epoch": 0.6662288282060145, "grad_norm": 0.7140657868241397, "learning_rate": 7.778491088510273e-06, "loss": 0.4352, "step": 9637 }, { "epoch": 0.6662979605945385, "grad_norm": 0.7129709532280988, "learning_rate": 7.778025696193519e-06, "loss": 0.2917, "step": 9638 }, { "epoch": 0.6663670929830625, "grad_norm": 0.7139961260247786, "learning_rate": 7.77756026905917e-06, "loss": 0.3429, "step": 9639 }, { "epoch": 0.6664362253715866, "grad_norm": 0.681304290721703, "learning_rate": 7.77709480711306e-06, "loss": 0.4526, "step": 9640 }, { "epoch": 0.6665053577601107, "grad_norm": 0.6340161352676803, "learning_rate": 7.77662931036102e-06, "loss": 0.3244, "step": 9641 }, { "epoch": 0.6665744901486347, "grad_norm": 0.6558882498445465, "learning_rate": 7.776163778808891e-06, "loss": 0.2863, "step": 9642 }, { "epoch": 0.6666436225371587, "grad_norm": 0.7277718615062599, "learning_rate": 7.775698212462501e-06, "loss": 0.377, "step": 9643 }, { "epoch": 0.6667127549256827, "grad_norm": 0.7926118748383719, "learning_rate": 7.775232611327687e-06, "loss": 0.3974, "step": 9644 }, { "epoch": 0.6667818873142067, "grad_norm": 0.8123104755338422, "learning_rate": 7.774766975410286e-06, "loss": 0.3498, "step": 9645 }, { "epoch": 0.6668510197027308, "grad_norm": 0.7007789332923783, "learning_rate": 7.774301304716132e-06, "loss": 0.2928, "step": 9646 }, { "epoch": 0.6669201520912548, "grad_norm": 0.7057471815351963, "learning_rate": 7.773835599251062e-06, "loss": 0.4653, "step": 9647 }, { "epoch": 0.6669892844797788, "grad_norm": 0.6659374744293193, "learning_rate": 7.773369859020912e-06, "loss": 0.2812, "step": 9648 }, { "epoch": 0.6670584168683028, "grad_norm": 0.6895819775805824, "learning_rate": 7.772904084031522e-06, "loss": 0.33, "step": 9649 }, { "epoch": 0.6671275492568268, "grad_norm": 0.6990196690964016, "learning_rate": 7.772438274288727e-06, "loss": 0.3678, "step": 9650 }, { "epoch": 0.6671966816453508, "grad_norm": 0.9594191659561111, "learning_rate": 7.771972429798366e-06, "loss": 0.4381, "step": 9651 }, { "epoch": 0.6672658140338749, "grad_norm": 0.637641631806831, "learning_rate": 7.771506550566277e-06, "loss": 0.2897, "step": 9652 }, { "epoch": 0.6673349464223989, "grad_norm": 0.6337737973994664, "learning_rate": 7.7710406365983e-06, "loss": 0.2513, "step": 9653 }, { "epoch": 0.6674040788109229, "grad_norm": 0.6927148140616576, "learning_rate": 7.770574687900273e-06, "loss": 0.3593, "step": 9654 }, { "epoch": 0.6674732111994469, "grad_norm": 0.7413857192511458, "learning_rate": 7.770108704478036e-06, "loss": 0.3497, "step": 9655 }, { "epoch": 0.6675423435879709, "grad_norm": 0.6633215206009901, "learning_rate": 7.769642686337432e-06, "loss": 0.2699, "step": 9656 }, { "epoch": 0.667611475976495, "grad_norm": 0.7580224139349514, "learning_rate": 7.769176633484299e-06, "loss": 0.3935, "step": 9657 }, { "epoch": 0.667680608365019, "grad_norm": 0.7412543127834456, "learning_rate": 7.768710545924477e-06, "loss": 0.4156, "step": 9658 }, { "epoch": 0.667749740753543, "grad_norm": 0.7324508014570587, "learning_rate": 7.768244423663812e-06, "loss": 0.3036, "step": 9659 }, { "epoch": 0.667818873142067, "grad_norm": 0.7574263699613522, "learning_rate": 7.76777826670814e-06, "loss": 0.2993, "step": 9660 }, { "epoch": 0.6678880055305911, "grad_norm": 0.6921054053300886, "learning_rate": 7.76731207506331e-06, "loss": 0.3889, "step": 9661 }, { "epoch": 0.6679571379191152, "grad_norm": 0.6922036666091211, "learning_rate": 7.766845848735158e-06, "loss": 0.4358, "step": 9662 }, { "epoch": 0.6680262703076392, "grad_norm": 0.8094923952401115, "learning_rate": 7.766379587729535e-06, "loss": 0.3101, "step": 9663 }, { "epoch": 0.6680954026961632, "grad_norm": 0.7657669811936086, "learning_rate": 7.765913292052278e-06, "loss": 0.3132, "step": 9664 }, { "epoch": 0.6681645350846872, "grad_norm": 0.7174861989090378, "learning_rate": 7.765446961709236e-06, "loss": 0.4235, "step": 9665 }, { "epoch": 0.6682336674732112, "grad_norm": 0.8668104003742657, "learning_rate": 7.764980596706249e-06, "loss": 0.3157, "step": 9666 }, { "epoch": 0.6683027998617352, "grad_norm": 0.6870016610454924, "learning_rate": 7.764514197049164e-06, "loss": 0.3857, "step": 9667 }, { "epoch": 0.6683719322502593, "grad_norm": 0.6994539749848596, "learning_rate": 7.764047762743829e-06, "loss": 0.3142, "step": 9668 }, { "epoch": 0.6684410646387833, "grad_norm": 0.7281296242949971, "learning_rate": 7.76358129379609e-06, "loss": 0.4601, "step": 9669 }, { "epoch": 0.6685101970273073, "grad_norm": 0.6227986501734328, "learning_rate": 7.763114790211787e-06, "loss": 0.3255, "step": 9670 }, { "epoch": 0.6685793294158313, "grad_norm": 0.6400130194844752, "learning_rate": 7.762648251996775e-06, "loss": 0.2719, "step": 9671 }, { "epoch": 0.6686484618043553, "grad_norm": 0.8134002502655613, "learning_rate": 7.762181679156895e-06, "loss": 0.3846, "step": 9672 }, { "epoch": 0.6687175941928793, "grad_norm": 0.6429475899083238, "learning_rate": 7.761715071697997e-06, "loss": 0.2922, "step": 9673 }, { "epoch": 0.6687867265814034, "grad_norm": 0.633123959117364, "learning_rate": 7.76124842962593e-06, "loss": 0.2519, "step": 9674 }, { "epoch": 0.6688558589699274, "grad_norm": 0.7205019840438233, "learning_rate": 7.76078175294654e-06, "loss": 0.3801, "step": 9675 }, { "epoch": 0.6689249913584514, "grad_norm": 0.695661469860621, "learning_rate": 7.760315041665678e-06, "loss": 0.4309, "step": 9676 }, { "epoch": 0.6689941237469754, "grad_norm": 0.6762223046178467, "learning_rate": 7.759848295789193e-06, "loss": 0.3538, "step": 9677 }, { "epoch": 0.6690632561354994, "grad_norm": 0.6279805872327232, "learning_rate": 7.759381515322936e-06, "loss": 0.2966, "step": 9678 }, { "epoch": 0.6691323885240235, "grad_norm": 0.6434045345249878, "learning_rate": 7.758914700272756e-06, "loss": 0.3647, "step": 9679 }, { "epoch": 0.6692015209125475, "grad_norm": 0.6320397606596382, "learning_rate": 7.758447850644503e-06, "loss": 0.3929, "step": 9680 }, { "epoch": 0.6692706533010716, "grad_norm": 0.6115090278590842, "learning_rate": 7.75798096644403e-06, "loss": 0.2733, "step": 9681 }, { "epoch": 0.6693397856895956, "grad_norm": 0.6780666875720153, "learning_rate": 7.757514047677186e-06, "loss": 0.3262, "step": 9682 }, { "epoch": 0.6694089180781196, "grad_norm": 0.6909966340354659, "learning_rate": 7.757047094349825e-06, "loss": 0.4253, "step": 9683 }, { "epoch": 0.6694780504666437, "grad_norm": 0.7319502912180612, "learning_rate": 7.7565801064678e-06, "loss": 0.3295, "step": 9684 }, { "epoch": 0.6695471828551677, "grad_norm": 0.6403047200120611, "learning_rate": 7.75611308403696e-06, "loss": 0.311, "step": 9685 }, { "epoch": 0.6696163152436917, "grad_norm": 0.7578453727912401, "learning_rate": 7.755646027063163e-06, "loss": 0.3985, "step": 9686 }, { "epoch": 0.6696854476322157, "grad_norm": 0.6675256011440025, "learning_rate": 7.75517893555226e-06, "loss": 0.3434, "step": 9687 }, { "epoch": 0.6697545800207397, "grad_norm": 0.6667634810980698, "learning_rate": 7.754711809510107e-06, "loss": 0.3846, "step": 9688 }, { "epoch": 0.6698237124092637, "grad_norm": 0.6365095017756135, "learning_rate": 7.754244648942554e-06, "loss": 0.3547, "step": 9689 }, { "epoch": 0.6698928447977878, "grad_norm": 0.7544915800312971, "learning_rate": 7.753777453855463e-06, "loss": 0.3609, "step": 9690 }, { "epoch": 0.6699619771863118, "grad_norm": 0.6619881413264674, "learning_rate": 7.753310224254685e-06, "loss": 0.3567, "step": 9691 }, { "epoch": 0.6700311095748358, "grad_norm": 0.6427414101981077, "learning_rate": 7.752842960146076e-06, "loss": 0.3444, "step": 9692 }, { "epoch": 0.6701002419633598, "grad_norm": 1.052274608857896, "learning_rate": 7.752375661535495e-06, "loss": 0.3559, "step": 9693 }, { "epoch": 0.6701693743518838, "grad_norm": 0.7323206828397679, "learning_rate": 7.751908328428797e-06, "loss": 0.403, "step": 9694 }, { "epoch": 0.6702385067404079, "grad_norm": 0.6777149783837837, "learning_rate": 7.751440960831837e-06, "loss": 0.3198, "step": 9695 }, { "epoch": 0.6703076391289319, "grad_norm": 0.6915790433157184, "learning_rate": 7.750973558750475e-06, "loss": 0.2916, "step": 9696 }, { "epoch": 0.6703767715174559, "grad_norm": 0.7402208707206231, "learning_rate": 7.75050612219057e-06, "loss": 0.4144, "step": 9697 }, { "epoch": 0.6704459039059799, "grad_norm": 0.7228329800836916, "learning_rate": 7.750038651157978e-06, "loss": 0.3752, "step": 9698 }, { "epoch": 0.6705150362945039, "grad_norm": 0.6223545151552529, "learning_rate": 7.74957114565856e-06, "loss": 0.2858, "step": 9699 }, { "epoch": 0.670584168683028, "grad_norm": 0.6671072187195792, "learning_rate": 7.749103605698175e-06, "loss": 0.3287, "step": 9700 }, { "epoch": 0.6706533010715521, "grad_norm": 0.7051819348743106, "learning_rate": 7.748636031282681e-06, "loss": 0.425, "step": 9701 }, { "epoch": 0.6707224334600761, "grad_norm": 0.713049265549843, "learning_rate": 7.74816842241794e-06, "loss": 0.3384, "step": 9702 }, { "epoch": 0.6707915658486001, "grad_norm": 1.1510499603473556, "learning_rate": 7.747700779109813e-06, "loss": 0.2858, "step": 9703 }, { "epoch": 0.6708606982371241, "grad_norm": 0.7036933693462235, "learning_rate": 7.74723310136416e-06, "loss": 0.378, "step": 9704 }, { "epoch": 0.6709298306256481, "grad_norm": 0.6656333713309215, "learning_rate": 7.746765389186843e-06, "loss": 0.3013, "step": 9705 }, { "epoch": 0.6709989630141722, "grad_norm": 0.7173009692577271, "learning_rate": 7.746297642583723e-06, "loss": 0.302, "step": 9706 }, { "epoch": 0.6710680954026962, "grad_norm": 0.6410080626136004, "learning_rate": 7.745829861560662e-06, "loss": 0.2815, "step": 9707 }, { "epoch": 0.6711372277912202, "grad_norm": 0.6601541388393223, "learning_rate": 7.745362046123526e-06, "loss": 0.3938, "step": 9708 }, { "epoch": 0.6712063601797442, "grad_norm": 0.6982954963358464, "learning_rate": 7.744894196278174e-06, "loss": 0.2969, "step": 9709 }, { "epoch": 0.6712754925682682, "grad_norm": 0.6443494893269658, "learning_rate": 7.744426312030473e-06, "loss": 0.2439, "step": 9710 }, { "epoch": 0.6713446249567923, "grad_norm": 0.9356214512245936, "learning_rate": 7.743958393386286e-06, "loss": 0.3817, "step": 9711 }, { "epoch": 0.6714137573453163, "grad_norm": 0.8708378808770139, "learning_rate": 7.743490440351476e-06, "loss": 0.375, "step": 9712 }, { "epoch": 0.6714828897338403, "grad_norm": 0.6401476053435022, "learning_rate": 7.743022452931909e-06, "loss": 0.2467, "step": 9713 }, { "epoch": 0.6715520221223643, "grad_norm": 0.7474346648213972, "learning_rate": 7.742554431133453e-06, "loss": 0.3307, "step": 9714 }, { "epoch": 0.6716211545108883, "grad_norm": 0.6889081550385311, "learning_rate": 7.74208637496197e-06, "loss": 0.4773, "step": 9715 }, { "epoch": 0.6716902868994123, "grad_norm": 0.7690085752740025, "learning_rate": 7.74161828442333e-06, "loss": 0.3825, "step": 9716 }, { "epoch": 0.6717594192879364, "grad_norm": 0.6405839381880177, "learning_rate": 7.741150159523393e-06, "loss": 0.2847, "step": 9717 }, { "epoch": 0.6718285516764604, "grad_norm": 0.8318883266708177, "learning_rate": 7.740682000268034e-06, "loss": 0.336, "step": 9718 }, { "epoch": 0.6718976840649844, "grad_norm": 0.6692872119979155, "learning_rate": 7.740213806663118e-06, "loss": 0.3945, "step": 9719 }, { "epoch": 0.6719668164535084, "grad_norm": 0.873313970976676, "learning_rate": 7.739745578714508e-06, "loss": 0.3186, "step": 9720 }, { "epoch": 0.6720359488420324, "grad_norm": 0.6320644286594882, "learning_rate": 7.73927731642808e-06, "loss": 0.2388, "step": 9721 }, { "epoch": 0.6721050812305566, "grad_norm": 0.6459410596101304, "learning_rate": 7.738809019809696e-06, "loss": 0.4138, "step": 9722 }, { "epoch": 0.6721742136190806, "grad_norm": 0.7148974486329344, "learning_rate": 7.738340688865233e-06, "loss": 0.339, "step": 9723 }, { "epoch": 0.6722433460076046, "grad_norm": 0.7056413039192923, "learning_rate": 7.737872323600553e-06, "loss": 0.2987, "step": 9724 }, { "epoch": 0.6723124783961286, "grad_norm": 0.7722073325346587, "learning_rate": 7.73740392402153e-06, "loss": 0.2863, "step": 9725 }, { "epoch": 0.6723816107846526, "grad_norm": 0.6636734706599534, "learning_rate": 7.736935490134034e-06, "loss": 0.352, "step": 9726 }, { "epoch": 0.6724507431731767, "grad_norm": 0.7070238349410252, "learning_rate": 7.736467021943936e-06, "loss": 0.2898, "step": 9727 }, { "epoch": 0.6725198755617007, "grad_norm": 0.7184701387244274, "learning_rate": 7.735998519457108e-06, "loss": 0.308, "step": 9728 }, { "epoch": 0.6725890079502247, "grad_norm": 0.7610298177129019, "learning_rate": 7.73552998267942e-06, "loss": 0.3853, "step": 9729 }, { "epoch": 0.6726581403387487, "grad_norm": 1.015206011331689, "learning_rate": 7.735061411616746e-06, "loss": 0.3823, "step": 9730 }, { "epoch": 0.6727272727272727, "grad_norm": 1.2697528152001376, "learning_rate": 7.734592806274959e-06, "loss": 0.2736, "step": 9731 }, { "epoch": 0.6727964051157967, "grad_norm": 0.6939305466715078, "learning_rate": 7.734124166659931e-06, "loss": 0.3054, "step": 9732 }, { "epoch": 0.6728655375043208, "grad_norm": 0.6790748578127478, "learning_rate": 7.733655492777535e-06, "loss": 0.4151, "step": 9733 }, { "epoch": 0.6729346698928448, "grad_norm": 0.6248280243805275, "learning_rate": 7.733186784633647e-06, "loss": 0.3134, "step": 9734 }, { "epoch": 0.6730038022813688, "grad_norm": 0.6934338613694021, "learning_rate": 7.732718042234137e-06, "loss": 0.3225, "step": 9735 }, { "epoch": 0.6730729346698928, "grad_norm": 0.7178415461129586, "learning_rate": 7.732249265584886e-06, "loss": 0.3864, "step": 9736 }, { "epoch": 0.6731420670584168, "grad_norm": 0.9120959619199768, "learning_rate": 7.731780454691767e-06, "loss": 0.4417, "step": 9737 }, { "epoch": 0.6732111994469409, "grad_norm": 0.661307060766059, "learning_rate": 7.731311609560654e-06, "loss": 0.2756, "step": 9738 }, { "epoch": 0.6732803318354649, "grad_norm": 0.5919536070092528, "learning_rate": 7.730842730197423e-06, "loss": 0.2412, "step": 9739 }, { "epoch": 0.6733494642239889, "grad_norm": 0.6517278061164259, "learning_rate": 7.730373816607952e-06, "loss": 0.4182, "step": 9740 }, { "epoch": 0.6734185966125129, "grad_norm": 0.7421072043023822, "learning_rate": 7.729904868798119e-06, "loss": 0.2878, "step": 9741 }, { "epoch": 0.673487729001037, "grad_norm": 0.5756927078583506, "learning_rate": 7.7294358867738e-06, "loss": 0.2848, "step": 9742 }, { "epoch": 0.6735568613895611, "grad_norm": 0.6983164075214824, "learning_rate": 7.728966870540873e-06, "loss": 0.3597, "step": 9743 }, { "epoch": 0.6736259937780851, "grad_norm": 0.7102530977250002, "learning_rate": 7.728497820105217e-06, "loss": 0.3955, "step": 9744 }, { "epoch": 0.6736951261666091, "grad_norm": 0.7295812024924573, "learning_rate": 7.728028735472709e-06, "loss": 0.3357, "step": 9745 }, { "epoch": 0.6737642585551331, "grad_norm": 0.6971722087314246, "learning_rate": 7.72755961664923e-06, "loss": 0.2982, "step": 9746 }, { "epoch": 0.6738333909436571, "grad_norm": 0.7025565219083878, "learning_rate": 7.727090463640658e-06, "loss": 0.4285, "step": 9747 }, { "epoch": 0.6739025233321811, "grad_norm": 0.6548613639606274, "learning_rate": 7.726621276452874e-06, "loss": 0.3143, "step": 9748 }, { "epoch": 0.6739716557207052, "grad_norm": 0.8167046716654606, "learning_rate": 7.726152055091757e-06, "loss": 0.2954, "step": 9749 }, { "epoch": 0.6740407881092292, "grad_norm": 0.6813445316169144, "learning_rate": 7.725682799563192e-06, "loss": 0.2841, "step": 9750 }, { "epoch": 0.6741099204977532, "grad_norm": 0.6418928969851467, "learning_rate": 7.725213509873055e-06, "loss": 0.3863, "step": 9751 }, { "epoch": 0.6741790528862772, "grad_norm": 0.7285056221184719, "learning_rate": 7.72474418602723e-06, "loss": 0.3365, "step": 9752 }, { "epoch": 0.6742481852748012, "grad_norm": 0.764224203906242, "learning_rate": 7.7242748280316e-06, "loss": 0.2628, "step": 9753 }, { "epoch": 0.6743173176633253, "grad_norm": 0.7584668009618086, "learning_rate": 7.723805435892046e-06, "loss": 0.3698, "step": 9754 }, { "epoch": 0.6743864500518493, "grad_norm": 0.7399481798651544, "learning_rate": 7.723336009614452e-06, "loss": 0.3546, "step": 9755 }, { "epoch": 0.6744555824403733, "grad_norm": 0.5909219674537955, "learning_rate": 7.722866549204701e-06, "loss": 0.2569, "step": 9756 }, { "epoch": 0.6745247148288973, "grad_norm": 0.7203188865358972, "learning_rate": 7.722397054668678e-06, "loss": 0.3028, "step": 9757 }, { "epoch": 0.6745938472174213, "grad_norm": 0.6533134308752249, "learning_rate": 7.721927526012267e-06, "loss": 0.4011, "step": 9758 }, { "epoch": 0.6746629796059453, "grad_norm": 0.6801200917500392, "learning_rate": 7.721457963241349e-06, "loss": 0.3228, "step": 9759 }, { "epoch": 0.6747321119944694, "grad_norm": 0.6457022645441425, "learning_rate": 7.720988366361814e-06, "loss": 0.2524, "step": 9760 }, { "epoch": 0.6748012443829934, "grad_norm": 0.6647482788337945, "learning_rate": 7.720518735379545e-06, "loss": 0.3927, "step": 9761 }, { "epoch": 0.6748703767715175, "grad_norm": 0.7059642793450265, "learning_rate": 7.720049070300427e-06, "loss": 0.4186, "step": 9762 }, { "epoch": 0.6749395091600415, "grad_norm": 0.6778089885442116, "learning_rate": 7.719579371130353e-06, "loss": 0.297, "step": 9763 }, { "epoch": 0.6750086415485655, "grad_norm": 0.8179672833827519, "learning_rate": 7.7191096378752e-06, "loss": 0.3726, "step": 9764 }, { "epoch": 0.6750777739370896, "grad_norm": 0.7077079660474932, "learning_rate": 7.718639870540862e-06, "loss": 0.4809, "step": 9765 }, { "epoch": 0.6751469063256136, "grad_norm": 0.7258431203938421, "learning_rate": 7.718170069133225e-06, "loss": 0.3289, "step": 9766 }, { "epoch": 0.6752160387141376, "grad_norm": 0.6079245028956457, "learning_rate": 7.717700233658176e-06, "loss": 0.2551, "step": 9767 }, { "epoch": 0.6752851711026616, "grad_norm": 0.6953660528724117, "learning_rate": 7.717230364121607e-06, "loss": 0.3332, "step": 9768 }, { "epoch": 0.6753543034911856, "grad_norm": 0.6752155436785908, "learning_rate": 7.716760460529402e-06, "loss": 0.3916, "step": 9769 }, { "epoch": 0.6754234358797097, "grad_norm": 1.1645825123003457, "learning_rate": 7.716290522887454e-06, "loss": 0.3081, "step": 9770 }, { "epoch": 0.6754925682682337, "grad_norm": 0.6964394776302705, "learning_rate": 7.715820551201652e-06, "loss": 0.274, "step": 9771 }, { "epoch": 0.6755617006567577, "grad_norm": 0.703777691710221, "learning_rate": 7.715350545477884e-06, "loss": 0.4531, "step": 9772 }, { "epoch": 0.6756308330452817, "grad_norm": 0.7286756242732484, "learning_rate": 7.714880505722044e-06, "loss": 0.3423, "step": 9773 }, { "epoch": 0.6756999654338057, "grad_norm": 0.588259838648078, "learning_rate": 7.714410431940022e-06, "loss": 0.3111, "step": 9774 }, { "epoch": 0.6757690978223297, "grad_norm": 0.7401566687732399, "learning_rate": 7.71394032413771e-06, "loss": 0.3245, "step": 9775 }, { "epoch": 0.6758382302108538, "grad_norm": 0.7114703211979961, "learning_rate": 7.713470182320997e-06, "loss": 0.4594, "step": 9776 }, { "epoch": 0.6759073625993778, "grad_norm": 0.6478127774747252, "learning_rate": 7.71300000649578e-06, "loss": 0.3215, "step": 9777 }, { "epoch": 0.6759764949879018, "grad_norm": 1.1470155620488622, "learning_rate": 7.712529796667949e-06, "loss": 0.3098, "step": 9778 }, { "epoch": 0.6760456273764258, "grad_norm": 0.8263096484773876, "learning_rate": 7.712059552843397e-06, "loss": 0.3606, "step": 9779 }, { "epoch": 0.6761147597649498, "grad_norm": 0.7057331090083068, "learning_rate": 7.71158927502802e-06, "loss": 0.4042, "step": 9780 }, { "epoch": 0.6761838921534739, "grad_norm": 0.7759126527168455, "learning_rate": 7.711118963227708e-06, "loss": 0.2984, "step": 9781 }, { "epoch": 0.6762530245419979, "grad_norm": 0.6975739945475717, "learning_rate": 7.71064861744836e-06, "loss": 0.3641, "step": 9782 }, { "epoch": 0.676322156930522, "grad_norm": 0.6534876134500153, "learning_rate": 7.710178237695867e-06, "loss": 0.3642, "step": 9783 }, { "epoch": 0.676391289319046, "grad_norm": 1.6941995777749839, "learning_rate": 7.709707823976127e-06, "loss": 0.2952, "step": 9784 }, { "epoch": 0.67646042170757, "grad_norm": 0.762651941037823, "learning_rate": 7.709237376295036e-06, "loss": 0.2767, "step": 9785 }, { "epoch": 0.676529554096094, "grad_norm": 0.6803957707820265, "learning_rate": 7.708766894658488e-06, "loss": 0.3711, "step": 9786 }, { "epoch": 0.6765986864846181, "grad_norm": 0.9270348905828627, "learning_rate": 7.708296379072383e-06, "loss": 0.3207, "step": 9787 }, { "epoch": 0.6766678188731421, "grad_norm": 0.6975305655114157, "learning_rate": 7.707825829542614e-06, "loss": 0.3034, "step": 9788 }, { "epoch": 0.6767369512616661, "grad_norm": 0.7335272356048788, "learning_rate": 7.707355246075084e-06, "loss": 0.3175, "step": 9789 }, { "epoch": 0.6768060836501901, "grad_norm": 0.788829315847754, "learning_rate": 7.706884628675684e-06, "loss": 0.465, "step": 9790 }, { "epoch": 0.6768752160387141, "grad_norm": 0.8784178318075202, "learning_rate": 7.706413977350317e-06, "loss": 0.3267, "step": 9791 }, { "epoch": 0.6769443484272382, "grad_norm": 0.766754479893676, "learning_rate": 7.70594329210488e-06, "loss": 0.3055, "step": 9792 }, { "epoch": 0.6770134808157622, "grad_norm": 0.6854618826916473, "learning_rate": 7.705472572945272e-06, "loss": 0.3545, "step": 9793 }, { "epoch": 0.6770826132042862, "grad_norm": 0.7387553785120937, "learning_rate": 7.705001819877396e-06, "loss": 0.4297, "step": 9794 }, { "epoch": 0.6771517455928102, "grad_norm": 0.6069968111296853, "learning_rate": 7.704531032907147e-06, "loss": 0.3063, "step": 9795 }, { "epoch": 0.6772208779813342, "grad_norm": 0.7596002914708875, "learning_rate": 7.70406021204043e-06, "loss": 0.4, "step": 9796 }, { "epoch": 0.6772900103698583, "grad_norm": 0.7495119322456122, "learning_rate": 7.703589357283144e-06, "loss": 0.3721, "step": 9797 }, { "epoch": 0.6773591427583823, "grad_norm": 0.6600442377520632, "learning_rate": 7.703118468641188e-06, "loss": 0.3184, "step": 9798 }, { "epoch": 0.6774282751469063, "grad_norm": 0.6533964187601652, "learning_rate": 7.702647546120468e-06, "loss": 0.2575, "step": 9799 }, { "epoch": 0.6774974075354303, "grad_norm": 0.664969646530706, "learning_rate": 7.702176589726882e-06, "loss": 0.3344, "step": 9800 }, { "epoch": 0.6775665399239543, "grad_norm": 0.6888369059090602, "learning_rate": 7.701705599466337e-06, "loss": 0.4006, "step": 9801 }, { "epoch": 0.6776356723124783, "grad_norm": 0.7038187511830293, "learning_rate": 7.701234575344732e-06, "loss": 0.3457, "step": 9802 }, { "epoch": 0.6777048047010025, "grad_norm": 0.6478105070523439, "learning_rate": 7.700763517367973e-06, "loss": 0.2503, "step": 9803 }, { "epoch": 0.6777739370895265, "grad_norm": 0.7853785565454091, "learning_rate": 7.700292425541963e-06, "loss": 0.3965, "step": 9804 }, { "epoch": 0.6778430694780505, "grad_norm": 0.6716673579369573, "learning_rate": 7.699821299872606e-06, "loss": 0.3509, "step": 9805 }, { "epoch": 0.6779122018665745, "grad_norm": 0.7011120888940575, "learning_rate": 7.699350140365807e-06, "loss": 0.2787, "step": 9806 }, { "epoch": 0.6779813342550985, "grad_norm": 0.6191920728055382, "learning_rate": 7.69887894702747e-06, "loss": 0.3134, "step": 9807 }, { "epoch": 0.6780504666436226, "grad_norm": 0.6695896491917442, "learning_rate": 7.698407719863505e-06, "loss": 0.3998, "step": 9808 }, { "epoch": 0.6781195990321466, "grad_norm": 0.638292540394632, "learning_rate": 7.697936458879813e-06, "loss": 0.2519, "step": 9809 }, { "epoch": 0.6781887314206706, "grad_norm": 0.7546057755546799, "learning_rate": 7.697465164082301e-06, "loss": 0.2811, "step": 9810 }, { "epoch": 0.6782578638091946, "grad_norm": 0.7645661552824866, "learning_rate": 7.696993835476879e-06, "loss": 0.3533, "step": 9811 }, { "epoch": 0.6783269961977186, "grad_norm": 0.7027088462412089, "learning_rate": 7.696522473069451e-06, "loss": 0.3686, "step": 9812 }, { "epoch": 0.6783961285862427, "grad_norm": 0.6708735388486529, "learning_rate": 7.696051076865928e-06, "loss": 0.2958, "step": 9813 }, { "epoch": 0.6784652609747667, "grad_norm": 0.6996042810497509, "learning_rate": 7.695579646872213e-06, "loss": 0.2698, "step": 9814 }, { "epoch": 0.6785343933632907, "grad_norm": 0.994377007306895, "learning_rate": 7.69510818309422e-06, "loss": 0.4817, "step": 9815 }, { "epoch": 0.6786035257518147, "grad_norm": 0.7350288285585866, "learning_rate": 7.694636685537855e-06, "loss": 0.3572, "step": 9816 }, { "epoch": 0.6786726581403387, "grad_norm": 0.6731362218135655, "learning_rate": 7.694165154209025e-06, "loss": 0.2917, "step": 9817 }, { "epoch": 0.6787417905288627, "grad_norm": 0.8675630453263077, "learning_rate": 7.693693589113647e-06, "loss": 0.3852, "step": 9818 }, { "epoch": 0.6788109229173868, "grad_norm": 0.6875765532138068, "learning_rate": 7.693221990257625e-06, "loss": 0.397, "step": 9819 }, { "epoch": 0.6788800553059108, "grad_norm": 0.8039117194077685, "learning_rate": 7.692750357646872e-06, "loss": 0.306, "step": 9820 }, { "epoch": 0.6789491876944348, "grad_norm": 1.191217684613082, "learning_rate": 7.692278691287297e-06, "loss": 0.2507, "step": 9821 }, { "epoch": 0.6790183200829588, "grad_norm": 0.6996682268544039, "learning_rate": 7.691806991184815e-06, "loss": 0.4262, "step": 9822 }, { "epoch": 0.6790874524714829, "grad_norm": 0.6865159981621733, "learning_rate": 7.691335257345338e-06, "loss": 0.3846, "step": 9823 }, { "epoch": 0.679156584860007, "grad_norm": 1.1880664564643486, "learning_rate": 7.690863489774774e-06, "loss": 0.3077, "step": 9824 }, { "epoch": 0.679225717248531, "grad_norm": 0.7809999321382296, "learning_rate": 7.690391688479037e-06, "loss": 0.3434, "step": 9825 }, { "epoch": 0.679294849637055, "grad_norm": 0.6896410862238326, "learning_rate": 7.689919853464044e-06, "loss": 0.4347, "step": 9826 }, { "epoch": 0.679363982025579, "grad_norm": 0.647763463674253, "learning_rate": 7.689447984735705e-06, "loss": 0.2862, "step": 9827 }, { "epoch": 0.679433114414103, "grad_norm": 2.050372108886137, "learning_rate": 7.688976082299935e-06, "loss": 0.3117, "step": 9828 }, { "epoch": 0.679502246802627, "grad_norm": 0.72292459693237, "learning_rate": 7.688504146162649e-06, "loss": 0.3738, "step": 9829 }, { "epoch": 0.6795713791911511, "grad_norm": 0.8708116500529759, "learning_rate": 7.68803217632976e-06, "loss": 0.3885, "step": 9830 }, { "epoch": 0.6796405115796751, "grad_norm": 0.8291411002436669, "learning_rate": 7.687560172807186e-06, "loss": 0.3114, "step": 9831 }, { "epoch": 0.6797096439681991, "grad_norm": 0.6502127594192458, "learning_rate": 7.68708813560084e-06, "loss": 0.3807, "step": 9832 }, { "epoch": 0.6797787763567231, "grad_norm": 0.8655265042422742, "learning_rate": 7.686616064716642e-06, "loss": 0.4019, "step": 9833 }, { "epoch": 0.6798479087452471, "grad_norm": 0.6888652621512686, "learning_rate": 7.686143960160504e-06, "loss": 0.3096, "step": 9834 }, { "epoch": 0.6799170411337712, "grad_norm": 0.6924766122012526, "learning_rate": 7.685671821938345e-06, "loss": 0.312, "step": 9835 }, { "epoch": 0.6799861735222952, "grad_norm": 0.6850770712677215, "learning_rate": 7.685199650056084e-06, "loss": 0.2918, "step": 9836 }, { "epoch": 0.6800553059108192, "grad_norm": 0.6991562188725593, "learning_rate": 7.684727444519638e-06, "loss": 0.3767, "step": 9837 }, { "epoch": 0.6801244382993432, "grad_norm": 0.6273699587401003, "learning_rate": 7.68425520533492e-06, "loss": 0.2985, "step": 9838 }, { "epoch": 0.6801935706878672, "grad_norm": 0.6988088468331795, "learning_rate": 7.683782932507857e-06, "loss": 0.3382, "step": 9839 }, { "epoch": 0.6802627030763913, "grad_norm": 0.7702836205737659, "learning_rate": 7.683310626044364e-06, "loss": 0.4651, "step": 9840 }, { "epoch": 0.6803318354649153, "grad_norm": 0.7662621715395361, "learning_rate": 7.682838285950362e-06, "loss": 0.3399, "step": 9841 }, { "epoch": 0.6804009678534393, "grad_norm": 0.6616363621092145, "learning_rate": 7.682365912231769e-06, "loss": 0.3008, "step": 9842 }, { "epoch": 0.6804701002419633, "grad_norm": 0.7259292440977526, "learning_rate": 7.681893504894504e-06, "loss": 0.2782, "step": 9843 }, { "epoch": 0.6805392326304874, "grad_norm": 0.6853186932567698, "learning_rate": 7.681421063944492e-06, "loss": 0.4555, "step": 9844 }, { "epoch": 0.6806083650190115, "grad_norm": 0.6618257417469411, "learning_rate": 7.680948589387654e-06, "loss": 0.2433, "step": 9845 }, { "epoch": 0.6806774974075355, "grad_norm": 0.6193224824333134, "learning_rate": 7.680476081229905e-06, "loss": 0.2778, "step": 9846 }, { "epoch": 0.6807466297960595, "grad_norm": 0.7905969488338218, "learning_rate": 7.680003539477176e-06, "loss": 0.3783, "step": 9847 }, { "epoch": 0.6808157621845835, "grad_norm": 0.6540618986262873, "learning_rate": 7.679530964135383e-06, "loss": 0.3129, "step": 9848 }, { "epoch": 0.6808848945731075, "grad_norm": 0.6060196007965093, "learning_rate": 7.679058355210452e-06, "loss": 0.2811, "step": 9849 }, { "epoch": 0.6809540269616315, "grad_norm": 0.7378559344090432, "learning_rate": 7.678585712708305e-06, "loss": 0.3264, "step": 9850 }, { "epoch": 0.6810231593501556, "grad_norm": 0.9163353969580489, "learning_rate": 7.678113036634867e-06, "loss": 0.3839, "step": 9851 }, { "epoch": 0.6810922917386796, "grad_norm": 1.4724463775744736, "learning_rate": 7.677640326996062e-06, "loss": 0.2623, "step": 9852 }, { "epoch": 0.6811614241272036, "grad_norm": 0.6705181100626847, "learning_rate": 7.677167583797812e-06, "loss": 0.3145, "step": 9853 }, { "epoch": 0.6812305565157276, "grad_norm": 0.715274217492676, "learning_rate": 7.676694807046046e-06, "loss": 0.4146, "step": 9854 }, { "epoch": 0.6812996889042516, "grad_norm": 0.7293350643351859, "learning_rate": 7.676221996746687e-06, "loss": 0.3934, "step": 9855 }, { "epoch": 0.6813688212927757, "grad_norm": 0.6494458046310683, "learning_rate": 7.67574915290566e-06, "loss": 0.3296, "step": 9856 }, { "epoch": 0.6814379536812997, "grad_norm": 0.6775568994146501, "learning_rate": 7.675276275528893e-06, "loss": 0.3335, "step": 9857 }, { "epoch": 0.6815070860698237, "grad_norm": 0.6954055493205558, "learning_rate": 7.674803364622311e-06, "loss": 0.3637, "step": 9858 }, { "epoch": 0.6815762184583477, "grad_norm": 0.922943633264535, "learning_rate": 7.674330420191845e-06, "loss": 0.2945, "step": 9859 }, { "epoch": 0.6816453508468717, "grad_norm": 0.6482818837505291, "learning_rate": 7.673857442243417e-06, "loss": 0.3017, "step": 9860 }, { "epoch": 0.6817144832353957, "grad_norm": 0.7535320655055777, "learning_rate": 7.673384430782959e-06, "loss": 0.4044, "step": 9861 }, { "epoch": 0.6817836156239198, "grad_norm": 0.6913823278588643, "learning_rate": 7.672911385816398e-06, "loss": 0.3533, "step": 9862 }, { "epoch": 0.6818527480124438, "grad_norm": 0.9976950563034053, "learning_rate": 7.672438307349663e-06, "loss": 0.3263, "step": 9863 }, { "epoch": 0.6819218804009679, "grad_norm": 0.775604203883369, "learning_rate": 7.671965195388683e-06, "loss": 0.3011, "step": 9864 }, { "epoch": 0.6819910127894919, "grad_norm": 0.7339441071692329, "learning_rate": 7.671492049939388e-06, "loss": 0.4628, "step": 9865 }, { "epoch": 0.6820601451780159, "grad_norm": 0.6701124864724497, "learning_rate": 7.671018871007707e-06, "loss": 0.3217, "step": 9866 }, { "epoch": 0.68212927756654, "grad_norm": 0.7777285623799478, "learning_rate": 7.67054565859957e-06, "loss": 0.2683, "step": 9867 }, { "epoch": 0.682198409955064, "grad_norm": 0.7282916866148006, "learning_rate": 7.67007241272091e-06, "loss": 0.4, "step": 9868 }, { "epoch": 0.682267542343588, "grad_norm": 0.6086685703788963, "learning_rate": 7.669599133377658e-06, "loss": 0.3435, "step": 9869 }, { "epoch": 0.682336674732112, "grad_norm": 0.5738072872485767, "learning_rate": 7.669125820575745e-06, "loss": 0.2689, "step": 9870 }, { "epoch": 0.682405807120636, "grad_norm": 0.6663229306529608, "learning_rate": 7.668652474321102e-06, "loss": 0.319, "step": 9871 }, { "epoch": 0.68247493950916, "grad_norm": 0.9394147858457026, "learning_rate": 7.668179094619662e-06, "loss": 0.43, "step": 9872 }, { "epoch": 0.6825440718976841, "grad_norm": 0.6944209925586557, "learning_rate": 7.667705681477359e-06, "loss": 0.3024, "step": 9873 }, { "epoch": 0.6826132042862081, "grad_norm": 0.6162025948384338, "learning_rate": 7.667232234900127e-06, "loss": 0.297, "step": 9874 }, { "epoch": 0.6826823366747321, "grad_norm": 0.6956153320887641, "learning_rate": 7.666758754893899e-06, "loss": 0.3649, "step": 9875 }, { "epoch": 0.6827514690632561, "grad_norm": 0.7354864978183533, "learning_rate": 7.666285241464608e-06, "loss": 0.4412, "step": 9876 }, { "epoch": 0.6828206014517801, "grad_norm": 0.7407725312617618, "learning_rate": 7.66581169461819e-06, "loss": 0.3117, "step": 9877 }, { "epoch": 0.6828897338403042, "grad_norm": 0.713703181756389, "learning_rate": 7.665338114360581e-06, "loss": 0.3168, "step": 9878 }, { "epoch": 0.6829588662288282, "grad_norm": 0.6841321182295796, "learning_rate": 7.664864500697714e-06, "loss": 0.4423, "step": 9879 }, { "epoch": 0.6830279986173522, "grad_norm": 0.66028406403197, "learning_rate": 7.664390853635528e-06, "loss": 0.335, "step": 9880 }, { "epoch": 0.6830971310058762, "grad_norm": 0.6583545916025287, "learning_rate": 7.663917173179956e-06, "loss": 0.3092, "step": 9881 }, { "epoch": 0.6831662633944002, "grad_norm": 0.7570599561541237, "learning_rate": 7.663443459336934e-06, "loss": 0.3624, "step": 9882 }, { "epoch": 0.6832353957829242, "grad_norm": 0.6755427614132494, "learning_rate": 7.662969712112404e-06, "loss": 0.4035, "step": 9883 }, { "epoch": 0.6833045281714484, "grad_norm": 0.6268141239057423, "learning_rate": 7.6624959315123e-06, "loss": 0.3384, "step": 9884 }, { "epoch": 0.6833736605599724, "grad_norm": 0.8235456363674336, "learning_rate": 7.662022117542563e-06, "loss": 0.2968, "step": 9885 }, { "epoch": 0.6834427929484964, "grad_norm": 0.742806477785081, "learning_rate": 7.661548270209127e-06, "loss": 0.342, "step": 9886 }, { "epoch": 0.6835119253370204, "grad_norm": 0.6882099459644708, "learning_rate": 7.661074389517934e-06, "loss": 0.3291, "step": 9887 }, { "epoch": 0.6835810577255444, "grad_norm": 0.6869564128585554, "learning_rate": 7.660600475474922e-06, "loss": 0.2575, "step": 9888 }, { "epoch": 0.6836501901140685, "grad_norm": 0.7569466142813093, "learning_rate": 7.66012652808603e-06, "loss": 0.3603, "step": 9889 }, { "epoch": 0.6837193225025925, "grad_norm": 0.6700712474901636, "learning_rate": 7.659652547357203e-06, "loss": 0.4424, "step": 9890 }, { "epoch": 0.6837884548911165, "grad_norm": 0.6993599556905006, "learning_rate": 7.659178533294374e-06, "loss": 0.2847, "step": 9891 }, { "epoch": 0.6838575872796405, "grad_norm": 0.6393121273747918, "learning_rate": 7.658704485903489e-06, "loss": 0.2588, "step": 9892 }, { "epoch": 0.6839267196681645, "grad_norm": 0.9308115904370967, "learning_rate": 7.658230405190488e-06, "loss": 0.3445, "step": 9893 }, { "epoch": 0.6839958520566886, "grad_norm": 0.6589215975399385, "learning_rate": 7.657756291161314e-06, "loss": 0.3511, "step": 9894 }, { "epoch": 0.6840649844452126, "grad_norm": 0.6381666003149165, "learning_rate": 7.657282143821905e-06, "loss": 0.2986, "step": 9895 }, { "epoch": 0.6841341168337366, "grad_norm": 0.691526013989083, "learning_rate": 7.656807963178208e-06, "loss": 0.334, "step": 9896 }, { "epoch": 0.6842032492222606, "grad_norm": 0.7184439981766849, "learning_rate": 7.656333749236165e-06, "loss": 0.3752, "step": 9897 }, { "epoch": 0.6842723816107846, "grad_norm": 0.6461320542070891, "learning_rate": 7.655859502001717e-06, "loss": 0.3378, "step": 9898 }, { "epoch": 0.6843415139993086, "grad_norm": 1.1366724827783872, "learning_rate": 7.655385221480812e-06, "loss": 0.3136, "step": 9899 }, { "epoch": 0.6844106463878327, "grad_norm": 0.688201358103605, "learning_rate": 7.654910907679389e-06, "loss": 0.3162, "step": 9900 }, { "epoch": 0.6844797787763567, "grad_norm": 0.87386202418146, "learning_rate": 7.654436560603399e-06, "loss": 0.4083, "step": 9901 }, { "epoch": 0.6845489111648807, "grad_norm": 0.6845473520033554, "learning_rate": 7.653962180258782e-06, "loss": 0.2784, "step": 9902 }, { "epoch": 0.6846180435534047, "grad_norm": 0.6341546473736951, "learning_rate": 7.653487766651486e-06, "loss": 0.2972, "step": 9903 }, { "epoch": 0.6846871759419287, "grad_norm": 0.6943281279054453, "learning_rate": 7.653013319787455e-06, "loss": 0.4002, "step": 9904 }, { "epoch": 0.6847563083304529, "grad_norm": 1.0670843007006905, "learning_rate": 7.652538839672638e-06, "loss": 0.383, "step": 9905 }, { "epoch": 0.6848254407189769, "grad_norm": 0.5877627207466328, "learning_rate": 7.652064326312979e-06, "loss": 0.3053, "step": 9906 }, { "epoch": 0.6848945731075009, "grad_norm": 0.6904744633167924, "learning_rate": 7.651589779714427e-06, "loss": 0.338, "step": 9907 }, { "epoch": 0.6849637054960249, "grad_norm": 0.6767727140416778, "learning_rate": 7.65111519988293e-06, "loss": 0.4022, "step": 9908 }, { "epoch": 0.6850328378845489, "grad_norm": 0.6643039614166433, "learning_rate": 7.650640586824433e-06, "loss": 0.3496, "step": 9909 }, { "epoch": 0.685101970273073, "grad_norm": 0.7381290561175629, "learning_rate": 7.65016594054489e-06, "loss": 0.2879, "step": 9910 }, { "epoch": 0.685171102661597, "grad_norm": 0.7176834354205558, "learning_rate": 7.649691261050243e-06, "loss": 0.3859, "step": 9911 }, { "epoch": 0.685240235050121, "grad_norm": 0.6810648141157277, "learning_rate": 7.649216548346446e-06, "loss": 0.4239, "step": 9912 }, { "epoch": 0.685309367438645, "grad_norm": 0.7055230843426116, "learning_rate": 7.648741802439448e-06, "loss": 0.3027, "step": 9913 }, { "epoch": 0.685378499827169, "grad_norm": 0.7965519892959139, "learning_rate": 7.648267023335196e-06, "loss": 0.3659, "step": 9914 }, { "epoch": 0.685447632215693, "grad_norm": 0.6632198544628309, "learning_rate": 7.647792211039646e-06, "loss": 0.3577, "step": 9915 }, { "epoch": 0.6855167646042171, "grad_norm": 0.857876634213789, "learning_rate": 7.647317365558744e-06, "loss": 0.3214, "step": 9916 }, { "epoch": 0.6855858969927411, "grad_norm": 0.5918218983038217, "learning_rate": 7.646842486898442e-06, "loss": 0.2649, "step": 9917 }, { "epoch": 0.6856550293812651, "grad_norm": 0.7185141300098188, "learning_rate": 7.646367575064696e-06, "loss": 0.3717, "step": 9918 }, { "epoch": 0.6857241617697891, "grad_norm": 0.6472941188354013, "learning_rate": 7.645892630063454e-06, "loss": 0.3838, "step": 9919 }, { "epoch": 0.6857932941583131, "grad_norm": 0.6127527401996624, "learning_rate": 7.645417651900669e-06, "loss": 0.266, "step": 9920 }, { "epoch": 0.6858624265468372, "grad_norm": 0.6413563242849738, "learning_rate": 7.644942640582294e-06, "loss": 0.2504, "step": 9921 }, { "epoch": 0.6859315589353612, "grad_norm": 0.6745404521233178, "learning_rate": 7.644467596114284e-06, "loss": 0.4009, "step": 9922 }, { "epoch": 0.6860006913238852, "grad_norm": 0.6869906263347705, "learning_rate": 7.643992518502592e-06, "loss": 0.3587, "step": 9923 }, { "epoch": 0.6860698237124092, "grad_norm": 0.6559403851129022, "learning_rate": 7.643517407753172e-06, "loss": 0.3026, "step": 9924 }, { "epoch": 0.6861389561009333, "grad_norm": 0.6940255320168315, "learning_rate": 7.64304226387198e-06, "loss": 0.3267, "step": 9925 }, { "epoch": 0.6862080884894574, "grad_norm": 0.6767438230038041, "learning_rate": 7.642567086864969e-06, "loss": 0.3837, "step": 9926 }, { "epoch": 0.6862772208779814, "grad_norm": 0.6310966444146158, "learning_rate": 7.642091876738094e-06, "loss": 0.2486, "step": 9927 }, { "epoch": 0.6863463532665054, "grad_norm": 0.6726401075280785, "learning_rate": 7.641616633497315e-06, "loss": 0.3666, "step": 9928 }, { "epoch": 0.6864154856550294, "grad_norm": 2.031572015453204, "learning_rate": 7.641141357148585e-06, "loss": 0.4384, "step": 9929 }, { "epoch": 0.6864846180435534, "grad_norm": 0.9609963947568959, "learning_rate": 7.64066604769786e-06, "loss": 0.4221, "step": 9930 }, { "epoch": 0.6865537504320774, "grad_norm": 0.7361358520861229, "learning_rate": 7.6401907051511e-06, "loss": 0.3608, "step": 9931 }, { "epoch": 0.6866228828206015, "grad_norm": 0.6842076244468577, "learning_rate": 7.63971532951426e-06, "loss": 0.3072, "step": 9932 }, { "epoch": 0.6866920152091255, "grad_norm": 0.6690645109734428, "learning_rate": 7.6392399207933e-06, "loss": 0.4353, "step": 9933 }, { "epoch": 0.6867611475976495, "grad_norm": 0.66726231391754, "learning_rate": 7.63876447899418e-06, "loss": 0.2913, "step": 9934 }, { "epoch": 0.6868302799861735, "grad_norm": 0.5875288409637653, "learning_rate": 7.638289004122849e-06, "loss": 0.2715, "step": 9935 }, { "epoch": 0.6868994123746975, "grad_norm": 0.6199616065940496, "learning_rate": 7.63781349618528e-06, "loss": 0.2992, "step": 9936 }, { "epoch": 0.6869685447632216, "grad_norm": 0.7224263094759062, "learning_rate": 7.637337955187425e-06, "loss": 0.4046, "step": 9937 }, { "epoch": 0.6870376771517456, "grad_norm": 0.7600017123340332, "learning_rate": 7.636862381135244e-06, "loss": 0.3008, "step": 9938 }, { "epoch": 0.6871068095402696, "grad_norm": 0.8553617651505793, "learning_rate": 7.636386774034698e-06, "loss": 0.3757, "step": 9939 }, { "epoch": 0.6871759419287936, "grad_norm": 0.8184170560435042, "learning_rate": 7.635911133891751e-06, "loss": 0.4251, "step": 9940 }, { "epoch": 0.6872450743173176, "grad_norm": 0.6940614982807911, "learning_rate": 7.63543546071236e-06, "loss": 0.3529, "step": 9941 }, { "epoch": 0.6873142067058416, "grad_norm": 0.7383843787033636, "learning_rate": 7.63495975450249e-06, "loss": 0.2925, "step": 9942 }, { "epoch": 0.6873833390943657, "grad_norm": 0.9022761049329815, "learning_rate": 7.634484015268102e-06, "loss": 0.3165, "step": 9943 }, { "epoch": 0.6874524714828897, "grad_norm": 0.7221030694481363, "learning_rate": 7.634008243015159e-06, "loss": 0.4071, "step": 9944 }, { "epoch": 0.6875216038714138, "grad_norm": 0.6732057314186337, "learning_rate": 7.63353243774962e-06, "loss": 0.2833, "step": 9945 }, { "epoch": 0.6875907362599378, "grad_norm": 0.6267763760236971, "learning_rate": 7.633056599477456e-06, "loss": 0.2946, "step": 9946 }, { "epoch": 0.6876598686484618, "grad_norm": 0.6843655216395854, "learning_rate": 7.632580728204624e-06, "loss": 0.3843, "step": 9947 }, { "epoch": 0.6877290010369859, "grad_norm": 0.6717321945885524, "learning_rate": 7.63210482393709e-06, "loss": 0.3205, "step": 9948 }, { "epoch": 0.6877981334255099, "grad_norm": 0.7656686215128836, "learning_rate": 7.63162888668082e-06, "loss": 0.3085, "step": 9949 }, { "epoch": 0.6878672658140339, "grad_norm": 0.695081791220949, "learning_rate": 7.631152916441777e-06, "loss": 0.3359, "step": 9950 }, { "epoch": 0.6879363982025579, "grad_norm": 0.6192885277089154, "learning_rate": 7.63067691322593e-06, "loss": 0.3625, "step": 9951 }, { "epoch": 0.6880055305910819, "grad_norm": 0.5855078456911206, "learning_rate": 7.63020087703924e-06, "loss": 0.2975, "step": 9952 }, { "epoch": 0.688074662979606, "grad_norm": 0.6980655319868917, "learning_rate": 7.629724807887679e-06, "loss": 0.2531, "step": 9953 }, { "epoch": 0.68814379536813, "grad_norm": 0.6659477762390361, "learning_rate": 7.629248705777208e-06, "loss": 0.3867, "step": 9954 }, { "epoch": 0.688212927756654, "grad_norm": 0.710320819013001, "learning_rate": 7.628772570713796e-06, "loss": 0.3724, "step": 9955 }, { "epoch": 0.688282060145178, "grad_norm": 0.6072588367788188, "learning_rate": 7.628296402703413e-06, "loss": 0.2714, "step": 9956 }, { "epoch": 0.688351192533702, "grad_norm": 0.6567841428745335, "learning_rate": 7.6278202017520255e-06, "loss": 0.2779, "step": 9957 }, { "epoch": 0.688420324922226, "grad_norm": 0.6731507780453374, "learning_rate": 7.627343967865599e-06, "loss": 0.4519, "step": 9958 }, { "epoch": 0.6884894573107501, "grad_norm": 1.3053057523053435, "learning_rate": 7.626867701050106e-06, "loss": 0.3658, "step": 9959 }, { "epoch": 0.6885585896992741, "grad_norm": 0.725916022697883, "learning_rate": 7.626391401311515e-06, "loss": 0.3105, "step": 9960 }, { "epoch": 0.6886277220877981, "grad_norm": 0.7039485640521915, "learning_rate": 7.625915068655793e-06, "loss": 0.3855, "step": 9961 }, { "epoch": 0.6886968544763221, "grad_norm": 0.7078998198436469, "learning_rate": 7.625438703088912e-06, "loss": 0.4501, "step": 9962 }, { "epoch": 0.6887659868648461, "grad_norm": 0.6346051377074514, "learning_rate": 7.624962304616843e-06, "loss": 0.2875, "step": 9963 }, { "epoch": 0.6888351192533702, "grad_norm": 0.6719234428153523, "learning_rate": 7.624485873245555e-06, "loss": 0.3076, "step": 9964 }, { "epoch": 0.6889042516418942, "grad_norm": 0.6885585790099111, "learning_rate": 7.624009408981021e-06, "loss": 0.3862, "step": 9965 }, { "epoch": 0.6889733840304183, "grad_norm": 0.7060526839783876, "learning_rate": 7.623532911829213e-06, "loss": 0.3113, "step": 9966 }, { "epoch": 0.6890425164189423, "grad_norm": 0.637563732756341, "learning_rate": 7.623056381796099e-06, "loss": 0.2584, "step": 9967 }, { "epoch": 0.6891116488074663, "grad_norm": 0.6660806718116015, "learning_rate": 7.622579818887657e-06, "loss": 0.3246, "step": 9968 }, { "epoch": 0.6891807811959904, "grad_norm": 0.7281098944866436, "learning_rate": 7.622103223109856e-06, "loss": 0.4074, "step": 9969 }, { "epoch": 0.6892499135845144, "grad_norm": 0.6027279815316605, "learning_rate": 7.62162659446867e-06, "loss": 0.238, "step": 9970 }, { "epoch": 0.6893190459730384, "grad_norm": 0.6965327929890288, "learning_rate": 7.621149932970073e-06, "loss": 0.296, "step": 9971 }, { "epoch": 0.6893881783615624, "grad_norm": 0.7783646575259145, "learning_rate": 7.62067323862004e-06, "loss": 0.3798, "step": 9972 }, { "epoch": 0.6894573107500864, "grad_norm": 2.839702349334829, "learning_rate": 7.620196511424545e-06, "loss": 0.3497, "step": 9973 }, { "epoch": 0.6895264431386104, "grad_norm": 0.624203599257033, "learning_rate": 7.6197197513895626e-06, "loss": 0.2603, "step": 9974 }, { "epoch": 0.6895955755271345, "grad_norm": 0.6929851069370297, "learning_rate": 7.619242958521067e-06, "loss": 0.3431, "step": 9975 }, { "epoch": 0.6896647079156585, "grad_norm": 0.7185098187035397, "learning_rate": 7.618766132825037e-06, "loss": 0.4789, "step": 9976 }, { "epoch": 0.6897338403041825, "grad_norm": 0.9228542608100158, "learning_rate": 7.618289274307444e-06, "loss": 0.3013, "step": 9977 }, { "epoch": 0.6898029726927065, "grad_norm": 0.61839850061396, "learning_rate": 7.61781238297427e-06, "loss": 0.2923, "step": 9978 }, { "epoch": 0.6898721050812305, "grad_norm": 0.6813750707602054, "learning_rate": 7.6173354588314895e-06, "loss": 0.392, "step": 9979 }, { "epoch": 0.6899412374697546, "grad_norm": 0.778003914244003, "learning_rate": 7.616858501885078e-06, "loss": 0.3355, "step": 9980 }, { "epoch": 0.6900103698582786, "grad_norm": 0.7531740938476562, "learning_rate": 7.616381512141018e-06, "loss": 0.2975, "step": 9981 }, { "epoch": 0.6900795022468026, "grad_norm": 0.6306404825613484, "learning_rate": 7.615904489605284e-06, "loss": 0.2928, "step": 9982 }, { "epoch": 0.6901486346353266, "grad_norm": 0.7531293175060295, "learning_rate": 7.615427434283855e-06, "loss": 0.3975, "step": 9983 }, { "epoch": 0.6902177670238506, "grad_norm": 0.7308279455670934, "learning_rate": 7.614950346182709e-06, "loss": 0.3081, "step": 9984 }, { "epoch": 0.6902868994123746, "grad_norm": 0.6372911680751022, "learning_rate": 7.614473225307829e-06, "loss": 0.2477, "step": 9985 }, { "epoch": 0.6903560318008988, "grad_norm": 0.7799127814136867, "learning_rate": 7.613996071665193e-06, "loss": 0.3805, "step": 9986 }, { "epoch": 0.6904251641894228, "grad_norm": 0.6615271155337874, "learning_rate": 7.61351888526078e-06, "loss": 0.3795, "step": 9987 }, { "epoch": 0.6904942965779468, "grad_norm": 0.7101349721844484, "learning_rate": 7.6130416661005716e-06, "loss": 0.3271, "step": 9988 }, { "epoch": 0.6905634289664708, "grad_norm": 0.6737338232287365, "learning_rate": 7.612564414190551e-06, "loss": 0.2835, "step": 9989 }, { "epoch": 0.6906325613549948, "grad_norm": 0.7489034190634509, "learning_rate": 7.612087129536696e-06, "loss": 0.4018, "step": 9990 }, { "epoch": 0.6907016937435189, "grad_norm": 0.6467190899539162, "learning_rate": 7.61160981214499e-06, "loss": 0.3593, "step": 9991 }, { "epoch": 0.6907708261320429, "grad_norm": 0.6907478875881408, "learning_rate": 7.6111324620214165e-06, "loss": 0.2502, "step": 9992 }, { "epoch": 0.6908399585205669, "grad_norm": 0.6908539023007135, "learning_rate": 7.610655079171958e-06, "loss": 0.3643, "step": 9993 }, { "epoch": 0.6909090909090909, "grad_norm": 0.6650310652725663, "learning_rate": 7.610177663602596e-06, "loss": 0.3918, "step": 9994 }, { "epoch": 0.6909782232976149, "grad_norm": 0.6509070970534055, "learning_rate": 7.609700215319316e-06, "loss": 0.3754, "step": 9995 }, { "epoch": 0.691047355686139, "grad_norm": 0.653965575384791, "learning_rate": 7.609222734328099e-06, "loss": 0.3048, "step": 9996 }, { "epoch": 0.691116488074663, "grad_norm": 0.6790499941763605, "learning_rate": 7.608745220634934e-06, "loss": 0.3711, "step": 9997 }, { "epoch": 0.691185620463187, "grad_norm": 0.7248618866360597, "learning_rate": 7.608267674245802e-06, "loss": 0.2932, "step": 9998 }, { "epoch": 0.691254752851711, "grad_norm": 0.6427963134333952, "learning_rate": 7.607790095166688e-06, "loss": 0.262, "step": 9999 }, { "epoch": 0.691323885240235, "grad_norm": 0.6733037934157247, "learning_rate": 7.607312483403582e-06, "loss": 0.3531, "step": 10000 }, { "epoch": 0.691393017628759, "grad_norm": 0.7551637729266357, "learning_rate": 7.606834838962465e-06, "loss": 0.4673, "step": 10001 }, { "epoch": 0.6914621500172831, "grad_norm": 0.6475444893059246, "learning_rate": 7.606357161849325e-06, "loss": 0.2925, "step": 10002 }, { "epoch": 0.6915312824058071, "grad_norm": 0.6817236275451456, "learning_rate": 7.60587945207015e-06, "loss": 0.3095, "step": 10003 }, { "epoch": 0.6916004147943311, "grad_norm": 0.7116196110306364, "learning_rate": 7.605401709630926e-06, "loss": 0.3662, "step": 10004 }, { "epoch": 0.6916695471828551, "grad_norm": 0.733887279139413, "learning_rate": 7.604923934537643e-06, "loss": 0.337, "step": 10005 }, { "epoch": 0.6917386795713792, "grad_norm": 0.673031761560593, "learning_rate": 7.604446126796284e-06, "loss": 0.3185, "step": 10006 }, { "epoch": 0.6918078119599033, "grad_norm": 0.7581695391019213, "learning_rate": 7.603968286412843e-06, "loss": 0.3671, "step": 10007 }, { "epoch": 0.6918769443484273, "grad_norm": 0.6706253170968481, "learning_rate": 7.603490413393306e-06, "loss": 0.4563, "step": 10008 }, { "epoch": 0.6919460767369513, "grad_norm": 0.802057804984809, "learning_rate": 7.6030125077436615e-06, "loss": 0.2628, "step": 10009 }, { "epoch": 0.6920152091254753, "grad_norm": 0.7300436854278151, "learning_rate": 7.602534569469902e-06, "loss": 0.3668, "step": 10010 }, { "epoch": 0.6920843415139993, "grad_norm": 0.6612191423173632, "learning_rate": 7.602056598578016e-06, "loss": 0.401, "step": 10011 }, { "epoch": 0.6921534739025234, "grad_norm": 0.7126500122012096, "learning_rate": 7.601578595073995e-06, "loss": 0.4455, "step": 10012 }, { "epoch": 0.6922226062910474, "grad_norm": 0.7742451960140383, "learning_rate": 7.6011005589638274e-06, "loss": 0.3003, "step": 10013 }, { "epoch": 0.6922917386795714, "grad_norm": 0.6636024338123107, "learning_rate": 7.6006224902535066e-06, "loss": 0.3164, "step": 10014 }, { "epoch": 0.6923608710680954, "grad_norm": 0.6931282568367606, "learning_rate": 7.600144388949024e-06, "loss": 0.4543, "step": 10015 }, { "epoch": 0.6924300034566194, "grad_norm": 0.7422875908215618, "learning_rate": 7.599666255056371e-06, "loss": 0.3831, "step": 10016 }, { "epoch": 0.6924991358451434, "grad_norm": 0.7679903457715233, "learning_rate": 7.5991880885815415e-06, "loss": 0.2555, "step": 10017 }, { "epoch": 0.6925682682336675, "grad_norm": 0.7003446264104449, "learning_rate": 7.598709889530529e-06, "loss": 0.3309, "step": 10018 }, { "epoch": 0.6926374006221915, "grad_norm": 0.6695748297381819, "learning_rate": 7.598231657909323e-06, "loss": 0.3789, "step": 10019 }, { "epoch": 0.6927065330107155, "grad_norm": 0.7944923642939096, "learning_rate": 7.597753393723921e-06, "loss": 0.3224, "step": 10020 }, { "epoch": 0.6927756653992395, "grad_norm": 0.6702050605157002, "learning_rate": 7.5972750969803165e-06, "loss": 0.2609, "step": 10021 }, { "epoch": 0.6928447977877635, "grad_norm": 0.6872577911251, "learning_rate": 7.596796767684505e-06, "loss": 0.3601, "step": 10022 }, { "epoch": 0.6929139301762876, "grad_norm": 0.6916675190039693, "learning_rate": 7.596318405842478e-06, "loss": 0.3289, "step": 10023 }, { "epoch": 0.6929830625648116, "grad_norm": 0.6830190665710615, "learning_rate": 7.595840011460233e-06, "loss": 0.3265, "step": 10024 }, { "epoch": 0.6930521949533356, "grad_norm": 0.7199591310447001, "learning_rate": 7.595361584543768e-06, "loss": 0.3977, "step": 10025 }, { "epoch": 0.6931213273418597, "grad_norm": 0.7182255435727846, "learning_rate": 7.594883125099075e-06, "loss": 0.4045, "step": 10026 }, { "epoch": 0.6931904597303837, "grad_norm": 0.6388846419613589, "learning_rate": 7.594404633132156e-06, "loss": 0.3272, "step": 10027 }, { "epoch": 0.6932595921189078, "grad_norm": 0.6372445112170944, "learning_rate": 7.593926108649001e-06, "loss": 0.2646, "step": 10028 }, { "epoch": 0.6933287245074318, "grad_norm": 0.6871708412835011, "learning_rate": 7.593447551655614e-06, "loss": 0.3822, "step": 10029 }, { "epoch": 0.6933978568959558, "grad_norm": 0.689921259904956, "learning_rate": 7.5929689621579876e-06, "loss": 0.3776, "step": 10030 }, { "epoch": 0.6934669892844798, "grad_norm": 0.7241674783182388, "learning_rate": 7.5924903401621245e-06, "loss": 0.266, "step": 10031 }, { "epoch": 0.6935361216730038, "grad_norm": 0.7253951698064606, "learning_rate": 7.592011685674021e-06, "loss": 0.2876, "step": 10032 }, { "epoch": 0.6936052540615278, "grad_norm": 0.9526837135074693, "learning_rate": 7.591532998699678e-06, "loss": 0.4054, "step": 10033 }, { "epoch": 0.6936743864500519, "grad_norm": 0.6484601576616055, "learning_rate": 7.591054279245091e-06, "loss": 0.2673, "step": 10034 }, { "epoch": 0.6937435188385759, "grad_norm": 0.6274246408919282, "learning_rate": 7.590575527316264e-06, "loss": 0.3057, "step": 10035 }, { "epoch": 0.6938126512270999, "grad_norm": 0.6759915930130919, "learning_rate": 7.590096742919197e-06, "loss": 0.2922, "step": 10036 }, { "epoch": 0.6938817836156239, "grad_norm": 0.7254483227743499, "learning_rate": 7.589617926059887e-06, "loss": 0.3731, "step": 10037 }, { "epoch": 0.6939509160041479, "grad_norm": 0.6920294781427486, "learning_rate": 7.589139076744341e-06, "loss": 0.3484, "step": 10038 }, { "epoch": 0.694020048392672, "grad_norm": 0.6265057845373414, "learning_rate": 7.588660194978554e-06, "loss": 0.3231, "step": 10039 }, { "epoch": 0.694089180781196, "grad_norm": 0.7146690942867675, "learning_rate": 7.588181280768534e-06, "loss": 0.4186, "step": 10040 }, { "epoch": 0.69415831316972, "grad_norm": 0.7196479385104663, "learning_rate": 7.587702334120277e-06, "loss": 0.3534, "step": 10041 }, { "epoch": 0.694227445558244, "grad_norm": 0.6479719633177043, "learning_rate": 7.587223355039792e-06, "loss": 0.3132, "step": 10042 }, { "epoch": 0.694296577946768, "grad_norm": 0.6726619105980034, "learning_rate": 7.586744343533079e-06, "loss": 0.3506, "step": 10043 }, { "epoch": 0.694365710335292, "grad_norm": 0.927457437802215, "learning_rate": 7.58626529960614e-06, "loss": 0.3983, "step": 10044 }, { "epoch": 0.6944348427238161, "grad_norm": 0.6624529400794879, "learning_rate": 7.585786223264982e-06, "loss": 0.2721, "step": 10045 }, { "epoch": 0.6945039751123401, "grad_norm": 0.6600055609715275, "learning_rate": 7.5853071145156095e-06, "loss": 0.2667, "step": 10046 }, { "epoch": 0.6945731075008642, "grad_norm": 0.6605480290354747, "learning_rate": 7.5848279733640255e-06, "loss": 0.4214, "step": 10047 }, { "epoch": 0.6946422398893882, "grad_norm": 0.6879468631053633, "learning_rate": 7.584348799816235e-06, "loss": 0.3474, "step": 10048 }, { "epoch": 0.6947113722779122, "grad_norm": 0.6687560112148736, "learning_rate": 7.583869593878243e-06, "loss": 0.2607, "step": 10049 }, { "epoch": 0.6947805046664363, "grad_norm": 0.7642820927531093, "learning_rate": 7.583390355556059e-06, "loss": 0.332, "step": 10050 }, { "epoch": 0.6948496370549603, "grad_norm": 0.643406071823625, "learning_rate": 7.582911084855686e-06, "loss": 0.4071, "step": 10051 }, { "epoch": 0.6949187694434843, "grad_norm": 0.7097440245278871, "learning_rate": 7.582431781783132e-06, "loss": 0.3162, "step": 10052 }, { "epoch": 0.6949879018320083, "grad_norm": 0.7026042304551784, "learning_rate": 7.581952446344405e-06, "loss": 0.32, "step": 10053 }, { "epoch": 0.6950570342205323, "grad_norm": 0.7189077472086058, "learning_rate": 7.581473078545511e-06, "loss": 0.3411, "step": 10054 }, { "epoch": 0.6951261666090564, "grad_norm": 0.6980300679608933, "learning_rate": 7.580993678392458e-06, "loss": 0.4128, "step": 10055 }, { "epoch": 0.6951952989975804, "grad_norm": 0.7579656949481445, "learning_rate": 7.580514245891257e-06, "loss": 0.3006, "step": 10056 }, { "epoch": 0.6952644313861044, "grad_norm": 0.9205491165428721, "learning_rate": 7.580034781047916e-06, "loss": 0.3367, "step": 10057 }, { "epoch": 0.6953335637746284, "grad_norm": 0.7542810874280862, "learning_rate": 7.579555283868441e-06, "loss": 0.4474, "step": 10058 }, { "epoch": 0.6954026961631524, "grad_norm": 0.757843219645859, "learning_rate": 7.579075754358845e-06, "loss": 0.3794, "step": 10059 }, { "epoch": 0.6954718285516764, "grad_norm": 0.6880709258439376, "learning_rate": 7.578596192525136e-06, "loss": 0.314, "step": 10060 }, { "epoch": 0.6955409609402005, "grad_norm": 0.8690718830039587, "learning_rate": 7.578116598373326e-06, "loss": 0.2937, "step": 10061 }, { "epoch": 0.6956100933287245, "grad_norm": 0.7501797097794939, "learning_rate": 7.577636971909424e-06, "loss": 0.419, "step": 10062 }, { "epoch": 0.6956792257172485, "grad_norm": 0.736970380352587, "learning_rate": 7.577157313139445e-06, "loss": 0.2773, "step": 10063 }, { "epoch": 0.6957483581057725, "grad_norm": 0.6244168310185402, "learning_rate": 7.5766776220693976e-06, "loss": 0.3019, "step": 10064 }, { "epoch": 0.6958174904942965, "grad_norm": 0.6924841776687126, "learning_rate": 7.576197898705293e-06, "loss": 0.4117, "step": 10065 }, { "epoch": 0.6958866228828205, "grad_norm": 0.6168708112596402, "learning_rate": 7.575718143053146e-06, "loss": 0.2996, "step": 10066 }, { "epoch": 0.6959557552713447, "grad_norm": 0.658738702724944, "learning_rate": 7.57523835511897e-06, "loss": 0.3058, "step": 10067 }, { "epoch": 0.6960248876598687, "grad_norm": 0.7575833028941341, "learning_rate": 7.5747585349087745e-06, "loss": 0.3457, "step": 10068 }, { "epoch": 0.6960940200483927, "grad_norm": 0.8170223520938084, "learning_rate": 7.5742786824285775e-06, "loss": 0.4106, "step": 10069 }, { "epoch": 0.6961631524369167, "grad_norm": 0.6666179527768731, "learning_rate": 7.573798797684391e-06, "loss": 0.2649, "step": 10070 }, { "epoch": 0.6962322848254408, "grad_norm": 0.6772590993212314, "learning_rate": 7.573318880682229e-06, "loss": 0.2524, "step": 10071 }, { "epoch": 0.6963014172139648, "grad_norm": 0.653507132990222, "learning_rate": 7.572838931428108e-06, "loss": 0.3466, "step": 10072 }, { "epoch": 0.6963705496024888, "grad_norm": 0.879174664410237, "learning_rate": 7.57235894992804e-06, "loss": 0.3897, "step": 10073 }, { "epoch": 0.6964396819910128, "grad_norm": 0.6814024602527883, "learning_rate": 7.5718789361880465e-06, "loss": 0.2481, "step": 10074 }, { "epoch": 0.6965088143795368, "grad_norm": 0.7149748740525566, "learning_rate": 7.571398890214139e-06, "loss": 0.3489, "step": 10075 }, { "epoch": 0.6965779467680608, "grad_norm": 0.7124314592547722, "learning_rate": 7.570918812012335e-06, "loss": 0.4345, "step": 10076 }, { "epoch": 0.6966470791565849, "grad_norm": 0.6664914234165752, "learning_rate": 7.5704387015886515e-06, "loss": 0.3118, "step": 10077 }, { "epoch": 0.6967162115451089, "grad_norm": 0.6742661638482893, "learning_rate": 7.5699585589491065e-06, "loss": 0.2647, "step": 10078 }, { "epoch": 0.6967853439336329, "grad_norm": 0.6895506771718654, "learning_rate": 7.5694783840997174e-06, "loss": 0.4049, "step": 10079 }, { "epoch": 0.6968544763221569, "grad_norm": 0.6838097811698269, "learning_rate": 7.5689981770465025e-06, "loss": 0.3337, "step": 10080 }, { "epoch": 0.6969236087106809, "grad_norm": 0.726075342477069, "learning_rate": 7.568517937795479e-06, "loss": 0.2925, "step": 10081 }, { "epoch": 0.696992741099205, "grad_norm": 0.7012507411524963, "learning_rate": 7.5680376663526675e-06, "loss": 0.3667, "step": 10082 }, { "epoch": 0.697061873487729, "grad_norm": 0.7213880151864959, "learning_rate": 7.567557362724085e-06, "loss": 0.4105, "step": 10083 }, { "epoch": 0.697131005876253, "grad_norm": 0.8427034934874731, "learning_rate": 7.5670770269157555e-06, "loss": 0.2777, "step": 10084 }, { "epoch": 0.697200138264777, "grad_norm": 0.6857918268230634, "learning_rate": 7.5665966589336955e-06, "loss": 0.2585, "step": 10085 }, { "epoch": 0.697269270653301, "grad_norm": 0.7530034771043062, "learning_rate": 7.566116258783928e-06, "loss": 0.3575, "step": 10086 }, { "epoch": 0.6973384030418252, "grad_norm": 0.7041111117998431, "learning_rate": 7.565635826472471e-06, "loss": 0.4037, "step": 10087 }, { "epoch": 0.6974075354303492, "grad_norm": 0.7020330211951639, "learning_rate": 7.565155362005348e-06, "loss": 0.3027, "step": 10088 }, { "epoch": 0.6974766678188732, "grad_norm": 0.7659101741807585, "learning_rate": 7.564674865388579e-06, "loss": 0.2839, "step": 10089 }, { "epoch": 0.6975458002073972, "grad_norm": 0.7716937256855271, "learning_rate": 7.564194336628188e-06, "loss": 0.3982, "step": 10090 }, { "epoch": 0.6976149325959212, "grad_norm": 0.6775614205461955, "learning_rate": 7.563713775730198e-06, "loss": 0.3203, "step": 10091 }, { "epoch": 0.6976840649844452, "grad_norm": 0.6440696110149514, "learning_rate": 7.563233182700631e-06, "loss": 0.2468, "step": 10092 }, { "epoch": 0.6977531973729693, "grad_norm": 0.7314594102444176, "learning_rate": 7.56275255754551e-06, "loss": 0.3287, "step": 10093 }, { "epoch": 0.6978223297614933, "grad_norm": 0.6803587928799951, "learning_rate": 7.562271900270859e-06, "loss": 0.3495, "step": 10094 }, { "epoch": 0.6978914621500173, "grad_norm": 0.6766361087775373, "learning_rate": 7.561791210882703e-06, "loss": 0.3061, "step": 10095 }, { "epoch": 0.6979605945385413, "grad_norm": 0.6598098341524963, "learning_rate": 7.561310489387065e-06, "loss": 0.2948, "step": 10096 }, { "epoch": 0.6980297269270653, "grad_norm": 0.7261157349222578, "learning_rate": 7.56082973578997e-06, "loss": 0.4031, "step": 10097 }, { "epoch": 0.6980988593155893, "grad_norm": 0.7080682993342053, "learning_rate": 7.560348950097446e-06, "loss": 0.359, "step": 10098 }, { "epoch": 0.6981679917041134, "grad_norm": 0.6833326878011804, "learning_rate": 7.559868132315517e-06, "loss": 0.287, "step": 10099 }, { "epoch": 0.6982371240926374, "grad_norm": 0.6704402433310016, "learning_rate": 7.5593872824502076e-06, "loss": 0.3142, "step": 10100 }, { "epoch": 0.6983062564811614, "grad_norm": 0.7140957697370088, "learning_rate": 7.558906400507547e-06, "loss": 0.4286, "step": 10101 }, { "epoch": 0.6983753888696854, "grad_norm": 0.6581534046705765, "learning_rate": 7.558425486493562e-06, "loss": 0.3419, "step": 10102 }, { "epoch": 0.6984445212582094, "grad_norm": 0.659019569411546, "learning_rate": 7.557944540414279e-06, "loss": 0.2942, "step": 10103 }, { "epoch": 0.6985136536467335, "grad_norm": 0.7056073548557217, "learning_rate": 7.557463562275727e-06, "loss": 0.3735, "step": 10104 }, { "epoch": 0.6985827860352575, "grad_norm": 0.6625739202978459, "learning_rate": 7.55698255208393e-06, "loss": 0.3858, "step": 10105 }, { "epoch": 0.6986519184237815, "grad_norm": 0.6076865836228524, "learning_rate": 7.556501509844922e-06, "loss": 0.2528, "step": 10106 }, { "epoch": 0.6987210508123055, "grad_norm": 0.6576436818908145, "learning_rate": 7.55602043556473e-06, "loss": 0.2756, "step": 10107 }, { "epoch": 0.6987901832008296, "grad_norm": 0.6685575151225754, "learning_rate": 7.5555393292493815e-06, "loss": 0.3815, "step": 10108 }, { "epoch": 0.6988593155893537, "grad_norm": 0.6686941801091784, "learning_rate": 7.55505819090491e-06, "loss": 0.3026, "step": 10109 }, { "epoch": 0.6989284479778777, "grad_norm": 0.6606635974864556, "learning_rate": 7.554577020537344e-06, "loss": 0.3047, "step": 10110 }, { "epoch": 0.6989975803664017, "grad_norm": 0.6824712806616199, "learning_rate": 7.554095818152713e-06, "loss": 0.3535, "step": 10111 }, { "epoch": 0.6990667127549257, "grad_norm": 1.2806300902803749, "learning_rate": 7.5536145837570484e-06, "loss": 0.3636, "step": 10112 }, { "epoch": 0.6991358451434497, "grad_norm": 0.6342726957991456, "learning_rate": 7.5531333173563825e-06, "loss": 0.2973, "step": 10113 }, { "epoch": 0.6992049775319737, "grad_norm": 0.6953766058331371, "learning_rate": 7.552652018956747e-06, "loss": 0.3174, "step": 10114 }, { "epoch": 0.6992741099204978, "grad_norm": 0.7176170456346864, "learning_rate": 7.552170688564175e-06, "loss": 0.4292, "step": 10115 }, { "epoch": 0.6993432423090218, "grad_norm": 0.7598664564470783, "learning_rate": 7.551689326184697e-06, "loss": 0.3015, "step": 10116 }, { "epoch": 0.6994123746975458, "grad_norm": 0.6481254662064795, "learning_rate": 7.551207931824347e-06, "loss": 0.2906, "step": 10117 }, { "epoch": 0.6994815070860698, "grad_norm": 0.7255373377607681, "learning_rate": 7.550726505489158e-06, "loss": 0.3084, "step": 10118 }, { "epoch": 0.6995506394745938, "grad_norm": 0.6438982373320383, "learning_rate": 7.550245047185163e-06, "loss": 0.4017, "step": 10119 }, { "epoch": 0.6996197718631179, "grad_norm": 3.1502118946187547, "learning_rate": 7.549763556918399e-06, "loss": 0.2598, "step": 10120 }, { "epoch": 0.6996889042516419, "grad_norm": 0.7808409127582392, "learning_rate": 7.5492820346949005e-06, "loss": 0.2236, "step": 10121 }, { "epoch": 0.6997580366401659, "grad_norm": 0.7004054456790674, "learning_rate": 7.548800480520698e-06, "loss": 0.3722, "step": 10122 }, { "epoch": 0.6998271690286899, "grad_norm": 0.7270271519113792, "learning_rate": 7.548318894401832e-06, "loss": 0.3552, "step": 10123 }, { "epoch": 0.6998963014172139, "grad_norm": 0.6377184599261053, "learning_rate": 7.547837276344337e-06, "loss": 0.2672, "step": 10124 }, { "epoch": 0.699965433805738, "grad_norm": 0.6955646051075923, "learning_rate": 7.547355626354247e-06, "loss": 0.3679, "step": 10125 }, { "epoch": 0.700034566194262, "grad_norm": 0.7157318027867706, "learning_rate": 7.5468739444376e-06, "loss": 0.4046, "step": 10126 }, { "epoch": 0.700103698582786, "grad_norm": 0.6880521250905559, "learning_rate": 7.546392230600435e-06, "loss": 0.3094, "step": 10127 }, { "epoch": 0.7001728309713101, "grad_norm": 0.6452971809416118, "learning_rate": 7.545910484848787e-06, "loss": 0.2924, "step": 10128 }, { "epoch": 0.7002419633598341, "grad_norm": 0.6956294823928929, "learning_rate": 7.545428707188693e-06, "loss": 0.443, "step": 10129 }, { "epoch": 0.7003110957483581, "grad_norm": 0.6295031854251878, "learning_rate": 7.544946897626193e-06, "loss": 0.3444, "step": 10130 }, { "epoch": 0.7003802281368822, "grad_norm": 0.7478269912080969, "learning_rate": 7.544465056167327e-06, "loss": 0.3422, "step": 10131 }, { "epoch": 0.7004493605254062, "grad_norm": 0.9145773084905712, "learning_rate": 7.543983182818131e-06, "loss": 0.3206, "step": 10132 }, { "epoch": 0.7005184929139302, "grad_norm": 0.6763405747221736, "learning_rate": 7.543501277584646e-06, "loss": 0.4415, "step": 10133 }, { "epoch": 0.7005876253024542, "grad_norm": 0.6883464543143132, "learning_rate": 7.543019340472911e-06, "loss": 0.3453, "step": 10134 }, { "epoch": 0.7006567576909782, "grad_norm": 0.6336861565678942, "learning_rate": 7.542537371488967e-06, "loss": 0.2929, "step": 10135 }, { "epoch": 0.7007258900795023, "grad_norm": 0.8691098184990294, "learning_rate": 7.542055370638857e-06, "loss": 0.3131, "step": 10136 }, { "epoch": 0.7007950224680263, "grad_norm": 0.7308960671034287, "learning_rate": 7.541573337928616e-06, "loss": 0.435, "step": 10137 }, { "epoch": 0.7008641548565503, "grad_norm": 0.8600795435945896, "learning_rate": 7.541091273364289e-06, "loss": 0.3113, "step": 10138 }, { "epoch": 0.7009332872450743, "grad_norm": 0.6730300335986553, "learning_rate": 7.54060917695192e-06, "loss": 0.3072, "step": 10139 }, { "epoch": 0.7010024196335983, "grad_norm": 0.6983312420998445, "learning_rate": 7.540127048697547e-06, "loss": 0.4495, "step": 10140 }, { "epoch": 0.7010715520221223, "grad_norm": 0.8434159850153424, "learning_rate": 7.539644888607215e-06, "loss": 0.3009, "step": 10141 }, { "epoch": 0.7011406844106464, "grad_norm": 0.6503857804566247, "learning_rate": 7.5391626966869665e-06, "loss": 0.2964, "step": 10142 }, { "epoch": 0.7012098167991704, "grad_norm": 0.7293521071788821, "learning_rate": 7.538680472942846e-06, "loss": 0.2877, "step": 10143 }, { "epoch": 0.7012789491876944, "grad_norm": 0.8321248718432639, "learning_rate": 7.5381982173808945e-06, "loss": 0.3917, "step": 10144 }, { "epoch": 0.7013480815762184, "grad_norm": 0.6252378102038476, "learning_rate": 7.537715930007159e-06, "loss": 0.2303, "step": 10145 }, { "epoch": 0.7014172139647424, "grad_norm": 0.7391485530883483, "learning_rate": 7.5372336108276835e-06, "loss": 0.2795, "step": 10146 }, { "epoch": 0.7014863463532665, "grad_norm": 0.7153042425419588, "learning_rate": 7.536751259848512e-06, "loss": 0.4713, "step": 10147 }, { "epoch": 0.7015554787417906, "grad_norm": 0.6615110484208203, "learning_rate": 7.53626887707569e-06, "loss": 0.3435, "step": 10148 }, { "epoch": 0.7016246111303146, "grad_norm": 0.6987185711782584, "learning_rate": 7.535786462515265e-06, "loss": 0.3581, "step": 10149 }, { "epoch": 0.7016937435188386, "grad_norm": 0.8063736195713895, "learning_rate": 7.535304016173281e-06, "loss": 0.2774, "step": 10150 }, { "epoch": 0.7017628759073626, "grad_norm": 0.7065121910775112, "learning_rate": 7.534821538055786e-06, "loss": 0.3874, "step": 10151 }, { "epoch": 0.7018320082958867, "grad_norm": 0.6250441516536233, "learning_rate": 7.534339028168827e-06, "loss": 0.3071, "step": 10152 }, { "epoch": 0.7019011406844107, "grad_norm": 0.6063142742019617, "learning_rate": 7.533856486518449e-06, "loss": 0.2888, "step": 10153 }, { "epoch": 0.7019702730729347, "grad_norm": 0.7755061104767808, "learning_rate": 7.533373913110704e-06, "loss": 0.4102, "step": 10154 }, { "epoch": 0.7020394054614587, "grad_norm": 0.7353081817074452, "learning_rate": 7.5328913079516375e-06, "loss": 0.3458, "step": 10155 }, { "epoch": 0.7021085378499827, "grad_norm": 1.7879814313815623, "learning_rate": 7.532408671047297e-06, "loss": 0.2826, "step": 10156 }, { "epoch": 0.7021776702385067, "grad_norm": 0.6720175948338817, "learning_rate": 7.531926002403736e-06, "loss": 0.2756, "step": 10157 }, { "epoch": 0.7022468026270308, "grad_norm": 0.6905128159503954, "learning_rate": 7.531443302026999e-06, "loss": 0.4346, "step": 10158 }, { "epoch": 0.7023159350155548, "grad_norm": 0.6826600538548152, "learning_rate": 7.530960569923138e-06, "loss": 0.3108, "step": 10159 }, { "epoch": 0.7023850674040788, "grad_norm": 0.7043770507622121, "learning_rate": 7.5304778060982045e-06, "loss": 0.2723, "step": 10160 }, { "epoch": 0.7024541997926028, "grad_norm": 0.6943215050594038, "learning_rate": 7.529995010558245e-06, "loss": 0.3367, "step": 10161 }, { "epoch": 0.7025233321811268, "grad_norm": 0.6677328966863427, "learning_rate": 7.529512183309316e-06, "loss": 0.3958, "step": 10162 }, { "epoch": 0.7025924645696509, "grad_norm": 0.678687556619208, "learning_rate": 7.529029324357463e-06, "loss": 0.3374, "step": 10163 }, { "epoch": 0.7026615969581749, "grad_norm": 0.6713334125007923, "learning_rate": 7.528546433708742e-06, "loss": 0.3705, "step": 10164 }, { "epoch": 0.7027307293466989, "grad_norm": 0.6808931130509203, "learning_rate": 7.528063511369203e-06, "loss": 0.4052, "step": 10165 }, { "epoch": 0.7027998617352229, "grad_norm": 0.6993454447329228, "learning_rate": 7.5275805573449e-06, "loss": 0.3182, "step": 10166 }, { "epoch": 0.7028689941237469, "grad_norm": 0.7254255503938115, "learning_rate": 7.527097571641886e-06, "loss": 0.2863, "step": 10167 }, { "epoch": 0.7029381265122709, "grad_norm": 0.7245874072766132, "learning_rate": 7.5266145542662124e-06, "loss": 0.3363, "step": 10168 }, { "epoch": 0.7030072589007951, "grad_norm": 0.7321211540721256, "learning_rate": 7.526131505223936e-06, "loss": 0.3646, "step": 10169 }, { "epoch": 0.7030763912893191, "grad_norm": 0.7224368848145254, "learning_rate": 7.525648424521108e-06, "loss": 0.3585, "step": 10170 }, { "epoch": 0.7031455236778431, "grad_norm": 0.6755803078564002, "learning_rate": 7.525165312163786e-06, "loss": 0.3188, "step": 10171 }, { "epoch": 0.7032146560663671, "grad_norm": 1.2559890929641597, "learning_rate": 7.524682168158021e-06, "loss": 0.4381, "step": 10172 }, { "epoch": 0.7032837884548911, "grad_norm": 0.6588986623236914, "learning_rate": 7.524198992509871e-06, "loss": 0.2942, "step": 10173 }, { "epoch": 0.7033529208434152, "grad_norm": 0.6271486743142242, "learning_rate": 7.523715785225392e-06, "loss": 0.2754, "step": 10174 }, { "epoch": 0.7034220532319392, "grad_norm": 0.901434610591903, "learning_rate": 7.5232325463106395e-06, "loss": 0.2908, "step": 10175 }, { "epoch": 0.7034911856204632, "grad_norm": 0.8362263139824795, "learning_rate": 7.522749275771668e-06, "loss": 0.3534, "step": 10176 }, { "epoch": 0.7035603180089872, "grad_norm": 0.675550340644507, "learning_rate": 7.522265973614539e-06, "loss": 0.2784, "step": 10177 }, { "epoch": 0.7036294503975112, "grad_norm": 2.047312607024591, "learning_rate": 7.521782639845305e-06, "loss": 0.2605, "step": 10178 }, { "epoch": 0.7036985827860353, "grad_norm": 0.7161692561164944, "learning_rate": 7.521299274470028e-06, "loss": 0.4399, "step": 10179 }, { "epoch": 0.7037677151745593, "grad_norm": 0.7038257531625048, "learning_rate": 7.520815877494763e-06, "loss": 0.3572, "step": 10180 }, { "epoch": 0.7038368475630833, "grad_norm": 0.7023205506669037, "learning_rate": 7.520332448925569e-06, "loss": 0.2821, "step": 10181 }, { "epoch": 0.7039059799516073, "grad_norm": 0.7565933000287831, "learning_rate": 7.519848988768507e-06, "loss": 0.3289, "step": 10182 }, { "epoch": 0.7039751123401313, "grad_norm": 0.7189624747496517, "learning_rate": 7.5193654970296335e-06, "loss": 0.3779, "step": 10183 }, { "epoch": 0.7040442447286553, "grad_norm": 0.687211037890812, "learning_rate": 7.51888197371501e-06, "loss": 0.3108, "step": 10184 }, { "epoch": 0.7041133771171794, "grad_norm": 0.8182904998644746, "learning_rate": 7.518398418830696e-06, "loss": 0.3339, "step": 10185 }, { "epoch": 0.7041825095057034, "grad_norm": 0.6910519785025357, "learning_rate": 7.517914832382751e-06, "loss": 0.3271, "step": 10186 }, { "epoch": 0.7042516418942274, "grad_norm": 0.7148199624893394, "learning_rate": 7.517431214377239e-06, "loss": 0.3981, "step": 10187 }, { "epoch": 0.7043207742827514, "grad_norm": 0.7068465956291395, "learning_rate": 7.5169475648202186e-06, "loss": 0.2943, "step": 10188 }, { "epoch": 0.7043899066712755, "grad_norm": 0.6970369612155068, "learning_rate": 7.5164638837177515e-06, "loss": 0.2855, "step": 10189 }, { "epoch": 0.7044590390597996, "grad_norm": 0.7032039583469677, "learning_rate": 7.515980171075901e-06, "loss": 0.4042, "step": 10190 }, { "epoch": 0.7045281714483236, "grad_norm": 0.624664841545353, "learning_rate": 7.515496426900728e-06, "loss": 0.2748, "step": 10191 }, { "epoch": 0.7045973038368476, "grad_norm": 0.6216312515419444, "learning_rate": 7.515012651198299e-06, "loss": 0.2714, "step": 10192 }, { "epoch": 0.7046664362253716, "grad_norm": 0.7432296017749696, "learning_rate": 7.5145288439746735e-06, "loss": 0.3479, "step": 10193 }, { "epoch": 0.7047355686138956, "grad_norm": 0.7252340377831287, "learning_rate": 7.514045005235915e-06, "loss": 0.4268, "step": 10194 }, { "epoch": 0.7048047010024197, "grad_norm": 0.6671148550247221, "learning_rate": 7.5135611349880895e-06, "loss": 0.2798, "step": 10195 }, { "epoch": 0.7048738333909437, "grad_norm": 0.7055856583452699, "learning_rate": 7.513077233237261e-06, "loss": 0.2983, "step": 10196 }, { "epoch": 0.7049429657794677, "grad_norm": 0.7323283075564293, "learning_rate": 7.512593299989493e-06, "loss": 0.4268, "step": 10197 }, { "epoch": 0.7050120981679917, "grad_norm": 0.6535024762949435, "learning_rate": 7.512109335250853e-06, "loss": 0.351, "step": 10198 }, { "epoch": 0.7050812305565157, "grad_norm": 0.7423933468633005, "learning_rate": 7.5116253390274065e-06, "loss": 0.2527, "step": 10199 }, { "epoch": 0.7051503629450397, "grad_norm": 0.6935047091862928, "learning_rate": 7.511141311325219e-06, "loss": 0.3031, "step": 10200 }, { "epoch": 0.7052194953335638, "grad_norm": 0.9810490273191128, "learning_rate": 7.510657252150353e-06, "loss": 0.4042, "step": 10201 }, { "epoch": 0.7052886277220878, "grad_norm": 0.709672918444857, "learning_rate": 7.510173161508882e-06, "loss": 0.3474, "step": 10202 }, { "epoch": 0.7053577601106118, "grad_norm": 0.6768155242149033, "learning_rate": 7.509689039406869e-06, "loss": 0.2956, "step": 10203 }, { "epoch": 0.7054268924991358, "grad_norm": 0.6659885513698883, "learning_rate": 7.509204885850382e-06, "loss": 0.3876, "step": 10204 }, { "epoch": 0.7054960248876598, "grad_norm": 0.7529074870694797, "learning_rate": 7.508720700845491e-06, "loss": 0.3424, "step": 10205 }, { "epoch": 0.7055651572761839, "grad_norm": 0.6969769993705784, "learning_rate": 7.5082364843982615e-06, "loss": 0.3277, "step": 10206 }, { "epoch": 0.7056342896647079, "grad_norm": 0.661003121390241, "learning_rate": 7.507752236514765e-06, "loss": 0.3303, "step": 10207 }, { "epoch": 0.7057034220532319, "grad_norm": 1.5998720208359, "learning_rate": 7.5072679572010685e-06, "loss": 0.3793, "step": 10208 }, { "epoch": 0.705772554441756, "grad_norm": 0.6393221511950352, "learning_rate": 7.506783646463244e-06, "loss": 0.3646, "step": 10209 }, { "epoch": 0.70584168683028, "grad_norm": 0.6801291385897603, "learning_rate": 7.50629930430736e-06, "loss": 0.2547, "step": 10210 }, { "epoch": 0.705910819218804, "grad_norm": 0.736369080039932, "learning_rate": 7.505814930739486e-06, "loss": 0.3181, "step": 10211 }, { "epoch": 0.7059799516073281, "grad_norm": 0.6467108622910112, "learning_rate": 7.505330525765694e-06, "loss": 0.3735, "step": 10212 }, { "epoch": 0.7060490839958521, "grad_norm": 0.614687403101091, "learning_rate": 7.504846089392055e-06, "loss": 0.2913, "step": 10213 }, { "epoch": 0.7061182163843761, "grad_norm": 0.7457875771322922, "learning_rate": 7.504361621624639e-06, "loss": 0.3722, "step": 10214 }, { "epoch": 0.7061873487729001, "grad_norm": 0.8481282089416597, "learning_rate": 7.503877122469521e-06, "loss": 0.4093, "step": 10215 }, { "epoch": 0.7062564811614241, "grad_norm": 0.6593988055769714, "learning_rate": 7.503392591932771e-06, "loss": 0.3567, "step": 10216 }, { "epoch": 0.7063256135499482, "grad_norm": 0.6006699668577699, "learning_rate": 7.502908030020463e-06, "loss": 0.3003, "step": 10217 }, { "epoch": 0.7063947459384722, "grad_norm": 0.668650955758018, "learning_rate": 7.502423436738668e-06, "loss": 0.3213, "step": 10218 }, { "epoch": 0.7064638783269962, "grad_norm": 0.6589787766626787, "learning_rate": 7.5019388120934615e-06, "loss": 0.3889, "step": 10219 }, { "epoch": 0.7065330107155202, "grad_norm": 0.6569135921058495, "learning_rate": 7.501454156090917e-06, "loss": 0.306, "step": 10220 }, { "epoch": 0.7066021431040442, "grad_norm": 0.6896758754002343, "learning_rate": 7.500969468737109e-06, "loss": 0.3067, "step": 10221 }, { "epoch": 0.7066712754925683, "grad_norm": 0.6491522132113517, "learning_rate": 7.500484750038113e-06, "loss": 0.4328, "step": 10222 }, { "epoch": 0.7067404078810923, "grad_norm": 0.6740546580028816, "learning_rate": 7.500000000000001e-06, "loss": 0.3621, "step": 10223 }, { "epoch": 0.7068095402696163, "grad_norm": 0.58939646290829, "learning_rate": 7.499515218628852e-06, "loss": 0.2973, "step": 10224 }, { "epoch": 0.7068786726581403, "grad_norm": 0.6867909102958567, "learning_rate": 7.499030405930738e-06, "loss": 0.3561, "step": 10225 }, { "epoch": 0.7069478050466643, "grad_norm": 0.681767532974822, "learning_rate": 7.498545561911739e-06, "loss": 0.4291, "step": 10226 }, { "epoch": 0.7070169374351883, "grad_norm": 0.6783784228368539, "learning_rate": 7.498060686577932e-06, "loss": 0.2816, "step": 10227 }, { "epoch": 0.7070860698237124, "grad_norm": 0.8132576550202355, "learning_rate": 7.497575779935391e-06, "loss": 0.2955, "step": 10228 }, { "epoch": 0.7071552022122364, "grad_norm": 0.761790436066296, "learning_rate": 7.497090841990193e-06, "loss": 0.4143, "step": 10229 }, { "epoch": 0.7072243346007605, "grad_norm": 0.73521734610379, "learning_rate": 7.496605872748419e-06, "loss": 0.3655, "step": 10230 }, { "epoch": 0.7072934669892845, "grad_norm": 0.6651726081954226, "learning_rate": 7.496120872216146e-06, "loss": 0.3276, "step": 10231 }, { "epoch": 0.7073625993778085, "grad_norm": 0.6668506811451099, "learning_rate": 7.495635840399451e-06, "loss": 0.293, "step": 10232 }, { "epoch": 0.7074317317663326, "grad_norm": 0.7000584348292158, "learning_rate": 7.495150777304416e-06, "loss": 0.4217, "step": 10233 }, { "epoch": 0.7075008641548566, "grad_norm": 0.6124342471410776, "learning_rate": 7.4946656829371185e-06, "loss": 0.3104, "step": 10234 }, { "epoch": 0.7075699965433806, "grad_norm": 0.6069656434712177, "learning_rate": 7.494180557303638e-06, "loss": 0.2982, "step": 10235 }, { "epoch": 0.7076391289319046, "grad_norm": 0.6552104667907644, "learning_rate": 7.493695400410056e-06, "loss": 0.3696, "step": 10236 }, { "epoch": 0.7077082613204286, "grad_norm": 0.7198584377859334, "learning_rate": 7.493210212262453e-06, "loss": 0.3493, "step": 10237 }, { "epoch": 0.7077773937089527, "grad_norm": 0.6444474175085604, "learning_rate": 7.492724992866908e-06, "loss": 0.2948, "step": 10238 }, { "epoch": 0.7078465260974767, "grad_norm": 0.7129505858306313, "learning_rate": 7.492239742229503e-06, "loss": 0.3253, "step": 10239 }, { "epoch": 0.7079156584860007, "grad_norm": 0.671222636332012, "learning_rate": 7.491754460356322e-06, "loss": 0.3573, "step": 10240 }, { "epoch": 0.7079847908745247, "grad_norm": 0.6516002819292983, "learning_rate": 7.491269147253446e-06, "loss": 0.3331, "step": 10241 }, { "epoch": 0.7080539232630487, "grad_norm": 0.593870365639376, "learning_rate": 7.490783802926955e-06, "loss": 0.2766, "step": 10242 }, { "epoch": 0.7081230556515727, "grad_norm": 0.6835488049028878, "learning_rate": 7.490298427382936e-06, "loss": 0.3468, "step": 10243 }, { "epoch": 0.7081921880400968, "grad_norm": 0.6843388387970564, "learning_rate": 7.4898130206274685e-06, "loss": 0.3991, "step": 10244 }, { "epoch": 0.7082613204286208, "grad_norm": 0.6995255901101745, "learning_rate": 7.48932758266664e-06, "loss": 0.2799, "step": 10245 }, { "epoch": 0.7083304528171448, "grad_norm": 0.6704323826804232, "learning_rate": 7.488842113506532e-06, "loss": 0.2558, "step": 10246 }, { "epoch": 0.7083995852056688, "grad_norm": 0.7734801727111091, "learning_rate": 7.488356613153229e-06, "loss": 0.3964, "step": 10247 }, { "epoch": 0.7084687175941928, "grad_norm": 0.70648329863949, "learning_rate": 7.487871081612816e-06, "loss": 0.3314, "step": 10248 }, { "epoch": 0.7085378499827168, "grad_norm": 0.6860717738627219, "learning_rate": 7.487385518891381e-06, "loss": 0.3092, "step": 10249 }, { "epoch": 0.708606982371241, "grad_norm": 0.8524809184169366, "learning_rate": 7.486899924995006e-06, "loss": 0.2909, "step": 10250 }, { "epoch": 0.708676114759765, "grad_norm": 0.6925521352180829, "learning_rate": 7.486414299929778e-06, "loss": 0.3797, "step": 10251 }, { "epoch": 0.708745247148289, "grad_norm": 0.6686413036449106, "learning_rate": 7.485928643701785e-06, "loss": 0.2814, "step": 10252 }, { "epoch": 0.708814379536813, "grad_norm": 0.6295663190907546, "learning_rate": 7.485442956317113e-06, "loss": 0.2619, "step": 10253 }, { "epoch": 0.708883511925337, "grad_norm": 0.7296949970928942, "learning_rate": 7.484957237781848e-06, "loss": 0.3762, "step": 10254 }, { "epoch": 0.7089526443138611, "grad_norm": 0.6815085948191646, "learning_rate": 7.4844714881020785e-06, "loss": 0.372, "step": 10255 }, { "epoch": 0.7090217767023851, "grad_norm": 0.6216697525322615, "learning_rate": 7.483985707283894e-06, "loss": 0.2606, "step": 10256 }, { "epoch": 0.7090909090909091, "grad_norm": 0.6913443069678535, "learning_rate": 7.48349989533338e-06, "loss": 0.3405, "step": 10257 }, { "epoch": 0.7091600414794331, "grad_norm": 0.6727034164700408, "learning_rate": 7.483014052256627e-06, "loss": 0.408, "step": 10258 }, { "epoch": 0.7092291738679571, "grad_norm": 0.6916206801546683, "learning_rate": 7.482528178059726e-06, "loss": 0.3559, "step": 10259 }, { "epoch": 0.7092983062564812, "grad_norm": 0.6551096208985526, "learning_rate": 7.482042272748763e-06, "loss": 0.3104, "step": 10260 }, { "epoch": 0.7093674386450052, "grad_norm": 0.6555048829827349, "learning_rate": 7.481556336329829e-06, "loss": 0.3961, "step": 10261 }, { "epoch": 0.7094365710335292, "grad_norm": 0.7320827972495888, "learning_rate": 7.481070368809016e-06, "loss": 0.3573, "step": 10262 }, { "epoch": 0.7095057034220532, "grad_norm": 0.673096772186988, "learning_rate": 7.480584370192413e-06, "loss": 0.3016, "step": 10263 }, { "epoch": 0.7095748358105772, "grad_norm": 0.6950678329826201, "learning_rate": 7.480098340486113e-06, "loss": 0.3181, "step": 10264 }, { "epoch": 0.7096439681991012, "grad_norm": 0.7221928158187759, "learning_rate": 7.479612279696203e-06, "loss": 0.464, "step": 10265 }, { "epoch": 0.7097131005876253, "grad_norm": 0.6473675165752849, "learning_rate": 7.479126187828781e-06, "loss": 0.3637, "step": 10266 }, { "epoch": 0.7097822329761493, "grad_norm": 0.6263514505006523, "learning_rate": 7.478640064889936e-06, "loss": 0.3029, "step": 10267 }, { "epoch": 0.7098513653646733, "grad_norm": 0.9515293480704969, "learning_rate": 7.478153910885761e-06, "loss": 0.3002, "step": 10268 }, { "epoch": 0.7099204977531973, "grad_norm": 0.7521229613362113, "learning_rate": 7.477667725822349e-06, "loss": 0.3959, "step": 10269 }, { "epoch": 0.7099896301417215, "grad_norm": 0.7200454754039696, "learning_rate": 7.4771815097057935e-06, "loss": 0.3012, "step": 10270 }, { "epoch": 0.7100587625302455, "grad_norm": 2.36404242293321, "learning_rate": 7.476695262542188e-06, "loss": 0.3255, "step": 10271 }, { "epoch": 0.7101278949187695, "grad_norm": 0.6395403676892955, "learning_rate": 7.476208984337627e-06, "loss": 0.4017, "step": 10272 }, { "epoch": 0.7101970273072935, "grad_norm": 0.721213653781501, "learning_rate": 7.475722675098207e-06, "loss": 0.3011, "step": 10273 }, { "epoch": 0.7102661596958175, "grad_norm": 0.7169465715541321, "learning_rate": 7.47523633483002e-06, "loss": 0.3005, "step": 10274 }, { "epoch": 0.7103352920843415, "grad_norm": 0.6702690014842453, "learning_rate": 7.474749963539164e-06, "loss": 0.2833, "step": 10275 }, { "epoch": 0.7104044244728656, "grad_norm": 0.6523442792030689, "learning_rate": 7.474263561231733e-06, "loss": 0.4088, "step": 10276 }, { "epoch": 0.7104735568613896, "grad_norm": 0.6579063846075173, "learning_rate": 7.473777127913824e-06, "loss": 0.2502, "step": 10277 }, { "epoch": 0.7105426892499136, "grad_norm": 0.6056011345580711, "learning_rate": 7.473290663591534e-06, "loss": 0.2628, "step": 10278 }, { "epoch": 0.7106118216384376, "grad_norm": 1.1559193097300366, "learning_rate": 7.472804168270959e-06, "loss": 0.4066, "step": 10279 }, { "epoch": 0.7106809540269616, "grad_norm": 0.6448276544113756, "learning_rate": 7.472317641958195e-06, "loss": 0.3444, "step": 10280 }, { "epoch": 0.7107500864154856, "grad_norm": 0.6469058187151014, "learning_rate": 7.471831084659343e-06, "loss": 0.2412, "step": 10281 }, { "epoch": 0.7108192188040097, "grad_norm": 0.6607007556710187, "learning_rate": 7.4713444963804995e-06, "loss": 0.3128, "step": 10282 }, { "epoch": 0.7108883511925337, "grad_norm": 0.6840709682111024, "learning_rate": 7.470857877127762e-06, "loss": 0.3957, "step": 10283 }, { "epoch": 0.7109574835810577, "grad_norm": 0.6229857915924271, "learning_rate": 7.470371226907232e-06, "loss": 0.2597, "step": 10284 }, { "epoch": 0.7110266159695817, "grad_norm": 0.5860613313556208, "learning_rate": 7.469884545725005e-06, "loss": 0.2654, "step": 10285 }, { "epoch": 0.7110957483581057, "grad_norm": 0.7047675265155406, "learning_rate": 7.4693978335871845e-06, "loss": 0.4083, "step": 10286 }, { "epoch": 0.7111648807466298, "grad_norm": 0.6748039159996797, "learning_rate": 7.468911090499869e-06, "loss": 0.3514, "step": 10287 }, { "epoch": 0.7112340131351538, "grad_norm": 0.7190494052359612, "learning_rate": 7.468424316469158e-06, "loss": 0.2754, "step": 10288 }, { "epoch": 0.7113031455236778, "grad_norm": 0.6356433626515915, "learning_rate": 7.467937511501154e-06, "loss": 0.3413, "step": 10289 }, { "epoch": 0.7113722779122018, "grad_norm": 0.7884884759394039, "learning_rate": 7.467450675601958e-06, "loss": 0.389, "step": 10290 }, { "epoch": 0.7114414103007259, "grad_norm": 0.7876475786544446, "learning_rate": 7.466963808777671e-06, "loss": 0.278, "step": 10291 }, { "epoch": 0.71151054268925, "grad_norm": 0.6606024141048497, "learning_rate": 7.466476911034394e-06, "loss": 0.2882, "step": 10292 }, { "epoch": 0.711579675077774, "grad_norm": 0.7494445327919865, "learning_rate": 7.465989982378232e-06, "loss": 0.3429, "step": 10293 }, { "epoch": 0.711648807466298, "grad_norm": 0.6317699587891146, "learning_rate": 7.465503022815286e-06, "loss": 0.3785, "step": 10294 }, { "epoch": 0.711717939854822, "grad_norm": 0.6163101202295628, "learning_rate": 7.4650160323516595e-06, "loss": 0.2354, "step": 10295 }, { "epoch": 0.711787072243346, "grad_norm": 1.2455294162953816, "learning_rate": 7.464529010993454e-06, "loss": 0.279, "step": 10296 }, { "epoch": 0.71185620463187, "grad_norm": 0.7193445094858266, "learning_rate": 7.464041958746778e-06, "loss": 0.4166, "step": 10297 }, { "epoch": 0.7119253370203941, "grad_norm": 0.714067269608513, "learning_rate": 7.463554875617733e-06, "loss": 0.3602, "step": 10298 }, { "epoch": 0.7119944694089181, "grad_norm": 1.2061551327772897, "learning_rate": 7.4630677616124235e-06, "loss": 0.2619, "step": 10299 }, { "epoch": 0.7120636017974421, "grad_norm": 1.1734616815969887, "learning_rate": 7.462580616736954e-06, "loss": 0.361, "step": 10300 }, { "epoch": 0.7121327341859661, "grad_norm": 0.8540289254417357, "learning_rate": 7.462093440997433e-06, "loss": 0.4279, "step": 10301 }, { "epoch": 0.7122018665744901, "grad_norm": 0.6751916048859076, "learning_rate": 7.461606234399965e-06, "loss": 0.2725, "step": 10302 }, { "epoch": 0.7122709989630142, "grad_norm": 0.6987875735015159, "learning_rate": 7.461118996950653e-06, "loss": 0.2776, "step": 10303 }, { "epoch": 0.7123401313515382, "grad_norm": 0.6819827578663172, "learning_rate": 7.4606317286556095e-06, "loss": 0.4223, "step": 10304 }, { "epoch": 0.7124092637400622, "grad_norm": 0.663484427097117, "learning_rate": 7.460144429520937e-06, "loss": 0.3624, "step": 10305 }, { "epoch": 0.7124783961285862, "grad_norm": 0.7980159473091354, "learning_rate": 7.459657099552744e-06, "loss": 0.2748, "step": 10306 }, { "epoch": 0.7125475285171102, "grad_norm": 0.7063424906964233, "learning_rate": 7.45916973875714e-06, "loss": 0.3358, "step": 10307 }, { "epoch": 0.7126166609056342, "grad_norm": 0.7136573193233141, "learning_rate": 7.45868234714023e-06, "loss": 0.4341, "step": 10308 }, { "epoch": 0.7126857932941583, "grad_norm": 0.6953238975533297, "learning_rate": 7.458194924708126e-06, "loss": 0.3144, "step": 10309 }, { "epoch": 0.7127549256826823, "grad_norm": 0.8132780469976432, "learning_rate": 7.457707471466935e-06, "loss": 0.3015, "step": 10310 }, { "epoch": 0.7128240580712064, "grad_norm": 0.6567978354660693, "learning_rate": 7.457219987422767e-06, "loss": 0.3835, "step": 10311 }, { "epoch": 0.7128931904597304, "grad_norm": 0.6852248661557929, "learning_rate": 7.456732472581731e-06, "loss": 0.4293, "step": 10312 }, { "epoch": 0.7129623228482544, "grad_norm": 0.8526006439619825, "learning_rate": 7.456244926949939e-06, "loss": 0.2861, "step": 10313 }, { "epoch": 0.7130314552367785, "grad_norm": 0.6430615532636, "learning_rate": 7.455757350533499e-06, "loss": 0.2678, "step": 10314 }, { "epoch": 0.7131005876253025, "grad_norm": 0.6786234908542625, "learning_rate": 7.455269743338523e-06, "loss": 0.41, "step": 10315 }, { "epoch": 0.7131697200138265, "grad_norm": 0.7414831783148056, "learning_rate": 7.454782105371123e-06, "loss": 0.3587, "step": 10316 }, { "epoch": 0.7132388524023505, "grad_norm": 1.4183370343488977, "learning_rate": 7.454294436637411e-06, "loss": 0.2927, "step": 10317 }, { "epoch": 0.7133079847908745, "grad_norm": 0.7441297852998081, "learning_rate": 7.4538067371434966e-06, "loss": 0.3483, "step": 10318 }, { "epoch": 0.7133771171793986, "grad_norm": 0.7864967749042246, "learning_rate": 7.453319006895494e-06, "loss": 0.435, "step": 10319 }, { "epoch": 0.7134462495679226, "grad_norm": 0.6616725576123135, "learning_rate": 7.4528312458995166e-06, "loss": 0.288, "step": 10320 }, { "epoch": 0.7135153819564466, "grad_norm": 0.6679576740635728, "learning_rate": 7.452343454161676e-06, "loss": 0.3409, "step": 10321 }, { "epoch": 0.7135845143449706, "grad_norm": 0.7453713900786233, "learning_rate": 7.451855631688087e-06, "loss": 0.3813, "step": 10322 }, { "epoch": 0.7136536467334946, "grad_norm": 0.8515344280000989, "learning_rate": 7.451367778484863e-06, "loss": 0.3748, "step": 10323 }, { "epoch": 0.7137227791220186, "grad_norm": 0.7731028644964267, "learning_rate": 7.450879894558119e-06, "loss": 0.2722, "step": 10324 }, { "epoch": 0.7137919115105427, "grad_norm": 0.7850373131284777, "learning_rate": 7.450391979913968e-06, "loss": 0.3846, "step": 10325 }, { "epoch": 0.7138610438990667, "grad_norm": 0.7542495379072286, "learning_rate": 7.449904034558527e-06, "loss": 0.3788, "step": 10326 }, { "epoch": 0.7139301762875907, "grad_norm": 0.6679788698386153, "learning_rate": 7.449416058497911e-06, "loss": 0.2865, "step": 10327 }, { "epoch": 0.7139993086761147, "grad_norm": 0.8127513779219645, "learning_rate": 7.448928051738237e-06, "loss": 0.2793, "step": 10328 }, { "epoch": 0.7140684410646387, "grad_norm": 0.8029429489270953, "learning_rate": 7.44844001428562e-06, "loss": 0.3933, "step": 10329 }, { "epoch": 0.7141375734531628, "grad_norm": 0.7924797969159187, "learning_rate": 7.4479519461461755e-06, "loss": 0.4386, "step": 10330 }, { "epoch": 0.7142067058416869, "grad_norm": 0.7060751009387378, "learning_rate": 7.447463847326023e-06, "loss": 0.2755, "step": 10331 }, { "epoch": 0.7142758382302109, "grad_norm": 0.7179077857386553, "learning_rate": 7.446975717831277e-06, "loss": 0.3023, "step": 10332 }, { "epoch": 0.7143449706187349, "grad_norm": 0.7383269832248314, "learning_rate": 7.446487557668059e-06, "loss": 0.4502, "step": 10333 }, { "epoch": 0.7144141030072589, "grad_norm": 0.7573876738382671, "learning_rate": 7.445999366842487e-06, "loss": 0.3389, "step": 10334 }, { "epoch": 0.714483235395783, "grad_norm": 0.6401786175132457, "learning_rate": 7.445511145360675e-06, "loss": 0.2647, "step": 10335 }, { "epoch": 0.714552367784307, "grad_norm": 0.7558759992239594, "learning_rate": 7.445022893228746e-06, "loss": 0.4209, "step": 10336 }, { "epoch": 0.714621500172831, "grad_norm": 0.805925856606801, "learning_rate": 7.444534610452819e-06, "loss": 0.4186, "step": 10337 }, { "epoch": 0.714690632561355, "grad_norm": 0.7505265539168803, "learning_rate": 7.444046297039013e-06, "loss": 0.2892, "step": 10338 }, { "epoch": 0.714759764949879, "grad_norm": 0.7033083957968168, "learning_rate": 7.443557952993448e-06, "loss": 0.2866, "step": 10339 }, { "epoch": 0.714828897338403, "grad_norm": 0.7362585337814568, "learning_rate": 7.443069578322244e-06, "loss": 0.3994, "step": 10340 }, { "epoch": 0.7148980297269271, "grad_norm": 0.6901695088042198, "learning_rate": 7.442581173031525e-06, "loss": 0.2985, "step": 10341 }, { "epoch": 0.7149671621154511, "grad_norm": 0.6905952783259556, "learning_rate": 7.442092737127408e-06, "loss": 0.2768, "step": 10342 }, { "epoch": 0.7150362945039751, "grad_norm": 0.7829147852720129, "learning_rate": 7.441604270616016e-06, "loss": 0.2988, "step": 10343 }, { "epoch": 0.7151054268924991, "grad_norm": 0.7403523303046213, "learning_rate": 7.441115773503473e-06, "loss": 0.3924, "step": 10344 }, { "epoch": 0.7151745592810231, "grad_norm": 0.6884406615286419, "learning_rate": 7.440627245795902e-06, "loss": 0.2652, "step": 10345 }, { "epoch": 0.7152436916695472, "grad_norm": 0.6356174216880851, "learning_rate": 7.440138687499421e-06, "loss": 0.3067, "step": 10346 }, { "epoch": 0.7153128240580712, "grad_norm": 0.9961111269470143, "learning_rate": 7.439650098620157e-06, "loss": 0.4025, "step": 10347 }, { "epoch": 0.7153819564465952, "grad_norm": 0.6777124174332493, "learning_rate": 7.439161479164232e-06, "loss": 0.2984, "step": 10348 }, { "epoch": 0.7154510888351192, "grad_norm": 0.7129181696405177, "learning_rate": 7.438672829137772e-06, "loss": 0.2768, "step": 10349 }, { "epoch": 0.7155202212236432, "grad_norm": 0.6833889732840205, "learning_rate": 7.4381841485468986e-06, "loss": 0.3727, "step": 10350 }, { "epoch": 0.7155893536121672, "grad_norm": 0.6561151075161628, "learning_rate": 7.437695437397741e-06, "loss": 0.4273, "step": 10351 }, { "epoch": 0.7156584860006914, "grad_norm": 0.5886755463268267, "learning_rate": 7.437206695696419e-06, "loss": 0.2875, "step": 10352 }, { "epoch": 0.7157276183892154, "grad_norm": 0.6754110618542756, "learning_rate": 7.43671792344906e-06, "loss": 0.2837, "step": 10353 }, { "epoch": 0.7157967507777394, "grad_norm": 0.7283922784058728, "learning_rate": 7.436229120661791e-06, "loss": 0.4186, "step": 10354 }, { "epoch": 0.7158658831662634, "grad_norm": 0.6729877308332955, "learning_rate": 7.4357402873407384e-06, "loss": 0.4268, "step": 10355 }, { "epoch": 0.7159350155547874, "grad_norm": 0.7632415407484382, "learning_rate": 7.435251423492027e-06, "loss": 0.3163, "step": 10356 }, { "epoch": 0.7160041479433115, "grad_norm": 0.6484468167179426, "learning_rate": 7.434762529121786e-06, "loss": 0.3134, "step": 10357 }, { "epoch": 0.7160732803318355, "grad_norm": 0.7341835759685714, "learning_rate": 7.434273604236141e-06, "loss": 0.4139, "step": 10358 }, { "epoch": 0.7161424127203595, "grad_norm": 0.6215168723552018, "learning_rate": 7.4337846488412215e-06, "loss": 0.2873, "step": 10359 }, { "epoch": 0.7162115451088835, "grad_norm": 0.6832220429347277, "learning_rate": 7.433295662943153e-06, "loss": 0.2838, "step": 10360 }, { "epoch": 0.7162806774974075, "grad_norm": 1.0750939658421559, "learning_rate": 7.432806646548067e-06, "loss": 0.3546, "step": 10361 }, { "epoch": 0.7163498098859316, "grad_norm": 0.7063076208795782, "learning_rate": 7.432317599662091e-06, "loss": 0.3676, "step": 10362 }, { "epoch": 0.7164189422744556, "grad_norm": 0.6119606620318476, "learning_rate": 7.4318285222913536e-06, "loss": 0.3031, "step": 10363 }, { "epoch": 0.7164880746629796, "grad_norm": 0.6923050656105242, "learning_rate": 7.431339414441987e-06, "loss": 0.3035, "step": 10364 }, { "epoch": 0.7165572070515036, "grad_norm": 0.7574308392016027, "learning_rate": 7.4308502761201185e-06, "loss": 0.4439, "step": 10365 }, { "epoch": 0.7166263394400276, "grad_norm": 0.7344707912430416, "learning_rate": 7.430361107331881e-06, "loss": 0.3258, "step": 10366 }, { "epoch": 0.7166954718285516, "grad_norm": 0.6954328146715428, "learning_rate": 7.429871908083402e-06, "loss": 0.3506, "step": 10367 }, { "epoch": 0.7167646042170757, "grad_norm": 0.7582529171054504, "learning_rate": 7.429382678380818e-06, "loss": 0.3643, "step": 10368 }, { "epoch": 0.7168337366055997, "grad_norm": 0.6876431283250949, "learning_rate": 7.428893418230256e-06, "loss": 0.3724, "step": 10369 }, { "epoch": 0.7169028689941237, "grad_norm": 0.670196497020044, "learning_rate": 7.428404127637849e-06, "loss": 0.3234, "step": 10370 }, { "epoch": 0.7169720013826477, "grad_norm": 0.6893203623437801, "learning_rate": 7.427914806609731e-06, "loss": 0.2847, "step": 10371 }, { "epoch": 0.7170411337711718, "grad_norm": 0.7950130619686637, "learning_rate": 7.427425455152033e-06, "loss": 0.3984, "step": 10372 }, { "epoch": 0.7171102661596959, "grad_norm": 0.7005156429504849, "learning_rate": 7.4269360732708895e-06, "loss": 0.3773, "step": 10373 }, { "epoch": 0.7171793985482199, "grad_norm": 0.6619869098058327, "learning_rate": 7.426446660972433e-06, "loss": 0.2881, "step": 10374 }, { "epoch": 0.7172485309367439, "grad_norm": 0.7899272232487259, "learning_rate": 7.425957218262798e-06, "loss": 0.3123, "step": 10375 }, { "epoch": 0.7173176633252679, "grad_norm": 0.6183391935874172, "learning_rate": 7.425467745148119e-06, "loss": 0.3971, "step": 10376 }, { "epoch": 0.7173867957137919, "grad_norm": 0.6644653274769777, "learning_rate": 7.4249782416345295e-06, "loss": 0.3439, "step": 10377 }, { "epoch": 0.717455928102316, "grad_norm": 0.7993313081168856, "learning_rate": 7.424488707728165e-06, "loss": 0.2966, "step": 10378 }, { "epoch": 0.71752506049084, "grad_norm": 0.7048807540240702, "learning_rate": 7.423999143435161e-06, "loss": 0.3828, "step": 10379 }, { "epoch": 0.717594192879364, "grad_norm": 0.6552408093736762, "learning_rate": 7.423509548761655e-06, "loss": 0.4223, "step": 10380 }, { "epoch": 0.717663325267888, "grad_norm": 0.6036895890862397, "learning_rate": 7.423019923713781e-06, "loss": 0.2935, "step": 10381 }, { "epoch": 0.717732457656412, "grad_norm": 0.6622426308117912, "learning_rate": 7.422530268297676e-06, "loss": 0.276, "step": 10382 }, { "epoch": 0.717801590044936, "grad_norm": 0.6563656468700814, "learning_rate": 7.422040582519479e-06, "loss": 0.44, "step": 10383 }, { "epoch": 0.7178707224334601, "grad_norm": 0.7029513595684689, "learning_rate": 7.421550866385323e-06, "loss": 0.3533, "step": 10384 }, { "epoch": 0.7179398548219841, "grad_norm": 0.6097347647553234, "learning_rate": 7.421061119901349e-06, "loss": 0.2756, "step": 10385 }, { "epoch": 0.7180089872105081, "grad_norm": 0.9715433145040961, "learning_rate": 7.420571343073696e-06, "loss": 0.342, "step": 10386 }, { "epoch": 0.7180781195990321, "grad_norm": 0.688330791109503, "learning_rate": 7.4200815359084995e-06, "loss": 0.4617, "step": 10387 }, { "epoch": 0.7181472519875561, "grad_norm": 0.6427565927082195, "learning_rate": 7.419591698411901e-06, "loss": 0.2573, "step": 10388 }, { "epoch": 0.7182163843760802, "grad_norm": 0.6462005828189324, "learning_rate": 7.419101830590037e-06, "loss": 0.2786, "step": 10389 }, { "epoch": 0.7182855167646042, "grad_norm": 0.7426974557156567, "learning_rate": 7.41861193244905e-06, "loss": 0.4178, "step": 10390 }, { "epoch": 0.7183546491531282, "grad_norm": 0.6829651312061399, "learning_rate": 7.418122003995079e-06, "loss": 0.2951, "step": 10391 }, { "epoch": 0.7184237815416523, "grad_norm": 0.6339754148811968, "learning_rate": 7.417632045234262e-06, "loss": 0.2594, "step": 10392 }, { "epoch": 0.7184929139301763, "grad_norm": 0.6251325324773168, "learning_rate": 7.417142056172743e-06, "loss": 0.357, "step": 10393 }, { "epoch": 0.7185620463187004, "grad_norm": 0.6911788475995496, "learning_rate": 7.4166520368166615e-06, "loss": 0.4079, "step": 10394 }, { "epoch": 0.7186311787072244, "grad_norm": 0.6919314941360054, "learning_rate": 7.41616198717216e-06, "loss": 0.2908, "step": 10395 }, { "epoch": 0.7187003110957484, "grad_norm": 0.6221997339446197, "learning_rate": 7.4156719072453785e-06, "loss": 0.2592, "step": 10396 }, { "epoch": 0.7187694434842724, "grad_norm": 0.6741589894314444, "learning_rate": 7.415181797042462e-06, "loss": 0.4178, "step": 10397 }, { "epoch": 0.7188385758727964, "grad_norm": 0.7144863693974729, "learning_rate": 7.4146916565695515e-06, "loss": 0.3051, "step": 10398 }, { "epoch": 0.7189077082613204, "grad_norm": 0.6152421187927261, "learning_rate": 7.4142014858327895e-06, "loss": 0.3109, "step": 10399 }, { "epoch": 0.7189768406498445, "grad_norm": 0.6461347161369934, "learning_rate": 7.413711284838321e-06, "loss": 0.2889, "step": 10400 }, { "epoch": 0.7190459730383685, "grad_norm": 0.7477527295272727, "learning_rate": 7.41322105359229e-06, "loss": 0.443, "step": 10401 }, { "epoch": 0.7191151054268925, "grad_norm": 0.7985102355278388, "learning_rate": 7.4127307921008385e-06, "loss": 0.2869, "step": 10402 }, { "epoch": 0.7191842378154165, "grad_norm": 0.8587538524988816, "learning_rate": 7.412240500370112e-06, "loss": 0.256, "step": 10403 }, { "epoch": 0.7192533702039405, "grad_norm": 0.6685474768309376, "learning_rate": 7.411750178406255e-06, "loss": 0.3281, "step": 10404 }, { "epoch": 0.7193225025924646, "grad_norm": 1.0105848021588424, "learning_rate": 7.411259826215416e-06, "loss": 0.3039, "step": 10405 }, { "epoch": 0.7193916349809886, "grad_norm": 0.7211928682510723, "learning_rate": 7.410769443803736e-06, "loss": 0.3239, "step": 10406 }, { "epoch": 0.7194607673695126, "grad_norm": 0.7198298262472665, "learning_rate": 7.4102790311773635e-06, "loss": 0.3791, "step": 10407 }, { "epoch": 0.7195298997580366, "grad_norm": 0.6275748394529328, "learning_rate": 7.409788588342445e-06, "loss": 0.3581, "step": 10408 }, { "epoch": 0.7195990321465606, "grad_norm": 0.6258451483183417, "learning_rate": 7.409298115305127e-06, "loss": 0.3103, "step": 10409 }, { "epoch": 0.7196681645350846, "grad_norm": 0.748351524249009, "learning_rate": 7.408807612071556e-06, "loss": 0.3312, "step": 10410 }, { "epoch": 0.7197372969236087, "grad_norm": 0.776516428850508, "learning_rate": 7.4083170786478804e-06, "loss": 0.3969, "step": 10411 }, { "epoch": 0.7198064293121328, "grad_norm": 0.6838929749072353, "learning_rate": 7.40782651504025e-06, "loss": 0.4077, "step": 10412 }, { "epoch": 0.7198755617006568, "grad_norm": 0.6742198644466489, "learning_rate": 7.40733592125481e-06, "loss": 0.2393, "step": 10413 }, { "epoch": 0.7199446940891808, "grad_norm": 0.6631239794485686, "learning_rate": 7.40684529729771e-06, "loss": 0.2961, "step": 10414 }, { "epoch": 0.7200138264777048, "grad_norm": 0.7077525259824998, "learning_rate": 7.4063546431751e-06, "loss": 0.4305, "step": 10415 }, { "epoch": 0.7200829588662289, "grad_norm": 0.6337904366938262, "learning_rate": 7.405863958893129e-06, "loss": 0.3582, "step": 10416 }, { "epoch": 0.7201520912547529, "grad_norm": 0.6462667306067155, "learning_rate": 7.405373244457945e-06, "loss": 0.3017, "step": 10417 }, { "epoch": 0.7202212236432769, "grad_norm": 0.7146607658140167, "learning_rate": 7.404882499875702e-06, "loss": 0.3591, "step": 10418 }, { "epoch": 0.7202903560318009, "grad_norm": 0.6697940833617763, "learning_rate": 7.404391725152547e-06, "loss": 0.398, "step": 10419 }, { "epoch": 0.7203594884203249, "grad_norm": 0.7061943784174203, "learning_rate": 7.403900920294633e-06, "loss": 0.255, "step": 10420 }, { "epoch": 0.720428620808849, "grad_norm": 0.6750362865051761, "learning_rate": 7.403410085308112e-06, "loss": 0.274, "step": 10421 }, { "epoch": 0.720497753197373, "grad_norm": 0.8518374934406382, "learning_rate": 7.402919220199134e-06, "loss": 0.4186, "step": 10422 }, { "epoch": 0.720566885585897, "grad_norm": 1.1975851445409686, "learning_rate": 7.402428324973852e-06, "loss": 0.3473, "step": 10423 }, { "epoch": 0.720636017974421, "grad_norm": 0.6035426187751067, "learning_rate": 7.401937399638417e-06, "loss": 0.2921, "step": 10424 }, { "epoch": 0.720705150362945, "grad_norm": 0.7017554382513848, "learning_rate": 7.4014464441989845e-06, "loss": 0.3316, "step": 10425 }, { "epoch": 0.720774282751469, "grad_norm": 0.7012083715826436, "learning_rate": 7.400955458661706e-06, "loss": 0.4103, "step": 10426 }, { "epoch": 0.7208434151399931, "grad_norm": 0.7446164738022695, "learning_rate": 7.400464443032734e-06, "loss": 0.2961, "step": 10427 }, { "epoch": 0.7209125475285171, "grad_norm": 0.7857986637866774, "learning_rate": 7.399973397318226e-06, "loss": 0.2771, "step": 10428 }, { "epoch": 0.7209816799170411, "grad_norm": 0.7377988494284237, "learning_rate": 7.3994823215243335e-06, "loss": 0.4328, "step": 10429 }, { "epoch": 0.7210508123055651, "grad_norm": 0.6326005501656312, "learning_rate": 7.398991215657214e-06, "loss": 0.3561, "step": 10430 }, { "epoch": 0.7211199446940891, "grad_norm": 0.6855719350068122, "learning_rate": 7.398500079723019e-06, "loss": 0.3207, "step": 10431 }, { "epoch": 0.7211890770826132, "grad_norm": 0.7008960914237014, "learning_rate": 7.398008913727905e-06, "loss": 0.3025, "step": 10432 }, { "epoch": 0.7212582094711373, "grad_norm": 0.678291221586732, "learning_rate": 7.397517717678031e-06, "loss": 0.372, "step": 10433 }, { "epoch": 0.7213273418596613, "grad_norm": 1.365616099702106, "learning_rate": 7.397026491579549e-06, "loss": 0.305, "step": 10434 }, { "epoch": 0.7213964742481853, "grad_norm": 0.6350635395594368, "learning_rate": 7.3965352354386185e-06, "loss": 0.3361, "step": 10435 }, { "epoch": 0.7214656066367093, "grad_norm": 0.6589514961520807, "learning_rate": 7.396043949261395e-06, "loss": 0.4326, "step": 10436 }, { "epoch": 0.7215347390252334, "grad_norm": 0.6536139645723347, "learning_rate": 7.3955526330540386e-06, "loss": 0.3528, "step": 10437 }, { "epoch": 0.7216038714137574, "grad_norm": 0.6669643364443791, "learning_rate": 7.395061286822702e-06, "loss": 0.3109, "step": 10438 }, { "epoch": 0.7216730038022814, "grad_norm": 0.6758374988399904, "learning_rate": 7.3945699105735495e-06, "loss": 0.2962, "step": 10439 }, { "epoch": 0.7217421361908054, "grad_norm": 0.7234242786319858, "learning_rate": 7.394078504312734e-06, "loss": 0.417, "step": 10440 }, { "epoch": 0.7218112685793294, "grad_norm": 0.8323899161663537, "learning_rate": 7.393587068046418e-06, "loss": 0.3345, "step": 10441 }, { "epoch": 0.7218804009678534, "grad_norm": 0.5860596625631088, "learning_rate": 7.393095601780759e-06, "loss": 0.3107, "step": 10442 }, { "epoch": 0.7219495333563775, "grad_norm": 0.6849827730583006, "learning_rate": 7.392604105521918e-06, "loss": 0.3311, "step": 10443 }, { "epoch": 0.7220186657449015, "grad_norm": 0.6546048280047758, "learning_rate": 7.392112579276054e-06, "loss": 0.4114, "step": 10444 }, { "epoch": 0.7220877981334255, "grad_norm": 0.7103363050596785, "learning_rate": 7.391621023049327e-06, "loss": 0.2553, "step": 10445 }, { "epoch": 0.7221569305219495, "grad_norm": 0.746179121933364, "learning_rate": 7.3911294368479e-06, "loss": 0.268, "step": 10446 }, { "epoch": 0.7222260629104735, "grad_norm": 0.686798232706914, "learning_rate": 7.3906378206779326e-06, "loss": 0.4258, "step": 10447 }, { "epoch": 0.7222951952989975, "grad_norm": 0.6676669110104253, "learning_rate": 7.390146174545585e-06, "loss": 0.3197, "step": 10448 }, { "epoch": 0.7223643276875216, "grad_norm": 0.5891762572624233, "learning_rate": 7.389654498457021e-06, "loss": 0.2673, "step": 10449 }, { "epoch": 0.7224334600760456, "grad_norm": 1.0856180128497255, "learning_rate": 7.389162792418403e-06, "loss": 0.2926, "step": 10450 }, { "epoch": 0.7225025924645696, "grad_norm": 0.6725662062575032, "learning_rate": 7.3886710564358934e-06, "loss": 0.4408, "step": 10451 }, { "epoch": 0.7225717248530936, "grad_norm": 0.6786729298298431, "learning_rate": 7.388179290515653e-06, "loss": 0.3293, "step": 10452 }, { "epoch": 0.7226408572416178, "grad_norm": 0.6764460819859954, "learning_rate": 7.387687494663849e-06, "loss": 0.2931, "step": 10453 }, { "epoch": 0.7227099896301418, "grad_norm": 0.755334932063058, "learning_rate": 7.387195668886642e-06, "loss": 0.4326, "step": 10454 }, { "epoch": 0.7227791220186658, "grad_norm": 0.6546619491964528, "learning_rate": 7.386703813190199e-06, "loss": 0.335, "step": 10455 }, { "epoch": 0.7228482544071898, "grad_norm": 0.7161325658759276, "learning_rate": 7.386211927580681e-06, "loss": 0.3366, "step": 10456 }, { "epoch": 0.7229173867957138, "grad_norm": 0.6952785255708874, "learning_rate": 7.3857200120642555e-06, "loss": 0.3215, "step": 10457 }, { "epoch": 0.7229865191842378, "grad_norm": 1.1552796387406634, "learning_rate": 7.385228066647087e-06, "loss": 0.412, "step": 10458 }, { "epoch": 0.7230556515727619, "grad_norm": 0.7351750194991643, "learning_rate": 7.384736091335342e-06, "loss": 0.303, "step": 10459 }, { "epoch": 0.7231247839612859, "grad_norm": 0.7524951387360808, "learning_rate": 7.384244086135186e-06, "loss": 0.2492, "step": 10460 }, { "epoch": 0.7231939163498099, "grad_norm": 0.7187666708167461, "learning_rate": 7.383752051052785e-06, "loss": 0.3452, "step": 10461 }, { "epoch": 0.7232630487383339, "grad_norm": 0.6763245017813901, "learning_rate": 7.3832599860943065e-06, "loss": 0.3669, "step": 10462 }, { "epoch": 0.7233321811268579, "grad_norm": 0.5925516142480434, "learning_rate": 7.382767891265916e-06, "loss": 0.3037, "step": 10463 }, { "epoch": 0.723401313515382, "grad_norm": 0.6875738368720983, "learning_rate": 7.3822757665737835e-06, "loss": 0.3405, "step": 10464 }, { "epoch": 0.723470445903906, "grad_norm": 0.6668428763612801, "learning_rate": 7.381783612024075e-06, "loss": 0.3865, "step": 10465 }, { "epoch": 0.72353957829243, "grad_norm": 0.6251827757485701, "learning_rate": 7.381291427622958e-06, "loss": 0.2764, "step": 10466 }, { "epoch": 0.723608710680954, "grad_norm": 0.6800954440326419, "learning_rate": 7.380799213376604e-06, "loss": 0.3023, "step": 10467 }, { "epoch": 0.723677843069478, "grad_norm": 0.6738370922403908, "learning_rate": 7.380306969291181e-06, "loss": 0.3133, "step": 10468 }, { "epoch": 0.723746975458002, "grad_norm": 0.6167466810165826, "learning_rate": 7.379814695372858e-06, "loss": 0.3569, "step": 10469 }, { "epoch": 0.7238161078465261, "grad_norm": 0.6399941332709291, "learning_rate": 7.379322391627803e-06, "loss": 0.2959, "step": 10470 }, { "epoch": 0.7238852402350501, "grad_norm": 0.9761068841736719, "learning_rate": 7.37883005806219e-06, "loss": 0.2925, "step": 10471 }, { "epoch": 0.7239543726235741, "grad_norm": 0.7235088077515482, "learning_rate": 7.3783376946821875e-06, "loss": 0.4607, "step": 10472 }, { "epoch": 0.7240235050120982, "grad_norm": 0.7954388670231164, "learning_rate": 7.377845301493964e-06, "loss": 0.3577, "step": 10473 }, { "epoch": 0.7240926374006222, "grad_norm": 0.68829065283015, "learning_rate": 7.377352878503695e-06, "loss": 0.3181, "step": 10474 }, { "epoch": 0.7241617697891463, "grad_norm": 0.6572444270615281, "learning_rate": 7.37686042571755e-06, "loss": 0.3183, "step": 10475 }, { "epoch": 0.7242309021776703, "grad_norm": 0.685196787701464, "learning_rate": 7.3763679431417015e-06, "loss": 0.4443, "step": 10476 }, { "epoch": 0.7243000345661943, "grad_norm": 0.7071153679406983, "learning_rate": 7.375875430782321e-06, "loss": 0.333, "step": 10477 }, { "epoch": 0.7243691669547183, "grad_norm": 0.7723391017285589, "learning_rate": 7.375382888645583e-06, "loss": 0.2873, "step": 10478 }, { "epoch": 0.7244382993432423, "grad_norm": 0.714009469621736, "learning_rate": 7.3748903167376585e-06, "loss": 0.354, "step": 10479 }, { "epoch": 0.7245074317317663, "grad_norm": 0.6606151507351877, "learning_rate": 7.374397715064721e-06, "loss": 0.3039, "step": 10480 }, { "epoch": 0.7245765641202904, "grad_norm": 0.7188448995585687, "learning_rate": 7.373905083632948e-06, "loss": 0.3041, "step": 10481 }, { "epoch": 0.7246456965088144, "grad_norm": 0.7478210998883619, "learning_rate": 7.373412422448509e-06, "loss": 0.3473, "step": 10482 }, { "epoch": 0.7247148288973384, "grad_norm": 0.6694242470571912, "learning_rate": 7.372919731517583e-06, "loss": 0.3755, "step": 10483 }, { "epoch": 0.7247839612858624, "grad_norm": 0.6328599351359048, "learning_rate": 7.37242701084634e-06, "loss": 0.2753, "step": 10484 }, { "epoch": 0.7248530936743864, "grad_norm": 0.8796413758715096, "learning_rate": 7.37193426044096e-06, "loss": 0.2705, "step": 10485 }, { "epoch": 0.7249222260629105, "grad_norm": 0.7084787174570367, "learning_rate": 7.371441480307617e-06, "loss": 0.3653, "step": 10486 }, { "epoch": 0.7249913584514345, "grad_norm": 0.6609899723045589, "learning_rate": 7.3709486704524864e-06, "loss": 0.4013, "step": 10487 }, { "epoch": 0.7250604908399585, "grad_norm": 0.6679774897998855, "learning_rate": 7.370455830881746e-06, "loss": 0.247, "step": 10488 }, { "epoch": 0.7251296232284825, "grad_norm": 0.7019071052323321, "learning_rate": 7.3699629616015725e-06, "loss": 0.3047, "step": 10489 }, { "epoch": 0.7251987556170065, "grad_norm": 0.6695310918091772, "learning_rate": 7.369470062618141e-06, "loss": 0.3709, "step": 10490 }, { "epoch": 0.7252678880055305, "grad_norm": 0.7995124371800997, "learning_rate": 7.368977133937632e-06, "loss": 0.3894, "step": 10491 }, { "epoch": 0.7253370203940546, "grad_norm": 0.690983491258567, "learning_rate": 7.368484175566222e-06, "loss": 0.3529, "step": 10492 }, { "epoch": 0.7254061527825786, "grad_norm": 0.6529175711018008, "learning_rate": 7.36799118751009e-06, "loss": 0.2798, "step": 10493 }, { "epoch": 0.7254752851711027, "grad_norm": 0.6736993322545191, "learning_rate": 7.3674981697754136e-06, "loss": 0.3753, "step": 10494 }, { "epoch": 0.7255444175596267, "grad_norm": 0.9793153856442061, "learning_rate": 7.367005122368373e-06, "loss": 0.2757, "step": 10495 }, { "epoch": 0.7256135499481507, "grad_norm": 0.735134883264484, "learning_rate": 7.366512045295147e-06, "loss": 0.3046, "step": 10496 }, { "epoch": 0.7256826823366748, "grad_norm": 0.6826250555945973, "learning_rate": 7.366018938561916e-06, "loss": 0.4364, "step": 10497 }, { "epoch": 0.7257518147251988, "grad_norm": 0.6964358529402926, "learning_rate": 7.36552580217486e-06, "loss": 0.3766, "step": 10498 }, { "epoch": 0.7258209471137228, "grad_norm": 0.6097595262092114, "learning_rate": 7.365032636140159e-06, "loss": 0.3016, "step": 10499 }, { "epoch": 0.7258900795022468, "grad_norm": 0.7259477898876016, "learning_rate": 7.364539440463996e-06, "loss": 0.325, "step": 10500 }, { "epoch": 0.7259592118907708, "grad_norm": 0.7320442079189586, "learning_rate": 7.3640462151525495e-06, "loss": 0.384, "step": 10501 }, { "epoch": 0.7260283442792949, "grad_norm": 0.6426727945081258, "learning_rate": 7.363552960212002e-06, "loss": 0.2865, "step": 10502 }, { "epoch": 0.7260974766678189, "grad_norm": 0.5902229120555528, "learning_rate": 7.363059675648537e-06, "loss": 0.2653, "step": 10503 }, { "epoch": 0.7261666090563429, "grad_norm": 0.6940601250817057, "learning_rate": 7.362566361468336e-06, "loss": 0.3765, "step": 10504 }, { "epoch": 0.7262357414448669, "grad_norm": 1.0025099971897369, "learning_rate": 7.362073017677582e-06, "loss": 0.3222, "step": 10505 }, { "epoch": 0.7263048738333909, "grad_norm": 0.6810559981976502, "learning_rate": 7.361579644282457e-06, "loss": 0.3144, "step": 10506 }, { "epoch": 0.726374006221915, "grad_norm": 0.6281067132018509, "learning_rate": 7.361086241289147e-06, "loss": 0.2538, "step": 10507 }, { "epoch": 0.726443138610439, "grad_norm": 0.6659730310664277, "learning_rate": 7.360592808703835e-06, "loss": 0.3872, "step": 10508 }, { "epoch": 0.726512270998963, "grad_norm": 0.6531252611452865, "learning_rate": 7.360099346532703e-06, "loss": 0.2981, "step": 10509 }, { "epoch": 0.726581403387487, "grad_norm": 0.712793905942306, "learning_rate": 7.3596058547819385e-06, "loss": 0.3064, "step": 10510 }, { "epoch": 0.726650535776011, "grad_norm": 0.7140843291971698, "learning_rate": 7.359112333457725e-06, "loss": 0.4019, "step": 10511 }, { "epoch": 0.726719668164535, "grad_norm": 0.661754317622832, "learning_rate": 7.358618782566249e-06, "loss": 0.4365, "step": 10512 }, { "epoch": 0.7267888005530591, "grad_norm": 0.7075296550528402, "learning_rate": 7.358125202113695e-06, "loss": 0.3045, "step": 10513 }, { "epoch": 0.7268579329415832, "grad_norm": 0.6910532979994577, "learning_rate": 7.35763159210625e-06, "loss": 0.333, "step": 10514 }, { "epoch": 0.7269270653301072, "grad_norm": 0.6792156318498656, "learning_rate": 7.357137952550101e-06, "loss": 0.4208, "step": 10515 }, { "epoch": 0.7269961977186312, "grad_norm": 0.6938836561779461, "learning_rate": 7.356644283451433e-06, "loss": 0.3375, "step": 10516 }, { "epoch": 0.7270653301071552, "grad_norm": 0.6753627670965584, "learning_rate": 7.356150584816435e-06, "loss": 0.2993, "step": 10517 }, { "epoch": 0.7271344624956793, "grad_norm": 0.6289105548293713, "learning_rate": 7.355656856651295e-06, "loss": 0.2809, "step": 10518 }, { "epoch": 0.7272035948842033, "grad_norm": 0.6766480920704853, "learning_rate": 7.3551630989622e-06, "loss": 0.3668, "step": 10519 }, { "epoch": 0.7272727272727273, "grad_norm": 0.7074476847693946, "learning_rate": 7.354669311755338e-06, "loss": 0.3171, "step": 10520 }, { "epoch": 0.7273418596612513, "grad_norm": 0.6295829573200978, "learning_rate": 7.3541754950369e-06, "loss": 0.2742, "step": 10521 }, { "epoch": 0.7274109920497753, "grad_norm": 0.962077423137082, "learning_rate": 7.353681648813071e-06, "loss": 0.4423, "step": 10522 }, { "epoch": 0.7274801244382993, "grad_norm": 0.6286190223669066, "learning_rate": 7.353187773090043e-06, "loss": 0.2983, "step": 10523 }, { "epoch": 0.7275492568268234, "grad_norm": 0.6020765020556379, "learning_rate": 7.3526938678740065e-06, "loss": 0.2935, "step": 10524 }, { "epoch": 0.7276183892153474, "grad_norm": 0.6280873384182996, "learning_rate": 7.352199933171152e-06, "loss": 0.3094, "step": 10525 }, { "epoch": 0.7276875216038714, "grad_norm": 0.6760103019768391, "learning_rate": 7.351705968987668e-06, "loss": 0.4499, "step": 10526 }, { "epoch": 0.7277566539923954, "grad_norm": 0.6583500532631869, "learning_rate": 7.351211975329745e-06, "loss": 0.2977, "step": 10527 }, { "epoch": 0.7278257863809194, "grad_norm": 0.6295642874855849, "learning_rate": 7.350717952203577e-06, "loss": 0.3024, "step": 10528 }, { "epoch": 0.7278949187694435, "grad_norm": 0.6493268514591833, "learning_rate": 7.350223899615357e-06, "loss": 0.3939, "step": 10529 }, { "epoch": 0.7279640511579675, "grad_norm": 0.6961261800930922, "learning_rate": 7.349729817571271e-06, "loss": 0.4131, "step": 10530 }, { "epoch": 0.7280331835464915, "grad_norm": 0.8209542699715314, "learning_rate": 7.349235706077516e-06, "loss": 0.2808, "step": 10531 }, { "epoch": 0.7281023159350155, "grad_norm": 0.6815807676947382, "learning_rate": 7.348741565140285e-06, "loss": 0.3781, "step": 10532 }, { "epoch": 0.7281714483235395, "grad_norm": 0.6595826229474255, "learning_rate": 7.348247394765769e-06, "loss": 0.4055, "step": 10533 }, { "epoch": 0.7282405807120637, "grad_norm": 0.6398737534279455, "learning_rate": 7.347753194960162e-06, "loss": 0.3053, "step": 10534 }, { "epoch": 0.7283097131005877, "grad_norm": 0.8311174229092237, "learning_rate": 7.347258965729659e-06, "loss": 0.346, "step": 10535 }, { "epoch": 0.7283788454891117, "grad_norm": 0.7342198031448368, "learning_rate": 7.346764707080453e-06, "loss": 0.3968, "step": 10536 }, { "epoch": 0.7284479778776357, "grad_norm": 0.6887960717041107, "learning_rate": 7.346270419018742e-06, "loss": 0.3868, "step": 10537 }, { "epoch": 0.7285171102661597, "grad_norm": 0.6098502186694451, "learning_rate": 7.345776101550716e-06, "loss": 0.2968, "step": 10538 }, { "epoch": 0.7285862426546837, "grad_norm": 0.6694594329717264, "learning_rate": 7.345281754682574e-06, "loss": 0.3437, "step": 10539 }, { "epoch": 0.7286553750432078, "grad_norm": 0.6806412583012811, "learning_rate": 7.3447873784205105e-06, "loss": 0.3891, "step": 10540 }, { "epoch": 0.7287245074317318, "grad_norm": 0.6193558864104673, "learning_rate": 7.34429297277072e-06, "loss": 0.281, "step": 10541 }, { "epoch": 0.7287936398202558, "grad_norm": 0.5974528718749395, "learning_rate": 7.343798537739401e-06, "loss": 0.2468, "step": 10542 }, { "epoch": 0.7288627722087798, "grad_norm": 0.6880456505732251, "learning_rate": 7.343304073332752e-06, "loss": 0.3513, "step": 10543 }, { "epoch": 0.7289319045973038, "grad_norm": 0.8056527515274812, "learning_rate": 7.342809579556968e-06, "loss": 0.3847, "step": 10544 }, { "epoch": 0.7290010369858279, "grad_norm": 0.6686230210518705, "learning_rate": 7.342315056418246e-06, "loss": 0.3177, "step": 10545 }, { "epoch": 0.7290701693743519, "grad_norm": 0.6450132393217171, "learning_rate": 7.341820503922785e-06, "loss": 0.3338, "step": 10546 }, { "epoch": 0.7291393017628759, "grad_norm": 0.7878918783182495, "learning_rate": 7.341325922076783e-06, "loss": 0.4018, "step": 10547 }, { "epoch": 0.7292084341513999, "grad_norm": 0.6610169259914008, "learning_rate": 7.340831310886438e-06, "loss": 0.2913, "step": 10548 }, { "epoch": 0.7292775665399239, "grad_norm": 0.60128825725745, "learning_rate": 7.340336670357951e-06, "loss": 0.2803, "step": 10549 }, { "epoch": 0.729346698928448, "grad_norm": 0.7117167668828543, "learning_rate": 7.339842000497521e-06, "loss": 0.3299, "step": 10550 }, { "epoch": 0.729415831316972, "grad_norm": 0.7566914231932087, "learning_rate": 7.339347301311346e-06, "loss": 0.5071, "step": 10551 }, { "epoch": 0.729484963705496, "grad_norm": 0.6620015342480364, "learning_rate": 7.338852572805627e-06, "loss": 0.3503, "step": 10552 }, { "epoch": 0.72955409609402, "grad_norm": 1.2618716499028297, "learning_rate": 7.338357814986565e-06, "loss": 0.2712, "step": 10553 }, { "epoch": 0.729623228482544, "grad_norm": 0.743300903717378, "learning_rate": 7.337863027860362e-06, "loss": 0.4501, "step": 10554 }, { "epoch": 0.7296923608710681, "grad_norm": 0.7820625545146307, "learning_rate": 7.337368211433217e-06, "loss": 0.3457, "step": 10555 }, { "epoch": 0.7297614932595922, "grad_norm": 0.6609014805568492, "learning_rate": 7.336873365711333e-06, "loss": 0.3085, "step": 10556 }, { "epoch": 0.7298306256481162, "grad_norm": 0.6755942758638747, "learning_rate": 7.3363784907009124e-06, "loss": 0.2928, "step": 10557 }, { "epoch": 0.7298997580366402, "grad_norm": 0.6444468540968357, "learning_rate": 7.335883586408157e-06, "loss": 0.3747, "step": 10558 }, { "epoch": 0.7299688904251642, "grad_norm": 0.6596353206239004, "learning_rate": 7.335388652839267e-06, "loss": 0.29, "step": 10559 }, { "epoch": 0.7300380228136882, "grad_norm": 0.7374899067251339, "learning_rate": 7.334893690000449e-06, "loss": 0.2837, "step": 10560 }, { "epoch": 0.7301071552022123, "grad_norm": 0.6995582132908384, "learning_rate": 7.334398697897907e-06, "loss": 0.3422, "step": 10561 }, { "epoch": 0.7301762875907363, "grad_norm": 1.0308388487720668, "learning_rate": 7.333903676537841e-06, "loss": 0.3631, "step": 10562 }, { "epoch": 0.7302454199792603, "grad_norm": 0.6452963193148251, "learning_rate": 7.3334086259264594e-06, "loss": 0.3007, "step": 10563 }, { "epoch": 0.7303145523677843, "grad_norm": 0.7392200798940605, "learning_rate": 7.3329135460699644e-06, "loss": 0.3501, "step": 10564 }, { "epoch": 0.7303836847563083, "grad_norm": 0.7444539501581853, "learning_rate": 7.332418436974561e-06, "loss": 0.4242, "step": 10565 }, { "epoch": 0.7304528171448323, "grad_norm": 0.6426549687297329, "learning_rate": 7.3319232986464535e-06, "loss": 0.2976, "step": 10566 }, { "epoch": 0.7305219495333564, "grad_norm": 0.643009194491292, "learning_rate": 7.331428131091851e-06, "loss": 0.2436, "step": 10567 }, { "epoch": 0.7305910819218804, "grad_norm": 0.8311897953599248, "learning_rate": 7.330932934316957e-06, "loss": 0.3093, "step": 10568 }, { "epoch": 0.7306602143104044, "grad_norm": 0.6691912607949331, "learning_rate": 7.330437708327977e-06, "loss": 0.3824, "step": 10569 }, { "epoch": 0.7307293466989284, "grad_norm": 0.6404896855654387, "learning_rate": 7.329942453131121e-06, "loss": 0.2799, "step": 10570 }, { "epoch": 0.7307984790874524, "grad_norm": 0.5891888116890653, "learning_rate": 7.329447168732594e-06, "loss": 0.255, "step": 10571 }, { "epoch": 0.7308676114759765, "grad_norm": 0.6913279097283791, "learning_rate": 7.328951855138604e-06, "loss": 0.4031, "step": 10572 }, { "epoch": 0.7309367438645005, "grad_norm": 0.6441231386057503, "learning_rate": 7.3284565123553574e-06, "loss": 0.3216, "step": 10573 }, { "epoch": 0.7310058762530245, "grad_norm": 0.6468794077080703, "learning_rate": 7.327961140389065e-06, "loss": 0.3169, "step": 10574 }, { "epoch": 0.7310750086415486, "grad_norm": 0.7336123774193902, "learning_rate": 7.3274657392459335e-06, "loss": 0.3345, "step": 10575 }, { "epoch": 0.7311441410300726, "grad_norm": 0.6416437620689057, "learning_rate": 7.326970308932173e-06, "loss": 0.4308, "step": 10576 }, { "epoch": 0.7312132734185967, "grad_norm": 0.6249263374755856, "learning_rate": 7.326474849453992e-06, "loss": 0.3371, "step": 10577 }, { "epoch": 0.7312824058071207, "grad_norm": 0.7489528407993558, "learning_rate": 7.3259793608176014e-06, "loss": 0.2804, "step": 10578 }, { "epoch": 0.7313515381956447, "grad_norm": 0.6819699312369748, "learning_rate": 7.3254838430292094e-06, "loss": 0.4088, "step": 10579 }, { "epoch": 0.7314206705841687, "grad_norm": 0.610413648947613, "learning_rate": 7.324988296095027e-06, "loss": 0.3608, "step": 10580 }, { "epoch": 0.7314898029726927, "grad_norm": 0.7535700045039797, "learning_rate": 7.324492720021267e-06, "loss": 0.3219, "step": 10581 }, { "epoch": 0.7315589353612167, "grad_norm": 0.654918395797976, "learning_rate": 7.323997114814138e-06, "loss": 0.2556, "step": 10582 }, { "epoch": 0.7316280677497408, "grad_norm": 0.6497493946061503, "learning_rate": 7.323501480479853e-06, "loss": 0.3532, "step": 10583 }, { "epoch": 0.7316972001382648, "grad_norm": 1.4858384650768819, "learning_rate": 7.323005817024622e-06, "loss": 0.3243, "step": 10584 }, { "epoch": 0.7317663325267888, "grad_norm": 0.6720505671485013, "learning_rate": 7.322510124454661e-06, "loss": 0.2594, "step": 10585 }, { "epoch": 0.7318354649153128, "grad_norm": 0.718146151269264, "learning_rate": 7.3220144027761785e-06, "loss": 0.3613, "step": 10586 }, { "epoch": 0.7319045973038368, "grad_norm": 0.6829145860642826, "learning_rate": 7.321518651995388e-06, "loss": 0.4051, "step": 10587 }, { "epoch": 0.7319737296923609, "grad_norm": 0.6347024787404809, "learning_rate": 7.321022872118507e-06, "loss": 0.3346, "step": 10588 }, { "epoch": 0.7320428620808849, "grad_norm": 0.6839577563708813, "learning_rate": 7.320527063151745e-06, "loss": 0.3036, "step": 10589 }, { "epoch": 0.7321119944694089, "grad_norm": 0.7213571025899126, "learning_rate": 7.320031225101318e-06, "loss": 0.437, "step": 10590 }, { "epoch": 0.7321811268579329, "grad_norm": 3.0337603905809014, "learning_rate": 7.319535357973439e-06, "loss": 0.3409, "step": 10591 }, { "epoch": 0.7322502592464569, "grad_norm": 0.6691364877234307, "learning_rate": 7.319039461774324e-06, "loss": 0.2707, "step": 10592 }, { "epoch": 0.7323193916349809, "grad_norm": 0.67955797032855, "learning_rate": 7.31854353651019e-06, "loss": 0.3435, "step": 10593 }, { "epoch": 0.732388524023505, "grad_norm": 0.9250383923618238, "learning_rate": 7.318047582187248e-06, "loss": 0.3887, "step": 10594 }, { "epoch": 0.7324576564120291, "grad_norm": 0.6608042823706727, "learning_rate": 7.317551598811716e-06, "loss": 0.3086, "step": 10595 }, { "epoch": 0.7325267888005531, "grad_norm": 0.6521803746380409, "learning_rate": 7.317055586389812e-06, "loss": 0.2761, "step": 10596 }, { "epoch": 0.7325959211890771, "grad_norm": 0.6429929050404538, "learning_rate": 7.31655954492775e-06, "loss": 0.359, "step": 10597 }, { "epoch": 0.7326650535776011, "grad_norm": 0.6885915736200233, "learning_rate": 7.316063474431748e-06, "loss": 0.3639, "step": 10598 }, { "epoch": 0.7327341859661252, "grad_norm": 0.7050839931546304, "learning_rate": 7.315567374908026e-06, "loss": 0.2715, "step": 10599 }, { "epoch": 0.7328033183546492, "grad_norm": 0.639083434254577, "learning_rate": 7.3150712463627975e-06, "loss": 0.3195, "step": 10600 }, { "epoch": 0.7328724507431732, "grad_norm": 0.7464749991468189, "learning_rate": 7.314575088802283e-06, "loss": 0.3778, "step": 10601 }, { "epoch": 0.7329415831316972, "grad_norm": 0.7149160587836212, "learning_rate": 7.3140789022326985e-06, "loss": 0.2825, "step": 10602 }, { "epoch": 0.7330107155202212, "grad_norm": 0.7254093524258252, "learning_rate": 7.313582686660265e-06, "loss": 0.2671, "step": 10603 }, { "epoch": 0.7330798479087453, "grad_norm": 0.6944161676355355, "learning_rate": 7.313086442091204e-06, "loss": 0.3166, "step": 10604 }, { "epoch": 0.7331489802972693, "grad_norm": 0.697368750502295, "learning_rate": 7.312590168531729e-06, "loss": 0.3849, "step": 10605 }, { "epoch": 0.7332181126857933, "grad_norm": 0.6520988831520282, "learning_rate": 7.312093865988065e-06, "loss": 0.2409, "step": 10606 }, { "epoch": 0.7332872450743173, "grad_norm": 0.6882463442702079, "learning_rate": 7.31159753446643e-06, "loss": 0.3745, "step": 10607 }, { "epoch": 0.7333563774628413, "grad_norm": 0.7439933909680161, "learning_rate": 7.311101173973046e-06, "loss": 0.4644, "step": 10608 }, { "epoch": 0.7334255098513653, "grad_norm": 0.7614812348237379, "learning_rate": 7.310604784514132e-06, "loss": 0.3253, "step": 10609 }, { "epoch": 0.7334946422398894, "grad_norm": 0.7226541693855322, "learning_rate": 7.310108366095912e-06, "loss": 0.2808, "step": 10610 }, { "epoch": 0.7335637746284134, "grad_norm": 0.8914353573204975, "learning_rate": 7.309611918724606e-06, "loss": 0.3247, "step": 10611 }, { "epoch": 0.7336329070169374, "grad_norm": 0.7682136361697188, "learning_rate": 7.309115442406435e-06, "loss": 0.3738, "step": 10612 }, { "epoch": 0.7337020394054614, "grad_norm": 0.687345884344431, "learning_rate": 7.308618937147623e-06, "loss": 0.2791, "step": 10613 }, { "epoch": 0.7337711717939854, "grad_norm": 0.7564732260768965, "learning_rate": 7.308122402954393e-06, "loss": 0.3931, "step": 10614 }, { "epoch": 0.7338403041825095, "grad_norm": 0.6882207341234026, "learning_rate": 7.307625839832967e-06, "loss": 0.4481, "step": 10615 }, { "epoch": 0.7339094365710336, "grad_norm": 0.7145779709856815, "learning_rate": 7.307129247789569e-06, "loss": 0.4165, "step": 10616 }, { "epoch": 0.7339785689595576, "grad_norm": 0.7005345137223592, "learning_rate": 7.306632626830425e-06, "loss": 0.2945, "step": 10617 }, { "epoch": 0.7340477013480816, "grad_norm": 0.6616708639974435, "learning_rate": 7.306135976961757e-06, "loss": 0.3118, "step": 10618 }, { "epoch": 0.7341168337366056, "grad_norm": 0.6596044464505418, "learning_rate": 7.305639298189788e-06, "loss": 0.4261, "step": 10619 }, { "epoch": 0.7341859661251297, "grad_norm": 0.8054302792627751, "learning_rate": 7.305142590520746e-06, "loss": 0.4061, "step": 10620 }, { "epoch": 0.7342550985136537, "grad_norm": 0.7857013097314067, "learning_rate": 7.304645853960857e-06, "loss": 0.2902, "step": 10621 }, { "epoch": 0.7343242309021777, "grad_norm": 0.7094769747887397, "learning_rate": 7.304149088516342e-06, "loss": 0.3457, "step": 10622 }, { "epoch": 0.7343933632907017, "grad_norm": 0.7118598379344373, "learning_rate": 7.303652294193433e-06, "loss": 0.3213, "step": 10623 }, { "epoch": 0.7344624956792257, "grad_norm": 0.6759112998106207, "learning_rate": 7.303155470998352e-06, "loss": 0.2837, "step": 10624 }, { "epoch": 0.7345316280677497, "grad_norm": 0.7488510838787809, "learning_rate": 7.302658618937328e-06, "loss": 0.2815, "step": 10625 }, { "epoch": 0.7346007604562738, "grad_norm": 0.6388215585311628, "learning_rate": 7.302161738016587e-06, "loss": 0.3659, "step": 10626 }, { "epoch": 0.7346698928447978, "grad_norm": 0.8488596790110613, "learning_rate": 7.301664828242356e-06, "loss": 0.2938, "step": 10627 }, { "epoch": 0.7347390252333218, "grad_norm": 0.646112698155992, "learning_rate": 7.301167889620866e-06, "loss": 0.2695, "step": 10628 }, { "epoch": 0.7348081576218458, "grad_norm": 0.6566664898921224, "learning_rate": 7.300670922158341e-06, "loss": 0.3488, "step": 10629 }, { "epoch": 0.7348772900103698, "grad_norm": 0.7583032931937441, "learning_rate": 7.300173925861013e-06, "loss": 0.3713, "step": 10630 }, { "epoch": 0.7349464223988939, "grad_norm": 0.7040231561722197, "learning_rate": 7.29967690073511e-06, "loss": 0.3003, "step": 10631 }, { "epoch": 0.7350155547874179, "grad_norm": 0.6911898328801218, "learning_rate": 7.299179846786862e-06, "loss": 0.3047, "step": 10632 }, { "epoch": 0.7350846871759419, "grad_norm": 0.6786708820780776, "learning_rate": 7.2986827640224945e-06, "loss": 0.4194, "step": 10633 }, { "epoch": 0.7351538195644659, "grad_norm": 0.8245786463079883, "learning_rate": 7.298185652448242e-06, "loss": 0.373, "step": 10634 }, { "epoch": 0.7352229519529899, "grad_norm": 0.6245038870721779, "learning_rate": 7.297688512070335e-06, "loss": 0.252, "step": 10635 }, { "epoch": 0.735292084341514, "grad_norm": 0.7272185616350523, "learning_rate": 7.2971913428950025e-06, "loss": 0.3806, "step": 10636 }, { "epoch": 0.7353612167300381, "grad_norm": 0.6671842716428144, "learning_rate": 7.296694144928474e-06, "loss": 0.4259, "step": 10637 }, { "epoch": 0.7354303491185621, "grad_norm": 0.6836140610143804, "learning_rate": 7.296196918176986e-06, "loss": 0.3153, "step": 10638 }, { "epoch": 0.7354994815070861, "grad_norm": 0.7194483097676297, "learning_rate": 7.295699662646766e-06, "loss": 0.3164, "step": 10639 }, { "epoch": 0.7355686138956101, "grad_norm": 0.7047543770025747, "learning_rate": 7.295202378344046e-06, "loss": 0.3664, "step": 10640 }, { "epoch": 0.7356377462841341, "grad_norm": 0.6690935201573448, "learning_rate": 7.294705065275063e-06, "loss": 0.3187, "step": 10641 }, { "epoch": 0.7357068786726582, "grad_norm": 0.6471933854103049, "learning_rate": 7.294207723446045e-06, "loss": 0.3304, "step": 10642 }, { "epoch": 0.7357760110611822, "grad_norm": 0.7672143671498871, "learning_rate": 7.2937103528632295e-06, "loss": 0.3021, "step": 10643 }, { "epoch": 0.7358451434497062, "grad_norm": 0.743822938865298, "learning_rate": 7.293212953532845e-06, "loss": 0.4372, "step": 10644 }, { "epoch": 0.7359142758382302, "grad_norm": 0.6472421270911121, "learning_rate": 7.2927155254611315e-06, "loss": 0.2852, "step": 10645 }, { "epoch": 0.7359834082267542, "grad_norm": 0.6988664840799202, "learning_rate": 7.2922180686543195e-06, "loss": 0.3137, "step": 10646 }, { "epoch": 0.7360525406152783, "grad_norm": 0.7258304669174176, "learning_rate": 7.2917205831186446e-06, "loss": 0.4275, "step": 10647 }, { "epoch": 0.7361216730038023, "grad_norm": 0.7798411500745833, "learning_rate": 7.291223068860341e-06, "loss": 0.3819, "step": 10648 }, { "epoch": 0.7361908053923263, "grad_norm": 0.9431266725087656, "learning_rate": 7.2907255258856455e-06, "loss": 0.2909, "step": 10649 }, { "epoch": 0.7362599377808503, "grad_norm": 0.7089995112397981, "learning_rate": 7.290227954200795e-06, "loss": 0.349, "step": 10650 }, { "epoch": 0.7363290701693743, "grad_norm": 0.7914656064140562, "learning_rate": 7.289730353812021e-06, "loss": 0.4248, "step": 10651 }, { "epoch": 0.7363982025578983, "grad_norm": 0.728344492297195, "learning_rate": 7.289232724725566e-06, "loss": 0.2859, "step": 10652 }, { "epoch": 0.7364673349464224, "grad_norm": 0.6518999067481641, "learning_rate": 7.288735066947663e-06, "loss": 0.2658, "step": 10653 }, { "epoch": 0.7365364673349464, "grad_norm": 0.6779632568363154, "learning_rate": 7.2882373804845516e-06, "loss": 0.3615, "step": 10654 }, { "epoch": 0.7366055997234704, "grad_norm": 0.8019868019199394, "learning_rate": 7.287739665342467e-06, "loss": 0.3911, "step": 10655 }, { "epoch": 0.7366747321119945, "grad_norm": 0.7258678141982958, "learning_rate": 7.287241921527648e-06, "loss": 0.3149, "step": 10656 }, { "epoch": 0.7367438645005185, "grad_norm": 0.732420734361609, "learning_rate": 7.286744149046334e-06, "loss": 0.3582, "step": 10657 }, { "epoch": 0.7368129968890426, "grad_norm": 0.62609655398309, "learning_rate": 7.286246347904763e-06, "loss": 0.3627, "step": 10658 }, { "epoch": 0.7368821292775666, "grad_norm": 0.6762473652017599, "learning_rate": 7.2857485181091745e-06, "loss": 0.3846, "step": 10659 }, { "epoch": 0.7369512616660906, "grad_norm": 0.6056516626697347, "learning_rate": 7.285250659665807e-06, "loss": 0.288, "step": 10660 }, { "epoch": 0.7370203940546146, "grad_norm": 0.7907604898747667, "learning_rate": 7.284752772580901e-06, "loss": 0.372, "step": 10661 }, { "epoch": 0.7370895264431386, "grad_norm": 0.7607198353548426, "learning_rate": 7.284254856860695e-06, "loss": 0.3728, "step": 10662 }, { "epoch": 0.7371586588316626, "grad_norm": 0.6698798688671236, "learning_rate": 7.283756912511432e-06, "loss": 0.2972, "step": 10663 }, { "epoch": 0.7372277912201867, "grad_norm": 0.7141914037813069, "learning_rate": 7.283258939539351e-06, "loss": 0.2754, "step": 10664 }, { "epoch": 0.7372969236087107, "grad_norm": 0.6528446801228659, "learning_rate": 7.2827609379506946e-06, "loss": 0.372, "step": 10665 }, { "epoch": 0.7373660559972347, "grad_norm": 1.4399758004367045, "learning_rate": 7.282262907751703e-06, "loss": 0.4178, "step": 10666 }, { "epoch": 0.7374351883857587, "grad_norm": 0.6414619882596799, "learning_rate": 7.281764848948618e-06, "loss": 0.2941, "step": 10667 }, { "epoch": 0.7375043207742827, "grad_norm": 0.6493597001728137, "learning_rate": 7.281266761547685e-06, "loss": 0.3275, "step": 10668 }, { "epoch": 0.7375734531628068, "grad_norm": 0.664598035640656, "learning_rate": 7.2807686455551415e-06, "loss": 0.349, "step": 10669 }, { "epoch": 0.7376425855513308, "grad_norm": 0.6695599989194008, "learning_rate": 7.280270500977234e-06, "loss": 0.3238, "step": 10670 }, { "epoch": 0.7377117179398548, "grad_norm": 0.680087301528884, "learning_rate": 7.279772327820207e-06, "loss": 0.2709, "step": 10671 }, { "epoch": 0.7377808503283788, "grad_norm": 0.7258926347525033, "learning_rate": 7.279274126090301e-06, "loss": 0.3557, "step": 10672 }, { "epoch": 0.7378499827169028, "grad_norm": 0.7500550968625972, "learning_rate": 7.2787758957937605e-06, "loss": 0.3361, "step": 10673 }, { "epoch": 0.7379191151054268, "grad_norm": 0.6806928967464975, "learning_rate": 7.278277636936832e-06, "loss": 0.2817, "step": 10674 }, { "epoch": 0.7379882474939509, "grad_norm": 0.701862635720472, "learning_rate": 7.277779349525759e-06, "loss": 0.3427, "step": 10675 }, { "epoch": 0.7380573798824749, "grad_norm": 0.6322284531773074, "learning_rate": 7.277281033566785e-06, "loss": 0.4195, "step": 10676 }, { "epoch": 0.738126512270999, "grad_norm": 0.6631833217695008, "learning_rate": 7.27678268906616e-06, "loss": 0.2602, "step": 10677 }, { "epoch": 0.738195644659523, "grad_norm": 0.8099177559921826, "learning_rate": 7.276284316030127e-06, "loss": 0.2897, "step": 10678 }, { "epoch": 0.738264777048047, "grad_norm": 0.6947061154154265, "learning_rate": 7.27578591446493e-06, "loss": 0.4391, "step": 10679 }, { "epoch": 0.7383339094365711, "grad_norm": 0.77877180102435, "learning_rate": 7.275287484376818e-06, "loss": 0.352, "step": 10680 }, { "epoch": 0.7384030418250951, "grad_norm": 0.5666639315618991, "learning_rate": 7.27478902577204e-06, "loss": 0.2492, "step": 10681 }, { "epoch": 0.7384721742136191, "grad_norm": 0.6730408027004207, "learning_rate": 7.274290538656841e-06, "loss": 0.3567, "step": 10682 }, { "epoch": 0.7385413066021431, "grad_norm": 0.6863247843642948, "learning_rate": 7.273792023037466e-06, "loss": 0.3924, "step": 10683 }, { "epoch": 0.7386104389906671, "grad_norm": 0.6031197137542182, "learning_rate": 7.273293478920167e-06, "loss": 0.2587, "step": 10684 }, { "epoch": 0.7386795713791912, "grad_norm": 0.6019664148365196, "learning_rate": 7.272794906311192e-06, "loss": 0.2563, "step": 10685 }, { "epoch": 0.7387487037677152, "grad_norm": 0.6904321652157712, "learning_rate": 7.272296305216788e-06, "loss": 0.3467, "step": 10686 }, { "epoch": 0.7388178361562392, "grad_norm": 0.6633259257460585, "learning_rate": 7.271797675643205e-06, "loss": 0.4075, "step": 10687 }, { "epoch": 0.7388869685447632, "grad_norm": 0.6999476158265433, "learning_rate": 7.2712990175966935e-06, "loss": 0.3022, "step": 10688 }, { "epoch": 0.7389561009332872, "grad_norm": 0.7217397751190647, "learning_rate": 7.270800331083501e-06, "loss": 0.3515, "step": 10689 }, { "epoch": 0.7390252333218112, "grad_norm": 0.8842054197138293, "learning_rate": 7.270301616109878e-06, "loss": 0.4438, "step": 10690 }, { "epoch": 0.7390943657103353, "grad_norm": 0.6207286871944387, "learning_rate": 7.269802872682078e-06, "loss": 0.3092, "step": 10691 }, { "epoch": 0.7391634980988593, "grad_norm": 0.7345011980536715, "learning_rate": 7.269304100806348e-06, "loss": 0.2973, "step": 10692 }, { "epoch": 0.7392326304873833, "grad_norm": 0.6323334916392968, "learning_rate": 7.268805300488941e-06, "loss": 0.272, "step": 10693 }, { "epoch": 0.7393017628759073, "grad_norm": 0.680784080943903, "learning_rate": 7.268306471736109e-06, "loss": 0.3619, "step": 10694 }, { "epoch": 0.7393708952644313, "grad_norm": 0.6856835262474861, "learning_rate": 7.2678076145541035e-06, "loss": 0.283, "step": 10695 }, { "epoch": 0.7394400276529554, "grad_norm": 0.6951125452959191, "learning_rate": 7.267308728949176e-06, "loss": 0.3559, "step": 10696 }, { "epoch": 0.7395091600414795, "grad_norm": 2.3735910344096656, "learning_rate": 7.266809814927579e-06, "loss": 0.4264, "step": 10697 }, { "epoch": 0.7395782924300035, "grad_norm": 0.7111803906926099, "learning_rate": 7.2663108724955675e-06, "loss": 0.381, "step": 10698 }, { "epoch": 0.7396474248185275, "grad_norm": 0.6551313450777377, "learning_rate": 7.265811901659394e-06, "loss": 0.2428, "step": 10699 }, { "epoch": 0.7397165572070515, "grad_norm": 0.804596615971727, "learning_rate": 7.2653129024253115e-06, "loss": 0.3457, "step": 10700 }, { "epoch": 0.7397856895955756, "grad_norm": 0.7628123295945975, "learning_rate": 7.264813874799573e-06, "loss": 0.405, "step": 10701 }, { "epoch": 0.7398548219840996, "grad_norm": 0.658769042254586, "learning_rate": 7.264314818788436e-06, "loss": 0.3429, "step": 10702 }, { "epoch": 0.7399239543726236, "grad_norm": 0.5786614693537215, "learning_rate": 7.263815734398153e-06, "loss": 0.2711, "step": 10703 }, { "epoch": 0.7399930867611476, "grad_norm": 0.7112075562656601, "learning_rate": 7.2633166216349795e-06, "loss": 0.4041, "step": 10704 }, { "epoch": 0.7400622191496716, "grad_norm": 0.7079023825693769, "learning_rate": 7.262817480505171e-06, "loss": 0.4069, "step": 10705 }, { "epoch": 0.7401313515381956, "grad_norm": 0.8098682910002348, "learning_rate": 7.262318311014984e-06, "loss": 0.2477, "step": 10706 }, { "epoch": 0.7402004839267197, "grad_norm": 0.747907775121882, "learning_rate": 7.261819113170675e-06, "loss": 0.3047, "step": 10707 }, { "epoch": 0.7402696163152437, "grad_norm": 0.7428041572343396, "learning_rate": 7.261319886978498e-06, "loss": 0.4289, "step": 10708 }, { "epoch": 0.7403387487037677, "grad_norm": 0.6852703305091031, "learning_rate": 7.260820632444714e-06, "loss": 0.3372, "step": 10709 }, { "epoch": 0.7404078810922917, "grad_norm": 0.6932364042216851, "learning_rate": 7.260321349575577e-06, "loss": 0.2958, "step": 10710 }, { "epoch": 0.7404770134808157, "grad_norm": 0.6994232819988419, "learning_rate": 7.259822038377345e-06, "loss": 0.3637, "step": 10711 }, { "epoch": 0.7405461458693398, "grad_norm": 0.683046464353382, "learning_rate": 7.259322698856276e-06, "loss": 0.3393, "step": 10712 }, { "epoch": 0.7406152782578638, "grad_norm": 0.6722961908453916, "learning_rate": 7.25882333101863e-06, "loss": 0.2903, "step": 10713 }, { "epoch": 0.7406844106463878, "grad_norm": 0.6121687940304432, "learning_rate": 7.258323934870665e-06, "loss": 0.2547, "step": 10714 }, { "epoch": 0.7407535430349118, "grad_norm": 0.8336497646114189, "learning_rate": 7.257824510418639e-06, "loss": 0.4079, "step": 10715 }, { "epoch": 0.7408226754234358, "grad_norm": 0.722656539084492, "learning_rate": 7.257325057668812e-06, "loss": 0.2728, "step": 10716 }, { "epoch": 0.74089180781196, "grad_norm": 0.616675052062139, "learning_rate": 7.256825576627444e-06, "loss": 0.3218, "step": 10717 }, { "epoch": 0.740960940200484, "grad_norm": 0.6411566841630214, "learning_rate": 7.256326067300795e-06, "loss": 0.2798, "step": 10718 }, { "epoch": 0.741030072589008, "grad_norm": 0.7082721302072191, "learning_rate": 7.255826529695124e-06, "loss": 0.386, "step": 10719 }, { "epoch": 0.741099204977532, "grad_norm": 0.8271684404595923, "learning_rate": 7.255326963816696e-06, "loss": 0.3228, "step": 10720 }, { "epoch": 0.741168337366056, "grad_norm": 0.6616967905340386, "learning_rate": 7.2548273696717675e-06, "loss": 0.3517, "step": 10721 }, { "epoch": 0.74123746975458, "grad_norm": 0.7029465483514823, "learning_rate": 7.2543277472666e-06, "loss": 0.4004, "step": 10722 }, { "epoch": 0.7413066021431041, "grad_norm": 0.8721834127831765, "learning_rate": 7.25382809660746e-06, "loss": 0.3339, "step": 10723 }, { "epoch": 0.7413757345316281, "grad_norm": 0.6344396364665631, "learning_rate": 7.253328417700607e-06, "loss": 0.2159, "step": 10724 }, { "epoch": 0.7414448669201521, "grad_norm": 0.6953644682210961, "learning_rate": 7.252828710552302e-06, "loss": 0.3365, "step": 10725 }, { "epoch": 0.7415139993086761, "grad_norm": 0.8218925374532546, "learning_rate": 7.25232897516881e-06, "loss": 0.3567, "step": 10726 }, { "epoch": 0.7415831316972001, "grad_norm": 0.6042424479946916, "learning_rate": 7.251829211556394e-06, "loss": 0.3053, "step": 10727 }, { "epoch": 0.7416522640857242, "grad_norm": 0.6856967253238274, "learning_rate": 7.251329419721318e-06, "loss": 0.2857, "step": 10728 }, { "epoch": 0.7417213964742482, "grad_norm": 0.6869028270573796, "learning_rate": 7.250829599669844e-06, "loss": 0.3512, "step": 10729 }, { "epoch": 0.7417905288627722, "grad_norm": 0.6663209123699846, "learning_rate": 7.2503297514082384e-06, "loss": 0.3717, "step": 10730 }, { "epoch": 0.7418596612512962, "grad_norm": 0.6575787133634289, "learning_rate": 7.249829874942766e-06, "loss": 0.3221, "step": 10731 }, { "epoch": 0.7419287936398202, "grad_norm": 0.6582082070922627, "learning_rate": 7.2493299702796906e-06, "loss": 0.343, "step": 10732 }, { "epoch": 0.7419979260283442, "grad_norm": 0.6563824326505862, "learning_rate": 7.248830037425276e-06, "loss": 0.367, "step": 10733 }, { "epoch": 0.7420670584168683, "grad_norm": 0.6277376192587638, "learning_rate": 7.248330076385793e-06, "loss": 0.2969, "step": 10734 }, { "epoch": 0.7421361908053923, "grad_norm": 0.6719000535781625, "learning_rate": 7.247830087167505e-06, "loss": 0.2556, "step": 10735 }, { "epoch": 0.7422053231939163, "grad_norm": 0.689768419346194, "learning_rate": 7.2473300697766764e-06, "loss": 0.3689, "step": 10736 }, { "epoch": 0.7422744555824403, "grad_norm": 0.669533481322061, "learning_rate": 7.246830024219577e-06, "loss": 0.4134, "step": 10737 }, { "epoch": 0.7423435879709644, "grad_norm": 0.634053529122585, "learning_rate": 7.246329950502472e-06, "loss": 0.3119, "step": 10738 }, { "epoch": 0.7424127203594885, "grad_norm": 0.7121790592454937, "learning_rate": 7.245829848631632e-06, "loss": 0.3619, "step": 10739 }, { "epoch": 0.7424818527480125, "grad_norm": 0.7285455946155974, "learning_rate": 7.245329718613318e-06, "loss": 0.399, "step": 10740 }, { "epoch": 0.7425509851365365, "grad_norm": 0.684495798204373, "learning_rate": 7.244829560453807e-06, "loss": 0.3014, "step": 10741 }, { "epoch": 0.7426201175250605, "grad_norm": 0.6104111613135174, "learning_rate": 7.244329374159364e-06, "loss": 0.2937, "step": 10742 }, { "epoch": 0.7426892499135845, "grad_norm": 0.6542394494436049, "learning_rate": 7.243829159736255e-06, "loss": 0.3348, "step": 10743 }, { "epoch": 0.7427583823021086, "grad_norm": 0.6604476445253348, "learning_rate": 7.243328917190754e-06, "loss": 0.3554, "step": 10744 }, { "epoch": 0.7428275146906326, "grad_norm": 0.6300204088383698, "learning_rate": 7.242828646529127e-06, "loss": 0.2929, "step": 10745 }, { "epoch": 0.7428966470791566, "grad_norm": 0.6841326237160771, "learning_rate": 7.242328347757645e-06, "loss": 0.2509, "step": 10746 }, { "epoch": 0.7429657794676806, "grad_norm": 0.6998678253421039, "learning_rate": 7.24182802088258e-06, "loss": 0.4305, "step": 10747 }, { "epoch": 0.7430349118562046, "grad_norm": 0.690564717379415, "learning_rate": 7.241327665910201e-06, "loss": 0.4073, "step": 10748 }, { "epoch": 0.7431040442447286, "grad_norm": 0.688209086107557, "learning_rate": 7.2408272828467795e-06, "loss": 0.2981, "step": 10749 }, { "epoch": 0.7431731766332527, "grad_norm": 0.7143530480039206, "learning_rate": 7.240326871698587e-06, "loss": 0.3248, "step": 10750 }, { "epoch": 0.7432423090217767, "grad_norm": 0.6609735452251432, "learning_rate": 7.239826432471895e-06, "loss": 0.4469, "step": 10751 }, { "epoch": 0.7433114414103007, "grad_norm": 0.6576112046566644, "learning_rate": 7.239325965172976e-06, "loss": 0.2826, "step": 10752 }, { "epoch": 0.7433805737988247, "grad_norm": 0.6163862022730003, "learning_rate": 7.2388254698081025e-06, "loss": 0.2589, "step": 10753 }, { "epoch": 0.7434497061873487, "grad_norm": 0.796939078008215, "learning_rate": 7.238324946383546e-06, "loss": 0.4321, "step": 10754 }, { "epoch": 0.7435188385758728, "grad_norm": 1.1071496497580426, "learning_rate": 7.237824394905583e-06, "loss": 0.3689, "step": 10755 }, { "epoch": 0.7435879709643968, "grad_norm": 0.895454724068807, "learning_rate": 7.237323815380483e-06, "loss": 0.3059, "step": 10756 }, { "epoch": 0.7436571033529208, "grad_norm": 0.6877190113095455, "learning_rate": 7.23682320781452e-06, "loss": 0.3264, "step": 10757 }, { "epoch": 0.7437262357414449, "grad_norm": 0.717643901256715, "learning_rate": 7.236322572213974e-06, "loss": 0.4341, "step": 10758 }, { "epoch": 0.7437953681299689, "grad_norm": 0.7047533183476422, "learning_rate": 7.235821908585113e-06, "loss": 0.3364, "step": 10759 }, { "epoch": 0.743864500518493, "grad_norm": 0.7345446553098348, "learning_rate": 7.2353212169342145e-06, "loss": 0.2939, "step": 10760 }, { "epoch": 0.743933632907017, "grad_norm": 1.297726178665044, "learning_rate": 7.234820497267554e-06, "loss": 0.3225, "step": 10761 }, { "epoch": 0.744002765295541, "grad_norm": 0.6810484077576174, "learning_rate": 7.234319749591408e-06, "loss": 0.3894, "step": 10762 }, { "epoch": 0.744071897684065, "grad_norm": 0.7303061096575189, "learning_rate": 7.233818973912049e-06, "loss": 0.2887, "step": 10763 }, { "epoch": 0.744141030072589, "grad_norm": 0.6934292175194398, "learning_rate": 7.233318170235757e-06, "loss": 0.2738, "step": 10764 }, { "epoch": 0.744210162461113, "grad_norm": 0.7106166265761654, "learning_rate": 7.232817338568806e-06, "loss": 0.4314, "step": 10765 }, { "epoch": 0.7442792948496371, "grad_norm": 0.7208090683909051, "learning_rate": 7.232316478917476e-06, "loss": 0.324, "step": 10766 }, { "epoch": 0.7443484272381611, "grad_norm": 0.6668609268871658, "learning_rate": 7.2318155912880415e-06, "loss": 0.2519, "step": 10767 }, { "epoch": 0.7444175596266851, "grad_norm": 0.6864639188132786, "learning_rate": 7.231314675686782e-06, "loss": 0.3888, "step": 10768 }, { "epoch": 0.7444866920152091, "grad_norm": 0.791001244176711, "learning_rate": 7.230813732119974e-06, "loss": 0.4368, "step": 10769 }, { "epoch": 0.7445558244037331, "grad_norm": 0.8904754971763646, "learning_rate": 7.230312760593899e-06, "loss": 0.3129, "step": 10770 }, { "epoch": 0.7446249567922572, "grad_norm": 0.7371732506160799, "learning_rate": 7.229811761114831e-06, "loss": 0.3417, "step": 10771 }, { "epoch": 0.7446940891807812, "grad_norm": 0.6619593126063004, "learning_rate": 7.229310733689053e-06, "loss": 0.356, "step": 10772 }, { "epoch": 0.7447632215693052, "grad_norm": 0.667930530545699, "learning_rate": 7.228809678322843e-06, "loss": 0.3513, "step": 10773 }, { "epoch": 0.7448323539578292, "grad_norm": 0.6731283883497293, "learning_rate": 7.228308595022483e-06, "loss": 0.3131, "step": 10774 }, { "epoch": 0.7449014863463532, "grad_norm": 0.7013157602445939, "learning_rate": 7.227807483794249e-06, "loss": 0.3115, "step": 10775 }, { "epoch": 0.7449706187348772, "grad_norm": 0.701168276141986, "learning_rate": 7.227306344644426e-06, "loss": 0.3907, "step": 10776 }, { "epoch": 0.7450397511234013, "grad_norm": 0.6758519375009631, "learning_rate": 7.226805177579291e-06, "loss": 0.2917, "step": 10777 }, { "epoch": 0.7451088835119254, "grad_norm": 0.6873130436958462, "learning_rate": 7.226303982605129e-06, "loss": 0.2799, "step": 10778 }, { "epoch": 0.7451780159004494, "grad_norm": 1.2028928931588418, "learning_rate": 7.225802759728218e-06, "loss": 0.3889, "step": 10779 }, { "epoch": 0.7452471482889734, "grad_norm": 0.6939646991447947, "learning_rate": 7.225301508954842e-06, "loss": 0.3008, "step": 10780 }, { "epoch": 0.7453162806774974, "grad_norm": 0.6689489917569407, "learning_rate": 7.224800230291282e-06, "loss": 0.3092, "step": 10781 }, { "epoch": 0.7453854130660215, "grad_norm": 0.8271742105340957, "learning_rate": 7.224298923743822e-06, "loss": 0.2975, "step": 10782 }, { "epoch": 0.7454545454545455, "grad_norm": 0.6726363826869886, "learning_rate": 7.223797589318745e-06, "loss": 0.4452, "step": 10783 }, { "epoch": 0.7455236778430695, "grad_norm": 0.7385619973445763, "learning_rate": 7.223296227022334e-06, "loss": 0.2367, "step": 10784 }, { "epoch": 0.7455928102315935, "grad_norm": 0.6226364064879235, "learning_rate": 7.222794836860871e-06, "loss": 0.3007, "step": 10785 }, { "epoch": 0.7456619426201175, "grad_norm": 0.732248045595136, "learning_rate": 7.222293418840642e-06, "loss": 0.4172, "step": 10786 }, { "epoch": 0.7457310750086416, "grad_norm": 0.653505047995183, "learning_rate": 7.221791972967931e-06, "loss": 0.4077, "step": 10787 }, { "epoch": 0.7458002073971656, "grad_norm": 0.6176123505622119, "learning_rate": 7.221290499249023e-06, "loss": 0.2794, "step": 10788 }, { "epoch": 0.7458693397856896, "grad_norm": 0.7009560337636126, "learning_rate": 7.220788997690203e-06, "loss": 0.3477, "step": 10789 }, { "epoch": 0.7459384721742136, "grad_norm": 0.8034372475182332, "learning_rate": 7.220287468297755e-06, "loss": 0.4436, "step": 10790 }, { "epoch": 0.7460076045627376, "grad_norm": 0.6751265839095753, "learning_rate": 7.219785911077967e-06, "loss": 0.3017, "step": 10791 }, { "epoch": 0.7460767369512616, "grad_norm": 0.7028544850179281, "learning_rate": 7.2192843260371235e-06, "loss": 0.2858, "step": 10792 }, { "epoch": 0.7461458693397857, "grad_norm": 0.7051519135240846, "learning_rate": 7.21878271318151e-06, "loss": 0.3962, "step": 10793 }, { "epoch": 0.7462150017283097, "grad_norm": 0.6446470831465642, "learning_rate": 7.218281072517417e-06, "loss": 0.3768, "step": 10794 }, { "epoch": 0.7462841341168337, "grad_norm": 0.7115858058702148, "learning_rate": 7.217779404051129e-06, "loss": 0.3256, "step": 10795 }, { "epoch": 0.7463532665053577, "grad_norm": 0.6331528475334568, "learning_rate": 7.217277707788933e-06, "loss": 0.3024, "step": 10796 }, { "epoch": 0.7464223988938817, "grad_norm": 0.6864300829633351, "learning_rate": 7.216775983737118e-06, "loss": 0.4433, "step": 10797 }, { "epoch": 0.7464915312824059, "grad_norm": 0.8910501108808718, "learning_rate": 7.2162742319019716e-06, "loss": 0.3101, "step": 10798 }, { "epoch": 0.7465606636709299, "grad_norm": 0.7002801404633687, "learning_rate": 7.215772452289784e-06, "loss": 0.2961, "step": 10799 }, { "epoch": 0.7466297960594539, "grad_norm": 0.6952813802487682, "learning_rate": 7.215270644906842e-06, "loss": 0.314, "step": 10800 }, { "epoch": 0.7466989284479779, "grad_norm": 0.7125598132823757, "learning_rate": 7.214768809759437e-06, "loss": 0.3822, "step": 10801 }, { "epoch": 0.7467680608365019, "grad_norm": 0.7220549543248769, "learning_rate": 7.214266946853857e-06, "loss": 0.2523, "step": 10802 }, { "epoch": 0.746837193225026, "grad_norm": 0.6369182238706046, "learning_rate": 7.213765056196392e-06, "loss": 0.2609, "step": 10803 }, { "epoch": 0.74690632561355, "grad_norm": 0.7123599813354861, "learning_rate": 7.213263137793332e-06, "loss": 0.3671, "step": 10804 }, { "epoch": 0.746975458002074, "grad_norm": 0.7558339768132678, "learning_rate": 7.2127611916509695e-06, "loss": 0.3923, "step": 10805 }, { "epoch": 0.747044590390598, "grad_norm": 0.6553922300401527, "learning_rate": 7.212259217775594e-06, "loss": 0.2972, "step": 10806 }, { "epoch": 0.747113722779122, "grad_norm": 0.6997007021758049, "learning_rate": 7.2117572161734986e-06, "loss": 0.3319, "step": 10807 }, { "epoch": 0.747182855167646, "grad_norm": 0.6534874540603111, "learning_rate": 7.2112551868509705e-06, "loss": 0.3897, "step": 10808 }, { "epoch": 0.7472519875561701, "grad_norm": 0.664260156351809, "learning_rate": 7.210753129814307e-06, "loss": 0.315, "step": 10809 }, { "epoch": 0.7473211199446941, "grad_norm": 0.6642570402737606, "learning_rate": 7.2102510450697985e-06, "loss": 0.2928, "step": 10810 }, { "epoch": 0.7473902523332181, "grad_norm": 0.6260520212576982, "learning_rate": 7.209748932623735e-06, "loss": 0.3399, "step": 10811 }, { "epoch": 0.7474593847217421, "grad_norm": 0.6431165551051335, "learning_rate": 7.209246792482415e-06, "loss": 0.3969, "step": 10812 }, { "epoch": 0.7475285171102661, "grad_norm": 0.5786728089203736, "learning_rate": 7.208744624652128e-06, "loss": 0.3038, "step": 10813 }, { "epoch": 0.7475976494987902, "grad_norm": 0.6686932891583464, "learning_rate": 7.208242429139167e-06, "loss": 0.2566, "step": 10814 }, { "epoch": 0.7476667818873142, "grad_norm": 0.688600342730391, "learning_rate": 7.20774020594983e-06, "loss": 0.4369, "step": 10815 }, { "epoch": 0.7477359142758382, "grad_norm": 0.8405538802561119, "learning_rate": 7.207237955090409e-06, "loss": 0.3205, "step": 10816 }, { "epoch": 0.7478050466643622, "grad_norm": 0.6274579650122019, "learning_rate": 7.2067356765672006e-06, "loss": 0.2449, "step": 10817 }, { "epoch": 0.7478741790528862, "grad_norm": 0.7090356160931989, "learning_rate": 7.206233370386496e-06, "loss": 0.3351, "step": 10818 }, { "epoch": 0.7479433114414104, "grad_norm": 0.6264305477638091, "learning_rate": 7.2057310365545955e-06, "loss": 0.3406, "step": 10819 }, { "epoch": 0.7480124438299344, "grad_norm": 1.0637755824793123, "learning_rate": 7.2052286750777915e-06, "loss": 0.2641, "step": 10820 }, { "epoch": 0.7480815762184584, "grad_norm": 0.7402388159897239, "learning_rate": 7.20472628596238e-06, "loss": 0.2565, "step": 10821 }, { "epoch": 0.7481507086069824, "grad_norm": 0.6973964998400559, "learning_rate": 7.2042238692146615e-06, "loss": 0.4496, "step": 10822 }, { "epoch": 0.7482198409955064, "grad_norm": 0.7390381902632956, "learning_rate": 7.20372142484093e-06, "loss": 0.3704, "step": 10823 }, { "epoch": 0.7482889733840304, "grad_norm": 0.6302536420536093, "learning_rate": 7.203218952847483e-06, "loss": 0.2761, "step": 10824 }, { "epoch": 0.7483581057725545, "grad_norm": 0.8525373710624796, "learning_rate": 7.202716453240617e-06, "loss": 0.3202, "step": 10825 }, { "epoch": 0.7484272381610785, "grad_norm": 0.7186075095391816, "learning_rate": 7.2022139260266334e-06, "loss": 0.3724, "step": 10826 }, { "epoch": 0.7484963705496025, "grad_norm": 0.6578110708696766, "learning_rate": 7.201711371211827e-06, "loss": 0.2981, "step": 10827 }, { "epoch": 0.7485655029381265, "grad_norm": 0.6533957414435616, "learning_rate": 7.201208788802496e-06, "loss": 0.2927, "step": 10828 }, { "epoch": 0.7486346353266505, "grad_norm": 0.6860358276704058, "learning_rate": 7.2007061788049436e-06, "loss": 0.3426, "step": 10829 }, { "epoch": 0.7487037677151746, "grad_norm": 0.6940176481502852, "learning_rate": 7.200203541225466e-06, "loss": 0.3248, "step": 10830 }, { "epoch": 0.7487729001036986, "grad_norm": 0.7057418004358296, "learning_rate": 7.199700876070363e-06, "loss": 0.3371, "step": 10831 }, { "epoch": 0.7488420324922226, "grad_norm": 0.7326222253371384, "learning_rate": 7.1991981833459345e-06, "loss": 0.2897, "step": 10832 }, { "epoch": 0.7489111648807466, "grad_norm": 0.6838583303081545, "learning_rate": 7.198695463058483e-06, "loss": 0.361, "step": 10833 }, { "epoch": 0.7489802972692706, "grad_norm": 0.6395082459268315, "learning_rate": 7.198192715214307e-06, "loss": 0.2636, "step": 10834 }, { "epoch": 0.7490494296577946, "grad_norm": 0.7593047857413432, "learning_rate": 7.197689939819708e-06, "loss": 0.276, "step": 10835 }, { "epoch": 0.7491185620463187, "grad_norm": 0.7177385832354176, "learning_rate": 7.197187136880987e-06, "loss": 0.3179, "step": 10836 }, { "epoch": 0.7491876944348427, "grad_norm": 0.6999177895186462, "learning_rate": 7.196684306404448e-06, "loss": 0.4034, "step": 10837 }, { "epoch": 0.7492568268233667, "grad_norm": 0.6194891253882939, "learning_rate": 7.19618144839639e-06, "loss": 0.299, "step": 10838 }, { "epoch": 0.7493259592118908, "grad_norm": 0.6332297249126181, "learning_rate": 7.195678562863117e-06, "loss": 0.3088, "step": 10839 }, { "epoch": 0.7493950916004148, "grad_norm": 0.7075715483183094, "learning_rate": 7.195175649810932e-06, "loss": 0.4094, "step": 10840 }, { "epoch": 0.7494642239889389, "grad_norm": 0.6588790283070575, "learning_rate": 7.194672709246137e-06, "loss": 0.3782, "step": 10841 }, { "epoch": 0.7495333563774629, "grad_norm": 0.6129061413055257, "learning_rate": 7.194169741175037e-06, "loss": 0.2706, "step": 10842 }, { "epoch": 0.7496024887659869, "grad_norm": 0.6238630785072696, "learning_rate": 7.193666745603933e-06, "loss": 0.2873, "step": 10843 }, { "epoch": 0.7496716211545109, "grad_norm": 0.994983278826479, "learning_rate": 7.1931637225391326e-06, "loss": 0.3925, "step": 10844 }, { "epoch": 0.7497407535430349, "grad_norm": 0.6728638424104947, "learning_rate": 7.19266067198694e-06, "loss": 0.2742, "step": 10845 }, { "epoch": 0.749809885931559, "grad_norm": 0.6805329189338853, "learning_rate": 7.192157593953657e-06, "loss": 0.2612, "step": 10846 }, { "epoch": 0.749879018320083, "grad_norm": 0.7058195984034089, "learning_rate": 7.191654488445591e-06, "loss": 0.3982, "step": 10847 }, { "epoch": 0.749948150708607, "grad_norm": 0.6655847422663194, "learning_rate": 7.191151355469048e-06, "loss": 0.3507, "step": 10848 }, { "epoch": 0.750017283097131, "grad_norm": 0.6338230735061695, "learning_rate": 7.190648195030332e-06, "loss": 0.286, "step": 10849 }, { "epoch": 0.750086415485655, "grad_norm": 0.7244107133021945, "learning_rate": 7.19014500713575e-06, "loss": 0.2939, "step": 10850 }, { "epoch": 0.750155547874179, "grad_norm": 0.750713212268061, "learning_rate": 7.189641791791611e-06, "loss": 0.4117, "step": 10851 }, { "epoch": 0.7502246802627031, "grad_norm": 0.7144789688105437, "learning_rate": 7.189138549004218e-06, "loss": 0.3345, "step": 10852 }, { "epoch": 0.7502938126512271, "grad_norm": 0.6386483350297434, "learning_rate": 7.18863527877988e-06, "loss": 0.2443, "step": 10853 }, { "epoch": 0.7503629450397511, "grad_norm": 0.721928018010897, "learning_rate": 7.188131981124904e-06, "loss": 0.3969, "step": 10854 }, { "epoch": 0.7504320774282751, "grad_norm": 0.6988448462765405, "learning_rate": 7.1876286560456e-06, "loss": 0.3338, "step": 10855 }, { "epoch": 0.7505012098167991, "grad_norm": 0.6098869679397882, "learning_rate": 7.1871253035482744e-06, "loss": 0.29, "step": 10856 }, { "epoch": 0.7505703422053231, "grad_norm": 0.775513082515217, "learning_rate": 7.186621923639235e-06, "loss": 0.3591, "step": 10857 }, { "epoch": 0.7506394745938472, "grad_norm": 0.6984731316308455, "learning_rate": 7.1861185163247935e-06, "loss": 0.3891, "step": 10858 }, { "epoch": 0.7507086069823713, "grad_norm": 0.7486267027229273, "learning_rate": 7.185615081611258e-06, "loss": 0.3113, "step": 10859 }, { "epoch": 0.7507777393708953, "grad_norm": 0.6243216127000255, "learning_rate": 7.185111619504937e-06, "loss": 0.2538, "step": 10860 }, { "epoch": 0.7508468717594193, "grad_norm": 0.8142828855610642, "learning_rate": 7.184608130012141e-06, "loss": 0.4675, "step": 10861 }, { "epoch": 0.7509160041479434, "grad_norm": 0.7219762408582939, "learning_rate": 7.184104613139181e-06, "loss": 0.4502, "step": 10862 }, { "epoch": 0.7509851365364674, "grad_norm": 0.6840105164677004, "learning_rate": 7.183601068892367e-06, "loss": 0.3017, "step": 10863 }, { "epoch": 0.7510542689249914, "grad_norm": 0.6501704393319724, "learning_rate": 7.183097497278011e-06, "loss": 0.2715, "step": 10864 }, { "epoch": 0.7511234013135154, "grad_norm": 0.8207948641998749, "learning_rate": 7.182593898302423e-06, "loss": 0.3914, "step": 10865 }, { "epoch": 0.7511925337020394, "grad_norm": 0.6558829744301639, "learning_rate": 7.182090271971918e-06, "loss": 0.2578, "step": 10866 }, { "epoch": 0.7512616660905634, "grad_norm": 0.6649662768215691, "learning_rate": 7.181586618292803e-06, "loss": 0.3037, "step": 10867 }, { "epoch": 0.7513307984790875, "grad_norm": 0.6808457821642431, "learning_rate": 7.181082937271393e-06, "loss": 0.4106, "step": 10868 }, { "epoch": 0.7513999308676115, "grad_norm": 1.4435750197395436, "learning_rate": 7.180579228914002e-06, "loss": 0.3147, "step": 10869 }, { "epoch": 0.7514690632561355, "grad_norm": 0.7068417834421855, "learning_rate": 7.180075493226941e-06, "loss": 0.3003, "step": 10870 }, { "epoch": 0.7515381956446595, "grad_norm": 0.7000280916426772, "learning_rate": 7.179571730216523e-06, "loss": 0.2932, "step": 10871 }, { "epoch": 0.7516073280331835, "grad_norm": 0.6837140859203803, "learning_rate": 7.179067939889066e-06, "loss": 0.3606, "step": 10872 }, { "epoch": 0.7516764604217075, "grad_norm": 0.660109173346949, "learning_rate": 7.1785641222508796e-06, "loss": 0.3537, "step": 10873 }, { "epoch": 0.7517455928102316, "grad_norm": 0.7131340220754092, "learning_rate": 7.17806027730828e-06, "loss": 0.3035, "step": 10874 }, { "epoch": 0.7518147251987556, "grad_norm": 0.6294583241530773, "learning_rate": 7.17755640506758e-06, "loss": 0.2653, "step": 10875 }, { "epoch": 0.7518838575872796, "grad_norm": 0.9218604101742806, "learning_rate": 7.177052505535098e-06, "loss": 0.3806, "step": 10876 }, { "epoch": 0.7519529899758036, "grad_norm": 0.7513675771408614, "learning_rate": 7.176548578717149e-06, "loss": 0.2791, "step": 10877 }, { "epoch": 0.7520221223643276, "grad_norm": 0.6755711092085439, "learning_rate": 7.176044624620046e-06, "loss": 0.2346, "step": 10878 }, { "epoch": 0.7520912547528517, "grad_norm": 0.6260098145452896, "learning_rate": 7.175540643250107e-06, "loss": 0.3493, "step": 10879 }, { "epoch": 0.7521603871413758, "grad_norm": 0.653656105110441, "learning_rate": 7.17503663461365e-06, "loss": 0.3116, "step": 10880 }, { "epoch": 0.7522295195298998, "grad_norm": 0.9649258430596364, "learning_rate": 7.174532598716989e-06, "loss": 0.2716, "step": 10881 }, { "epoch": 0.7522986519184238, "grad_norm": 0.915457931511692, "learning_rate": 7.174028535566443e-06, "loss": 0.2914, "step": 10882 }, { "epoch": 0.7523677843069478, "grad_norm": 0.7144365224018118, "learning_rate": 7.17352444516833e-06, "loss": 0.3791, "step": 10883 }, { "epoch": 0.7524369166954719, "grad_norm": 0.7046489931424511, "learning_rate": 7.173020327528966e-06, "loss": 0.3148, "step": 10884 }, { "epoch": 0.7525060490839959, "grad_norm": 0.7778650619868326, "learning_rate": 7.172516182654668e-06, "loss": 0.3016, "step": 10885 }, { "epoch": 0.7525751814725199, "grad_norm": 0.6659258978674719, "learning_rate": 7.172012010551758e-06, "loss": 0.3636, "step": 10886 }, { "epoch": 0.7526443138610439, "grad_norm": 0.7250181784902384, "learning_rate": 7.171507811226555e-06, "loss": 0.3457, "step": 10887 }, { "epoch": 0.7527134462495679, "grad_norm": 0.5942047836679903, "learning_rate": 7.171003584685377e-06, "loss": 0.2591, "step": 10888 }, { "epoch": 0.752782578638092, "grad_norm": 0.8219490756547447, "learning_rate": 7.170499330934542e-06, "loss": 0.2886, "step": 10889 }, { "epoch": 0.752851711026616, "grad_norm": 0.7103981809943031, "learning_rate": 7.169995049980371e-06, "loss": 0.366, "step": 10890 }, { "epoch": 0.75292084341514, "grad_norm": 0.6672884266372963, "learning_rate": 7.169490741829186e-06, "loss": 0.3162, "step": 10891 }, { "epoch": 0.752989975803664, "grad_norm": 0.6798811455346327, "learning_rate": 7.168986406487305e-06, "loss": 0.2604, "step": 10892 }, { "epoch": 0.753059108192188, "grad_norm": 0.6837786948685507, "learning_rate": 7.168482043961051e-06, "loss": 0.2911, "step": 10893 }, { "epoch": 0.753128240580712, "grad_norm": 0.7217250420990993, "learning_rate": 7.167977654256744e-06, "loss": 0.4297, "step": 10894 }, { "epoch": 0.7531973729692361, "grad_norm": 0.6668091648598129, "learning_rate": 7.167473237380706e-06, "loss": 0.333, "step": 10895 }, { "epoch": 0.7532665053577601, "grad_norm": 0.6621399321888997, "learning_rate": 7.166968793339259e-06, "loss": 0.248, "step": 10896 }, { "epoch": 0.7533356377462841, "grad_norm": 0.8329219672939688, "learning_rate": 7.166464322138724e-06, "loss": 0.3964, "step": 10897 }, { "epoch": 0.7534047701348081, "grad_norm": 0.6980098842358808, "learning_rate": 7.165959823785427e-06, "loss": 0.3569, "step": 10898 }, { "epoch": 0.7534739025233321, "grad_norm": 0.6162275879125351, "learning_rate": 7.165455298285689e-06, "loss": 0.2922, "step": 10899 }, { "epoch": 0.7535430349118563, "grad_norm": 0.7230423145994386, "learning_rate": 7.164950745645831e-06, "loss": 0.3462, "step": 10900 }, { "epoch": 0.7536121673003803, "grad_norm": 1.187607471841789, "learning_rate": 7.1644461658721805e-06, "loss": 0.3797, "step": 10901 }, { "epoch": 0.7536812996889043, "grad_norm": 0.6994345224679835, "learning_rate": 7.163941558971062e-06, "loss": 0.2801, "step": 10902 }, { "epoch": 0.7537504320774283, "grad_norm": 0.6374322131579375, "learning_rate": 7.163436924948793e-06, "loss": 0.2706, "step": 10903 }, { "epoch": 0.7538195644659523, "grad_norm": 0.6398615150300475, "learning_rate": 7.162932263811706e-06, "loss": 0.3653, "step": 10904 }, { "epoch": 0.7538886968544763, "grad_norm": 0.6250438917679639, "learning_rate": 7.162427575566123e-06, "loss": 0.3221, "step": 10905 }, { "epoch": 0.7539578292430004, "grad_norm": 0.6572074969969318, "learning_rate": 7.16192286021837e-06, "loss": 0.3348, "step": 10906 }, { "epoch": 0.7540269616315244, "grad_norm": 0.6378973646126427, "learning_rate": 7.16141811777477e-06, "loss": 0.3257, "step": 10907 }, { "epoch": 0.7540960940200484, "grad_norm": 0.7156112820640796, "learning_rate": 7.160913348241653e-06, "loss": 0.4195, "step": 10908 }, { "epoch": 0.7541652264085724, "grad_norm": 0.7941268846822044, "learning_rate": 7.160408551625342e-06, "loss": 0.3379, "step": 10909 }, { "epoch": 0.7542343587970964, "grad_norm": 0.6719205465432516, "learning_rate": 7.159903727932165e-06, "loss": 0.2916, "step": 10910 }, { "epoch": 0.7543034911856205, "grad_norm": 0.7248776987527144, "learning_rate": 7.159398877168451e-06, "loss": 0.3132, "step": 10911 }, { "epoch": 0.7543726235741445, "grad_norm": 0.6454737776711057, "learning_rate": 7.158893999340525e-06, "loss": 0.3781, "step": 10912 }, { "epoch": 0.7544417559626685, "grad_norm": 0.8659536650991926, "learning_rate": 7.158389094454715e-06, "loss": 0.2919, "step": 10913 }, { "epoch": 0.7545108883511925, "grad_norm": 0.6702668549336867, "learning_rate": 7.15788416251735e-06, "loss": 0.305, "step": 10914 }, { "epoch": 0.7545800207397165, "grad_norm": 0.8076184054677543, "learning_rate": 7.157379203534758e-06, "loss": 0.3834, "step": 10915 }, { "epoch": 0.7546491531282405, "grad_norm": 0.6466072601446591, "learning_rate": 7.156874217513268e-06, "loss": 0.3194, "step": 10916 }, { "epoch": 0.7547182855167646, "grad_norm": 0.6619465878303926, "learning_rate": 7.156369204459207e-06, "loss": 0.2925, "step": 10917 }, { "epoch": 0.7547874179052886, "grad_norm": 0.7422709181854465, "learning_rate": 7.155864164378907e-06, "loss": 0.3938, "step": 10918 }, { "epoch": 0.7548565502938126, "grad_norm": 0.7187319148057341, "learning_rate": 7.155359097278698e-06, "loss": 0.3954, "step": 10919 }, { "epoch": 0.7549256826823367, "grad_norm": 0.613279446903531, "learning_rate": 7.15485400316491e-06, "loss": 0.2549, "step": 10920 }, { "epoch": 0.7549948150708607, "grad_norm": 0.6717642596452756, "learning_rate": 7.15434888204387e-06, "loss": 0.2809, "step": 10921 }, { "epoch": 0.7550639474593848, "grad_norm": 0.673637126163773, "learning_rate": 7.1538437339219125e-06, "loss": 0.4168, "step": 10922 }, { "epoch": 0.7551330798479088, "grad_norm": 0.6328244906774648, "learning_rate": 7.153338558805368e-06, "loss": 0.3167, "step": 10923 }, { "epoch": 0.7552022122364328, "grad_norm": 0.643480201837461, "learning_rate": 7.1528333567005655e-06, "loss": 0.3309, "step": 10924 }, { "epoch": 0.7552713446249568, "grad_norm": 0.7350707565892575, "learning_rate": 7.1523281276138416e-06, "loss": 0.3161, "step": 10925 }, { "epoch": 0.7553404770134808, "grad_norm": 0.7707383545213118, "learning_rate": 7.151822871551525e-06, "loss": 0.3878, "step": 10926 }, { "epoch": 0.7554096094020049, "grad_norm": 0.6681162112770962, "learning_rate": 7.151317588519948e-06, "loss": 0.2554, "step": 10927 }, { "epoch": 0.7554787417905289, "grad_norm": 0.6900687919651761, "learning_rate": 7.150812278525443e-06, "loss": 0.2813, "step": 10928 }, { "epoch": 0.7555478741790529, "grad_norm": 0.8112537864763074, "learning_rate": 7.150306941574346e-06, "loss": 0.3882, "step": 10929 }, { "epoch": 0.7556170065675769, "grad_norm": 0.7027176424017401, "learning_rate": 7.14980157767299e-06, "loss": 0.3654, "step": 10930 }, { "epoch": 0.7556861389561009, "grad_norm": 0.6980188926212412, "learning_rate": 7.149296186827707e-06, "loss": 0.3155, "step": 10931 }, { "epoch": 0.755755271344625, "grad_norm": 0.6849790143440267, "learning_rate": 7.14879076904483e-06, "loss": 0.301, "step": 10932 }, { "epoch": 0.755824403733149, "grad_norm": 0.7113269975842293, "learning_rate": 7.148285324330697e-06, "loss": 0.4067, "step": 10933 }, { "epoch": 0.755893536121673, "grad_norm": 0.6952201795650942, "learning_rate": 7.147779852691641e-06, "loss": 0.2923, "step": 10934 }, { "epoch": 0.755962668510197, "grad_norm": 0.9797211397682658, "learning_rate": 7.1472743541339965e-06, "loss": 0.3252, "step": 10935 }, { "epoch": 0.756031800898721, "grad_norm": 1.0107251325056419, "learning_rate": 7.146768828664101e-06, "loss": 0.3644, "step": 10936 }, { "epoch": 0.756100933287245, "grad_norm": 0.7397311500389198, "learning_rate": 7.14626327628829e-06, "loss": 0.3639, "step": 10937 }, { "epoch": 0.756170065675769, "grad_norm": 0.8060230376806423, "learning_rate": 7.145757697012898e-06, "loss": 0.2888, "step": 10938 }, { "epoch": 0.7562391980642931, "grad_norm": 0.6524545965592538, "learning_rate": 7.145252090844261e-06, "loss": 0.2856, "step": 10939 }, { "epoch": 0.7563083304528171, "grad_norm": 0.7919232711762603, "learning_rate": 7.1447464577887205e-06, "loss": 0.4265, "step": 10940 }, { "epoch": 0.7563774628413412, "grad_norm": 0.7399651021407232, "learning_rate": 7.144240797852609e-06, "loss": 0.3068, "step": 10941 }, { "epoch": 0.7564465952298652, "grad_norm": 0.8234431218456242, "learning_rate": 7.143735111042265e-06, "loss": 0.2936, "step": 10942 }, { "epoch": 0.7565157276183893, "grad_norm": 0.6617238612505405, "learning_rate": 7.143229397364029e-06, "loss": 0.3912, "step": 10943 }, { "epoch": 0.7565848600069133, "grad_norm": 0.6851727195088279, "learning_rate": 7.142723656824237e-06, "loss": 0.4081, "step": 10944 }, { "epoch": 0.7566539923954373, "grad_norm": 0.6878111281392918, "learning_rate": 7.142217889429226e-06, "loss": 0.3297, "step": 10945 }, { "epoch": 0.7567231247839613, "grad_norm": 0.6405402298508438, "learning_rate": 7.141712095185338e-06, "loss": 0.2478, "step": 10946 }, { "epoch": 0.7567922571724853, "grad_norm": 0.6652640049870786, "learning_rate": 7.141206274098911e-06, "loss": 0.4307, "step": 10947 }, { "epoch": 0.7568613895610093, "grad_norm": 0.6481167144684497, "learning_rate": 7.140700426176285e-06, "loss": 0.3648, "step": 10948 }, { "epoch": 0.7569305219495334, "grad_norm": 0.8555413084130244, "learning_rate": 7.140194551423798e-06, "loss": 0.3622, "step": 10949 }, { "epoch": 0.7569996543380574, "grad_norm": 0.7330854188100099, "learning_rate": 7.139688649847791e-06, "loss": 0.3636, "step": 10950 }, { "epoch": 0.7570687867265814, "grad_norm": 0.6330359304072195, "learning_rate": 7.139182721454607e-06, "loss": 0.3904, "step": 10951 }, { "epoch": 0.7571379191151054, "grad_norm": 0.6554992499114789, "learning_rate": 7.138676766250585e-06, "loss": 0.3085, "step": 10952 }, { "epoch": 0.7572070515036294, "grad_norm": 0.8290350626971719, "learning_rate": 7.1381707842420655e-06, "loss": 0.2807, "step": 10953 }, { "epoch": 0.7572761838921535, "grad_norm": 0.7053048417276049, "learning_rate": 7.137664775435392e-06, "loss": 0.4243, "step": 10954 }, { "epoch": 0.7573453162806775, "grad_norm": 0.6911274424074392, "learning_rate": 7.1371587398369045e-06, "loss": 0.3761, "step": 10955 }, { "epoch": 0.7574144486692015, "grad_norm": 0.8982238691692472, "learning_rate": 7.136652677452947e-06, "loss": 0.3544, "step": 10956 }, { "epoch": 0.7574835810577255, "grad_norm": 0.7473604521622683, "learning_rate": 7.136146588289861e-06, "loss": 0.3164, "step": 10957 }, { "epoch": 0.7575527134462495, "grad_norm": 0.6511945148030492, "learning_rate": 7.135640472353989e-06, "loss": 0.3986, "step": 10958 }, { "epoch": 0.7576218458347735, "grad_norm": 0.7033696989814928, "learning_rate": 7.135134329651675e-06, "loss": 0.3283, "step": 10959 }, { "epoch": 0.7576909782232976, "grad_norm": 0.761117900378964, "learning_rate": 7.134628160189262e-06, "loss": 0.3192, "step": 10960 }, { "epoch": 0.7577601106118217, "grad_norm": 0.7479976071027151, "learning_rate": 7.134121963973095e-06, "loss": 0.3068, "step": 10961 }, { "epoch": 0.7578292430003457, "grad_norm": 0.6994160551951395, "learning_rate": 7.133615741009519e-06, "loss": 0.4613, "step": 10962 }, { "epoch": 0.7578983753888697, "grad_norm": 0.6371637402394228, "learning_rate": 7.1331094913048744e-06, "loss": 0.2523, "step": 10963 }, { "epoch": 0.7579675077773937, "grad_norm": 0.704304845955971, "learning_rate": 7.132603214865511e-06, "loss": 0.3354, "step": 10964 }, { "epoch": 0.7580366401659178, "grad_norm": 0.7548453368338242, "learning_rate": 7.132096911697772e-06, "loss": 0.4322, "step": 10965 }, { "epoch": 0.7581057725544418, "grad_norm": 0.6567960847113151, "learning_rate": 7.1315905818080034e-06, "loss": 0.3118, "step": 10966 }, { "epoch": 0.7581749049429658, "grad_norm": 0.6128887531108453, "learning_rate": 7.13108422520255e-06, "loss": 0.3006, "step": 10967 }, { "epoch": 0.7582440373314898, "grad_norm": 0.673646198199333, "learning_rate": 7.13057784188776e-06, "loss": 0.3324, "step": 10968 }, { "epoch": 0.7583131697200138, "grad_norm": 0.6907518854852771, "learning_rate": 7.130071431869978e-06, "loss": 0.4038, "step": 10969 }, { "epoch": 0.7583823021085379, "grad_norm": 0.6105753235880106, "learning_rate": 7.129564995155551e-06, "loss": 0.2925, "step": 10970 }, { "epoch": 0.7584514344970619, "grad_norm": 0.9922288247977352, "learning_rate": 7.129058531750829e-06, "loss": 0.3003, "step": 10971 }, { "epoch": 0.7585205668855859, "grad_norm": 0.7086655481533054, "learning_rate": 7.128552041662158e-06, "loss": 0.3888, "step": 10972 }, { "epoch": 0.7585896992741099, "grad_norm": 0.9642632006156541, "learning_rate": 7.1280455248958845e-06, "loss": 0.3166, "step": 10973 }, { "epoch": 0.7586588316626339, "grad_norm": 0.6711203385631225, "learning_rate": 7.127538981458357e-06, "loss": 0.3162, "step": 10974 }, { "epoch": 0.7587279640511579, "grad_norm": 0.8022051741580237, "learning_rate": 7.127032411355927e-06, "loss": 0.3127, "step": 10975 }, { "epoch": 0.758797096439682, "grad_norm": 0.6760268339231869, "learning_rate": 7.1265258145949425e-06, "loss": 0.4108, "step": 10976 }, { "epoch": 0.758866228828206, "grad_norm": 1.0882337801209787, "learning_rate": 7.12601919118175e-06, "loss": 0.2774, "step": 10977 }, { "epoch": 0.75893536121673, "grad_norm": 0.6138617558495657, "learning_rate": 7.1255125411227015e-06, "loss": 0.2304, "step": 10978 }, { "epoch": 0.759004493605254, "grad_norm": 0.6579388063672671, "learning_rate": 7.125005864424146e-06, "loss": 0.4355, "step": 10979 }, { "epoch": 0.759073625993778, "grad_norm": 0.8595563855279648, "learning_rate": 7.124499161092434e-06, "loss": 0.2954, "step": 10980 }, { "epoch": 0.7591427583823022, "grad_norm": 0.692456927688475, "learning_rate": 7.123992431133917e-06, "loss": 0.2918, "step": 10981 }, { "epoch": 0.7592118907708262, "grad_norm": 0.9567122872064723, "learning_rate": 7.1234856745549455e-06, "loss": 0.3752, "step": 10982 }, { "epoch": 0.7592810231593502, "grad_norm": 0.6996011116255244, "learning_rate": 7.12297889136187e-06, "loss": 0.3794, "step": 10983 }, { "epoch": 0.7593501555478742, "grad_norm": 0.672201336467138, "learning_rate": 7.122472081561044e-06, "loss": 0.2755, "step": 10984 }, { "epoch": 0.7594192879363982, "grad_norm": 0.7093741754343956, "learning_rate": 7.121965245158816e-06, "loss": 0.2606, "step": 10985 }, { "epoch": 0.7594884203249223, "grad_norm": 0.701970574333703, "learning_rate": 7.121458382161542e-06, "loss": 0.377, "step": 10986 }, { "epoch": 0.7595575527134463, "grad_norm": 0.6373059641221263, "learning_rate": 7.120951492575573e-06, "loss": 0.4013, "step": 10987 }, { "epoch": 0.7596266851019703, "grad_norm": 0.6544904643750293, "learning_rate": 7.1204445764072615e-06, "loss": 0.3319, "step": 10988 }, { "epoch": 0.7596958174904943, "grad_norm": 0.7750522749946873, "learning_rate": 7.119937633662962e-06, "loss": 0.3142, "step": 10989 }, { "epoch": 0.7597649498790183, "grad_norm": 1.596906614809802, "learning_rate": 7.119430664349028e-06, "loss": 0.4247, "step": 10990 }, { "epoch": 0.7598340822675423, "grad_norm": 0.7290264790490901, "learning_rate": 7.118923668471811e-06, "loss": 0.4725, "step": 10991 }, { "epoch": 0.7599032146560664, "grad_norm": 0.6433575875776064, "learning_rate": 7.118416646037668e-06, "loss": 0.277, "step": 10992 }, { "epoch": 0.7599723470445904, "grad_norm": 0.6162291010939432, "learning_rate": 7.117909597052953e-06, "loss": 0.3394, "step": 10993 }, { "epoch": 0.7600414794331144, "grad_norm": 0.6695601665267338, "learning_rate": 7.117402521524022e-06, "loss": 0.3243, "step": 10994 }, { "epoch": 0.7601106118216384, "grad_norm": 0.6440104815968457, "learning_rate": 7.116895419457228e-06, "loss": 0.2774, "step": 10995 }, { "epoch": 0.7601797442101624, "grad_norm": 0.6545587143851961, "learning_rate": 7.116388290858927e-06, "loss": 0.2479, "step": 10996 }, { "epoch": 0.7602488765986865, "grad_norm": 0.6545850111753796, "learning_rate": 7.115881135735476e-06, "loss": 0.3572, "step": 10997 }, { "epoch": 0.7603180089872105, "grad_norm": 0.708042331168689, "learning_rate": 7.115373954093232e-06, "loss": 0.357, "step": 10998 }, { "epoch": 0.7603871413757345, "grad_norm": 0.7024699260027262, "learning_rate": 7.114866745938549e-06, "loss": 0.2939, "step": 10999 }, { "epoch": 0.7604562737642585, "grad_norm": 0.6733105472671198, "learning_rate": 7.114359511277786e-06, "loss": 0.3147, "step": 11000 }, { "epoch": 0.7605254061527825, "grad_norm": 0.621629383740373, "learning_rate": 7.1138522501173005e-06, "loss": 0.3767, "step": 11001 }, { "epoch": 0.7605945385413067, "grad_norm": 0.6435821521277438, "learning_rate": 7.1133449624634486e-06, "loss": 0.3267, "step": 11002 }, { "epoch": 0.7606636709298307, "grad_norm": 0.7031947388631227, "learning_rate": 7.112837648322589e-06, "loss": 0.2703, "step": 11003 }, { "epoch": 0.7607328033183547, "grad_norm": 0.7111578452066346, "learning_rate": 7.112330307701081e-06, "loss": 0.3798, "step": 11004 }, { "epoch": 0.7608019357068787, "grad_norm": 0.68680245452685, "learning_rate": 7.111822940605282e-06, "loss": 0.2943, "step": 11005 }, { "epoch": 0.7608710680954027, "grad_norm": 0.654518467866905, "learning_rate": 7.111315547041549e-06, "loss": 0.2543, "step": 11006 }, { "epoch": 0.7609402004839267, "grad_norm": 0.6955050523882697, "learning_rate": 7.110808127016246e-06, "loss": 0.2937, "step": 11007 }, { "epoch": 0.7610093328724508, "grad_norm": 0.724075722801596, "learning_rate": 7.110300680535729e-06, "loss": 0.3856, "step": 11008 }, { "epoch": 0.7610784652609748, "grad_norm": 0.6426762504006984, "learning_rate": 7.109793207606359e-06, "loss": 0.2766, "step": 11009 }, { "epoch": 0.7611475976494988, "grad_norm": 0.7070129925709553, "learning_rate": 7.109285708234497e-06, "loss": 0.2745, "step": 11010 }, { "epoch": 0.7612167300380228, "grad_norm": 0.7589926673870779, "learning_rate": 7.108778182426503e-06, "loss": 0.4172, "step": 11011 }, { "epoch": 0.7612858624265468, "grad_norm": 0.8122906548478505, "learning_rate": 7.108270630188737e-06, "loss": 0.3684, "step": 11012 }, { "epoch": 0.7613549948150709, "grad_norm": 0.6299144410634078, "learning_rate": 7.107763051527561e-06, "loss": 0.2498, "step": 11013 }, { "epoch": 0.7614241272035949, "grad_norm": 0.6911857306102712, "learning_rate": 7.107255446449336e-06, "loss": 0.2973, "step": 11014 }, { "epoch": 0.7614932595921189, "grad_norm": 0.7288084142405276, "learning_rate": 7.106747814960425e-06, "loss": 0.4435, "step": 11015 }, { "epoch": 0.7615623919806429, "grad_norm": 0.6361684567792693, "learning_rate": 7.106240157067191e-06, "loss": 0.2976, "step": 11016 }, { "epoch": 0.7616315243691669, "grad_norm": 0.6913511437699102, "learning_rate": 7.105732472775994e-06, "loss": 0.2949, "step": 11017 }, { "epoch": 0.7617006567576909, "grad_norm": 0.765080313051298, "learning_rate": 7.105224762093198e-06, "loss": 0.3923, "step": 11018 }, { "epoch": 0.761769789146215, "grad_norm": 0.6564380539991038, "learning_rate": 7.104717025025167e-06, "loss": 0.38, "step": 11019 }, { "epoch": 0.761838921534739, "grad_norm": 0.6098903057789145, "learning_rate": 7.1042092615782644e-06, "loss": 0.3462, "step": 11020 }, { "epoch": 0.761908053923263, "grad_norm": 0.6759089915985627, "learning_rate": 7.103701471758854e-06, "loss": 0.3118, "step": 11021 }, { "epoch": 0.7619771863117871, "grad_norm": 0.7074931467901747, "learning_rate": 7.103193655573299e-06, "loss": 0.3572, "step": 11022 }, { "epoch": 0.7620463187003111, "grad_norm": 0.7036678786597935, "learning_rate": 7.1026858130279655e-06, "loss": 0.3667, "step": 11023 }, { "epoch": 0.7621154510888352, "grad_norm": 0.6193324510787015, "learning_rate": 7.102177944129218e-06, "loss": 0.3074, "step": 11024 }, { "epoch": 0.7621845834773592, "grad_norm": 0.6656673867752158, "learning_rate": 7.101670048883422e-06, "loss": 0.3227, "step": 11025 }, { "epoch": 0.7622537158658832, "grad_norm": 0.8317786171524026, "learning_rate": 7.101162127296941e-06, "loss": 0.4092, "step": 11026 }, { "epoch": 0.7623228482544072, "grad_norm": 0.6909419340909754, "learning_rate": 7.100654179376143e-06, "loss": 0.3468, "step": 11027 }, { "epoch": 0.7623919806429312, "grad_norm": 0.7002583324317306, "learning_rate": 7.1001462051273935e-06, "loss": 0.2532, "step": 11028 }, { "epoch": 0.7624611130314553, "grad_norm": 0.6878164371025159, "learning_rate": 7.09963820455706e-06, "loss": 0.3748, "step": 11029 }, { "epoch": 0.7625302454199793, "grad_norm": 0.697304927170354, "learning_rate": 7.099130177671509e-06, "loss": 0.3556, "step": 11030 }, { "epoch": 0.7625993778085033, "grad_norm": 0.6486947267301092, "learning_rate": 7.098622124477106e-06, "loss": 0.3416, "step": 11031 }, { "epoch": 0.7626685101970273, "grad_norm": 0.7711659913805378, "learning_rate": 7.09811404498022e-06, "loss": 0.3465, "step": 11032 }, { "epoch": 0.7627376425855513, "grad_norm": 0.6193842779425138, "learning_rate": 7.097605939187219e-06, "loss": 0.3772, "step": 11033 }, { "epoch": 0.7628067749740753, "grad_norm": 0.6565639454842721, "learning_rate": 7.097097807104471e-06, "loss": 0.2521, "step": 11034 }, { "epoch": 0.7628759073625994, "grad_norm": 0.8225120860084792, "learning_rate": 7.096589648738343e-06, "loss": 0.3046, "step": 11035 }, { "epoch": 0.7629450397511234, "grad_norm": 0.7151246415921177, "learning_rate": 7.0960814640952084e-06, "loss": 0.3554, "step": 11036 }, { "epoch": 0.7630141721396474, "grad_norm": 0.8170235087256731, "learning_rate": 7.09557325318143e-06, "loss": 0.3426, "step": 11037 }, { "epoch": 0.7630833045281714, "grad_norm": 0.6677122918697104, "learning_rate": 7.0950650160033816e-06, "loss": 0.2474, "step": 11038 }, { "epoch": 0.7631524369166954, "grad_norm": 0.6529756714434047, "learning_rate": 7.094556752567433e-06, "loss": 0.3101, "step": 11039 }, { "epoch": 0.7632215693052194, "grad_norm": 0.6228774969137146, "learning_rate": 7.094048462879953e-06, "loss": 0.3429, "step": 11040 }, { "epoch": 0.7632907016937435, "grad_norm": 0.6691153414295832, "learning_rate": 7.093540146947312e-06, "loss": 0.3433, "step": 11041 }, { "epoch": 0.7633598340822676, "grad_norm": 0.6850503309938016, "learning_rate": 7.093031804775882e-06, "loss": 0.2924, "step": 11042 }, { "epoch": 0.7634289664707916, "grad_norm": 0.677172068730386, "learning_rate": 7.092523436372033e-06, "loss": 0.3199, "step": 11043 }, { "epoch": 0.7634980988593156, "grad_norm": 0.6860402754835924, "learning_rate": 7.092015041742137e-06, "loss": 0.4122, "step": 11044 }, { "epoch": 0.7635672312478397, "grad_norm": 1.2044804709362147, "learning_rate": 7.091506620892566e-06, "loss": 0.2795, "step": 11045 }, { "epoch": 0.7636363636363637, "grad_norm": 1.0600005485870863, "learning_rate": 7.090998173829692e-06, "loss": 0.2647, "step": 11046 }, { "epoch": 0.7637054960248877, "grad_norm": 0.6612806337291656, "learning_rate": 7.090489700559888e-06, "loss": 0.3883, "step": 11047 }, { "epoch": 0.7637746284134117, "grad_norm": 0.7315000030213766, "learning_rate": 7.089981201089526e-06, "loss": 0.2819, "step": 11048 }, { "epoch": 0.7638437608019357, "grad_norm": 0.7521710649382878, "learning_rate": 7.089472675424977e-06, "loss": 0.2807, "step": 11049 }, { "epoch": 0.7639128931904597, "grad_norm": 0.8750303260276281, "learning_rate": 7.08896412357262e-06, "loss": 0.3303, "step": 11050 }, { "epoch": 0.7639820255789838, "grad_norm": 0.6569267333181411, "learning_rate": 7.088455545538825e-06, "loss": 0.3831, "step": 11051 }, { "epoch": 0.7640511579675078, "grad_norm": 0.6828021721494494, "learning_rate": 7.087946941329966e-06, "loss": 0.2573, "step": 11052 }, { "epoch": 0.7641202903560318, "grad_norm": 0.6614681178783093, "learning_rate": 7.087438310952418e-06, "loss": 0.2217, "step": 11053 }, { "epoch": 0.7641894227445558, "grad_norm": 0.6814831060706554, "learning_rate": 7.086929654412558e-06, "loss": 0.3792, "step": 11054 }, { "epoch": 0.7642585551330798, "grad_norm": 0.7471778774569413, "learning_rate": 7.086420971716756e-06, "loss": 0.4042, "step": 11055 }, { "epoch": 0.7643276875216038, "grad_norm": 0.6586595379784519, "learning_rate": 7.085912262871393e-06, "loss": 0.251, "step": 11056 }, { "epoch": 0.7643968199101279, "grad_norm": 0.6870963876026098, "learning_rate": 7.085403527882842e-06, "loss": 0.2933, "step": 11057 }, { "epoch": 0.7644659522986519, "grad_norm": 0.6802596772467804, "learning_rate": 7.084894766757477e-06, "loss": 0.4314, "step": 11058 }, { "epoch": 0.7645350846871759, "grad_norm": 2.3990806702438907, "learning_rate": 7.084385979501678e-06, "loss": 0.2846, "step": 11059 }, { "epoch": 0.7646042170756999, "grad_norm": 0.5953980864587552, "learning_rate": 7.0838771661218216e-06, "loss": 0.2357, "step": 11060 }, { "epoch": 0.7646733494642239, "grad_norm": 0.7233136701946763, "learning_rate": 7.083368326624284e-06, "loss": 0.4159, "step": 11061 }, { "epoch": 0.764742481852748, "grad_norm": 1.3403179585573122, "learning_rate": 7.082859461015441e-06, "loss": 0.3842, "step": 11062 }, { "epoch": 0.7648116142412721, "grad_norm": 0.6769087935902449, "learning_rate": 7.0823505693016715e-06, "loss": 0.3429, "step": 11063 }, { "epoch": 0.7648807466297961, "grad_norm": 0.6416375758097125, "learning_rate": 7.081841651489354e-06, "loss": 0.2742, "step": 11064 }, { "epoch": 0.7649498790183201, "grad_norm": 0.6465832515282869, "learning_rate": 7.081332707584868e-06, "loss": 0.3995, "step": 11065 }, { "epoch": 0.7650190114068441, "grad_norm": 0.7293353479305903, "learning_rate": 7.08082373759459e-06, "loss": 0.342, "step": 11066 }, { "epoch": 0.7650881437953682, "grad_norm": 0.6787838443035333, "learning_rate": 7.0803147415248985e-06, "loss": 0.238, "step": 11067 }, { "epoch": 0.7651572761838922, "grad_norm": 0.6776141146647126, "learning_rate": 7.079805719382176e-06, "loss": 0.374, "step": 11068 }, { "epoch": 0.7652264085724162, "grad_norm": 1.1424545902599905, "learning_rate": 7.079296671172799e-06, "loss": 0.3752, "step": 11069 }, { "epoch": 0.7652955409609402, "grad_norm": 0.646650976869644, "learning_rate": 7.078787596903149e-06, "loss": 0.3142, "step": 11070 }, { "epoch": 0.7653646733494642, "grad_norm": 0.9477746684260077, "learning_rate": 7.078278496579608e-06, "loss": 0.2968, "step": 11071 }, { "epoch": 0.7654338057379882, "grad_norm": 0.7379252388700481, "learning_rate": 7.077769370208554e-06, "loss": 0.3802, "step": 11072 }, { "epoch": 0.7655029381265123, "grad_norm": 0.654512868536872, "learning_rate": 7.077260217796369e-06, "loss": 0.3194, "step": 11073 }, { "epoch": 0.7655720705150363, "grad_norm": 0.8091646246417683, "learning_rate": 7.076751039349433e-06, "loss": 0.2871, "step": 11074 }, { "epoch": 0.7656412029035603, "grad_norm": 0.7472789015598821, "learning_rate": 7.07624183487413e-06, "loss": 0.382, "step": 11075 }, { "epoch": 0.7657103352920843, "grad_norm": 0.7148072347397252, "learning_rate": 7.0757326043768415e-06, "loss": 0.4375, "step": 11076 }, { "epoch": 0.7657794676806083, "grad_norm": 0.7146331711771828, "learning_rate": 7.075223347863948e-06, "loss": 0.2755, "step": 11077 }, { "epoch": 0.7658486000691324, "grad_norm": 0.7240246772577644, "learning_rate": 7.074714065341834e-06, "loss": 0.3073, "step": 11078 }, { "epoch": 0.7659177324576564, "grad_norm": 0.7251578296654301, "learning_rate": 7.07420475681688e-06, "loss": 0.4005, "step": 11079 }, { "epoch": 0.7659868648461804, "grad_norm": 1.5666120033141657, "learning_rate": 7.0736954222954725e-06, "loss": 0.3792, "step": 11080 }, { "epoch": 0.7660559972347044, "grad_norm": 0.8487260002863508, "learning_rate": 7.073186061783993e-06, "loss": 0.2578, "step": 11081 }, { "epoch": 0.7661251296232284, "grad_norm": 0.6772266283423709, "learning_rate": 7.072676675288827e-06, "loss": 0.3283, "step": 11082 }, { "epoch": 0.7661942620117526, "grad_norm": 0.7676707173021051, "learning_rate": 7.072167262816357e-06, "loss": 0.4008, "step": 11083 }, { "epoch": 0.7662633944002766, "grad_norm": 1.8418051625679264, "learning_rate": 7.071657824372967e-06, "loss": 0.2459, "step": 11084 }, { "epoch": 0.7663325267888006, "grad_norm": 0.7066502907711422, "learning_rate": 7.071148359965045e-06, "loss": 0.2627, "step": 11085 }, { "epoch": 0.7664016591773246, "grad_norm": 0.6422613847432261, "learning_rate": 7.070638869598973e-06, "loss": 0.3466, "step": 11086 }, { "epoch": 0.7664707915658486, "grad_norm": 0.7089904878935755, "learning_rate": 7.070129353281139e-06, "loss": 0.3882, "step": 11087 }, { "epoch": 0.7665399239543726, "grad_norm": 0.712589899625456, "learning_rate": 7.0696198110179275e-06, "loss": 0.3184, "step": 11088 }, { "epoch": 0.7666090563428967, "grad_norm": 0.7709227765313335, "learning_rate": 7.069110242815725e-06, "loss": 0.3303, "step": 11089 }, { "epoch": 0.7666781887314207, "grad_norm": 0.7194154680446845, "learning_rate": 7.068600648680919e-06, "loss": 0.3684, "step": 11090 }, { "epoch": 0.7667473211199447, "grad_norm": 0.6528900845619311, "learning_rate": 7.0680910286198945e-06, "loss": 0.3428, "step": 11091 }, { "epoch": 0.7668164535084687, "grad_norm": 0.6650333909752718, "learning_rate": 7.067581382639039e-06, "loss": 0.3354, "step": 11092 }, { "epoch": 0.7668855858969927, "grad_norm": 0.7158053823866607, "learning_rate": 7.067071710744742e-06, "loss": 0.3505, "step": 11093 }, { "epoch": 0.7669547182855168, "grad_norm": 0.7329164546765758, "learning_rate": 7.066562012943389e-06, "loss": 0.3923, "step": 11094 }, { "epoch": 0.7670238506740408, "grad_norm": 0.6660453664699393, "learning_rate": 7.06605228924137e-06, "loss": 0.277, "step": 11095 }, { "epoch": 0.7670929830625648, "grad_norm": 0.6316267866017214, "learning_rate": 7.065542539645073e-06, "loss": 0.2765, "step": 11096 }, { "epoch": 0.7671621154510888, "grad_norm": 0.7067505155677364, "learning_rate": 7.065032764160885e-06, "loss": 0.3536, "step": 11097 }, { "epoch": 0.7672312478396128, "grad_norm": 0.8024695366774295, "learning_rate": 7.064522962795197e-06, "loss": 0.3061, "step": 11098 }, { "epoch": 0.7673003802281368, "grad_norm": 0.6652853690543837, "learning_rate": 7.064013135554397e-06, "loss": 0.2826, "step": 11099 }, { "epoch": 0.7673695126166609, "grad_norm": 0.7225836303416658, "learning_rate": 7.063503282444878e-06, "loss": 0.3633, "step": 11100 }, { "epoch": 0.7674386450051849, "grad_norm": 0.6954474342571166, "learning_rate": 7.062993403473026e-06, "loss": 0.4119, "step": 11101 }, { "epoch": 0.7675077773937089, "grad_norm": 0.7699161010942326, "learning_rate": 7.062483498645234e-06, "loss": 0.3608, "step": 11102 }, { "epoch": 0.767576909782233, "grad_norm": 0.6430913525752293, "learning_rate": 7.0619735679678925e-06, "loss": 0.2952, "step": 11103 }, { "epoch": 0.767646042170757, "grad_norm": 0.763628438286239, "learning_rate": 7.061463611447391e-06, "loss": 0.3534, "step": 11104 }, { "epoch": 0.7677151745592811, "grad_norm": 0.7195821222080129, "learning_rate": 7.060953629090123e-06, "loss": 0.2993, "step": 11105 }, { "epoch": 0.7677843069478051, "grad_norm": 0.6094632790965541, "learning_rate": 7.0604436209024785e-06, "loss": 0.3417, "step": 11106 }, { "epoch": 0.7678534393363291, "grad_norm": 0.681463872307587, "learning_rate": 7.05993358689085e-06, "loss": 0.3298, "step": 11107 }, { "epoch": 0.7679225717248531, "grad_norm": 0.6815309665530502, "learning_rate": 7.059423527061631e-06, "loss": 0.3936, "step": 11108 }, { "epoch": 0.7679917041133771, "grad_norm": 1.1692464925208805, "learning_rate": 7.058913441421213e-06, "loss": 0.3077, "step": 11109 }, { "epoch": 0.7680608365019012, "grad_norm": 0.6619534199057899, "learning_rate": 7.058403329975989e-06, "loss": 0.2529, "step": 11110 }, { "epoch": 0.7681299688904252, "grad_norm": 0.6394279782936673, "learning_rate": 7.057893192732353e-06, "loss": 0.3398, "step": 11111 }, { "epoch": 0.7681991012789492, "grad_norm": 0.6316611568657732, "learning_rate": 7.057383029696697e-06, "loss": 0.3822, "step": 11112 }, { "epoch": 0.7682682336674732, "grad_norm": 0.6998089282233163, "learning_rate": 7.056872840875417e-06, "loss": 0.277, "step": 11113 }, { "epoch": 0.7683373660559972, "grad_norm": 0.7202881416085242, "learning_rate": 7.0563626262749076e-06, "loss": 0.2729, "step": 11114 }, { "epoch": 0.7684064984445212, "grad_norm": 0.6983668879061332, "learning_rate": 7.0558523859015624e-06, "loss": 0.4039, "step": 11115 }, { "epoch": 0.7684756308330453, "grad_norm": 0.7779299052455148, "learning_rate": 7.055342119761774e-06, "loss": 0.3651, "step": 11116 }, { "epoch": 0.7685447632215693, "grad_norm": 0.6514045447103984, "learning_rate": 7.054831827861943e-06, "loss": 0.3741, "step": 11117 }, { "epoch": 0.7686138956100933, "grad_norm": 0.6573823129231644, "learning_rate": 7.0543215102084614e-06, "loss": 0.3253, "step": 11118 }, { "epoch": 0.7686830279986173, "grad_norm": 0.7065343833679039, "learning_rate": 7.053811166807725e-06, "loss": 0.4154, "step": 11119 }, { "epoch": 0.7687521603871413, "grad_norm": 0.7419034838158043, "learning_rate": 7.0533007976661305e-06, "loss": 0.2988, "step": 11120 }, { "epoch": 0.7688212927756654, "grad_norm": 0.8528092981548059, "learning_rate": 7.052790402790075e-06, "loss": 0.267, "step": 11121 }, { "epoch": 0.7688904251641894, "grad_norm": 0.6966028825299084, "learning_rate": 7.052279982185957e-06, "loss": 0.4051, "step": 11122 }, { "epoch": 0.7689595575527134, "grad_norm": 0.6858949612720003, "learning_rate": 7.051769535860169e-06, "loss": 0.409, "step": 11123 }, { "epoch": 0.7690286899412375, "grad_norm": 0.6063821829502759, "learning_rate": 7.051259063819112e-06, "loss": 0.2711, "step": 11124 }, { "epoch": 0.7690978223297615, "grad_norm": 0.7491529136940545, "learning_rate": 7.050748566069185e-06, "loss": 0.3869, "step": 11125 }, { "epoch": 0.7691669547182856, "grad_norm": 0.7422106696267692, "learning_rate": 7.0502380426167836e-06, "loss": 0.4291, "step": 11126 }, { "epoch": 0.7692360871068096, "grad_norm": 0.6742277243829472, "learning_rate": 7.049727493468307e-06, "loss": 0.3002, "step": 11127 }, { "epoch": 0.7693052194953336, "grad_norm": 0.7113302735131217, "learning_rate": 7.049216918630154e-06, "loss": 0.2881, "step": 11128 }, { "epoch": 0.7693743518838576, "grad_norm": 0.7063099209081344, "learning_rate": 7.0487063181087245e-06, "loss": 0.3689, "step": 11129 }, { "epoch": 0.7694434842723816, "grad_norm": 1.115896659726928, "learning_rate": 7.048195691910417e-06, "loss": 0.349, "step": 11130 }, { "epoch": 0.7695126166609056, "grad_norm": 0.6722227446404102, "learning_rate": 7.047685040041632e-06, "loss": 0.3113, "step": 11131 }, { "epoch": 0.7695817490494297, "grad_norm": 0.6867646446995992, "learning_rate": 7.047174362508769e-06, "loss": 0.3325, "step": 11132 }, { "epoch": 0.7696508814379537, "grad_norm": 0.8276746397618256, "learning_rate": 7.046663659318229e-06, "loss": 0.4337, "step": 11133 }, { "epoch": 0.7697200138264777, "grad_norm": 0.6251088671673557, "learning_rate": 7.046152930476411e-06, "loss": 0.2409, "step": 11134 }, { "epoch": 0.7697891462150017, "grad_norm": 0.6595676914377083, "learning_rate": 7.045642175989718e-06, "loss": 0.3256, "step": 11135 }, { "epoch": 0.7698582786035257, "grad_norm": 0.689422544778825, "learning_rate": 7.045131395864553e-06, "loss": 0.3305, "step": 11136 }, { "epoch": 0.7699274109920498, "grad_norm": 0.6958881311423948, "learning_rate": 7.044620590107314e-06, "loss": 0.3486, "step": 11137 }, { "epoch": 0.7699965433805738, "grad_norm": 0.8164633534554703, "learning_rate": 7.044109758724405e-06, "loss": 0.2843, "step": 11138 }, { "epoch": 0.7700656757690978, "grad_norm": 0.6287479098264026, "learning_rate": 7.0435989017222285e-06, "loss": 0.262, "step": 11139 }, { "epoch": 0.7701348081576218, "grad_norm": 1.0220412973101851, "learning_rate": 7.043088019107186e-06, "loss": 0.412, "step": 11140 }, { "epoch": 0.7702039405461458, "grad_norm": 0.722918486400006, "learning_rate": 7.042577110885681e-06, "loss": 0.2989, "step": 11141 }, { "epoch": 0.7702730729346698, "grad_norm": 0.792133711916955, "learning_rate": 7.042066177064117e-06, "loss": 0.306, "step": 11142 }, { "epoch": 0.7703422053231939, "grad_norm": 0.6867593046118606, "learning_rate": 7.0415552176489e-06, "loss": 0.3012, "step": 11143 }, { "epoch": 0.770411337711718, "grad_norm": 0.7252417722284885, "learning_rate": 7.041044232646429e-06, "loss": 0.4433, "step": 11144 }, { "epoch": 0.770480470100242, "grad_norm": 0.6224258445481509, "learning_rate": 7.040533222063113e-06, "loss": 0.2655, "step": 11145 }, { "epoch": 0.770549602488766, "grad_norm": 0.7056013365103044, "learning_rate": 7.040022185905353e-06, "loss": 0.2923, "step": 11146 }, { "epoch": 0.77061873487729, "grad_norm": 0.6821260010478761, "learning_rate": 7.039511124179557e-06, "loss": 0.4191, "step": 11147 }, { "epoch": 0.7706878672658141, "grad_norm": 0.6617332977686535, "learning_rate": 7.039000036892126e-06, "loss": 0.3354, "step": 11148 }, { "epoch": 0.7707569996543381, "grad_norm": 0.6870284049551959, "learning_rate": 7.038488924049471e-06, "loss": 0.3226, "step": 11149 }, { "epoch": 0.7708261320428621, "grad_norm": 0.7173863936483567, "learning_rate": 7.037977785657994e-06, "loss": 0.3723, "step": 11150 }, { "epoch": 0.7708952644313861, "grad_norm": 0.7013208535367917, "learning_rate": 7.037466621724101e-06, "loss": 0.4073, "step": 11151 }, { "epoch": 0.7709643968199101, "grad_norm": 0.6549129899712633, "learning_rate": 7.036955432254202e-06, "loss": 0.308, "step": 11152 }, { "epoch": 0.7710335292084342, "grad_norm": 0.6585314906929295, "learning_rate": 7.036444217254702e-06, "loss": 0.2997, "step": 11153 }, { "epoch": 0.7711026615969582, "grad_norm": 0.65986336963285, "learning_rate": 7.035932976732008e-06, "loss": 0.3909, "step": 11154 }, { "epoch": 0.7711717939854822, "grad_norm": 0.7899262825115718, "learning_rate": 7.035421710692525e-06, "loss": 0.3439, "step": 11155 }, { "epoch": 0.7712409263740062, "grad_norm": 0.6849941246143549, "learning_rate": 7.034910419142666e-06, "loss": 0.3064, "step": 11156 }, { "epoch": 0.7713100587625302, "grad_norm": 0.7285347579945824, "learning_rate": 7.0343991020888346e-06, "loss": 0.3028, "step": 11157 }, { "epoch": 0.7713791911510542, "grad_norm": 0.6588297621061278, "learning_rate": 7.033887759537441e-06, "loss": 0.3996, "step": 11158 }, { "epoch": 0.7714483235395783, "grad_norm": 0.6631871847852725, "learning_rate": 7.033376391494895e-06, "loss": 0.3535, "step": 11159 }, { "epoch": 0.7715174559281023, "grad_norm": 0.6811249385120653, "learning_rate": 7.032864997967603e-06, "loss": 0.328, "step": 11160 }, { "epoch": 0.7715865883166263, "grad_norm": 0.6414945549047253, "learning_rate": 7.032353578961979e-06, "loss": 0.3407, "step": 11161 }, { "epoch": 0.7716557207051503, "grad_norm": 0.6181966324651458, "learning_rate": 7.0318421344844265e-06, "loss": 0.3833, "step": 11162 }, { "epoch": 0.7717248530936743, "grad_norm": 0.66959589433631, "learning_rate": 7.03133066454136e-06, "loss": 0.2748, "step": 11163 }, { "epoch": 0.7717939854821985, "grad_norm": 0.6527583660375579, "learning_rate": 7.0308191691391904e-06, "loss": 0.3071, "step": 11164 }, { "epoch": 0.7718631178707225, "grad_norm": 0.7168676290639149, "learning_rate": 7.0303076482843245e-06, "loss": 0.4226, "step": 11165 }, { "epoch": 0.7719322502592465, "grad_norm": 0.6464672847353399, "learning_rate": 7.029796101983176e-06, "loss": 0.3193, "step": 11166 }, { "epoch": 0.7720013826477705, "grad_norm": 0.6279534856491557, "learning_rate": 7.029284530242157e-06, "loss": 0.321, "step": 11167 }, { "epoch": 0.7720705150362945, "grad_norm": 0.6709058923751682, "learning_rate": 7.0287729330676766e-06, "loss": 0.3415, "step": 11168 }, { "epoch": 0.7721396474248186, "grad_norm": 0.7242524288185003, "learning_rate": 7.028261310466148e-06, "loss": 0.3967, "step": 11169 }, { "epoch": 0.7722087798133426, "grad_norm": 0.6684356180238806, "learning_rate": 7.027749662443984e-06, "loss": 0.2853, "step": 11170 }, { "epoch": 0.7722779122018666, "grad_norm": 0.6568461256938954, "learning_rate": 7.027237989007596e-06, "loss": 0.2912, "step": 11171 }, { "epoch": 0.7723470445903906, "grad_norm": 0.7102991754316097, "learning_rate": 7.026726290163398e-06, "loss": 0.4428, "step": 11172 }, { "epoch": 0.7724161769789146, "grad_norm": 0.6734476704077055, "learning_rate": 7.0262145659178025e-06, "loss": 0.363, "step": 11173 }, { "epoch": 0.7724853093674386, "grad_norm": 0.7503382545465364, "learning_rate": 7.025702816277223e-06, "loss": 0.2981, "step": 11174 }, { "epoch": 0.7725544417559627, "grad_norm": 1.5015292000551381, "learning_rate": 7.025191041248075e-06, "loss": 0.3734, "step": 11175 }, { "epoch": 0.7726235741444867, "grad_norm": 0.6975674412383973, "learning_rate": 7.02467924083677e-06, "loss": 0.4077, "step": 11176 }, { "epoch": 0.7726927065330107, "grad_norm": 0.7090348077227739, "learning_rate": 7.024167415049726e-06, "loss": 0.3012, "step": 11177 }, { "epoch": 0.7727618389215347, "grad_norm": 0.63092251900169, "learning_rate": 7.023655563893354e-06, "loss": 0.3043, "step": 11178 }, { "epoch": 0.7728309713100587, "grad_norm": 0.6946885717317184, "learning_rate": 7.023143687374071e-06, "loss": 0.3363, "step": 11179 }, { "epoch": 0.7729001036985828, "grad_norm": 0.7035625821717252, "learning_rate": 7.0226317854982924e-06, "loss": 0.3934, "step": 11180 }, { "epoch": 0.7729692360871068, "grad_norm": 0.72895158924918, "learning_rate": 7.022119858272434e-06, "loss": 0.3008, "step": 11181 }, { "epoch": 0.7730383684756308, "grad_norm": 0.6534088281915994, "learning_rate": 7.021607905702912e-06, "loss": 0.2168, "step": 11182 }, { "epoch": 0.7731075008641548, "grad_norm": 0.6191516694502721, "learning_rate": 7.021095927796142e-06, "loss": 0.3476, "step": 11183 }, { "epoch": 0.7731766332526789, "grad_norm": 0.6787984333499505, "learning_rate": 7.020583924558543e-06, "loss": 0.329, "step": 11184 }, { "epoch": 0.773245765641203, "grad_norm": 0.6622201809313837, "learning_rate": 7.0200718959965285e-06, "loss": 0.3198, "step": 11185 }, { "epoch": 0.773314898029727, "grad_norm": 0.7341525352856837, "learning_rate": 7.019559842116519e-06, "loss": 0.3158, "step": 11186 }, { "epoch": 0.773384030418251, "grad_norm": 0.9028912366702294, "learning_rate": 7.01904776292493e-06, "loss": 0.4401, "step": 11187 }, { "epoch": 0.773453162806775, "grad_norm": 0.6035902815124579, "learning_rate": 7.018535658428181e-06, "loss": 0.2574, "step": 11188 }, { "epoch": 0.773522295195299, "grad_norm": 0.6488892334743532, "learning_rate": 7.018023528632691e-06, "loss": 0.3599, "step": 11189 }, { "epoch": 0.773591427583823, "grad_norm": 0.7995108706382078, "learning_rate": 7.017511373544877e-06, "loss": 0.3957, "step": 11190 }, { "epoch": 0.7736605599723471, "grad_norm": 0.6741136344649697, "learning_rate": 7.0169991931711555e-06, "loss": 0.3493, "step": 11191 }, { "epoch": 0.7737296923608711, "grad_norm": 0.6566405563026104, "learning_rate": 7.0164869875179505e-06, "loss": 0.2613, "step": 11192 }, { "epoch": 0.7737988247493951, "grad_norm": 0.6742782710942752, "learning_rate": 7.01597475659168e-06, "loss": 0.3903, "step": 11193 }, { "epoch": 0.7738679571379191, "grad_norm": 0.6415791881681248, "learning_rate": 7.015462500398763e-06, "loss": 0.3645, "step": 11194 }, { "epoch": 0.7739370895264431, "grad_norm": 0.643275349999541, "learning_rate": 7.014950218945621e-06, "loss": 0.2828, "step": 11195 }, { "epoch": 0.7740062219149672, "grad_norm": 0.658055386316051, "learning_rate": 7.014437912238674e-06, "loss": 0.2948, "step": 11196 }, { "epoch": 0.7740753543034912, "grad_norm": 0.7386119007846895, "learning_rate": 7.013925580284342e-06, "loss": 0.381, "step": 11197 }, { "epoch": 0.7741444866920152, "grad_norm": 0.7413550911042587, "learning_rate": 7.0134132230890476e-06, "loss": 0.3489, "step": 11198 }, { "epoch": 0.7742136190805392, "grad_norm": 0.6357214040136017, "learning_rate": 7.01290084065921e-06, "loss": 0.308, "step": 11199 }, { "epoch": 0.7742827514690632, "grad_norm": 0.6716600978217641, "learning_rate": 7.0123884330012535e-06, "loss": 0.3334, "step": 11200 }, { "epoch": 0.7743518838575872, "grad_norm": 0.6296958189685594, "learning_rate": 7.011876000121599e-06, "loss": 0.3775, "step": 11201 }, { "epoch": 0.7744210162461113, "grad_norm": 0.6819023578007523, "learning_rate": 7.01136354202667e-06, "loss": 0.3329, "step": 11202 }, { "epoch": 0.7744901486346353, "grad_norm": 0.6589136081136413, "learning_rate": 7.010851058722889e-06, "loss": 0.2704, "step": 11203 }, { "epoch": 0.7745592810231593, "grad_norm": 0.774630131458299, "learning_rate": 7.010338550216678e-06, "loss": 0.3904, "step": 11204 }, { "epoch": 0.7746284134116834, "grad_norm": 0.6823603491124911, "learning_rate": 7.009826016514458e-06, "loss": 0.3175, "step": 11205 }, { "epoch": 0.7746975458002074, "grad_norm": 0.6549489085215932, "learning_rate": 7.009313457622658e-06, "loss": 0.3822, "step": 11206 }, { "epoch": 0.7747666781887315, "grad_norm": 0.7109757193206794, "learning_rate": 7.008800873547701e-06, "loss": 0.3206, "step": 11207 }, { "epoch": 0.7748358105772555, "grad_norm": 0.6738031965670113, "learning_rate": 7.0082882642960085e-06, "loss": 0.3841, "step": 11208 }, { "epoch": 0.7749049429657795, "grad_norm": 0.7269488384810385, "learning_rate": 7.0077756298740055e-06, "loss": 0.3166, "step": 11209 }, { "epoch": 0.7749740753543035, "grad_norm": 0.6933244932648869, "learning_rate": 7.00726297028812e-06, "loss": 0.3108, "step": 11210 }, { "epoch": 0.7750432077428275, "grad_norm": 0.6447544300459803, "learning_rate": 7.006750285544775e-06, "loss": 0.3117, "step": 11211 }, { "epoch": 0.7751123401313516, "grad_norm": 0.6596180509265812, "learning_rate": 7.006237575650394e-06, "loss": 0.4169, "step": 11212 }, { "epoch": 0.7751814725198756, "grad_norm": 0.6953867210165933, "learning_rate": 7.005724840611408e-06, "loss": 0.3017, "step": 11213 }, { "epoch": 0.7752506049083996, "grad_norm": 0.7418902453729626, "learning_rate": 7.005212080434241e-06, "loss": 0.3384, "step": 11214 }, { "epoch": 0.7753197372969236, "grad_norm": 0.7347859776262905, "learning_rate": 7.004699295125316e-06, "loss": 0.3973, "step": 11215 }, { "epoch": 0.7753888696854476, "grad_norm": 0.6388498900515548, "learning_rate": 7.004186484691065e-06, "loss": 0.2822, "step": 11216 }, { "epoch": 0.7754580020739716, "grad_norm": 0.6794943458920896, "learning_rate": 7.003673649137912e-06, "loss": 0.2676, "step": 11217 }, { "epoch": 0.7755271344624957, "grad_norm": 0.7699491912725325, "learning_rate": 7.003160788472287e-06, "loss": 0.3665, "step": 11218 }, { "epoch": 0.7755962668510197, "grad_norm": 0.6943544206118846, "learning_rate": 7.002647902700613e-06, "loss": 0.3583, "step": 11219 }, { "epoch": 0.7756653992395437, "grad_norm": 0.7105468311432216, "learning_rate": 7.002134991829325e-06, "loss": 0.2566, "step": 11220 }, { "epoch": 0.7757345316280677, "grad_norm": 0.7305623950083583, "learning_rate": 7.001622055864846e-06, "loss": 0.3223, "step": 11221 }, { "epoch": 0.7758036640165917, "grad_norm": 0.6506153884781452, "learning_rate": 7.001109094813607e-06, "loss": 0.3783, "step": 11222 }, { "epoch": 0.7758727964051157, "grad_norm": 0.9820496642955219, "learning_rate": 7.000596108682036e-06, "loss": 0.4204, "step": 11223 }, { "epoch": 0.7759419287936398, "grad_norm": 0.6294057666746853, "learning_rate": 7.0000830974765646e-06, "loss": 0.2748, "step": 11224 }, { "epoch": 0.7760110611821639, "grad_norm": 0.7610658191211898, "learning_rate": 6.9995700612036195e-06, "loss": 0.3644, "step": 11225 }, { "epoch": 0.7760801935706879, "grad_norm": 0.6949736179119922, "learning_rate": 6.9990569998696335e-06, "loss": 0.38, "step": 11226 }, { "epoch": 0.7761493259592119, "grad_norm": 0.673294890414384, "learning_rate": 6.998543913481034e-06, "loss": 0.3207, "step": 11227 }, { "epoch": 0.776218458347736, "grad_norm": 0.7012140951325149, "learning_rate": 6.998030802044254e-06, "loss": 0.3047, "step": 11228 }, { "epoch": 0.77628759073626, "grad_norm": 0.7668349668125884, "learning_rate": 6.997517665565723e-06, "loss": 0.4195, "step": 11229 }, { "epoch": 0.776356723124784, "grad_norm": 0.7472760659562654, "learning_rate": 6.997004504051873e-06, "loss": 0.2897, "step": 11230 }, { "epoch": 0.776425855513308, "grad_norm": 0.6393379927034576, "learning_rate": 6.996491317509135e-06, "loss": 0.275, "step": 11231 }, { "epoch": 0.776494987901832, "grad_norm": 0.849505799430763, "learning_rate": 6.995978105943942e-06, "loss": 0.323, "step": 11232 }, { "epoch": 0.776564120290356, "grad_norm": 0.7048611193477122, "learning_rate": 6.995464869362725e-06, "loss": 0.4224, "step": 11233 }, { "epoch": 0.7766332526788801, "grad_norm": 0.6496046521825428, "learning_rate": 6.9949516077719175e-06, "loss": 0.2882, "step": 11234 }, { "epoch": 0.7767023850674041, "grad_norm": 0.6322467330683799, "learning_rate": 6.994438321177951e-06, "loss": 0.256, "step": 11235 }, { "epoch": 0.7767715174559281, "grad_norm": 0.7905266286433202, "learning_rate": 6.99392500958726e-06, "loss": 0.3416, "step": 11236 }, { "epoch": 0.7768406498444521, "grad_norm": 0.714247466594727, "learning_rate": 6.993411673006276e-06, "loss": 0.3809, "step": 11237 }, { "epoch": 0.7769097822329761, "grad_norm": 0.6669709635772556, "learning_rate": 6.992898311441436e-06, "loss": 0.3131, "step": 11238 }, { "epoch": 0.7769789146215001, "grad_norm": 0.6294805349287447, "learning_rate": 6.992384924899171e-06, "loss": 0.2703, "step": 11239 }, { "epoch": 0.7770480470100242, "grad_norm": 0.7721213094333353, "learning_rate": 6.991871513385916e-06, "loss": 0.4192, "step": 11240 }, { "epoch": 0.7771171793985482, "grad_norm": 0.7256296912507381, "learning_rate": 6.991358076908107e-06, "loss": 0.3474, "step": 11241 }, { "epoch": 0.7771863117870722, "grad_norm": 0.5749948972210652, "learning_rate": 6.990844615472179e-06, "loss": 0.2602, "step": 11242 }, { "epoch": 0.7772554441755962, "grad_norm": 0.7013379052899361, "learning_rate": 6.990331129084566e-06, "loss": 0.3854, "step": 11243 }, { "epoch": 0.7773245765641202, "grad_norm": 0.6732452584585361, "learning_rate": 6.989817617751703e-06, "loss": 0.3913, "step": 11244 }, { "epoch": 0.7773937089526444, "grad_norm": 0.6398436930251025, "learning_rate": 6.989304081480028e-06, "loss": 0.307, "step": 11245 }, { "epoch": 0.7774628413411684, "grad_norm": 0.6470113618910657, "learning_rate": 6.988790520275976e-06, "loss": 0.2695, "step": 11246 }, { "epoch": 0.7775319737296924, "grad_norm": 1.8080657876859296, "learning_rate": 6.9882769341459845e-06, "loss": 0.4273, "step": 11247 }, { "epoch": 0.7776011061182164, "grad_norm": 0.6159568063049109, "learning_rate": 6.98776332309649e-06, "loss": 0.355, "step": 11248 }, { "epoch": 0.7776702385067404, "grad_norm": 1.0128988619259147, "learning_rate": 6.9872496871339304e-06, "loss": 0.2987, "step": 11249 }, { "epoch": 0.7777393708952645, "grad_norm": 0.6961785737434416, "learning_rate": 6.986736026264742e-06, "loss": 0.3292, "step": 11250 }, { "epoch": 0.7778085032837885, "grad_norm": 0.6452263922052021, "learning_rate": 6.986222340495361e-06, "loss": 0.4021, "step": 11251 }, { "epoch": 0.7778776356723125, "grad_norm": 0.763795121432892, "learning_rate": 6.98570862983223e-06, "loss": 0.3265, "step": 11252 }, { "epoch": 0.7779467680608365, "grad_norm": 0.658662813974223, "learning_rate": 6.985194894281784e-06, "loss": 0.2727, "step": 11253 }, { "epoch": 0.7780159004493605, "grad_norm": 0.7150602652721368, "learning_rate": 6.984681133850462e-06, "loss": 0.3925, "step": 11254 }, { "epoch": 0.7780850328378845, "grad_norm": 0.7543641106919109, "learning_rate": 6.984167348544704e-06, "loss": 0.3651, "step": 11255 }, { "epoch": 0.7781541652264086, "grad_norm": 0.5966249487884933, "learning_rate": 6.98365353837095e-06, "loss": 0.2673, "step": 11256 }, { "epoch": 0.7782232976149326, "grad_norm": 0.8641835786453836, "learning_rate": 6.98313970333564e-06, "loss": 0.3747, "step": 11257 }, { "epoch": 0.7782924300034566, "grad_norm": 0.6716781375610825, "learning_rate": 6.9826258434452095e-06, "loss": 0.414, "step": 11258 }, { "epoch": 0.7783615623919806, "grad_norm": 0.7148318348915464, "learning_rate": 6.9821119587061045e-06, "loss": 0.2869, "step": 11259 }, { "epoch": 0.7784306947805046, "grad_norm": 0.6483608044059815, "learning_rate": 6.981598049124764e-06, "loss": 0.3475, "step": 11260 }, { "epoch": 0.7784998271690287, "grad_norm": 0.6733479933944386, "learning_rate": 6.981084114707627e-06, "loss": 0.295, "step": 11261 }, { "epoch": 0.7785689595575527, "grad_norm": 0.6797092607449425, "learning_rate": 6.980570155461135e-06, "loss": 0.3855, "step": 11262 }, { "epoch": 0.7786380919460767, "grad_norm": 0.6783432085530414, "learning_rate": 6.980056171391732e-06, "loss": 0.2483, "step": 11263 }, { "epoch": 0.7787072243346007, "grad_norm": 0.6801640470839511, "learning_rate": 6.979542162505859e-06, "loss": 0.2372, "step": 11264 }, { "epoch": 0.7787763567231247, "grad_norm": 0.6925397362093583, "learning_rate": 6.9790281288099574e-06, "loss": 0.345, "step": 11265 }, { "epoch": 0.7788454891116489, "grad_norm": 0.6705056525971256, "learning_rate": 6.9785140703104695e-06, "loss": 0.3637, "step": 11266 }, { "epoch": 0.7789146215001729, "grad_norm": 0.7143168448672752, "learning_rate": 6.977999987013838e-06, "loss": 0.2682, "step": 11267 }, { "epoch": 0.7789837538886969, "grad_norm": 0.6348149993084726, "learning_rate": 6.977485878926508e-06, "loss": 0.3274, "step": 11268 }, { "epoch": 0.7790528862772209, "grad_norm": 0.860164082219953, "learning_rate": 6.976971746054919e-06, "loss": 0.4419, "step": 11269 }, { "epoch": 0.7791220186657449, "grad_norm": 0.7208565359093634, "learning_rate": 6.9764575884055195e-06, "loss": 0.3278, "step": 11270 }, { "epoch": 0.779191151054269, "grad_norm": 0.7045546461536669, "learning_rate": 6.9759434059847506e-06, "loss": 0.3108, "step": 11271 }, { "epoch": 0.779260283442793, "grad_norm": 0.7149767249933758, "learning_rate": 6.975429198799057e-06, "loss": 0.4673, "step": 11272 }, { "epoch": 0.779329415831317, "grad_norm": 0.6872199554121701, "learning_rate": 6.974914966854884e-06, "loss": 0.3983, "step": 11273 }, { "epoch": 0.779398548219841, "grad_norm": 0.6096941222769184, "learning_rate": 6.974400710158677e-06, "loss": 0.2973, "step": 11274 }, { "epoch": 0.779467680608365, "grad_norm": 0.6961214560488221, "learning_rate": 6.973886428716881e-06, "loss": 0.3672, "step": 11275 }, { "epoch": 0.779536812996889, "grad_norm": 0.6653274016950593, "learning_rate": 6.973372122535939e-06, "loss": 0.4167, "step": 11276 }, { "epoch": 0.7796059453854131, "grad_norm": 0.6442412499410297, "learning_rate": 6.9728577916223e-06, "loss": 0.3103, "step": 11277 }, { "epoch": 0.7796750777739371, "grad_norm": 0.7099658286247318, "learning_rate": 6.97234343598241e-06, "loss": 0.2563, "step": 11278 }, { "epoch": 0.7797442101624611, "grad_norm": 0.7077647934783668, "learning_rate": 6.971829055622714e-06, "loss": 0.3356, "step": 11279 }, { "epoch": 0.7798133425509851, "grad_norm": 0.6472898512920835, "learning_rate": 6.971314650549661e-06, "loss": 0.3835, "step": 11280 }, { "epoch": 0.7798824749395091, "grad_norm": 0.5911339492947046, "learning_rate": 6.970800220769696e-06, "loss": 0.299, "step": 11281 }, { "epoch": 0.7799516073280331, "grad_norm": 0.6450273347647432, "learning_rate": 6.970285766289266e-06, "loss": 0.2551, "step": 11282 }, { "epoch": 0.7800207397165572, "grad_norm": 0.6439518509483579, "learning_rate": 6.9697712871148215e-06, "loss": 0.3463, "step": 11283 }, { "epoch": 0.7800898721050812, "grad_norm": 0.6433116420934704, "learning_rate": 6.969256783252809e-06, "loss": 0.2996, "step": 11284 }, { "epoch": 0.7801590044936052, "grad_norm": 0.877319605333404, "learning_rate": 6.968742254709677e-06, "loss": 0.3419, "step": 11285 }, { "epoch": 0.7802281368821293, "grad_norm": 0.6596478769905781, "learning_rate": 6.968227701491873e-06, "loss": 0.3468, "step": 11286 }, { "epoch": 0.7802972692706533, "grad_norm": 0.7003825907894536, "learning_rate": 6.967713123605849e-06, "loss": 0.404, "step": 11287 }, { "epoch": 0.7803664016591774, "grad_norm": 0.5798852240511839, "learning_rate": 6.967198521058051e-06, "loss": 0.2551, "step": 11288 }, { "epoch": 0.7804355340477014, "grad_norm": 0.7311864358444865, "learning_rate": 6.966683893854932e-06, "loss": 0.3146, "step": 11289 }, { "epoch": 0.7805046664362254, "grad_norm": 0.6846782349996559, "learning_rate": 6.966169242002938e-06, "loss": 0.3882, "step": 11290 }, { "epoch": 0.7805737988247494, "grad_norm": 0.7179348238952468, "learning_rate": 6.9656545655085225e-06, "loss": 0.3354, "step": 11291 }, { "epoch": 0.7806429312132734, "grad_norm": 0.5617099768848236, "learning_rate": 6.965139864378135e-06, "loss": 0.2408, "step": 11292 }, { "epoch": 0.7807120636017975, "grad_norm": 0.6988466183615404, "learning_rate": 6.964625138618226e-06, "loss": 0.3752, "step": 11293 }, { "epoch": 0.7807811959903215, "grad_norm": 0.6467391795565952, "learning_rate": 6.964110388235248e-06, "loss": 0.4004, "step": 11294 }, { "epoch": 0.7808503283788455, "grad_norm": 0.6668521653399961, "learning_rate": 6.963595613235649e-06, "loss": 0.3341, "step": 11295 }, { "epoch": 0.7809194607673695, "grad_norm": 0.6355539653369728, "learning_rate": 6.963080813625885e-06, "loss": 0.2709, "step": 11296 }, { "epoch": 0.7809885931558935, "grad_norm": 0.6609901481123911, "learning_rate": 6.962565989412405e-06, "loss": 0.3234, "step": 11297 }, { "epoch": 0.7810577255444175, "grad_norm": 0.6877462115189635, "learning_rate": 6.9620511406016645e-06, "loss": 0.3912, "step": 11298 }, { "epoch": 0.7811268579329416, "grad_norm": 0.7123442341483018, "learning_rate": 6.961536267200113e-06, "loss": 0.2884, "step": 11299 }, { "epoch": 0.7811959903214656, "grad_norm": 0.9782432340415025, "learning_rate": 6.961021369214205e-06, "loss": 0.3554, "step": 11300 }, { "epoch": 0.7812651227099896, "grad_norm": 0.671218866438583, "learning_rate": 6.960506446650395e-06, "loss": 0.3702, "step": 11301 }, { "epoch": 0.7813342550985136, "grad_norm": 0.6551555350981564, "learning_rate": 6.959991499515134e-06, "loss": 0.2923, "step": 11302 }, { "epoch": 0.7814033874870376, "grad_norm": 0.7112447092989378, "learning_rate": 6.959476527814878e-06, "loss": 0.2997, "step": 11303 }, { "epoch": 0.7814725198755617, "grad_norm": 0.7258912169225056, "learning_rate": 6.95896153155608e-06, "loss": 0.43, "step": 11304 }, { "epoch": 0.7815416522640857, "grad_norm": 0.6751024809968924, "learning_rate": 6.958446510745195e-06, "loss": 0.3515, "step": 11305 }, { "epoch": 0.7816107846526098, "grad_norm": 0.8793295926572652, "learning_rate": 6.957931465388679e-06, "loss": 0.2726, "step": 11306 }, { "epoch": 0.7816799170411338, "grad_norm": 0.7292265468704984, "learning_rate": 6.957416395492985e-06, "loss": 0.3437, "step": 11307 }, { "epoch": 0.7817490494296578, "grad_norm": 0.6650943167239928, "learning_rate": 6.95690130106457e-06, "loss": 0.3527, "step": 11308 }, { "epoch": 0.7818181818181819, "grad_norm": 0.6669401590683773, "learning_rate": 6.95638618210989e-06, "loss": 0.3088, "step": 11309 }, { "epoch": 0.7818873142067059, "grad_norm": 0.7893464955740578, "learning_rate": 6.9558710386354e-06, "loss": 0.298, "step": 11310 }, { "epoch": 0.7819564465952299, "grad_norm": 0.9987895386648082, "learning_rate": 6.955355870647556e-06, "loss": 0.3375, "step": 11311 }, { "epoch": 0.7820255789837539, "grad_norm": 0.6822877655684837, "learning_rate": 6.954840678152818e-06, "loss": 0.3248, "step": 11312 }, { "epoch": 0.7820947113722779, "grad_norm": 0.6969786093930223, "learning_rate": 6.954325461157639e-06, "loss": 0.3076, "step": 11313 }, { "epoch": 0.782163843760802, "grad_norm": 0.6213682418190262, "learning_rate": 6.953810219668479e-06, "loss": 0.3217, "step": 11314 }, { "epoch": 0.782232976149326, "grad_norm": 0.6583187157385055, "learning_rate": 6.953294953691794e-06, "loss": 0.4219, "step": 11315 }, { "epoch": 0.78230210853785, "grad_norm": 0.9101563368022629, "learning_rate": 6.952779663234043e-06, "loss": 0.4423, "step": 11316 }, { "epoch": 0.782371240926374, "grad_norm": 0.6256503817387418, "learning_rate": 6.952264348301683e-06, "loss": 0.2954, "step": 11317 }, { "epoch": 0.782440373314898, "grad_norm": 0.8297049494863815, "learning_rate": 6.951749008901173e-06, "loss": 0.3445, "step": 11318 }, { "epoch": 0.782509505703422, "grad_norm": 0.6818455404433597, "learning_rate": 6.951233645038972e-06, "loss": 0.4021, "step": 11319 }, { "epoch": 0.782578638091946, "grad_norm": 0.7668833652306145, "learning_rate": 6.950718256721543e-06, "loss": 0.2902, "step": 11320 }, { "epoch": 0.7826477704804701, "grad_norm": 0.660634393807732, "learning_rate": 6.950202843955338e-06, "loss": 0.3114, "step": 11321 }, { "epoch": 0.7827169028689941, "grad_norm": 0.6715547778151371, "learning_rate": 6.949687406746821e-06, "loss": 0.377, "step": 11322 }, { "epoch": 0.7827860352575181, "grad_norm": 0.6296329655223505, "learning_rate": 6.949171945102454e-06, "loss": 0.3052, "step": 11323 }, { "epoch": 0.7828551676460421, "grad_norm": 0.6249778630566882, "learning_rate": 6.948656459028694e-06, "loss": 0.2857, "step": 11324 }, { "epoch": 0.7829243000345661, "grad_norm": 2.1114692326298923, "learning_rate": 6.948140948532003e-06, "loss": 0.3211, "step": 11325 }, { "epoch": 0.7829934324230902, "grad_norm": 0.6927531601734949, "learning_rate": 6.9476254136188405e-06, "loss": 0.3909, "step": 11326 }, { "epoch": 0.7830625648116143, "grad_norm": 0.6406312598822614, "learning_rate": 6.947109854295671e-06, "loss": 0.2873, "step": 11327 }, { "epoch": 0.7831316972001383, "grad_norm": 0.65203203862694, "learning_rate": 6.946594270568952e-06, "loss": 0.263, "step": 11328 }, { "epoch": 0.7832008295886623, "grad_norm": 0.697859895556061, "learning_rate": 6.946078662445149e-06, "loss": 0.3424, "step": 11329 }, { "epoch": 0.7832699619771863, "grad_norm": 0.7247235044186507, "learning_rate": 6.945563029930723e-06, "loss": 0.313, "step": 11330 }, { "epoch": 0.7833390943657104, "grad_norm": 0.6831549450367996, "learning_rate": 6.945047373032137e-06, "loss": 0.3395, "step": 11331 }, { "epoch": 0.7834082267542344, "grad_norm": 0.670578183076864, "learning_rate": 6.9445316917558524e-06, "loss": 0.3051, "step": 11332 }, { "epoch": 0.7834773591427584, "grad_norm": 0.6713906408078393, "learning_rate": 6.944015986108332e-06, "loss": 0.4472, "step": 11333 }, { "epoch": 0.7835464915312824, "grad_norm": 0.7702638447193855, "learning_rate": 6.943500256096043e-06, "loss": 0.3042, "step": 11334 }, { "epoch": 0.7836156239198064, "grad_norm": 0.6525318509463656, "learning_rate": 6.942984501725446e-06, "loss": 0.3036, "step": 11335 }, { "epoch": 0.7836847563083305, "grad_norm": 0.9981633673941246, "learning_rate": 6.942468723003004e-06, "loss": 0.3716, "step": 11336 }, { "epoch": 0.7837538886968545, "grad_norm": 0.8470395111672949, "learning_rate": 6.941952919935184e-06, "loss": 0.4285, "step": 11337 }, { "epoch": 0.7838230210853785, "grad_norm": 0.6502500960018969, "learning_rate": 6.941437092528452e-06, "loss": 0.2717, "step": 11338 }, { "epoch": 0.7838921534739025, "grad_norm": 1.0205022002172097, "learning_rate": 6.940921240789268e-06, "loss": 0.3383, "step": 11339 }, { "epoch": 0.7839612858624265, "grad_norm": 0.9790418350502494, "learning_rate": 6.9404053647241e-06, "loss": 0.429, "step": 11340 }, { "epoch": 0.7840304182509505, "grad_norm": 0.6404673182770451, "learning_rate": 6.939889464339415e-06, "loss": 0.2572, "step": 11341 }, { "epoch": 0.7840995506394746, "grad_norm": 0.6136193215673871, "learning_rate": 6.939373539641676e-06, "loss": 0.2803, "step": 11342 }, { "epoch": 0.7841686830279986, "grad_norm": 0.655705605419115, "learning_rate": 6.938857590637351e-06, "loss": 0.3294, "step": 11343 }, { "epoch": 0.7842378154165226, "grad_norm": 0.9293280328882937, "learning_rate": 6.938341617332907e-06, "loss": 0.3943, "step": 11344 }, { "epoch": 0.7843069478050466, "grad_norm": 0.6947539824184168, "learning_rate": 6.93782561973481e-06, "loss": 0.2662, "step": 11345 }, { "epoch": 0.7843760801935706, "grad_norm": 0.7056452724548027, "learning_rate": 6.937309597849525e-06, "loss": 0.2596, "step": 11346 }, { "epoch": 0.7844452125820948, "grad_norm": 0.6326755265178278, "learning_rate": 6.936793551683524e-06, "loss": 0.3488, "step": 11347 }, { "epoch": 0.7845143449706188, "grad_norm": 2.2976843782421605, "learning_rate": 6.936277481243271e-06, "loss": 0.2743, "step": 11348 }, { "epoch": 0.7845834773591428, "grad_norm": 0.7449254428133477, "learning_rate": 6.935761386535236e-06, "loss": 0.2971, "step": 11349 }, { "epoch": 0.7846526097476668, "grad_norm": 0.7020561312552738, "learning_rate": 6.935245267565884e-06, "loss": 0.3797, "step": 11350 }, { "epoch": 0.7847217421361908, "grad_norm": 0.7753292901665972, "learning_rate": 6.934729124341689e-06, "loss": 0.446, "step": 11351 }, { "epoch": 0.7847908745247149, "grad_norm": 0.6554865898726295, "learning_rate": 6.934212956869116e-06, "loss": 0.3375, "step": 11352 }, { "epoch": 0.7848600069132389, "grad_norm": 0.7425184432161508, "learning_rate": 6.9336967651546346e-06, "loss": 0.27, "step": 11353 }, { "epoch": 0.7849291393017629, "grad_norm": 0.7215537475919921, "learning_rate": 6.9331805492047164e-06, "loss": 0.3928, "step": 11354 }, { "epoch": 0.7849982716902869, "grad_norm": 0.7449619255456433, "learning_rate": 6.932664309025831e-06, "loss": 0.4012, "step": 11355 }, { "epoch": 0.7850674040788109, "grad_norm": 0.6573953675696773, "learning_rate": 6.932148044624445e-06, "loss": 0.2666, "step": 11356 }, { "epoch": 0.785136536467335, "grad_norm": 0.6429996370807224, "learning_rate": 6.931631756007032e-06, "loss": 0.3006, "step": 11357 }, { "epoch": 0.785205668855859, "grad_norm": 0.6723091315464111, "learning_rate": 6.931115443180063e-06, "loss": 0.428, "step": 11358 }, { "epoch": 0.785274801244383, "grad_norm": 0.6582199258675051, "learning_rate": 6.930599106150008e-06, "loss": 0.3082, "step": 11359 }, { "epoch": 0.785343933632907, "grad_norm": 0.6446143827157833, "learning_rate": 6.930082744923338e-06, "loss": 0.2528, "step": 11360 }, { "epoch": 0.785413066021431, "grad_norm": 0.8254615767978469, "learning_rate": 6.929566359506524e-06, "loss": 0.3061, "step": 11361 }, { "epoch": 0.785482198409955, "grad_norm": 0.6699815635468434, "learning_rate": 6.929049949906041e-06, "loss": 0.4219, "step": 11362 }, { "epoch": 0.785551330798479, "grad_norm": 0.6287379705395446, "learning_rate": 6.928533516128359e-06, "loss": 0.2914, "step": 11363 }, { "epoch": 0.7856204631870031, "grad_norm": 0.9489170328486487, "learning_rate": 6.92801705817995e-06, "loss": 0.2439, "step": 11364 }, { "epoch": 0.7856895955755271, "grad_norm": 0.8573824144295586, "learning_rate": 6.927500576067289e-06, "loss": 0.4061, "step": 11365 }, { "epoch": 0.7857587279640511, "grad_norm": 0.6885441651204628, "learning_rate": 6.926984069796848e-06, "loss": 0.3114, "step": 11366 }, { "epoch": 0.7858278603525752, "grad_norm": 0.6570057803657825, "learning_rate": 6.9264675393751015e-06, "loss": 0.288, "step": 11367 }, { "epoch": 0.7858969927410993, "grad_norm": 0.6782873634813001, "learning_rate": 6.9259509848085195e-06, "loss": 0.2874, "step": 11368 }, { "epoch": 0.7859661251296233, "grad_norm": 0.6607854893325269, "learning_rate": 6.9254344061035806e-06, "loss": 0.3354, "step": 11369 }, { "epoch": 0.7860352575181473, "grad_norm": 0.7405652256957126, "learning_rate": 6.924917803266758e-06, "loss": 0.294, "step": 11370 }, { "epoch": 0.7861043899066713, "grad_norm": 0.6390263916158032, "learning_rate": 6.924401176304525e-06, "loss": 0.2682, "step": 11371 }, { "epoch": 0.7861735222951953, "grad_norm": 0.7983324373802511, "learning_rate": 6.923884525223358e-06, "loss": 0.4074, "step": 11372 }, { "epoch": 0.7862426546837193, "grad_norm": 0.6800704640969581, "learning_rate": 6.923367850029732e-06, "loss": 0.3624, "step": 11373 }, { "epoch": 0.7863117870722434, "grad_norm": 0.6711573898819895, "learning_rate": 6.922851150730122e-06, "loss": 0.292, "step": 11374 }, { "epoch": 0.7863809194607674, "grad_norm": 0.664265749737879, "learning_rate": 6.922334427331003e-06, "loss": 0.2783, "step": 11375 }, { "epoch": 0.7864500518492914, "grad_norm": 0.6673873313979332, "learning_rate": 6.921817679838855e-06, "loss": 0.4245, "step": 11376 }, { "epoch": 0.7865191842378154, "grad_norm": 0.6672207472251721, "learning_rate": 6.921300908260151e-06, "loss": 0.2843, "step": 11377 }, { "epoch": 0.7865883166263394, "grad_norm": 0.7302657219312183, "learning_rate": 6.920784112601369e-06, "loss": 0.301, "step": 11378 }, { "epoch": 0.7866574490148635, "grad_norm": 0.7611464137083475, "learning_rate": 6.9202672928689855e-06, "loss": 0.3716, "step": 11379 }, { "epoch": 0.7867265814033875, "grad_norm": 0.7345846979493166, "learning_rate": 6.919750449069477e-06, "loss": 0.353, "step": 11380 }, { "epoch": 0.7867957137919115, "grad_norm": 0.6810020137879395, "learning_rate": 6.919233581209326e-06, "loss": 0.3387, "step": 11381 }, { "epoch": 0.7868648461804355, "grad_norm": 1.0209039764234806, "learning_rate": 6.918716689295004e-06, "loss": 0.3046, "step": 11382 }, { "epoch": 0.7869339785689595, "grad_norm": 0.7169721970847946, "learning_rate": 6.918199773332994e-06, "loss": 0.3968, "step": 11383 }, { "epoch": 0.7870031109574835, "grad_norm": 0.8581470282175249, "learning_rate": 6.917682833329773e-06, "loss": 0.3234, "step": 11384 }, { "epoch": 0.7870722433460076, "grad_norm": 0.7595397322646098, "learning_rate": 6.91716586929182e-06, "loss": 0.313, "step": 11385 }, { "epoch": 0.7871413757345316, "grad_norm": 0.6902701880357821, "learning_rate": 6.916648881225613e-06, "loss": 0.3232, "step": 11386 }, { "epoch": 0.7872105081230556, "grad_norm": 0.7606004104215847, "learning_rate": 6.9161318691376336e-06, "loss": 0.4305, "step": 11387 }, { "epoch": 0.7872796405115797, "grad_norm": 0.6374241888510588, "learning_rate": 6.915614833034361e-06, "loss": 0.2642, "step": 11388 }, { "epoch": 0.7873487729001037, "grad_norm": 0.9165341302322851, "learning_rate": 6.915097772922273e-06, "loss": 0.2755, "step": 11389 }, { "epoch": 0.7874179052886278, "grad_norm": 0.6682539483650833, "learning_rate": 6.9145806888078535e-06, "loss": 0.3714, "step": 11390 }, { "epoch": 0.7874870376771518, "grad_norm": 0.7150303835628048, "learning_rate": 6.914063580697583e-06, "loss": 0.3389, "step": 11391 }, { "epoch": 0.7875561700656758, "grad_norm": 0.8186383976320539, "learning_rate": 6.913546448597941e-06, "loss": 0.2748, "step": 11392 }, { "epoch": 0.7876253024541998, "grad_norm": 0.7647180215663801, "learning_rate": 6.913029292515407e-06, "loss": 0.3922, "step": 11393 }, { "epoch": 0.7876944348427238, "grad_norm": 0.6838112995356384, "learning_rate": 6.912512112456466e-06, "loss": 0.3851, "step": 11394 }, { "epoch": 0.7877635672312479, "grad_norm": 0.5959714757292522, "learning_rate": 6.9119949084276e-06, "loss": 0.2783, "step": 11395 }, { "epoch": 0.7878326996197719, "grad_norm": 0.7008304520187225, "learning_rate": 6.911477680435288e-06, "loss": 0.2566, "step": 11396 }, { "epoch": 0.7879018320082959, "grad_norm": 0.7273540105509296, "learning_rate": 6.910960428486015e-06, "loss": 0.4754, "step": 11397 }, { "epoch": 0.7879709643968199, "grad_norm": 0.7137505301626721, "learning_rate": 6.9104431525862635e-06, "loss": 0.3725, "step": 11398 }, { "epoch": 0.7880400967853439, "grad_norm": 0.6402917323386683, "learning_rate": 6.909925852742517e-06, "loss": 0.2581, "step": 11399 }, { "epoch": 0.7881092291738679, "grad_norm": 0.699053817629997, "learning_rate": 6.909408528961257e-06, "loss": 0.2987, "step": 11400 }, { "epoch": 0.788178361562392, "grad_norm": 0.6620408674905597, "learning_rate": 6.908891181248969e-06, "loss": 0.4154, "step": 11401 }, { "epoch": 0.788247493950916, "grad_norm": 0.6299575914987304, "learning_rate": 6.908373809612138e-06, "loss": 0.31, "step": 11402 }, { "epoch": 0.78831662633944, "grad_norm": 0.6951431557420671, "learning_rate": 6.907856414057246e-06, "loss": 0.3222, "step": 11403 }, { "epoch": 0.788385758727964, "grad_norm": 0.7174472530452192, "learning_rate": 6.907338994590777e-06, "loss": 0.4468, "step": 11404 }, { "epoch": 0.788454891116488, "grad_norm": 0.7030160115156523, "learning_rate": 6.90682155121922e-06, "loss": 0.3742, "step": 11405 }, { "epoch": 0.788524023505012, "grad_norm": 0.8001369140567745, "learning_rate": 6.906304083949057e-06, "loss": 0.3255, "step": 11406 }, { "epoch": 0.7885931558935361, "grad_norm": 0.6749224404949009, "learning_rate": 6.905786592786774e-06, "loss": 0.3047, "step": 11407 }, { "epoch": 0.7886622882820602, "grad_norm": 0.6416622514375482, "learning_rate": 6.905269077738857e-06, "loss": 0.3459, "step": 11408 }, { "epoch": 0.7887314206705842, "grad_norm": 0.7325870729846338, "learning_rate": 6.9047515388117935e-06, "loss": 0.3002, "step": 11409 }, { "epoch": 0.7888005530591082, "grad_norm": 0.6619818205022662, "learning_rate": 6.904233976012067e-06, "loss": 0.3165, "step": 11410 }, { "epoch": 0.7888696854476323, "grad_norm": 0.813369733216132, "learning_rate": 6.903716389346167e-06, "loss": 0.3359, "step": 11411 }, { "epoch": 0.7889388178361563, "grad_norm": 0.7430702033786538, "learning_rate": 6.903198778820578e-06, "loss": 0.4188, "step": 11412 }, { "epoch": 0.7890079502246803, "grad_norm": 0.6551126901176414, "learning_rate": 6.90268114444179e-06, "loss": 0.3091, "step": 11413 }, { "epoch": 0.7890770826132043, "grad_norm": 0.6613664184259616, "learning_rate": 6.902163486216288e-06, "loss": 0.3268, "step": 11414 }, { "epoch": 0.7891462150017283, "grad_norm": 0.6466750442469538, "learning_rate": 6.901645804150562e-06, "loss": 0.3889, "step": 11415 }, { "epoch": 0.7892153473902523, "grad_norm": 0.6722903759775776, "learning_rate": 6.9011280982510995e-06, "loss": 0.3026, "step": 11416 }, { "epoch": 0.7892844797787764, "grad_norm": 0.767127644437313, "learning_rate": 6.900610368524389e-06, "loss": 0.2854, "step": 11417 }, { "epoch": 0.7893536121673004, "grad_norm": 0.681554778741737, "learning_rate": 6.9000926149769186e-06, "loss": 0.3975, "step": 11418 }, { "epoch": 0.7894227445558244, "grad_norm": 0.746819733878102, "learning_rate": 6.89957483761518e-06, "loss": 0.3997, "step": 11419 }, { "epoch": 0.7894918769443484, "grad_norm": 0.7013391683894475, "learning_rate": 6.89905703644566e-06, "loss": 0.2886, "step": 11420 }, { "epoch": 0.7895610093328724, "grad_norm": 0.7315241677911867, "learning_rate": 6.8985392114748485e-06, "loss": 0.2569, "step": 11421 }, { "epoch": 0.7896301417213965, "grad_norm": 0.6854464136424133, "learning_rate": 6.898021362709235e-06, "loss": 0.4621, "step": 11422 }, { "epoch": 0.7896992741099205, "grad_norm": 0.7065391058035627, "learning_rate": 6.897503490155313e-06, "loss": 0.3533, "step": 11423 }, { "epoch": 0.7897684064984445, "grad_norm": 0.6609530902543129, "learning_rate": 6.8969855938195695e-06, "loss": 0.2827, "step": 11424 }, { "epoch": 0.7898375388869685, "grad_norm": 0.7228204411029009, "learning_rate": 6.896467673708498e-06, "loss": 0.3715, "step": 11425 }, { "epoch": 0.7899066712754925, "grad_norm": 0.7320912915536908, "learning_rate": 6.895949729828589e-06, "loss": 0.4094, "step": 11426 }, { "epoch": 0.7899758036640165, "grad_norm": 0.7201795639705415, "learning_rate": 6.895431762186333e-06, "loss": 0.3287, "step": 11427 }, { "epoch": 0.7900449360525407, "grad_norm": 0.7665382210206554, "learning_rate": 6.8949137707882205e-06, "loss": 0.2978, "step": 11428 }, { "epoch": 0.7901140684410647, "grad_norm": 0.6481926502942772, "learning_rate": 6.894395755640748e-06, "loss": 0.4001, "step": 11429 }, { "epoch": 0.7901832008295887, "grad_norm": 0.6249278739843992, "learning_rate": 6.8938777167504045e-06, "loss": 0.3307, "step": 11430 }, { "epoch": 0.7902523332181127, "grad_norm": 0.7351139083598093, "learning_rate": 6.893359654123685e-06, "loss": 0.3164, "step": 11431 }, { "epoch": 0.7903214656066367, "grad_norm": 0.6861169792266433, "learning_rate": 6.892841567767079e-06, "loss": 0.2871, "step": 11432 }, { "epoch": 0.7903905979951608, "grad_norm": 0.7242947040053156, "learning_rate": 6.8923234576870826e-06, "loss": 0.44, "step": 11433 }, { "epoch": 0.7904597303836848, "grad_norm": 0.7041607968524335, "learning_rate": 6.891805323890188e-06, "loss": 0.2876, "step": 11434 }, { "epoch": 0.7905288627722088, "grad_norm": 0.7105317501596311, "learning_rate": 6.891287166382891e-06, "loss": 0.3027, "step": 11435 }, { "epoch": 0.7905979951607328, "grad_norm": 0.5893378268965161, "learning_rate": 6.890768985171684e-06, "loss": 0.3279, "step": 11436 }, { "epoch": 0.7906671275492568, "grad_norm": 0.650312527534888, "learning_rate": 6.890250780263061e-06, "loss": 0.3442, "step": 11437 }, { "epoch": 0.7907362599377808, "grad_norm": 0.5988147507657969, "learning_rate": 6.889732551663519e-06, "loss": 0.2543, "step": 11438 }, { "epoch": 0.7908053923263049, "grad_norm": 0.7082350366521355, "learning_rate": 6.8892142993795516e-06, "loss": 0.3492, "step": 11439 }, { "epoch": 0.7908745247148289, "grad_norm": 0.7120196561346711, "learning_rate": 6.8886960234176545e-06, "loss": 0.4492, "step": 11440 }, { "epoch": 0.7909436571033529, "grad_norm": 0.6654209703579995, "learning_rate": 6.888177723784324e-06, "loss": 0.3008, "step": 11441 }, { "epoch": 0.7910127894918769, "grad_norm": 0.6988379931982769, "learning_rate": 6.887659400486054e-06, "loss": 0.3242, "step": 11442 }, { "epoch": 0.7910819218804009, "grad_norm": 0.7374196222612697, "learning_rate": 6.8871410535293425e-06, "loss": 0.3493, "step": 11443 }, { "epoch": 0.791151054268925, "grad_norm": 0.689616792765786, "learning_rate": 6.886622682920686e-06, "loss": 0.4104, "step": 11444 }, { "epoch": 0.791220186657449, "grad_norm": 0.6924835772893791, "learning_rate": 6.886104288666582e-06, "loss": 0.3125, "step": 11445 }, { "epoch": 0.791289319045973, "grad_norm": 0.6366670629300761, "learning_rate": 6.885585870773524e-06, "loss": 0.3018, "step": 11446 }, { "epoch": 0.791358451434497, "grad_norm": 0.7829374188791842, "learning_rate": 6.885067429248014e-06, "loss": 0.3858, "step": 11447 }, { "epoch": 0.791427583823021, "grad_norm": 0.6711584519980297, "learning_rate": 6.8845489640965466e-06, "loss": 0.3474, "step": 11448 }, { "epoch": 0.7914967162115452, "grad_norm": 0.67355385538693, "learning_rate": 6.8840304753256225e-06, "loss": 0.3109, "step": 11449 }, { "epoch": 0.7915658486000692, "grad_norm": 0.6585316601451173, "learning_rate": 6.883511962941738e-06, "loss": 0.3385, "step": 11450 }, { "epoch": 0.7916349809885932, "grad_norm": 0.6517800014593036, "learning_rate": 6.882993426951392e-06, "loss": 0.4327, "step": 11451 }, { "epoch": 0.7917041133771172, "grad_norm": 0.7293527826975242, "learning_rate": 6.882474867361084e-06, "loss": 0.3139, "step": 11452 }, { "epoch": 0.7917732457656412, "grad_norm": 0.7556857225536483, "learning_rate": 6.8819562841773125e-06, "loss": 0.2796, "step": 11453 }, { "epoch": 0.7918423781541652, "grad_norm": 1.0538375015464343, "learning_rate": 6.8814376774065784e-06, "loss": 0.3626, "step": 11454 }, { "epoch": 0.7919115105426893, "grad_norm": 0.7520339529503527, "learning_rate": 6.88091904705538e-06, "loss": 0.3457, "step": 11455 }, { "epoch": 0.7919806429312133, "grad_norm": 0.6186671640296248, "learning_rate": 6.880400393130219e-06, "loss": 0.3011, "step": 11456 }, { "epoch": 0.7920497753197373, "grad_norm": 0.626128459442993, "learning_rate": 6.879881715637593e-06, "loss": 0.331, "step": 11457 }, { "epoch": 0.7921189077082613, "grad_norm": 0.6472291502577189, "learning_rate": 6.879363014584005e-06, "loss": 0.3861, "step": 11458 }, { "epoch": 0.7921880400967853, "grad_norm": 0.6493889518518087, "learning_rate": 6.878844289975957e-06, "loss": 0.3255, "step": 11459 }, { "epoch": 0.7922571724853094, "grad_norm": 0.6174021717455662, "learning_rate": 6.878325541819947e-06, "loss": 0.2623, "step": 11460 }, { "epoch": 0.7923263048738334, "grad_norm": 0.6791383877111892, "learning_rate": 6.8778067701224795e-06, "loss": 0.3766, "step": 11461 }, { "epoch": 0.7923954372623574, "grad_norm": 0.6352039838248296, "learning_rate": 6.877287974890055e-06, "loss": 0.3634, "step": 11462 }, { "epoch": 0.7924645696508814, "grad_norm": 0.6747763607425993, "learning_rate": 6.876769156129175e-06, "loss": 0.2711, "step": 11463 }, { "epoch": 0.7925337020394054, "grad_norm": 0.7038980803183776, "learning_rate": 6.876250313846343e-06, "loss": 0.2928, "step": 11464 }, { "epoch": 0.7926028344279294, "grad_norm": 0.7435178231682114, "learning_rate": 6.875731448048062e-06, "loss": 0.39, "step": 11465 }, { "epoch": 0.7926719668164535, "grad_norm": 0.7481940801252083, "learning_rate": 6.875212558740836e-06, "loss": 0.3286, "step": 11466 }, { "epoch": 0.7927410992049775, "grad_norm": 0.6169599200241215, "learning_rate": 6.8746936459311655e-06, "loss": 0.2534, "step": 11467 }, { "epoch": 0.7928102315935015, "grad_norm": 0.8808038951875089, "learning_rate": 6.8741747096255565e-06, "loss": 0.3493, "step": 11468 }, { "epoch": 0.7928793639820256, "grad_norm": 0.7254395234034438, "learning_rate": 6.873655749830511e-06, "loss": 0.3807, "step": 11469 }, { "epoch": 0.7929484963705496, "grad_norm": 0.6015171811167979, "learning_rate": 6.873136766552535e-06, "loss": 0.3183, "step": 11470 }, { "epoch": 0.7930176287590737, "grad_norm": 0.7070146104748333, "learning_rate": 6.872617759798133e-06, "loss": 0.2703, "step": 11471 }, { "epoch": 0.7930867611475977, "grad_norm": 0.7064758020919627, "learning_rate": 6.872098729573809e-06, "loss": 0.4055, "step": 11472 }, { "epoch": 0.7931558935361217, "grad_norm": 0.6367380105676015, "learning_rate": 6.871579675886067e-06, "loss": 0.329, "step": 11473 }, { "epoch": 0.7932250259246457, "grad_norm": 0.682158670743908, "learning_rate": 6.871060598741416e-06, "loss": 0.2607, "step": 11474 }, { "epoch": 0.7932941583131697, "grad_norm": 0.6991940090221119, "learning_rate": 6.870541498146357e-06, "loss": 0.3556, "step": 11475 }, { "epoch": 0.7933632907016938, "grad_norm": 0.6465685459515964, "learning_rate": 6.870022374107402e-06, "loss": 0.4252, "step": 11476 }, { "epoch": 0.7934324230902178, "grad_norm": 0.8188568117451783, "learning_rate": 6.869503226631051e-06, "loss": 0.3329, "step": 11477 }, { "epoch": 0.7935015554787418, "grad_norm": 0.6522016342792356, "learning_rate": 6.868984055723813e-06, "loss": 0.2584, "step": 11478 }, { "epoch": 0.7935706878672658, "grad_norm": 0.7392029984535482, "learning_rate": 6.868464861392196e-06, "loss": 0.3761, "step": 11479 }, { "epoch": 0.7936398202557898, "grad_norm": 0.7172878555072734, "learning_rate": 6.867945643642706e-06, "loss": 0.4092, "step": 11480 }, { "epoch": 0.7937089526443138, "grad_norm": 0.6825030755709146, "learning_rate": 6.867426402481851e-06, "loss": 0.2843, "step": 11481 }, { "epoch": 0.7937780850328379, "grad_norm": 0.7492184822184936, "learning_rate": 6.866907137916139e-06, "loss": 0.2921, "step": 11482 }, { "epoch": 0.7938472174213619, "grad_norm": 0.6905335638915404, "learning_rate": 6.866387849952077e-06, "loss": 0.4065, "step": 11483 }, { "epoch": 0.7939163498098859, "grad_norm": 0.9220306307591727, "learning_rate": 6.8658685385961755e-06, "loss": 0.2756, "step": 11484 }, { "epoch": 0.7939854821984099, "grad_norm": 0.8443190123545088, "learning_rate": 6.865349203854941e-06, "loss": 0.2509, "step": 11485 }, { "epoch": 0.7940546145869339, "grad_norm": 0.653309070599074, "learning_rate": 6.864829845734882e-06, "loss": 0.3377, "step": 11486 }, { "epoch": 0.794123746975458, "grad_norm": 0.6385925804017, "learning_rate": 6.864310464242509e-06, "loss": 0.3237, "step": 11487 }, { "epoch": 0.794192879363982, "grad_norm": 0.637323956722456, "learning_rate": 6.863791059384333e-06, "loss": 0.253, "step": 11488 }, { "epoch": 0.7942620117525061, "grad_norm": 0.6238987339640547, "learning_rate": 6.863271631166861e-06, "loss": 0.2794, "step": 11489 }, { "epoch": 0.7943311441410301, "grad_norm": 0.7030607712203393, "learning_rate": 6.862752179596604e-06, "loss": 0.4081, "step": 11490 }, { "epoch": 0.7944002765295541, "grad_norm": 0.755680490880416, "learning_rate": 6.8622327046800745e-06, "loss": 0.3076, "step": 11491 }, { "epoch": 0.7944694089180782, "grad_norm": 0.5943939736653815, "learning_rate": 6.861713206423778e-06, "loss": 0.2942, "step": 11492 }, { "epoch": 0.7945385413066022, "grad_norm": 0.6711861511699616, "learning_rate": 6.861193684834231e-06, "loss": 0.3468, "step": 11493 }, { "epoch": 0.7946076736951262, "grad_norm": 0.7700741396530202, "learning_rate": 6.860674139917942e-06, "loss": 0.4659, "step": 11494 }, { "epoch": 0.7946768060836502, "grad_norm": 0.7272245499771701, "learning_rate": 6.860154571681424e-06, "loss": 0.2974, "step": 11495 }, { "epoch": 0.7947459384721742, "grad_norm": 0.7176350698295063, "learning_rate": 6.8596349801311865e-06, "loss": 0.2846, "step": 11496 }, { "epoch": 0.7948150708606982, "grad_norm": 0.7177406657969942, "learning_rate": 6.859115365273746e-06, "loss": 0.3805, "step": 11497 }, { "epoch": 0.7948842032492223, "grad_norm": 0.6395739657737611, "learning_rate": 6.85859572711561e-06, "loss": 0.3283, "step": 11498 }, { "epoch": 0.7949533356377463, "grad_norm": 0.6418546250934944, "learning_rate": 6.858076065663292e-06, "loss": 0.3196, "step": 11499 }, { "epoch": 0.7950224680262703, "grad_norm": 0.7851512826143088, "learning_rate": 6.857556380923309e-06, "loss": 0.3257, "step": 11500 }, { "epoch": 0.7950916004147943, "grad_norm": 0.6200338142461387, "learning_rate": 6.857036672902171e-06, "loss": 0.3444, "step": 11501 }, { "epoch": 0.7951607328033183, "grad_norm": 0.5957001299973631, "learning_rate": 6.8565169416063924e-06, "loss": 0.227, "step": 11502 }, { "epoch": 0.7952298651918424, "grad_norm": 0.6971654411926147, "learning_rate": 6.8559971870424865e-06, "loss": 0.3194, "step": 11503 }, { "epoch": 0.7952989975803664, "grad_norm": 0.7397870000202763, "learning_rate": 6.855477409216969e-06, "loss": 0.3701, "step": 11504 }, { "epoch": 0.7953681299688904, "grad_norm": 0.6855400992711548, "learning_rate": 6.854957608136352e-06, "loss": 0.3467, "step": 11505 }, { "epoch": 0.7954372623574144, "grad_norm": 0.6154535729522137, "learning_rate": 6.854437783807153e-06, "loss": 0.2599, "step": 11506 }, { "epoch": 0.7955063947459384, "grad_norm": 0.7000557686681557, "learning_rate": 6.8539179362358855e-06, "loss": 0.3112, "step": 11507 }, { "epoch": 0.7955755271344624, "grad_norm": 0.7837152832474998, "learning_rate": 6.853398065429064e-06, "loss": 0.4117, "step": 11508 }, { "epoch": 0.7956446595229865, "grad_norm": 0.6444041802693788, "learning_rate": 6.852878171393207e-06, "loss": 0.3042, "step": 11509 }, { "epoch": 0.7957137919115106, "grad_norm": 0.6628851947585425, "learning_rate": 6.8523582541348275e-06, "loss": 0.2891, "step": 11510 }, { "epoch": 0.7957829243000346, "grad_norm": 1.366145297501699, "learning_rate": 6.851838313660444e-06, "loss": 0.3514, "step": 11511 }, { "epoch": 0.7958520566885586, "grad_norm": 0.6863826435099838, "learning_rate": 6.851318349976571e-06, "loss": 0.3359, "step": 11512 }, { "epoch": 0.7959211890770826, "grad_norm": 0.6595085296151686, "learning_rate": 6.850798363089726e-06, "loss": 0.324, "step": 11513 }, { "epoch": 0.7959903214656067, "grad_norm": 0.6661345724990939, "learning_rate": 6.850278353006427e-06, "loss": 0.2723, "step": 11514 }, { "epoch": 0.7960594538541307, "grad_norm": 0.6486987610739261, "learning_rate": 6.849758319733192e-06, "loss": 0.4053, "step": 11515 }, { "epoch": 0.7961285862426547, "grad_norm": 1.4398322573767148, "learning_rate": 6.849238263276537e-06, "loss": 0.3267, "step": 11516 }, { "epoch": 0.7961977186311787, "grad_norm": 0.6839846487450411, "learning_rate": 6.848718183642978e-06, "loss": 0.3058, "step": 11517 }, { "epoch": 0.7962668510197027, "grad_norm": 0.6598031298854777, "learning_rate": 6.848198080839038e-06, "loss": 0.308, "step": 11518 }, { "epoch": 0.7963359834082268, "grad_norm": 0.6905833766502142, "learning_rate": 6.8476779548712335e-06, "loss": 0.3905, "step": 11519 }, { "epoch": 0.7964051157967508, "grad_norm": 0.6602272748419948, "learning_rate": 6.847157805746083e-06, "loss": 0.265, "step": 11520 }, { "epoch": 0.7964742481852748, "grad_norm": 0.6215067484525139, "learning_rate": 6.846637633470104e-06, "loss": 0.2866, "step": 11521 }, { "epoch": 0.7965433805737988, "grad_norm": 0.6818626577616825, "learning_rate": 6.8461174380498194e-06, "loss": 0.3388, "step": 11522 }, { "epoch": 0.7966125129623228, "grad_norm": 0.665168974550904, "learning_rate": 6.845597219491749e-06, "loss": 0.3664, "step": 11523 }, { "epoch": 0.7966816453508468, "grad_norm": 0.6955536055331402, "learning_rate": 6.845076977802408e-06, "loss": 0.2998, "step": 11524 }, { "epoch": 0.7967507777393709, "grad_norm": 0.6776278037006449, "learning_rate": 6.844556712988321e-06, "loss": 0.308, "step": 11525 }, { "epoch": 0.7968199101278949, "grad_norm": 0.6769680458677281, "learning_rate": 6.844036425056006e-06, "loss": 0.3863, "step": 11526 }, { "epoch": 0.7968890425164189, "grad_norm": 0.7020210264077641, "learning_rate": 6.843516114011988e-06, "loss": 0.3299, "step": 11527 }, { "epoch": 0.7969581749049429, "grad_norm": 0.628220299175315, "learning_rate": 6.842995779862783e-06, "loss": 0.3238, "step": 11528 }, { "epoch": 0.7970273072934669, "grad_norm": 0.676247291947048, "learning_rate": 6.8424754226149156e-06, "loss": 0.3725, "step": 11529 }, { "epoch": 0.7970964396819911, "grad_norm": 0.6709239160251022, "learning_rate": 6.841955042274908e-06, "loss": 0.3246, "step": 11530 }, { "epoch": 0.7971655720705151, "grad_norm": 0.6262006160531405, "learning_rate": 6.841434638849279e-06, "loss": 0.2843, "step": 11531 }, { "epoch": 0.7972347044590391, "grad_norm": 0.7096281077325367, "learning_rate": 6.840914212344554e-06, "loss": 0.2776, "step": 11532 }, { "epoch": 0.7973038368475631, "grad_norm": 0.6841843815052133, "learning_rate": 6.840393762767256e-06, "loss": 0.4475, "step": 11533 }, { "epoch": 0.7973729692360871, "grad_norm": 0.671846528305501, "learning_rate": 6.8398732901239055e-06, "loss": 0.3306, "step": 11534 }, { "epoch": 0.7974421016246112, "grad_norm": 0.616743852676669, "learning_rate": 6.839352794421025e-06, "loss": 0.2538, "step": 11535 }, { "epoch": 0.7975112340131352, "grad_norm": 0.8323930261182548, "learning_rate": 6.838832275665142e-06, "loss": 0.3632, "step": 11536 }, { "epoch": 0.7975803664016592, "grad_norm": 0.6898238019941032, "learning_rate": 6.838311733862777e-06, "loss": 0.3407, "step": 11537 }, { "epoch": 0.7976494987901832, "grad_norm": 0.7957697744290438, "learning_rate": 6.837791169020455e-06, "loss": 0.2645, "step": 11538 }, { "epoch": 0.7977186311787072, "grad_norm": 0.6606744070918763, "learning_rate": 6.837270581144699e-06, "loss": 0.2903, "step": 11539 }, { "epoch": 0.7977877635672312, "grad_norm": 0.6623592192396263, "learning_rate": 6.836749970242037e-06, "loss": 0.4298, "step": 11540 }, { "epoch": 0.7978568959557553, "grad_norm": 0.6716333199188779, "learning_rate": 6.836229336318992e-06, "loss": 0.3036, "step": 11541 }, { "epoch": 0.7979260283442793, "grad_norm": 0.6419058882411319, "learning_rate": 6.835708679382089e-06, "loss": 0.2554, "step": 11542 }, { "epoch": 0.7979951607328033, "grad_norm": 0.7715609416262893, "learning_rate": 6.835187999437854e-06, "loss": 0.342, "step": 11543 }, { "epoch": 0.7980642931213273, "grad_norm": 0.668315692876793, "learning_rate": 6.834667296492812e-06, "loss": 0.3529, "step": 11544 }, { "epoch": 0.7981334255098513, "grad_norm": 1.0486204885275991, "learning_rate": 6.834146570553488e-06, "loss": 0.2857, "step": 11545 }, { "epoch": 0.7982025578983754, "grad_norm": 0.634188674678784, "learning_rate": 6.833625821626412e-06, "loss": 0.3317, "step": 11546 }, { "epoch": 0.7982716902868994, "grad_norm": 0.7188421129083142, "learning_rate": 6.833105049718108e-06, "loss": 0.4327, "step": 11547 }, { "epoch": 0.7983408226754234, "grad_norm": 0.6196632749304526, "learning_rate": 6.832584254835103e-06, "loss": 0.3184, "step": 11548 }, { "epoch": 0.7984099550639474, "grad_norm": 0.6121364364865397, "learning_rate": 6.832063436983925e-06, "loss": 0.2825, "step": 11549 }, { "epoch": 0.7984790874524715, "grad_norm": 0.6708799834795504, "learning_rate": 6.831542596171101e-06, "loss": 0.3177, "step": 11550 }, { "epoch": 0.7985482198409956, "grad_norm": 0.6508958933167677, "learning_rate": 6.8310217324031605e-06, "loss": 0.389, "step": 11551 }, { "epoch": 0.7986173522295196, "grad_norm": 0.6135470914304476, "learning_rate": 6.8305008456866274e-06, "loss": 0.3139, "step": 11552 }, { "epoch": 0.7986864846180436, "grad_norm": 0.894211394706071, "learning_rate": 6.829979936028035e-06, "loss": 0.3151, "step": 11553 }, { "epoch": 0.7987556170065676, "grad_norm": 0.6617670170114868, "learning_rate": 6.82945900343391e-06, "loss": 0.3655, "step": 11554 }, { "epoch": 0.7988247493950916, "grad_norm": 0.6532768563745672, "learning_rate": 6.82893804791078e-06, "loss": 0.3437, "step": 11555 }, { "epoch": 0.7988938817836156, "grad_norm": 0.5999346301744476, "learning_rate": 6.828417069465177e-06, "loss": 0.2834, "step": 11556 }, { "epoch": 0.7989630141721397, "grad_norm": 1.4125545342334298, "learning_rate": 6.827896068103628e-06, "loss": 0.3482, "step": 11557 }, { "epoch": 0.7990321465606637, "grad_norm": 0.7659424157484769, "learning_rate": 6.827375043832663e-06, "loss": 0.401, "step": 11558 }, { "epoch": 0.7991012789491877, "grad_norm": 0.7106825449599344, "learning_rate": 6.826853996658813e-06, "loss": 0.237, "step": 11559 }, { "epoch": 0.7991704113377117, "grad_norm": 0.6210207791072468, "learning_rate": 6.82633292658861e-06, "loss": 0.2772, "step": 11560 }, { "epoch": 0.7992395437262357, "grad_norm": 0.663730615783114, "learning_rate": 6.825811833628583e-06, "loss": 0.3397, "step": 11561 }, { "epoch": 0.7993086761147598, "grad_norm": 0.6525100617909162, "learning_rate": 6.825290717785262e-06, "loss": 0.3826, "step": 11562 }, { "epoch": 0.7993778085032838, "grad_norm": 0.6107578880464629, "learning_rate": 6.824769579065179e-06, "loss": 0.3009, "step": 11563 }, { "epoch": 0.7994469408918078, "grad_norm": 0.6694424671979711, "learning_rate": 6.824248417474866e-06, "loss": 0.3331, "step": 11564 }, { "epoch": 0.7995160732803318, "grad_norm": 0.7367264582768245, "learning_rate": 6.823727233020855e-06, "loss": 0.3606, "step": 11565 }, { "epoch": 0.7995852056688558, "grad_norm": 0.7135471655985984, "learning_rate": 6.8232060257096776e-06, "loss": 0.3046, "step": 11566 }, { "epoch": 0.7996543380573798, "grad_norm": 0.6268311562631074, "learning_rate": 6.822684795547866e-06, "loss": 0.2872, "step": 11567 }, { "epoch": 0.7997234704459039, "grad_norm": 0.6975788542709821, "learning_rate": 6.822163542541953e-06, "loss": 0.3359, "step": 11568 }, { "epoch": 0.7997926028344279, "grad_norm": 0.7101634272750922, "learning_rate": 6.8216422666984725e-06, "loss": 0.3796, "step": 11569 }, { "epoch": 0.799861735222952, "grad_norm": 0.7139105370202329, "learning_rate": 6.821120968023956e-06, "loss": 0.3503, "step": 11570 }, { "epoch": 0.799930867611476, "grad_norm": 0.6146928655592626, "learning_rate": 6.8205996465249395e-06, "loss": 0.283, "step": 11571 }, { "epoch": 0.8, "grad_norm": 0.7042145102767547, "learning_rate": 6.8200783022079555e-06, "loss": 0.4454, "step": 11572 }, { "epoch": 0.8000691323885241, "grad_norm": 0.7074325751131537, "learning_rate": 6.8195569350795365e-06, "loss": 0.3626, "step": 11573 }, { "epoch": 0.8001382647770481, "grad_norm": 0.6218599113217107, "learning_rate": 6.81903554514622e-06, "loss": 0.252, "step": 11574 }, { "epoch": 0.8002073971655721, "grad_norm": 0.6829285471293979, "learning_rate": 6.818514132414538e-06, "loss": 0.2821, "step": 11575 }, { "epoch": 0.8002765295540961, "grad_norm": 0.6897391708982579, "learning_rate": 6.817992696891027e-06, "loss": 0.3888, "step": 11576 }, { "epoch": 0.8003456619426201, "grad_norm": 0.7390910481567287, "learning_rate": 6.817471238582222e-06, "loss": 0.2821, "step": 11577 }, { "epoch": 0.8004147943311442, "grad_norm": 0.6911134761419395, "learning_rate": 6.816949757494659e-06, "loss": 0.2971, "step": 11578 }, { "epoch": 0.8004839267196682, "grad_norm": 0.7107324685980078, "learning_rate": 6.816428253634873e-06, "loss": 0.3712, "step": 11579 }, { "epoch": 0.8005530591081922, "grad_norm": 0.7729228444694445, "learning_rate": 6.8159067270094e-06, "loss": 0.3998, "step": 11580 }, { "epoch": 0.8006221914967162, "grad_norm": 0.6680980225621596, "learning_rate": 6.815385177624776e-06, "loss": 0.3258, "step": 11581 }, { "epoch": 0.8006913238852402, "grad_norm": 0.6699708839797007, "learning_rate": 6.814863605487539e-06, "loss": 0.3023, "step": 11582 }, { "epoch": 0.8007604562737642, "grad_norm": 0.6691748524035656, "learning_rate": 6.8143420106042266e-06, "loss": 0.4052, "step": 11583 }, { "epoch": 0.8008295886622883, "grad_norm": 0.8157057169558121, "learning_rate": 6.813820392981374e-06, "loss": 0.2493, "step": 11584 }, { "epoch": 0.8008987210508123, "grad_norm": 0.6654209265438782, "learning_rate": 6.8132987526255205e-06, "loss": 0.3029, "step": 11585 }, { "epoch": 0.8009678534393363, "grad_norm": 0.6413672676066404, "learning_rate": 6.812777089543202e-06, "loss": 0.3612, "step": 11586 }, { "epoch": 0.8010369858278603, "grad_norm": 0.6642448961640933, "learning_rate": 6.812255403740959e-06, "loss": 0.3301, "step": 11587 }, { "epoch": 0.8011061182163843, "grad_norm": 0.5680471442923434, "learning_rate": 6.811733695225326e-06, "loss": 0.2479, "step": 11588 }, { "epoch": 0.8011752506049084, "grad_norm": 0.6954874029556727, "learning_rate": 6.811211964002848e-06, "loss": 0.2535, "step": 11589 }, { "epoch": 0.8012443829934324, "grad_norm": 0.7032359863638905, "learning_rate": 6.810690210080058e-06, "loss": 0.4109, "step": 11590 }, { "epoch": 0.8013135153819565, "grad_norm": 0.642138818486489, "learning_rate": 6.810168433463499e-06, "loss": 0.3762, "step": 11591 }, { "epoch": 0.8013826477704805, "grad_norm": 0.6186959889757261, "learning_rate": 6.809646634159707e-06, "loss": 0.2489, "step": 11592 }, { "epoch": 0.8014517801590045, "grad_norm": 0.7154500828348794, "learning_rate": 6.809124812175225e-06, "loss": 0.3652, "step": 11593 }, { "epoch": 0.8015209125475286, "grad_norm": 0.7453089058265984, "learning_rate": 6.8086029675165925e-06, "loss": 0.4389, "step": 11594 }, { "epoch": 0.8015900449360526, "grad_norm": 1.0270097402867477, "learning_rate": 6.808081100190348e-06, "loss": 0.291, "step": 11595 }, { "epoch": 0.8016591773245766, "grad_norm": 0.7344761115724769, "learning_rate": 6.807559210203035e-06, "loss": 0.334, "step": 11596 }, { "epoch": 0.8017283097131006, "grad_norm": 0.7104955622599695, "learning_rate": 6.8070372975611935e-06, "loss": 0.3481, "step": 11597 }, { "epoch": 0.8017974421016246, "grad_norm": 0.7110610266444288, "learning_rate": 6.8065153622713635e-06, "loss": 0.3361, "step": 11598 }, { "epoch": 0.8018665744901486, "grad_norm": 0.6220304933045279, "learning_rate": 6.805993404340085e-06, "loss": 0.222, "step": 11599 }, { "epoch": 0.8019357068786727, "grad_norm": 0.7561151072411824, "learning_rate": 6.805471423773905e-06, "loss": 0.3466, "step": 11600 }, { "epoch": 0.8020048392671967, "grad_norm": 0.8679128852605361, "learning_rate": 6.804949420579362e-06, "loss": 0.363, "step": 11601 }, { "epoch": 0.8020739716557207, "grad_norm": 0.5584458424488894, "learning_rate": 6.804427394762999e-06, "loss": 0.2783, "step": 11602 }, { "epoch": 0.8021431040442447, "grad_norm": 0.6500247969380089, "learning_rate": 6.803905346331358e-06, "loss": 0.3385, "step": 11603 }, { "epoch": 0.8022122364327687, "grad_norm": 0.7714114584039329, "learning_rate": 6.803383275290984e-06, "loss": 0.4176, "step": 11604 }, { "epoch": 0.8022813688212928, "grad_norm": 0.6265236386540224, "learning_rate": 6.802861181648419e-06, "loss": 0.341, "step": 11605 }, { "epoch": 0.8023505012098168, "grad_norm": 0.6901110870271827, "learning_rate": 6.802339065410205e-06, "loss": 0.2373, "step": 11606 }, { "epoch": 0.8024196335983408, "grad_norm": 0.6321784755855524, "learning_rate": 6.8018169265828896e-06, "loss": 0.3044, "step": 11607 }, { "epoch": 0.8024887659868648, "grad_norm": 0.6781862516559196, "learning_rate": 6.801294765173012e-06, "loss": 0.379, "step": 11608 }, { "epoch": 0.8025578983753888, "grad_norm": 0.659394522015371, "learning_rate": 6.80077258118712e-06, "loss": 0.2785, "step": 11609 }, { "epoch": 0.8026270307639128, "grad_norm": 0.6235811812033316, "learning_rate": 6.800250374631758e-06, "loss": 0.259, "step": 11610 }, { "epoch": 0.802696163152437, "grad_norm": 0.6443932043246577, "learning_rate": 6.79972814551347e-06, "loss": 0.3179, "step": 11611 }, { "epoch": 0.802765295540961, "grad_norm": 0.6347895337190433, "learning_rate": 6.799205893838801e-06, "loss": 0.365, "step": 11612 }, { "epoch": 0.802834427929485, "grad_norm": 1.2073924518912196, "learning_rate": 6.798683619614297e-06, "loss": 0.265, "step": 11613 }, { "epoch": 0.802903560318009, "grad_norm": 0.6306335952185341, "learning_rate": 6.798161322846504e-06, "loss": 0.3019, "step": 11614 }, { "epoch": 0.802972692706533, "grad_norm": 0.6327600572786912, "learning_rate": 6.797639003541967e-06, "loss": 0.3915, "step": 11615 }, { "epoch": 0.8030418250950571, "grad_norm": 0.670643380102479, "learning_rate": 6.7971166617072334e-06, "loss": 0.2995, "step": 11616 }, { "epoch": 0.8031109574835811, "grad_norm": 0.6089174904999342, "learning_rate": 6.796594297348851e-06, "loss": 0.2874, "step": 11617 }, { "epoch": 0.8031800898721051, "grad_norm": 0.6895642121886638, "learning_rate": 6.796071910473363e-06, "loss": 0.3429, "step": 11618 }, { "epoch": 0.8032492222606291, "grad_norm": 0.6960405468142588, "learning_rate": 6.795549501087321e-06, "loss": 0.3749, "step": 11619 }, { "epoch": 0.8033183546491531, "grad_norm": 0.6018700900760714, "learning_rate": 6.795027069197269e-06, "loss": 0.3008, "step": 11620 }, { "epoch": 0.8033874870376772, "grad_norm": 0.9290021445116905, "learning_rate": 6.794504614809756e-06, "loss": 0.2573, "step": 11621 }, { "epoch": 0.8034566194262012, "grad_norm": 0.7574645293221302, "learning_rate": 6.79398213793133e-06, "loss": 0.4509, "step": 11622 }, { "epoch": 0.8035257518147252, "grad_norm": 0.6588458583647928, "learning_rate": 6.793459638568539e-06, "loss": 0.3236, "step": 11623 }, { "epoch": 0.8035948842032492, "grad_norm": 1.5576094364186133, "learning_rate": 6.7929371167279335e-06, "loss": 0.2818, "step": 11624 }, { "epoch": 0.8036640165917732, "grad_norm": 1.298002870916002, "learning_rate": 6.79241457241606e-06, "loss": 0.3305, "step": 11625 }, { "epoch": 0.8037331489802972, "grad_norm": 0.9423813738409585, "learning_rate": 6.791892005639469e-06, "loss": 0.4074, "step": 11626 }, { "epoch": 0.8038022813688213, "grad_norm": 0.6449378029484931, "learning_rate": 6.791369416404708e-06, "loss": 0.292, "step": 11627 }, { "epoch": 0.8038714137573453, "grad_norm": 0.6995527122313742, "learning_rate": 6.790846804718331e-06, "loss": 0.402, "step": 11628 }, { "epoch": 0.8039405461458693, "grad_norm": 0.7099961882771798, "learning_rate": 6.790324170586883e-06, "loss": 0.3563, "step": 11629 }, { "epoch": 0.8040096785343933, "grad_norm": 0.6232575489103894, "learning_rate": 6.789801514016916e-06, "loss": 0.3258, "step": 11630 }, { "epoch": 0.8040788109229174, "grad_norm": 0.6216419197837737, "learning_rate": 6.789278835014983e-06, "loss": 0.3239, "step": 11631 }, { "epoch": 0.8041479433114415, "grad_norm": 1.1017709687100385, "learning_rate": 6.788756133587633e-06, "loss": 0.3144, "step": 11632 }, { "epoch": 0.8042170756999655, "grad_norm": 0.6849590775389288, "learning_rate": 6.788233409741416e-06, "loss": 0.4864, "step": 11633 }, { "epoch": 0.8042862080884895, "grad_norm": 0.7281670866254276, "learning_rate": 6.787710663482884e-06, "loss": 0.3483, "step": 11634 }, { "epoch": 0.8043553404770135, "grad_norm": 0.7007501313848759, "learning_rate": 6.78718789481859e-06, "loss": 0.3274, "step": 11635 }, { "epoch": 0.8044244728655375, "grad_norm": 0.6668954912612178, "learning_rate": 6.786665103755085e-06, "loss": 0.3541, "step": 11636 }, { "epoch": 0.8044936052540616, "grad_norm": 0.6290411675131099, "learning_rate": 6.7861422902989205e-06, "loss": 0.3535, "step": 11637 }, { "epoch": 0.8045627376425856, "grad_norm": 0.6774718902297214, "learning_rate": 6.78561945445665e-06, "loss": 0.3373, "step": 11638 }, { "epoch": 0.8046318700311096, "grad_norm": 0.650031254317476, "learning_rate": 6.785096596234827e-06, "loss": 0.2479, "step": 11639 }, { "epoch": 0.8047010024196336, "grad_norm": 0.6776405731881073, "learning_rate": 6.7845737156400035e-06, "loss": 0.4205, "step": 11640 }, { "epoch": 0.8047701348081576, "grad_norm": 0.7414191555215893, "learning_rate": 6.784050812678733e-06, "loss": 0.3486, "step": 11641 }, { "epoch": 0.8048392671966816, "grad_norm": 0.8495845495200304, "learning_rate": 6.78352788735757e-06, "loss": 0.3145, "step": 11642 }, { "epoch": 0.8049083995852057, "grad_norm": 0.6342869771823468, "learning_rate": 6.783004939683068e-06, "loss": 0.2878, "step": 11643 }, { "epoch": 0.8049775319737297, "grad_norm": 0.6731055218562296, "learning_rate": 6.782481969661781e-06, "loss": 0.3653, "step": 11644 }, { "epoch": 0.8050466643622537, "grad_norm": 0.7589112297891986, "learning_rate": 6.781958977300262e-06, "loss": 0.3307, "step": 11645 }, { "epoch": 0.8051157967507777, "grad_norm": 0.7154345896294965, "learning_rate": 6.7814359626050685e-06, "loss": 0.3125, "step": 11646 }, { "epoch": 0.8051849291393017, "grad_norm": 0.7185201025863336, "learning_rate": 6.780912925582754e-06, "loss": 0.4118, "step": 11647 }, { "epoch": 0.8052540615278257, "grad_norm": 0.6993652202717107, "learning_rate": 6.780389866239874e-06, "loss": 0.3424, "step": 11648 }, { "epoch": 0.8053231939163498, "grad_norm": 0.7150359031449904, "learning_rate": 6.779866784582985e-06, "loss": 0.2973, "step": 11649 }, { "epoch": 0.8053923263048738, "grad_norm": 0.6664685687860364, "learning_rate": 6.779343680618642e-06, "loss": 0.2597, "step": 11650 }, { "epoch": 0.8054614586933978, "grad_norm": 0.6089902771287093, "learning_rate": 6.7788205543534e-06, "loss": 0.3777, "step": 11651 }, { "epoch": 0.8055305910819219, "grad_norm": 0.649914006880249, "learning_rate": 6.778297405793818e-06, "loss": 0.3426, "step": 11652 }, { "epoch": 0.805599723470446, "grad_norm": 0.646578273310672, "learning_rate": 6.777774234946452e-06, "loss": 0.2775, "step": 11653 }, { "epoch": 0.80566885585897, "grad_norm": 0.711511563004529, "learning_rate": 6.777251041817857e-06, "loss": 0.3551, "step": 11654 }, { "epoch": 0.805737988247494, "grad_norm": 0.6694045575874075, "learning_rate": 6.776727826414592e-06, "loss": 0.3782, "step": 11655 }, { "epoch": 0.805807120636018, "grad_norm": 0.656014628854187, "learning_rate": 6.776204588743216e-06, "loss": 0.2942, "step": 11656 }, { "epoch": 0.805876253024542, "grad_norm": 0.812293867973541, "learning_rate": 6.775681328810284e-06, "loss": 0.3064, "step": 11657 }, { "epoch": 0.805945385413066, "grad_norm": 0.7803798167324407, "learning_rate": 6.775158046622356e-06, "loss": 0.4022, "step": 11658 }, { "epoch": 0.8060145178015901, "grad_norm": 0.6476461507927348, "learning_rate": 6.774634742185989e-06, "loss": 0.2851, "step": 11659 }, { "epoch": 0.8060836501901141, "grad_norm": 0.6636413330196173, "learning_rate": 6.774111415507743e-06, "loss": 0.2894, "step": 11660 }, { "epoch": 0.8061527825786381, "grad_norm": 0.6307748988906505, "learning_rate": 6.773588066594177e-06, "loss": 0.378, "step": 11661 }, { "epoch": 0.8062219149671621, "grad_norm": 0.8331282662895964, "learning_rate": 6.77306469545185e-06, "loss": 0.3643, "step": 11662 }, { "epoch": 0.8062910473556861, "grad_norm": 0.753431394900975, "learning_rate": 6.772541302087319e-06, "loss": 0.2758, "step": 11663 }, { "epoch": 0.8063601797442101, "grad_norm": 0.670639935785085, "learning_rate": 6.772017886507147e-06, "loss": 0.2994, "step": 11664 }, { "epoch": 0.8064293121327342, "grad_norm": 0.6439552503953219, "learning_rate": 6.771494448717895e-06, "loss": 0.3876, "step": 11665 }, { "epoch": 0.8064984445212582, "grad_norm": 0.6882180904925932, "learning_rate": 6.770970988726118e-06, "loss": 0.3195, "step": 11666 }, { "epoch": 0.8065675769097822, "grad_norm": 0.653664480796752, "learning_rate": 6.770447506538383e-06, "loss": 0.2581, "step": 11667 }, { "epoch": 0.8066367092983062, "grad_norm": 0.7868500561400859, "learning_rate": 6.769924002161248e-06, "loss": 0.3592, "step": 11668 }, { "epoch": 0.8067058416868302, "grad_norm": 0.7511226418234642, "learning_rate": 6.769400475601273e-06, "loss": 0.3496, "step": 11669 }, { "epoch": 0.8067749740753543, "grad_norm": 0.6161045360574394, "learning_rate": 6.76887692686502e-06, "loss": 0.2565, "step": 11670 }, { "epoch": 0.8068441064638783, "grad_norm": 0.6398401415529037, "learning_rate": 6.768353355959053e-06, "loss": 0.2639, "step": 11671 }, { "epoch": 0.8069132388524024, "grad_norm": 0.699673540002414, "learning_rate": 6.767829762889933e-06, "loss": 0.4211, "step": 11672 }, { "epoch": 0.8069823712409264, "grad_norm": 0.6785229417890398, "learning_rate": 6.767306147664219e-06, "loss": 0.3349, "step": 11673 }, { "epoch": 0.8070515036294504, "grad_norm": 0.6171077498575553, "learning_rate": 6.766782510288479e-06, "loss": 0.2313, "step": 11674 }, { "epoch": 0.8071206360179745, "grad_norm": 0.7451341958725659, "learning_rate": 6.766258850769273e-06, "loss": 0.2818, "step": 11675 }, { "epoch": 0.8071897684064985, "grad_norm": 0.6419408215976941, "learning_rate": 6.765735169113162e-06, "loss": 0.3941, "step": 11676 }, { "epoch": 0.8072589007950225, "grad_norm": 1.059683721172362, "learning_rate": 6.765211465326715e-06, "loss": 0.3483, "step": 11677 }, { "epoch": 0.8073280331835465, "grad_norm": 0.60866586590924, "learning_rate": 6.76468773941649e-06, "loss": 0.273, "step": 11678 }, { "epoch": 0.8073971655720705, "grad_norm": 0.7072980483045325, "learning_rate": 6.764163991389055e-06, "loss": 0.3357, "step": 11679 }, { "epoch": 0.8074662979605945, "grad_norm": 0.6942754213309718, "learning_rate": 6.76364022125097e-06, "loss": 0.3968, "step": 11680 }, { "epoch": 0.8075354303491186, "grad_norm": 0.7160810319471619, "learning_rate": 6.763116429008805e-06, "loss": 0.2491, "step": 11681 }, { "epoch": 0.8076045627376426, "grad_norm": 0.7284839650450543, "learning_rate": 6.76259261466912e-06, "loss": 0.3136, "step": 11682 }, { "epoch": 0.8076736951261666, "grad_norm": 0.6684956658757163, "learning_rate": 6.762068778238483e-06, "loss": 0.4311, "step": 11683 }, { "epoch": 0.8077428275146906, "grad_norm": 1.1052734884133397, "learning_rate": 6.761544919723458e-06, "loss": 0.303, "step": 11684 }, { "epoch": 0.8078119599032146, "grad_norm": 0.5889062954900196, "learning_rate": 6.7610210391306116e-06, "loss": 0.303, "step": 11685 }, { "epoch": 0.8078810922917387, "grad_norm": 0.709923771403207, "learning_rate": 6.76049713646651e-06, "loss": 0.4209, "step": 11686 }, { "epoch": 0.8079502246802627, "grad_norm": 0.6364529008428969, "learning_rate": 6.759973211737716e-06, "loss": 0.3915, "step": 11687 }, { "epoch": 0.8080193570687867, "grad_norm": 0.6474458372766908, "learning_rate": 6.759449264950801e-06, "loss": 0.2804, "step": 11688 }, { "epoch": 0.8080884894573107, "grad_norm": 0.7508472847548311, "learning_rate": 6.7589252961123285e-06, "loss": 0.278, "step": 11689 }, { "epoch": 0.8081576218458347, "grad_norm": 0.6648612089087736, "learning_rate": 6.758401305228867e-06, "loss": 0.3937, "step": 11690 }, { "epoch": 0.8082267542343587, "grad_norm": 0.668955231188687, "learning_rate": 6.757877292306981e-06, "loss": 0.3253, "step": 11691 }, { "epoch": 0.8082958866228829, "grad_norm": 0.600213004669064, "learning_rate": 6.757353257353243e-06, "loss": 0.2972, "step": 11692 }, { "epoch": 0.8083650190114069, "grad_norm": 0.7409565501071353, "learning_rate": 6.756829200374216e-06, "loss": 0.2948, "step": 11693 }, { "epoch": 0.8084341513999309, "grad_norm": 0.695939340893488, "learning_rate": 6.756305121376471e-06, "loss": 0.3292, "step": 11694 }, { "epoch": 0.8085032837884549, "grad_norm": 0.748678831799322, "learning_rate": 6.7557810203665755e-06, "loss": 0.286, "step": 11695 }, { "epoch": 0.808572416176979, "grad_norm": 0.6448546120310958, "learning_rate": 6.755256897351099e-06, "loss": 0.2778, "step": 11696 }, { "epoch": 0.808641548565503, "grad_norm": 0.7622265384471321, "learning_rate": 6.754732752336609e-06, "loss": 0.3634, "step": 11697 }, { "epoch": 0.808710680954027, "grad_norm": 0.6423509222485599, "learning_rate": 6.754208585329674e-06, "loss": 0.3381, "step": 11698 }, { "epoch": 0.808779813342551, "grad_norm": 0.6054233694252634, "learning_rate": 6.753684396336867e-06, "loss": 0.2769, "step": 11699 }, { "epoch": 0.808848945731075, "grad_norm": 0.7113554089646846, "learning_rate": 6.753160185364755e-06, "loss": 0.3317, "step": 11700 }, { "epoch": 0.808918078119599, "grad_norm": 0.7004380744672662, "learning_rate": 6.7526359524199085e-06, "loss": 0.3402, "step": 11701 }, { "epoch": 0.8089872105081231, "grad_norm": 0.687617223565979, "learning_rate": 6.752111697508899e-06, "loss": 0.3042, "step": 11702 }, { "epoch": 0.8090563428966471, "grad_norm": 0.8325757627554236, "learning_rate": 6.751587420638295e-06, "loss": 0.2744, "step": 11703 }, { "epoch": 0.8091254752851711, "grad_norm": 0.6655798260537364, "learning_rate": 6.7510631218146695e-06, "loss": 0.3556, "step": 11704 }, { "epoch": 0.8091946076736951, "grad_norm": 0.7064107667008225, "learning_rate": 6.750538801044591e-06, "loss": 0.3844, "step": 11705 }, { "epoch": 0.8092637400622191, "grad_norm": 0.8805495083130629, "learning_rate": 6.750014458334634e-06, "loss": 0.2875, "step": 11706 }, { "epoch": 0.8093328724507431, "grad_norm": 0.7006485196238952, "learning_rate": 6.7494900936913685e-06, "loss": 0.27, "step": 11707 }, { "epoch": 0.8094020048392672, "grad_norm": 0.6796163557054449, "learning_rate": 6.748965707121367e-06, "loss": 0.4409, "step": 11708 }, { "epoch": 0.8094711372277912, "grad_norm": 0.6600663338124464, "learning_rate": 6.748441298631201e-06, "loss": 0.2886, "step": 11709 }, { "epoch": 0.8095402696163152, "grad_norm": 0.5929616933165587, "learning_rate": 6.7479168682274444e-06, "loss": 0.286, "step": 11710 }, { "epoch": 0.8096094020048392, "grad_norm": 0.7582009782521659, "learning_rate": 6.7473924159166685e-06, "loss": 0.3533, "step": 11711 }, { "epoch": 0.8096785343933632, "grad_norm": 0.6387638456882819, "learning_rate": 6.7468679417054465e-06, "loss": 0.4451, "step": 11712 }, { "epoch": 0.8097476667818874, "grad_norm": 0.6124290382937896, "learning_rate": 6.746343445600353e-06, "loss": 0.261, "step": 11713 }, { "epoch": 0.8098167991704114, "grad_norm": 0.6225199606181573, "learning_rate": 6.745818927607961e-06, "loss": 0.2265, "step": 11714 }, { "epoch": 0.8098859315589354, "grad_norm": 0.7014068756168329, "learning_rate": 6.745294387734844e-06, "loss": 0.4009, "step": 11715 }, { "epoch": 0.8099550639474594, "grad_norm": 0.6696832386591566, "learning_rate": 6.744769825987576e-06, "loss": 0.3487, "step": 11716 }, { "epoch": 0.8100241963359834, "grad_norm": 0.6851540169730248, "learning_rate": 6.744245242372734e-06, "loss": 0.2855, "step": 11717 }, { "epoch": 0.8100933287245075, "grad_norm": 0.8437945709631653, "learning_rate": 6.743720636896889e-06, "loss": 0.3278, "step": 11718 }, { "epoch": 0.8101624611130315, "grad_norm": 0.8417434567319113, "learning_rate": 6.743196009566616e-06, "loss": 0.4126, "step": 11719 }, { "epoch": 0.8102315935015555, "grad_norm": 0.6683212282498626, "learning_rate": 6.742671360388494e-06, "loss": 0.3343, "step": 11720 }, { "epoch": 0.8103007258900795, "grad_norm": 0.6646107012533979, "learning_rate": 6.742146689369097e-06, "loss": 0.2897, "step": 11721 }, { "epoch": 0.8103698582786035, "grad_norm": 0.6830791419501799, "learning_rate": 6.741621996514999e-06, "loss": 0.3847, "step": 11722 }, { "epoch": 0.8104389906671275, "grad_norm": 0.6583081158470274, "learning_rate": 6.741097281832777e-06, "loss": 0.3148, "step": 11723 }, { "epoch": 0.8105081230556516, "grad_norm": 0.688453757866989, "learning_rate": 6.740572545329008e-06, "loss": 0.2386, "step": 11724 }, { "epoch": 0.8105772554441756, "grad_norm": 1.0150890230900147, "learning_rate": 6.74004778701027e-06, "loss": 0.316, "step": 11725 }, { "epoch": 0.8106463878326996, "grad_norm": 0.6506222260023857, "learning_rate": 6.739523006883137e-06, "loss": 0.3391, "step": 11726 }, { "epoch": 0.8107155202212236, "grad_norm": 0.6237010634388918, "learning_rate": 6.738998204954187e-06, "loss": 0.2628, "step": 11727 }, { "epoch": 0.8107846526097476, "grad_norm": 0.8413882183535732, "learning_rate": 6.738473381229997e-06, "loss": 0.287, "step": 11728 }, { "epoch": 0.8108537849982717, "grad_norm": 0.75325727429128, "learning_rate": 6.737948535717148e-06, "loss": 0.3339, "step": 11729 }, { "epoch": 0.8109229173867957, "grad_norm": 0.6362226603931416, "learning_rate": 6.737423668422214e-06, "loss": 0.3637, "step": 11730 }, { "epoch": 0.8109920497753197, "grad_norm": 0.6292502488354345, "learning_rate": 6.736898779351776e-06, "loss": 0.286, "step": 11731 }, { "epoch": 0.8110611821638437, "grad_norm": 0.6730294480112121, "learning_rate": 6.73637386851241e-06, "loss": 0.2992, "step": 11732 }, { "epoch": 0.8111303145523678, "grad_norm": 0.9166436141603677, "learning_rate": 6.735848935910699e-06, "loss": 0.4041, "step": 11733 }, { "epoch": 0.8111994469408919, "grad_norm": 0.678004130253949, "learning_rate": 6.735323981553216e-06, "loss": 0.2905, "step": 11734 }, { "epoch": 0.8112685793294159, "grad_norm": 0.6783553788675875, "learning_rate": 6.734799005446545e-06, "loss": 0.2741, "step": 11735 }, { "epoch": 0.8113377117179399, "grad_norm": 0.6354611280242137, "learning_rate": 6.734274007597264e-06, "loss": 0.3196, "step": 11736 }, { "epoch": 0.8114068441064639, "grad_norm": 0.6891716810426499, "learning_rate": 6.733748988011954e-06, "loss": 0.395, "step": 11737 }, { "epoch": 0.8114759764949879, "grad_norm": 0.6828472011886133, "learning_rate": 6.733223946697194e-06, "loss": 0.2883, "step": 11738 }, { "epoch": 0.811545108883512, "grad_norm": 0.8040356065694052, "learning_rate": 6.732698883659565e-06, "loss": 0.3689, "step": 11739 }, { "epoch": 0.811614241272036, "grad_norm": 0.6948702175198228, "learning_rate": 6.732173798905649e-06, "loss": 0.3918, "step": 11740 }, { "epoch": 0.81168337366056, "grad_norm": 0.7006565812488901, "learning_rate": 6.731648692442023e-06, "loss": 0.2583, "step": 11741 }, { "epoch": 0.811752506049084, "grad_norm": 0.6719219290229688, "learning_rate": 6.731123564275274e-06, "loss": 0.2602, "step": 11742 }, { "epoch": 0.811821638437608, "grad_norm": 0.6732357082788671, "learning_rate": 6.73059841441198e-06, "loss": 0.3286, "step": 11743 }, { "epoch": 0.811890770826132, "grad_norm": 0.634006106315409, "learning_rate": 6.73007324285872e-06, "loss": 0.3869, "step": 11744 }, { "epoch": 0.811959903214656, "grad_norm": 0.9530428052684946, "learning_rate": 6.729548049622082e-06, "loss": 0.2974, "step": 11745 }, { "epoch": 0.8120290356031801, "grad_norm": 0.6648678850065274, "learning_rate": 6.7290228347086454e-06, "loss": 0.3141, "step": 11746 }, { "epoch": 0.8120981679917041, "grad_norm": 0.6552019994784872, "learning_rate": 6.7284975981249924e-06, "loss": 0.395, "step": 11747 }, { "epoch": 0.8121673003802281, "grad_norm": 0.6396552425561721, "learning_rate": 6.727972339877706e-06, "loss": 0.332, "step": 11748 }, { "epoch": 0.8122364327687521, "grad_norm": 0.7262362849638619, "learning_rate": 6.727447059973372e-06, "loss": 0.3008, "step": 11749 }, { "epoch": 0.8123055651572761, "grad_norm": 0.7690191830925508, "learning_rate": 6.726921758418572e-06, "loss": 0.3231, "step": 11750 }, { "epoch": 0.8123746975458002, "grad_norm": 0.6274302188062483, "learning_rate": 6.726396435219888e-06, "loss": 0.3596, "step": 11751 }, { "epoch": 0.8124438299343242, "grad_norm": 0.6788305275220406, "learning_rate": 6.725871090383905e-06, "loss": 0.28, "step": 11752 }, { "epoch": 0.8125129623228483, "grad_norm": 0.96638935834202, "learning_rate": 6.725345723917209e-06, "loss": 0.2576, "step": 11753 }, { "epoch": 0.8125820947113723, "grad_norm": 0.6543336073549441, "learning_rate": 6.724820335826383e-06, "loss": 0.3584, "step": 11754 }, { "epoch": 0.8126512270998963, "grad_norm": 0.7197292064043701, "learning_rate": 6.7242949261180135e-06, "loss": 0.3988, "step": 11755 }, { "epoch": 0.8127203594884204, "grad_norm": 0.903204844194296, "learning_rate": 6.723769494798683e-06, "loss": 0.2635, "step": 11756 }, { "epoch": 0.8127894918769444, "grad_norm": 0.6558432094280439, "learning_rate": 6.7232440418749785e-06, "loss": 0.2824, "step": 11757 }, { "epoch": 0.8128586242654684, "grad_norm": 0.7264405749706339, "learning_rate": 6.722718567353485e-06, "loss": 0.4226, "step": 11758 }, { "epoch": 0.8129277566539924, "grad_norm": 0.7149246540730105, "learning_rate": 6.7221930712407875e-06, "loss": 0.396, "step": 11759 }, { "epoch": 0.8129968890425164, "grad_norm": 0.8643851037559769, "learning_rate": 6.721667553543476e-06, "loss": 0.3124, "step": 11760 }, { "epoch": 0.8130660214310405, "grad_norm": 0.6460414652179416, "learning_rate": 6.7211420142681326e-06, "loss": 0.3473, "step": 11761 }, { "epoch": 0.8131351538195645, "grad_norm": 0.6695449954703256, "learning_rate": 6.720616453421345e-06, "loss": 0.4146, "step": 11762 }, { "epoch": 0.8132042862080885, "grad_norm": 0.6808711564125072, "learning_rate": 6.720090871009701e-06, "loss": 0.2773, "step": 11763 }, { "epoch": 0.8132734185966125, "grad_norm": 0.6950132922066434, "learning_rate": 6.719565267039788e-06, "loss": 0.3822, "step": 11764 }, { "epoch": 0.8133425509851365, "grad_norm": 0.680850202196793, "learning_rate": 6.719039641518192e-06, "loss": 0.404, "step": 11765 }, { "epoch": 0.8134116833736605, "grad_norm": 0.658342184519434, "learning_rate": 6.718513994451503e-06, "loss": 0.2951, "step": 11766 }, { "epoch": 0.8134808157621846, "grad_norm": 0.7201993848332041, "learning_rate": 6.7179883258463085e-06, "loss": 0.2684, "step": 11767 }, { "epoch": 0.8135499481507086, "grad_norm": 0.7656902565986251, "learning_rate": 6.717462635709196e-06, "loss": 0.3221, "step": 11768 }, { "epoch": 0.8136190805392326, "grad_norm": 0.6623280496921389, "learning_rate": 6.716936924046753e-06, "loss": 0.3785, "step": 11769 }, { "epoch": 0.8136882129277566, "grad_norm": 0.708660789939908, "learning_rate": 6.71641119086557e-06, "loss": 0.3268, "step": 11770 }, { "epoch": 0.8137573453162806, "grad_norm": 0.7209969275790176, "learning_rate": 6.715885436172236e-06, "loss": 0.2911, "step": 11771 }, { "epoch": 0.8138264777048047, "grad_norm": 0.7366151622453405, "learning_rate": 6.7153596599733395e-06, "loss": 0.4163, "step": 11772 }, { "epoch": 0.8138956100933287, "grad_norm": 0.7486022325724141, "learning_rate": 6.714833862275473e-06, "loss": 0.4164, "step": 11773 }, { "epoch": 0.8139647424818528, "grad_norm": 0.6239760677077982, "learning_rate": 6.714308043085222e-06, "loss": 0.2523, "step": 11774 }, { "epoch": 0.8140338748703768, "grad_norm": 0.7122259184229836, "learning_rate": 6.713782202409182e-06, "loss": 0.2644, "step": 11775 }, { "epoch": 0.8141030072589008, "grad_norm": 0.6173352802753446, "learning_rate": 6.7132563402539364e-06, "loss": 0.3892, "step": 11776 }, { "epoch": 0.8141721396474249, "grad_norm": 0.7312277769791578, "learning_rate": 6.712730456626083e-06, "loss": 0.3235, "step": 11777 }, { "epoch": 0.8142412720359489, "grad_norm": 0.6208599965392368, "learning_rate": 6.7122045515322086e-06, "loss": 0.274, "step": 11778 }, { "epoch": 0.8143104044244729, "grad_norm": 0.6541677224393091, "learning_rate": 6.711678624978907e-06, "loss": 0.3686, "step": 11779 }, { "epoch": 0.8143795368129969, "grad_norm": 0.6937308217601298, "learning_rate": 6.711152676972767e-06, "loss": 0.3597, "step": 11780 }, { "epoch": 0.8144486692015209, "grad_norm": 0.6634536567746798, "learning_rate": 6.7106267075203825e-06, "loss": 0.2936, "step": 11781 }, { "epoch": 0.8145178015900449, "grad_norm": 0.6910907854101017, "learning_rate": 6.710100716628345e-06, "loss": 0.3403, "step": 11782 }, { "epoch": 0.814586933978569, "grad_norm": 0.6779057048325265, "learning_rate": 6.709574704303245e-06, "loss": 0.4015, "step": 11783 }, { "epoch": 0.814656066367093, "grad_norm": 0.7110374253287606, "learning_rate": 6.709048670551679e-06, "loss": 0.3235, "step": 11784 }, { "epoch": 0.814725198755617, "grad_norm": 0.9361020495563696, "learning_rate": 6.708522615380239e-06, "loss": 0.2591, "step": 11785 }, { "epoch": 0.814794331144141, "grad_norm": 0.6988507206295277, "learning_rate": 6.707996538795515e-06, "loss": 0.3832, "step": 11786 }, { "epoch": 0.814863463532665, "grad_norm": 0.6470184446173344, "learning_rate": 6.707470440804103e-06, "loss": 0.3841, "step": 11787 }, { "epoch": 0.814932595921189, "grad_norm": 0.6413165444212496, "learning_rate": 6.706944321412596e-06, "loss": 0.3199, "step": 11788 }, { "epoch": 0.8150017283097131, "grad_norm": 1.668062482236563, "learning_rate": 6.706418180627587e-06, "loss": 0.2854, "step": 11789 }, { "epoch": 0.8150708606982371, "grad_norm": 0.6484050520113881, "learning_rate": 6.705892018455673e-06, "loss": 0.3733, "step": 11790 }, { "epoch": 0.8151399930867611, "grad_norm": 0.6803333736947422, "learning_rate": 6.705365834903447e-06, "loss": 0.2814, "step": 11791 }, { "epoch": 0.8152091254752851, "grad_norm": 0.6427574658344997, "learning_rate": 6.704839629977503e-06, "loss": 0.3532, "step": 11792 }, { "epoch": 0.8152782578638091, "grad_norm": 0.70719564367468, "learning_rate": 6.704313403684437e-06, "loss": 0.3315, "step": 11793 }, { "epoch": 0.8153473902523333, "grad_norm": 0.6293299114580237, "learning_rate": 6.7037871560308435e-06, "loss": 0.3496, "step": 11794 }, { "epoch": 0.8154165226408573, "grad_norm": 0.6318171079075163, "learning_rate": 6.703260887023319e-06, "loss": 0.3179, "step": 11795 }, { "epoch": 0.8154856550293813, "grad_norm": 0.7853440528960912, "learning_rate": 6.702734596668458e-06, "loss": 0.2692, "step": 11796 }, { "epoch": 0.8155547874179053, "grad_norm": 0.869063022595652, "learning_rate": 6.702208284972859e-06, "loss": 0.4415, "step": 11797 }, { "epoch": 0.8156239198064293, "grad_norm": 0.6009959755026463, "learning_rate": 6.7016819519431144e-06, "loss": 0.3128, "step": 11798 }, { "epoch": 0.8156930521949534, "grad_norm": 0.6254199085965134, "learning_rate": 6.701155597585826e-06, "loss": 0.2507, "step": 11799 }, { "epoch": 0.8157621845834774, "grad_norm": 0.6780679153562734, "learning_rate": 6.700629221907586e-06, "loss": 0.3296, "step": 11800 }, { "epoch": 0.8158313169720014, "grad_norm": 0.6383796591160618, "learning_rate": 6.7001028249149935e-06, "loss": 0.3779, "step": 11801 }, { "epoch": 0.8159004493605254, "grad_norm": 0.71249477019985, "learning_rate": 6.699576406614647e-06, "loss": 0.2467, "step": 11802 }, { "epoch": 0.8159695817490494, "grad_norm": 0.5998870497418781, "learning_rate": 6.6990499670131445e-06, "loss": 0.2971, "step": 11803 }, { "epoch": 0.8160387141375735, "grad_norm": 0.6555260888130167, "learning_rate": 6.698523506117082e-06, "loss": 0.3414, "step": 11804 }, { "epoch": 0.8161078465260975, "grad_norm": 0.6981993537415235, "learning_rate": 6.697997023933058e-06, "loss": 0.3613, "step": 11805 }, { "epoch": 0.8161769789146215, "grad_norm": 0.6801378572564697, "learning_rate": 6.697470520467672e-06, "loss": 0.3362, "step": 11806 }, { "epoch": 0.8162461113031455, "grad_norm": 0.710841899183791, "learning_rate": 6.6969439957275215e-06, "loss": 0.3076, "step": 11807 }, { "epoch": 0.8163152436916695, "grad_norm": 0.7074534122227562, "learning_rate": 6.696417449719207e-06, "loss": 0.4092, "step": 11808 }, { "epoch": 0.8163843760801935, "grad_norm": 0.7337834159784569, "learning_rate": 6.6958908824493276e-06, "loss": 0.3406, "step": 11809 }, { "epoch": 0.8164535084687176, "grad_norm": 0.6673632805162643, "learning_rate": 6.695364293924481e-06, "loss": 0.285, "step": 11810 }, { "epoch": 0.8165226408572416, "grad_norm": 0.751844208670645, "learning_rate": 6.69483768415127e-06, "loss": 0.388, "step": 11811 }, { "epoch": 0.8165917732457656, "grad_norm": 0.710863373271739, "learning_rate": 6.6943110531362924e-06, "loss": 0.3383, "step": 11812 }, { "epoch": 0.8166609056342896, "grad_norm": 0.656381719604319, "learning_rate": 6.693784400886149e-06, "loss": 0.3167, "step": 11813 }, { "epoch": 0.8167300380228137, "grad_norm": 0.5994939424292057, "learning_rate": 6.693257727407442e-06, "loss": 0.3174, "step": 11814 }, { "epoch": 0.8167991704113378, "grad_norm": 1.5849550151836331, "learning_rate": 6.69273103270677e-06, "loss": 0.4676, "step": 11815 }, { "epoch": 0.8168683027998618, "grad_norm": 0.6185979123332778, "learning_rate": 6.692204316790736e-06, "loss": 0.286, "step": 11816 }, { "epoch": 0.8169374351883858, "grad_norm": 0.628179661422065, "learning_rate": 6.691677579665941e-06, "loss": 0.2816, "step": 11817 }, { "epoch": 0.8170065675769098, "grad_norm": 0.9401065444980233, "learning_rate": 6.691150821338984e-06, "loss": 0.2998, "step": 11818 }, { "epoch": 0.8170756999654338, "grad_norm": 0.7348039157198889, "learning_rate": 6.690624041816471e-06, "loss": 0.4244, "step": 11819 }, { "epoch": 0.8171448323539579, "grad_norm": 0.745910959736205, "learning_rate": 6.690097241105004e-06, "loss": 0.2604, "step": 11820 }, { "epoch": 0.8172139647424819, "grad_norm": 0.6667046640700823, "learning_rate": 6.689570419211185e-06, "loss": 0.3141, "step": 11821 }, { "epoch": 0.8172830971310059, "grad_norm": 0.67808676332171, "learning_rate": 6.689043576141612e-06, "loss": 0.4182, "step": 11822 }, { "epoch": 0.8173522295195299, "grad_norm": 0.685053839758732, "learning_rate": 6.6885167119028945e-06, "loss": 0.3537, "step": 11823 }, { "epoch": 0.8174213619080539, "grad_norm": 0.5753144075863545, "learning_rate": 6.687989826501634e-06, "loss": 0.2469, "step": 11824 }, { "epoch": 0.8174904942965779, "grad_norm": 0.6957965884951192, "learning_rate": 6.687462919944432e-06, "loss": 0.3818, "step": 11825 }, { "epoch": 0.817559626685102, "grad_norm": 0.6391980812368682, "learning_rate": 6.686935992237894e-06, "loss": 0.4223, "step": 11826 }, { "epoch": 0.817628759073626, "grad_norm": 0.7617106474990284, "learning_rate": 6.6864090433886246e-06, "loss": 0.2984, "step": 11827 }, { "epoch": 0.81769789146215, "grad_norm": 0.6884080783810851, "learning_rate": 6.685882073403227e-06, "loss": 0.2904, "step": 11828 }, { "epoch": 0.817767023850674, "grad_norm": 0.6554798002715806, "learning_rate": 6.685355082288305e-06, "loss": 0.3257, "step": 11829 }, { "epoch": 0.817836156239198, "grad_norm": 0.652461090008212, "learning_rate": 6.6848280700504665e-06, "loss": 0.3549, "step": 11830 }, { "epoch": 0.817905288627722, "grad_norm": 0.6411567605160893, "learning_rate": 6.684301036696313e-06, "loss": 0.3139, "step": 11831 }, { "epoch": 0.8179744210162461, "grad_norm": 0.6217166972125959, "learning_rate": 6.683773982232453e-06, "loss": 0.2725, "step": 11832 }, { "epoch": 0.8180435534047701, "grad_norm": 0.639858427350802, "learning_rate": 6.68324690666549e-06, "loss": 0.3915, "step": 11833 }, { "epoch": 0.8181126857932941, "grad_norm": 0.6417949475839357, "learning_rate": 6.682719810002033e-06, "loss": 0.3131, "step": 11834 }, { "epoch": 0.8181818181818182, "grad_norm": 0.6211126729622831, "learning_rate": 6.682192692248685e-06, "loss": 0.2485, "step": 11835 }, { "epoch": 0.8182509505703423, "grad_norm": 0.6902208590864455, "learning_rate": 6.6816655534120525e-06, "loss": 0.3384, "step": 11836 }, { "epoch": 0.8183200829588663, "grad_norm": 0.9759301741476072, "learning_rate": 6.681138393498744e-06, "loss": 0.3752, "step": 11837 }, { "epoch": 0.8183892153473903, "grad_norm": 0.6174060458549797, "learning_rate": 6.680611212515367e-06, "loss": 0.2512, "step": 11838 }, { "epoch": 0.8184583477359143, "grad_norm": 0.6729829730997287, "learning_rate": 6.680084010468526e-06, "loss": 0.3179, "step": 11839 }, { "epoch": 0.8185274801244383, "grad_norm": 0.6506148318238286, "learning_rate": 6.679556787364831e-06, "loss": 0.3916, "step": 11840 }, { "epoch": 0.8185966125129623, "grad_norm": 0.7064321065863616, "learning_rate": 6.679029543210888e-06, "loss": 0.4046, "step": 11841 }, { "epoch": 0.8186657449014864, "grad_norm": 0.966879486696029, "learning_rate": 6.678502278013308e-06, "loss": 0.2863, "step": 11842 }, { "epoch": 0.8187348772900104, "grad_norm": 0.7221405800742825, "learning_rate": 6.677974991778695e-06, "loss": 0.3058, "step": 11843 }, { "epoch": 0.8188040096785344, "grad_norm": 0.655000033250203, "learning_rate": 6.677447684513662e-06, "loss": 0.3828, "step": 11844 }, { "epoch": 0.8188731420670584, "grad_norm": 0.7707882800622714, "learning_rate": 6.676920356224816e-06, "loss": 0.3033, "step": 11845 }, { "epoch": 0.8189422744555824, "grad_norm": 0.7571866866914582, "learning_rate": 6.676393006918764e-06, "loss": 0.288, "step": 11846 }, { "epoch": 0.8190114068441064, "grad_norm": 0.99748008161519, "learning_rate": 6.6758656366021195e-06, "loss": 0.4082, "step": 11847 }, { "epoch": 0.8190805392326305, "grad_norm": 0.6436330592050328, "learning_rate": 6.6753382452814885e-06, "loss": 0.2954, "step": 11848 }, { "epoch": 0.8191496716211545, "grad_norm": 0.5976904914492726, "learning_rate": 6.674810832963483e-06, "loss": 0.3025, "step": 11849 }, { "epoch": 0.8192188040096785, "grad_norm": 8.635392658513613, "learning_rate": 6.674283399654712e-06, "loss": 0.3005, "step": 11850 }, { "epoch": 0.8192879363982025, "grad_norm": 0.6781564767908754, "learning_rate": 6.6737559453617865e-06, "loss": 0.3532, "step": 11851 }, { "epoch": 0.8193570687867265, "grad_norm": 0.7015070882887071, "learning_rate": 6.673228470091318e-06, "loss": 0.2716, "step": 11852 }, { "epoch": 0.8194262011752506, "grad_norm": 0.6095570366866646, "learning_rate": 6.672700973849915e-06, "loss": 0.2908, "step": 11853 }, { "epoch": 0.8194953335637746, "grad_norm": 0.691047328903637, "learning_rate": 6.672173456644191e-06, "loss": 0.3656, "step": 11854 }, { "epoch": 0.8195644659522987, "grad_norm": 0.731516038023684, "learning_rate": 6.671645918480758e-06, "loss": 0.335, "step": 11855 }, { "epoch": 0.8196335983408227, "grad_norm": 0.7515187018223402, "learning_rate": 6.6711183593662255e-06, "loss": 0.2502, "step": 11856 }, { "epoch": 0.8197027307293467, "grad_norm": 0.6663584672647351, "learning_rate": 6.670590779307207e-06, "loss": 0.2692, "step": 11857 }, { "epoch": 0.8197718631178708, "grad_norm": 0.6379903741223473, "learning_rate": 6.670063178310313e-06, "loss": 0.3762, "step": 11858 }, { "epoch": 0.8198409955063948, "grad_norm": 0.6970011898576615, "learning_rate": 6.669535556382159e-06, "loss": 0.2906, "step": 11859 }, { "epoch": 0.8199101278949188, "grad_norm": 0.6161462314983556, "learning_rate": 6.669007913529357e-06, "loss": 0.2862, "step": 11860 }, { "epoch": 0.8199792602834428, "grad_norm": 0.6867952266235992, "learning_rate": 6.668480249758518e-06, "loss": 0.3806, "step": 11861 }, { "epoch": 0.8200483926719668, "grad_norm": 0.8614122250130138, "learning_rate": 6.667952565076257e-06, "loss": 0.3773, "step": 11862 }, { "epoch": 0.8201175250604908, "grad_norm": 1.7836190236397975, "learning_rate": 6.6674248594891865e-06, "loss": 0.2523, "step": 11863 }, { "epoch": 0.8201866574490149, "grad_norm": 0.7136469743074266, "learning_rate": 6.666897133003922e-06, "loss": 0.2832, "step": 11864 }, { "epoch": 0.8202557898375389, "grad_norm": 0.7085138854716022, "learning_rate": 6.666369385627076e-06, "loss": 0.408, "step": 11865 }, { "epoch": 0.8203249222260629, "grad_norm": 0.6717236095857806, "learning_rate": 6.665841617365264e-06, "loss": 0.3229, "step": 11866 }, { "epoch": 0.8203940546145869, "grad_norm": 0.6220601397715283, "learning_rate": 6.665313828225099e-06, "loss": 0.2862, "step": 11867 }, { "epoch": 0.8204631870031109, "grad_norm": 0.6599840375176892, "learning_rate": 6.664786018213198e-06, "loss": 0.2949, "step": 11868 }, { "epoch": 0.820532319391635, "grad_norm": 0.6779027371438445, "learning_rate": 6.6642581873361754e-06, "loss": 0.4103, "step": 11869 }, { "epoch": 0.820601451780159, "grad_norm": 0.6337668578147001, "learning_rate": 6.663730335600647e-06, "loss": 0.2867, "step": 11870 }, { "epoch": 0.820670584168683, "grad_norm": 0.8327628604114335, "learning_rate": 6.6632024630132266e-06, "loss": 0.2737, "step": 11871 }, { "epoch": 0.820739716557207, "grad_norm": 0.7588832393387869, "learning_rate": 6.662674569580532e-06, "loss": 0.3551, "step": 11872 }, { "epoch": 0.820808848945731, "grad_norm": 0.7452749277643829, "learning_rate": 6.6621466553091785e-06, "loss": 0.4124, "step": 11873 }, { "epoch": 0.820877981334255, "grad_norm": 0.7098340076022253, "learning_rate": 6.661618720205783e-06, "loss": 0.299, "step": 11874 }, { "epoch": 0.8209471137227792, "grad_norm": 0.7439838434425058, "learning_rate": 6.661090764276963e-06, "loss": 0.3873, "step": 11875 }, { "epoch": 0.8210162461113032, "grad_norm": 0.6891369442550876, "learning_rate": 6.660562787529335e-06, "loss": 0.3979, "step": 11876 }, { "epoch": 0.8210853784998272, "grad_norm": 0.7089274317179074, "learning_rate": 6.6600347899695136e-06, "loss": 0.2811, "step": 11877 }, { "epoch": 0.8211545108883512, "grad_norm": 0.7345778182632811, "learning_rate": 6.6595067716041205e-06, "loss": 0.2724, "step": 11878 }, { "epoch": 0.8212236432768752, "grad_norm": 0.7401705460217913, "learning_rate": 6.658978732439771e-06, "loss": 0.3961, "step": 11879 }, { "epoch": 0.8212927756653993, "grad_norm": 0.7038953442821962, "learning_rate": 6.658450672483084e-06, "loss": 0.3689, "step": 11880 }, { "epoch": 0.8213619080539233, "grad_norm": 0.6311333372969253, "learning_rate": 6.657922591740677e-06, "loss": 0.292, "step": 11881 }, { "epoch": 0.8214310404424473, "grad_norm": 0.6603324508526185, "learning_rate": 6.6573944902191665e-06, "loss": 0.3228, "step": 11882 }, { "epoch": 0.8215001728309713, "grad_norm": 0.7189643542446533, "learning_rate": 6.656866367925176e-06, "loss": 0.455, "step": 11883 }, { "epoch": 0.8215693052194953, "grad_norm": 0.6469964783014251, "learning_rate": 6.6563382248653245e-06, "loss": 0.3464, "step": 11884 }, { "epoch": 0.8216384376080194, "grad_norm": 1.3310753761233587, "learning_rate": 6.6558100610462275e-06, "loss": 0.2586, "step": 11885 }, { "epoch": 0.8217075699965434, "grad_norm": 0.7583177435294083, "learning_rate": 6.655281876474506e-06, "loss": 0.4294, "step": 11886 }, { "epoch": 0.8217767023850674, "grad_norm": 0.7079869711483463, "learning_rate": 6.6547536711567794e-06, "loss": 0.3901, "step": 11887 }, { "epoch": 0.8218458347735914, "grad_norm": 0.6898764880746371, "learning_rate": 6.65422544509967e-06, "loss": 0.2738, "step": 11888 }, { "epoch": 0.8219149671621154, "grad_norm": 0.7697430507062527, "learning_rate": 6.653697198309795e-06, "loss": 0.3221, "step": 11889 }, { "epoch": 0.8219840995506394, "grad_norm": 0.7411366314902649, "learning_rate": 6.653168930793779e-06, "loss": 0.4724, "step": 11890 }, { "epoch": 0.8220532319391635, "grad_norm": 0.632105328318011, "learning_rate": 6.65264064255824e-06, "loss": 0.3292, "step": 11891 }, { "epoch": 0.8221223643276875, "grad_norm": 0.6163585120235813, "learning_rate": 6.652112333609799e-06, "loss": 0.2674, "step": 11892 }, { "epoch": 0.8221914967162115, "grad_norm": 0.6763017229365913, "learning_rate": 6.6515840039550784e-06, "loss": 0.3205, "step": 11893 }, { "epoch": 0.8222606291047355, "grad_norm": 0.6991876448752994, "learning_rate": 6.651055653600701e-06, "loss": 0.3627, "step": 11894 }, { "epoch": 0.8223297614932595, "grad_norm": 0.6148488542722588, "learning_rate": 6.650527282553286e-06, "loss": 0.2453, "step": 11895 }, { "epoch": 0.8223988938817837, "grad_norm": 0.6601786414383227, "learning_rate": 6.649998890819458e-06, "loss": 0.3231, "step": 11896 }, { "epoch": 0.8224680262703077, "grad_norm": 0.7169934146904987, "learning_rate": 6.6494704784058375e-06, "loss": 0.4129, "step": 11897 }, { "epoch": 0.8225371586588317, "grad_norm": 0.614093944238276, "learning_rate": 6.648942045319049e-06, "loss": 0.2864, "step": 11898 }, { "epoch": 0.8226062910473557, "grad_norm": 0.6554208399478559, "learning_rate": 6.648413591565714e-06, "loss": 0.2693, "step": 11899 }, { "epoch": 0.8226754234358797, "grad_norm": 0.8282054162775496, "learning_rate": 6.647885117152456e-06, "loss": 0.2666, "step": 11900 }, { "epoch": 0.8227445558244038, "grad_norm": 0.7436002195194876, "learning_rate": 6.6473566220859e-06, "loss": 0.3854, "step": 11901 }, { "epoch": 0.8228136882129278, "grad_norm": 0.860309414376669, "learning_rate": 6.646828106372668e-06, "loss": 0.2831, "step": 11902 }, { "epoch": 0.8228828206014518, "grad_norm": 0.641548081262971, "learning_rate": 6.6462995700193845e-06, "loss": 0.2211, "step": 11903 }, { "epoch": 0.8229519529899758, "grad_norm": 0.664622688431914, "learning_rate": 6.645771013032673e-06, "loss": 0.3814, "step": 11904 }, { "epoch": 0.8230210853784998, "grad_norm": 1.1173312662866473, "learning_rate": 6.64524243541916e-06, "loss": 0.3675, "step": 11905 }, { "epoch": 0.8230902177670238, "grad_norm": 0.6783905450125832, "learning_rate": 6.64471383718547e-06, "loss": 0.2937, "step": 11906 }, { "epoch": 0.8231593501555479, "grad_norm": 0.7039743420548029, "learning_rate": 6.644185218338225e-06, "loss": 0.2931, "step": 11907 }, { "epoch": 0.8232284825440719, "grad_norm": 0.7644974863032555, "learning_rate": 6.643656578884054e-06, "loss": 0.4068, "step": 11908 }, { "epoch": 0.8232976149325959, "grad_norm": 0.6229674576176576, "learning_rate": 6.6431279188295805e-06, "loss": 0.2823, "step": 11909 }, { "epoch": 0.8233667473211199, "grad_norm": 0.6719611427460419, "learning_rate": 6.642599238181431e-06, "loss": 0.2554, "step": 11910 }, { "epoch": 0.8234358797096439, "grad_norm": 0.7024747565449967, "learning_rate": 6.642070536946232e-06, "loss": 0.4098, "step": 11911 }, { "epoch": 0.823505012098168, "grad_norm": 0.7292939736940961, "learning_rate": 6.641541815130608e-06, "loss": 0.3881, "step": 11912 }, { "epoch": 0.823574144486692, "grad_norm": 0.6414796875981733, "learning_rate": 6.641013072741187e-06, "loss": 0.295, "step": 11913 }, { "epoch": 0.823643276875216, "grad_norm": 0.7473401508094967, "learning_rate": 6.640484309784596e-06, "loss": 0.3351, "step": 11914 }, { "epoch": 0.82371240926374, "grad_norm": 0.7216301748981486, "learning_rate": 6.639955526267462e-06, "loss": 0.3851, "step": 11915 }, { "epoch": 0.8237815416522641, "grad_norm": 0.6538891214650281, "learning_rate": 6.639426722196411e-06, "loss": 0.3528, "step": 11916 }, { "epoch": 0.8238506740407882, "grad_norm": 0.6161853017526523, "learning_rate": 6.638897897578074e-06, "loss": 0.2877, "step": 11917 }, { "epoch": 0.8239198064293122, "grad_norm": 0.7188518398771732, "learning_rate": 6.638369052419074e-06, "loss": 0.3417, "step": 11918 }, { "epoch": 0.8239889388178362, "grad_norm": 0.6675298000328606, "learning_rate": 6.637840186726044e-06, "loss": 0.3587, "step": 11919 }, { "epoch": 0.8240580712063602, "grad_norm": 0.6078757517983696, "learning_rate": 6.637311300505609e-06, "loss": 0.2566, "step": 11920 }, { "epoch": 0.8241272035948842, "grad_norm": 0.7482433246992888, "learning_rate": 6.636782393764399e-06, "loss": 0.2482, "step": 11921 }, { "epoch": 0.8241963359834082, "grad_norm": 0.7953144092833513, "learning_rate": 6.636253466509042e-06, "loss": 0.3896, "step": 11922 }, { "epoch": 0.8242654683719323, "grad_norm": 0.8026687908405895, "learning_rate": 6.635724518746168e-06, "loss": 0.4423, "step": 11923 }, { "epoch": 0.8243346007604563, "grad_norm": 0.6836044989564903, "learning_rate": 6.635195550482407e-06, "loss": 0.2727, "step": 11924 }, { "epoch": 0.8244037331489803, "grad_norm": 1.060973324605638, "learning_rate": 6.634666561724386e-06, "loss": 0.3015, "step": 11925 }, { "epoch": 0.8244728655375043, "grad_norm": 0.8157545955922186, "learning_rate": 6.634137552478739e-06, "loss": 0.3521, "step": 11926 }, { "epoch": 0.8245419979260283, "grad_norm": 0.6781700453933497, "learning_rate": 6.633608522752093e-06, "loss": 0.2965, "step": 11927 }, { "epoch": 0.8246111303145524, "grad_norm": 0.8429814721589532, "learning_rate": 6.633079472551081e-06, "loss": 0.2962, "step": 11928 }, { "epoch": 0.8246802627030764, "grad_norm": 0.7364092672459394, "learning_rate": 6.6325504018823294e-06, "loss": 0.3677, "step": 11929 }, { "epoch": 0.8247493950916004, "grad_norm": 0.6837027974253223, "learning_rate": 6.632021310752474e-06, "loss": 0.3999, "step": 11930 }, { "epoch": 0.8248185274801244, "grad_norm": 0.7658381886091906, "learning_rate": 6.631492199168144e-06, "loss": 0.2447, "step": 11931 }, { "epoch": 0.8248876598686484, "grad_norm": 0.6848758116170113, "learning_rate": 6.63096306713597e-06, "loss": 0.3745, "step": 11932 }, { "epoch": 0.8249567922571724, "grad_norm": 0.7026683461052005, "learning_rate": 6.630433914662584e-06, "loss": 0.4255, "step": 11933 }, { "epoch": 0.8250259246456965, "grad_norm": 0.6008842978627804, "learning_rate": 6.62990474175462e-06, "loss": 0.2654, "step": 11934 }, { "epoch": 0.8250950570342205, "grad_norm": 0.6100335951026089, "learning_rate": 6.629375548418708e-06, "loss": 0.303, "step": 11935 }, { "epoch": 0.8251641894227446, "grad_norm": 0.6770918382923332, "learning_rate": 6.628846334661481e-06, "loss": 0.3523, "step": 11936 }, { "epoch": 0.8252333218112686, "grad_norm": 0.6988396703412999, "learning_rate": 6.628317100489572e-06, "loss": 0.3481, "step": 11937 }, { "epoch": 0.8253024541997926, "grad_norm": 0.5994288095987624, "learning_rate": 6.6277878459096144e-06, "loss": 0.3212, "step": 11938 }, { "epoch": 0.8253715865883167, "grad_norm": 0.6873897763539102, "learning_rate": 6.6272585709282426e-06, "loss": 0.3373, "step": 11939 }, { "epoch": 0.8254407189768407, "grad_norm": 0.739729881483052, "learning_rate": 6.626729275552085e-06, "loss": 0.4618, "step": 11940 }, { "epoch": 0.8255098513653647, "grad_norm": 0.6384695532173548, "learning_rate": 6.626199959787783e-06, "loss": 0.3506, "step": 11941 }, { "epoch": 0.8255789837538887, "grad_norm": 0.6245840371243431, "learning_rate": 6.6256706236419635e-06, "loss": 0.2176, "step": 11942 }, { "epoch": 0.8256481161424127, "grad_norm": 0.6632179118157198, "learning_rate": 6.625141267121265e-06, "loss": 0.3738, "step": 11943 }, { "epoch": 0.8257172485309368, "grad_norm": 0.7695435216272157, "learning_rate": 6.624611890232321e-06, "loss": 0.4457, "step": 11944 }, { "epoch": 0.8257863809194608, "grad_norm": 0.7798252151229482, "learning_rate": 6.624082492981767e-06, "loss": 0.2881, "step": 11945 }, { "epoch": 0.8258555133079848, "grad_norm": 0.6170056439401997, "learning_rate": 6.623553075376235e-06, "loss": 0.2778, "step": 11946 }, { "epoch": 0.8259246456965088, "grad_norm": 0.6935513058738448, "learning_rate": 6.6230236374223645e-06, "loss": 0.4332, "step": 11947 }, { "epoch": 0.8259937780850328, "grad_norm": 0.6384246947674582, "learning_rate": 6.622494179126788e-06, "loss": 0.34, "step": 11948 }, { "epoch": 0.8260629104735568, "grad_norm": 0.7156203320361446, "learning_rate": 6.621964700496144e-06, "loss": 0.2823, "step": 11949 }, { "epoch": 0.8261320428620809, "grad_norm": 0.6084225305890244, "learning_rate": 6.6214352015370646e-06, "loss": 0.2808, "step": 11950 }, { "epoch": 0.8262011752506049, "grad_norm": 0.671650549651987, "learning_rate": 6.62090568225619e-06, "loss": 0.4285, "step": 11951 }, { "epoch": 0.8262703076391289, "grad_norm": 0.7543231229355972, "learning_rate": 6.620376142660156e-06, "loss": 0.2648, "step": 11952 }, { "epoch": 0.8263394400276529, "grad_norm": 0.6714434625349759, "learning_rate": 6.619846582755597e-06, "loss": 0.3328, "step": 11953 }, { "epoch": 0.8264085724161769, "grad_norm": 0.6787084511740298, "learning_rate": 6.6193170025491526e-06, "loss": 0.4169, "step": 11954 }, { "epoch": 0.826477704804701, "grad_norm": 0.7142447300478598, "learning_rate": 6.61878740204746e-06, "loss": 0.3393, "step": 11955 }, { "epoch": 0.8265468371932251, "grad_norm": 0.7010545391553448, "learning_rate": 6.618257781257156e-06, "loss": 0.2796, "step": 11956 }, { "epoch": 0.8266159695817491, "grad_norm": 0.6534625908481696, "learning_rate": 6.617728140184877e-06, "loss": 0.3075, "step": 11957 }, { "epoch": 0.8266851019702731, "grad_norm": 0.646454990367996, "learning_rate": 6.617198478837265e-06, "loss": 0.405, "step": 11958 }, { "epoch": 0.8267542343587971, "grad_norm": 0.6484964423683394, "learning_rate": 6.616668797220955e-06, "loss": 0.2935, "step": 11959 }, { "epoch": 0.8268233667473212, "grad_norm": 0.7124718875674226, "learning_rate": 6.616139095342586e-06, "loss": 0.2793, "step": 11960 }, { "epoch": 0.8268924991358452, "grad_norm": 0.9092979067403659, "learning_rate": 6.615609373208798e-06, "loss": 0.3895, "step": 11961 }, { "epoch": 0.8269616315243692, "grad_norm": 0.6633298984300597, "learning_rate": 6.615079630826231e-06, "loss": 0.3614, "step": 11962 }, { "epoch": 0.8270307639128932, "grad_norm": 0.7256010421591871, "learning_rate": 6.614549868201522e-06, "loss": 0.3642, "step": 11963 }, { "epoch": 0.8270998963014172, "grad_norm": 0.7338833240448318, "learning_rate": 6.614020085341311e-06, "loss": 0.2702, "step": 11964 }, { "epoch": 0.8271690286899412, "grad_norm": 0.7163890604773331, "learning_rate": 6.613490282252239e-06, "loss": 0.37, "step": 11965 }, { "epoch": 0.8272381610784653, "grad_norm": 0.6828779643539727, "learning_rate": 6.612960458940947e-06, "loss": 0.3733, "step": 11966 }, { "epoch": 0.8273072934669893, "grad_norm": 0.6434304991627847, "learning_rate": 6.612430615414074e-06, "loss": 0.3011, "step": 11967 }, { "epoch": 0.8273764258555133, "grad_norm": 0.7035374243301464, "learning_rate": 6.611900751678258e-06, "loss": 0.296, "step": 11968 }, { "epoch": 0.8274455582440373, "grad_norm": 0.7181355296364339, "learning_rate": 6.611370867740145e-06, "loss": 0.4385, "step": 11969 }, { "epoch": 0.8275146906325613, "grad_norm": 0.6906610637384988, "learning_rate": 6.610840963606374e-06, "loss": 0.3198, "step": 11970 }, { "epoch": 0.8275838230210854, "grad_norm": 0.6535904427628488, "learning_rate": 6.610311039283586e-06, "loss": 0.3165, "step": 11971 }, { "epoch": 0.8276529554096094, "grad_norm": 0.719488750181997, "learning_rate": 6.6097810947784225e-06, "loss": 0.3901, "step": 11972 }, { "epoch": 0.8277220877981334, "grad_norm": 0.6247760219702648, "learning_rate": 6.609251130097525e-06, "loss": 0.3125, "step": 11973 }, { "epoch": 0.8277912201866574, "grad_norm": 0.693360819675996, "learning_rate": 6.608721145247538e-06, "loss": 0.2668, "step": 11974 }, { "epoch": 0.8278603525751814, "grad_norm": 0.6990728345372039, "learning_rate": 6.608191140235101e-06, "loss": 0.3339, "step": 11975 }, { "epoch": 0.8279294849637054, "grad_norm": 0.6034945513105543, "learning_rate": 6.607661115066858e-06, "loss": 0.3653, "step": 11976 }, { "epoch": 0.8279986173522296, "grad_norm": 0.6752803063467923, "learning_rate": 6.607131069749453e-06, "loss": 0.268, "step": 11977 }, { "epoch": 0.8280677497407536, "grad_norm": 0.6724509472187074, "learning_rate": 6.606601004289527e-06, "loss": 0.3448, "step": 11978 }, { "epoch": 0.8281368821292776, "grad_norm": 0.693140160152975, "learning_rate": 6.606070918693726e-06, "loss": 0.3717, "step": 11979 }, { "epoch": 0.8282060145178016, "grad_norm": 0.6665795366405242, "learning_rate": 6.6055408129686905e-06, "loss": 0.3281, "step": 11980 }, { "epoch": 0.8282751469063256, "grad_norm": 0.6788720114031563, "learning_rate": 6.605010687121067e-06, "loss": 0.3082, "step": 11981 }, { "epoch": 0.8283442792948497, "grad_norm": 0.8051339933455042, "learning_rate": 6.604480541157498e-06, "loss": 0.3019, "step": 11982 }, { "epoch": 0.8284134116833737, "grad_norm": 0.7803589852881821, "learning_rate": 6.603950375084629e-06, "loss": 0.4173, "step": 11983 }, { "epoch": 0.8284825440718977, "grad_norm": 0.8528126271065234, "learning_rate": 6.603420188909105e-06, "loss": 0.3678, "step": 11984 }, { "epoch": 0.8285516764604217, "grad_norm": 0.6476980507049049, "learning_rate": 6.60288998263757e-06, "loss": 0.2784, "step": 11985 }, { "epoch": 0.8286208088489457, "grad_norm": 0.7206187044844501, "learning_rate": 6.60235975627667e-06, "loss": 0.3165, "step": 11986 }, { "epoch": 0.8286899412374698, "grad_norm": 0.6825807399511573, "learning_rate": 6.6018295098330495e-06, "loss": 0.328, "step": 11987 }, { "epoch": 0.8287590736259938, "grad_norm": 0.6943887416352404, "learning_rate": 6.601299243313355e-06, "loss": 0.2838, "step": 11988 }, { "epoch": 0.8288282060145178, "grad_norm": 0.718300048161815, "learning_rate": 6.600768956724232e-06, "loss": 0.2743, "step": 11989 }, { "epoch": 0.8288973384030418, "grad_norm": 0.6378594569630802, "learning_rate": 6.600238650072327e-06, "loss": 0.3796, "step": 11990 }, { "epoch": 0.8289664707915658, "grad_norm": 0.6932855424465206, "learning_rate": 6.599708323364286e-06, "loss": 0.3372, "step": 11991 }, { "epoch": 0.8290356031800898, "grad_norm": 0.7701529116475297, "learning_rate": 6.599177976606757e-06, "loss": 0.3228, "step": 11992 }, { "epoch": 0.8291047355686139, "grad_norm": 0.6353687049112106, "learning_rate": 6.598647609806383e-06, "loss": 0.3022, "step": 11993 }, { "epoch": 0.8291738679571379, "grad_norm": 0.7085708946770289, "learning_rate": 6.5981172229698176e-06, "loss": 0.4158, "step": 11994 }, { "epoch": 0.8292430003456619, "grad_norm": 0.7165269133784621, "learning_rate": 6.597586816103704e-06, "loss": 0.2835, "step": 11995 }, { "epoch": 0.8293121327341859, "grad_norm": 0.6726378212797608, "learning_rate": 6.597056389214689e-06, "loss": 0.2858, "step": 11996 }, { "epoch": 0.82938126512271, "grad_norm": 0.708210479517155, "learning_rate": 6.596525942309423e-06, "loss": 0.3793, "step": 11997 }, { "epoch": 0.8294503975112341, "grad_norm": 0.8029071016282403, "learning_rate": 6.5959954753945545e-06, "loss": 0.3227, "step": 11998 }, { "epoch": 0.8295195298997581, "grad_norm": 0.602782168944798, "learning_rate": 6.59546498847673e-06, "loss": 0.3132, "step": 11999 }, { "epoch": 0.8295886622882821, "grad_norm": 0.6977426776915356, "learning_rate": 6.594934481562598e-06, "loss": 0.2864, "step": 12000 }, { "epoch": 0.8296577946768061, "grad_norm": 0.7047328304615667, "learning_rate": 6.59440395465881e-06, "loss": 0.3667, "step": 12001 }, { "epoch": 0.8297269270653301, "grad_norm": 0.6218643823507717, "learning_rate": 6.593873407772015e-06, "loss": 0.2541, "step": 12002 }, { "epoch": 0.8297960594538542, "grad_norm": 0.6841703729383548, "learning_rate": 6.593342840908858e-06, "loss": 0.3084, "step": 12003 }, { "epoch": 0.8298651918423782, "grad_norm": 0.7091980235729758, "learning_rate": 6.5928122540759945e-06, "loss": 0.4696, "step": 12004 }, { "epoch": 0.8299343242309022, "grad_norm": 0.6590552348534333, "learning_rate": 6.592281647280071e-06, "loss": 0.3856, "step": 12005 }, { "epoch": 0.8300034566194262, "grad_norm": 0.8341459572677801, "learning_rate": 6.591751020527739e-06, "loss": 0.2408, "step": 12006 }, { "epoch": 0.8300725890079502, "grad_norm": 0.6996206895756005, "learning_rate": 6.591220373825648e-06, "loss": 0.3023, "step": 12007 }, { "epoch": 0.8301417213964742, "grad_norm": 0.6879335474558006, "learning_rate": 6.590689707180451e-06, "loss": 0.4092, "step": 12008 }, { "epoch": 0.8302108537849983, "grad_norm": 0.6525322096736881, "learning_rate": 6.5901590205987956e-06, "loss": 0.3295, "step": 12009 }, { "epoch": 0.8302799861735223, "grad_norm": 0.642400185791166, "learning_rate": 6.589628314087334e-06, "loss": 0.3001, "step": 12010 }, { "epoch": 0.8303491185620463, "grad_norm": 0.6620351107097693, "learning_rate": 6.589097587652719e-06, "loss": 0.2739, "step": 12011 }, { "epoch": 0.8304182509505703, "grad_norm": 0.7090694674424689, "learning_rate": 6.588566841301602e-06, "loss": 0.3441, "step": 12012 }, { "epoch": 0.8304873833390943, "grad_norm": 0.6953989576623768, "learning_rate": 6.588036075040634e-06, "loss": 0.3119, "step": 12013 }, { "epoch": 0.8305565157276183, "grad_norm": 0.7158048257626887, "learning_rate": 6.587505288876468e-06, "loss": 0.3138, "step": 12014 }, { "epoch": 0.8306256481161424, "grad_norm": 0.6575826555735812, "learning_rate": 6.5869744828157564e-06, "loss": 0.3612, "step": 12015 }, { "epoch": 0.8306947805046664, "grad_norm": 0.6787118669750911, "learning_rate": 6.586443656865153e-06, "loss": 0.3241, "step": 12016 }, { "epoch": 0.8307639128931905, "grad_norm": 0.6409265996295999, "learning_rate": 6.585912811031306e-06, "loss": 0.3023, "step": 12017 }, { "epoch": 0.8308330452817145, "grad_norm": 0.7600807961560412, "learning_rate": 6.585381945320874e-06, "loss": 0.3238, "step": 12018 }, { "epoch": 0.8309021776702386, "grad_norm": 0.6463172007162604, "learning_rate": 6.58485105974051e-06, "loss": 0.4268, "step": 12019 }, { "epoch": 0.8309713100587626, "grad_norm": 0.6632238209931551, "learning_rate": 6.584320154296866e-06, "loss": 0.2319, "step": 12020 }, { "epoch": 0.8310404424472866, "grad_norm": 0.7177122570582704, "learning_rate": 6.583789228996594e-06, "loss": 0.3106, "step": 12021 }, { "epoch": 0.8311095748358106, "grad_norm": 0.6702905454217833, "learning_rate": 6.5832582838463525e-06, "loss": 0.3618, "step": 12022 }, { "epoch": 0.8311787072243346, "grad_norm": 0.6686001560917548, "learning_rate": 6.582727318852793e-06, "loss": 0.3261, "step": 12023 }, { "epoch": 0.8312478396128586, "grad_norm": 0.6791016521744405, "learning_rate": 6.58219633402257e-06, "loss": 0.2628, "step": 12024 }, { "epoch": 0.8313169720013827, "grad_norm": 0.7291728458952962, "learning_rate": 6.581665329362341e-06, "loss": 0.2741, "step": 12025 }, { "epoch": 0.8313861043899067, "grad_norm": 0.785570785657095, "learning_rate": 6.58113430487876e-06, "loss": 0.4078, "step": 12026 }, { "epoch": 0.8314552367784307, "grad_norm": 0.6271428455285538, "learning_rate": 6.580603260578482e-06, "loss": 0.3218, "step": 12027 }, { "epoch": 0.8315243691669547, "grad_norm": 0.6054808314521581, "learning_rate": 6.580072196468161e-06, "loss": 0.239, "step": 12028 }, { "epoch": 0.8315935015554787, "grad_norm": 1.3211330990858707, "learning_rate": 6.579541112554456e-06, "loss": 0.392, "step": 12029 }, { "epoch": 0.8316626339440027, "grad_norm": 0.6917408842775967, "learning_rate": 6.5790100088440225e-06, "loss": 0.3916, "step": 12030 }, { "epoch": 0.8317317663325268, "grad_norm": 0.6796405597998096, "learning_rate": 6.578478885343515e-06, "loss": 0.3181, "step": 12031 }, { "epoch": 0.8318008987210508, "grad_norm": 0.6777617456056595, "learning_rate": 6.577947742059594e-06, "loss": 0.2612, "step": 12032 }, { "epoch": 0.8318700311095748, "grad_norm": 0.6713614923768269, "learning_rate": 6.577416578998912e-06, "loss": 0.4053, "step": 12033 }, { "epoch": 0.8319391634980988, "grad_norm": 0.6568205986841119, "learning_rate": 6.576885396168129e-06, "loss": 0.3245, "step": 12034 }, { "epoch": 0.8320082958866228, "grad_norm": 0.6479664721274789, "learning_rate": 6.576354193573901e-06, "loss": 0.2702, "step": 12035 }, { "epoch": 0.8320774282751469, "grad_norm": 0.6565825413317916, "learning_rate": 6.575822971222887e-06, "loss": 0.3098, "step": 12036 }, { "epoch": 0.8321465606636709, "grad_norm": 0.7230372754278493, "learning_rate": 6.575291729121744e-06, "loss": 0.3501, "step": 12037 }, { "epoch": 0.832215693052195, "grad_norm": 0.6837858938690176, "learning_rate": 6.5747604672771305e-06, "loss": 0.2722, "step": 12038 }, { "epoch": 0.832284825440719, "grad_norm": 0.6906952010550044, "learning_rate": 6.574229185695704e-06, "loss": 0.3562, "step": 12039 }, { "epoch": 0.832353957829243, "grad_norm": 0.70223432386126, "learning_rate": 6.573697884384125e-06, "loss": 0.4312, "step": 12040 }, { "epoch": 0.8324230902177671, "grad_norm": 0.6697636281395587, "learning_rate": 6.573166563349051e-06, "loss": 0.2712, "step": 12041 }, { "epoch": 0.8324922226062911, "grad_norm": 0.5936446998900986, "learning_rate": 6.572635222597141e-06, "loss": 0.256, "step": 12042 }, { "epoch": 0.8325613549948151, "grad_norm": 0.70501601226325, "learning_rate": 6.572103862135056e-06, "loss": 0.3797, "step": 12043 }, { "epoch": 0.8326304873833391, "grad_norm": 0.6660970302307505, "learning_rate": 6.571572481969455e-06, "loss": 0.4114, "step": 12044 }, { "epoch": 0.8326996197718631, "grad_norm": 0.6621735588734183, "learning_rate": 6.571041082106997e-06, "loss": 0.3689, "step": 12045 }, { "epoch": 0.8327687521603871, "grad_norm": 0.6281684284324167, "learning_rate": 6.570509662554342e-06, "loss": 0.2465, "step": 12046 }, { "epoch": 0.8328378845489112, "grad_norm": 0.6599179341524954, "learning_rate": 6.569978223318152e-06, "loss": 0.422, "step": 12047 }, { "epoch": 0.8329070169374352, "grad_norm": 0.7055098439893888, "learning_rate": 6.5694467644050875e-06, "loss": 0.3399, "step": 12048 }, { "epoch": 0.8329761493259592, "grad_norm": 0.7697965115680492, "learning_rate": 6.568915285821806e-06, "loss": 0.2544, "step": 12049 }, { "epoch": 0.8330452817144832, "grad_norm": 0.6900127584764, "learning_rate": 6.568383787574972e-06, "loss": 0.3718, "step": 12050 }, { "epoch": 0.8331144141030072, "grad_norm": 0.6878242966740272, "learning_rate": 6.567852269671248e-06, "loss": 0.3567, "step": 12051 }, { "epoch": 0.8331835464915313, "grad_norm": 0.6585712014893097, "learning_rate": 6.567320732117293e-06, "loss": 0.2733, "step": 12052 }, { "epoch": 0.8332526788800553, "grad_norm": 0.7946170675331481, "learning_rate": 6.566789174919769e-06, "loss": 0.2549, "step": 12053 }, { "epoch": 0.8333218112685793, "grad_norm": 0.6782153916868227, "learning_rate": 6.566257598085338e-06, "loss": 0.3871, "step": 12054 }, { "epoch": 0.8333909436571033, "grad_norm": 0.6626236957701088, "learning_rate": 6.5657260016206645e-06, "loss": 0.3311, "step": 12055 }, { "epoch": 0.8334600760456273, "grad_norm": 0.7783676410235013, "learning_rate": 6.56519438553241e-06, "loss": 0.3421, "step": 12056 }, { "epoch": 0.8335292084341513, "grad_norm": 0.6798264807677329, "learning_rate": 6.5646627498272356e-06, "loss": 0.3133, "step": 12057 }, { "epoch": 0.8335983408226755, "grad_norm": 0.9724276391809804, "learning_rate": 6.564131094511808e-06, "loss": 0.43, "step": 12058 }, { "epoch": 0.8336674732111995, "grad_norm": 0.7351853462942721, "learning_rate": 6.5635994195927865e-06, "loss": 0.3384, "step": 12059 }, { "epoch": 0.8337366055997235, "grad_norm": 0.6536913651898311, "learning_rate": 6.563067725076837e-06, "loss": 0.2757, "step": 12060 }, { "epoch": 0.8338057379882475, "grad_norm": 0.6957272602401485, "learning_rate": 6.562536010970624e-06, "loss": 0.3608, "step": 12061 }, { "epoch": 0.8338748703767715, "grad_norm": 0.8034728460423576, "learning_rate": 6.562004277280809e-06, "loss": 0.367, "step": 12062 }, { "epoch": 0.8339440027652956, "grad_norm": 1.0855377098248455, "learning_rate": 6.5614725240140595e-06, "loss": 0.2722, "step": 12063 }, { "epoch": 0.8340131351538196, "grad_norm": 0.6312414979408875, "learning_rate": 6.560940751177036e-06, "loss": 0.2875, "step": 12064 }, { "epoch": 0.8340822675423436, "grad_norm": 0.6787744240453413, "learning_rate": 6.560408958776409e-06, "loss": 0.36, "step": 12065 }, { "epoch": 0.8341513999308676, "grad_norm": 0.6092504400938715, "learning_rate": 6.559877146818839e-06, "loss": 0.2921, "step": 12066 }, { "epoch": 0.8342205323193916, "grad_norm": 0.705048605213657, "learning_rate": 6.559345315310992e-06, "loss": 0.2945, "step": 12067 }, { "epoch": 0.8342896647079157, "grad_norm": 1.0946575049703564, "learning_rate": 6.558813464259535e-06, "loss": 0.4216, "step": 12068 }, { "epoch": 0.8343587970964397, "grad_norm": 0.7202054846945992, "learning_rate": 6.558281593671133e-06, "loss": 0.4395, "step": 12069 }, { "epoch": 0.8344279294849637, "grad_norm": 0.6248130029331784, "learning_rate": 6.557749703552453e-06, "loss": 0.2755, "step": 12070 }, { "epoch": 0.8344970618734877, "grad_norm": 0.7965781261780862, "learning_rate": 6.557217793910157e-06, "loss": 0.2968, "step": 12071 }, { "epoch": 0.8345661942620117, "grad_norm": 0.9244889246212525, "learning_rate": 6.556685864750919e-06, "loss": 0.3819, "step": 12072 }, { "epoch": 0.8346353266505357, "grad_norm": 0.673863258097381, "learning_rate": 6.5561539160814e-06, "loss": 0.336, "step": 12073 }, { "epoch": 0.8347044590390598, "grad_norm": 0.9889441282287127, "learning_rate": 6.55562194790827e-06, "loss": 0.3034, "step": 12074 }, { "epoch": 0.8347735914275838, "grad_norm": 0.7355796725060815, "learning_rate": 6.5550899602381934e-06, "loss": 0.2847, "step": 12075 }, { "epoch": 0.8348427238161078, "grad_norm": 0.6275985967649719, "learning_rate": 6.55455795307784e-06, "loss": 0.4249, "step": 12076 }, { "epoch": 0.8349118562046318, "grad_norm": 0.6345451718533294, "learning_rate": 6.554025926433875e-06, "loss": 0.2683, "step": 12077 }, { "epoch": 0.834980988593156, "grad_norm": 0.6503896290936195, "learning_rate": 6.553493880312971e-06, "loss": 0.2226, "step": 12078 }, { "epoch": 0.83505012098168, "grad_norm": 0.6453471140449336, "learning_rate": 6.5529618147217936e-06, "loss": 0.3505, "step": 12079 }, { "epoch": 0.835119253370204, "grad_norm": 1.1963396689525454, "learning_rate": 6.5524297296670095e-06, "loss": 0.3894, "step": 12080 }, { "epoch": 0.835188385758728, "grad_norm": 0.6623443516383147, "learning_rate": 6.551897625155289e-06, "loss": 0.2819, "step": 12081 }, { "epoch": 0.835257518147252, "grad_norm": 0.7422480861192101, "learning_rate": 6.551365501193303e-06, "loss": 0.3318, "step": 12082 }, { "epoch": 0.835326650535776, "grad_norm": 0.6895081340647315, "learning_rate": 6.550833357787718e-06, "loss": 0.3422, "step": 12083 }, { "epoch": 0.8353957829243001, "grad_norm": 0.6307696256269801, "learning_rate": 6.550301194945204e-06, "loss": 0.3036, "step": 12084 }, { "epoch": 0.8354649153128241, "grad_norm": 0.6782011338885128, "learning_rate": 6.549769012672431e-06, "loss": 0.3109, "step": 12085 }, { "epoch": 0.8355340477013481, "grad_norm": 0.6650921616586258, "learning_rate": 6.549236810976071e-06, "loss": 0.3137, "step": 12086 }, { "epoch": 0.8356031800898721, "grad_norm": 0.6901166646530388, "learning_rate": 6.548704589862791e-06, "loss": 0.406, "step": 12087 }, { "epoch": 0.8356723124783961, "grad_norm": 0.6987643419128011, "learning_rate": 6.548172349339263e-06, "loss": 0.3057, "step": 12088 }, { "epoch": 0.8357414448669201, "grad_norm": 0.6032379642320155, "learning_rate": 6.547640089412158e-06, "loss": 0.3411, "step": 12089 }, { "epoch": 0.8358105772554442, "grad_norm": 0.7459199733703196, "learning_rate": 6.5471078100881445e-06, "loss": 0.4441, "step": 12090 }, { "epoch": 0.8358797096439682, "grad_norm": 0.7502950440151267, "learning_rate": 6.546575511373897e-06, "loss": 0.3377, "step": 12091 }, { "epoch": 0.8359488420324922, "grad_norm": 0.6995659255541357, "learning_rate": 6.5460431932760844e-06, "loss": 0.2591, "step": 12092 }, { "epoch": 0.8360179744210162, "grad_norm": 0.7428304366423536, "learning_rate": 6.54551085580138e-06, "loss": 0.3475, "step": 12093 }, { "epoch": 0.8360871068095402, "grad_norm": 0.5774465941860233, "learning_rate": 6.544978498956456e-06, "loss": 0.4037, "step": 12094 }, { "epoch": 0.8361562391980643, "grad_norm": 0.6432986923501228, "learning_rate": 6.544446122747982e-06, "loss": 0.2879, "step": 12095 }, { "epoch": 0.8362253715865883, "grad_norm": 0.6820551492806568, "learning_rate": 6.5439137271826335e-06, "loss": 0.317, "step": 12096 }, { "epoch": 0.8362945039751123, "grad_norm": 0.6257031970710655, "learning_rate": 6.543381312267081e-06, "loss": 0.3619, "step": 12097 }, { "epoch": 0.8363636363636363, "grad_norm": 0.6867326219301747, "learning_rate": 6.542848878007999e-06, "loss": 0.3366, "step": 12098 }, { "epoch": 0.8364327687521604, "grad_norm": 0.6437337796474968, "learning_rate": 6.542316424412057e-06, "loss": 0.2405, "step": 12099 }, { "epoch": 0.8365019011406845, "grad_norm": 0.6444322827639898, "learning_rate": 6.541783951485934e-06, "loss": 0.2932, "step": 12100 }, { "epoch": 0.8365710335292085, "grad_norm": 0.7593305648960776, "learning_rate": 6.5412514592363e-06, "loss": 0.4021, "step": 12101 }, { "epoch": 0.8366401659177325, "grad_norm": 0.6692604094864631, "learning_rate": 6.540718947669827e-06, "loss": 0.296, "step": 12102 }, { "epoch": 0.8367092983062565, "grad_norm": 0.7334258111422158, "learning_rate": 6.540186416793194e-06, "loss": 0.3291, "step": 12103 }, { "epoch": 0.8367784306947805, "grad_norm": 0.6782818816643149, "learning_rate": 6.539653866613073e-06, "loss": 0.4161, "step": 12104 }, { "epoch": 0.8368475630833045, "grad_norm": 0.7807082388321367, "learning_rate": 6.539121297136137e-06, "loss": 0.3332, "step": 12105 }, { "epoch": 0.8369166954718286, "grad_norm": 0.7306042315668257, "learning_rate": 6.538588708369062e-06, "loss": 0.3103, "step": 12106 }, { "epoch": 0.8369858278603526, "grad_norm": 0.6565062000268352, "learning_rate": 6.538056100318525e-06, "loss": 0.3129, "step": 12107 }, { "epoch": 0.8370549602488766, "grad_norm": 0.7321383232286782, "learning_rate": 6.5375234729912e-06, "loss": 0.3979, "step": 12108 }, { "epoch": 0.8371240926374006, "grad_norm": 0.7128029683908562, "learning_rate": 6.536990826393761e-06, "loss": 0.3233, "step": 12109 }, { "epoch": 0.8371932250259246, "grad_norm": 0.6242806658572342, "learning_rate": 6.5364581605328835e-06, "loss": 0.2771, "step": 12110 }, { "epoch": 0.8372623574144487, "grad_norm": 0.6594792176611333, "learning_rate": 6.535925475415246e-06, "loss": 0.3086, "step": 12111 }, { "epoch": 0.8373314898029727, "grad_norm": 0.6243689023691977, "learning_rate": 6.535392771047524e-06, "loss": 0.3371, "step": 12112 }, { "epoch": 0.8374006221914967, "grad_norm": 0.7116774687335995, "learning_rate": 6.534860047436393e-06, "loss": 0.3076, "step": 12113 }, { "epoch": 0.8374697545800207, "grad_norm": 0.6368032885843745, "learning_rate": 6.5343273045885306e-06, "loss": 0.2892, "step": 12114 }, { "epoch": 0.8375388869685447, "grad_norm": 0.8432249455841279, "learning_rate": 6.533794542510614e-06, "loss": 0.4436, "step": 12115 }, { "epoch": 0.8376080193570687, "grad_norm": 0.7497918483954337, "learning_rate": 6.53326176120932e-06, "loss": 0.3401, "step": 12116 }, { "epoch": 0.8376771517455928, "grad_norm": 1.309012087454456, "learning_rate": 6.532728960691326e-06, "loss": 0.2989, "step": 12117 }, { "epoch": 0.8377462841341168, "grad_norm": 0.6593705119706358, "learning_rate": 6.532196140963309e-06, "loss": 0.2929, "step": 12118 }, { "epoch": 0.8378154165226409, "grad_norm": 0.7285007920472266, "learning_rate": 6.531663302031949e-06, "loss": 0.4573, "step": 12119 }, { "epoch": 0.8378845489111649, "grad_norm": 0.6181421950571353, "learning_rate": 6.531130443903921e-06, "loss": 0.266, "step": 12120 }, { "epoch": 0.837953681299689, "grad_norm": 0.6759854130986527, "learning_rate": 6.530597566585907e-06, "loss": 0.3122, "step": 12121 }, { "epoch": 0.838022813688213, "grad_norm": 0.6891479106884165, "learning_rate": 6.530064670084584e-06, "loss": 0.4511, "step": 12122 }, { "epoch": 0.838091946076737, "grad_norm": 0.6425838541964083, "learning_rate": 6.5295317544066306e-06, "loss": 0.335, "step": 12123 }, { "epoch": 0.838161078465261, "grad_norm": 1.2088830294609383, "learning_rate": 6.528998819558725e-06, "loss": 0.3408, "step": 12124 }, { "epoch": 0.838230210853785, "grad_norm": 1.0924079433598368, "learning_rate": 6.528465865547549e-06, "loss": 0.2568, "step": 12125 }, { "epoch": 0.838299343242309, "grad_norm": 0.7969233989442334, "learning_rate": 6.52793289237978e-06, "loss": 0.4165, "step": 12126 }, { "epoch": 0.838368475630833, "grad_norm": 0.7795495936553362, "learning_rate": 6.5273999000621e-06, "loss": 0.2812, "step": 12127 }, { "epoch": 0.8384376080193571, "grad_norm": 0.7715887927413293, "learning_rate": 6.526866888601188e-06, "loss": 0.253, "step": 12128 }, { "epoch": 0.8385067404078811, "grad_norm": 0.7435551970120641, "learning_rate": 6.526333858003724e-06, "loss": 0.3967, "step": 12129 }, { "epoch": 0.8385758727964051, "grad_norm": 0.7934698966917996, "learning_rate": 6.525800808276389e-06, "loss": 0.3345, "step": 12130 }, { "epoch": 0.8386450051849291, "grad_norm": 0.6466082422619546, "learning_rate": 6.525267739425864e-06, "loss": 0.281, "step": 12131 }, { "epoch": 0.8387141375734531, "grad_norm": 0.6589612670347444, "learning_rate": 6.524734651458829e-06, "loss": 0.2799, "step": 12132 }, { "epoch": 0.8387832699619772, "grad_norm": 0.853895494196303, "learning_rate": 6.524201544381968e-06, "loss": 0.4262, "step": 12133 }, { "epoch": 0.8388524023505012, "grad_norm": 0.6530605579946172, "learning_rate": 6.523668418201959e-06, "loss": 0.2693, "step": 12134 }, { "epoch": 0.8389215347390252, "grad_norm": 0.7027589718199966, "learning_rate": 6.523135272925485e-06, "loss": 0.2771, "step": 12135 }, { "epoch": 0.8389906671275492, "grad_norm": 0.7100658458104132, "learning_rate": 6.52260210855923e-06, "loss": 0.3287, "step": 12136 }, { "epoch": 0.8390597995160732, "grad_norm": 0.6818275393421308, "learning_rate": 6.522068925109874e-06, "loss": 0.3711, "step": 12137 }, { "epoch": 0.8391289319045973, "grad_norm": 0.6847829490380963, "learning_rate": 6.5215357225841e-06, "loss": 0.2522, "step": 12138 }, { "epoch": 0.8391980642931214, "grad_norm": 0.7072115766771683, "learning_rate": 6.5210025009885926e-06, "loss": 0.3583, "step": 12139 }, { "epoch": 0.8392671966816454, "grad_norm": 0.7006946648730767, "learning_rate": 6.520469260330032e-06, "loss": 0.3679, "step": 12140 }, { "epoch": 0.8393363290701694, "grad_norm": 0.6804477625057318, "learning_rate": 6.519936000615104e-06, "loss": 0.2737, "step": 12141 }, { "epoch": 0.8394054614586934, "grad_norm": 0.6887249490145839, "learning_rate": 6.5194027218504895e-06, "loss": 0.2373, "step": 12142 }, { "epoch": 0.8394745938472175, "grad_norm": 0.6554741093980828, "learning_rate": 6.518869424042873e-06, "loss": 0.3522, "step": 12143 }, { "epoch": 0.8395437262357415, "grad_norm": 0.6852752695322252, "learning_rate": 6.518336107198939e-06, "loss": 0.4212, "step": 12144 }, { "epoch": 0.8396128586242655, "grad_norm": 0.6528537861240191, "learning_rate": 6.517802771325372e-06, "loss": 0.2818, "step": 12145 }, { "epoch": 0.8396819910127895, "grad_norm": 0.6822753087360433, "learning_rate": 6.517269416428856e-06, "loss": 0.2997, "step": 12146 }, { "epoch": 0.8397511234013135, "grad_norm": 0.7400846030192967, "learning_rate": 6.516736042516074e-06, "loss": 0.4013, "step": 12147 }, { "epoch": 0.8398202557898375, "grad_norm": 0.6502706844391135, "learning_rate": 6.5162026495937135e-06, "loss": 0.3319, "step": 12148 }, { "epoch": 0.8398893881783616, "grad_norm": 0.6195487694828524, "learning_rate": 6.515669237668459e-06, "loss": 0.3038, "step": 12149 }, { "epoch": 0.8399585205668856, "grad_norm": 0.722902636844067, "learning_rate": 6.515135806746995e-06, "loss": 0.3435, "step": 12150 }, { "epoch": 0.8400276529554096, "grad_norm": 0.6444077643948525, "learning_rate": 6.514602356836009e-06, "loss": 0.3846, "step": 12151 }, { "epoch": 0.8400967853439336, "grad_norm": 0.6777723298504427, "learning_rate": 6.514068887942182e-06, "loss": 0.3148, "step": 12152 }, { "epoch": 0.8401659177324576, "grad_norm": 0.6672840196224837, "learning_rate": 6.513535400072206e-06, "loss": 0.2902, "step": 12153 }, { "epoch": 0.8402350501209817, "grad_norm": 0.6536351433205748, "learning_rate": 6.513001893232764e-06, "loss": 0.3569, "step": 12154 }, { "epoch": 0.8403041825095057, "grad_norm": 0.8111927064620134, "learning_rate": 6.5124683674305435e-06, "loss": 0.3635, "step": 12155 }, { "epoch": 0.8403733148980297, "grad_norm": 0.6370574911387805, "learning_rate": 6.511934822672232e-06, "loss": 0.2885, "step": 12156 }, { "epoch": 0.8404424472865537, "grad_norm": 0.6233610142845479, "learning_rate": 6.511401258964515e-06, "loss": 0.3077, "step": 12157 }, { "epoch": 0.8405115796750777, "grad_norm": 0.6476500221343862, "learning_rate": 6.510867676314081e-06, "loss": 0.3812, "step": 12158 }, { "epoch": 0.8405807120636017, "grad_norm": 0.6368404613796981, "learning_rate": 6.510334074727616e-06, "loss": 0.2978, "step": 12159 }, { "epoch": 0.8406498444521259, "grad_norm": 0.6674547212323649, "learning_rate": 6.509800454211809e-06, "loss": 0.3148, "step": 12160 }, { "epoch": 0.8407189768406499, "grad_norm": 0.6310397403171, "learning_rate": 6.509266814773348e-06, "loss": 0.3211, "step": 12161 }, { "epoch": 0.8407881092291739, "grad_norm": 0.6459046543847679, "learning_rate": 6.50873315641892e-06, "loss": 0.3527, "step": 12162 }, { "epoch": 0.8408572416176979, "grad_norm": 0.6183382667125075, "learning_rate": 6.508199479155214e-06, "loss": 0.2732, "step": 12163 }, { "epoch": 0.8409263740062219, "grad_norm": 0.6378898819710035, "learning_rate": 6.5076657829889204e-06, "loss": 0.3049, "step": 12164 }, { "epoch": 0.840995506394746, "grad_norm": 0.7308749416569214, "learning_rate": 6.507132067926727e-06, "loss": 0.3691, "step": 12165 }, { "epoch": 0.84106463878327, "grad_norm": 0.6730365313875929, "learning_rate": 6.506598333975321e-06, "loss": 0.3032, "step": 12166 }, { "epoch": 0.841133771171794, "grad_norm": 0.6724718478179481, "learning_rate": 6.506064581141395e-06, "loss": 0.2876, "step": 12167 }, { "epoch": 0.841202903560318, "grad_norm": 0.68110980131652, "learning_rate": 6.505530809431637e-06, "loss": 0.3204, "step": 12168 }, { "epoch": 0.841272035948842, "grad_norm": 0.695848637363696, "learning_rate": 6.504997018852736e-06, "loss": 0.3627, "step": 12169 }, { "epoch": 0.841341168337366, "grad_norm": 0.6561974233538325, "learning_rate": 6.5044632094113836e-06, "loss": 0.3191, "step": 12170 }, { "epoch": 0.8414103007258901, "grad_norm": 0.6626998903517717, "learning_rate": 6.50392938111427e-06, "loss": 0.2873, "step": 12171 }, { "epoch": 0.8414794331144141, "grad_norm": 0.6043292468824035, "learning_rate": 6.503395533968087e-06, "loss": 0.3918, "step": 12172 }, { "epoch": 0.8415485655029381, "grad_norm": 0.6366487586031321, "learning_rate": 6.502861667979521e-06, "loss": 0.3212, "step": 12173 }, { "epoch": 0.8416176978914621, "grad_norm": 0.7212739527823707, "learning_rate": 6.502327783155266e-06, "loss": 0.3188, "step": 12174 }, { "epoch": 0.8416868302799861, "grad_norm": 0.8668949188472558, "learning_rate": 6.501793879502017e-06, "loss": 0.2549, "step": 12175 }, { "epoch": 0.8417559626685102, "grad_norm": 0.638541686149167, "learning_rate": 6.501259957026459e-06, "loss": 0.3998, "step": 12176 }, { "epoch": 0.8418250950570342, "grad_norm": 0.6265939250812331, "learning_rate": 6.500726015735286e-06, "loss": 0.322, "step": 12177 }, { "epoch": 0.8418942274455582, "grad_norm": 0.670939300091197, "learning_rate": 6.500192055635191e-06, "loss": 0.2804, "step": 12178 }, { "epoch": 0.8419633598340822, "grad_norm": 1.1874031845333246, "learning_rate": 6.499658076732868e-06, "loss": 0.4393, "step": 12179 }, { "epoch": 0.8420324922226063, "grad_norm": 0.7091871834097356, "learning_rate": 6.499124079035005e-06, "loss": 0.355, "step": 12180 }, { "epoch": 0.8421016246111304, "grad_norm": 0.6214966759894982, "learning_rate": 6.498590062548298e-06, "loss": 0.2721, "step": 12181 }, { "epoch": 0.8421707569996544, "grad_norm": 0.6744267723249897, "learning_rate": 6.498056027279438e-06, "loss": 0.3193, "step": 12182 }, { "epoch": 0.8422398893881784, "grad_norm": 0.6762988197898813, "learning_rate": 6.4975219732351194e-06, "loss": 0.4668, "step": 12183 }, { "epoch": 0.8423090217767024, "grad_norm": 0.6535189949362863, "learning_rate": 6.4969879004220345e-06, "loss": 0.2565, "step": 12184 }, { "epoch": 0.8423781541652264, "grad_norm": 0.6654614845314311, "learning_rate": 6.49645380884688e-06, "loss": 0.3175, "step": 12185 }, { "epoch": 0.8424472865537505, "grad_norm": 0.7004763135032653, "learning_rate": 6.495919698516346e-06, "loss": 0.4342, "step": 12186 }, { "epoch": 0.8425164189422745, "grad_norm": 0.6477728788761888, "learning_rate": 6.495385569437129e-06, "loss": 0.3382, "step": 12187 }, { "epoch": 0.8425855513307985, "grad_norm": 0.5876437977486204, "learning_rate": 6.494851421615921e-06, "loss": 0.2438, "step": 12188 }, { "epoch": 0.8426546837193225, "grad_norm": 0.8019743479560109, "learning_rate": 6.494317255059419e-06, "loss": 0.3498, "step": 12189 }, { "epoch": 0.8427238161078465, "grad_norm": 0.6837514939931678, "learning_rate": 6.493783069774318e-06, "loss": 0.4279, "step": 12190 }, { "epoch": 0.8427929484963705, "grad_norm": 0.6065403588665352, "learning_rate": 6.49324886576731e-06, "loss": 0.2709, "step": 12191 }, { "epoch": 0.8428620808848946, "grad_norm": 0.6917750587040636, "learning_rate": 6.492714643045094e-06, "loss": 0.2459, "step": 12192 }, { "epoch": 0.8429312132734186, "grad_norm": 0.7146615479364169, "learning_rate": 6.492180401614363e-06, "loss": 0.3199, "step": 12193 }, { "epoch": 0.8430003456619426, "grad_norm": 0.6405770148551394, "learning_rate": 6.491646141481815e-06, "loss": 0.3535, "step": 12194 }, { "epoch": 0.8430694780504666, "grad_norm": 0.6593790186402178, "learning_rate": 6.491111862654143e-06, "loss": 0.3362, "step": 12195 }, { "epoch": 0.8431386104389906, "grad_norm": 0.6277666597150511, "learning_rate": 6.490577565138044e-06, "loss": 0.3012, "step": 12196 }, { "epoch": 0.8432077428275147, "grad_norm": 0.6969743873206683, "learning_rate": 6.490043248940217e-06, "loss": 0.4164, "step": 12197 }, { "epoch": 0.8432768752160387, "grad_norm": 0.7197196725971672, "learning_rate": 6.489508914067357e-06, "loss": 0.318, "step": 12198 }, { "epoch": 0.8433460076045627, "grad_norm": 0.5978351264338183, "learning_rate": 6.488974560526159e-06, "loss": 0.2853, "step": 12199 }, { "epoch": 0.8434151399930868, "grad_norm": 0.6790933041711461, "learning_rate": 6.488440188323324e-06, "loss": 0.3167, "step": 12200 }, { "epoch": 0.8434842723816108, "grad_norm": 0.6934921175944393, "learning_rate": 6.487905797465547e-06, "loss": 0.3919, "step": 12201 }, { "epoch": 0.8435534047701349, "grad_norm": 0.6438285532772182, "learning_rate": 6.4873713879595244e-06, "loss": 0.2839, "step": 12202 }, { "epoch": 0.8436225371586589, "grad_norm": 0.6685900237149832, "learning_rate": 6.486836959811958e-06, "loss": 0.3221, "step": 12203 }, { "epoch": 0.8436916695471829, "grad_norm": 0.7236308937141916, "learning_rate": 6.486302513029543e-06, "loss": 0.3964, "step": 12204 }, { "epoch": 0.8437608019357069, "grad_norm": 0.7545509477850181, "learning_rate": 6.485768047618979e-06, "loss": 0.2953, "step": 12205 }, { "epoch": 0.8438299343242309, "grad_norm": 0.6430976257093869, "learning_rate": 6.485233563586963e-06, "loss": 0.3031, "step": 12206 }, { "epoch": 0.8438990667127549, "grad_norm": 0.6474421135745988, "learning_rate": 6.484699060940196e-06, "loss": 0.3111, "step": 12207 }, { "epoch": 0.843968199101279, "grad_norm": 0.6711154126836588, "learning_rate": 6.484164539685376e-06, "loss": 0.3918, "step": 12208 }, { "epoch": 0.844037331489803, "grad_norm": 0.8066870746910505, "learning_rate": 6.4836299998292005e-06, "loss": 0.3198, "step": 12209 }, { "epoch": 0.844106463878327, "grad_norm": 0.6608371195038791, "learning_rate": 6.483095441378371e-06, "loss": 0.295, "step": 12210 }, { "epoch": 0.844175596266851, "grad_norm": 0.7231174006581625, "learning_rate": 6.482560864339587e-06, "loss": 0.3374, "step": 12211 }, { "epoch": 0.844244728655375, "grad_norm": 0.6389767783371513, "learning_rate": 6.482026268719549e-06, "loss": 0.3774, "step": 12212 }, { "epoch": 0.844313861043899, "grad_norm": 0.6310796292343944, "learning_rate": 6.481491654524956e-06, "loss": 0.2467, "step": 12213 }, { "epoch": 0.8443829934324231, "grad_norm": 0.6818106452119711, "learning_rate": 6.480957021762509e-06, "loss": 0.298, "step": 12214 }, { "epoch": 0.8444521258209471, "grad_norm": 0.6540897164590699, "learning_rate": 6.480422370438909e-06, "loss": 0.4056, "step": 12215 }, { "epoch": 0.8445212582094711, "grad_norm": 0.7840568757002635, "learning_rate": 6.4798877005608555e-06, "loss": 0.2837, "step": 12216 }, { "epoch": 0.8445903905979951, "grad_norm": 1.5259818253170399, "learning_rate": 6.479353012135051e-06, "loss": 0.2589, "step": 12217 }, { "epoch": 0.8446595229865191, "grad_norm": 0.6534352078926957, "learning_rate": 6.478818305168197e-06, "loss": 0.3591, "step": 12218 }, { "epoch": 0.8447286553750432, "grad_norm": 0.7247165552922726, "learning_rate": 6.478283579666994e-06, "loss": 0.3835, "step": 12219 }, { "epoch": 0.8447977877635672, "grad_norm": 0.7139982793296885, "learning_rate": 6.4777488356381445e-06, "loss": 0.2599, "step": 12220 }, { "epoch": 0.8448669201520913, "grad_norm": 0.6702193930180778, "learning_rate": 6.477214073088352e-06, "loss": 0.2993, "step": 12221 }, { "epoch": 0.8449360525406153, "grad_norm": 0.6541671881959293, "learning_rate": 6.476679292024316e-06, "loss": 0.3448, "step": 12222 }, { "epoch": 0.8450051849291393, "grad_norm": 0.7146187014408785, "learning_rate": 6.47614449245274e-06, "loss": 0.4163, "step": 12223 }, { "epoch": 0.8450743173176634, "grad_norm": 0.6692904614417708, "learning_rate": 6.4756096743803276e-06, "loss": 0.2879, "step": 12224 }, { "epoch": 0.8451434497061874, "grad_norm": 0.6744572616697563, "learning_rate": 6.475074837813782e-06, "loss": 0.2492, "step": 12225 }, { "epoch": 0.8452125820947114, "grad_norm": 1.3734464534454316, "learning_rate": 6.474539982759803e-06, "loss": 0.3805, "step": 12226 }, { "epoch": 0.8452817144832354, "grad_norm": 0.6702741227440127, "learning_rate": 6.474005109225099e-06, "loss": 0.2734, "step": 12227 }, { "epoch": 0.8453508468717594, "grad_norm": 0.6713495936738338, "learning_rate": 6.473470217216371e-06, "loss": 0.3029, "step": 12228 }, { "epoch": 0.8454199792602834, "grad_norm": 0.6581269560221452, "learning_rate": 6.472935306740325e-06, "loss": 0.3975, "step": 12229 }, { "epoch": 0.8454891116488075, "grad_norm": 0.6701522521969127, "learning_rate": 6.472400377803661e-06, "loss": 0.3323, "step": 12230 }, { "epoch": 0.8455582440373315, "grad_norm": 0.6923165240412529, "learning_rate": 6.471865430413087e-06, "loss": 0.2664, "step": 12231 }, { "epoch": 0.8456273764258555, "grad_norm": 0.6871938785867627, "learning_rate": 6.471330464575306e-06, "loss": 0.2873, "step": 12232 }, { "epoch": 0.8456965088143795, "grad_norm": 0.6730723089438224, "learning_rate": 6.4707954802970245e-06, "loss": 0.4347, "step": 12233 }, { "epoch": 0.8457656412029035, "grad_norm": 0.8967397411840515, "learning_rate": 6.470260477584944e-06, "loss": 0.2587, "step": 12234 }, { "epoch": 0.8458347735914276, "grad_norm": 0.6833708763431782, "learning_rate": 6.469725456445775e-06, "loss": 0.2919, "step": 12235 }, { "epoch": 0.8459039059799516, "grad_norm": 0.6302773096827521, "learning_rate": 6.469190416886218e-06, "loss": 0.2776, "step": 12236 }, { "epoch": 0.8459730383684756, "grad_norm": 0.6887709766145631, "learning_rate": 6.468655358912982e-06, "loss": 0.3737, "step": 12237 }, { "epoch": 0.8460421707569996, "grad_norm": 0.6257068736162932, "learning_rate": 6.468120282532772e-06, "loss": 0.2588, "step": 12238 }, { "epoch": 0.8461113031455236, "grad_norm": 0.7498305995398882, "learning_rate": 6.467585187752294e-06, "loss": 0.2911, "step": 12239 }, { "epoch": 0.8461804355340476, "grad_norm": 0.7042963248887056, "learning_rate": 6.467050074578256e-06, "loss": 0.3819, "step": 12240 }, { "epoch": 0.8462495679225718, "grad_norm": 0.6238130117637255, "learning_rate": 6.466514943017361e-06, "loss": 0.3092, "step": 12241 }, { "epoch": 0.8463187003110958, "grad_norm": 0.6518414847429136, "learning_rate": 6.4659797930763205e-06, "loss": 0.2755, "step": 12242 }, { "epoch": 0.8463878326996198, "grad_norm": 1.636587504113172, "learning_rate": 6.4654446247618384e-06, "loss": 0.3375, "step": 12243 }, { "epoch": 0.8464569650881438, "grad_norm": 0.7117262448682419, "learning_rate": 6.464909438080623e-06, "loss": 0.3989, "step": 12244 }, { "epoch": 0.8465260974766678, "grad_norm": 0.759428669244431, "learning_rate": 6.464374233039383e-06, "loss": 0.2949, "step": 12245 }, { "epoch": 0.8465952298651919, "grad_norm": 0.6769455581059941, "learning_rate": 6.463839009644824e-06, "loss": 0.332, "step": 12246 }, { "epoch": 0.8466643622537159, "grad_norm": 0.7082576492281847, "learning_rate": 6.463303767903657e-06, "loss": 0.3514, "step": 12247 }, { "epoch": 0.8467334946422399, "grad_norm": 0.7008281681391537, "learning_rate": 6.462768507822586e-06, "loss": 0.2759, "step": 12248 }, { "epoch": 0.8468026270307639, "grad_norm": 0.6257750855984915, "learning_rate": 6.462233229408325e-06, "loss": 0.297, "step": 12249 }, { "epoch": 0.8468717594192879, "grad_norm": 0.7364070643247682, "learning_rate": 6.461697932667579e-06, "loss": 0.3622, "step": 12250 }, { "epoch": 0.846940891807812, "grad_norm": 0.7226590745065024, "learning_rate": 6.461162617607058e-06, "loss": 0.3873, "step": 12251 }, { "epoch": 0.847010024196336, "grad_norm": 1.1885682972181852, "learning_rate": 6.460627284233469e-06, "loss": 0.297, "step": 12252 }, { "epoch": 0.84707915658486, "grad_norm": 0.6466801390206148, "learning_rate": 6.460091932553527e-06, "loss": 0.267, "step": 12253 }, { "epoch": 0.847148288973384, "grad_norm": 0.6806394094314577, "learning_rate": 6.4595565625739365e-06, "loss": 0.3899, "step": 12254 }, { "epoch": 0.847217421361908, "grad_norm": 0.7086752106518968, "learning_rate": 6.4590211743014085e-06, "loss": 0.3556, "step": 12255 }, { "epoch": 0.847286553750432, "grad_norm": 0.7402314560416825, "learning_rate": 6.4584857677426546e-06, "loss": 0.2552, "step": 12256 }, { "epoch": 0.8473556861389561, "grad_norm": 0.734484713442535, "learning_rate": 6.457950342904386e-06, "loss": 0.3056, "step": 12257 }, { "epoch": 0.8474248185274801, "grad_norm": 0.6730483617939507, "learning_rate": 6.457414899793311e-06, "loss": 0.4193, "step": 12258 }, { "epoch": 0.8474939509160041, "grad_norm": 0.6994699310671474, "learning_rate": 6.456879438416139e-06, "loss": 0.2861, "step": 12259 }, { "epoch": 0.8475630833045281, "grad_norm": 0.5750831977280085, "learning_rate": 6.456343958779585e-06, "loss": 0.2744, "step": 12260 }, { "epoch": 0.8476322156930522, "grad_norm": 0.6889515552981692, "learning_rate": 6.4558084608903585e-06, "loss": 0.3445, "step": 12261 }, { "epoch": 0.8477013480815763, "grad_norm": 0.7763564638513886, "learning_rate": 6.455272944755169e-06, "loss": 0.4404, "step": 12262 }, { "epoch": 0.8477704804701003, "grad_norm": 0.668986949771761, "learning_rate": 6.454737410380731e-06, "loss": 0.264, "step": 12263 }, { "epoch": 0.8478396128586243, "grad_norm": 0.6409460565931412, "learning_rate": 6.454201857773756e-06, "loss": 0.276, "step": 12264 }, { "epoch": 0.8479087452471483, "grad_norm": 0.6847359228759505, "learning_rate": 6.4536662869409575e-06, "loss": 0.4431, "step": 12265 }, { "epoch": 0.8479778776356723, "grad_norm": 1.6559811422396384, "learning_rate": 6.453130697889044e-06, "loss": 0.3455, "step": 12266 }, { "epoch": 0.8480470100241964, "grad_norm": 0.7460044215600015, "learning_rate": 6.452595090624731e-06, "loss": 0.2552, "step": 12267 }, { "epoch": 0.8481161424127204, "grad_norm": 0.7170544781997258, "learning_rate": 6.452059465154732e-06, "loss": 0.3412, "step": 12268 }, { "epoch": 0.8481852748012444, "grad_norm": 0.6919800176679842, "learning_rate": 6.451523821485758e-06, "loss": 0.3539, "step": 12269 }, { "epoch": 0.8482544071897684, "grad_norm": 0.6070971600147992, "learning_rate": 6.450988159624523e-06, "loss": 0.252, "step": 12270 }, { "epoch": 0.8483235395782924, "grad_norm": 0.9695487127176101, "learning_rate": 6.450452479577741e-06, "loss": 0.3188, "step": 12271 }, { "epoch": 0.8483926719668164, "grad_norm": 0.6975525102523105, "learning_rate": 6.449916781352126e-06, "loss": 0.4059, "step": 12272 }, { "epoch": 0.8484618043553405, "grad_norm": 0.7350527151805205, "learning_rate": 6.44938106495439e-06, "loss": 0.3012, "step": 12273 }, { "epoch": 0.8485309367438645, "grad_norm": 0.6661093789269376, "learning_rate": 6.448845330391252e-06, "loss": 0.2731, "step": 12274 }, { "epoch": 0.8486000691323885, "grad_norm": 0.7095015932556831, "learning_rate": 6.448309577669421e-06, "loss": 0.3883, "step": 12275 }, { "epoch": 0.8486692015209125, "grad_norm": 0.6551007430873097, "learning_rate": 6.447773806795614e-06, "loss": 0.384, "step": 12276 }, { "epoch": 0.8487383339094365, "grad_norm": 0.9590148512102298, "learning_rate": 6.447238017776547e-06, "loss": 0.2707, "step": 12277 }, { "epoch": 0.8488074662979606, "grad_norm": 0.6781094964379853, "learning_rate": 6.446702210618934e-06, "loss": 0.259, "step": 12278 }, { "epoch": 0.8488765986864846, "grad_norm": 0.7055087111050335, "learning_rate": 6.446166385329491e-06, "loss": 0.3974, "step": 12279 }, { "epoch": 0.8489457310750086, "grad_norm": 0.7751936500012255, "learning_rate": 6.445630541914932e-06, "loss": 0.3733, "step": 12280 }, { "epoch": 0.8490148634635326, "grad_norm": 0.6475761540942553, "learning_rate": 6.445094680381975e-06, "loss": 0.2898, "step": 12281 }, { "epoch": 0.8490839958520567, "grad_norm": 0.6349410498322894, "learning_rate": 6.444558800737335e-06, "loss": 0.2729, "step": 12282 }, { "epoch": 0.8491531282405808, "grad_norm": 0.7440927159834937, "learning_rate": 6.444022902987727e-06, "loss": 0.3688, "step": 12283 }, { "epoch": 0.8492222606291048, "grad_norm": 2.6137095039052514, "learning_rate": 6.443486987139869e-06, "loss": 0.2638, "step": 12284 }, { "epoch": 0.8492913930176288, "grad_norm": 0.8521496652070223, "learning_rate": 6.44295105320048e-06, "loss": 0.2471, "step": 12285 }, { "epoch": 0.8493605254061528, "grad_norm": 0.680415520821561, "learning_rate": 6.4424151011762735e-06, "loss": 0.3661, "step": 12286 }, { "epoch": 0.8494296577946768, "grad_norm": 0.6711435107496044, "learning_rate": 6.441879131073967e-06, "loss": 0.3539, "step": 12287 }, { "epoch": 0.8494987901832008, "grad_norm": 0.6250652500194933, "learning_rate": 6.441343142900278e-06, "loss": 0.2588, "step": 12288 }, { "epoch": 0.8495679225717249, "grad_norm": 0.624508281144131, "learning_rate": 6.440807136661927e-06, "loss": 0.3181, "step": 12289 }, { "epoch": 0.8496370549602489, "grad_norm": 0.6677217577906136, "learning_rate": 6.440271112365628e-06, "loss": 0.3671, "step": 12290 }, { "epoch": 0.8497061873487729, "grad_norm": 0.7200939519269224, "learning_rate": 6.439735070018103e-06, "loss": 0.3302, "step": 12291 }, { "epoch": 0.8497753197372969, "grad_norm": 0.609773264979216, "learning_rate": 6.439199009626066e-06, "loss": 0.2263, "step": 12292 }, { "epoch": 0.8498444521258209, "grad_norm": 0.6887638529048365, "learning_rate": 6.438662931196239e-06, "loss": 0.342, "step": 12293 }, { "epoch": 0.849913584514345, "grad_norm": 0.6323057893628921, "learning_rate": 6.438126834735339e-06, "loss": 0.393, "step": 12294 }, { "epoch": 0.849982716902869, "grad_norm": 0.6677696407604902, "learning_rate": 6.437590720250087e-06, "loss": 0.3116, "step": 12295 }, { "epoch": 0.850051849291393, "grad_norm": 0.6597268874013648, "learning_rate": 6.437054587747199e-06, "loss": 0.3042, "step": 12296 }, { "epoch": 0.850120981679917, "grad_norm": 0.658985817553737, "learning_rate": 6.4365184372333965e-06, "loss": 0.3776, "step": 12297 }, { "epoch": 0.850190114068441, "grad_norm": 0.6444207299062857, "learning_rate": 6.435982268715399e-06, "loss": 0.3492, "step": 12298 }, { "epoch": 0.850259246456965, "grad_norm": 0.7459238219074787, "learning_rate": 6.435446082199927e-06, "loss": 0.287, "step": 12299 }, { "epoch": 0.8503283788454891, "grad_norm": 0.6233322343646983, "learning_rate": 6.434909877693701e-06, "loss": 0.2449, "step": 12300 }, { "epoch": 0.8503975112340131, "grad_norm": 0.6479660629434046, "learning_rate": 6.434373655203437e-06, "loss": 0.4402, "step": 12301 }, { "epoch": 0.8504666436225372, "grad_norm": 0.6494530474498884, "learning_rate": 6.433837414735862e-06, "loss": 0.2969, "step": 12302 }, { "epoch": 0.8505357760110612, "grad_norm": 0.6071912178917498, "learning_rate": 6.433301156297693e-06, "loss": 0.2141, "step": 12303 }, { "epoch": 0.8506049083995852, "grad_norm": 0.7791604072223982, "learning_rate": 6.4327648798956525e-06, "loss": 0.3883, "step": 12304 }, { "epoch": 0.8506740407881093, "grad_norm": 0.7349382356474261, "learning_rate": 6.43222858553646e-06, "loss": 0.4145, "step": 12305 }, { "epoch": 0.8507431731766333, "grad_norm": 0.6005171189502483, "learning_rate": 6.431692273226838e-06, "loss": 0.283, "step": 12306 }, { "epoch": 0.8508123055651573, "grad_norm": 0.650279540107994, "learning_rate": 6.43115594297351e-06, "loss": 0.2682, "step": 12307 }, { "epoch": 0.8508814379536813, "grad_norm": 0.8500869485829274, "learning_rate": 6.430619594783194e-06, "loss": 0.3932, "step": 12308 }, { "epoch": 0.8509505703422053, "grad_norm": 0.7115007703283311, "learning_rate": 6.4300832286626155e-06, "loss": 0.283, "step": 12309 }, { "epoch": 0.8510197027307294, "grad_norm": 0.6560406892164016, "learning_rate": 6.429546844618497e-06, "loss": 0.2917, "step": 12310 }, { "epoch": 0.8510888351192534, "grad_norm": 0.6243363717359826, "learning_rate": 6.4290104426575595e-06, "loss": 0.3005, "step": 12311 }, { "epoch": 0.8511579675077774, "grad_norm": 0.6445739516258919, "learning_rate": 6.428474022786524e-06, "loss": 0.2923, "step": 12312 }, { "epoch": 0.8512270998963014, "grad_norm": 1.8431266582846215, "learning_rate": 6.427937585012119e-06, "loss": 0.3095, "step": 12313 }, { "epoch": 0.8512962322848254, "grad_norm": 0.7514345839482639, "learning_rate": 6.427401129341062e-06, "loss": 0.3085, "step": 12314 }, { "epoch": 0.8513653646733494, "grad_norm": 0.7632415665800547, "learning_rate": 6.4268646557800805e-06, "loss": 0.3596, "step": 12315 }, { "epoch": 0.8514344970618735, "grad_norm": 0.6958900238294182, "learning_rate": 6.426328164335897e-06, "loss": 0.3084, "step": 12316 }, { "epoch": 0.8515036294503975, "grad_norm": 0.6091983921169956, "learning_rate": 6.425791655015234e-06, "loss": 0.2573, "step": 12317 }, { "epoch": 0.8515727618389215, "grad_norm": 0.6308555253570551, "learning_rate": 6.425255127824819e-06, "loss": 0.2965, "step": 12318 }, { "epoch": 0.8516418942274455, "grad_norm": 0.7654321985808222, "learning_rate": 6.424718582771373e-06, "loss": 0.4516, "step": 12319 }, { "epoch": 0.8517110266159695, "grad_norm": 0.6642107094449808, "learning_rate": 6.4241820198616234e-06, "loss": 0.2821, "step": 12320 }, { "epoch": 0.8517801590044936, "grad_norm": 0.5937573615616387, "learning_rate": 6.423645439102293e-06, "loss": 0.277, "step": 12321 }, { "epoch": 0.8518492913930177, "grad_norm": 0.7092060731427952, "learning_rate": 6.423108840500107e-06, "loss": 0.3525, "step": 12322 }, { "epoch": 0.8519184237815417, "grad_norm": 0.6862508333570555, "learning_rate": 6.422572224061791e-06, "loss": 0.3805, "step": 12323 }, { "epoch": 0.8519875561700657, "grad_norm": 0.7331976751231979, "learning_rate": 6.422035589794073e-06, "loss": 0.2783, "step": 12324 }, { "epoch": 0.8520566885585897, "grad_norm": 0.6979043710294225, "learning_rate": 6.421498937703676e-06, "loss": 0.2914, "step": 12325 }, { "epoch": 0.8521258209471138, "grad_norm": 0.8033610742054489, "learning_rate": 6.420962267797325e-06, "loss": 0.441, "step": 12326 }, { "epoch": 0.8521949533356378, "grad_norm": 0.6570671806049648, "learning_rate": 6.420425580081749e-06, "loss": 0.331, "step": 12327 }, { "epoch": 0.8522640857241618, "grad_norm": 0.6367231598292975, "learning_rate": 6.4198888745636736e-06, "loss": 0.2353, "step": 12328 }, { "epoch": 0.8523332181126858, "grad_norm": 0.7469585585273271, "learning_rate": 6.419352151249824e-06, "loss": 0.4535, "step": 12329 }, { "epoch": 0.8524023505012098, "grad_norm": 0.6471741339268957, "learning_rate": 6.418815410146928e-06, "loss": 0.3466, "step": 12330 }, { "epoch": 0.8524714828897338, "grad_norm": 0.5811028559136096, "learning_rate": 6.418278651261714e-06, "loss": 0.2061, "step": 12331 }, { "epoch": 0.8525406152782579, "grad_norm": 0.8714270962248287, "learning_rate": 6.417741874600908e-06, "loss": 0.2611, "step": 12332 }, { "epoch": 0.8526097476667819, "grad_norm": 0.6285024826036075, "learning_rate": 6.417205080171237e-06, "loss": 0.3646, "step": 12333 }, { "epoch": 0.8526788800553059, "grad_norm": 0.7022578151515262, "learning_rate": 6.41666826797943e-06, "loss": 0.2895, "step": 12334 }, { "epoch": 0.8527480124438299, "grad_norm": 0.6935939125866183, "learning_rate": 6.416131438032214e-06, "loss": 0.2588, "step": 12335 }, { "epoch": 0.8528171448323539, "grad_norm": 0.8327218471614382, "learning_rate": 6.415594590336318e-06, "loss": 0.3842, "step": 12336 }, { "epoch": 0.852886277220878, "grad_norm": 0.7296229598583169, "learning_rate": 6.415057724898469e-06, "loss": 0.3856, "step": 12337 }, { "epoch": 0.852955409609402, "grad_norm": 0.7391935982579506, "learning_rate": 6.414520841725398e-06, "loss": 0.3074, "step": 12338 }, { "epoch": 0.853024541997926, "grad_norm": 1.1804347727152156, "learning_rate": 6.413983940823832e-06, "loss": 0.3149, "step": 12339 }, { "epoch": 0.85309367438645, "grad_norm": 0.7192001020990894, "learning_rate": 6.413447022200502e-06, "loss": 0.3796, "step": 12340 }, { "epoch": 0.853162806774974, "grad_norm": 0.5943797188700973, "learning_rate": 6.412910085862134e-06, "loss": 0.2935, "step": 12341 }, { "epoch": 0.8532319391634982, "grad_norm": 0.7038816550666565, "learning_rate": 6.412373131815461e-06, "loss": 0.2712, "step": 12342 }, { "epoch": 0.8533010715520222, "grad_norm": 0.7456657675208851, "learning_rate": 6.41183616006721e-06, "loss": 0.3003, "step": 12343 }, { "epoch": 0.8533702039405462, "grad_norm": 0.6578488716708408, "learning_rate": 6.411299170624113e-06, "loss": 0.3872, "step": 12344 }, { "epoch": 0.8534393363290702, "grad_norm": 0.6485076880630796, "learning_rate": 6.410762163492899e-06, "loss": 0.2925, "step": 12345 }, { "epoch": 0.8535084687175942, "grad_norm": 0.64423000309923, "learning_rate": 6.410225138680301e-06, "loss": 0.3382, "step": 12346 }, { "epoch": 0.8535776011061182, "grad_norm": 0.6921619273713352, "learning_rate": 6.409688096193044e-06, "loss": 0.4073, "step": 12347 }, { "epoch": 0.8536467334946423, "grad_norm": 0.6867076404599882, "learning_rate": 6.409151036037864e-06, "loss": 0.3703, "step": 12348 }, { "epoch": 0.8537158658831663, "grad_norm": 0.9922044429834549, "learning_rate": 6.4086139582214914e-06, "loss": 0.2323, "step": 12349 }, { "epoch": 0.8537849982716903, "grad_norm": 0.6982545163072491, "learning_rate": 6.4080768627506565e-06, "loss": 0.355, "step": 12350 }, { "epoch": 0.8538541306602143, "grad_norm": 0.6331529277632207, "learning_rate": 6.407539749632089e-06, "loss": 0.4441, "step": 12351 }, { "epoch": 0.8539232630487383, "grad_norm": 0.6363856424922897, "learning_rate": 6.407002618872524e-06, "loss": 0.2639, "step": 12352 }, { "epoch": 0.8539923954372624, "grad_norm": 0.7562450682076348, "learning_rate": 6.406465470478693e-06, "loss": 0.3017, "step": 12353 }, { "epoch": 0.8540615278257864, "grad_norm": 0.7584745183756623, "learning_rate": 6.405928304457325e-06, "loss": 0.3774, "step": 12354 }, { "epoch": 0.8541306602143104, "grad_norm": 0.6966308860083699, "learning_rate": 6.405391120815156e-06, "loss": 0.3424, "step": 12355 }, { "epoch": 0.8541997926028344, "grad_norm": 0.6536525822587985, "learning_rate": 6.404853919558918e-06, "loss": 0.3023, "step": 12356 }, { "epoch": 0.8542689249913584, "grad_norm": 0.748483635425155, "learning_rate": 6.404316700695344e-06, "loss": 0.2688, "step": 12357 }, { "epoch": 0.8543380573798824, "grad_norm": 0.8584980316452373, "learning_rate": 6.403779464231163e-06, "loss": 0.3744, "step": 12358 }, { "epoch": 0.8544071897684065, "grad_norm": 0.664422431215246, "learning_rate": 6.403242210173114e-06, "loss": 0.3151, "step": 12359 }, { "epoch": 0.8544763221569305, "grad_norm": 0.5938564272192011, "learning_rate": 6.402704938527927e-06, "loss": 0.2708, "step": 12360 }, { "epoch": 0.8545454545454545, "grad_norm": 0.6681539485177184, "learning_rate": 6.402167649302337e-06, "loss": 0.3016, "step": 12361 }, { "epoch": 0.8546145869339785, "grad_norm": 0.7398646958324183, "learning_rate": 6.401630342503078e-06, "loss": 0.4021, "step": 12362 }, { "epoch": 0.8546837193225026, "grad_norm": 0.6334278493036692, "learning_rate": 6.401093018136885e-06, "loss": 0.276, "step": 12363 }, { "epoch": 0.8547528517110267, "grad_norm": 0.7017595519817841, "learning_rate": 6.40055567621049e-06, "loss": 0.3262, "step": 12364 }, { "epoch": 0.8548219840995507, "grad_norm": 0.6469251324362431, "learning_rate": 6.40001831673063e-06, "loss": 0.4106, "step": 12365 }, { "epoch": 0.8548911164880747, "grad_norm": 0.8060412006831549, "learning_rate": 6.399480939704038e-06, "loss": 0.2919, "step": 12366 }, { "epoch": 0.8549602488765987, "grad_norm": 0.7521714567059694, "learning_rate": 6.39894354513745e-06, "loss": 0.2553, "step": 12367 }, { "epoch": 0.8550293812651227, "grad_norm": 0.6848048037630257, "learning_rate": 6.398406133037602e-06, "loss": 0.3113, "step": 12368 }, { "epoch": 0.8550985136536468, "grad_norm": 0.666514549935682, "learning_rate": 6.3978687034112276e-06, "loss": 0.3955, "step": 12369 }, { "epoch": 0.8551676460421708, "grad_norm": 0.6174094486623435, "learning_rate": 6.397331256265063e-06, "loss": 0.2686, "step": 12370 }, { "epoch": 0.8552367784306948, "grad_norm": 0.7445770270550554, "learning_rate": 6.396793791605847e-06, "loss": 0.3112, "step": 12371 }, { "epoch": 0.8553059108192188, "grad_norm": 0.8687645945291897, "learning_rate": 6.396256309440312e-06, "loss": 0.3609, "step": 12372 }, { "epoch": 0.8553750432077428, "grad_norm": 0.6980485994601106, "learning_rate": 6.3957188097751965e-06, "loss": 0.3961, "step": 12373 }, { "epoch": 0.8554441755962668, "grad_norm": 0.6202239297822407, "learning_rate": 6.395181292617236e-06, "loss": 0.3211, "step": 12374 }, { "epoch": 0.8555133079847909, "grad_norm": 0.7113268211240911, "learning_rate": 6.394643757973169e-06, "loss": 0.3178, "step": 12375 }, { "epoch": 0.8555824403733149, "grad_norm": 0.8663532463090471, "learning_rate": 6.39410620584973e-06, "loss": 0.428, "step": 12376 }, { "epoch": 0.8556515727618389, "grad_norm": 0.800755562536734, "learning_rate": 6.393568636253658e-06, "loss": 0.3679, "step": 12377 }, { "epoch": 0.8557207051503629, "grad_norm": 0.6070841039853004, "learning_rate": 6.393031049191689e-06, "loss": 0.2682, "step": 12378 }, { "epoch": 0.8557898375388869, "grad_norm": 0.6754607987621433, "learning_rate": 6.392493444670562e-06, "loss": 0.3445, "step": 12379 }, { "epoch": 0.855858969927411, "grad_norm": 0.7588734690460966, "learning_rate": 6.391955822697017e-06, "loss": 0.3575, "step": 12380 }, { "epoch": 0.855928102315935, "grad_norm": 0.8031615522863795, "learning_rate": 6.391418183277789e-06, "loss": 0.2305, "step": 12381 }, { "epoch": 0.855997234704459, "grad_norm": 0.6589731390751985, "learning_rate": 6.390880526419616e-06, "loss": 0.3084, "step": 12382 }, { "epoch": 0.8560663670929831, "grad_norm": 0.6918141242300432, "learning_rate": 6.390342852129237e-06, "loss": 0.3513, "step": 12383 }, { "epoch": 0.8561354994815071, "grad_norm": 0.8701109668131416, "learning_rate": 6.389805160413394e-06, "loss": 0.2932, "step": 12384 }, { "epoch": 0.8562046318700312, "grad_norm": 0.7266459721205293, "learning_rate": 6.389267451278822e-06, "loss": 0.3396, "step": 12385 }, { "epoch": 0.8562737642585552, "grad_norm": 0.6684733329821098, "learning_rate": 6.388729724732261e-06, "loss": 0.3425, "step": 12386 }, { "epoch": 0.8563428966470792, "grad_norm": 0.6648263573566653, "learning_rate": 6.388191980780454e-06, "loss": 0.3559, "step": 12387 }, { "epoch": 0.8564120290356032, "grad_norm": 0.5561448238489691, "learning_rate": 6.387654219430136e-06, "loss": 0.2711, "step": 12388 }, { "epoch": 0.8564811614241272, "grad_norm": 0.7689425163325058, "learning_rate": 6.38711644068805e-06, "loss": 0.3376, "step": 12389 }, { "epoch": 0.8565502938126512, "grad_norm": 0.7188278438909377, "learning_rate": 6.386578644560932e-06, "loss": 0.3728, "step": 12390 }, { "epoch": 0.8566194262011753, "grad_norm": 0.7482264717973698, "learning_rate": 6.386040831055526e-06, "loss": 0.3185, "step": 12391 }, { "epoch": 0.8566885585896993, "grad_norm": 0.6834420586705353, "learning_rate": 6.385503000178575e-06, "loss": 0.3137, "step": 12392 }, { "epoch": 0.8567576909782233, "grad_norm": 1.0049935504650511, "learning_rate": 6.384965151936813e-06, "loss": 0.3021, "step": 12393 }, { "epoch": 0.8568268233667473, "grad_norm": 0.719234375940106, "learning_rate": 6.384427286336985e-06, "loss": 0.3551, "step": 12394 }, { "epoch": 0.8568959557552713, "grad_norm": 0.7686392995350053, "learning_rate": 6.383889403385833e-06, "loss": 0.2584, "step": 12395 }, { "epoch": 0.8569650881437954, "grad_norm": 0.748103144905704, "learning_rate": 6.383351503090095e-06, "loss": 0.2785, "step": 12396 }, { "epoch": 0.8570342205323194, "grad_norm": 0.6663658709819542, "learning_rate": 6.382813585456516e-06, "loss": 0.3953, "step": 12397 }, { "epoch": 0.8571033529208434, "grad_norm": 0.6579147164141317, "learning_rate": 6.382275650491836e-06, "loss": 0.3596, "step": 12398 }, { "epoch": 0.8571724853093674, "grad_norm": 0.6144450574908985, "learning_rate": 6.381737698202797e-06, "loss": 0.2821, "step": 12399 }, { "epoch": 0.8572416176978914, "grad_norm": 0.6517046157642117, "learning_rate": 6.3811997285961425e-06, "loss": 0.3069, "step": 12400 }, { "epoch": 0.8573107500864154, "grad_norm": 0.6455572522008819, "learning_rate": 6.380661741678613e-06, "loss": 0.3957, "step": 12401 }, { "epoch": 0.8573798824749395, "grad_norm": 0.6428065156472098, "learning_rate": 6.380123737456955e-06, "loss": 0.2926, "step": 12402 }, { "epoch": 0.8574490148634636, "grad_norm": 0.7489799957980188, "learning_rate": 6.379585715937907e-06, "loss": 0.3292, "step": 12403 }, { "epoch": 0.8575181472519876, "grad_norm": 0.7033304873880825, "learning_rate": 6.379047677128214e-06, "loss": 0.3428, "step": 12404 }, { "epoch": 0.8575872796405116, "grad_norm": 0.7043874605751861, "learning_rate": 6.37850962103462e-06, "loss": 0.2989, "step": 12405 }, { "epoch": 0.8576564120290356, "grad_norm": 0.6312423301606751, "learning_rate": 6.377971547663868e-06, "loss": 0.278, "step": 12406 }, { "epoch": 0.8577255444175597, "grad_norm": 0.8347727644345803, "learning_rate": 6.377433457022701e-06, "loss": 0.3484, "step": 12407 }, { "epoch": 0.8577946768060837, "grad_norm": 0.705773372758531, "learning_rate": 6.376895349117865e-06, "loss": 0.4194, "step": 12408 }, { "epoch": 0.8578638091946077, "grad_norm": 0.7271878471706771, "learning_rate": 6.376357223956104e-06, "loss": 0.2978, "step": 12409 }, { "epoch": 0.8579329415831317, "grad_norm": 0.7202952367324487, "learning_rate": 6.375819081544159e-06, "loss": 0.3216, "step": 12410 }, { "epoch": 0.8580020739716557, "grad_norm": 0.6724821938292308, "learning_rate": 6.375280921888777e-06, "loss": 0.3434, "step": 12411 }, { "epoch": 0.8580712063601798, "grad_norm": 0.6487181531732331, "learning_rate": 6.374742744996705e-06, "loss": 0.3298, "step": 12412 }, { "epoch": 0.8581403387487038, "grad_norm": 0.6653659965008663, "learning_rate": 6.374204550874684e-06, "loss": 0.292, "step": 12413 }, { "epoch": 0.8582094711372278, "grad_norm": 0.7063012275945192, "learning_rate": 6.373666339529462e-06, "loss": 0.2878, "step": 12414 }, { "epoch": 0.8582786035257518, "grad_norm": 0.710103017495951, "learning_rate": 6.373128110967783e-06, "loss": 0.4425, "step": 12415 }, { "epoch": 0.8583477359142758, "grad_norm": 0.7618311458866508, "learning_rate": 6.372589865196396e-06, "loss": 0.2942, "step": 12416 }, { "epoch": 0.8584168683027998, "grad_norm": 0.6173860987892458, "learning_rate": 6.372051602222043e-06, "loss": 0.2902, "step": 12417 }, { "epoch": 0.8584860006913239, "grad_norm": 0.5513481458226752, "learning_rate": 6.371513322051471e-06, "loss": 0.2617, "step": 12418 }, { "epoch": 0.8585551330798479, "grad_norm": 0.70173455866358, "learning_rate": 6.3709750246914284e-06, "loss": 0.4374, "step": 12419 }, { "epoch": 0.8586242654683719, "grad_norm": 0.6038057649487293, "learning_rate": 6.37043671014866e-06, "loss": 0.254, "step": 12420 }, { "epoch": 0.8586933978568959, "grad_norm": 0.763632657683077, "learning_rate": 6.369898378429913e-06, "loss": 0.3315, "step": 12421 }, { "epoch": 0.8587625302454199, "grad_norm": 0.7929189493513942, "learning_rate": 6.369360029541934e-06, "loss": 0.3893, "step": 12422 }, { "epoch": 0.858831662633944, "grad_norm": 0.6502379254011188, "learning_rate": 6.368821663491472e-06, "loss": 0.3383, "step": 12423 }, { "epoch": 0.8589007950224681, "grad_norm": 0.8406752765131572, "learning_rate": 6.368283280285272e-06, "loss": 0.3301, "step": 12424 }, { "epoch": 0.8589699274109921, "grad_norm": 0.6611131134927082, "learning_rate": 6.3677448799300836e-06, "loss": 0.2832, "step": 12425 }, { "epoch": 0.8590390597995161, "grad_norm": 0.6894907645773377, "learning_rate": 6.367206462432653e-06, "loss": 0.3766, "step": 12426 }, { "epoch": 0.8591081921880401, "grad_norm": 0.6680906749802932, "learning_rate": 6.366668027799732e-06, "loss": 0.2595, "step": 12427 }, { "epoch": 0.8591773245765641, "grad_norm": 0.6276529203563775, "learning_rate": 6.366129576038064e-06, "loss": 0.3473, "step": 12428 }, { "epoch": 0.8592464569650882, "grad_norm": 0.6788300403747993, "learning_rate": 6.3655911071543986e-06, "loss": 0.4052, "step": 12429 }, { "epoch": 0.8593155893536122, "grad_norm": 0.6785725082404754, "learning_rate": 6.3650526211554875e-06, "loss": 0.3352, "step": 12430 }, { "epoch": 0.8593847217421362, "grad_norm": 0.6380478903779284, "learning_rate": 6.364514118048077e-06, "loss": 0.2965, "step": 12431 }, { "epoch": 0.8594538541306602, "grad_norm": 0.707497684287512, "learning_rate": 6.363975597838917e-06, "loss": 0.3484, "step": 12432 }, { "epoch": 0.8595229865191842, "grad_norm": 0.7076699590296499, "learning_rate": 6.363437060534758e-06, "loss": 0.426, "step": 12433 }, { "epoch": 0.8595921189077083, "grad_norm": 0.8338449695733047, "learning_rate": 6.362898506142349e-06, "loss": 0.3726, "step": 12434 }, { "epoch": 0.8596612512962323, "grad_norm": 0.6136471075268058, "learning_rate": 6.362359934668439e-06, "loss": 0.2655, "step": 12435 }, { "epoch": 0.8597303836847563, "grad_norm": 0.7169480624269281, "learning_rate": 6.361821346119776e-06, "loss": 0.3773, "step": 12436 }, { "epoch": 0.8597995160732803, "grad_norm": 0.9902343276591887, "learning_rate": 6.3612827405031164e-06, "loss": 0.4088, "step": 12437 }, { "epoch": 0.8598686484618043, "grad_norm": 0.6130096543638722, "learning_rate": 6.360744117825205e-06, "loss": 0.2899, "step": 12438 }, { "epoch": 0.8599377808503283, "grad_norm": 0.6823281648874078, "learning_rate": 6.360205478092796e-06, "loss": 0.288, "step": 12439 }, { "epoch": 0.8600069132388524, "grad_norm": 0.7083579445973427, "learning_rate": 6.359666821312636e-06, "loss": 0.4038, "step": 12440 }, { "epoch": 0.8600760456273764, "grad_norm": 0.728032336476844, "learning_rate": 6.35912814749148e-06, "loss": 0.319, "step": 12441 }, { "epoch": 0.8601451780159004, "grad_norm": 0.6367881726124803, "learning_rate": 6.35858945663608e-06, "loss": 0.2489, "step": 12442 }, { "epoch": 0.8602143104044244, "grad_norm": 0.7524304422252821, "learning_rate": 6.358050748753183e-06, "loss": 0.3222, "step": 12443 }, { "epoch": 0.8602834427929485, "grad_norm": 0.6479978450195994, "learning_rate": 6.357512023849545e-06, "loss": 0.381, "step": 12444 }, { "epoch": 0.8603525751814726, "grad_norm": 0.725265668227606, "learning_rate": 6.356973281931916e-06, "loss": 0.3071, "step": 12445 }, { "epoch": 0.8604217075699966, "grad_norm": 0.7111728872685947, "learning_rate": 6.356434523007049e-06, "loss": 0.2413, "step": 12446 }, { "epoch": 0.8604908399585206, "grad_norm": 0.6120134939465585, "learning_rate": 6.355895747081694e-06, "loss": 0.3651, "step": 12447 }, { "epoch": 0.8605599723470446, "grad_norm": 0.6751303314964118, "learning_rate": 6.355356954162608e-06, "loss": 0.3562, "step": 12448 }, { "epoch": 0.8606291047355686, "grad_norm": 0.6129260685161402, "learning_rate": 6.35481814425654e-06, "loss": 0.275, "step": 12449 }, { "epoch": 0.8606982371240927, "grad_norm": 0.6596333038347589, "learning_rate": 6.354279317370245e-06, "loss": 0.2715, "step": 12450 }, { "epoch": 0.8607673695126167, "grad_norm": 0.7398335601142154, "learning_rate": 6.353740473510475e-06, "loss": 0.3947, "step": 12451 }, { "epoch": 0.8608365019011407, "grad_norm": 0.7155982271813532, "learning_rate": 6.353201612683983e-06, "loss": 0.2407, "step": 12452 }, { "epoch": 0.8609056342896647, "grad_norm": 0.6493086256422056, "learning_rate": 6.352662734897525e-06, "loss": 0.2539, "step": 12453 }, { "epoch": 0.8609747666781887, "grad_norm": 0.6695993969646232, "learning_rate": 6.352123840157854e-06, "loss": 0.3467, "step": 12454 }, { "epoch": 0.8610438990667127, "grad_norm": 0.7923778009863877, "learning_rate": 6.351584928471723e-06, "loss": 0.3631, "step": 12455 }, { "epoch": 0.8611130314552368, "grad_norm": 0.6449544432103514, "learning_rate": 6.351045999845887e-06, "loss": 0.2628, "step": 12456 }, { "epoch": 0.8611821638437608, "grad_norm": 0.7687655664419022, "learning_rate": 6.350507054287101e-06, "loss": 0.295, "step": 12457 }, { "epoch": 0.8612512962322848, "grad_norm": 0.7117715984501803, "learning_rate": 6.349968091802118e-06, "loss": 0.4264, "step": 12458 }, { "epoch": 0.8613204286208088, "grad_norm": 0.6714843986704225, "learning_rate": 6.349429112397695e-06, "loss": 0.3176, "step": 12459 }, { "epoch": 0.8613895610093328, "grad_norm": 0.6327762685856206, "learning_rate": 6.348890116080586e-06, "loss": 0.2723, "step": 12460 }, { "epoch": 0.8614586933978569, "grad_norm": 0.6960921504576575, "learning_rate": 6.348351102857545e-06, "loss": 0.3434, "step": 12461 }, { "epoch": 0.8615278257863809, "grad_norm": 0.6995813047574251, "learning_rate": 6.347812072735331e-06, "loss": 0.4629, "step": 12462 }, { "epoch": 0.8615969581749049, "grad_norm": 0.5880021281275375, "learning_rate": 6.347273025720698e-06, "loss": 0.3156, "step": 12463 }, { "epoch": 0.861666090563429, "grad_norm": 0.6638651035797445, "learning_rate": 6.346733961820402e-06, "loss": 0.3221, "step": 12464 }, { "epoch": 0.861735222951953, "grad_norm": 0.6559866743351673, "learning_rate": 6.346194881041198e-06, "loss": 0.395, "step": 12465 }, { "epoch": 0.8618043553404771, "grad_norm": 0.5895850923948828, "learning_rate": 6.345655783389845e-06, "loss": 0.2503, "step": 12466 }, { "epoch": 0.8618734877290011, "grad_norm": 0.7402509643794454, "learning_rate": 6.345116668873097e-06, "loss": 0.2823, "step": 12467 }, { "epoch": 0.8619426201175251, "grad_norm": 0.64397250700377, "learning_rate": 6.344577537497712e-06, "loss": 0.3342, "step": 12468 }, { "epoch": 0.8620117525060491, "grad_norm": 0.6616971687887742, "learning_rate": 6.344038389270448e-06, "loss": 0.3705, "step": 12469 }, { "epoch": 0.8620808848945731, "grad_norm": 0.6091398818740046, "learning_rate": 6.343499224198061e-06, "loss": 0.321, "step": 12470 }, { "epoch": 0.8621500172830971, "grad_norm": 0.7827502860088348, "learning_rate": 6.342960042287309e-06, "loss": 0.2468, "step": 12471 }, { "epoch": 0.8622191496716212, "grad_norm": 0.6907166946376541, "learning_rate": 6.342420843544949e-06, "loss": 0.3819, "step": 12472 }, { "epoch": 0.8622882820601452, "grad_norm": 0.6533668250791975, "learning_rate": 6.341881627977741e-06, "loss": 0.3628, "step": 12473 }, { "epoch": 0.8623574144486692, "grad_norm": 0.5971507230427077, "learning_rate": 6.341342395592441e-06, "loss": 0.2865, "step": 12474 }, { "epoch": 0.8624265468371932, "grad_norm": 1.0166543926459952, "learning_rate": 6.340803146395804e-06, "loss": 0.301, "step": 12475 }, { "epoch": 0.8624956792257172, "grad_norm": 0.6957465573190253, "learning_rate": 6.340263880394596e-06, "loss": 0.4461, "step": 12476 }, { "epoch": 0.8625648116142413, "grad_norm": 0.6304259426086513, "learning_rate": 6.3397245975955715e-06, "loss": 0.2794, "step": 12477 }, { "epoch": 0.8626339440027653, "grad_norm": 0.6007273746061514, "learning_rate": 6.339185298005489e-06, "loss": 0.3098, "step": 12478 }, { "epoch": 0.8627030763912893, "grad_norm": 0.6680046793098644, "learning_rate": 6.3386459816311094e-06, "loss": 0.3712, "step": 12479 }, { "epoch": 0.8627722087798133, "grad_norm": 0.7622700139354449, "learning_rate": 6.338106648479192e-06, "loss": 0.2927, "step": 12480 }, { "epoch": 0.8628413411683373, "grad_norm": 0.761197797846736, "learning_rate": 6.337567298556495e-06, "loss": 0.3352, "step": 12481 }, { "epoch": 0.8629104735568613, "grad_norm": 0.6819203352376323, "learning_rate": 6.3370279318697794e-06, "loss": 0.3194, "step": 12482 }, { "epoch": 0.8629796059453854, "grad_norm": 0.6897991279182026, "learning_rate": 6.336488548425804e-06, "loss": 0.4665, "step": 12483 }, { "epoch": 0.8630487383339094, "grad_norm": 0.8191512270905272, "learning_rate": 6.33594914823133e-06, "loss": 0.3876, "step": 12484 }, { "epoch": 0.8631178707224335, "grad_norm": 0.6506204843426344, "learning_rate": 6.335409731293118e-06, "loss": 0.2846, "step": 12485 }, { "epoch": 0.8631870031109575, "grad_norm": 0.7644642133521787, "learning_rate": 6.334870297617928e-06, "loss": 0.4273, "step": 12486 }, { "epoch": 0.8632561354994815, "grad_norm": 0.6775026252014652, "learning_rate": 6.33433084721252e-06, "loss": 0.3074, "step": 12487 }, { "epoch": 0.8633252678880056, "grad_norm": 0.6443227899837023, "learning_rate": 6.333791380083657e-06, "loss": 0.2376, "step": 12488 }, { "epoch": 0.8633944002765296, "grad_norm": 0.6511584969347819, "learning_rate": 6.333251896238098e-06, "loss": 0.2893, "step": 12489 }, { "epoch": 0.8634635326650536, "grad_norm": 0.6260252234039494, "learning_rate": 6.3327123956826075e-06, "loss": 0.354, "step": 12490 }, { "epoch": 0.8635326650535776, "grad_norm": 0.6121419600753836, "learning_rate": 6.332172878423945e-06, "loss": 0.2991, "step": 12491 }, { "epoch": 0.8636017974421016, "grad_norm": 0.7098806805826611, "learning_rate": 6.331633344468873e-06, "loss": 0.324, "step": 12492 }, { "epoch": 0.8636709298306257, "grad_norm": 0.7414212954686127, "learning_rate": 6.331093793824152e-06, "loss": 0.326, "step": 12493 }, { "epoch": 0.8637400622191497, "grad_norm": 0.6859076036342832, "learning_rate": 6.330554226496548e-06, "loss": 0.3858, "step": 12494 }, { "epoch": 0.8638091946076737, "grad_norm": 0.6032788689792659, "learning_rate": 6.33001464249282e-06, "loss": 0.2713, "step": 12495 }, { "epoch": 0.8638783269961977, "grad_norm": 0.7133197004055021, "learning_rate": 6.329475041819731e-06, "loss": 0.2478, "step": 12496 }, { "epoch": 0.8639474593847217, "grad_norm": 1.0919471031266432, "learning_rate": 6.3289354244840466e-06, "loss": 0.3311, "step": 12497 }, { "epoch": 0.8640165917732457, "grad_norm": 0.6875078189827394, "learning_rate": 6.328395790492527e-06, "loss": 0.3616, "step": 12498 }, { "epoch": 0.8640857241617698, "grad_norm": 0.802850241448441, "learning_rate": 6.327856139851939e-06, "loss": 0.2764, "step": 12499 }, { "epoch": 0.8641548565502938, "grad_norm": 0.648935819100127, "learning_rate": 6.327316472569041e-06, "loss": 0.3274, "step": 12500 }, { "epoch": 0.8642239889388178, "grad_norm": 0.7044502934343054, "learning_rate": 6.326776788650601e-06, "loss": 0.3732, "step": 12501 }, { "epoch": 0.8642931213273418, "grad_norm": 0.6488636765375, "learning_rate": 6.326237088103383e-06, "loss": 0.248, "step": 12502 }, { "epoch": 0.8643622537158658, "grad_norm": 0.7464945455400719, "learning_rate": 6.325697370934148e-06, "loss": 0.3219, "step": 12503 }, { "epoch": 0.8644313861043899, "grad_norm": 0.7242831030564476, "learning_rate": 6.325157637149663e-06, "loss": 0.3856, "step": 12504 }, { "epoch": 0.864500518492914, "grad_norm": 0.7172316043884583, "learning_rate": 6.324617886756692e-06, "loss": 0.316, "step": 12505 }, { "epoch": 0.864569650881438, "grad_norm": 0.7969214928427051, "learning_rate": 6.324078119761999e-06, "loss": 0.3191, "step": 12506 }, { "epoch": 0.864638783269962, "grad_norm": 1.3297630891701855, "learning_rate": 6.32353833617235e-06, "loss": 0.3057, "step": 12507 }, { "epoch": 0.864707915658486, "grad_norm": 0.6661758615635831, "learning_rate": 6.3229985359945104e-06, "loss": 0.3974, "step": 12508 }, { "epoch": 0.86477704804701, "grad_norm": 0.8935173934438252, "learning_rate": 6.322458719235245e-06, "loss": 0.2729, "step": 12509 }, { "epoch": 0.8648461804355341, "grad_norm": 0.7407090533551817, "learning_rate": 6.3219188859013195e-06, "loss": 0.313, "step": 12510 }, { "epoch": 0.8649153128240581, "grad_norm": 0.6139512931658937, "learning_rate": 6.3213790359995e-06, "loss": 0.2941, "step": 12511 }, { "epoch": 0.8649844452125821, "grad_norm": 0.674473712756955, "learning_rate": 6.320839169536551e-06, "loss": 0.3318, "step": 12512 }, { "epoch": 0.8650535776011061, "grad_norm": 0.6389648997579703, "learning_rate": 6.320299286519243e-06, "loss": 0.298, "step": 12513 }, { "epoch": 0.8651227099896301, "grad_norm": 0.6971894831254359, "learning_rate": 6.319759386954336e-06, "loss": 0.3137, "step": 12514 }, { "epoch": 0.8651918423781542, "grad_norm": 0.7222673396466631, "learning_rate": 6.3192194708486035e-06, "loss": 0.395, "step": 12515 }, { "epoch": 0.8652609747666782, "grad_norm": 0.6508806743637692, "learning_rate": 6.3186795382088074e-06, "loss": 0.2836, "step": 12516 }, { "epoch": 0.8653301071552022, "grad_norm": 0.6315918008299662, "learning_rate": 6.318139589041717e-06, "loss": 0.2576, "step": 12517 }, { "epoch": 0.8653992395437262, "grad_norm": 0.6733162291981167, "learning_rate": 6.317599623354099e-06, "loss": 0.3567, "step": 12518 }, { "epoch": 0.8654683719322502, "grad_norm": 0.6422886029206449, "learning_rate": 6.317059641152722e-06, "loss": 0.4195, "step": 12519 }, { "epoch": 0.8655375043207743, "grad_norm": 0.6171253804800828, "learning_rate": 6.316519642444352e-06, "loss": 0.2997, "step": 12520 }, { "epoch": 0.8656066367092983, "grad_norm": 0.6035643547869776, "learning_rate": 6.315979627235757e-06, "loss": 0.3098, "step": 12521 }, { "epoch": 0.8656757690978223, "grad_norm": 0.7200392113819589, "learning_rate": 6.3154395955337064e-06, "loss": 0.3466, "step": 12522 }, { "epoch": 0.8657449014863463, "grad_norm": 0.7752871729843547, "learning_rate": 6.314899547344969e-06, "loss": 0.322, "step": 12523 }, { "epoch": 0.8658140338748703, "grad_norm": 0.6263796976312597, "learning_rate": 6.314359482676312e-06, "loss": 0.2753, "step": 12524 }, { "epoch": 0.8658831662633945, "grad_norm": 0.6799727607834894, "learning_rate": 6.313819401534503e-06, "loss": 0.3703, "step": 12525 }, { "epoch": 0.8659522986519185, "grad_norm": 0.6912096001604238, "learning_rate": 6.313279303926314e-06, "loss": 0.4649, "step": 12526 }, { "epoch": 0.8660214310404425, "grad_norm": 0.6698572005268709, "learning_rate": 6.312739189858512e-06, "loss": 0.2814, "step": 12527 }, { "epoch": 0.8660905634289665, "grad_norm": 0.6317350037294458, "learning_rate": 6.312199059337868e-06, "loss": 0.268, "step": 12528 }, { "epoch": 0.8661596958174905, "grad_norm": 0.6624825208002063, "learning_rate": 6.311658912371148e-06, "loss": 0.3483, "step": 12529 }, { "epoch": 0.8662288282060145, "grad_norm": 0.6177986954151625, "learning_rate": 6.311118748965126e-06, "loss": 0.3354, "step": 12530 }, { "epoch": 0.8662979605945386, "grad_norm": 0.5852439018413841, "learning_rate": 6.310578569126571e-06, "loss": 0.2772, "step": 12531 }, { "epoch": 0.8663670929830626, "grad_norm": 0.6774777551357859, "learning_rate": 6.31003837286225e-06, "loss": 0.3574, "step": 12532 }, { "epoch": 0.8664362253715866, "grad_norm": 0.6665599046089564, "learning_rate": 6.309498160178939e-06, "loss": 0.4689, "step": 12533 }, { "epoch": 0.8665053577601106, "grad_norm": 0.8130728500167231, "learning_rate": 6.308957931083404e-06, "loss": 0.2653, "step": 12534 }, { "epoch": 0.8665744901486346, "grad_norm": 0.6412129678264259, "learning_rate": 6.3084176855824174e-06, "loss": 0.2954, "step": 12535 }, { "epoch": 0.8666436225371587, "grad_norm": 0.767291013469895, "learning_rate": 6.3078774236827485e-06, "loss": 0.3542, "step": 12536 }, { "epoch": 0.8667127549256827, "grad_norm": 0.6922812539752161, "learning_rate": 6.307337145391172e-06, "loss": 0.3604, "step": 12537 }, { "epoch": 0.8667818873142067, "grad_norm": 0.6381315419174286, "learning_rate": 6.306796850714458e-06, "loss": 0.3218, "step": 12538 }, { "epoch": 0.8668510197027307, "grad_norm": 1.4986493529036424, "learning_rate": 6.3062565396593746e-06, "loss": 0.2647, "step": 12539 }, { "epoch": 0.8669201520912547, "grad_norm": 0.7369854763771342, "learning_rate": 6.3057162122327e-06, "loss": 0.4109, "step": 12540 }, { "epoch": 0.8669892844797787, "grad_norm": 1.151582516468076, "learning_rate": 6.305175868441201e-06, "loss": 0.3114, "step": 12541 }, { "epoch": 0.8670584168683028, "grad_norm": 0.6044547709631612, "learning_rate": 6.304635508291652e-06, "loss": 0.2681, "step": 12542 }, { "epoch": 0.8671275492568268, "grad_norm": 0.7451965071446518, "learning_rate": 6.304095131790825e-06, "loss": 0.3144, "step": 12543 }, { "epoch": 0.8671966816453508, "grad_norm": 0.6853427026070597, "learning_rate": 6.303554738945493e-06, "loss": 0.4408, "step": 12544 }, { "epoch": 0.8672658140338748, "grad_norm": 3.375182929752515, "learning_rate": 6.303014329762429e-06, "loss": 0.3011, "step": 12545 }, { "epoch": 0.867334946422399, "grad_norm": 0.8924805086899894, "learning_rate": 6.302473904248404e-06, "loss": 0.319, "step": 12546 }, { "epoch": 0.867404078810923, "grad_norm": 0.7523005192578857, "learning_rate": 6.3019334624101946e-06, "loss": 0.4489, "step": 12547 }, { "epoch": 0.867473211199447, "grad_norm": 0.6514001022588145, "learning_rate": 6.301393004254572e-06, "loss": 0.3272, "step": 12548 }, { "epoch": 0.867542343587971, "grad_norm": 0.6351658317642919, "learning_rate": 6.300852529788311e-06, "loss": 0.2548, "step": 12549 }, { "epoch": 0.867611475976495, "grad_norm": 0.6520557233713034, "learning_rate": 6.300312039018185e-06, "loss": 0.3252, "step": 12550 }, { "epoch": 0.867680608365019, "grad_norm": 0.7549952315613402, "learning_rate": 6.299771531950968e-06, "loss": 0.3909, "step": 12551 }, { "epoch": 0.867749740753543, "grad_norm": 0.6622095947154187, "learning_rate": 6.299231008593434e-06, "loss": 0.3159, "step": 12552 }, { "epoch": 0.8678188731420671, "grad_norm": 0.638372514208187, "learning_rate": 6.298690468952357e-06, "loss": 0.3086, "step": 12553 }, { "epoch": 0.8678880055305911, "grad_norm": 0.7077187998273065, "learning_rate": 6.298149913034515e-06, "loss": 0.4215, "step": 12554 }, { "epoch": 0.8679571379191151, "grad_norm": 0.6668571912247562, "learning_rate": 6.29760934084668e-06, "loss": 0.3407, "step": 12555 }, { "epoch": 0.8680262703076391, "grad_norm": 0.6454352787136962, "learning_rate": 6.297068752395625e-06, "loss": 0.293, "step": 12556 }, { "epoch": 0.8680954026961631, "grad_norm": 0.6877095439009014, "learning_rate": 6.29652814768813e-06, "loss": 0.3367, "step": 12557 }, { "epoch": 0.8681645350846872, "grad_norm": 0.8423299678891634, "learning_rate": 6.295987526730969e-06, "loss": 0.3951, "step": 12558 }, { "epoch": 0.8682336674732112, "grad_norm": 0.620523945836674, "learning_rate": 6.295446889530916e-06, "loss": 0.2936, "step": 12559 }, { "epoch": 0.8683027998617352, "grad_norm": 0.6488481508655767, "learning_rate": 6.294906236094749e-06, "loss": 0.2597, "step": 12560 }, { "epoch": 0.8683719322502592, "grad_norm": 0.6499102611083581, "learning_rate": 6.2943655664292425e-06, "loss": 0.3532, "step": 12561 }, { "epoch": 0.8684410646387832, "grad_norm": 0.7001200859695065, "learning_rate": 6.293824880541174e-06, "loss": 0.405, "step": 12562 }, { "epoch": 0.8685101970273073, "grad_norm": 0.6327038535170807, "learning_rate": 6.293284178437319e-06, "loss": 0.279, "step": 12563 }, { "epoch": 0.8685793294158313, "grad_norm": 0.6177062178752727, "learning_rate": 6.292743460124454e-06, "loss": 0.2444, "step": 12564 }, { "epoch": 0.8686484618043553, "grad_norm": 0.7203329180801031, "learning_rate": 6.292202725609359e-06, "loss": 0.426, "step": 12565 }, { "epoch": 0.8687175941928794, "grad_norm": 0.6610097331556448, "learning_rate": 6.291661974898808e-06, "loss": 0.3113, "step": 12566 }, { "epoch": 0.8687867265814034, "grad_norm": 0.6717520362173882, "learning_rate": 6.291121207999578e-06, "loss": 0.2395, "step": 12567 }, { "epoch": 0.8688558589699275, "grad_norm": 0.917006855542997, "learning_rate": 6.2905804249184485e-06, "loss": 0.2637, "step": 12568 }, { "epoch": 0.8689249913584515, "grad_norm": 0.7058001735069638, "learning_rate": 6.290039625662197e-06, "loss": 0.4322, "step": 12569 }, { "epoch": 0.8689941237469755, "grad_norm": 0.6939120174121042, "learning_rate": 6.2894988102376e-06, "loss": 0.3029, "step": 12570 }, { "epoch": 0.8690632561354995, "grad_norm": 0.6775573258866939, "learning_rate": 6.288957978651436e-06, "loss": 0.3246, "step": 12571 }, { "epoch": 0.8691323885240235, "grad_norm": 0.7790207997620022, "learning_rate": 6.288417130910487e-06, "loss": 0.3252, "step": 12572 }, { "epoch": 0.8692015209125475, "grad_norm": 0.6229716230472173, "learning_rate": 6.2878762670215265e-06, "loss": 0.2998, "step": 12573 }, { "epoch": 0.8692706533010716, "grad_norm": 0.6616108694532777, "learning_rate": 6.2873353869913345e-06, "loss": 0.3193, "step": 12574 }, { "epoch": 0.8693397856895956, "grad_norm": 0.7006535113744352, "learning_rate": 6.286794490826692e-06, "loss": 0.3697, "step": 12575 }, { "epoch": 0.8694089180781196, "grad_norm": 0.8188430721085032, "learning_rate": 6.286253578534377e-06, "loss": 0.4207, "step": 12576 }, { "epoch": 0.8694780504666436, "grad_norm": 0.6571031400831232, "learning_rate": 6.2857126501211695e-06, "loss": 0.3012, "step": 12577 }, { "epoch": 0.8695471828551676, "grad_norm": 0.6432282673544676, "learning_rate": 6.285171705593847e-06, "loss": 0.3028, "step": 12578 }, { "epoch": 0.8696163152436917, "grad_norm": 0.8818921645203028, "learning_rate": 6.28463074495919e-06, "loss": 0.3992, "step": 12579 }, { "epoch": 0.8696854476322157, "grad_norm": 0.6381168355775036, "learning_rate": 6.284089768223981e-06, "loss": 0.3442, "step": 12580 }, { "epoch": 0.8697545800207397, "grad_norm": 0.6696674086081812, "learning_rate": 6.283548775394998e-06, "loss": 0.2466, "step": 12581 }, { "epoch": 0.8698237124092637, "grad_norm": 0.6574219502941906, "learning_rate": 6.28300776647902e-06, "loss": 0.329, "step": 12582 }, { "epoch": 0.8698928447977877, "grad_norm": 0.6359661767266652, "learning_rate": 6.282466741482831e-06, "loss": 0.3331, "step": 12583 }, { "epoch": 0.8699619771863117, "grad_norm": 0.6652680592332159, "learning_rate": 6.28192570041321e-06, "loss": 0.309, "step": 12584 }, { "epoch": 0.8700311095748358, "grad_norm": 0.5911007445404182, "learning_rate": 6.281384643276937e-06, "loss": 0.294, "step": 12585 }, { "epoch": 0.8701002419633599, "grad_norm": 0.7159490446257322, "learning_rate": 6.2808435700807954e-06, "loss": 0.3328, "step": 12586 }, { "epoch": 0.8701693743518839, "grad_norm": 0.7210890899776916, "learning_rate": 6.280302480831565e-06, "loss": 0.3794, "step": 12587 }, { "epoch": 0.8702385067404079, "grad_norm": 0.6666554611649471, "learning_rate": 6.2797613755360285e-06, "loss": 0.2595, "step": 12588 }, { "epoch": 0.8703076391289319, "grad_norm": 0.6235534629640448, "learning_rate": 6.279220254200965e-06, "loss": 0.2589, "step": 12589 }, { "epoch": 0.870376771517456, "grad_norm": 0.6971136801855451, "learning_rate": 6.278679116833159e-06, "loss": 0.33, "step": 12590 }, { "epoch": 0.87044590390598, "grad_norm": 0.6806186323724389, "learning_rate": 6.278137963439395e-06, "loss": 0.2217, "step": 12591 }, { "epoch": 0.870515036294504, "grad_norm": 0.5967123681673596, "learning_rate": 6.2775967940264485e-06, "loss": 0.3257, "step": 12592 }, { "epoch": 0.870584168683028, "grad_norm": 0.6268686070049843, "learning_rate": 6.277055608601108e-06, "loss": 0.2878, "step": 12593 }, { "epoch": 0.870653301071552, "grad_norm": 0.7133905872168482, "learning_rate": 6.276514407170156e-06, "loss": 0.405, "step": 12594 }, { "epoch": 0.870722433460076, "grad_norm": 0.5966021313752528, "learning_rate": 6.275973189740372e-06, "loss": 0.2133, "step": 12595 }, { "epoch": 0.8707915658486001, "grad_norm": 0.7917790162612567, "learning_rate": 6.275431956318541e-06, "loss": 0.2802, "step": 12596 }, { "epoch": 0.8708606982371241, "grad_norm": 0.6777489630390671, "learning_rate": 6.274890706911447e-06, "loss": 0.4101, "step": 12597 }, { "epoch": 0.8709298306256481, "grad_norm": 0.7326126755818468, "learning_rate": 6.274349441525874e-06, "loss": 0.3405, "step": 12598 }, { "epoch": 0.8709989630141721, "grad_norm": 0.7260576335193006, "learning_rate": 6.273808160168604e-06, "loss": 0.2409, "step": 12599 }, { "epoch": 0.8710680954026961, "grad_norm": 0.6638093514753971, "learning_rate": 6.273266862846422e-06, "loss": 0.3194, "step": 12600 }, { "epoch": 0.8711372277912202, "grad_norm": 0.8883843020843073, "learning_rate": 6.272725549566112e-06, "loss": 0.3872, "step": 12601 }, { "epoch": 0.8712063601797442, "grad_norm": 0.6045115635529907, "learning_rate": 6.272184220334459e-06, "loss": 0.2855, "step": 12602 }, { "epoch": 0.8712754925682682, "grad_norm": 0.8012841537204111, "learning_rate": 6.271642875158247e-06, "loss": 0.272, "step": 12603 }, { "epoch": 0.8713446249567922, "grad_norm": 0.7447870856906362, "learning_rate": 6.271101514044261e-06, "loss": 0.3516, "step": 12604 }, { "epoch": 0.8714137573453162, "grad_norm": 2.289012743564057, "learning_rate": 6.270560136999286e-06, "loss": 0.3055, "step": 12605 }, { "epoch": 0.8714828897338402, "grad_norm": 1.105509600283833, "learning_rate": 6.270018744030107e-06, "loss": 0.2554, "step": 12606 }, { "epoch": 0.8715520221223644, "grad_norm": 0.6726756541949409, "learning_rate": 6.26947733514351e-06, "loss": 0.3111, "step": 12607 }, { "epoch": 0.8716211545108884, "grad_norm": 0.663182456355063, "learning_rate": 6.268935910346279e-06, "loss": 0.3497, "step": 12608 }, { "epoch": 0.8716902868994124, "grad_norm": 0.6623808253367605, "learning_rate": 6.268394469645203e-06, "loss": 0.2913, "step": 12609 }, { "epoch": 0.8717594192879364, "grad_norm": 0.5833916135197246, "learning_rate": 6.267853013047063e-06, "loss": 0.2781, "step": 12610 }, { "epoch": 0.8718285516764605, "grad_norm": 0.6781688395600068, "learning_rate": 6.26731154055865e-06, "loss": 0.424, "step": 12611 }, { "epoch": 0.8718976840649845, "grad_norm": 0.6500798725339304, "learning_rate": 6.266770052186747e-06, "loss": 0.3762, "step": 12612 }, { "epoch": 0.8719668164535085, "grad_norm": 0.6723513267044534, "learning_rate": 6.266228547938142e-06, "loss": 0.288, "step": 12613 }, { "epoch": 0.8720359488420325, "grad_norm": 0.8208424513680725, "learning_rate": 6.265687027819623e-06, "loss": 0.3243, "step": 12614 }, { "epoch": 0.8721050812305565, "grad_norm": 0.7690630062866598, "learning_rate": 6.265145491837975e-06, "loss": 0.4522, "step": 12615 }, { "epoch": 0.8721742136190805, "grad_norm": 0.654797268996779, "learning_rate": 6.264603939999987e-06, "loss": 0.3338, "step": 12616 }, { "epoch": 0.8722433460076046, "grad_norm": 0.7652379456031817, "learning_rate": 6.2640623723124435e-06, "loss": 0.2457, "step": 12617 }, { "epoch": 0.8723124783961286, "grad_norm": 0.6891614330377848, "learning_rate": 6.263520788782136e-06, "loss": 0.3337, "step": 12618 }, { "epoch": 0.8723816107846526, "grad_norm": 0.7193656926084263, "learning_rate": 6.26297918941585e-06, "loss": 0.4363, "step": 12619 }, { "epoch": 0.8724507431731766, "grad_norm": 0.6590890631662071, "learning_rate": 6.262437574220372e-06, "loss": 0.2993, "step": 12620 }, { "epoch": 0.8725198755617006, "grad_norm": 0.6837110065777442, "learning_rate": 6.261895943202494e-06, "loss": 0.3101, "step": 12621 }, { "epoch": 0.8725890079502246, "grad_norm": 0.694437246922249, "learning_rate": 6.261354296369002e-06, "loss": 0.423, "step": 12622 }, { "epoch": 0.8726581403387487, "grad_norm": 0.782346040282182, "learning_rate": 6.260812633726685e-06, "loss": 0.35, "step": 12623 }, { "epoch": 0.8727272727272727, "grad_norm": 0.6405313890538801, "learning_rate": 6.26027095528233e-06, "loss": 0.3017, "step": 12624 }, { "epoch": 0.8727964051157967, "grad_norm": 0.6604015204212622, "learning_rate": 6.259729261042728e-06, "loss": 0.3043, "step": 12625 }, { "epoch": 0.8728655375043207, "grad_norm": 0.6746331486193331, "learning_rate": 6.259187551014669e-06, "loss": 0.3989, "step": 12626 }, { "epoch": 0.8729346698928449, "grad_norm": 0.6637938911083906, "learning_rate": 6.258645825204942e-06, "loss": 0.2637, "step": 12627 }, { "epoch": 0.8730038022813689, "grad_norm": 0.7046680014396833, "learning_rate": 6.258104083620333e-06, "loss": 0.2976, "step": 12628 }, { "epoch": 0.8730729346698929, "grad_norm": 0.6309491535448601, "learning_rate": 6.257562326267637e-06, "loss": 0.3364, "step": 12629 }, { "epoch": 0.8731420670584169, "grad_norm": 0.6947069653443656, "learning_rate": 6.257020553153641e-06, "loss": 0.4161, "step": 12630 }, { "epoch": 0.8732111994469409, "grad_norm": 1.2215962841928412, "learning_rate": 6.256478764285133e-06, "loss": 0.2984, "step": 12631 }, { "epoch": 0.8732803318354649, "grad_norm": 0.7533831581162269, "learning_rate": 6.255936959668909e-06, "loss": 0.2879, "step": 12632 }, { "epoch": 0.873349464223989, "grad_norm": 0.7595704496002242, "learning_rate": 6.255395139311755e-06, "loss": 0.4271, "step": 12633 }, { "epoch": 0.873418596612513, "grad_norm": 0.603015419360671, "learning_rate": 6.254853303220465e-06, "loss": 0.2563, "step": 12634 }, { "epoch": 0.873487729001037, "grad_norm": 0.6654701935522281, "learning_rate": 6.254311451401826e-06, "loss": 0.3387, "step": 12635 }, { "epoch": 0.873556861389561, "grad_norm": 0.669779745157286, "learning_rate": 6.253769583862632e-06, "loss": 0.3849, "step": 12636 }, { "epoch": 0.873625993778085, "grad_norm": 0.692517266770181, "learning_rate": 6.2532277006096755e-06, "loss": 0.3722, "step": 12637 }, { "epoch": 0.873695126166609, "grad_norm": 0.6267492828140608, "learning_rate": 6.252685801649744e-06, "loss": 0.2501, "step": 12638 }, { "epoch": 0.8737642585551331, "grad_norm": 1.3671155083677293, "learning_rate": 6.252143886989633e-06, "loss": 0.3126, "step": 12639 }, { "epoch": 0.8738333909436571, "grad_norm": 0.6802508932111738, "learning_rate": 6.251601956636132e-06, "loss": 0.4004, "step": 12640 }, { "epoch": 0.8739025233321811, "grad_norm": 0.665184213255445, "learning_rate": 6.251060010596035e-06, "loss": 0.3548, "step": 12641 }, { "epoch": 0.8739716557207051, "grad_norm": 0.7271017164883292, "learning_rate": 6.250518048876132e-06, "loss": 0.304, "step": 12642 }, { "epoch": 0.8740407881092291, "grad_norm": 0.6637621484952118, "learning_rate": 6.249976071483217e-06, "loss": 0.3265, "step": 12643 }, { "epoch": 0.8741099204977532, "grad_norm": 0.6505948033730353, "learning_rate": 6.249434078424084e-06, "loss": 0.384, "step": 12644 }, { "epoch": 0.8741790528862772, "grad_norm": 0.8765173772006759, "learning_rate": 6.248892069705523e-06, "loss": 0.2628, "step": 12645 }, { "epoch": 0.8742481852748012, "grad_norm": 0.6728229700729278, "learning_rate": 6.248350045334331e-06, "loss": 0.2764, "step": 12646 }, { "epoch": 0.8743173176633253, "grad_norm": 0.6967532808763391, "learning_rate": 6.247808005317296e-06, "loss": 0.3628, "step": 12647 }, { "epoch": 0.8743864500518493, "grad_norm": 0.7584726046701814, "learning_rate": 6.247265949661216e-06, "loss": 0.3051, "step": 12648 }, { "epoch": 0.8744555824403734, "grad_norm": 0.6528826571104436, "learning_rate": 6.246723878372884e-06, "loss": 0.252, "step": 12649 }, { "epoch": 0.8745247148288974, "grad_norm": 0.7998547110993142, "learning_rate": 6.246181791459092e-06, "loss": 0.3521, "step": 12650 }, { "epoch": 0.8745938472174214, "grad_norm": 0.7870179226490799, "learning_rate": 6.2456396889266345e-06, "loss": 0.3865, "step": 12651 }, { "epoch": 0.8746629796059454, "grad_norm": 1.002194089658567, "learning_rate": 6.245097570782308e-06, "loss": 0.3119, "step": 12652 }, { "epoch": 0.8747321119944694, "grad_norm": 0.625709878851742, "learning_rate": 6.244555437032904e-06, "loss": 0.3019, "step": 12653 }, { "epoch": 0.8748012443829934, "grad_norm": 0.729798870902409, "learning_rate": 6.24401328768522e-06, "loss": 0.4458, "step": 12654 }, { "epoch": 0.8748703767715175, "grad_norm": 0.7383629750237345, "learning_rate": 6.2434711227460485e-06, "loss": 0.3312, "step": 12655 }, { "epoch": 0.8749395091600415, "grad_norm": 0.6155006460142453, "learning_rate": 6.242928942222185e-06, "loss": 0.2772, "step": 12656 }, { "epoch": 0.8750086415485655, "grad_norm": 0.7060304197530333, "learning_rate": 6.242386746120427e-06, "loss": 0.3368, "step": 12657 }, { "epoch": 0.8750777739370895, "grad_norm": 0.6402835496994311, "learning_rate": 6.241844534447567e-06, "loss": 0.3997, "step": 12658 }, { "epoch": 0.8751469063256135, "grad_norm": 0.6541301636843948, "learning_rate": 6.241302307210403e-06, "loss": 0.3088, "step": 12659 }, { "epoch": 0.8752160387141376, "grad_norm": 0.7101375066580299, "learning_rate": 6.2407600644157284e-06, "loss": 0.2469, "step": 12660 }, { "epoch": 0.8752851711026616, "grad_norm": 0.8091974205831539, "learning_rate": 6.240217806070341e-06, "loss": 0.4058, "step": 12661 }, { "epoch": 0.8753543034911856, "grad_norm": 0.639411225466008, "learning_rate": 6.2396755321810375e-06, "loss": 0.3466, "step": 12662 }, { "epoch": 0.8754234358797096, "grad_norm": 0.7379783099958896, "learning_rate": 6.2391332427546125e-06, "loss": 0.3172, "step": 12663 }, { "epoch": 0.8754925682682336, "grad_norm": 0.7890138575194185, "learning_rate": 6.238590937797864e-06, "loss": 0.3318, "step": 12664 }, { "epoch": 0.8755617006567576, "grad_norm": 0.8310792140752084, "learning_rate": 6.238048617317589e-06, "loss": 0.4398, "step": 12665 }, { "epoch": 0.8756308330452817, "grad_norm": 0.6671008577915446, "learning_rate": 6.237506281320584e-06, "loss": 0.318, "step": 12666 }, { "epoch": 0.8756999654338057, "grad_norm": 0.622913176487906, "learning_rate": 6.236963929813645e-06, "loss": 0.2544, "step": 12667 }, { "epoch": 0.8757690978223298, "grad_norm": 0.759059460816099, "learning_rate": 6.236421562803572e-06, "loss": 0.3774, "step": 12668 }, { "epoch": 0.8758382302108538, "grad_norm": 2.2489440306192137, "learning_rate": 6.2358791802971605e-06, "loss": 0.4333, "step": 12669 }, { "epoch": 0.8759073625993778, "grad_norm": 0.6433173833570313, "learning_rate": 6.235336782301209e-06, "loss": 0.3009, "step": 12670 }, { "epoch": 0.8759764949879019, "grad_norm": 0.6926753953122954, "learning_rate": 6.2347943688225155e-06, "loss": 0.2777, "step": 12671 }, { "epoch": 0.8760456273764259, "grad_norm": 0.923918767878148, "learning_rate": 6.2342519398678784e-06, "loss": 0.359, "step": 12672 }, { "epoch": 0.8761147597649499, "grad_norm": 0.7140586276225058, "learning_rate": 6.2337094954440955e-06, "loss": 0.3279, "step": 12673 }, { "epoch": 0.8761838921534739, "grad_norm": 0.5955418077523685, "learning_rate": 6.233167035557967e-06, "loss": 0.2533, "step": 12674 }, { "epoch": 0.8762530245419979, "grad_norm": 0.6423634578820023, "learning_rate": 6.2326245602162894e-06, "loss": 0.3227, "step": 12675 }, { "epoch": 0.876322156930522, "grad_norm": 0.6335491453347236, "learning_rate": 6.232082069425862e-06, "loss": 0.3514, "step": 12676 }, { "epoch": 0.876391289319046, "grad_norm": 0.7375209040803823, "learning_rate": 6.231539563193485e-06, "loss": 0.2734, "step": 12677 }, { "epoch": 0.87646042170757, "grad_norm": 1.4665324661787738, "learning_rate": 6.230997041525958e-06, "loss": 0.2807, "step": 12678 }, { "epoch": 0.876529554096094, "grad_norm": 0.6815375610766823, "learning_rate": 6.230454504430079e-06, "loss": 0.3263, "step": 12679 }, { "epoch": 0.876598686484618, "grad_norm": 0.7760214821547305, "learning_rate": 6.229911951912649e-06, "loss": 0.2924, "step": 12680 }, { "epoch": 0.876667818873142, "grad_norm": 0.9086143901796551, "learning_rate": 6.229369383980468e-06, "loss": 0.2361, "step": 12681 }, { "epoch": 0.8767369512616661, "grad_norm": 0.7105342003121622, "learning_rate": 6.228826800640335e-06, "loss": 0.3048, "step": 12682 }, { "epoch": 0.8768060836501901, "grad_norm": 0.7537517781814237, "learning_rate": 6.228284201899052e-06, "loss": 0.3636, "step": 12683 }, { "epoch": 0.8768752160387141, "grad_norm": 0.6288445649667954, "learning_rate": 6.227741587763416e-06, "loss": 0.3164, "step": 12684 }, { "epoch": 0.8769443484272381, "grad_norm": 0.6536265062673762, "learning_rate": 6.227198958240232e-06, "loss": 0.2667, "step": 12685 }, { "epoch": 0.8770134808157621, "grad_norm": 0.665798133366459, "learning_rate": 6.2266563133363e-06, "loss": 0.353, "step": 12686 }, { "epoch": 0.8770826132042862, "grad_norm": 1.3822352176856052, "learning_rate": 6.22611365305842e-06, "loss": 0.379, "step": 12687 }, { "epoch": 0.8771517455928103, "grad_norm": 0.6790983338590182, "learning_rate": 6.225570977413392e-06, "loss": 0.2914, "step": 12688 }, { "epoch": 0.8772208779813343, "grad_norm": 0.6851326136687549, "learning_rate": 6.22502828640802e-06, "loss": 0.3432, "step": 12689 }, { "epoch": 0.8772900103698583, "grad_norm": 0.7590563364047405, "learning_rate": 6.224485580049104e-06, "loss": 0.3724, "step": 12690 }, { "epoch": 0.8773591427583823, "grad_norm": 2.028208735706837, "learning_rate": 6.223942858343446e-06, "loss": 0.2681, "step": 12691 }, { "epoch": 0.8774282751469064, "grad_norm": 0.6473052756827301, "learning_rate": 6.22340012129785e-06, "loss": 0.239, "step": 12692 }, { "epoch": 0.8774974075354304, "grad_norm": 0.6554934812966866, "learning_rate": 6.222857368919116e-06, "loss": 0.3121, "step": 12693 }, { "epoch": 0.8775665399239544, "grad_norm": 0.7123027968139278, "learning_rate": 6.222314601214048e-06, "loss": 0.3923, "step": 12694 }, { "epoch": 0.8776356723124784, "grad_norm": 1.5200029954531762, "learning_rate": 6.221771818189446e-06, "loss": 0.3233, "step": 12695 }, { "epoch": 0.8777048047010024, "grad_norm": 0.737605484890375, "learning_rate": 6.221229019852115e-06, "loss": 0.3015, "step": 12696 }, { "epoch": 0.8777739370895264, "grad_norm": 0.6934406055663538, "learning_rate": 6.220686206208859e-06, "loss": 0.4569, "step": 12697 }, { "epoch": 0.8778430694780505, "grad_norm": 0.7064185849358267, "learning_rate": 6.220143377266479e-06, "loss": 0.3779, "step": 12698 }, { "epoch": 0.8779122018665745, "grad_norm": 0.7450291925451727, "learning_rate": 6.219600533031778e-06, "loss": 0.2373, "step": 12699 }, { "epoch": 0.8779813342550985, "grad_norm": 0.8010801233638788, "learning_rate": 6.2190576735115626e-06, "loss": 0.3474, "step": 12700 }, { "epoch": 0.8780504666436225, "grad_norm": 0.7515196419044522, "learning_rate": 6.218514798712634e-06, "loss": 0.3873, "step": 12701 }, { "epoch": 0.8781195990321465, "grad_norm": 0.6818825292060731, "learning_rate": 6.217971908641797e-06, "loss": 0.2581, "step": 12702 }, { "epoch": 0.8781887314206706, "grad_norm": 0.644601910748762, "learning_rate": 6.217429003305856e-06, "loss": 0.2932, "step": 12703 }, { "epoch": 0.8782578638091946, "grad_norm": 0.7828938518213182, "learning_rate": 6.216886082711615e-06, "loss": 0.3472, "step": 12704 }, { "epoch": 0.8783269961977186, "grad_norm": 0.6857158433336591, "learning_rate": 6.216343146865878e-06, "loss": 0.3575, "step": 12705 }, { "epoch": 0.8783961285862426, "grad_norm": 0.6701558856927239, "learning_rate": 6.21580019577545e-06, "loss": 0.2597, "step": 12706 }, { "epoch": 0.8784652609747666, "grad_norm": 0.6360610915539061, "learning_rate": 6.215257229447137e-06, "loss": 0.2842, "step": 12707 }, { "epoch": 0.8785343933632908, "grad_norm": 0.6631576894489709, "learning_rate": 6.214714247887744e-06, "loss": 0.4129, "step": 12708 }, { "epoch": 0.8786035257518148, "grad_norm": 0.6780593951508671, "learning_rate": 6.214171251104074e-06, "loss": 0.2829, "step": 12709 }, { "epoch": 0.8786726581403388, "grad_norm": 0.6657912763919392, "learning_rate": 6.213628239102935e-06, "loss": 0.2888, "step": 12710 }, { "epoch": 0.8787417905288628, "grad_norm": 0.6093902299995134, "learning_rate": 6.2130852118911325e-06, "loss": 0.3306, "step": 12711 }, { "epoch": 0.8788109229173868, "grad_norm": 0.7490762388602318, "learning_rate": 6.212542169475472e-06, "loss": 0.4139, "step": 12712 }, { "epoch": 0.8788800553059108, "grad_norm": 0.694562560274104, "learning_rate": 6.211999111862756e-06, "loss": 0.3175, "step": 12713 }, { "epoch": 0.8789491876944349, "grad_norm": 0.6463785834093664, "learning_rate": 6.211456039059798e-06, "loss": 0.2662, "step": 12714 }, { "epoch": 0.8790183200829589, "grad_norm": 0.6263086630402229, "learning_rate": 6.210912951073399e-06, "loss": 0.3535, "step": 12715 }, { "epoch": 0.8790874524714829, "grad_norm": 1.0286955689720376, "learning_rate": 6.210369847910367e-06, "loss": 0.2817, "step": 12716 }, { "epoch": 0.8791565848600069, "grad_norm": 0.6557014876076837, "learning_rate": 6.209826729577508e-06, "loss": 0.3198, "step": 12717 }, { "epoch": 0.8792257172485309, "grad_norm": 0.6796928049630989, "learning_rate": 6.209283596081632e-06, "loss": 0.3392, "step": 12718 }, { "epoch": 0.879294849637055, "grad_norm": 0.6779031021393679, "learning_rate": 6.208740447429543e-06, "loss": 0.4039, "step": 12719 }, { "epoch": 0.879363982025579, "grad_norm": 0.6059086128836192, "learning_rate": 6.208197283628049e-06, "loss": 0.2895, "step": 12720 }, { "epoch": 0.879433114414103, "grad_norm": 1.2971389099127155, "learning_rate": 6.20765410468396e-06, "loss": 0.2638, "step": 12721 }, { "epoch": 0.879502246802627, "grad_norm": 0.6610904537183392, "learning_rate": 6.20711091060408e-06, "loss": 0.3714, "step": 12722 }, { "epoch": 0.879571379191151, "grad_norm": 0.7032374353995946, "learning_rate": 6.206567701395221e-06, "loss": 0.3714, "step": 12723 }, { "epoch": 0.879640511579675, "grad_norm": 2.5382853854676437, "learning_rate": 6.2060244770641874e-06, "loss": 0.2488, "step": 12724 }, { "epoch": 0.8797096439681991, "grad_norm": 0.625660077546477, "learning_rate": 6.20548123761779e-06, "loss": 0.32, "step": 12725 }, { "epoch": 0.8797787763567231, "grad_norm": 0.6358438078073378, "learning_rate": 6.204937983062836e-06, "loss": 0.3761, "step": 12726 }, { "epoch": 0.8798479087452471, "grad_norm": 0.7063732523734719, "learning_rate": 6.2043947134061345e-06, "loss": 0.3438, "step": 12727 }, { "epoch": 0.8799170411337712, "grad_norm": 0.8140190112002206, "learning_rate": 6.203851428654496e-06, "loss": 0.2546, "step": 12728 }, { "epoch": 0.8799861735222952, "grad_norm": 0.6833882262703842, "learning_rate": 6.2033081288147265e-06, "loss": 0.3726, "step": 12729 }, { "epoch": 0.8800553059108193, "grad_norm": 0.6921291043315643, "learning_rate": 6.2027648138936395e-06, "loss": 0.307, "step": 12730 }, { "epoch": 0.8801244382993433, "grad_norm": 0.7183035280302354, "learning_rate": 6.2022214838980385e-06, "loss": 0.2849, "step": 12731 }, { "epoch": 0.8801935706878673, "grad_norm": 0.7461133138265594, "learning_rate": 6.2016781388347395e-06, "loss": 0.371, "step": 12732 }, { "epoch": 0.8802627030763913, "grad_norm": 0.7367506156984509, "learning_rate": 6.201134778710548e-06, "loss": 0.3541, "step": 12733 }, { "epoch": 0.8803318354649153, "grad_norm": 0.6676701537107924, "learning_rate": 6.200591403532275e-06, "loss": 0.3493, "step": 12734 }, { "epoch": 0.8804009678534394, "grad_norm": 0.6617509917014087, "learning_rate": 6.200048013306733e-06, "loss": 0.3359, "step": 12735 }, { "epoch": 0.8804701002419634, "grad_norm": 0.6487273168049286, "learning_rate": 6.199504608040731e-06, "loss": 0.3611, "step": 12736 }, { "epoch": 0.8805392326304874, "grad_norm": 0.689305415584373, "learning_rate": 6.198961187741079e-06, "loss": 0.3476, "step": 12737 }, { "epoch": 0.8806083650190114, "grad_norm": 0.7156753040115983, "learning_rate": 6.198417752414586e-06, "loss": 0.339, "step": 12738 }, { "epoch": 0.8806774974075354, "grad_norm": 0.6769676682741879, "learning_rate": 6.197874302068069e-06, "loss": 0.3527, "step": 12739 }, { "epoch": 0.8807466297960594, "grad_norm": 0.7342574835502056, "learning_rate": 6.1973308367083326e-06, "loss": 0.4222, "step": 12740 }, { "epoch": 0.8808157621845835, "grad_norm": 0.7743133958237282, "learning_rate": 6.196787356342191e-06, "loss": 0.3116, "step": 12741 }, { "epoch": 0.8808848945731075, "grad_norm": 0.7357646992262381, "learning_rate": 6.1962438609764565e-06, "loss": 0.3004, "step": 12742 }, { "epoch": 0.8809540269616315, "grad_norm": 0.7262110169010488, "learning_rate": 6.195700350617939e-06, "loss": 0.3271, "step": 12743 }, { "epoch": 0.8810231593501555, "grad_norm": 0.6944220157971412, "learning_rate": 6.195156825273452e-06, "loss": 0.3737, "step": 12744 }, { "epoch": 0.8810922917386795, "grad_norm": 0.6297216910181691, "learning_rate": 6.194613284949806e-06, "loss": 0.2793, "step": 12745 }, { "epoch": 0.8811614241272036, "grad_norm": 1.3966237972291635, "learning_rate": 6.194069729653817e-06, "loss": 0.2605, "step": 12746 }, { "epoch": 0.8812305565157276, "grad_norm": 0.6910945490163481, "learning_rate": 6.193526159392294e-06, "loss": 0.3812, "step": 12747 }, { "epoch": 0.8812996889042516, "grad_norm": 0.676655852261262, "learning_rate": 6.192982574172048e-06, "loss": 0.3375, "step": 12748 }, { "epoch": 0.8813688212927757, "grad_norm": 0.6974336063607265, "learning_rate": 6.192438973999897e-06, "loss": 0.3128, "step": 12749 }, { "epoch": 0.8814379536812997, "grad_norm": 0.6565617774802807, "learning_rate": 6.19189535888265e-06, "loss": 0.3459, "step": 12750 }, { "epoch": 0.8815070860698238, "grad_norm": 0.6427879532478615, "learning_rate": 6.191351728827123e-06, "loss": 0.3406, "step": 12751 }, { "epoch": 0.8815762184583478, "grad_norm": 0.7160764169200851, "learning_rate": 6.190808083840126e-06, "loss": 0.3133, "step": 12752 }, { "epoch": 0.8816453508468718, "grad_norm": 0.7098952735230292, "learning_rate": 6.190264423928477e-06, "loss": 0.2737, "step": 12753 }, { "epoch": 0.8817144832353958, "grad_norm": 0.6496565499115015, "learning_rate": 6.189720749098988e-06, "loss": 0.371, "step": 12754 }, { "epoch": 0.8817836156239198, "grad_norm": 0.6106691926054685, "learning_rate": 6.18917705935847e-06, "loss": 0.3177, "step": 12755 }, { "epoch": 0.8818527480124438, "grad_norm": 0.6552538777284825, "learning_rate": 6.188633354713741e-06, "loss": 0.2539, "step": 12756 }, { "epoch": 0.8819218804009679, "grad_norm": 0.6792438853242331, "learning_rate": 6.188089635171615e-06, "loss": 0.3261, "step": 12757 }, { "epoch": 0.8819910127894919, "grad_norm": 0.6530752015834449, "learning_rate": 6.187545900738905e-06, "loss": 0.4058, "step": 12758 }, { "epoch": 0.8820601451780159, "grad_norm": 0.7595740147211262, "learning_rate": 6.187002151422426e-06, "loss": 0.3182, "step": 12759 }, { "epoch": 0.8821292775665399, "grad_norm": 0.7993665158352833, "learning_rate": 6.186458387228993e-06, "loss": 0.3567, "step": 12760 }, { "epoch": 0.8821984099550639, "grad_norm": 0.7263316015823685, "learning_rate": 6.185914608165422e-06, "loss": 0.3227, "step": 12761 }, { "epoch": 0.882267542343588, "grad_norm": 0.6876723039230394, "learning_rate": 6.185370814238527e-06, "loss": 0.3599, "step": 12762 }, { "epoch": 0.882336674732112, "grad_norm": 0.6774846914787733, "learning_rate": 6.184827005455125e-06, "loss": 0.2611, "step": 12763 }, { "epoch": 0.882405807120636, "grad_norm": 0.7297962128436976, "learning_rate": 6.184283181822032e-06, "loss": 0.288, "step": 12764 }, { "epoch": 0.88247493950916, "grad_norm": 0.856495959439123, "learning_rate": 6.183739343346063e-06, "loss": 0.4115, "step": 12765 }, { "epoch": 0.882544071897684, "grad_norm": 0.6762937397786771, "learning_rate": 6.183195490034031e-06, "loss": 0.2871, "step": 12766 }, { "epoch": 0.882613204286208, "grad_norm": 0.7426009307209325, "learning_rate": 6.1826516218927565e-06, "loss": 0.2999, "step": 12767 }, { "epoch": 0.8826823366747321, "grad_norm": 0.6888147528856671, "learning_rate": 6.182107738929055e-06, "loss": 0.3306, "step": 12768 }, { "epoch": 0.8827514690632562, "grad_norm": 0.6758914258387219, "learning_rate": 6.1815638411497426e-06, "loss": 0.3916, "step": 12769 }, { "epoch": 0.8828206014517802, "grad_norm": 0.7565323165223327, "learning_rate": 6.181019928561635e-06, "loss": 0.2472, "step": 12770 }, { "epoch": 0.8828897338403042, "grad_norm": 0.6278546772258153, "learning_rate": 6.1804760011715505e-06, "loss": 0.2821, "step": 12771 }, { "epoch": 0.8829588662288282, "grad_norm": 0.7675702172593153, "learning_rate": 6.179932058986306e-06, "loss": 0.4218, "step": 12772 }, { "epoch": 0.8830279986173523, "grad_norm": 0.7959069589946443, "learning_rate": 6.179388102012719e-06, "loss": 0.3485, "step": 12773 }, { "epoch": 0.8830971310058763, "grad_norm": 0.6797929163501342, "learning_rate": 6.178844130257606e-06, "loss": 0.3618, "step": 12774 }, { "epoch": 0.8831662633944003, "grad_norm": 0.7049638769588178, "learning_rate": 6.178300143727788e-06, "loss": 0.3313, "step": 12775 }, { "epoch": 0.8832353957829243, "grad_norm": 0.6762593967041103, "learning_rate": 6.177756142430079e-06, "loss": 0.3929, "step": 12776 }, { "epoch": 0.8833045281714483, "grad_norm": 0.6872691643893034, "learning_rate": 6.177212126371297e-06, "loss": 0.2893, "step": 12777 }, { "epoch": 0.8833736605599724, "grad_norm": 0.6586758949938514, "learning_rate": 6.176668095558264e-06, "loss": 0.3105, "step": 12778 }, { "epoch": 0.8834427929484964, "grad_norm": 0.6791405810756855, "learning_rate": 6.176124049997796e-06, "loss": 0.375, "step": 12779 }, { "epoch": 0.8835119253370204, "grad_norm": 0.6523629282201213, "learning_rate": 6.175579989696711e-06, "loss": 0.3251, "step": 12780 }, { "epoch": 0.8835810577255444, "grad_norm": 0.5920687625136211, "learning_rate": 6.175035914661829e-06, "loss": 0.2245, "step": 12781 }, { "epoch": 0.8836501901140684, "grad_norm": 0.6633865547738187, "learning_rate": 6.17449182489997e-06, "loss": 0.3017, "step": 12782 }, { "epoch": 0.8837193225025924, "grad_norm": 0.7201249464673581, "learning_rate": 6.173947720417951e-06, "loss": 0.3911, "step": 12783 }, { "epoch": 0.8837884548911165, "grad_norm": 1.3566026741367212, "learning_rate": 6.173403601222592e-06, "loss": 0.3136, "step": 12784 }, { "epoch": 0.8838575872796405, "grad_norm": 0.7918879019029952, "learning_rate": 6.172859467320713e-06, "loss": 0.2521, "step": 12785 }, { "epoch": 0.8839267196681645, "grad_norm": 0.6837641717157539, "learning_rate": 6.172315318719134e-06, "loss": 0.3545, "step": 12786 }, { "epoch": 0.8839958520566885, "grad_norm": 0.7149526542183854, "learning_rate": 6.171771155424675e-06, "loss": 0.4213, "step": 12787 }, { "epoch": 0.8840649844452125, "grad_norm": 0.6910879052369644, "learning_rate": 6.171226977444154e-06, "loss": 0.3132, "step": 12788 }, { "epoch": 0.8841341168337367, "grad_norm": 0.9243696598035762, "learning_rate": 6.170682784784394e-06, "loss": 0.2695, "step": 12789 }, { "epoch": 0.8842032492222607, "grad_norm": 0.7355903706377376, "learning_rate": 6.170138577452215e-06, "loss": 0.3896, "step": 12790 }, { "epoch": 0.8842723816107847, "grad_norm": 0.6308566913728563, "learning_rate": 6.169594355454436e-06, "loss": 0.3223, "step": 12791 }, { "epoch": 0.8843415139993087, "grad_norm": 1.5133559993712355, "learning_rate": 6.16905011879788e-06, "loss": 0.2342, "step": 12792 }, { "epoch": 0.8844106463878327, "grad_norm": 0.6865337693459194, "learning_rate": 6.168505867489367e-06, "loss": 0.3384, "step": 12793 }, { "epoch": 0.8844797787763568, "grad_norm": 0.6451204653128939, "learning_rate": 6.167961601535718e-06, "loss": 0.3847, "step": 12794 }, { "epoch": 0.8845489111648808, "grad_norm": 0.6535911749339124, "learning_rate": 6.167417320943754e-06, "loss": 0.3018, "step": 12795 }, { "epoch": 0.8846180435534048, "grad_norm": 0.6896834303246, "learning_rate": 6.166873025720298e-06, "loss": 0.3304, "step": 12796 }, { "epoch": 0.8846871759419288, "grad_norm": 0.6512929182522614, "learning_rate": 6.166328715872171e-06, "loss": 0.3821, "step": 12797 }, { "epoch": 0.8847563083304528, "grad_norm": 0.6869931589338878, "learning_rate": 6.165784391406194e-06, "loss": 0.3256, "step": 12798 }, { "epoch": 0.8848254407189768, "grad_norm": 0.726337391678403, "learning_rate": 6.165240052329191e-06, "loss": 0.2756, "step": 12799 }, { "epoch": 0.8848945731075009, "grad_norm": 0.6759929928551415, "learning_rate": 6.1646956986479835e-06, "loss": 0.2819, "step": 12800 }, { "epoch": 0.8849637054960249, "grad_norm": 0.7484324799969326, "learning_rate": 6.164151330369394e-06, "loss": 0.4161, "step": 12801 }, { "epoch": 0.8850328378845489, "grad_norm": 0.6801623480454851, "learning_rate": 6.163606947500245e-06, "loss": 0.3923, "step": 12802 }, { "epoch": 0.8851019702730729, "grad_norm": 0.6989392269158564, "learning_rate": 6.163062550047359e-06, "loss": 0.2367, "step": 12803 }, { "epoch": 0.8851711026615969, "grad_norm": 0.6342464188941395, "learning_rate": 6.162518138017559e-06, "loss": 0.3442, "step": 12804 }, { "epoch": 0.885240235050121, "grad_norm": 1.070442885506593, "learning_rate": 6.161973711417669e-06, "loss": 0.3298, "step": 12805 }, { "epoch": 0.885309367438645, "grad_norm": 0.666012023501174, "learning_rate": 6.1614292702545144e-06, "loss": 0.3372, "step": 12806 }, { "epoch": 0.885378499827169, "grad_norm": 0.7131930425965017, "learning_rate": 6.160884814534916e-06, "loss": 0.4524, "step": 12807 }, { "epoch": 0.885447632215693, "grad_norm": 0.6708425144981677, "learning_rate": 6.160340344265695e-06, "loss": 0.3563, "step": 12808 }, { "epoch": 0.885516764604217, "grad_norm": 0.7054330216472793, "learning_rate": 6.159795859453681e-06, "loss": 0.3613, "step": 12809 }, { "epoch": 0.8855858969927412, "grad_norm": 0.5993277293005727, "learning_rate": 6.1592513601056955e-06, "loss": 0.3031, "step": 12810 }, { "epoch": 0.8856550293812652, "grad_norm": 0.6844308990315134, "learning_rate": 6.158706846228563e-06, "loss": 0.3873, "step": 12811 }, { "epoch": 0.8857241617697892, "grad_norm": 0.671277970323566, "learning_rate": 6.1581623178291065e-06, "loss": 0.3811, "step": 12812 }, { "epoch": 0.8857932941583132, "grad_norm": 0.6943006755464762, "learning_rate": 6.157617774914153e-06, "loss": 0.3043, "step": 12813 }, { "epoch": 0.8858624265468372, "grad_norm": 0.685602229299518, "learning_rate": 6.157073217490528e-06, "loss": 0.2645, "step": 12814 }, { "epoch": 0.8859315589353612, "grad_norm": 0.9482344694564739, "learning_rate": 6.156528645565053e-06, "loss": 0.4, "step": 12815 }, { "epoch": 0.8860006913238853, "grad_norm": 0.6702168105441352, "learning_rate": 6.155984059144556e-06, "loss": 0.336, "step": 12816 }, { "epoch": 0.8860698237124093, "grad_norm": 0.6892851214847807, "learning_rate": 6.155439458235862e-06, "loss": 0.3119, "step": 12817 }, { "epoch": 0.8861389561009333, "grad_norm": 0.7411753194558386, "learning_rate": 6.154894842845798e-06, "loss": 0.2957, "step": 12818 }, { "epoch": 0.8862080884894573, "grad_norm": 0.6819025381205328, "learning_rate": 6.154350212981185e-06, "loss": 0.375, "step": 12819 }, { "epoch": 0.8862772208779813, "grad_norm": 0.6224003774012744, "learning_rate": 6.153805568648854e-06, "loss": 0.2503, "step": 12820 }, { "epoch": 0.8863463532665053, "grad_norm": 0.6454434568852324, "learning_rate": 6.15326090985563e-06, "loss": 0.2911, "step": 12821 }, { "epoch": 0.8864154856550294, "grad_norm": 0.7062645945886646, "learning_rate": 6.152716236608337e-06, "loss": 0.4261, "step": 12822 }, { "epoch": 0.8864846180435534, "grad_norm": 0.7663375967247312, "learning_rate": 6.152171548913803e-06, "loss": 0.34, "step": 12823 }, { "epoch": 0.8865537504320774, "grad_norm": 0.6548413500542165, "learning_rate": 6.151626846778855e-06, "loss": 0.2974, "step": 12824 }, { "epoch": 0.8866228828206014, "grad_norm": 0.6703245441639158, "learning_rate": 6.1510821302103204e-06, "loss": 0.288, "step": 12825 }, { "epoch": 0.8866920152091254, "grad_norm": 0.6795310043206758, "learning_rate": 6.150537399215024e-06, "loss": 0.3735, "step": 12826 }, { "epoch": 0.8867611475976495, "grad_norm": 0.6734526676532794, "learning_rate": 6.149992653799796e-06, "loss": 0.2757, "step": 12827 }, { "epoch": 0.8868302799861735, "grad_norm": 1.08637754970536, "learning_rate": 6.149447893971463e-06, "loss": 0.2817, "step": 12828 }, { "epoch": 0.8868994123746975, "grad_norm": 0.7397554396367464, "learning_rate": 6.148903119736851e-06, "loss": 0.4001, "step": 12829 }, { "epoch": 0.8869685447632216, "grad_norm": 0.7269953397235923, "learning_rate": 6.148358331102788e-06, "loss": 0.3249, "step": 12830 }, { "epoch": 0.8870376771517456, "grad_norm": 0.6826009746457021, "learning_rate": 6.147813528076103e-06, "loss": 0.2656, "step": 12831 }, { "epoch": 0.8871068095402697, "grad_norm": 0.8671898933213368, "learning_rate": 6.147268710663624e-06, "loss": 0.3427, "step": 12832 }, { "epoch": 0.8871759419287937, "grad_norm": 0.7746569923834796, "learning_rate": 6.146723878872178e-06, "loss": 0.4237, "step": 12833 }, { "epoch": 0.8872450743173177, "grad_norm": 0.6346301290957082, "learning_rate": 6.146179032708597e-06, "loss": 0.3095, "step": 12834 }, { "epoch": 0.8873142067058417, "grad_norm": 0.6585023139313716, "learning_rate": 6.1456341721797065e-06, "loss": 0.3365, "step": 12835 }, { "epoch": 0.8873833390943657, "grad_norm": 0.6624181782938305, "learning_rate": 6.145089297292336e-06, "loss": 0.3487, "step": 12836 }, { "epoch": 0.8874524714828897, "grad_norm": 0.8439793512979856, "learning_rate": 6.1445444080533135e-06, "loss": 0.4361, "step": 12837 }, { "epoch": 0.8875216038714138, "grad_norm": 0.7542981754298738, "learning_rate": 6.14399950446947e-06, "loss": 0.3271, "step": 12838 }, { "epoch": 0.8875907362599378, "grad_norm": 0.7728997482206663, "learning_rate": 6.143454586547635e-06, "loss": 0.3277, "step": 12839 }, { "epoch": 0.8876598686484618, "grad_norm": 0.6744643365097452, "learning_rate": 6.142909654294638e-06, "loss": 0.3527, "step": 12840 }, { "epoch": 0.8877290010369858, "grad_norm": 0.719839031508409, "learning_rate": 6.1423647077173054e-06, "loss": 0.3111, "step": 12841 }, { "epoch": 0.8877981334255098, "grad_norm": 0.6278964551893588, "learning_rate": 6.141819746822472e-06, "loss": 0.2641, "step": 12842 }, { "epoch": 0.8878672658140339, "grad_norm": 0.6578896533760903, "learning_rate": 6.141274771616965e-06, "loss": 0.3221, "step": 12843 }, { "epoch": 0.8879363982025579, "grad_norm": 0.6650019292616718, "learning_rate": 6.140729782107615e-06, "loss": 0.3601, "step": 12844 }, { "epoch": 0.8880055305910819, "grad_norm": 0.6381391925893348, "learning_rate": 6.140184778301254e-06, "loss": 0.3008, "step": 12845 }, { "epoch": 0.8880746629796059, "grad_norm": 0.7748858902835528, "learning_rate": 6.139639760204711e-06, "loss": 0.3162, "step": 12846 }, { "epoch": 0.8881437953681299, "grad_norm": 0.6901951885321254, "learning_rate": 6.139094727824817e-06, "loss": 0.413, "step": 12847 }, { "epoch": 0.888212927756654, "grad_norm": 0.6791256022425346, "learning_rate": 6.138549681168403e-06, "loss": 0.3621, "step": 12848 }, { "epoch": 0.888282060145178, "grad_norm": 0.6710183064563887, "learning_rate": 6.138004620242302e-06, "loss": 0.2514, "step": 12849 }, { "epoch": 0.8883511925337021, "grad_norm": 0.7202182366658139, "learning_rate": 6.137459545053344e-06, "loss": 0.2721, "step": 12850 }, { "epoch": 0.8884203249222261, "grad_norm": 0.6151276035112809, "learning_rate": 6.136914455608358e-06, "loss": 0.3978, "step": 12851 }, { "epoch": 0.8884894573107501, "grad_norm": 0.5932456918851579, "learning_rate": 6.1363693519141795e-06, "loss": 0.2707, "step": 12852 }, { "epoch": 0.8885585896992741, "grad_norm": 0.6283864114960354, "learning_rate": 6.135824233977641e-06, "loss": 0.3102, "step": 12853 }, { "epoch": 0.8886277220877982, "grad_norm": 0.744991865755581, "learning_rate": 6.135279101805571e-06, "loss": 0.3134, "step": 12854 }, { "epoch": 0.8886968544763222, "grad_norm": 0.7105632096937765, "learning_rate": 6.134733955404803e-06, "loss": 0.4388, "step": 12855 }, { "epoch": 0.8887659868648462, "grad_norm": 0.6650774284037829, "learning_rate": 6.134188794782171e-06, "loss": 0.3039, "step": 12856 }, { "epoch": 0.8888351192533702, "grad_norm": 0.6629066646624753, "learning_rate": 6.133643619944507e-06, "loss": 0.2642, "step": 12857 }, { "epoch": 0.8889042516418942, "grad_norm": 0.6795750024375562, "learning_rate": 6.133098430898643e-06, "loss": 0.4119, "step": 12858 }, { "epoch": 0.8889733840304183, "grad_norm": 0.6320213745041533, "learning_rate": 6.13255322765141e-06, "loss": 0.2912, "step": 12859 }, { "epoch": 0.8890425164189423, "grad_norm": 0.8247378294422854, "learning_rate": 6.132008010209646e-06, "loss": 0.2484, "step": 12860 }, { "epoch": 0.8891116488074663, "grad_norm": 0.7441304157203026, "learning_rate": 6.13146277858018e-06, "loss": 0.3531, "step": 12861 }, { "epoch": 0.8891807811959903, "grad_norm": 0.7874457077321694, "learning_rate": 6.130917532769848e-06, "loss": 0.3208, "step": 12862 }, { "epoch": 0.8892499135845143, "grad_norm": 0.6856089270566464, "learning_rate": 6.130372272785483e-06, "loss": 0.2727, "step": 12863 }, { "epoch": 0.8893190459730383, "grad_norm": 1.5423500480749748, "learning_rate": 6.1298269986339195e-06, "loss": 0.319, "step": 12864 }, { "epoch": 0.8893881783615624, "grad_norm": 0.7033521668311951, "learning_rate": 6.12928171032199e-06, "loss": 0.4443, "step": 12865 }, { "epoch": 0.8894573107500864, "grad_norm": 0.6723011865430508, "learning_rate": 6.128736407856531e-06, "loss": 0.295, "step": 12866 }, { "epoch": 0.8895264431386104, "grad_norm": 1.0315578692946075, "learning_rate": 6.128191091244373e-06, "loss": 0.2494, "step": 12867 }, { "epoch": 0.8895955755271344, "grad_norm": 0.760792515530094, "learning_rate": 6.127645760492354e-06, "loss": 0.344, "step": 12868 }, { "epoch": 0.8896647079156584, "grad_norm": 0.7087832647874637, "learning_rate": 6.127100415607308e-06, "loss": 0.338, "step": 12869 }, { "epoch": 0.8897338403041825, "grad_norm": 0.6554508351631092, "learning_rate": 6.126555056596069e-06, "loss": 0.2864, "step": 12870 }, { "epoch": 0.8898029726927066, "grad_norm": 0.6735270255477928, "learning_rate": 6.126009683465475e-06, "loss": 0.2838, "step": 12871 }, { "epoch": 0.8898721050812306, "grad_norm": 0.6687973001918757, "learning_rate": 6.125464296222357e-06, "loss": 0.3556, "step": 12872 }, { "epoch": 0.8899412374697546, "grad_norm": 0.7989017721573156, "learning_rate": 6.124918894873554e-06, "loss": 0.3303, "step": 12873 }, { "epoch": 0.8900103698582786, "grad_norm": 0.632611779607768, "learning_rate": 6.1243734794259e-06, "loss": 0.2709, "step": 12874 }, { "epoch": 0.8900795022468027, "grad_norm": 0.6698375016641781, "learning_rate": 6.12382804988623e-06, "loss": 0.3144, "step": 12875 }, { "epoch": 0.8901486346353267, "grad_norm": 0.6942771977739796, "learning_rate": 6.12328260626138e-06, "loss": 0.3985, "step": 12876 }, { "epoch": 0.8902177670238507, "grad_norm": 0.7696471715814199, "learning_rate": 6.122737148558189e-06, "loss": 0.2615, "step": 12877 }, { "epoch": 0.8902868994123747, "grad_norm": 0.6651676430701573, "learning_rate": 6.122191676783491e-06, "loss": 0.2706, "step": 12878 }, { "epoch": 0.8903560318008987, "grad_norm": 0.7781531467383025, "learning_rate": 6.121646190944122e-06, "loss": 0.3893, "step": 12879 }, { "epoch": 0.8904251641894227, "grad_norm": 0.6877295036523442, "learning_rate": 6.12110069104692e-06, "loss": 0.3398, "step": 12880 }, { "epoch": 0.8904942965779468, "grad_norm": 0.6577498206861584, "learning_rate": 6.120555177098724e-06, "loss": 0.2998, "step": 12881 }, { "epoch": 0.8905634289664708, "grad_norm": 1.1305780760917923, "learning_rate": 6.120009649106366e-06, "loss": 0.3068, "step": 12882 }, { "epoch": 0.8906325613549948, "grad_norm": 0.6735866240976223, "learning_rate": 6.119464107076686e-06, "loss": 0.3756, "step": 12883 }, { "epoch": 0.8907016937435188, "grad_norm": 0.7207655959798241, "learning_rate": 6.118918551016521e-06, "loss": 0.2454, "step": 12884 }, { "epoch": 0.8907708261320428, "grad_norm": 0.8176611877421633, "learning_rate": 6.11837298093271e-06, "loss": 0.2341, "step": 12885 }, { "epoch": 0.8908399585205669, "grad_norm": 0.6579549283744092, "learning_rate": 6.117827396832088e-06, "loss": 0.3173, "step": 12886 }, { "epoch": 0.8909090909090909, "grad_norm": 0.7375390980732571, "learning_rate": 6.117281798721495e-06, "loss": 0.3669, "step": 12887 }, { "epoch": 0.8909782232976149, "grad_norm": 3.693901897444534, "learning_rate": 6.116736186607769e-06, "loss": 0.2538, "step": 12888 }, { "epoch": 0.8910473556861389, "grad_norm": 0.7342439967981133, "learning_rate": 6.1161905604977475e-06, "loss": 0.3226, "step": 12889 }, { "epoch": 0.8911164880746629, "grad_norm": 0.6581538036378158, "learning_rate": 6.11564492039827e-06, "loss": 0.3385, "step": 12890 }, { "epoch": 0.8911856204631871, "grad_norm": 0.6562674423282296, "learning_rate": 6.1150992663161735e-06, "loss": 0.3455, "step": 12891 }, { "epoch": 0.8912547528517111, "grad_norm": 0.6951136651533807, "learning_rate": 6.114553598258297e-06, "loss": 0.2777, "step": 12892 }, { "epoch": 0.8913238852402351, "grad_norm": 0.8619128984255715, "learning_rate": 6.114007916231481e-06, "loss": 0.3327, "step": 12893 }, { "epoch": 0.8913930176287591, "grad_norm": 0.6983464115425972, "learning_rate": 6.1134622202425646e-06, "loss": 0.3513, "step": 12894 }, { "epoch": 0.8914621500172831, "grad_norm": 0.6453742535147424, "learning_rate": 6.1129165102983855e-06, "loss": 0.2849, "step": 12895 }, { "epoch": 0.8915312824058071, "grad_norm": 0.6763196562861284, "learning_rate": 6.112370786405784e-06, "loss": 0.2498, "step": 12896 }, { "epoch": 0.8916004147943312, "grad_norm": 0.6625201551767458, "learning_rate": 6.111825048571599e-06, "loss": 0.3738, "step": 12897 }, { "epoch": 0.8916695471828552, "grad_norm": 0.7011315582682439, "learning_rate": 6.111279296802673e-06, "loss": 0.3572, "step": 12898 }, { "epoch": 0.8917386795713792, "grad_norm": 0.6471589500970563, "learning_rate": 6.110733531105844e-06, "loss": 0.2672, "step": 12899 }, { "epoch": 0.8918078119599032, "grad_norm": 0.7838262022664704, "learning_rate": 6.110187751487951e-06, "loss": 0.2938, "step": 12900 }, { "epoch": 0.8918769443484272, "grad_norm": 0.6778586998213424, "learning_rate": 6.1096419579558356e-06, "loss": 0.487, "step": 12901 }, { "epoch": 0.8919460767369513, "grad_norm": 2.2083493626519752, "learning_rate": 6.109096150516339e-06, "loss": 0.2679, "step": 12902 }, { "epoch": 0.8920152091254753, "grad_norm": 0.806986495496123, "learning_rate": 6.108550329176302e-06, "loss": 0.2998, "step": 12903 }, { "epoch": 0.8920843415139993, "grad_norm": 0.7219943284268288, "learning_rate": 6.108004493942565e-06, "loss": 0.386, "step": 12904 }, { "epoch": 0.8921534739025233, "grad_norm": 0.7524413645631272, "learning_rate": 6.107458644821968e-06, "loss": 0.3811, "step": 12905 }, { "epoch": 0.8922226062910473, "grad_norm": 0.6827652655710151, "learning_rate": 6.106912781821355e-06, "loss": 0.2694, "step": 12906 }, { "epoch": 0.8922917386795713, "grad_norm": 0.6417750771941889, "learning_rate": 6.106366904947564e-06, "loss": 0.2669, "step": 12907 }, { "epoch": 0.8923608710680954, "grad_norm": 0.6420254271088721, "learning_rate": 6.105821014207438e-06, "loss": 0.4553, "step": 12908 }, { "epoch": 0.8924300034566194, "grad_norm": 0.6171909610931788, "learning_rate": 6.10527510960782e-06, "loss": 0.2817, "step": 12909 }, { "epoch": 0.8924991358451434, "grad_norm": 0.5908491397033487, "learning_rate": 6.104729191155551e-06, "loss": 0.2374, "step": 12910 }, { "epoch": 0.8925682682336675, "grad_norm": 0.6895404876274819, "learning_rate": 6.1041832588574735e-06, "loss": 0.3363, "step": 12911 }, { "epoch": 0.8926374006221915, "grad_norm": 0.7435769367450419, "learning_rate": 6.103637312720427e-06, "loss": 0.4144, "step": 12912 }, { "epoch": 0.8927065330107156, "grad_norm": 0.6814681442379971, "learning_rate": 6.103091352751258e-06, "loss": 0.2578, "step": 12913 }, { "epoch": 0.8927756653992396, "grad_norm": 0.6875521429945319, "learning_rate": 6.102545378956808e-06, "loss": 0.2669, "step": 12914 }, { "epoch": 0.8928447977877636, "grad_norm": 0.6551760327390901, "learning_rate": 6.101999391343918e-06, "loss": 0.3741, "step": 12915 }, { "epoch": 0.8929139301762876, "grad_norm": 0.6309470334137026, "learning_rate": 6.101453389919432e-06, "loss": 0.2796, "step": 12916 }, { "epoch": 0.8929830625648116, "grad_norm": 0.6481691896752412, "learning_rate": 6.100907374690196e-06, "loss": 0.2768, "step": 12917 }, { "epoch": 0.8930521949533357, "grad_norm": 0.7013749224421044, "learning_rate": 6.10036134566305e-06, "loss": 0.2875, "step": 12918 }, { "epoch": 0.8931213273418597, "grad_norm": 0.634160448638431, "learning_rate": 6.099815302844836e-06, "loss": 0.3919, "step": 12919 }, { "epoch": 0.8931904597303837, "grad_norm": 0.6915169347598037, "learning_rate": 6.099269246242401e-06, "loss": 0.3231, "step": 12920 }, { "epoch": 0.8932595921189077, "grad_norm": 0.9294031436823919, "learning_rate": 6.0987231758625875e-06, "loss": 0.2373, "step": 12921 }, { "epoch": 0.8933287245074317, "grad_norm": 1.04249083006412, "learning_rate": 6.09817709171224e-06, "loss": 0.3138, "step": 12922 }, { "epoch": 0.8933978568959557, "grad_norm": 0.6654422814963897, "learning_rate": 6.097630993798202e-06, "loss": 0.3655, "step": 12923 }, { "epoch": 0.8934669892844798, "grad_norm": 0.6095804719163568, "learning_rate": 6.09708488212732e-06, "loss": 0.2777, "step": 12924 }, { "epoch": 0.8935361216730038, "grad_norm": 0.7135691663082504, "learning_rate": 6.096538756706436e-06, "loss": 0.2672, "step": 12925 }, { "epoch": 0.8936052540615278, "grad_norm": 0.9223949095553539, "learning_rate": 6.0959926175423945e-06, "loss": 0.4423, "step": 12926 }, { "epoch": 0.8936743864500518, "grad_norm": 0.6277535803565604, "learning_rate": 6.095446464642042e-06, "loss": 0.2626, "step": 12927 }, { "epoch": 0.8937435188385758, "grad_norm": 0.7552853815142696, "learning_rate": 6.094900298012223e-06, "loss": 0.2905, "step": 12928 }, { "epoch": 0.8938126512270999, "grad_norm": 0.6894691035715241, "learning_rate": 6.0943541176597835e-06, "loss": 0.3523, "step": 12929 }, { "epoch": 0.8938817836156239, "grad_norm": 0.7116680436886915, "learning_rate": 6.093807923591566e-06, "loss": 0.4046, "step": 12930 }, { "epoch": 0.8939509160041479, "grad_norm": 1.0760123259486534, "learning_rate": 6.09326171581442e-06, "loss": 0.2693, "step": 12931 }, { "epoch": 0.894020048392672, "grad_norm": 0.8651535927521894, "learning_rate": 6.092715494335189e-06, "loss": 0.3508, "step": 12932 }, { "epoch": 0.894089180781196, "grad_norm": 0.7724340789958676, "learning_rate": 6.09216925916072e-06, "loss": 0.3968, "step": 12933 }, { "epoch": 0.89415831316972, "grad_norm": 0.6284374782104061, "learning_rate": 6.091623010297857e-06, "loss": 0.3282, "step": 12934 }, { "epoch": 0.8942274455582441, "grad_norm": 0.6620248608632119, "learning_rate": 6.091076747753448e-06, "loss": 0.2671, "step": 12935 }, { "epoch": 0.8942965779467681, "grad_norm": 0.7675761654632512, "learning_rate": 6.09053047153434e-06, "loss": 0.3719, "step": 12936 }, { "epoch": 0.8943657103352921, "grad_norm": 0.806527753202469, "learning_rate": 6.089984181647378e-06, "loss": 0.3488, "step": 12937 }, { "epoch": 0.8944348427238161, "grad_norm": 0.6129793485898392, "learning_rate": 6.089437878099409e-06, "loss": 0.3103, "step": 12938 }, { "epoch": 0.8945039751123401, "grad_norm": 0.6530830141398309, "learning_rate": 6.088891560897283e-06, "loss": 0.2596, "step": 12939 }, { "epoch": 0.8945731075008642, "grad_norm": 0.6830254919924107, "learning_rate": 6.088345230047842e-06, "loss": 0.4326, "step": 12940 }, { "epoch": 0.8946422398893882, "grad_norm": 1.6664408537452728, "learning_rate": 6.087798885557936e-06, "loss": 0.2863, "step": 12941 }, { "epoch": 0.8947113722779122, "grad_norm": 0.6412349192991861, "learning_rate": 6.087252527434413e-06, "loss": 0.2322, "step": 12942 }, { "epoch": 0.8947805046664362, "grad_norm": 0.6704741489353919, "learning_rate": 6.086706155684118e-06, "loss": 0.3444, "step": 12943 }, { "epoch": 0.8948496370549602, "grad_norm": 0.7532497702823203, "learning_rate": 6.086159770313902e-06, "loss": 0.3969, "step": 12944 }, { "epoch": 0.8949187694434843, "grad_norm": 0.8171344718572989, "learning_rate": 6.085613371330612e-06, "loss": 0.2553, "step": 12945 }, { "epoch": 0.8949879018320083, "grad_norm": 0.7747826866953269, "learning_rate": 6.0850669587410956e-06, "loss": 0.2535, "step": 12946 }, { "epoch": 0.8950570342205323, "grad_norm": 0.6539470667218239, "learning_rate": 6.0845205325522006e-06, "loss": 0.3915, "step": 12947 }, { "epoch": 0.8951261666090563, "grad_norm": 0.6629211045088027, "learning_rate": 6.083974092770777e-06, "loss": 0.3452, "step": 12948 }, { "epoch": 0.8951952989975803, "grad_norm": 0.7325985744973466, "learning_rate": 6.083427639403671e-06, "loss": 0.3135, "step": 12949 }, { "epoch": 0.8952644313861043, "grad_norm": 0.8971187267857759, "learning_rate": 6.082881172457734e-06, "loss": 0.3125, "step": 12950 }, { "epoch": 0.8953335637746284, "grad_norm": 0.8474088379879171, "learning_rate": 6.082334691939815e-06, "loss": 0.3592, "step": 12951 }, { "epoch": 0.8954026961631525, "grad_norm": 0.5872747565952585, "learning_rate": 6.081788197856761e-06, "loss": 0.2573, "step": 12952 }, { "epoch": 0.8954718285516765, "grad_norm": 0.6541800068995148, "learning_rate": 6.081241690215424e-06, "loss": 0.2517, "step": 12953 }, { "epoch": 0.8955409609402005, "grad_norm": 0.7229974341490275, "learning_rate": 6.080695169022649e-06, "loss": 0.3402, "step": 12954 }, { "epoch": 0.8956100933287245, "grad_norm": 0.931766540029086, "learning_rate": 6.080148634285291e-06, "loss": 0.3626, "step": 12955 }, { "epoch": 0.8956792257172486, "grad_norm": 0.7532247045381352, "learning_rate": 6.079602086010197e-06, "loss": 0.3492, "step": 12956 }, { "epoch": 0.8957483581057726, "grad_norm": 0.6506929548907348, "learning_rate": 6.079055524204217e-06, "loss": 0.2894, "step": 12957 }, { "epoch": 0.8958174904942966, "grad_norm": 0.6632780741079927, "learning_rate": 6.0785089488742026e-06, "loss": 0.3754, "step": 12958 }, { "epoch": 0.8958866228828206, "grad_norm": 0.7766164259690008, "learning_rate": 6.0779623600270025e-06, "loss": 0.3334, "step": 12959 }, { "epoch": 0.8959557552713446, "grad_norm": 0.6431145650673897, "learning_rate": 6.0774157576694685e-06, "loss": 0.2923, "step": 12960 }, { "epoch": 0.8960248876598687, "grad_norm": 0.7431773651059653, "learning_rate": 6.076869141808449e-06, "loss": 0.3177, "step": 12961 }, { "epoch": 0.8960940200483927, "grad_norm": 0.7355160918162801, "learning_rate": 6.076322512450799e-06, "loss": 0.3407, "step": 12962 }, { "epoch": 0.8961631524369167, "grad_norm": 0.6593262960508266, "learning_rate": 6.075775869603365e-06, "loss": 0.3163, "step": 12963 }, { "epoch": 0.8962322848254407, "grad_norm": 0.7125548279653845, "learning_rate": 6.075229213273002e-06, "loss": 0.353, "step": 12964 }, { "epoch": 0.8963014172139647, "grad_norm": 0.6612133222141483, "learning_rate": 6.074682543466558e-06, "loss": 0.3835, "step": 12965 }, { "epoch": 0.8963705496024887, "grad_norm": 0.6145651076315349, "learning_rate": 6.074135860190887e-06, "loss": 0.2771, "step": 12966 }, { "epoch": 0.8964396819910128, "grad_norm": 0.6720337916652659, "learning_rate": 6.07358916345284e-06, "loss": 0.3542, "step": 12967 }, { "epoch": 0.8965088143795368, "grad_norm": 0.7366745932292109, "learning_rate": 6.073042453259268e-06, "loss": 0.329, "step": 12968 }, { "epoch": 0.8965779467680608, "grad_norm": 0.7126433928599925, "learning_rate": 6.072495729617025e-06, "loss": 0.3882, "step": 12969 }, { "epoch": 0.8966470791565848, "grad_norm": 0.9664338933110016, "learning_rate": 6.071948992532961e-06, "loss": 0.2932, "step": 12970 }, { "epoch": 0.8967162115451088, "grad_norm": 0.6661143187529424, "learning_rate": 6.071402242013931e-06, "loss": 0.2796, "step": 12971 }, { "epoch": 0.896785343933633, "grad_norm": 0.6305273400363571, "learning_rate": 6.070855478066784e-06, "loss": 0.3807, "step": 12972 }, { "epoch": 0.896854476322157, "grad_norm": 0.6559239035740524, "learning_rate": 6.070308700698375e-06, "loss": 0.324, "step": 12973 }, { "epoch": 0.896923608710681, "grad_norm": 0.6409049299346548, "learning_rate": 6.069761909915557e-06, "loss": 0.3059, "step": 12974 }, { "epoch": 0.896992741099205, "grad_norm": 0.6840242938732263, "learning_rate": 6.069215105725182e-06, "loss": 0.3465, "step": 12975 }, { "epoch": 0.897061873487729, "grad_norm": 0.7078983916377953, "learning_rate": 6.068668288134104e-06, "loss": 0.3943, "step": 12976 }, { "epoch": 0.897131005876253, "grad_norm": 0.6348889903538298, "learning_rate": 6.068121457149177e-06, "loss": 0.3222, "step": 12977 }, { "epoch": 0.8972001382647771, "grad_norm": 0.6326964605680276, "learning_rate": 6.067574612777252e-06, "loss": 0.3134, "step": 12978 }, { "epoch": 0.8972692706533011, "grad_norm": 0.6660884360791801, "learning_rate": 6.0670277550251845e-06, "loss": 0.3325, "step": 12979 }, { "epoch": 0.8973384030418251, "grad_norm": 0.7212612339972267, "learning_rate": 6.06648088389983e-06, "loss": 0.3628, "step": 12980 }, { "epoch": 0.8974075354303491, "grad_norm": 0.6254513132879702, "learning_rate": 6.06593399940804e-06, "loss": 0.2356, "step": 12981 }, { "epoch": 0.8974766678188731, "grad_norm": 0.7421714619803652, "learning_rate": 6.065387101556669e-06, "loss": 0.4, "step": 12982 }, { "epoch": 0.8975458002073972, "grad_norm": 0.6308276361332469, "learning_rate": 6.064840190352573e-06, "loss": 0.3867, "step": 12983 }, { "epoch": 0.8976149325959212, "grad_norm": 0.7432111982780092, "learning_rate": 6.0642932658026045e-06, "loss": 0.3267, "step": 12984 }, { "epoch": 0.8976840649844452, "grad_norm": 0.6301823853949057, "learning_rate": 6.06374632791362e-06, "loss": 0.3136, "step": 12985 }, { "epoch": 0.8977531973729692, "grad_norm": 0.6743710910284417, "learning_rate": 6.063199376692472e-06, "loss": 0.3424, "step": 12986 }, { "epoch": 0.8978223297614932, "grad_norm": 1.0463238117869402, "learning_rate": 6.0626524121460185e-06, "loss": 0.4034, "step": 12987 }, { "epoch": 0.8978914621500172, "grad_norm": 0.5986941717018512, "learning_rate": 6.062105434281113e-06, "loss": 0.2929, "step": 12988 }, { "epoch": 0.8979605945385413, "grad_norm": 0.6706430556562544, "learning_rate": 6.061558443104613e-06, "loss": 0.2441, "step": 12989 }, { "epoch": 0.8980297269270653, "grad_norm": 0.6579066925018255, "learning_rate": 6.06101143862337e-06, "loss": 0.4047, "step": 12990 }, { "epoch": 0.8980988593155893, "grad_norm": 0.8772526184251993, "learning_rate": 6.060464420844243e-06, "loss": 0.3331, "step": 12991 }, { "epoch": 0.8981679917041133, "grad_norm": 0.6288839333891352, "learning_rate": 6.059917389774087e-06, "loss": 0.2782, "step": 12992 }, { "epoch": 0.8982371240926375, "grad_norm": 0.6839118902024526, "learning_rate": 6.059370345419758e-06, "loss": 0.3011, "step": 12993 }, { "epoch": 0.8983062564811615, "grad_norm": 0.6530090577618033, "learning_rate": 6.058823287788112e-06, "loss": 0.3903, "step": 12994 }, { "epoch": 0.8983753888696855, "grad_norm": 0.711870458365136, "learning_rate": 6.058276216886007e-06, "loss": 0.3057, "step": 12995 }, { "epoch": 0.8984445212582095, "grad_norm": 0.6923273029791468, "learning_rate": 6.057729132720297e-06, "loss": 0.2678, "step": 12996 }, { "epoch": 0.8985136536467335, "grad_norm": 0.6687605795511816, "learning_rate": 6.057182035297839e-06, "loss": 0.3727, "step": 12997 }, { "epoch": 0.8985827860352575, "grad_norm": 0.667037070071461, "learning_rate": 6.056634924625492e-06, "loss": 0.2885, "step": 12998 }, { "epoch": 0.8986519184237816, "grad_norm": 0.6315486855603293, "learning_rate": 6.056087800710112e-06, "loss": 0.2687, "step": 12999 }, { "epoch": 0.8987210508123056, "grad_norm": 0.6567422081379217, "learning_rate": 6.055540663558557e-06, "loss": 0.3195, "step": 13000 }, { "epoch": 0.8987901832008296, "grad_norm": 0.6640626720428731, "learning_rate": 6.0549935131776805e-06, "loss": 0.4008, "step": 13001 }, { "epoch": 0.8988593155893536, "grad_norm": 0.6883539666960822, "learning_rate": 6.054446349574345e-06, "loss": 0.2762, "step": 13002 }, { "epoch": 0.8989284479778776, "grad_norm": 0.6760164728324926, "learning_rate": 6.0538991727554066e-06, "loss": 0.3215, "step": 13003 }, { "epoch": 0.8989975803664016, "grad_norm": 0.6410634943046619, "learning_rate": 6.0533519827277224e-06, "loss": 0.3671, "step": 13004 }, { "epoch": 0.8990667127549257, "grad_norm": 0.6473484635909447, "learning_rate": 6.0528047794981515e-06, "loss": 0.4105, "step": 13005 }, { "epoch": 0.8991358451434497, "grad_norm": 0.5926434944266487, "learning_rate": 6.052257563073551e-06, "loss": 0.3006, "step": 13006 }, { "epoch": 0.8992049775319737, "grad_norm": 0.6860259182361614, "learning_rate": 6.051710333460778e-06, "loss": 0.372, "step": 13007 }, { "epoch": 0.8992741099204977, "grad_norm": 0.7736587813614648, "learning_rate": 6.051163090666697e-06, "loss": 0.4034, "step": 13008 }, { "epoch": 0.8993432423090217, "grad_norm": 0.9027886080426413, "learning_rate": 6.050615834698159e-06, "loss": 0.265, "step": 13009 }, { "epoch": 0.8994123746975458, "grad_norm": 0.6620634381152176, "learning_rate": 6.0500685655620285e-06, "loss": 0.253, "step": 13010 }, { "epoch": 0.8994815070860698, "grad_norm": 0.7141445541473977, "learning_rate": 6.04952128326516e-06, "loss": 0.425, "step": 13011 }, { "epoch": 0.8995506394745938, "grad_norm": 0.6383145923145497, "learning_rate": 6.048973987814417e-06, "loss": 0.3497, "step": 13012 }, { "epoch": 0.8996197718631179, "grad_norm": 0.5814936320598012, "learning_rate": 6.048426679216657e-06, "loss": 0.2412, "step": 13013 }, { "epoch": 0.8996889042516419, "grad_norm": 0.6919094253871091, "learning_rate": 6.047879357478738e-06, "loss": 0.2788, "step": 13014 }, { "epoch": 0.899758036640166, "grad_norm": 0.727116938543503, "learning_rate": 6.0473320226075225e-06, "loss": 0.3713, "step": 13015 }, { "epoch": 0.89982716902869, "grad_norm": 0.6501809832768817, "learning_rate": 6.046784674609869e-06, "loss": 0.3083, "step": 13016 }, { "epoch": 0.899896301417214, "grad_norm": 0.6129676199117647, "learning_rate": 6.0462373134926386e-06, "loss": 0.3331, "step": 13017 }, { "epoch": 0.899965433805738, "grad_norm": 0.6922945977174052, "learning_rate": 6.045689939262689e-06, "loss": 0.3082, "step": 13018 }, { "epoch": 0.900034566194262, "grad_norm": 0.710338462966521, "learning_rate": 6.0451425519268826e-06, "loss": 0.4563, "step": 13019 }, { "epoch": 0.900103698582786, "grad_norm": 0.6156254773360682, "learning_rate": 6.044595151492081e-06, "loss": 0.328, "step": 13020 }, { "epoch": 0.9001728309713101, "grad_norm": 0.6601646405544226, "learning_rate": 6.044047737965141e-06, "loss": 0.2536, "step": 13021 }, { "epoch": 0.9002419633598341, "grad_norm": 0.6657910463088431, "learning_rate": 6.043500311352927e-06, "loss": 0.3769, "step": 13022 }, { "epoch": 0.9003110957483581, "grad_norm": 0.6975259147430398, "learning_rate": 6.042952871662299e-06, "loss": 0.3456, "step": 13023 }, { "epoch": 0.9003802281368821, "grad_norm": 0.8221510666455882, "learning_rate": 6.042405418900117e-06, "loss": 0.2514, "step": 13024 }, { "epoch": 0.9004493605254061, "grad_norm": 0.711598586172487, "learning_rate": 6.041857953073244e-06, "loss": 0.2778, "step": 13025 }, { "epoch": 0.9005184929139302, "grad_norm": 1.383718928920832, "learning_rate": 6.041310474188541e-06, "loss": 0.4207, "step": 13026 }, { "epoch": 0.9005876253024542, "grad_norm": 0.6447774481175292, "learning_rate": 6.040762982252869e-06, "loss": 0.3055, "step": 13027 }, { "epoch": 0.9006567576909782, "grad_norm": 0.80825046742486, "learning_rate": 6.040215477273091e-06, "loss": 0.2655, "step": 13028 }, { "epoch": 0.9007258900795022, "grad_norm": 0.6211144167587493, "learning_rate": 6.039667959256067e-06, "loss": 0.3174, "step": 13029 }, { "epoch": 0.9007950224680262, "grad_norm": 0.6852771797299833, "learning_rate": 6.039120428208663e-06, "loss": 0.3546, "step": 13030 }, { "epoch": 0.9008641548565502, "grad_norm": 0.64517984497823, "learning_rate": 6.0385728841377364e-06, "loss": 0.303, "step": 13031 }, { "epoch": 0.9009332872450743, "grad_norm": 0.7231759461172741, "learning_rate": 6.0380253270501525e-06, "loss": 0.2923, "step": 13032 }, { "epoch": 0.9010024196335984, "grad_norm": 0.6266851308173959, "learning_rate": 6.037477756952774e-06, "loss": 0.3803, "step": 13033 }, { "epoch": 0.9010715520221224, "grad_norm": 0.6073085887589379, "learning_rate": 6.036930173852465e-06, "loss": 0.2907, "step": 13034 }, { "epoch": 0.9011406844106464, "grad_norm": 0.6407116379298604, "learning_rate": 6.036382577756085e-06, "loss": 0.2435, "step": 13035 }, { "epoch": 0.9012098167991704, "grad_norm": 0.6456748692046212, "learning_rate": 6.035834968670497e-06, "loss": 0.3429, "step": 13036 }, { "epoch": 0.9012789491876945, "grad_norm": 0.6897712061752088, "learning_rate": 6.03528734660257e-06, "loss": 0.3815, "step": 13037 }, { "epoch": 0.9013480815762185, "grad_norm": 1.274771297370178, "learning_rate": 6.034739711559161e-06, "loss": 0.2596, "step": 13038 }, { "epoch": 0.9014172139647425, "grad_norm": 0.648244255308176, "learning_rate": 6.034192063547137e-06, "loss": 0.2694, "step": 13039 }, { "epoch": 0.9014863463532665, "grad_norm": 0.688404670809966, "learning_rate": 6.033644402573361e-06, "loss": 0.4297, "step": 13040 }, { "epoch": 0.9015554787417905, "grad_norm": 0.70134622815481, "learning_rate": 6.033096728644699e-06, "loss": 0.2975, "step": 13041 }, { "epoch": 0.9016246111303146, "grad_norm": 0.6406274429347038, "learning_rate": 6.032549041768011e-06, "loss": 0.277, "step": 13042 }, { "epoch": 0.9016937435188386, "grad_norm": 0.6650810815143741, "learning_rate": 6.032001341950162e-06, "loss": 0.3434, "step": 13043 }, { "epoch": 0.9017628759073626, "grad_norm": 0.6724746307548879, "learning_rate": 6.031453629198021e-06, "loss": 0.395, "step": 13044 }, { "epoch": 0.9018320082958866, "grad_norm": 0.6294375552938855, "learning_rate": 6.030905903518448e-06, "loss": 0.3431, "step": 13045 }, { "epoch": 0.9019011406844106, "grad_norm": 2.0619885336589823, "learning_rate": 6.030358164918308e-06, "loss": 0.3016, "step": 13046 }, { "epoch": 0.9019702730729346, "grad_norm": 0.6712176022296541, "learning_rate": 6.02981041340447e-06, "loss": 0.3271, "step": 13047 }, { "epoch": 0.9020394054614587, "grad_norm": 0.6713771087390744, "learning_rate": 6.029262648983795e-06, "loss": 0.3696, "step": 13048 }, { "epoch": 0.9021085378499827, "grad_norm": 0.6717561011484382, "learning_rate": 6.0287148716631495e-06, "loss": 0.2745, "step": 13049 }, { "epoch": 0.9021776702385067, "grad_norm": 0.7559785068347826, "learning_rate": 6.028167081449397e-06, "loss": 0.3047, "step": 13050 }, { "epoch": 0.9022468026270307, "grad_norm": 0.6537364997072197, "learning_rate": 6.027619278349408e-06, "loss": 0.4055, "step": 13051 }, { "epoch": 0.9023159350155547, "grad_norm": 0.697438661507441, "learning_rate": 6.0270714623700444e-06, "loss": 0.2706, "step": 13052 }, { "epoch": 0.9023850674040788, "grad_norm": 0.6309661651603624, "learning_rate": 6.026523633518172e-06, "loss": 0.3126, "step": 13053 }, { "epoch": 0.9024541997926029, "grad_norm": 0.8465459775274257, "learning_rate": 6.025975791800658e-06, "loss": 0.3736, "step": 13054 }, { "epoch": 0.9025233321811269, "grad_norm": 0.6645382315596222, "learning_rate": 6.025427937224369e-06, "loss": 0.3763, "step": 13055 }, { "epoch": 0.9025924645696509, "grad_norm": 0.5954921352131257, "learning_rate": 6.0248800697961704e-06, "loss": 0.2908, "step": 13056 }, { "epoch": 0.9026615969581749, "grad_norm": 0.6523510925717121, "learning_rate": 6.0243321895229275e-06, "loss": 0.3106, "step": 13057 }, { "epoch": 0.902730729346699, "grad_norm": 0.6831140259640391, "learning_rate": 6.023784296411511e-06, "loss": 0.3988, "step": 13058 }, { "epoch": 0.902799861735223, "grad_norm": 0.6543900674373374, "learning_rate": 6.023236390468786e-06, "loss": 0.2685, "step": 13059 }, { "epoch": 0.902868994123747, "grad_norm": 1.1256018284874036, "learning_rate": 6.022688471701616e-06, "loss": 0.2841, "step": 13060 }, { "epoch": 0.902938126512271, "grad_norm": 0.6737205932151913, "learning_rate": 6.022140540116873e-06, "loss": 0.3374, "step": 13061 }, { "epoch": 0.903007258900795, "grad_norm": 0.6505065982066787, "learning_rate": 6.021592595721422e-06, "loss": 0.382, "step": 13062 }, { "epoch": 0.903076391289319, "grad_norm": 0.6877778986013828, "learning_rate": 6.021044638522131e-06, "loss": 0.3267, "step": 13063 }, { "epoch": 0.9031455236778431, "grad_norm": 0.6739225619682683, "learning_rate": 6.020496668525867e-06, "loss": 0.3555, "step": 13064 }, { "epoch": 0.9032146560663671, "grad_norm": 0.6609753386333432, "learning_rate": 6.019948685739499e-06, "loss": 0.4008, "step": 13065 }, { "epoch": 0.9032837884548911, "grad_norm": 0.7179924598546545, "learning_rate": 6.0194006901698946e-06, "loss": 0.3477, "step": 13066 }, { "epoch": 0.9033529208434151, "grad_norm": 0.6421057340573414, "learning_rate": 6.0188526818239215e-06, "loss": 0.2365, "step": 13067 }, { "epoch": 0.9034220532319391, "grad_norm": 0.6391356493432631, "learning_rate": 6.018304660708447e-06, "loss": 0.2967, "step": 13068 }, { "epoch": 0.9034911856204632, "grad_norm": 0.7213507549491682, "learning_rate": 6.017756626830342e-06, "loss": 0.333, "step": 13069 }, { "epoch": 0.9035603180089872, "grad_norm": 0.6549916826359213, "learning_rate": 6.017208580196475e-06, "loss": 0.3276, "step": 13070 }, { "epoch": 0.9036294503975112, "grad_norm": 0.6495971132007603, "learning_rate": 6.016660520813713e-06, "loss": 0.2468, "step": 13071 }, { "epoch": 0.9036985827860352, "grad_norm": 0.6863788013810725, "learning_rate": 6.016112448688926e-06, "loss": 0.3549, "step": 13072 }, { "epoch": 0.9037677151745592, "grad_norm": 0.619352105975567, "learning_rate": 6.015564363828982e-06, "loss": 0.3339, "step": 13073 }, { "epoch": 0.9038368475630834, "grad_norm": 0.6065024504595495, "learning_rate": 6.015016266240751e-06, "loss": 0.2613, "step": 13074 }, { "epoch": 0.9039059799516074, "grad_norm": 0.6911417554497609, "learning_rate": 6.014468155931102e-06, "loss": 0.3126, "step": 13075 }, { "epoch": 0.9039751123401314, "grad_norm": 0.6573653167529336, "learning_rate": 6.013920032906907e-06, "loss": 0.4411, "step": 13076 }, { "epoch": 0.9040442447286554, "grad_norm": 0.7359021452523338, "learning_rate": 6.0133718971750325e-06, "loss": 0.2821, "step": 13077 }, { "epoch": 0.9041133771171794, "grad_norm": 0.5865823737048692, "learning_rate": 6.0128237487423494e-06, "loss": 0.2284, "step": 13078 }, { "epoch": 0.9041825095057034, "grad_norm": 0.8929450042104194, "learning_rate": 6.012275587615728e-06, "loss": 0.3237, "step": 13079 }, { "epoch": 0.9042516418942275, "grad_norm": 1.1440348683826835, "learning_rate": 6.01172741380204e-06, "loss": 0.3449, "step": 13080 }, { "epoch": 0.9043207742827515, "grad_norm": 0.5788183921776296, "learning_rate": 6.011179227308153e-06, "loss": 0.2096, "step": 13081 }, { "epoch": 0.9043899066712755, "grad_norm": 0.771019116016675, "learning_rate": 6.010631028140939e-06, "loss": 0.3417, "step": 13082 }, { "epoch": 0.9044590390597995, "grad_norm": 0.6452559696442777, "learning_rate": 6.01008281630727e-06, "loss": 0.3591, "step": 13083 }, { "epoch": 0.9045281714483235, "grad_norm": 0.7181996613433052, "learning_rate": 6.009534591814014e-06, "loss": 0.2671, "step": 13084 }, { "epoch": 0.9045973038368476, "grad_norm": 0.6577554082673678, "learning_rate": 6.008986354668043e-06, "loss": 0.2694, "step": 13085 }, { "epoch": 0.9046664362253716, "grad_norm": 0.8631369036503201, "learning_rate": 6.008438104876229e-06, "loss": 0.3598, "step": 13086 }, { "epoch": 0.9047355686138956, "grad_norm": 0.7220032266938294, "learning_rate": 6.007889842445444e-06, "loss": 0.3436, "step": 13087 }, { "epoch": 0.9048047010024196, "grad_norm": 0.6777364999041217, "learning_rate": 6.00734156738256e-06, "loss": 0.3685, "step": 13088 }, { "epoch": 0.9048738333909436, "grad_norm": 0.6721730587252065, "learning_rate": 6.006793279694444e-06, "loss": 0.2924, "step": 13089 }, { "epoch": 0.9049429657794676, "grad_norm": 0.696745127430973, "learning_rate": 6.0062449793879716e-06, "loss": 0.389, "step": 13090 }, { "epoch": 0.9050120981679917, "grad_norm": 0.6866451846517185, "learning_rate": 6.0056966664700155e-06, "loss": 0.3207, "step": 13091 }, { "epoch": 0.9050812305565157, "grad_norm": 0.637039623763438, "learning_rate": 6.005148340947445e-06, "loss": 0.2448, "step": 13092 }, { "epoch": 0.9051503629450397, "grad_norm": 0.6563467484491349, "learning_rate": 6.004600002827134e-06, "loss": 0.3248, "step": 13093 }, { "epoch": 0.9052194953335638, "grad_norm": 0.685099249092064, "learning_rate": 6.0040516521159565e-06, "loss": 0.4462, "step": 13094 }, { "epoch": 0.9052886277220878, "grad_norm": 0.733088046305556, "learning_rate": 6.003503288820782e-06, "loss": 0.2474, "step": 13095 }, { "epoch": 0.9053577601106119, "grad_norm": 0.6170742022766392, "learning_rate": 6.0029549129484845e-06, "loss": 0.306, "step": 13096 }, { "epoch": 0.9054268924991359, "grad_norm": 0.6701183769099209, "learning_rate": 6.002406524505937e-06, "loss": 0.3492, "step": 13097 }, { "epoch": 0.9054960248876599, "grad_norm": 0.7184026364045369, "learning_rate": 6.001858123500013e-06, "loss": 0.3877, "step": 13098 }, { "epoch": 0.9055651572761839, "grad_norm": 0.5939516091250576, "learning_rate": 6.001309709937585e-06, "loss": 0.28, "step": 13099 }, { "epoch": 0.9056342896647079, "grad_norm": 0.6428946858402704, "learning_rate": 6.0007612838255256e-06, "loss": 0.3009, "step": 13100 }, { "epoch": 0.905703422053232, "grad_norm": 2.1494103392556183, "learning_rate": 6.000212845170711e-06, "loss": 0.4126, "step": 13101 }, { "epoch": 0.905772554441756, "grad_norm": 0.603460950420995, "learning_rate": 5.999664393980014e-06, "loss": 0.2991, "step": 13102 }, { "epoch": 0.90584168683028, "grad_norm": 0.6233140188153892, "learning_rate": 5.999115930260305e-06, "loss": 0.262, "step": 13103 }, { "epoch": 0.905910819218804, "grad_norm": 0.6940328277221213, "learning_rate": 5.998567454018463e-06, "loss": 0.4259, "step": 13104 }, { "epoch": 0.905979951607328, "grad_norm": 0.6699276009277988, "learning_rate": 5.9980189652613606e-06, "loss": 0.3588, "step": 13105 }, { "epoch": 0.906049083995852, "grad_norm": 0.5884710026138645, "learning_rate": 5.99747046399587e-06, "loss": 0.2603, "step": 13106 }, { "epoch": 0.9061182163843761, "grad_norm": 0.6165928944021329, "learning_rate": 5.996921950228867e-06, "loss": 0.2462, "step": 13107 }, { "epoch": 0.9061873487729001, "grad_norm": 0.6366011197464436, "learning_rate": 5.9963734239672275e-06, "loss": 0.3892, "step": 13108 }, { "epoch": 0.9062564811614241, "grad_norm": 0.6420849494429538, "learning_rate": 5.9958248852178245e-06, "loss": 0.3259, "step": 13109 }, { "epoch": 0.9063256135499481, "grad_norm": 0.6310554722568875, "learning_rate": 5.995276333987534e-06, "loss": 0.261, "step": 13110 }, { "epoch": 0.9063947459384721, "grad_norm": 0.6662707637848593, "learning_rate": 5.994727770283231e-06, "loss": 0.3389, "step": 13111 }, { "epoch": 0.9064638783269962, "grad_norm": 0.7177662689381012, "learning_rate": 5.994179194111791e-06, "loss": 0.4036, "step": 13112 }, { "epoch": 0.9065330107155202, "grad_norm": 0.6253387850153032, "learning_rate": 5.993630605480089e-06, "loss": 0.2428, "step": 13113 }, { "epoch": 0.9066021431040443, "grad_norm": 0.6080899483493856, "learning_rate": 5.993082004394999e-06, "loss": 0.2979, "step": 13114 }, { "epoch": 0.9066712754925683, "grad_norm": 0.7056047734522432, "learning_rate": 5.9925333908634e-06, "loss": 0.4721, "step": 13115 }, { "epoch": 0.9067404078810923, "grad_norm": 0.7981517461365406, "learning_rate": 5.991984764892164e-06, "loss": 0.3292, "step": 13116 }, { "epoch": 0.9068095402696164, "grad_norm": 0.8250491410677853, "learning_rate": 5.99143612648817e-06, "loss": 0.2947, "step": 13117 }, { "epoch": 0.9068786726581404, "grad_norm": 0.748092684156215, "learning_rate": 5.9908874756582955e-06, "loss": 0.3344, "step": 13118 }, { "epoch": 0.9069478050466644, "grad_norm": 0.6644687461417234, "learning_rate": 5.990338812409414e-06, "loss": 0.3891, "step": 13119 }, { "epoch": 0.9070169374351884, "grad_norm": 0.6098587446280233, "learning_rate": 5.989790136748402e-06, "loss": 0.2236, "step": 13120 }, { "epoch": 0.9070860698237124, "grad_norm": 0.7680828394418173, "learning_rate": 5.989241448682137e-06, "loss": 0.2801, "step": 13121 }, { "epoch": 0.9071552022122364, "grad_norm": 0.6877813957712555, "learning_rate": 5.988692748217496e-06, "loss": 0.3708, "step": 13122 }, { "epoch": 0.9072243346007605, "grad_norm": 0.651712529648937, "learning_rate": 5.988144035361356e-06, "loss": 0.3164, "step": 13123 }, { "epoch": 0.9072934669892845, "grad_norm": 0.6366835230179653, "learning_rate": 5.987595310120595e-06, "loss": 0.2625, "step": 13124 }, { "epoch": 0.9073625993778085, "grad_norm": 0.6504574875181445, "learning_rate": 5.987046572502087e-06, "loss": 0.2932, "step": 13125 }, { "epoch": 0.9074317317663325, "grad_norm": 0.6508551437823971, "learning_rate": 5.986497822512713e-06, "loss": 0.3907, "step": 13126 }, { "epoch": 0.9075008641548565, "grad_norm": 0.6247514236901884, "learning_rate": 5.985949060159349e-06, "loss": 0.2481, "step": 13127 }, { "epoch": 0.9075699965433806, "grad_norm": 0.5977861555290913, "learning_rate": 5.985400285448873e-06, "loss": 0.2227, "step": 13128 }, { "epoch": 0.9076391289319046, "grad_norm": 0.6658265618087887, "learning_rate": 5.984851498388163e-06, "loss": 0.4138, "step": 13129 }, { "epoch": 0.9077082613204286, "grad_norm": 0.6435751884229594, "learning_rate": 5.984302698984098e-06, "loss": 0.3418, "step": 13130 }, { "epoch": 0.9077773937089526, "grad_norm": 0.6782039019262578, "learning_rate": 5.9837538872435545e-06, "loss": 0.2457, "step": 13131 }, { "epoch": 0.9078465260974766, "grad_norm": 0.8183902287926016, "learning_rate": 5.9832050631734105e-06, "loss": 0.2809, "step": 13132 }, { "epoch": 0.9079156584860006, "grad_norm": 0.661326914098253, "learning_rate": 5.9826562267805465e-06, "loss": 0.368, "step": 13133 }, { "epoch": 0.9079847908745247, "grad_norm": 0.6640265744963988, "learning_rate": 5.98210737807184e-06, "loss": 0.2806, "step": 13134 }, { "epoch": 0.9080539232630488, "grad_norm": 0.5741365019479855, "learning_rate": 5.98155851705417e-06, "loss": 0.2728, "step": 13135 }, { "epoch": 0.9081230556515728, "grad_norm": 0.7940797264634806, "learning_rate": 5.981009643734417e-06, "loss": 0.3955, "step": 13136 }, { "epoch": 0.9081921880400968, "grad_norm": 0.6692699971656785, "learning_rate": 5.980460758119459e-06, "loss": 0.3616, "step": 13137 }, { "epoch": 0.9082613204286208, "grad_norm": 0.8049014922780917, "learning_rate": 5.979911860216174e-06, "loss": 0.2984, "step": 13138 }, { "epoch": 0.9083304528171449, "grad_norm": 0.7202846428684198, "learning_rate": 5.979362950031443e-06, "loss": 0.3444, "step": 13139 }, { "epoch": 0.9083995852056689, "grad_norm": 0.7848956688399978, "learning_rate": 5.978814027572145e-06, "loss": 0.3757, "step": 13140 }, { "epoch": 0.9084687175941929, "grad_norm": 0.676531629770606, "learning_rate": 5.97826509284516e-06, "loss": 0.3153, "step": 13141 }, { "epoch": 0.9085378499827169, "grad_norm": 0.6393263298826065, "learning_rate": 5.977716145857366e-06, "loss": 0.3082, "step": 13142 }, { "epoch": 0.9086069823712409, "grad_norm": 0.6860223570587738, "learning_rate": 5.977167186615648e-06, "loss": 0.3238, "step": 13143 }, { "epoch": 0.908676114759765, "grad_norm": 0.679958692408316, "learning_rate": 5.9766182151268815e-06, "loss": 0.3901, "step": 13144 }, { "epoch": 0.908745247148289, "grad_norm": 0.7799129480361255, "learning_rate": 5.976069231397948e-06, "loss": 0.2806, "step": 13145 }, { "epoch": 0.908814379536813, "grad_norm": 0.5855363240275031, "learning_rate": 5.97552023543573e-06, "loss": 0.2561, "step": 13146 }, { "epoch": 0.908883511925337, "grad_norm": 0.6944232390630203, "learning_rate": 5.974971227247106e-06, "loss": 0.4099, "step": 13147 }, { "epoch": 0.908952644313861, "grad_norm": 1.3773941420172007, "learning_rate": 5.9744222068389576e-06, "loss": 0.3165, "step": 13148 }, { "epoch": 0.909021776702385, "grad_norm": 0.6424747588285189, "learning_rate": 5.973873174218165e-06, "loss": 0.2794, "step": 13149 }, { "epoch": 0.9090909090909091, "grad_norm": 0.7276201686037091, "learning_rate": 5.973324129391611e-06, "loss": 0.3805, "step": 13150 }, { "epoch": 0.9091600414794331, "grad_norm": 0.6529670700736129, "learning_rate": 5.9727750723661774e-06, "loss": 0.3918, "step": 13151 }, { "epoch": 0.9092291738679571, "grad_norm": 0.7215663776816623, "learning_rate": 5.972226003148742e-06, "loss": 0.2996, "step": 13152 }, { "epoch": 0.9092983062564811, "grad_norm": 0.5777748766638345, "learning_rate": 5.971676921746189e-06, "loss": 0.2533, "step": 13153 }, { "epoch": 0.9093674386450051, "grad_norm": 0.6250828001113877, "learning_rate": 5.971127828165401e-06, "loss": 0.3713, "step": 13154 }, { "epoch": 0.9094365710335293, "grad_norm": 0.7910923844127212, "learning_rate": 5.970578722413259e-06, "loss": 0.3322, "step": 13155 }, { "epoch": 0.9095057034220533, "grad_norm": 1.0949731437661543, "learning_rate": 5.970029604496642e-06, "loss": 0.2843, "step": 13156 }, { "epoch": 0.9095748358105773, "grad_norm": 0.6951799514272978, "learning_rate": 5.969480474422438e-06, "loss": 0.3106, "step": 13157 }, { "epoch": 0.9096439681991013, "grad_norm": 1.9738536412319634, "learning_rate": 5.968931332197525e-06, "loss": 0.3699, "step": 13158 }, { "epoch": 0.9097131005876253, "grad_norm": 0.6786023732363775, "learning_rate": 5.9683821778287865e-06, "loss": 0.3085, "step": 13159 }, { "epoch": 0.9097822329761494, "grad_norm": 0.6675406239007877, "learning_rate": 5.967833011323105e-06, "loss": 0.2811, "step": 13160 }, { "epoch": 0.9098513653646734, "grad_norm": 0.7413579876996962, "learning_rate": 5.967283832687366e-06, "loss": 0.387, "step": 13161 }, { "epoch": 0.9099204977531974, "grad_norm": 0.6792104192658328, "learning_rate": 5.96673464192845e-06, "loss": 0.3415, "step": 13162 }, { "epoch": 0.9099896301417214, "grad_norm": 0.6345945820901105, "learning_rate": 5.966185439053239e-06, "loss": 0.2947, "step": 13163 }, { "epoch": 0.9100587625302454, "grad_norm": 0.6245691877509261, "learning_rate": 5.965636224068618e-06, "loss": 0.27, "step": 13164 }, { "epoch": 0.9101278949187694, "grad_norm": 0.7381469609992157, "learning_rate": 5.965086996981472e-06, "loss": 0.3823, "step": 13165 }, { "epoch": 0.9101970273072935, "grad_norm": 0.7282454348771905, "learning_rate": 5.964537757798682e-06, "loss": 0.3375, "step": 13166 }, { "epoch": 0.9102661596958175, "grad_norm": 0.7600119046791951, "learning_rate": 5.963988506527131e-06, "loss": 0.3009, "step": 13167 }, { "epoch": 0.9103352920843415, "grad_norm": 0.6138781374314856, "learning_rate": 5.963439243173705e-06, "loss": 0.3611, "step": 13168 }, { "epoch": 0.9104044244728655, "grad_norm": 0.7299113319889846, "learning_rate": 5.962889967745289e-06, "loss": 0.4092, "step": 13169 }, { "epoch": 0.9104735568613895, "grad_norm": 0.63598996550793, "learning_rate": 5.962340680248764e-06, "loss": 0.2647, "step": 13170 }, { "epoch": 0.9105426892499136, "grad_norm": 0.6712398475940694, "learning_rate": 5.961791380691015e-06, "loss": 0.2222, "step": 13171 }, { "epoch": 0.9106118216384376, "grad_norm": 0.7731544137331972, "learning_rate": 5.96124206907893e-06, "loss": 0.41, "step": 13172 }, { "epoch": 0.9106809540269616, "grad_norm": 0.6923858020738964, "learning_rate": 5.960692745419391e-06, "loss": 0.3044, "step": 13173 }, { "epoch": 0.9107500864154856, "grad_norm": 0.6665663508962242, "learning_rate": 5.96014340971928e-06, "loss": 0.2956, "step": 13174 }, { "epoch": 0.9108192188040097, "grad_norm": 0.7251145117125162, "learning_rate": 5.959594061985487e-06, "loss": 0.2827, "step": 13175 }, { "epoch": 0.9108883511925338, "grad_norm": 0.6133729234915295, "learning_rate": 5.9590447022248955e-06, "loss": 0.3718, "step": 13176 }, { "epoch": 0.9109574835810578, "grad_norm": 0.7016619255356873, "learning_rate": 5.9584953304443895e-06, "loss": 0.289, "step": 13177 }, { "epoch": 0.9110266159695818, "grad_norm": 0.7919732249643504, "learning_rate": 5.957945946650854e-06, "loss": 0.3085, "step": 13178 }, { "epoch": 0.9110957483581058, "grad_norm": 0.6738858755100607, "learning_rate": 5.957396550851175e-06, "loss": 0.3808, "step": 13179 }, { "epoch": 0.9111648807466298, "grad_norm": 0.7374223269879323, "learning_rate": 5.9568471430522415e-06, "loss": 0.4598, "step": 13180 }, { "epoch": 0.9112340131351538, "grad_norm": 0.6040031939659333, "learning_rate": 5.956297723260934e-06, "loss": 0.2272, "step": 13181 }, { "epoch": 0.9113031455236779, "grad_norm": 0.6514470745031357, "learning_rate": 5.955748291484142e-06, "loss": 0.3165, "step": 13182 }, { "epoch": 0.9113722779122019, "grad_norm": 0.6909581039547132, "learning_rate": 5.955198847728752e-06, "loss": 0.3911, "step": 13183 }, { "epoch": 0.9114414103007259, "grad_norm": 0.6528832607137183, "learning_rate": 5.954649392001649e-06, "loss": 0.2539, "step": 13184 }, { "epoch": 0.9115105426892499, "grad_norm": 0.6218311042038264, "learning_rate": 5.954099924309717e-06, "loss": 0.2664, "step": 13185 }, { "epoch": 0.9115796750777739, "grad_norm": 0.6864742626292762, "learning_rate": 5.953550444659848e-06, "loss": 0.3556, "step": 13186 }, { "epoch": 0.911648807466298, "grad_norm": 1.7368169844965384, "learning_rate": 5.953000953058925e-06, "loss": 0.3644, "step": 13187 }, { "epoch": 0.911717939854822, "grad_norm": 0.6176884453707698, "learning_rate": 5.952451449513835e-06, "loss": 0.292, "step": 13188 }, { "epoch": 0.911787072243346, "grad_norm": 0.5859144189311968, "learning_rate": 5.951901934031466e-06, "loss": 0.2633, "step": 13189 }, { "epoch": 0.91185620463187, "grad_norm": 0.7542822649700873, "learning_rate": 5.951352406618706e-06, "loss": 0.3947, "step": 13190 }, { "epoch": 0.911925337020394, "grad_norm": 0.6867497281425492, "learning_rate": 5.95080286728244e-06, "loss": 0.3147, "step": 13191 }, { "epoch": 0.911994469408918, "grad_norm": 0.6305175814859898, "learning_rate": 5.950253316029557e-06, "loss": 0.3093, "step": 13192 }, { "epoch": 0.9120636017974421, "grad_norm": 0.6815346050382082, "learning_rate": 5.949703752866946e-06, "loss": 0.2952, "step": 13193 }, { "epoch": 0.9121327341859661, "grad_norm": 0.6852877685549953, "learning_rate": 5.949154177801491e-06, "loss": 0.411, "step": 13194 }, { "epoch": 0.9122018665744901, "grad_norm": 0.63776233295225, "learning_rate": 5.948604590840085e-06, "loss": 0.2768, "step": 13195 }, { "epoch": 0.9122709989630142, "grad_norm": 0.6878629941840689, "learning_rate": 5.94805499198961e-06, "loss": 0.2752, "step": 13196 }, { "epoch": 0.9123401313515382, "grad_norm": 0.6716153888268247, "learning_rate": 5.9475053812569595e-06, "loss": 0.3724, "step": 13197 }, { "epoch": 0.9124092637400623, "grad_norm": 0.6452732696275815, "learning_rate": 5.94695575864902e-06, "loss": 0.3612, "step": 13198 }, { "epoch": 0.9124783961285863, "grad_norm": 0.6206579922409601, "learning_rate": 5.946406124172679e-06, "loss": 0.2615, "step": 13199 }, { "epoch": 0.9125475285171103, "grad_norm": 0.7193039735272356, "learning_rate": 5.9458564778348286e-06, "loss": 0.3112, "step": 13200 }, { "epoch": 0.9126166609056343, "grad_norm": 0.7094855440633691, "learning_rate": 5.9453068196423524e-06, "loss": 0.4584, "step": 13201 }, { "epoch": 0.9126857932941583, "grad_norm": 0.6560868215933232, "learning_rate": 5.9447571496021436e-06, "loss": 0.314, "step": 13202 }, { "epoch": 0.9127549256826823, "grad_norm": 0.8631024254315186, "learning_rate": 5.944207467721089e-06, "loss": 0.2641, "step": 13203 }, { "epoch": 0.9128240580712064, "grad_norm": 0.6903528620589404, "learning_rate": 5.94365777400608e-06, "loss": 0.3932, "step": 13204 }, { "epoch": 0.9128931904597304, "grad_norm": 0.738501066312967, "learning_rate": 5.9431080684640054e-06, "loss": 0.3629, "step": 13205 }, { "epoch": 0.9129623228482544, "grad_norm": 0.5624041048424322, "learning_rate": 5.942558351101752e-06, "loss": 0.2387, "step": 13206 }, { "epoch": 0.9130314552367784, "grad_norm": 0.6702875338085283, "learning_rate": 5.942008621926214e-06, "loss": 0.29, "step": 13207 }, { "epoch": 0.9131005876253024, "grad_norm": 0.7491624920560999, "learning_rate": 5.941458880944277e-06, "loss": 0.3534, "step": 13208 }, { "epoch": 0.9131697200138265, "grad_norm": 0.6875476384796497, "learning_rate": 5.940909128162834e-06, "loss": 0.2844, "step": 13209 }, { "epoch": 0.9132388524023505, "grad_norm": 0.6069034510696689, "learning_rate": 5.940359363588774e-06, "loss": 0.2501, "step": 13210 }, { "epoch": 0.9133079847908745, "grad_norm": 0.8131889862522285, "learning_rate": 5.939809587228988e-06, "loss": 0.3303, "step": 13211 }, { "epoch": 0.9133771171793985, "grad_norm": 0.6840989686000118, "learning_rate": 5.939259799090365e-06, "loss": 0.342, "step": 13212 }, { "epoch": 0.9134462495679225, "grad_norm": 0.6179265904767528, "learning_rate": 5.938709999179797e-06, "loss": 0.254, "step": 13213 }, { "epoch": 0.9135153819564465, "grad_norm": 0.66153677405409, "learning_rate": 5.938160187504175e-06, "loss": 0.2715, "step": 13214 }, { "epoch": 0.9135845143449706, "grad_norm": 0.7058691749547391, "learning_rate": 5.93761036407039e-06, "loss": 0.3982, "step": 13215 }, { "epoch": 0.9136536467334947, "grad_norm": 0.6823908477617724, "learning_rate": 5.93706052888533e-06, "loss": 0.3211, "step": 13216 }, { "epoch": 0.9137227791220187, "grad_norm": 0.6244079066731452, "learning_rate": 5.936510681955889e-06, "loss": 0.29, "step": 13217 }, { "epoch": 0.9137919115105427, "grad_norm": 0.6717105580328685, "learning_rate": 5.93596082328896e-06, "loss": 0.2881, "step": 13218 }, { "epoch": 0.9138610438990667, "grad_norm": 0.6474307876031236, "learning_rate": 5.935410952891431e-06, "loss": 0.4139, "step": 13219 }, { "epoch": 0.9139301762875908, "grad_norm": 0.7374289077720665, "learning_rate": 5.934861070770194e-06, "loss": 0.2674, "step": 13220 }, { "epoch": 0.9139993086761148, "grad_norm": 0.7293967910874668, "learning_rate": 5.934311176932143e-06, "loss": 0.3468, "step": 13221 }, { "epoch": 0.9140684410646388, "grad_norm": 0.6484157058033431, "learning_rate": 5.933761271384169e-06, "loss": 0.343, "step": 13222 }, { "epoch": 0.9141375734531628, "grad_norm": 0.66077159927152, "learning_rate": 5.9332113541331634e-06, "loss": 0.3812, "step": 13223 }, { "epoch": 0.9142067058416868, "grad_norm": 0.7153654620855227, "learning_rate": 5.932661425186018e-06, "loss": 0.2335, "step": 13224 }, { "epoch": 0.9142758382302109, "grad_norm": 0.6762040459329004, "learning_rate": 5.932111484549628e-06, "loss": 0.3241, "step": 13225 }, { "epoch": 0.9143449706187349, "grad_norm": 0.6394926794063848, "learning_rate": 5.9315615322308824e-06, "loss": 0.3464, "step": 13226 }, { "epoch": 0.9144141030072589, "grad_norm": 0.6731710940615193, "learning_rate": 5.9310115682366755e-06, "loss": 0.276, "step": 13227 }, { "epoch": 0.9144832353957829, "grad_norm": 0.6696505988832746, "learning_rate": 5.930461592573901e-06, "loss": 0.3279, "step": 13228 }, { "epoch": 0.9145523677843069, "grad_norm": 0.7177882725053778, "learning_rate": 5.92991160524945e-06, "loss": 0.3665, "step": 13229 }, { "epoch": 0.914621500172831, "grad_norm": 0.8772434319326954, "learning_rate": 5.929361606270217e-06, "loss": 0.3473, "step": 13230 }, { "epoch": 0.914690632561355, "grad_norm": 1.1037750993138593, "learning_rate": 5.928811595643093e-06, "loss": 0.2543, "step": 13231 }, { "epoch": 0.914759764949879, "grad_norm": 0.6160843349147678, "learning_rate": 5.928261573374976e-06, "loss": 0.3071, "step": 13232 }, { "epoch": 0.914828897338403, "grad_norm": 0.8315869797057769, "learning_rate": 5.927711539472756e-06, "loss": 0.4678, "step": 13233 }, { "epoch": 0.914898029726927, "grad_norm": 0.6002362829538811, "learning_rate": 5.927161493943327e-06, "loss": 0.2609, "step": 13234 }, { "epoch": 0.914967162115451, "grad_norm": 0.6321498982725449, "learning_rate": 5.926611436793583e-06, "loss": 0.2725, "step": 13235 }, { "epoch": 0.9150362945039752, "grad_norm": 0.6588555347965847, "learning_rate": 5.926061368030419e-06, "loss": 0.2818, "step": 13236 }, { "epoch": 0.9151054268924992, "grad_norm": 0.713544577669128, "learning_rate": 5.925511287660728e-06, "loss": 0.3795, "step": 13237 }, { "epoch": 0.9151745592810232, "grad_norm": 0.8853252715588129, "learning_rate": 5.924961195691403e-06, "loss": 0.2336, "step": 13238 }, { "epoch": 0.9152436916695472, "grad_norm": 0.5927496191283274, "learning_rate": 5.924411092129343e-06, "loss": 0.2345, "step": 13239 }, { "epoch": 0.9153128240580712, "grad_norm": 0.6459567651135926, "learning_rate": 5.9238609769814385e-06, "loss": 0.336, "step": 13240 }, { "epoch": 0.9153819564465953, "grad_norm": 0.7000385196852068, "learning_rate": 5.923310850254586e-06, "loss": 0.2906, "step": 13241 }, { "epoch": 0.9154510888351193, "grad_norm": 0.6603992152844934, "learning_rate": 5.922760711955677e-06, "loss": 0.3609, "step": 13242 }, { "epoch": 0.9155202212236433, "grad_norm": 0.6582779548198443, "learning_rate": 5.9222105620916115e-06, "loss": 0.3127, "step": 13243 }, { "epoch": 0.9155893536121673, "grad_norm": 0.6582617628334557, "learning_rate": 5.921660400669282e-06, "loss": 0.3731, "step": 13244 }, { "epoch": 0.9156584860006913, "grad_norm": 0.6238727379157154, "learning_rate": 5.921110227695583e-06, "loss": 0.3131, "step": 13245 }, { "epoch": 0.9157276183892153, "grad_norm": 0.7859320610022387, "learning_rate": 5.920560043177413e-06, "loss": 0.2724, "step": 13246 }, { "epoch": 0.9157967507777394, "grad_norm": 0.6781790806714677, "learning_rate": 5.920009847121665e-06, "loss": 0.4037, "step": 13247 }, { "epoch": 0.9158658831662634, "grad_norm": 0.636938943150357, "learning_rate": 5.9194596395352355e-06, "loss": 0.2957, "step": 13248 }, { "epoch": 0.9159350155547874, "grad_norm": 0.630854217052723, "learning_rate": 5.91890942042502e-06, "loss": 0.286, "step": 13249 }, { "epoch": 0.9160041479433114, "grad_norm": 0.7493937480366645, "learning_rate": 5.918359189797915e-06, "loss": 0.2912, "step": 13250 }, { "epoch": 0.9160732803318354, "grad_norm": 0.7127483410431857, "learning_rate": 5.917808947660817e-06, "loss": 0.4028, "step": 13251 }, { "epoch": 0.9161424127203595, "grad_norm": 0.7272072905949357, "learning_rate": 5.91725869402062e-06, "loss": 0.3271, "step": 13252 }, { "epoch": 0.9162115451088835, "grad_norm": 0.6734908251774916, "learning_rate": 5.916708428884224e-06, "loss": 0.2941, "step": 13253 }, { "epoch": 0.9162806774974075, "grad_norm": 0.6309988198335463, "learning_rate": 5.916158152258522e-06, "loss": 0.3934, "step": 13254 }, { "epoch": 0.9163498098859315, "grad_norm": 0.6708642378632369, "learning_rate": 5.915607864150415e-06, "loss": 0.3287, "step": 13255 }, { "epoch": 0.9164189422744555, "grad_norm": 0.7854761663986312, "learning_rate": 5.915057564566794e-06, "loss": 0.2526, "step": 13256 }, { "epoch": 0.9164880746629797, "grad_norm": 0.7272624096136331, "learning_rate": 5.914507253514562e-06, "loss": 0.347, "step": 13257 }, { "epoch": 0.9165572070515037, "grad_norm": 0.6366593914869955, "learning_rate": 5.913956931000613e-06, "loss": 0.4102, "step": 13258 }, { "epoch": 0.9166263394400277, "grad_norm": 0.6461037563701145, "learning_rate": 5.913406597031845e-06, "loss": 0.2693, "step": 13259 }, { "epoch": 0.9166954718285517, "grad_norm": 0.6275518351989764, "learning_rate": 5.912856251615155e-06, "loss": 0.2669, "step": 13260 }, { "epoch": 0.9167646042170757, "grad_norm": 0.772992044290385, "learning_rate": 5.912305894757441e-06, "loss": 0.3675, "step": 13261 }, { "epoch": 0.9168337366055997, "grad_norm": 0.7961794191928218, "learning_rate": 5.911755526465601e-06, "loss": 0.3371, "step": 13262 }, { "epoch": 0.9169028689941238, "grad_norm": 0.6287807520626042, "learning_rate": 5.911205146746532e-06, "loss": 0.3058, "step": 13263 }, { "epoch": 0.9169720013826478, "grad_norm": 0.7311136594080161, "learning_rate": 5.910654755607132e-06, "loss": 0.3399, "step": 13264 }, { "epoch": 0.9170411337711718, "grad_norm": 0.6381780278721679, "learning_rate": 5.910104353054302e-06, "loss": 0.3822, "step": 13265 }, { "epoch": 0.9171102661596958, "grad_norm": 0.9111143947799979, "learning_rate": 5.909553939094937e-06, "loss": 0.3262, "step": 13266 }, { "epoch": 0.9171793985482198, "grad_norm": 0.6820381484944685, "learning_rate": 5.909003513735936e-06, "loss": 0.3368, "step": 13267 }, { "epoch": 0.9172485309367439, "grad_norm": 1.0896709954285437, "learning_rate": 5.908453076984199e-06, "loss": 0.312, "step": 13268 }, { "epoch": 0.9173176633252679, "grad_norm": 0.6430979434186571, "learning_rate": 5.907902628846623e-06, "loss": 0.3906, "step": 13269 }, { "epoch": 0.9173867957137919, "grad_norm": 0.701508013421759, "learning_rate": 5.907352169330107e-06, "loss": 0.2807, "step": 13270 }, { "epoch": 0.9174559281023159, "grad_norm": 0.6546424095900483, "learning_rate": 5.906801698441554e-06, "loss": 0.2549, "step": 13271 }, { "epoch": 0.9175250604908399, "grad_norm": 0.649627546841409, "learning_rate": 5.906251216187858e-06, "loss": 0.3782, "step": 13272 }, { "epoch": 0.917594192879364, "grad_norm": 0.6507574757767117, "learning_rate": 5.90570072257592e-06, "loss": 0.3404, "step": 13273 }, { "epoch": 0.917663325267888, "grad_norm": 0.7251734820511946, "learning_rate": 5.905150217612641e-06, "loss": 0.3466, "step": 13274 }, { "epoch": 0.917732457656412, "grad_norm": 0.7358041327529169, "learning_rate": 5.904599701304918e-06, "loss": 0.3003, "step": 13275 }, { "epoch": 0.917801590044936, "grad_norm": 0.6368003539756975, "learning_rate": 5.9040491736596524e-06, "loss": 0.3625, "step": 13276 }, { "epoch": 0.9178707224334601, "grad_norm": 0.6347112925554127, "learning_rate": 5.903498634683743e-06, "loss": 0.2824, "step": 13277 }, { "epoch": 0.9179398548219841, "grad_norm": 0.8338763675596231, "learning_rate": 5.9029480843840915e-06, "loss": 0.3999, "step": 13278 }, { "epoch": 0.9180089872105082, "grad_norm": 0.7206520541041284, "learning_rate": 5.902397522767598e-06, "loss": 0.3657, "step": 13279 }, { "epoch": 0.9180781195990322, "grad_norm": 0.7737707550550867, "learning_rate": 5.90184694984116e-06, "loss": 0.3774, "step": 13280 }, { "epoch": 0.9181472519875562, "grad_norm": 0.5878126688649985, "learning_rate": 5.901296365611681e-06, "loss": 0.2748, "step": 13281 }, { "epoch": 0.9182163843760802, "grad_norm": 0.6530781689118024, "learning_rate": 5.90074577008606e-06, "loss": 0.2466, "step": 13282 }, { "epoch": 0.9182855167646042, "grad_norm": 0.6444806692059242, "learning_rate": 5.9001951632712e-06, "loss": 0.4039, "step": 13283 }, { "epoch": 0.9183546491531283, "grad_norm": 0.748468739734452, "learning_rate": 5.899644545173997e-06, "loss": 0.2893, "step": 13284 }, { "epoch": 0.9184237815416523, "grad_norm": 0.6065968445996948, "learning_rate": 5.899093915801357e-06, "loss": 0.3251, "step": 13285 }, { "epoch": 0.9184929139301763, "grad_norm": 0.6469238335920812, "learning_rate": 5.89854327516018e-06, "loss": 0.3474, "step": 13286 }, { "epoch": 0.9185620463187003, "grad_norm": 0.7342655710783564, "learning_rate": 5.897992623257365e-06, "loss": 0.3329, "step": 13287 }, { "epoch": 0.9186311787072243, "grad_norm": 0.6096226635906334, "learning_rate": 5.8974419600998154e-06, "loss": 0.2913, "step": 13288 }, { "epoch": 0.9187003110957483, "grad_norm": 0.69676150834031, "learning_rate": 5.896891285694433e-06, "loss": 0.2789, "step": 13289 }, { "epoch": 0.9187694434842724, "grad_norm": 0.6505934173821885, "learning_rate": 5.896340600048118e-06, "loss": 0.3933, "step": 13290 }, { "epoch": 0.9188385758727964, "grad_norm": 0.7397040758457621, "learning_rate": 5.895789903167773e-06, "loss": 0.3783, "step": 13291 }, { "epoch": 0.9189077082613204, "grad_norm": 0.6820780140739433, "learning_rate": 5.895239195060302e-06, "loss": 0.2603, "step": 13292 }, { "epoch": 0.9189768406498444, "grad_norm": 0.663156084261359, "learning_rate": 5.894688475732604e-06, "loss": 0.3183, "step": 13293 }, { "epoch": 0.9190459730383684, "grad_norm": 0.9106434808677004, "learning_rate": 5.894137745191584e-06, "loss": 0.3349, "step": 13294 }, { "epoch": 0.9191151054268925, "grad_norm": 0.6803176068241238, "learning_rate": 5.89358700344414e-06, "loss": 0.3094, "step": 13295 }, { "epoch": 0.9191842378154165, "grad_norm": 0.8061609141932328, "learning_rate": 5.89303625049718e-06, "loss": 0.3434, "step": 13296 }, { "epoch": 0.9192533702039406, "grad_norm": 0.7639288641964507, "learning_rate": 5.892485486357605e-06, "loss": 0.4014, "step": 13297 }, { "epoch": 0.9193225025924646, "grad_norm": 0.6726631325791429, "learning_rate": 5.891934711032314e-06, "loss": 0.3299, "step": 13298 }, { "epoch": 0.9193916349809886, "grad_norm": 0.7407327527676382, "learning_rate": 5.891383924528215e-06, "loss": 0.2606, "step": 13299 }, { "epoch": 0.9194607673695127, "grad_norm": 0.6746957563190316, "learning_rate": 5.890833126852209e-06, "loss": 0.3454, "step": 13300 }, { "epoch": 0.9195298997580367, "grad_norm": 0.6367959815413048, "learning_rate": 5.8902823180112e-06, "loss": 0.4015, "step": 13301 }, { "epoch": 0.9195990321465607, "grad_norm": 0.6976583764010331, "learning_rate": 5.88973149801209e-06, "loss": 0.2688, "step": 13302 }, { "epoch": 0.9196681645350847, "grad_norm": 0.6690399131293248, "learning_rate": 5.889180666861784e-06, "loss": 0.2734, "step": 13303 }, { "epoch": 0.9197372969236087, "grad_norm": 0.6503893573961999, "learning_rate": 5.888629824567187e-06, "loss": 0.395, "step": 13304 }, { "epoch": 0.9198064293121327, "grad_norm": 0.749903852642582, "learning_rate": 5.888078971135198e-06, "loss": 0.3221, "step": 13305 }, { "epoch": 0.9198755617006568, "grad_norm": 0.6123872584819563, "learning_rate": 5.887528106572726e-06, "loss": 0.2867, "step": 13306 }, { "epoch": 0.9199446940891808, "grad_norm": 0.6706228024515829, "learning_rate": 5.886977230886673e-06, "loss": 0.269, "step": 13307 }, { "epoch": 0.9200138264777048, "grad_norm": 0.6758327234456684, "learning_rate": 5.886426344083942e-06, "loss": 0.3521, "step": 13308 }, { "epoch": 0.9200829588662288, "grad_norm": 0.7352668676527184, "learning_rate": 5.8858754461714385e-06, "loss": 0.2774, "step": 13309 }, { "epoch": 0.9201520912547528, "grad_norm": 0.8347224225111902, "learning_rate": 5.885324537156067e-06, "loss": 0.2895, "step": 13310 }, { "epoch": 0.9202212236432769, "grad_norm": 0.6769667412876278, "learning_rate": 5.884773617044735e-06, "loss": 0.3692, "step": 13311 }, { "epoch": 0.9202903560318009, "grad_norm": 0.7061061057698558, "learning_rate": 5.884222685844343e-06, "loss": 0.4186, "step": 13312 }, { "epoch": 0.9203594884203249, "grad_norm": 0.6804137155285379, "learning_rate": 5.883671743561796e-06, "loss": 0.2509, "step": 13313 }, { "epoch": 0.9204286208088489, "grad_norm": 0.6551542544403667, "learning_rate": 5.8831207902040035e-06, "loss": 0.3025, "step": 13314 }, { "epoch": 0.9204977531973729, "grad_norm": 0.6772810886036392, "learning_rate": 5.882569825777866e-06, "loss": 0.4204, "step": 13315 }, { "epoch": 0.9205668855858969, "grad_norm": 0.6688712652985586, "learning_rate": 5.882018850290291e-06, "loss": 0.31, "step": 13316 }, { "epoch": 0.920636017974421, "grad_norm": 0.7615659931069545, "learning_rate": 5.8814678637481846e-06, "loss": 0.3306, "step": 13317 }, { "epoch": 0.9207051503629451, "grad_norm": 0.6550710168535603, "learning_rate": 5.88091686615845e-06, "loss": 0.2844, "step": 13318 }, { "epoch": 0.9207742827514691, "grad_norm": 0.7391715543081684, "learning_rate": 5.880365857527997e-06, "loss": 0.3986, "step": 13319 }, { "epoch": 0.9208434151399931, "grad_norm": 0.6734485440441143, "learning_rate": 5.879814837863727e-06, "loss": 0.3254, "step": 13320 }, { "epoch": 0.9209125475285171, "grad_norm": 0.637323646303904, "learning_rate": 5.8792638071725485e-06, "loss": 0.3133, "step": 13321 }, { "epoch": 0.9209816799170412, "grad_norm": 0.7222409109709261, "learning_rate": 5.8787127654613684e-06, "loss": 0.4082, "step": 13322 }, { "epoch": 0.9210508123055652, "grad_norm": 0.6715922213712247, "learning_rate": 5.8781617127370905e-06, "loss": 0.3539, "step": 13323 }, { "epoch": 0.9211199446940892, "grad_norm": 0.6505673432300806, "learning_rate": 5.877610649006623e-06, "loss": 0.2747, "step": 13324 }, { "epoch": 0.9211890770826132, "grad_norm": 0.7551762880680272, "learning_rate": 5.877059574276873e-06, "loss": 0.3305, "step": 13325 }, { "epoch": 0.9212582094711372, "grad_norm": 0.5921071312627465, "learning_rate": 5.876508488554746e-06, "loss": 0.3682, "step": 13326 }, { "epoch": 0.9213273418596613, "grad_norm": 0.7046485149034872, "learning_rate": 5.875957391847149e-06, "loss": 0.3563, "step": 13327 }, { "epoch": 0.9213964742481853, "grad_norm": 0.6639248670478662, "learning_rate": 5.87540628416099e-06, "loss": 0.2805, "step": 13328 }, { "epoch": 0.9214656066367093, "grad_norm": 1.170809183998853, "learning_rate": 5.874855165503175e-06, "loss": 0.3563, "step": 13329 }, { "epoch": 0.9215347390252333, "grad_norm": 0.602757407001042, "learning_rate": 5.874304035880611e-06, "loss": 0.3612, "step": 13330 }, { "epoch": 0.9216038714137573, "grad_norm": 0.6441622604348786, "learning_rate": 5.873752895300207e-06, "loss": 0.3333, "step": 13331 }, { "epoch": 0.9216730038022813, "grad_norm": 0.5900455799618589, "learning_rate": 5.87320174376887e-06, "loss": 0.277, "step": 13332 }, { "epoch": 0.9217421361908054, "grad_norm": 0.7129589028317386, "learning_rate": 5.872650581293506e-06, "loss": 0.3371, "step": 13333 }, { "epoch": 0.9218112685793294, "grad_norm": 0.640159676329665, "learning_rate": 5.872099407881025e-06, "loss": 0.2817, "step": 13334 }, { "epoch": 0.9218804009678534, "grad_norm": 0.6700400979093097, "learning_rate": 5.871548223538335e-06, "loss": 0.2825, "step": 13335 }, { "epoch": 0.9219495333563774, "grad_norm": 0.6824986481429741, "learning_rate": 5.870997028272344e-06, "loss": 0.3965, "step": 13336 }, { "epoch": 0.9220186657449014, "grad_norm": 0.6773634419396933, "learning_rate": 5.8704458220899584e-06, "loss": 0.3498, "step": 13337 }, { "epoch": 0.9220877981334256, "grad_norm": 0.7249400260368118, "learning_rate": 5.8698946049980875e-06, "loss": 0.2872, "step": 13338 }, { "epoch": 0.9221569305219496, "grad_norm": 0.6452352669904444, "learning_rate": 5.869343377003641e-06, "loss": 0.312, "step": 13339 }, { "epoch": 0.9222260629104736, "grad_norm": 0.6686524444928632, "learning_rate": 5.8687921381135275e-06, "loss": 0.3532, "step": 13340 }, { "epoch": 0.9222951952989976, "grad_norm": 0.6445899056178415, "learning_rate": 5.8682408883346535e-06, "loss": 0.2676, "step": 13341 }, { "epoch": 0.9223643276875216, "grad_norm": 0.6967596078973654, "learning_rate": 5.867689627673929e-06, "loss": 0.3012, "step": 13342 }, { "epoch": 0.9224334600760457, "grad_norm": 0.6821652247252018, "learning_rate": 5.8671383561382645e-06, "loss": 0.3618, "step": 13343 }, { "epoch": 0.9225025924645697, "grad_norm": 0.8577367631691497, "learning_rate": 5.866587073734566e-06, "loss": 0.3898, "step": 13344 }, { "epoch": 0.9225717248530937, "grad_norm": 0.5926359058394182, "learning_rate": 5.8660357804697476e-06, "loss": 0.2619, "step": 13345 }, { "epoch": 0.9226408572416177, "grad_norm": 0.6464783767548992, "learning_rate": 5.865484476350716e-06, "loss": 0.254, "step": 13346 }, { "epoch": 0.9227099896301417, "grad_norm": 0.715093321158957, "learning_rate": 5.864933161384381e-06, "loss": 0.4055, "step": 13347 }, { "epoch": 0.9227791220186657, "grad_norm": 0.656009799468764, "learning_rate": 5.864381835577651e-06, "loss": 0.3316, "step": 13348 }, { "epoch": 0.9228482544071898, "grad_norm": 0.5998301095204743, "learning_rate": 5.8638304989374384e-06, "loss": 0.287, "step": 13349 }, { "epoch": 0.9229173867957138, "grad_norm": 0.7083552380248437, "learning_rate": 5.863279151470652e-06, "loss": 0.2374, "step": 13350 }, { "epoch": 0.9229865191842378, "grad_norm": 0.6359161480298344, "learning_rate": 5.862727793184201e-06, "loss": 0.3817, "step": 13351 }, { "epoch": 0.9230556515727618, "grad_norm": 0.6951157241678071, "learning_rate": 5.862176424084998e-06, "loss": 0.2958, "step": 13352 }, { "epoch": 0.9231247839612858, "grad_norm": 0.6837332772401515, "learning_rate": 5.861625044179952e-06, "loss": 0.2507, "step": 13353 }, { "epoch": 0.9231939163498099, "grad_norm": 0.6723148698925305, "learning_rate": 5.861073653475974e-06, "loss": 0.3676, "step": 13354 }, { "epoch": 0.9232630487383339, "grad_norm": 0.7036904008456868, "learning_rate": 5.860522251979972e-06, "loss": 0.374, "step": 13355 }, { "epoch": 0.9233321811268579, "grad_norm": 0.7436497634401986, "learning_rate": 5.859970839698862e-06, "loss": 0.2382, "step": 13356 }, { "epoch": 0.9234013135153819, "grad_norm": 0.6222160195626006, "learning_rate": 5.859419416639552e-06, "loss": 0.2637, "step": 13357 }, { "epoch": 0.923470445903906, "grad_norm": 0.7200872510734467, "learning_rate": 5.858867982808952e-06, "loss": 0.3793, "step": 13358 }, { "epoch": 0.92353957829243, "grad_norm": 0.6487849816446274, "learning_rate": 5.858316538213974e-06, "loss": 0.3569, "step": 13359 }, { "epoch": 0.9236087106809541, "grad_norm": 0.6701027377422016, "learning_rate": 5.857765082861532e-06, "loss": 0.2194, "step": 13360 }, { "epoch": 0.9236778430694781, "grad_norm": 0.6642850936965263, "learning_rate": 5.857213616758535e-06, "loss": 0.3451, "step": 13361 }, { "epoch": 0.9237469754580021, "grad_norm": 0.6757815502994647, "learning_rate": 5.856662139911893e-06, "loss": 0.3925, "step": 13362 }, { "epoch": 0.9238161078465261, "grad_norm": 1.0231834259564663, "learning_rate": 5.8561106523285215e-06, "loss": 0.3029, "step": 13363 }, { "epoch": 0.9238852402350501, "grad_norm": 0.7020587322583598, "learning_rate": 5.8555591540153315e-06, "loss": 0.2942, "step": 13364 }, { "epoch": 0.9239543726235742, "grad_norm": 0.6625921941413173, "learning_rate": 5.855007644979234e-06, "loss": 0.4174, "step": 13365 }, { "epoch": 0.9240235050120982, "grad_norm": 0.6995466983833474, "learning_rate": 5.854456125227139e-06, "loss": 0.2979, "step": 13366 }, { "epoch": 0.9240926374006222, "grad_norm": 0.5655873111603408, "learning_rate": 5.853904594765965e-06, "loss": 0.3097, "step": 13367 }, { "epoch": 0.9241617697891462, "grad_norm": 0.8608184622539174, "learning_rate": 5.853353053602619e-06, "loss": 0.3126, "step": 13368 }, { "epoch": 0.9242309021776702, "grad_norm": 0.6962836371079622, "learning_rate": 5.852801501744015e-06, "loss": 0.4166, "step": 13369 }, { "epoch": 0.9243000345661943, "grad_norm": 0.7133420177170573, "learning_rate": 5.852249939197065e-06, "loss": 0.3152, "step": 13370 }, { "epoch": 0.9243691669547183, "grad_norm": 0.6587405244084384, "learning_rate": 5.851698365968686e-06, "loss": 0.2883, "step": 13371 }, { "epoch": 0.9244382993432423, "grad_norm": 0.6553678437800561, "learning_rate": 5.851146782065786e-06, "loss": 0.3797, "step": 13372 }, { "epoch": 0.9245074317317663, "grad_norm": 0.6918447223958418, "learning_rate": 5.8505951874952804e-06, "loss": 0.2622, "step": 13373 }, { "epoch": 0.9245765641202903, "grad_norm": 0.6288890227783523, "learning_rate": 5.850043582264082e-06, "loss": 0.2412, "step": 13374 }, { "epoch": 0.9246456965088143, "grad_norm": 0.8011401491177419, "learning_rate": 5.8494919663791035e-06, "loss": 0.3733, "step": 13375 }, { "epoch": 0.9247148288973384, "grad_norm": 0.6778671177113162, "learning_rate": 5.84894033984726e-06, "loss": 0.3672, "step": 13376 }, { "epoch": 0.9247839612858624, "grad_norm": 0.590549250788257, "learning_rate": 5.848388702675465e-06, "loss": 0.2292, "step": 13377 }, { "epoch": 0.9248530936743864, "grad_norm": 0.7165559735049647, "learning_rate": 5.8478370548706305e-06, "loss": 0.2598, "step": 13378 }, { "epoch": 0.9249222260629105, "grad_norm": 0.6400589066151995, "learning_rate": 5.847285396439673e-06, "loss": 0.3318, "step": 13379 }, { "epoch": 0.9249913584514345, "grad_norm": 0.6746408542126687, "learning_rate": 5.846733727389503e-06, "loss": 0.3335, "step": 13380 }, { "epoch": 0.9250604908399586, "grad_norm": 0.6410033055851083, "learning_rate": 5.846182047727038e-06, "loss": 0.3163, "step": 13381 }, { "epoch": 0.9251296232284826, "grad_norm": 0.6545426876178961, "learning_rate": 5.845630357459191e-06, "loss": 0.2826, "step": 13382 }, { "epoch": 0.9251987556170066, "grad_norm": 0.6494418993813086, "learning_rate": 5.845078656592877e-06, "loss": 0.4276, "step": 13383 }, { "epoch": 0.9252678880055306, "grad_norm": 0.7003956023917408, "learning_rate": 5.84452694513501e-06, "loss": 0.3001, "step": 13384 }, { "epoch": 0.9253370203940546, "grad_norm": 1.0397311957461413, "learning_rate": 5.843975223092504e-06, "loss": 0.2861, "step": 13385 }, { "epoch": 0.9254061527825787, "grad_norm": 0.6742195222128246, "learning_rate": 5.843423490472275e-06, "loss": 0.4248, "step": 13386 }, { "epoch": 0.9254752851711027, "grad_norm": 2.588513673035932, "learning_rate": 5.842871747281238e-06, "loss": 0.3656, "step": 13387 }, { "epoch": 0.9255444175596267, "grad_norm": 0.60881504324836, "learning_rate": 5.842319993526307e-06, "loss": 0.2422, "step": 13388 }, { "epoch": 0.9256135499481507, "grad_norm": 0.6340912883447077, "learning_rate": 5.8417682292144e-06, "loss": 0.2952, "step": 13389 }, { "epoch": 0.9256826823366747, "grad_norm": 0.6943707738593675, "learning_rate": 5.84121645435243e-06, "loss": 0.3414, "step": 13390 }, { "epoch": 0.9257518147251987, "grad_norm": 0.6672064072052526, "learning_rate": 5.8406646689473115e-06, "loss": 0.3112, "step": 13391 }, { "epoch": 0.9258209471137228, "grad_norm": 0.6184548029365396, "learning_rate": 5.840112873005963e-06, "loss": 0.3129, "step": 13392 }, { "epoch": 0.9258900795022468, "grad_norm": 0.7245082169389989, "learning_rate": 5.839561066535298e-06, "loss": 0.2957, "step": 13393 }, { "epoch": 0.9259592118907708, "grad_norm": 0.6539055021918677, "learning_rate": 5.8390092495422335e-06, "loss": 0.3761, "step": 13394 }, { "epoch": 0.9260283442792948, "grad_norm": 0.6874370341816087, "learning_rate": 5.838457422033686e-06, "loss": 0.2949, "step": 13395 }, { "epoch": 0.9260974766678188, "grad_norm": 0.6250346895481587, "learning_rate": 5.83790558401657e-06, "loss": 0.2838, "step": 13396 }, { "epoch": 0.9261666090563428, "grad_norm": 0.6834936476887205, "learning_rate": 5.837353735497804e-06, "loss": 0.3982, "step": 13397 }, { "epoch": 0.9262357414448669, "grad_norm": 0.7129476979850893, "learning_rate": 5.8368018764843015e-06, "loss": 0.3249, "step": 13398 }, { "epoch": 0.926304873833391, "grad_norm": 0.7450340656840317, "learning_rate": 5.836250006982982e-06, "loss": 0.2738, "step": 13399 }, { "epoch": 0.926374006221915, "grad_norm": 0.8878379865579717, "learning_rate": 5.835698127000761e-06, "loss": 0.401, "step": 13400 }, { "epoch": 0.926443138610439, "grad_norm": 0.7150873385397287, "learning_rate": 5.835146236544556e-06, "loss": 0.3883, "step": 13401 }, { "epoch": 0.926512270998963, "grad_norm": 0.722182288413954, "learning_rate": 5.834594335621281e-06, "loss": 0.277, "step": 13402 }, { "epoch": 0.9265814033874871, "grad_norm": 0.6801224661170266, "learning_rate": 5.834042424237857e-06, "loss": 0.2321, "step": 13403 }, { "epoch": 0.9266505357760111, "grad_norm": 0.6710421432110621, "learning_rate": 5.8334905024012e-06, "loss": 0.3612, "step": 13404 }, { "epoch": 0.9267196681645351, "grad_norm": 0.6530257980780618, "learning_rate": 5.832938570118228e-06, "loss": 0.3479, "step": 13405 }, { "epoch": 0.9267888005530591, "grad_norm": 0.97878152394445, "learning_rate": 5.832386627395855e-06, "loss": 0.271, "step": 13406 }, { "epoch": 0.9268579329415831, "grad_norm": 0.7536456329236209, "learning_rate": 5.831834674241003e-06, "loss": 0.3097, "step": 13407 }, { "epoch": 0.9269270653301072, "grad_norm": 0.7360647187565502, "learning_rate": 5.8312827106605855e-06, "loss": 0.3869, "step": 13408 }, { "epoch": 0.9269961977186312, "grad_norm": 0.677796959437277, "learning_rate": 5.830730736661524e-06, "loss": 0.3324, "step": 13409 }, { "epoch": 0.9270653301071552, "grad_norm": 0.5993144391781786, "learning_rate": 5.830178752250736e-06, "loss": 0.2442, "step": 13410 }, { "epoch": 0.9271344624956792, "grad_norm": 0.7045476703090036, "learning_rate": 5.829626757435138e-06, "loss": 0.3434, "step": 13411 }, { "epoch": 0.9272035948842032, "grad_norm": 0.6987221139144008, "learning_rate": 5.82907475222165e-06, "loss": 0.4333, "step": 13412 }, { "epoch": 0.9272727272727272, "grad_norm": 0.6092309739599286, "learning_rate": 5.8285227366171885e-06, "loss": 0.3009, "step": 13413 }, { "epoch": 0.9273418596612513, "grad_norm": 0.7056583120544693, "learning_rate": 5.8279707106286736e-06, "loss": 0.3151, "step": 13414 }, { "epoch": 0.9274109920497753, "grad_norm": 0.6572592123068433, "learning_rate": 5.8274186742630224e-06, "loss": 0.4017, "step": 13415 }, { "epoch": 0.9274801244382993, "grad_norm": 0.631733981795605, "learning_rate": 5.8268666275271554e-06, "loss": 0.2927, "step": 13416 }, { "epoch": 0.9275492568268233, "grad_norm": 0.5761323291110338, "learning_rate": 5.826314570427991e-06, "loss": 0.2592, "step": 13417 }, { "epoch": 0.9276183892153473, "grad_norm": 0.6850684039786114, "learning_rate": 5.82576250297245e-06, "loss": 0.3322, "step": 13418 }, { "epoch": 0.9276875216038715, "grad_norm": 0.7185523571649417, "learning_rate": 5.8252104251674455e-06, "loss": 0.4292, "step": 13419 }, { "epoch": 0.9277566539923955, "grad_norm": 0.7207712632303434, "learning_rate": 5.824658337019904e-06, "loss": 0.287, "step": 13420 }, { "epoch": 0.9278257863809195, "grad_norm": 0.713521050764188, "learning_rate": 5.824106238536742e-06, "loss": 0.2967, "step": 13421 }, { "epoch": 0.9278949187694435, "grad_norm": 0.6686438600321014, "learning_rate": 5.823554129724877e-06, "loss": 0.4401, "step": 13422 }, { "epoch": 0.9279640511579675, "grad_norm": 0.7047999096273251, "learning_rate": 5.823002010591232e-06, "loss": 0.2987, "step": 13423 }, { "epoch": 0.9280331835464916, "grad_norm": 0.654549375452007, "learning_rate": 5.822449881142725e-06, "loss": 0.2367, "step": 13424 }, { "epoch": 0.9281023159350156, "grad_norm": 0.8262193778012408, "learning_rate": 5.821897741386278e-06, "loss": 0.3315, "step": 13425 }, { "epoch": 0.9281714483235396, "grad_norm": 0.6292387539296665, "learning_rate": 5.821345591328807e-06, "loss": 0.3453, "step": 13426 }, { "epoch": 0.9282405807120636, "grad_norm": 0.6376969039913625, "learning_rate": 5.820793430977237e-06, "loss": 0.3029, "step": 13427 }, { "epoch": 0.9283097131005876, "grad_norm": 0.7759516125215885, "learning_rate": 5.820241260338485e-06, "loss": 0.2637, "step": 13428 }, { "epoch": 0.9283788454891116, "grad_norm": 0.7018613138118001, "learning_rate": 5.819689079419473e-06, "loss": 0.3776, "step": 13429 }, { "epoch": 0.9284479778776357, "grad_norm": 0.6949644851981909, "learning_rate": 5.819136888227121e-06, "loss": 0.3132, "step": 13430 }, { "epoch": 0.9285171102661597, "grad_norm": 0.6482251648300963, "learning_rate": 5.81858468676835e-06, "loss": 0.2598, "step": 13431 }, { "epoch": 0.9285862426546837, "grad_norm": 0.7433559099679535, "learning_rate": 5.81803247505008e-06, "loss": 0.2904, "step": 13432 }, { "epoch": 0.9286553750432077, "grad_norm": 0.6879250822035713, "learning_rate": 5.817480253079233e-06, "loss": 0.4021, "step": 13433 }, { "epoch": 0.9287245074317317, "grad_norm": 0.6605354888528122, "learning_rate": 5.816928020862731e-06, "loss": 0.3109, "step": 13434 }, { "epoch": 0.9287936398202558, "grad_norm": 0.6386322158149058, "learning_rate": 5.816375778407494e-06, "loss": 0.2693, "step": 13435 }, { "epoch": 0.9288627722087798, "grad_norm": 0.7007807226473558, "learning_rate": 5.815823525720443e-06, "loss": 0.3306, "step": 13436 }, { "epoch": 0.9289319045973038, "grad_norm": 0.9386436263148591, "learning_rate": 5.815271262808498e-06, "loss": 0.3919, "step": 13437 }, { "epoch": 0.9290010369858278, "grad_norm": 0.6543420220001585, "learning_rate": 5.814718989678586e-06, "loss": 0.259, "step": 13438 }, { "epoch": 0.9290701693743518, "grad_norm": 0.6936697985112165, "learning_rate": 5.814166706337624e-06, "loss": 0.3352, "step": 13439 }, { "epoch": 0.929139301762876, "grad_norm": 0.6514147421026206, "learning_rate": 5.813614412792534e-06, "loss": 0.3658, "step": 13440 }, { "epoch": 0.9292084341514, "grad_norm": 0.6501099232625669, "learning_rate": 5.813062109050242e-06, "loss": 0.2976, "step": 13441 }, { "epoch": 0.929277566539924, "grad_norm": 0.6389592260001047, "learning_rate": 5.8125097951176655e-06, "loss": 0.3294, "step": 13442 }, { "epoch": 0.929346698928448, "grad_norm": 0.6563499691467718, "learning_rate": 5.8119574710017305e-06, "loss": 0.4294, "step": 13443 }, { "epoch": 0.929415831316972, "grad_norm": 0.710359973074032, "learning_rate": 5.811405136709355e-06, "loss": 0.4077, "step": 13444 }, { "epoch": 0.929484963705496, "grad_norm": 0.6268934433495509, "learning_rate": 5.810852792247465e-06, "loss": 0.2961, "step": 13445 }, { "epoch": 0.9295540960940201, "grad_norm": 0.711691046031895, "learning_rate": 5.810300437622984e-06, "loss": 0.3197, "step": 13446 }, { "epoch": 0.9296232284825441, "grad_norm": 0.7710272428498882, "learning_rate": 5.809748072842831e-06, "loss": 0.4209, "step": 13447 }, { "epoch": 0.9296923608710681, "grad_norm": 0.7214077829252531, "learning_rate": 5.809195697913932e-06, "loss": 0.3416, "step": 13448 }, { "epoch": 0.9297614932595921, "grad_norm": 0.7047885988923387, "learning_rate": 5.808643312843209e-06, "loss": 0.2593, "step": 13449 }, { "epoch": 0.9298306256481161, "grad_norm": 0.6891695649887497, "learning_rate": 5.808090917637586e-06, "loss": 0.2904, "step": 13450 }, { "epoch": 0.9298997580366402, "grad_norm": 0.7658318093480032, "learning_rate": 5.807538512303983e-06, "loss": 0.4295, "step": 13451 }, { "epoch": 0.9299688904251642, "grad_norm": 0.7885217171357143, "learning_rate": 5.8069860968493276e-06, "loss": 0.3386, "step": 13452 }, { "epoch": 0.9300380228136882, "grad_norm": 0.7436499541382612, "learning_rate": 5.806433671280541e-06, "loss": 0.2882, "step": 13453 }, { "epoch": 0.9301071552022122, "grad_norm": 0.6371936496101988, "learning_rate": 5.805881235604548e-06, "loss": 0.3537, "step": 13454 }, { "epoch": 0.9301762875907362, "grad_norm": 0.7056552229503281, "learning_rate": 5.805328789828271e-06, "loss": 0.4137, "step": 13455 }, { "epoch": 0.9302454199792602, "grad_norm": 0.6177163576658841, "learning_rate": 5.804776333958636e-06, "loss": 0.2935, "step": 13456 }, { "epoch": 0.9303145523677843, "grad_norm": 0.9605045425758018, "learning_rate": 5.804223868002566e-06, "loss": 0.3082, "step": 13457 }, { "epoch": 0.9303836847563083, "grad_norm": 1.0184224817687726, "learning_rate": 5.803671391966983e-06, "loss": 0.3761, "step": 13458 }, { "epoch": 0.9304528171448323, "grad_norm": 0.6751145071434669, "learning_rate": 5.803118905858814e-06, "loss": 0.3205, "step": 13459 }, { "epoch": 0.9305219495333564, "grad_norm": 0.7355707067761025, "learning_rate": 5.802566409684984e-06, "loss": 0.2582, "step": 13460 }, { "epoch": 0.9305910819218804, "grad_norm": 0.7200346365578982, "learning_rate": 5.802013903452417e-06, "loss": 0.3646, "step": 13461 }, { "epoch": 0.9306602143104045, "grad_norm": 0.7919120100332548, "learning_rate": 5.8014613871680345e-06, "loss": 0.3872, "step": 13462 }, { "epoch": 0.9307293466989285, "grad_norm": 0.6262966866424104, "learning_rate": 5.800908860838765e-06, "loss": 0.2701, "step": 13463 }, { "epoch": 0.9307984790874525, "grad_norm": 0.6539465736007897, "learning_rate": 5.800356324471532e-06, "loss": 0.2415, "step": 13464 }, { "epoch": 0.9308676114759765, "grad_norm": 0.7314197835978133, "learning_rate": 5.799803778073261e-06, "loss": 0.3877, "step": 13465 }, { "epoch": 0.9309367438645005, "grad_norm": 0.6679544693094798, "learning_rate": 5.799251221650877e-06, "loss": 0.2694, "step": 13466 }, { "epoch": 0.9310058762530246, "grad_norm": 0.7036687556992616, "learning_rate": 5.798698655211304e-06, "loss": 0.2607, "step": 13467 }, { "epoch": 0.9310750086415486, "grad_norm": 0.6179337858791696, "learning_rate": 5.798146078761471e-06, "loss": 0.3157, "step": 13468 }, { "epoch": 0.9311441410300726, "grad_norm": 0.7157425466048725, "learning_rate": 5.797593492308299e-06, "loss": 0.3669, "step": 13469 }, { "epoch": 0.9312132734185966, "grad_norm": 0.6933319357211141, "learning_rate": 5.797040895858717e-06, "loss": 0.2724, "step": 13470 }, { "epoch": 0.9312824058071206, "grad_norm": 0.7459033176158772, "learning_rate": 5.796488289419652e-06, "loss": 0.3029, "step": 13471 }, { "epoch": 0.9313515381956446, "grad_norm": 0.7106465743526493, "learning_rate": 5.795935672998025e-06, "loss": 0.3829, "step": 13472 }, { "epoch": 0.9314206705841687, "grad_norm": 0.6449351122157021, "learning_rate": 5.795383046600764e-06, "loss": 0.3965, "step": 13473 }, { "epoch": 0.9314898029726927, "grad_norm": 0.7449235269357437, "learning_rate": 5.794830410234798e-06, "loss": 0.2775, "step": 13474 }, { "epoch": 0.9315589353612167, "grad_norm": 0.6822322428502032, "learning_rate": 5.79427776390705e-06, "loss": 0.2756, "step": 13475 }, { "epoch": 0.9316280677497407, "grad_norm": 0.7785915440548883, "learning_rate": 5.793725107624449e-06, "loss": 0.3826, "step": 13476 }, { "epoch": 0.9316972001382647, "grad_norm": 0.6493504822016575, "learning_rate": 5.79317244139392e-06, "loss": 0.2644, "step": 13477 }, { "epoch": 0.9317663325267888, "grad_norm": 0.8399860347372821, "learning_rate": 5.792619765222388e-06, "loss": 0.2626, "step": 13478 }, { "epoch": 0.9318354649153128, "grad_norm": 0.8416840330979986, "learning_rate": 5.792067079116782e-06, "loss": 0.36, "step": 13479 }, { "epoch": 0.9319045973038369, "grad_norm": 0.6602263657516818, "learning_rate": 5.7915143830840305e-06, "loss": 0.393, "step": 13480 }, { "epoch": 0.9319737296923609, "grad_norm": 0.6665133669586385, "learning_rate": 5.790961677131059e-06, "loss": 0.3049, "step": 13481 }, { "epoch": 0.9320428620808849, "grad_norm": 0.7118503835584407, "learning_rate": 5.790408961264793e-06, "loss": 0.2964, "step": 13482 }, { "epoch": 0.932111994469409, "grad_norm": 0.7145460192052969, "learning_rate": 5.789856235492161e-06, "loss": 0.3508, "step": 13483 }, { "epoch": 0.932181126857933, "grad_norm": 1.1735250849577794, "learning_rate": 5.789303499820091e-06, "loss": 0.2923, "step": 13484 }, { "epoch": 0.932250259246457, "grad_norm": 0.7055695649671159, "learning_rate": 5.78875075425551e-06, "loss": 0.2551, "step": 13485 }, { "epoch": 0.932319391634981, "grad_norm": 0.6513478989658502, "learning_rate": 5.788197998805346e-06, "loss": 0.335, "step": 13486 }, { "epoch": 0.932388524023505, "grad_norm": 0.6910057451038987, "learning_rate": 5.7876452334765275e-06, "loss": 0.3788, "step": 13487 }, { "epoch": 0.932457656412029, "grad_norm": 0.5721935144891275, "learning_rate": 5.7870924582759815e-06, "loss": 0.2292, "step": 13488 }, { "epoch": 0.9325267888005531, "grad_norm": 0.6638069906095071, "learning_rate": 5.786539673210636e-06, "loss": 0.2891, "step": 13489 }, { "epoch": 0.9325959211890771, "grad_norm": 0.7311697535567997, "learning_rate": 5.785986878287417e-06, "loss": 0.3742, "step": 13490 }, { "epoch": 0.9326650535776011, "grad_norm": 0.7005360423598469, "learning_rate": 5.785434073513259e-06, "loss": 0.3024, "step": 13491 }, { "epoch": 0.9327341859661251, "grad_norm": 0.6245904524021377, "learning_rate": 5.784881258895084e-06, "loss": 0.2453, "step": 13492 }, { "epoch": 0.9328033183546491, "grad_norm": 0.7086443465241391, "learning_rate": 5.784328434439823e-06, "loss": 0.34, "step": 13493 }, { "epoch": 0.9328724507431732, "grad_norm": 0.7557763633257022, "learning_rate": 5.7837756001544065e-06, "loss": 0.3273, "step": 13494 }, { "epoch": 0.9329415831316972, "grad_norm": 0.9691018557247956, "learning_rate": 5.783222756045761e-06, "loss": 0.2746, "step": 13495 }, { "epoch": 0.9330107155202212, "grad_norm": 0.7031493806462272, "learning_rate": 5.782669902120816e-06, "loss": 0.2472, "step": 13496 }, { "epoch": 0.9330798479087452, "grad_norm": 0.7136493042582055, "learning_rate": 5.782117038386499e-06, "loss": 0.4402, "step": 13497 }, { "epoch": 0.9331489802972692, "grad_norm": 0.7373428857624738, "learning_rate": 5.7815641648497425e-06, "loss": 0.4032, "step": 13498 }, { "epoch": 0.9332181126857932, "grad_norm": 0.6489397650778137, "learning_rate": 5.7810112815174736e-06, "loss": 0.2913, "step": 13499 }, { "epoch": 0.9332872450743174, "grad_norm": 0.6451377096845875, "learning_rate": 5.780458388396622e-06, "loss": 0.3367, "step": 13500 }, { "epoch": 0.9333563774628414, "grad_norm": 0.6140914101470333, "learning_rate": 5.7799054854941165e-06, "loss": 0.4084, "step": 13501 }, { "epoch": 0.9334255098513654, "grad_norm": 0.9304193660882425, "learning_rate": 5.7793525728168875e-06, "loss": 0.2854, "step": 13502 }, { "epoch": 0.9334946422398894, "grad_norm": 0.6502512006229048, "learning_rate": 5.778799650371866e-06, "loss": 0.3251, "step": 13503 }, { "epoch": 0.9335637746284134, "grad_norm": 0.7365298432798902, "learning_rate": 5.77824671816598e-06, "loss": 0.4282, "step": 13504 }, { "epoch": 0.9336329070169375, "grad_norm": 0.6852601366057935, "learning_rate": 5.77769377620616e-06, "loss": 0.3533, "step": 13505 }, { "epoch": 0.9337020394054615, "grad_norm": 0.640558826155749, "learning_rate": 5.777140824499337e-06, "loss": 0.328, "step": 13506 }, { "epoch": 0.9337711717939855, "grad_norm": 0.6552411507375605, "learning_rate": 5.776587863052441e-06, "loss": 0.3019, "step": 13507 }, { "epoch": 0.9338403041825095, "grad_norm": 0.6427542286817388, "learning_rate": 5.776034891872401e-06, "loss": 0.4178, "step": 13508 }, { "epoch": 0.9339094365710335, "grad_norm": 0.6269093139973093, "learning_rate": 5.7754819109661485e-06, "loss": 0.2859, "step": 13509 }, { "epoch": 0.9339785689595576, "grad_norm": 0.6338340464750336, "learning_rate": 5.774928920340614e-06, "loss": 0.2911, "step": 13510 }, { "epoch": 0.9340477013480816, "grad_norm": 0.7093697374275635, "learning_rate": 5.774375920002729e-06, "loss": 0.3756, "step": 13511 }, { "epoch": 0.9341168337366056, "grad_norm": 0.6636254867718114, "learning_rate": 5.7738229099594246e-06, "loss": 0.3616, "step": 13512 }, { "epoch": 0.9341859661251296, "grad_norm": 0.5919233330966243, "learning_rate": 5.77326989021763e-06, "loss": 0.2408, "step": 13513 }, { "epoch": 0.9342550985136536, "grad_norm": 0.6924541333062122, "learning_rate": 5.7727168607842775e-06, "loss": 0.3056, "step": 13514 }, { "epoch": 0.9343242309021776, "grad_norm": 0.6345541951177841, "learning_rate": 5.772163821666298e-06, "loss": 0.3842, "step": 13515 }, { "epoch": 0.9343933632907017, "grad_norm": 0.6368198437436662, "learning_rate": 5.7716107728706224e-06, "loss": 0.3087, "step": 13516 }, { "epoch": 0.9344624956792257, "grad_norm": 0.6103137082436736, "learning_rate": 5.771057714404185e-06, "loss": 0.2906, "step": 13517 }, { "epoch": 0.9345316280677497, "grad_norm": 0.6838198325306054, "learning_rate": 5.770504646273914e-06, "loss": 0.3594, "step": 13518 }, { "epoch": 0.9346007604562737, "grad_norm": 0.9473350195380538, "learning_rate": 5.76995156848674e-06, "loss": 0.4559, "step": 13519 }, { "epoch": 0.9346698928447977, "grad_norm": 0.6448817264949006, "learning_rate": 5.7693984810496e-06, "loss": 0.2728, "step": 13520 }, { "epoch": 0.9347390252333219, "grad_norm": 0.6627669493491758, "learning_rate": 5.768845383969421e-06, "loss": 0.2645, "step": 13521 }, { "epoch": 0.9348081576218459, "grad_norm": 0.6529695678119402, "learning_rate": 5.768292277253138e-06, "loss": 0.3861, "step": 13522 }, { "epoch": 0.9348772900103699, "grad_norm": 0.6310935853069602, "learning_rate": 5.767739160907683e-06, "loss": 0.2777, "step": 13523 }, { "epoch": 0.9349464223988939, "grad_norm": 0.6496751412425756, "learning_rate": 5.767186034939988e-06, "loss": 0.3012, "step": 13524 }, { "epoch": 0.9350155547874179, "grad_norm": 2.1263468085146404, "learning_rate": 5.7666328993569846e-06, "loss": 0.3195, "step": 13525 }, { "epoch": 0.935084687175942, "grad_norm": 0.7079888641201357, "learning_rate": 5.766079754165604e-06, "loss": 0.4005, "step": 13526 }, { "epoch": 0.935153819564466, "grad_norm": 0.6348753861653661, "learning_rate": 5.765526599372784e-06, "loss": 0.3189, "step": 13527 }, { "epoch": 0.93522295195299, "grad_norm": 0.6930050490578826, "learning_rate": 5.764973434985452e-06, "loss": 0.2815, "step": 13528 }, { "epoch": 0.935292084341514, "grad_norm": 0.6906373349651117, "learning_rate": 5.764420261010544e-06, "loss": 0.3791, "step": 13529 }, { "epoch": 0.935361216730038, "grad_norm": 0.7276870101602924, "learning_rate": 5.763867077454992e-06, "loss": 0.4127, "step": 13530 }, { "epoch": 0.935430349118562, "grad_norm": 0.6665861556906836, "learning_rate": 5.763313884325729e-06, "loss": 0.2715, "step": 13531 }, { "epoch": 0.9354994815070861, "grad_norm": 0.91471623826868, "learning_rate": 5.762760681629691e-06, "loss": 0.2876, "step": 13532 }, { "epoch": 0.9355686138956101, "grad_norm": 0.6651371597324841, "learning_rate": 5.762207469373806e-06, "loss": 0.4022, "step": 13533 }, { "epoch": 0.9356377462841341, "grad_norm": 0.6151199337444222, "learning_rate": 5.7616542475650125e-06, "loss": 0.2909, "step": 13534 }, { "epoch": 0.9357068786726581, "grad_norm": 0.6443823557404815, "learning_rate": 5.761101016210242e-06, "loss": 0.2606, "step": 13535 }, { "epoch": 0.9357760110611821, "grad_norm": 0.6892788314414415, "learning_rate": 5.760547775316427e-06, "loss": 0.2997, "step": 13536 }, { "epoch": 0.9358451434497062, "grad_norm": 0.6801891838002568, "learning_rate": 5.759994524890505e-06, "loss": 0.4262, "step": 13537 }, { "epoch": 0.9359142758382302, "grad_norm": 0.6385740375223813, "learning_rate": 5.759441264939407e-06, "loss": 0.2552, "step": 13538 }, { "epoch": 0.9359834082267542, "grad_norm": 0.643890994325788, "learning_rate": 5.758887995470068e-06, "loss": 0.2409, "step": 13539 }, { "epoch": 0.9360525406152782, "grad_norm": 0.6497464308691268, "learning_rate": 5.758334716489422e-06, "loss": 0.3973, "step": 13540 }, { "epoch": 0.9361216730038023, "grad_norm": 0.6540610603657881, "learning_rate": 5.757781428004404e-06, "loss": 0.3337, "step": 13541 }, { "epoch": 0.9361908053923264, "grad_norm": 0.6621213271440949, "learning_rate": 5.757228130021949e-06, "loss": 0.3146, "step": 13542 }, { "epoch": 0.9362599377808504, "grad_norm": 0.6834035460597756, "learning_rate": 5.756674822548989e-06, "loss": 0.2875, "step": 13543 }, { "epoch": 0.9363290701693744, "grad_norm": 0.6375606612308004, "learning_rate": 5.756121505592462e-06, "loss": 0.325, "step": 13544 }, { "epoch": 0.9363982025578984, "grad_norm": 0.5694867606363463, "learning_rate": 5.7555681791593e-06, "loss": 0.2589, "step": 13545 }, { "epoch": 0.9364673349464224, "grad_norm": 0.8069674135474992, "learning_rate": 5.755014843256439e-06, "loss": 0.3158, "step": 13546 }, { "epoch": 0.9365364673349464, "grad_norm": 0.7100032182115859, "learning_rate": 5.754461497890815e-06, "loss": 0.4344, "step": 13547 }, { "epoch": 0.9366055997234705, "grad_norm": 0.631905687257955, "learning_rate": 5.753908143069363e-06, "loss": 0.3489, "step": 13548 }, { "epoch": 0.9366747321119945, "grad_norm": 0.6508229248678786, "learning_rate": 5.753354778799017e-06, "loss": 0.3052, "step": 13549 }, { "epoch": 0.9367438645005185, "grad_norm": 0.6322071266949942, "learning_rate": 5.7528014050867134e-06, "loss": 0.3378, "step": 13550 }, { "epoch": 0.9368129968890425, "grad_norm": 0.6136247594596396, "learning_rate": 5.7522480219393885e-06, "loss": 0.3842, "step": 13551 }, { "epoch": 0.9368821292775665, "grad_norm": 0.6746364861774912, "learning_rate": 5.751694629363976e-06, "loss": 0.2884, "step": 13552 }, { "epoch": 0.9369512616660906, "grad_norm": 0.6571378142361647, "learning_rate": 5.751141227367414e-06, "loss": 0.2941, "step": 13553 }, { "epoch": 0.9370203940546146, "grad_norm": 0.6838666163661788, "learning_rate": 5.750587815956636e-06, "loss": 0.3682, "step": 13554 }, { "epoch": 0.9370895264431386, "grad_norm": 0.6844307196925358, "learning_rate": 5.750034395138579e-06, "loss": 0.3522, "step": 13555 }, { "epoch": 0.9371586588316626, "grad_norm": 0.6466879336520032, "learning_rate": 5.749480964920181e-06, "loss": 0.232, "step": 13556 }, { "epoch": 0.9372277912201866, "grad_norm": 0.6979148683770728, "learning_rate": 5.748927525308374e-06, "loss": 0.2934, "step": 13557 }, { "epoch": 0.9372969236087106, "grad_norm": 0.7455271891055522, "learning_rate": 5.748374076310099e-06, "loss": 0.4256, "step": 13558 }, { "epoch": 0.9373660559972347, "grad_norm": 0.7058641394057621, "learning_rate": 5.74782061793229e-06, "loss": 0.3677, "step": 13559 }, { "epoch": 0.9374351883857587, "grad_norm": 0.6588351622628449, "learning_rate": 5.7472671501818856e-06, "loss": 0.2515, "step": 13560 }, { "epoch": 0.9375043207742828, "grad_norm": 0.6527800538774963, "learning_rate": 5.746713673065818e-06, "loss": 0.2999, "step": 13561 }, { "epoch": 0.9375734531628068, "grad_norm": 0.7282540408188577, "learning_rate": 5.746160186591029e-06, "loss": 0.3762, "step": 13562 }, { "epoch": 0.9376425855513308, "grad_norm": 0.6491416652162775, "learning_rate": 5.745606690764454e-06, "loss": 0.3131, "step": 13563 }, { "epoch": 0.9377117179398549, "grad_norm": 0.7771263270390483, "learning_rate": 5.745053185593028e-06, "loss": 0.2876, "step": 13564 }, { "epoch": 0.9377808503283789, "grad_norm": 0.6473327682651611, "learning_rate": 5.744499671083692e-06, "loss": 0.3726, "step": 13565 }, { "epoch": 0.9378499827169029, "grad_norm": 0.7139820124683379, "learning_rate": 5.74394614724338e-06, "loss": 0.3477, "step": 13566 }, { "epoch": 0.9379191151054269, "grad_norm": 0.6714613230318943, "learning_rate": 5.743392614079031e-06, "loss": 0.2858, "step": 13567 }, { "epoch": 0.9379882474939509, "grad_norm": 0.6685523021021611, "learning_rate": 5.742839071597581e-06, "loss": 0.3092, "step": 13568 }, { "epoch": 0.938057379882475, "grad_norm": 0.652308160475966, "learning_rate": 5.74228551980597e-06, "loss": 0.3203, "step": 13569 }, { "epoch": 0.938126512270999, "grad_norm": 0.6260186990719205, "learning_rate": 5.7417319587111355e-06, "loss": 0.2856, "step": 13570 }, { "epoch": 0.938195644659523, "grad_norm": 0.5692190729207707, "learning_rate": 5.741178388320014e-06, "loss": 0.2875, "step": 13571 }, { "epoch": 0.938264777048047, "grad_norm": 0.7499750123648047, "learning_rate": 5.740624808639544e-06, "loss": 0.4099, "step": 13572 }, { "epoch": 0.938333909436571, "grad_norm": 0.7257349888840958, "learning_rate": 5.740071219676663e-06, "loss": 0.3148, "step": 13573 }, { "epoch": 0.938403041825095, "grad_norm": 0.6610966995519199, "learning_rate": 5.739517621438311e-06, "loss": 0.2698, "step": 13574 }, { "epoch": 0.9384721742136191, "grad_norm": 0.6330964038584991, "learning_rate": 5.738964013931424e-06, "loss": 0.2932, "step": 13575 }, { "epoch": 0.9385413066021431, "grad_norm": 0.7987735514020505, "learning_rate": 5.738410397162943e-06, "loss": 0.4342, "step": 13576 }, { "epoch": 0.9386104389906671, "grad_norm": 0.682663948124669, "learning_rate": 5.737856771139806e-06, "loss": 0.3268, "step": 13577 }, { "epoch": 0.9386795713791911, "grad_norm": 0.6985323962150479, "learning_rate": 5.73730313586895e-06, "loss": 0.2812, "step": 13578 }, { "epoch": 0.9387487037677151, "grad_norm": 0.6719573133491211, "learning_rate": 5.736749491357314e-06, "loss": 0.3556, "step": 13579 }, { "epoch": 0.9388178361562391, "grad_norm": 0.6547425946476484, "learning_rate": 5.73619583761184e-06, "loss": 0.3314, "step": 13580 }, { "epoch": 0.9388869685447632, "grad_norm": 0.6352392554044494, "learning_rate": 5.735642174639464e-06, "loss": 0.3025, "step": 13581 }, { "epoch": 0.9389561009332873, "grad_norm": 0.7101513027161974, "learning_rate": 5.735088502447126e-06, "loss": 0.3571, "step": 13582 }, { "epoch": 0.9390252333218113, "grad_norm": 0.6443332455645235, "learning_rate": 5.734534821041766e-06, "loss": 0.4025, "step": 13583 }, { "epoch": 0.9390943657103353, "grad_norm": 0.650450697591239, "learning_rate": 5.733981130430323e-06, "loss": 0.2655, "step": 13584 }, { "epoch": 0.9391634980988594, "grad_norm": 0.6900785329105988, "learning_rate": 5.7334274306197365e-06, "loss": 0.2835, "step": 13585 }, { "epoch": 0.9392326304873834, "grad_norm": 1.124007469297111, "learning_rate": 5.732873721616945e-06, "loss": 0.3825, "step": 13586 }, { "epoch": 0.9393017628759074, "grad_norm": 0.6644427204194352, "learning_rate": 5.732320003428889e-06, "loss": 0.3998, "step": 13587 }, { "epoch": 0.9393708952644314, "grad_norm": 0.6774682810641398, "learning_rate": 5.73176627606251e-06, "loss": 0.2211, "step": 13588 }, { "epoch": 0.9394400276529554, "grad_norm": 0.6249116330685318, "learning_rate": 5.731212539524746e-06, "loss": 0.245, "step": 13589 }, { "epoch": 0.9395091600414794, "grad_norm": 0.7390505874526148, "learning_rate": 5.7306587938225376e-06, "loss": 0.395, "step": 13590 }, { "epoch": 0.9395782924300035, "grad_norm": 0.6877763611240042, "learning_rate": 5.730105038962825e-06, "loss": 0.3963, "step": 13591 }, { "epoch": 0.9396474248185275, "grad_norm": 0.7422435651300899, "learning_rate": 5.72955127495255e-06, "loss": 0.2828, "step": 13592 }, { "epoch": 0.9397165572070515, "grad_norm": 0.6610039094983808, "learning_rate": 5.72899750179865e-06, "loss": 0.3631, "step": 13593 }, { "epoch": 0.9397856895955755, "grad_norm": 0.6694931781332863, "learning_rate": 5.728443719508068e-06, "loss": 0.4148, "step": 13594 }, { "epoch": 0.9398548219840995, "grad_norm": 0.6717370382957749, "learning_rate": 5.727889928087744e-06, "loss": 0.2969, "step": 13595 }, { "epoch": 0.9399239543726235, "grad_norm": 0.6082762665806114, "learning_rate": 5.727336127544619e-06, "loss": 0.305, "step": 13596 }, { "epoch": 0.9399930867611476, "grad_norm": 0.7608539304944648, "learning_rate": 5.726782317885633e-06, "loss": 0.4109, "step": 13597 }, { "epoch": 0.9400622191496716, "grad_norm": 0.7205633635953165, "learning_rate": 5.726228499117728e-06, "loss": 0.3677, "step": 13598 }, { "epoch": 0.9401313515381956, "grad_norm": 0.7311545097588374, "learning_rate": 5.7256746712478454e-06, "loss": 0.2821, "step": 13599 }, { "epoch": 0.9402004839267196, "grad_norm": 0.6570071896360793, "learning_rate": 5.725120834282924e-06, "loss": 0.3862, "step": 13600 }, { "epoch": 0.9402696163152436, "grad_norm": 0.7945830658633272, "learning_rate": 5.724566988229908e-06, "loss": 0.3546, "step": 13601 }, { "epoch": 0.9403387487037678, "grad_norm": 0.6715024505105447, "learning_rate": 5.724013133095738e-06, "loss": 0.3277, "step": 13602 }, { "epoch": 0.9404078810922918, "grad_norm": 0.8991989682811489, "learning_rate": 5.7234592688873555e-06, "loss": 0.287, "step": 13603 }, { "epoch": 0.9404770134808158, "grad_norm": 0.6375302762641004, "learning_rate": 5.722905395611703e-06, "loss": 0.354, "step": 13604 }, { "epoch": 0.9405461458693398, "grad_norm": 0.7201057577297977, "learning_rate": 5.7223515132757205e-06, "loss": 0.326, "step": 13605 }, { "epoch": 0.9406152782578638, "grad_norm": 0.7063572135647567, "learning_rate": 5.721797621886351e-06, "loss": 0.2633, "step": 13606 }, { "epoch": 0.9406844106463879, "grad_norm": 0.605742594195879, "learning_rate": 5.721243721450536e-06, "loss": 0.2812, "step": 13607 }, { "epoch": 0.9407535430349119, "grad_norm": 0.8670092534538321, "learning_rate": 5.720689811975218e-06, "loss": 0.4022, "step": 13608 }, { "epoch": 0.9408226754234359, "grad_norm": 0.620315380990182, "learning_rate": 5.7201358934673404e-06, "loss": 0.313, "step": 13609 }, { "epoch": 0.9408918078119599, "grad_norm": 0.647709996798095, "learning_rate": 5.719581965933843e-06, "loss": 0.2534, "step": 13610 }, { "epoch": 0.9409609402004839, "grad_norm": 0.6657217300397648, "learning_rate": 5.71902802938167e-06, "loss": 0.3232, "step": 13611 }, { "epoch": 0.941030072589008, "grad_norm": 0.6831122543846819, "learning_rate": 5.718474083817765e-06, "loss": 0.4047, "step": 13612 }, { "epoch": 0.941099204977532, "grad_norm": 0.6297982001330428, "learning_rate": 5.71792012924907e-06, "loss": 0.2777, "step": 13613 }, { "epoch": 0.941168337366056, "grad_norm": 0.6722697214489535, "learning_rate": 5.717366165682525e-06, "loss": 0.2777, "step": 13614 }, { "epoch": 0.94123746975458, "grad_norm": 0.7143599355628555, "learning_rate": 5.716812193125077e-06, "loss": 0.358, "step": 13615 }, { "epoch": 0.941306602143104, "grad_norm": 0.7606949496245887, "learning_rate": 5.7162582115836675e-06, "loss": 0.3439, "step": 13616 }, { "epoch": 0.941375734531628, "grad_norm": 0.646150936147227, "learning_rate": 5.715704221065238e-06, "loss": 0.3002, "step": 13617 }, { "epoch": 0.9414448669201521, "grad_norm": 0.703711403029536, "learning_rate": 5.715150221576734e-06, "loss": 0.3332, "step": 13618 }, { "epoch": 0.9415139993086761, "grad_norm": 0.9018477257306402, "learning_rate": 5.714596213125098e-06, "loss": 0.3785, "step": 13619 }, { "epoch": 0.9415831316972001, "grad_norm": 0.6973749607150759, "learning_rate": 5.714042195717276e-06, "loss": 0.2926, "step": 13620 }, { "epoch": 0.9416522640857241, "grad_norm": 0.7231810511172879, "learning_rate": 5.713488169360206e-06, "loss": 0.2653, "step": 13621 }, { "epoch": 0.9417213964742482, "grad_norm": 0.7338234007965126, "learning_rate": 5.712934134060837e-06, "loss": 0.3417, "step": 13622 }, { "epoch": 0.9417905288627723, "grad_norm": 0.6574576316978311, "learning_rate": 5.712380089826112e-06, "loss": 0.3494, "step": 13623 }, { "epoch": 0.9418596612512963, "grad_norm": 0.6806747478541021, "learning_rate": 5.711826036662972e-06, "loss": 0.3059, "step": 13624 }, { "epoch": 0.9419287936398203, "grad_norm": 0.7720690337915929, "learning_rate": 5.711271974578363e-06, "loss": 0.331, "step": 13625 }, { "epoch": 0.9419979260283443, "grad_norm": 0.7137741636282366, "learning_rate": 5.7107179035792296e-06, "loss": 0.4603, "step": 13626 }, { "epoch": 0.9420670584168683, "grad_norm": 0.7608395474560582, "learning_rate": 5.710163823672515e-06, "loss": 0.3121, "step": 13627 }, { "epoch": 0.9421361908053923, "grad_norm": 0.778879159919473, "learning_rate": 5.709609734865165e-06, "loss": 0.2946, "step": 13628 }, { "epoch": 0.9422053231939164, "grad_norm": 0.7212615694658409, "learning_rate": 5.709055637164123e-06, "loss": 0.3474, "step": 13629 }, { "epoch": 0.9422744555824404, "grad_norm": 0.6561721603831491, "learning_rate": 5.708501530576335e-06, "loss": 0.343, "step": 13630 }, { "epoch": 0.9423435879709644, "grad_norm": 0.6176376188387765, "learning_rate": 5.7079474151087435e-06, "loss": 0.312, "step": 13631 }, { "epoch": 0.9424127203594884, "grad_norm": 0.6976825675531179, "learning_rate": 5.707393290768294e-06, "loss": 0.284, "step": 13632 }, { "epoch": 0.9424818527480124, "grad_norm": 0.7562229853575252, "learning_rate": 5.706839157561933e-06, "loss": 0.429, "step": 13633 }, { "epoch": 0.9425509851365365, "grad_norm": 0.7165579275063171, "learning_rate": 5.706285015496605e-06, "loss": 0.3138, "step": 13634 }, { "epoch": 0.9426201175250605, "grad_norm": 0.617656136640689, "learning_rate": 5.705730864579253e-06, "loss": 0.2367, "step": 13635 }, { "epoch": 0.9426892499135845, "grad_norm": 0.7200325674756306, "learning_rate": 5.705176704816825e-06, "loss": 0.3866, "step": 13636 }, { "epoch": 0.9427583823021085, "grad_norm": 0.6738216708184709, "learning_rate": 5.704622536216267e-06, "loss": 0.395, "step": 13637 }, { "epoch": 0.9428275146906325, "grad_norm": 0.6491667784850401, "learning_rate": 5.704068358784521e-06, "loss": 0.2553, "step": 13638 }, { "epoch": 0.9428966470791565, "grad_norm": 0.6357709696652067, "learning_rate": 5.703514172528535e-06, "loss": 0.3293, "step": 13639 }, { "epoch": 0.9429657794676806, "grad_norm": 0.6044734777837701, "learning_rate": 5.702959977455253e-06, "loss": 0.3277, "step": 13640 }, { "epoch": 0.9430349118562046, "grad_norm": 0.6808454756906767, "learning_rate": 5.702405773571626e-06, "loss": 0.3194, "step": 13641 }, { "epoch": 0.9431040442447286, "grad_norm": 0.6191290775096162, "learning_rate": 5.701851560884593e-06, "loss": 0.252, "step": 13642 }, { "epoch": 0.9431731766332527, "grad_norm": 0.6217557135159534, "learning_rate": 5.701297339401103e-06, "loss": 0.306, "step": 13643 }, { "epoch": 0.9432423090217767, "grad_norm": 1.0254641651951122, "learning_rate": 5.700743109128104e-06, "loss": 0.3692, "step": 13644 }, { "epoch": 0.9433114414103008, "grad_norm": 0.7617335441176679, "learning_rate": 5.7001888700725405e-06, "loss": 0.2776, "step": 13645 }, { "epoch": 0.9433805737988248, "grad_norm": 0.6477751397618824, "learning_rate": 5.699634622241358e-06, "loss": 0.2834, "step": 13646 }, { "epoch": 0.9434497061873488, "grad_norm": 0.6984647321422568, "learning_rate": 5.699080365641504e-06, "loss": 0.3763, "step": 13647 }, { "epoch": 0.9435188385758728, "grad_norm": 0.6943326897435946, "learning_rate": 5.6985261002799266e-06, "loss": 0.3762, "step": 13648 }, { "epoch": 0.9435879709643968, "grad_norm": 0.6939780851910673, "learning_rate": 5.697971826163571e-06, "loss": 0.3475, "step": 13649 }, { "epoch": 0.9436571033529209, "grad_norm": 0.7500027040322501, "learning_rate": 5.697417543299382e-06, "loss": 0.2938, "step": 13650 }, { "epoch": 0.9437262357414449, "grad_norm": 0.8572962204240042, "learning_rate": 5.69686325169431e-06, "loss": 0.3879, "step": 13651 }, { "epoch": 0.9437953681299689, "grad_norm": 0.6791552960830786, "learning_rate": 5.6963089513553015e-06, "loss": 0.2544, "step": 13652 }, { "epoch": 0.9438645005184929, "grad_norm": 0.6391806940099239, "learning_rate": 5.695754642289302e-06, "loss": 0.2515, "step": 13653 }, { "epoch": 0.9439336329070169, "grad_norm": 0.7930481302326444, "learning_rate": 5.69520032450326e-06, "loss": 0.4063, "step": 13654 }, { "epoch": 0.944002765295541, "grad_norm": 0.6310908787642201, "learning_rate": 5.694645998004122e-06, "loss": 0.385, "step": 13655 }, { "epoch": 0.944071897684065, "grad_norm": 0.6565373675867616, "learning_rate": 5.6940916627988375e-06, "loss": 0.2891, "step": 13656 }, { "epoch": 0.944141030072589, "grad_norm": 0.7013369161229427, "learning_rate": 5.6935373188943514e-06, "loss": 0.3506, "step": 13657 }, { "epoch": 0.944210162461113, "grad_norm": 0.7446838419912407, "learning_rate": 5.692982966297613e-06, "loss": 0.3763, "step": 13658 }, { "epoch": 0.944279294849637, "grad_norm": 0.6802945311139258, "learning_rate": 5.692428605015571e-06, "loss": 0.2969, "step": 13659 }, { "epoch": 0.944348427238161, "grad_norm": 0.6298508424279775, "learning_rate": 5.69187423505517e-06, "loss": 0.2621, "step": 13660 }, { "epoch": 0.944417559626685, "grad_norm": 0.6624275152848235, "learning_rate": 5.691319856423361e-06, "loss": 0.3585, "step": 13661 }, { "epoch": 0.9444866920152091, "grad_norm": 0.691021653617344, "learning_rate": 5.690765469127092e-06, "loss": 0.399, "step": 13662 }, { "epoch": 0.9445558244037332, "grad_norm": 0.6348047996806891, "learning_rate": 5.6902110731733106e-06, "loss": 0.2664, "step": 13663 }, { "epoch": 0.9446249567922572, "grad_norm": 0.6633451953717419, "learning_rate": 5.689656668568964e-06, "loss": 0.2833, "step": 13664 }, { "epoch": 0.9446940891807812, "grad_norm": 0.6915095486343513, "learning_rate": 5.6891022553210015e-06, "loss": 0.3573, "step": 13665 }, { "epoch": 0.9447632215693053, "grad_norm": 0.6366505591981106, "learning_rate": 5.6885478334363745e-06, "loss": 0.2602, "step": 13666 }, { "epoch": 0.9448323539578293, "grad_norm": 0.6080934850810307, "learning_rate": 5.6879934029220276e-06, "loss": 0.2456, "step": 13667 }, { "epoch": 0.9449014863463533, "grad_norm": 1.1453027262745317, "learning_rate": 5.687438963784911e-06, "loss": 0.3893, "step": 13668 }, { "epoch": 0.9449706187348773, "grad_norm": 0.7511641695447175, "learning_rate": 5.686884516031973e-06, "loss": 0.414, "step": 13669 }, { "epoch": 0.9450397511234013, "grad_norm": 0.6303339670889453, "learning_rate": 5.686330059670166e-06, "loss": 0.2685, "step": 13670 }, { "epoch": 0.9451088835119253, "grad_norm": 0.609349325428188, "learning_rate": 5.685775594706434e-06, "loss": 0.3091, "step": 13671 }, { "epoch": 0.9451780159004494, "grad_norm": 0.6511301833857849, "learning_rate": 5.685221121147729e-06, "loss": 0.3698, "step": 13672 }, { "epoch": 0.9452471482889734, "grad_norm": 0.654549686825192, "learning_rate": 5.684666639001002e-06, "loss": 0.3697, "step": 13673 }, { "epoch": 0.9453162806774974, "grad_norm": 0.6338921247412174, "learning_rate": 5.684112148273198e-06, "loss": 0.2453, "step": 13674 }, { "epoch": 0.9453854130660214, "grad_norm": 0.6812575684682683, "learning_rate": 5.6835576489712704e-06, "loss": 0.2741, "step": 13675 }, { "epoch": 0.9454545454545454, "grad_norm": 0.6512979996409406, "learning_rate": 5.683003141102168e-06, "loss": 0.3344, "step": 13676 }, { "epoch": 0.9455236778430695, "grad_norm": 0.6635340125961269, "learning_rate": 5.682448624672839e-06, "loss": 0.2589, "step": 13677 }, { "epoch": 0.9455928102315935, "grad_norm": 0.6453365797839099, "learning_rate": 5.681894099690235e-06, "loss": 0.2951, "step": 13678 }, { "epoch": 0.9456619426201175, "grad_norm": 0.6852829596662573, "learning_rate": 5.6813395661613054e-06, "loss": 0.4026, "step": 13679 }, { "epoch": 0.9457310750086415, "grad_norm": 0.6672465416745184, "learning_rate": 5.680785024093001e-06, "loss": 0.3397, "step": 13680 }, { "epoch": 0.9458002073971655, "grad_norm": 0.673408195237644, "learning_rate": 5.680230473492269e-06, "loss": 0.2575, "step": 13681 }, { "epoch": 0.9458693397856895, "grad_norm": 0.7997614777223971, "learning_rate": 5.679675914366063e-06, "loss": 0.3205, "step": 13682 }, { "epoch": 0.9459384721742137, "grad_norm": 0.6695081540054396, "learning_rate": 5.679121346721334e-06, "loss": 0.3551, "step": 13683 }, { "epoch": 0.9460076045627377, "grad_norm": 0.6456532166637277, "learning_rate": 5.678566770565028e-06, "loss": 0.2698, "step": 13684 }, { "epoch": 0.9460767369512617, "grad_norm": 0.67678239632572, "learning_rate": 5.6780121859041e-06, "loss": 0.2491, "step": 13685 }, { "epoch": 0.9461458693397857, "grad_norm": 0.647527261741262, "learning_rate": 5.677457592745499e-06, "loss": 0.3048, "step": 13686 }, { "epoch": 0.9462150017283097, "grad_norm": 0.8125474593761055, "learning_rate": 5.676902991096177e-06, "loss": 0.3561, "step": 13687 }, { "epoch": 0.9462841341168338, "grad_norm": 0.6207606111185948, "learning_rate": 5.676348380963082e-06, "loss": 0.2284, "step": 13688 }, { "epoch": 0.9463532665053578, "grad_norm": 0.765008616652415, "learning_rate": 5.675793762353169e-06, "loss": 0.3444, "step": 13689 }, { "epoch": 0.9464223988938818, "grad_norm": 0.7008243308837867, "learning_rate": 5.675239135273387e-06, "loss": 0.3643, "step": 13690 }, { "epoch": 0.9464915312824058, "grad_norm": 0.6796794442411619, "learning_rate": 5.674684499730687e-06, "loss": 0.3061, "step": 13691 }, { "epoch": 0.9465606636709298, "grad_norm": 1.259846884761337, "learning_rate": 5.67412985573202e-06, "loss": 0.2967, "step": 13692 }, { "epoch": 0.9466297960594539, "grad_norm": 0.7333193933249659, "learning_rate": 5.67357520328434e-06, "loss": 0.2842, "step": 13693 }, { "epoch": 0.9466989284479779, "grad_norm": 0.6709038046544317, "learning_rate": 5.673020542394595e-06, "loss": 0.3749, "step": 13694 }, { "epoch": 0.9467680608365019, "grad_norm": 0.6341586934330468, "learning_rate": 5.67246587306974e-06, "loss": 0.3062, "step": 13695 }, { "epoch": 0.9468371932250259, "grad_norm": 0.692749005845278, "learning_rate": 5.671911195316724e-06, "loss": 0.2731, "step": 13696 }, { "epoch": 0.9469063256135499, "grad_norm": 0.6532749533211557, "learning_rate": 5.6713565091425016e-06, "loss": 0.4354, "step": 13697 }, { "epoch": 0.9469754580020739, "grad_norm": 0.6347751347481367, "learning_rate": 5.670801814554023e-06, "loss": 0.3129, "step": 13698 }, { "epoch": 0.947044590390598, "grad_norm": 0.6798888634765322, "learning_rate": 5.67024711155824e-06, "loss": 0.3235, "step": 13699 }, { "epoch": 0.947113722779122, "grad_norm": 0.8665063105744337, "learning_rate": 5.669692400162106e-06, "loss": 0.3963, "step": 13700 }, { "epoch": 0.947182855167646, "grad_norm": 0.7909424242575629, "learning_rate": 5.669137680372574e-06, "loss": 0.4167, "step": 13701 }, { "epoch": 0.94725198755617, "grad_norm": 0.6151633428533069, "learning_rate": 5.668582952196594e-06, "loss": 0.2733, "step": 13702 }, { "epoch": 0.947321119944694, "grad_norm": 0.5465187278461064, "learning_rate": 5.668028215641119e-06, "loss": 0.2392, "step": 13703 }, { "epoch": 0.9473902523332182, "grad_norm": 0.7786899826394205, "learning_rate": 5.667473470713105e-06, "loss": 0.4201, "step": 13704 }, { "epoch": 0.9474593847217422, "grad_norm": 0.7898998444490487, "learning_rate": 5.666918717419501e-06, "loss": 0.3222, "step": 13705 }, { "epoch": 0.9475285171102662, "grad_norm": 0.6337598311227105, "learning_rate": 5.66636395576726e-06, "loss": 0.275, "step": 13706 }, { "epoch": 0.9475976494987902, "grad_norm": 0.6531453502183912, "learning_rate": 5.665809185763338e-06, "loss": 0.324, "step": 13707 }, { "epoch": 0.9476667818873142, "grad_norm": 0.6615504932552669, "learning_rate": 5.6652544074146846e-06, "loss": 0.4025, "step": 13708 }, { "epoch": 0.9477359142758383, "grad_norm": 0.9375586948675984, "learning_rate": 5.664699620728255e-06, "loss": 0.2806, "step": 13709 }, { "epoch": 0.9478050466643623, "grad_norm": 0.6182032753938842, "learning_rate": 5.664144825711001e-06, "loss": 0.2779, "step": 13710 }, { "epoch": 0.9478741790528863, "grad_norm": 0.7534739081228427, "learning_rate": 5.663590022369878e-06, "loss": 0.3998, "step": 13711 }, { "epoch": 0.9479433114414103, "grad_norm": 0.6685885102028439, "learning_rate": 5.663035210711838e-06, "loss": 0.4209, "step": 13712 }, { "epoch": 0.9480124438299343, "grad_norm": 0.6929169023962156, "learning_rate": 5.662480390743835e-06, "loss": 0.2764, "step": 13713 }, { "epoch": 0.9480815762184583, "grad_norm": 0.5580545217422751, "learning_rate": 5.661925562472821e-06, "loss": 0.2536, "step": 13714 }, { "epoch": 0.9481507086069824, "grad_norm": 0.9264767301305052, "learning_rate": 5.661370725905753e-06, "loss": 0.4496, "step": 13715 }, { "epoch": 0.9482198409955064, "grad_norm": 0.6966386874422339, "learning_rate": 5.660815881049583e-06, "loss": 0.3252, "step": 13716 }, { "epoch": 0.9482889733840304, "grad_norm": 0.7235181457556508, "learning_rate": 5.660261027911264e-06, "loss": 0.2878, "step": 13717 }, { "epoch": 0.9483581057725544, "grad_norm": 0.709478751646311, "learning_rate": 5.659706166497753e-06, "loss": 0.3534, "step": 13718 }, { "epoch": 0.9484272381610784, "grad_norm": 0.8029716717346299, "learning_rate": 5.659151296816002e-06, "loss": 0.3535, "step": 13719 }, { "epoch": 0.9484963705496025, "grad_norm": 0.817698391234445, "learning_rate": 5.6585964188729655e-06, "loss": 0.304, "step": 13720 }, { "epoch": 0.9485655029381265, "grad_norm": 0.7227514259376452, "learning_rate": 5.658041532675598e-06, "loss": 0.2774, "step": 13721 }, { "epoch": 0.9486346353266505, "grad_norm": 0.6965773509662753, "learning_rate": 5.657486638230854e-06, "loss": 0.4138, "step": 13722 }, { "epoch": 0.9487037677151745, "grad_norm": 0.6681459624606656, "learning_rate": 5.6569317355456896e-06, "loss": 0.2968, "step": 13723 }, { "epoch": 0.9487729001036986, "grad_norm": 0.5977551429460781, "learning_rate": 5.656376824627056e-06, "loss": 0.216, "step": 13724 }, { "epoch": 0.9488420324922227, "grad_norm": 0.6701019377294868, "learning_rate": 5.655821905481913e-06, "loss": 0.305, "step": 13725 }, { "epoch": 0.9489111648807467, "grad_norm": 0.6585866148850711, "learning_rate": 5.655266978117212e-06, "loss": 0.3825, "step": 13726 }, { "epoch": 0.9489802972692707, "grad_norm": 0.6825948808662453, "learning_rate": 5.654712042539909e-06, "loss": 0.2923, "step": 13727 }, { "epoch": 0.9490494296577947, "grad_norm": 0.6028985525328778, "learning_rate": 5.654157098756957e-06, "loss": 0.2295, "step": 13728 }, { "epoch": 0.9491185620463187, "grad_norm": 0.7053036775506856, "learning_rate": 5.653602146775314e-06, "loss": 0.3331, "step": 13729 }, { "epoch": 0.9491876944348427, "grad_norm": 0.7426546562237614, "learning_rate": 5.653047186601935e-06, "loss": 0.3834, "step": 13730 }, { "epoch": 0.9492568268233668, "grad_norm": 0.6708441751710411, "learning_rate": 5.652492218243776e-06, "loss": 0.258, "step": 13731 }, { "epoch": 0.9493259592118908, "grad_norm": 0.6551014009943125, "learning_rate": 5.65193724170779e-06, "loss": 0.3243, "step": 13732 }, { "epoch": 0.9493950916004148, "grad_norm": 0.6414162625651277, "learning_rate": 5.651382257000934e-06, "loss": 0.3925, "step": 13733 }, { "epoch": 0.9494642239889388, "grad_norm": 0.6359104425349931, "learning_rate": 5.650827264130165e-06, "loss": 0.3054, "step": 13734 }, { "epoch": 0.9495333563774628, "grad_norm": 0.6356678005474385, "learning_rate": 5.6502722631024366e-06, "loss": 0.2814, "step": 13735 }, { "epoch": 0.9496024887659869, "grad_norm": 0.6665869700598364, "learning_rate": 5.649717253924706e-06, "loss": 0.3705, "step": 13736 }, { "epoch": 0.9496716211545109, "grad_norm": 0.8054481109167871, "learning_rate": 5.649162236603931e-06, "loss": 0.3886, "step": 13737 }, { "epoch": 0.9497407535430349, "grad_norm": 0.6067867468154671, "learning_rate": 5.648607211147063e-06, "loss": 0.2702, "step": 13738 }, { "epoch": 0.9498098859315589, "grad_norm": 0.5898061602865324, "learning_rate": 5.6480521775610634e-06, "loss": 0.2873, "step": 13739 }, { "epoch": 0.9498790183200829, "grad_norm": 0.7919035262206765, "learning_rate": 5.647497135852884e-06, "loss": 0.3972, "step": 13740 }, { "epoch": 0.9499481507086069, "grad_norm": 0.6843832922090322, "learning_rate": 5.646942086029486e-06, "loss": 0.3023, "step": 13741 }, { "epoch": 0.950017283097131, "grad_norm": 0.67710919300314, "learning_rate": 5.64638702809782e-06, "loss": 0.2921, "step": 13742 }, { "epoch": 0.950086415485655, "grad_norm": 0.6448157339065473, "learning_rate": 5.64583196206485e-06, "loss": 0.3296, "step": 13743 }, { "epoch": 0.9501555478741791, "grad_norm": 0.6543045292070208, "learning_rate": 5.645276887937526e-06, "loss": 0.3965, "step": 13744 }, { "epoch": 0.9502246802627031, "grad_norm": 1.5076609750879917, "learning_rate": 5.644721805722809e-06, "loss": 0.2792, "step": 13745 }, { "epoch": 0.9502938126512271, "grad_norm": 0.7427162428374444, "learning_rate": 5.644166715427655e-06, "loss": 0.3386, "step": 13746 }, { "epoch": 0.9503629450397512, "grad_norm": 0.8219944199393057, "learning_rate": 5.64361161705902e-06, "loss": 0.3514, "step": 13747 }, { "epoch": 0.9504320774282752, "grad_norm": 0.6362586598993173, "learning_rate": 5.643056510623863e-06, "loss": 0.3592, "step": 13748 }, { "epoch": 0.9505012098167992, "grad_norm": 0.7010148812374702, "learning_rate": 5.642501396129139e-06, "loss": 0.3546, "step": 13749 }, { "epoch": 0.9505703422053232, "grad_norm": 0.7993665395904379, "learning_rate": 5.641946273581807e-06, "loss": 0.3044, "step": 13750 }, { "epoch": 0.9506394745938472, "grad_norm": 0.6733276680200216, "learning_rate": 5.641391142988825e-06, "loss": 0.383, "step": 13751 }, { "epoch": 0.9507086069823713, "grad_norm": 0.7009398627631703, "learning_rate": 5.640836004357149e-06, "loss": 0.3103, "step": 13752 }, { "epoch": 0.9507777393708953, "grad_norm": 0.6400287563479142, "learning_rate": 5.6402808576937364e-06, "loss": 0.3009, "step": 13753 }, { "epoch": 0.9508468717594193, "grad_norm": 0.7581921157841797, "learning_rate": 5.639725703005547e-06, "loss": 0.3332, "step": 13754 }, { "epoch": 0.9509160041479433, "grad_norm": 0.703669752688285, "learning_rate": 5.6391705402995375e-06, "loss": 0.3841, "step": 13755 }, { "epoch": 0.9509851365364673, "grad_norm": 0.5983041707999333, "learning_rate": 5.638615369582665e-06, "loss": 0.2627, "step": 13756 }, { "epoch": 0.9510542689249913, "grad_norm": 0.6620268420835357, "learning_rate": 5.638060190861889e-06, "loss": 0.3307, "step": 13757 }, { "epoch": 0.9511234013135154, "grad_norm": 0.7562834138489521, "learning_rate": 5.637505004144168e-06, "loss": 0.3722, "step": 13758 }, { "epoch": 0.9511925337020394, "grad_norm": 0.8196121163556219, "learning_rate": 5.636949809436459e-06, "loss": 0.2607, "step": 13759 }, { "epoch": 0.9512616660905634, "grad_norm": 0.8407887803290482, "learning_rate": 5.636394606745719e-06, "loss": 0.2676, "step": 13760 }, { "epoch": 0.9513307984790874, "grad_norm": 0.721635179051231, "learning_rate": 5.635839396078912e-06, "loss": 0.3911, "step": 13761 }, { "epoch": 0.9513999308676114, "grad_norm": 2.433749959797972, "learning_rate": 5.63528417744299e-06, "loss": 0.3486, "step": 13762 }, { "epoch": 0.9514690632561354, "grad_norm": 0.673927827632099, "learning_rate": 5.634728950844914e-06, "loss": 0.2724, "step": 13763 }, { "epoch": 0.9515381956446595, "grad_norm": 0.6582947242864661, "learning_rate": 5.6341737162916454e-06, "loss": 0.2377, "step": 13764 }, { "epoch": 0.9516073280331836, "grad_norm": 0.6841311596282752, "learning_rate": 5.633618473790141e-06, "loss": 0.3274, "step": 13765 }, { "epoch": 0.9516764604217076, "grad_norm": 0.6800766045569967, "learning_rate": 5.633063223347359e-06, "loss": 0.2969, "step": 13766 }, { "epoch": 0.9517455928102316, "grad_norm": 0.7421393726663769, "learning_rate": 5.632507964970259e-06, "loss": 0.3035, "step": 13767 }, { "epoch": 0.9518147251987557, "grad_norm": 0.8576653975378568, "learning_rate": 5.6319526986658015e-06, "loss": 0.344, "step": 13768 }, { "epoch": 0.9518838575872797, "grad_norm": 0.6877234419144654, "learning_rate": 5.631397424440943e-06, "loss": 0.4801, "step": 13769 }, { "epoch": 0.9519529899758037, "grad_norm": 0.690245567099877, "learning_rate": 5.630842142302645e-06, "loss": 0.2964, "step": 13770 }, { "epoch": 0.9520221223643277, "grad_norm": 0.6122033819003226, "learning_rate": 5.630286852257867e-06, "loss": 0.3033, "step": 13771 }, { "epoch": 0.9520912547528517, "grad_norm": 0.6443016637436932, "learning_rate": 5.629731554313569e-06, "loss": 0.3767, "step": 13772 }, { "epoch": 0.9521603871413757, "grad_norm": 0.6468371203281083, "learning_rate": 5.629176248476709e-06, "loss": 0.3886, "step": 13773 }, { "epoch": 0.9522295195298998, "grad_norm": 0.6760606828715642, "learning_rate": 5.628620934754246e-06, "loss": 0.2936, "step": 13774 }, { "epoch": 0.9522986519184238, "grad_norm": 0.6872936948068402, "learning_rate": 5.6280656131531444e-06, "loss": 0.343, "step": 13775 }, { "epoch": 0.9523677843069478, "grad_norm": 0.6423785327827265, "learning_rate": 5.62751028368036e-06, "loss": 0.3734, "step": 13776 }, { "epoch": 0.9524369166954718, "grad_norm": 0.69539383642382, "learning_rate": 5.626954946342852e-06, "loss": 0.2832, "step": 13777 }, { "epoch": 0.9525060490839958, "grad_norm": 0.6104382655141959, "learning_rate": 5.626399601147584e-06, "loss": 0.2565, "step": 13778 }, { "epoch": 0.9525751814725198, "grad_norm": 0.6783764752330419, "learning_rate": 5.625844248101516e-06, "loss": 0.3227, "step": 13779 }, { "epoch": 0.9526443138610439, "grad_norm": 0.649022570624918, "learning_rate": 5.625288887211606e-06, "loss": 0.3219, "step": 13780 }, { "epoch": 0.9527134462495679, "grad_norm": 0.6712223108928902, "learning_rate": 5.624733518484816e-06, "loss": 0.2923, "step": 13781 }, { "epoch": 0.9527825786380919, "grad_norm": 0.7434207376155216, "learning_rate": 5.624178141928106e-06, "loss": 0.323, "step": 13782 }, { "epoch": 0.9528517110266159, "grad_norm": 0.7290550849339851, "learning_rate": 5.623622757548438e-06, "loss": 0.3914, "step": 13783 }, { "epoch": 0.9529208434151399, "grad_norm": 0.7374925169213221, "learning_rate": 5.623067365352771e-06, "loss": 0.2834, "step": 13784 }, { "epoch": 0.9529899758036641, "grad_norm": 0.6144181310770599, "learning_rate": 5.622511965348065e-06, "loss": 0.2919, "step": 13785 }, { "epoch": 0.9530591081921881, "grad_norm": 0.79479198170831, "learning_rate": 5.621956557541284e-06, "loss": 0.3356, "step": 13786 }, { "epoch": 0.9531282405807121, "grad_norm": 0.655553938740568, "learning_rate": 5.621401141939387e-06, "loss": 0.335, "step": 13787 }, { "epoch": 0.9531973729692361, "grad_norm": 0.6831192095454164, "learning_rate": 5.620845718549336e-06, "loss": 0.2586, "step": 13788 }, { "epoch": 0.9532665053577601, "grad_norm": 0.7121841428686493, "learning_rate": 5.620290287378092e-06, "loss": 0.3166, "step": 13789 }, { "epoch": 0.9533356377462842, "grad_norm": 0.7537570402950977, "learning_rate": 5.619734848432615e-06, "loss": 0.3929, "step": 13790 }, { "epoch": 0.9534047701348082, "grad_norm": 0.741985375893265, "learning_rate": 5.61917940171987e-06, "loss": 0.2846, "step": 13791 }, { "epoch": 0.9534739025233322, "grad_norm": 0.5456494036598265, "learning_rate": 5.618623947246813e-06, "loss": 0.275, "step": 13792 }, { "epoch": 0.9535430349118562, "grad_norm": 0.6457694613518338, "learning_rate": 5.618068485020411e-06, "loss": 0.3201, "step": 13793 }, { "epoch": 0.9536121673003802, "grad_norm": 0.5767726668920928, "learning_rate": 5.617513015047623e-06, "loss": 0.3049, "step": 13794 }, { "epoch": 0.9536812996889042, "grad_norm": 0.7407330351823749, "learning_rate": 5.616957537335412e-06, "loss": 0.3211, "step": 13795 }, { "epoch": 0.9537504320774283, "grad_norm": 0.566398071028175, "learning_rate": 5.616402051890738e-06, "loss": 0.2315, "step": 13796 }, { "epoch": 0.9538195644659523, "grad_norm": 0.6635540859397289, "learning_rate": 5.615846558720566e-06, "loss": 0.3655, "step": 13797 }, { "epoch": 0.9538886968544763, "grad_norm": 0.6215876736110647, "learning_rate": 5.615291057831855e-06, "loss": 0.3596, "step": 13798 }, { "epoch": 0.9539578292430003, "grad_norm": 0.6406853429253262, "learning_rate": 5.614735549231568e-06, "loss": 0.2904, "step": 13799 }, { "epoch": 0.9540269616315243, "grad_norm": 1.202958213940182, "learning_rate": 5.614180032926669e-06, "loss": 0.2923, "step": 13800 }, { "epoch": 0.9540960940200484, "grad_norm": 0.7076505941277716, "learning_rate": 5.61362450892412e-06, "loss": 0.4396, "step": 13801 }, { "epoch": 0.9541652264085724, "grad_norm": 0.6830714906513177, "learning_rate": 5.613068977230882e-06, "loss": 0.3152, "step": 13802 }, { "epoch": 0.9542343587970964, "grad_norm": 0.6535309031793572, "learning_rate": 5.612513437853919e-06, "loss": 0.2243, "step": 13803 }, { "epoch": 0.9543034911856204, "grad_norm": 0.6403607479704441, "learning_rate": 5.611957890800193e-06, "loss": 0.391, "step": 13804 }, { "epoch": 0.9543726235741445, "grad_norm": 0.7784083291392888, "learning_rate": 5.611402336076666e-06, "loss": 0.3389, "step": 13805 }, { "epoch": 0.9544417559626686, "grad_norm": 0.6950701152340343, "learning_rate": 5.610846773690303e-06, "loss": 0.2634, "step": 13806 }, { "epoch": 0.9545108883511926, "grad_norm": 0.6375543590289791, "learning_rate": 5.610291203648065e-06, "loss": 0.294, "step": 13807 }, { "epoch": 0.9545800207397166, "grad_norm": 0.9785205607435381, "learning_rate": 5.609735625956917e-06, "loss": 0.3506, "step": 13808 }, { "epoch": 0.9546491531282406, "grad_norm": 0.6295165829248462, "learning_rate": 5.609180040623819e-06, "loss": 0.2439, "step": 13809 }, { "epoch": 0.9547182855167646, "grad_norm": 0.6162265463495427, "learning_rate": 5.6086244476557376e-06, "loss": 0.262, "step": 13810 }, { "epoch": 0.9547874179052886, "grad_norm": 0.662279521905591, "learning_rate": 5.608068847059635e-06, "loss": 0.37, "step": 13811 }, { "epoch": 0.9548565502938127, "grad_norm": 0.7052263637297742, "learning_rate": 5.607513238842474e-06, "loss": 0.3739, "step": 13812 }, { "epoch": 0.9549256826823367, "grad_norm": 0.9885081965613151, "learning_rate": 5.606957623011219e-06, "loss": 0.3656, "step": 13813 }, { "epoch": 0.9549948150708607, "grad_norm": 0.7841228933586983, "learning_rate": 5.6064019995728335e-06, "loss": 0.2934, "step": 13814 }, { "epoch": 0.9550639474593847, "grad_norm": 0.6703929248118565, "learning_rate": 5.6058463685342815e-06, "loss": 0.3764, "step": 13815 }, { "epoch": 0.9551330798479087, "grad_norm": 0.7127486898194736, "learning_rate": 5.605290729902525e-06, "loss": 0.3223, "step": 13816 }, { "epoch": 0.9552022122364328, "grad_norm": 0.6434418739696521, "learning_rate": 5.604735083684531e-06, "loss": 0.2702, "step": 13817 }, { "epoch": 0.9552713446249568, "grad_norm": 0.6922666736327009, "learning_rate": 5.6041794298872605e-06, "loss": 0.3079, "step": 13818 }, { "epoch": 0.9553404770134808, "grad_norm": 0.6863577492177936, "learning_rate": 5.603623768517681e-06, "loss": 0.3851, "step": 13819 }, { "epoch": 0.9554096094020048, "grad_norm": 0.5854867138575015, "learning_rate": 5.603068099582752e-06, "loss": 0.2558, "step": 13820 }, { "epoch": 0.9554787417905288, "grad_norm": 0.6443520996032917, "learning_rate": 5.602512423089443e-06, "loss": 0.2972, "step": 13821 }, { "epoch": 0.9555478741790528, "grad_norm": 0.661506012721565, "learning_rate": 5.601956739044715e-06, "loss": 0.3967, "step": 13822 }, { "epoch": 0.9556170065675769, "grad_norm": 0.7213425874490971, "learning_rate": 5.601401047455534e-06, "loss": 0.3069, "step": 13823 }, { "epoch": 0.9556861389561009, "grad_norm": 0.6965309407855957, "learning_rate": 5.6008453483288635e-06, "loss": 0.2518, "step": 13824 }, { "epoch": 0.9557552713446249, "grad_norm": 0.6677297359294978, "learning_rate": 5.600289641671669e-06, "loss": 0.2873, "step": 13825 }, { "epoch": 0.955824403733149, "grad_norm": 0.6899225039388615, "learning_rate": 5.599733927490916e-06, "loss": 0.4437, "step": 13826 }, { "epoch": 0.955893536121673, "grad_norm": 0.6298097547001904, "learning_rate": 5.5991782057935675e-06, "loss": 0.3584, "step": 13827 }, { "epoch": 0.9559626685101971, "grad_norm": 0.7771209165245417, "learning_rate": 5.59862247658659e-06, "loss": 0.2717, "step": 13828 }, { "epoch": 0.9560318008987211, "grad_norm": 0.717371518877571, "learning_rate": 5.598066739876947e-06, "loss": 0.3369, "step": 13829 }, { "epoch": 0.9561009332872451, "grad_norm": 0.621908927435388, "learning_rate": 5.597510995671606e-06, "loss": 0.294, "step": 13830 }, { "epoch": 0.9561700656757691, "grad_norm": 0.6659749346706477, "learning_rate": 5.59695524397753e-06, "loss": 0.2899, "step": 13831 }, { "epoch": 0.9562391980642931, "grad_norm": 0.603188785404784, "learning_rate": 5.596399484801686e-06, "loss": 0.2628, "step": 13832 }, { "epoch": 0.9563083304528172, "grad_norm": 0.6322914111426328, "learning_rate": 5.595843718151038e-06, "loss": 0.3786, "step": 13833 }, { "epoch": 0.9563774628413412, "grad_norm": 0.7548847853990941, "learning_rate": 5.595287944032552e-06, "loss": 0.2886, "step": 13834 }, { "epoch": 0.9564465952298652, "grad_norm": 0.8980518292531017, "learning_rate": 5.594732162453196e-06, "loss": 0.3113, "step": 13835 }, { "epoch": 0.9565157276183892, "grad_norm": 0.7088226564725475, "learning_rate": 5.594176373419933e-06, "loss": 0.3054, "step": 13836 }, { "epoch": 0.9565848600069132, "grad_norm": 0.6535568625973224, "learning_rate": 5.593620576939729e-06, "loss": 0.3438, "step": 13837 }, { "epoch": 0.9566539923954372, "grad_norm": 0.6215750740348629, "learning_rate": 5.593064773019549e-06, "loss": 0.2838, "step": 13838 }, { "epoch": 0.9567231247839613, "grad_norm": 0.6827022533213187, "learning_rate": 5.592508961666363e-06, "loss": 0.3136, "step": 13839 }, { "epoch": 0.9567922571724853, "grad_norm": 1.0373145335810232, "learning_rate": 5.591953142887133e-06, "loss": 0.4384, "step": 13840 }, { "epoch": 0.9568613895610093, "grad_norm": 0.734367510117255, "learning_rate": 5.591397316688827e-06, "loss": 0.3343, "step": 13841 }, { "epoch": 0.9569305219495333, "grad_norm": 0.6820859315463492, "learning_rate": 5.590841483078412e-06, "loss": 0.2891, "step": 13842 }, { "epoch": 0.9569996543380573, "grad_norm": 0.7290454979113062, "learning_rate": 5.590285642062852e-06, "loss": 0.3236, "step": 13843 }, { "epoch": 0.9570687867265814, "grad_norm": 0.7094098915928818, "learning_rate": 5.589729793649117e-06, "loss": 0.3947, "step": 13844 }, { "epoch": 0.9571379191151054, "grad_norm": 0.6877222742071334, "learning_rate": 5.589173937844169e-06, "loss": 0.3338, "step": 13845 }, { "epoch": 0.9572070515036295, "grad_norm": 0.7011926598776397, "learning_rate": 5.588618074654978e-06, "loss": 0.2806, "step": 13846 }, { "epoch": 0.9572761838921535, "grad_norm": 0.6381583517409989, "learning_rate": 5.58806220408851e-06, "loss": 0.3802, "step": 13847 }, { "epoch": 0.9573453162806775, "grad_norm": 0.7623466491651618, "learning_rate": 5.587506326151731e-06, "loss": 0.3292, "step": 13848 }, { "epoch": 0.9574144486692016, "grad_norm": 0.6640115851245337, "learning_rate": 5.586950440851609e-06, "loss": 0.2346, "step": 13849 }, { "epoch": 0.9574835810577256, "grad_norm": 0.7753158086516296, "learning_rate": 5.586394548195112e-06, "loss": 0.3761, "step": 13850 }, { "epoch": 0.9575527134462496, "grad_norm": 0.6214618561436535, "learning_rate": 5.5858386481892055e-06, "loss": 0.3826, "step": 13851 }, { "epoch": 0.9576218458347736, "grad_norm": 0.6728350631162358, "learning_rate": 5.585282740840855e-06, "loss": 0.2923, "step": 13852 }, { "epoch": 0.9576909782232976, "grad_norm": 0.6290383123650342, "learning_rate": 5.584726826157032e-06, "loss": 0.2491, "step": 13853 }, { "epoch": 0.9577601106118216, "grad_norm": 0.7030941486239503, "learning_rate": 5.584170904144701e-06, "loss": 0.3574, "step": 13854 }, { "epoch": 0.9578292430003457, "grad_norm": 0.7640653027196722, "learning_rate": 5.5836149748108304e-06, "loss": 0.3554, "step": 13855 }, { "epoch": 0.9578983753888697, "grad_norm": 0.6409974458715789, "learning_rate": 5.583059038162386e-06, "loss": 0.2837, "step": 13856 }, { "epoch": 0.9579675077773937, "grad_norm": 0.6291744002301759, "learning_rate": 5.58250309420634e-06, "loss": 0.2975, "step": 13857 }, { "epoch": 0.9580366401659177, "grad_norm": 0.66354890275264, "learning_rate": 5.581947142949655e-06, "loss": 0.3664, "step": 13858 }, { "epoch": 0.9581057725544417, "grad_norm": 0.7248260880657447, "learning_rate": 5.5813911843993005e-06, "loss": 0.2848, "step": 13859 }, { "epoch": 0.9581749049429658, "grad_norm": 0.6842994062797985, "learning_rate": 5.580835218562247e-06, "loss": 0.264, "step": 13860 }, { "epoch": 0.9582440373314898, "grad_norm": 0.7063254439865091, "learning_rate": 5.58027924544546e-06, "loss": 0.4124, "step": 13861 }, { "epoch": 0.9583131697200138, "grad_norm": 0.6807723255252136, "learning_rate": 5.5797232650559085e-06, "loss": 0.3593, "step": 13862 }, { "epoch": 0.9583823021085378, "grad_norm": 0.7683704529925097, "learning_rate": 5.579167277400559e-06, "loss": 0.2667, "step": 13863 }, { "epoch": 0.9584514344970618, "grad_norm": 0.6518122256543442, "learning_rate": 5.578611282486383e-06, "loss": 0.2942, "step": 13864 }, { "epoch": 0.9585205668855858, "grad_norm": 0.6963314387237394, "learning_rate": 5.578055280320347e-06, "loss": 0.3779, "step": 13865 }, { "epoch": 0.95858969927411, "grad_norm": 0.7461879688692852, "learning_rate": 5.577499270909417e-06, "loss": 0.3164, "step": 13866 }, { "epoch": 0.958658831662634, "grad_norm": 0.7190674307958371, "learning_rate": 5.576943254260567e-06, "loss": 0.3041, "step": 13867 }, { "epoch": 0.958727964051158, "grad_norm": 0.6630942846338507, "learning_rate": 5.5763872303807635e-06, "loss": 0.3162, "step": 13868 }, { "epoch": 0.958797096439682, "grad_norm": 0.7497172104844592, "learning_rate": 5.575831199276973e-06, "loss": 0.3562, "step": 13869 }, { "epoch": 0.958866228828206, "grad_norm": 0.6745598094467998, "learning_rate": 5.575275160956166e-06, "loss": 0.2887, "step": 13870 }, { "epoch": 0.9589353612167301, "grad_norm": 0.6376986612261694, "learning_rate": 5.574719115425313e-06, "loss": 0.2478, "step": 13871 }, { "epoch": 0.9590044936052541, "grad_norm": 0.6696673916599041, "learning_rate": 5.5741630626913805e-06, "loss": 0.3611, "step": 13872 }, { "epoch": 0.9590736259937781, "grad_norm": 0.6397981964117526, "learning_rate": 5.573607002761338e-06, "loss": 0.3361, "step": 13873 }, { "epoch": 0.9591427583823021, "grad_norm": 0.7055735360543993, "learning_rate": 5.573050935642156e-06, "loss": 0.2609, "step": 13874 }, { "epoch": 0.9592118907708261, "grad_norm": 1.140884200442919, "learning_rate": 5.572494861340804e-06, "loss": 0.295, "step": 13875 }, { "epoch": 0.9592810231593502, "grad_norm": 0.6484892527230126, "learning_rate": 5.571938779864249e-06, "loss": 0.3968, "step": 13876 }, { "epoch": 0.9593501555478742, "grad_norm": 0.6350171192098456, "learning_rate": 5.5713826912194626e-06, "loss": 0.2829, "step": 13877 }, { "epoch": 0.9594192879363982, "grad_norm": 0.643077892542861, "learning_rate": 5.570826595413414e-06, "loss": 0.3486, "step": 13878 }, { "epoch": 0.9594884203249222, "grad_norm": 0.6360408484582524, "learning_rate": 5.570270492453073e-06, "loss": 0.3639, "step": 13879 }, { "epoch": 0.9595575527134462, "grad_norm": 0.6943911022803101, "learning_rate": 5.569714382345408e-06, "loss": 0.3611, "step": 13880 }, { "epoch": 0.9596266851019702, "grad_norm": 0.6300881663830917, "learning_rate": 5.569158265097391e-06, "loss": 0.2605, "step": 13881 }, { "epoch": 0.9596958174904943, "grad_norm": 0.6697855454729501, "learning_rate": 5.568602140715991e-06, "loss": 0.2607, "step": 13882 }, { "epoch": 0.9597649498790183, "grad_norm": 0.6609214941810968, "learning_rate": 5.568046009208178e-06, "loss": 0.3508, "step": 13883 }, { "epoch": 0.9598340822675423, "grad_norm": 0.61920146028999, "learning_rate": 5.56748987058092e-06, "loss": 0.2829, "step": 13884 }, { "epoch": 0.9599032146560663, "grad_norm": 0.6257002701877143, "learning_rate": 5.566933724841192e-06, "loss": 0.247, "step": 13885 }, { "epoch": 0.9599723470445904, "grad_norm": 0.7269086005163902, "learning_rate": 5.56637757199596e-06, "loss": 0.3426, "step": 13886 }, { "epoch": 0.9600414794331145, "grad_norm": 0.672258150143168, "learning_rate": 5.5658214120521945e-06, "loss": 0.3742, "step": 13887 }, { "epoch": 0.9601106118216385, "grad_norm": 0.6580031485786025, "learning_rate": 5.5652652450168694e-06, "loss": 0.276, "step": 13888 }, { "epoch": 0.9601797442101625, "grad_norm": 0.655238501340919, "learning_rate": 5.564709070896953e-06, "loss": 0.3706, "step": 13889 }, { "epoch": 0.9602488765986865, "grad_norm": 0.6455619475488371, "learning_rate": 5.564152889699416e-06, "loss": 0.363, "step": 13890 }, { "epoch": 0.9603180089872105, "grad_norm": 0.9225334475374606, "learning_rate": 5.563596701431229e-06, "loss": 0.3112, "step": 13891 }, { "epoch": 0.9603871413757346, "grad_norm": 0.6983810903686097, "learning_rate": 5.563040506099363e-06, "loss": 0.2718, "step": 13892 }, { "epoch": 0.9604562737642586, "grad_norm": 0.6462592923813434, "learning_rate": 5.56248430371079e-06, "loss": 0.3092, "step": 13893 }, { "epoch": 0.9605254061527826, "grad_norm": 0.729635379664179, "learning_rate": 5.561928094272478e-06, "loss": 0.377, "step": 13894 }, { "epoch": 0.9605945385413066, "grad_norm": 0.7048993317878374, "learning_rate": 5.561371877791402e-06, "loss": 0.3259, "step": 13895 }, { "epoch": 0.9606636709298306, "grad_norm": 0.6719785927330425, "learning_rate": 5.560815654274531e-06, "loss": 0.2754, "step": 13896 }, { "epoch": 0.9607328033183546, "grad_norm": 0.6890405002164607, "learning_rate": 5.560259423728837e-06, "loss": 0.3734, "step": 13897 }, { "epoch": 0.9608019357068787, "grad_norm": 0.5886734853557201, "learning_rate": 5.559703186161289e-06, "loss": 0.2484, "step": 13898 }, { "epoch": 0.9608710680954027, "grad_norm": 0.7690631516704846, "learning_rate": 5.559146941578862e-06, "loss": 0.28, "step": 13899 }, { "epoch": 0.9609402004839267, "grad_norm": 0.7030987489147645, "learning_rate": 5.5585906899885264e-06, "loss": 0.3029, "step": 13900 }, { "epoch": 0.9610093328724507, "grad_norm": 0.5772073707264389, "learning_rate": 5.558034431397253e-06, "loss": 0.3947, "step": 13901 }, { "epoch": 0.9610784652609747, "grad_norm": 0.6566806759010188, "learning_rate": 5.557478165812012e-06, "loss": 0.2773, "step": 13902 }, { "epoch": 0.9611475976494988, "grad_norm": 0.6859021907337007, "learning_rate": 5.55692189323978e-06, "loss": 0.3646, "step": 13903 }, { "epoch": 0.9612167300380228, "grad_norm": 0.7029949174306076, "learning_rate": 5.556365613687524e-06, "loss": 0.4473, "step": 13904 }, { "epoch": 0.9612858624265468, "grad_norm": 0.6551843785782544, "learning_rate": 5.5558093271622185e-06, "loss": 0.3625, "step": 13905 }, { "epoch": 0.9613549948150708, "grad_norm": 0.632166495639845, "learning_rate": 5.555253033670834e-06, "loss": 0.2578, "step": 13906 }, { "epoch": 0.9614241272035949, "grad_norm": 0.6677305209903928, "learning_rate": 5.554696733220346e-06, "loss": 0.2743, "step": 13907 }, { "epoch": 0.961493259592119, "grad_norm": 0.6424453196285727, "learning_rate": 5.5541404258177225e-06, "loss": 0.3446, "step": 13908 }, { "epoch": 0.961562391980643, "grad_norm": 0.6892046499597263, "learning_rate": 5.553584111469937e-06, "loss": 0.2987, "step": 13909 }, { "epoch": 0.961631524369167, "grad_norm": 0.8883172299468595, "learning_rate": 5.553027790183965e-06, "loss": 0.3383, "step": 13910 }, { "epoch": 0.961700656757691, "grad_norm": 0.659377043234735, "learning_rate": 5.552471461966775e-06, "loss": 0.3226, "step": 13911 }, { "epoch": 0.961769789146215, "grad_norm": 0.8648882121710615, "learning_rate": 5.55191512682534e-06, "loss": 0.3918, "step": 13912 }, { "epoch": 0.961838921534739, "grad_norm": 0.6539028429590578, "learning_rate": 5.551358784766636e-06, "loss": 0.2663, "step": 13913 }, { "epoch": 0.9619080539232631, "grad_norm": 0.6720707332037178, "learning_rate": 5.550802435797633e-06, "loss": 0.2785, "step": 13914 }, { "epoch": 0.9619771863117871, "grad_norm": 0.6982689398155953, "learning_rate": 5.5502460799253045e-06, "loss": 0.4229, "step": 13915 }, { "epoch": 0.9620463187003111, "grad_norm": 0.6992754283714753, "learning_rate": 5.549689717156623e-06, "loss": 0.3371, "step": 13916 }, { "epoch": 0.9621154510888351, "grad_norm": 0.6317281975979024, "learning_rate": 5.549133347498561e-06, "loss": 0.327, "step": 13917 }, { "epoch": 0.9621845834773591, "grad_norm": 0.7010118870364295, "learning_rate": 5.548576970958094e-06, "loss": 0.2832, "step": 13918 }, { "epoch": 0.9622537158658832, "grad_norm": 0.6798624709905386, "learning_rate": 5.548020587542193e-06, "loss": 0.4048, "step": 13919 }, { "epoch": 0.9623228482544072, "grad_norm": 0.6352729987464668, "learning_rate": 5.547464197257831e-06, "loss": 0.3311, "step": 13920 }, { "epoch": 0.9623919806429312, "grad_norm": 0.6190813178718743, "learning_rate": 5.546907800111983e-06, "loss": 0.2478, "step": 13921 }, { "epoch": 0.9624611130314552, "grad_norm": 0.6669329221760875, "learning_rate": 5.546351396111623e-06, "loss": 0.3618, "step": 13922 }, { "epoch": 0.9625302454199792, "grad_norm": 0.6893501603024812, "learning_rate": 5.545794985263721e-06, "loss": 0.3409, "step": 13923 }, { "epoch": 0.9625993778085032, "grad_norm": 0.8278704766414829, "learning_rate": 5.545238567575255e-06, "loss": 0.2408, "step": 13924 }, { "epoch": 0.9626685101970273, "grad_norm": 0.6858487962670293, "learning_rate": 5.544682143053196e-06, "loss": 0.3104, "step": 13925 }, { "epoch": 0.9627376425855513, "grad_norm": 0.6532037499992012, "learning_rate": 5.544125711704518e-06, "loss": 0.3758, "step": 13926 }, { "epoch": 0.9628067749740754, "grad_norm": 0.6087645314167142, "learning_rate": 5.543569273536194e-06, "loss": 0.2807, "step": 13927 }, { "epoch": 0.9628759073625994, "grad_norm": 0.6863101360639816, "learning_rate": 5.543012828555201e-06, "loss": 0.2898, "step": 13928 }, { "epoch": 0.9629450397511234, "grad_norm": 0.6349513418264278, "learning_rate": 5.542456376768511e-06, "loss": 0.3234, "step": 13929 }, { "epoch": 0.9630141721396475, "grad_norm": 0.7874013772965385, "learning_rate": 5.5418999181830965e-06, "loss": 0.3464, "step": 13930 }, { "epoch": 0.9630833045281715, "grad_norm": 0.6407999522159498, "learning_rate": 5.541343452805936e-06, "loss": 0.3194, "step": 13931 }, { "epoch": 0.9631524369166955, "grad_norm": 0.9816456610069008, "learning_rate": 5.540786980644001e-06, "loss": 0.2827, "step": 13932 }, { "epoch": 0.9632215693052195, "grad_norm": 0.6696497366085136, "learning_rate": 5.540230501704266e-06, "loss": 0.3746, "step": 13933 }, { "epoch": 0.9632907016937435, "grad_norm": 0.6909813090622932, "learning_rate": 5.539674015993704e-06, "loss": 0.3419, "step": 13934 }, { "epoch": 0.9633598340822676, "grad_norm": 0.7087523927629438, "learning_rate": 5.539117523519294e-06, "loss": 0.2925, "step": 13935 }, { "epoch": 0.9634289664707916, "grad_norm": 0.6546596407283457, "learning_rate": 5.538561024288008e-06, "loss": 0.3109, "step": 13936 }, { "epoch": 0.9634980988593156, "grad_norm": 0.6602821305181129, "learning_rate": 5.538004518306819e-06, "loss": 0.3627, "step": 13937 }, { "epoch": 0.9635672312478396, "grad_norm": 0.581911029415749, "learning_rate": 5.537448005582704e-06, "loss": 0.2424, "step": 13938 }, { "epoch": 0.9636363636363636, "grad_norm": 0.6714656461468572, "learning_rate": 5.536891486122638e-06, "loss": 0.328, "step": 13939 }, { "epoch": 0.9637054960248876, "grad_norm": 2.0700615215792175, "learning_rate": 5.536334959933594e-06, "loss": 0.3907, "step": 13940 }, { "epoch": 0.9637746284134117, "grad_norm": 0.6429479838441821, "learning_rate": 5.535778427022549e-06, "loss": 0.3148, "step": 13941 }, { "epoch": 0.9638437608019357, "grad_norm": 0.6213439829447386, "learning_rate": 5.535221887396478e-06, "loss": 0.2778, "step": 13942 }, { "epoch": 0.9639128931904597, "grad_norm": 0.7044225334586612, "learning_rate": 5.534665341062355e-06, "loss": 0.3543, "step": 13943 }, { "epoch": 0.9639820255789837, "grad_norm": 0.8409277167520647, "learning_rate": 5.534108788027156e-06, "loss": 0.4132, "step": 13944 }, { "epoch": 0.9640511579675077, "grad_norm": 0.6459491514794186, "learning_rate": 5.533552228297857e-06, "loss": 0.2878, "step": 13945 }, { "epoch": 0.9641202903560318, "grad_norm": 0.6387877846109088, "learning_rate": 5.532995661881434e-06, "loss": 0.2643, "step": 13946 }, { "epoch": 0.9641894227445559, "grad_norm": 0.6860869300281184, "learning_rate": 5.532439088784859e-06, "loss": 0.3889, "step": 13947 }, { "epoch": 0.9642585551330799, "grad_norm": 0.6410062870104396, "learning_rate": 5.531882509015112e-06, "loss": 0.3206, "step": 13948 }, { "epoch": 0.9643276875216039, "grad_norm": 0.6722095984123169, "learning_rate": 5.531325922579167e-06, "loss": 0.2782, "step": 13949 }, { "epoch": 0.9643968199101279, "grad_norm": 0.8490159931058107, "learning_rate": 5.530769329484e-06, "loss": 0.3076, "step": 13950 }, { "epoch": 0.964465952298652, "grad_norm": 0.6183894766287987, "learning_rate": 5.530212729736585e-06, "loss": 0.324, "step": 13951 }, { "epoch": 0.964535084687176, "grad_norm": 0.6175873902482905, "learning_rate": 5.529656123343901e-06, "loss": 0.2567, "step": 13952 }, { "epoch": 0.9646042170757, "grad_norm": 0.7181392363818841, "learning_rate": 5.529099510312923e-06, "loss": 0.2519, "step": 13953 }, { "epoch": 0.964673349464224, "grad_norm": 0.6465827119437113, "learning_rate": 5.528542890650626e-06, "loss": 0.4125, "step": 13954 }, { "epoch": 0.964742481852748, "grad_norm": 0.6791248017182413, "learning_rate": 5.527986264363987e-06, "loss": 0.3392, "step": 13955 }, { "epoch": 0.964811614241272, "grad_norm": 0.560027971533912, "learning_rate": 5.527429631459983e-06, "loss": 0.2431, "step": 13956 }, { "epoch": 0.9648807466297961, "grad_norm": 0.7013914394401113, "learning_rate": 5.52687299194559e-06, "loss": 0.3161, "step": 13957 }, { "epoch": 0.9649498790183201, "grad_norm": 0.6056996028687046, "learning_rate": 5.5263163458277835e-06, "loss": 0.372, "step": 13958 }, { "epoch": 0.9650190114068441, "grad_norm": 0.6650736726098222, "learning_rate": 5.525759693113541e-06, "loss": 0.2877, "step": 13959 }, { "epoch": 0.9650881437953681, "grad_norm": 0.6383754428436721, "learning_rate": 5.525203033809839e-06, "loss": 0.3011, "step": 13960 }, { "epoch": 0.9651572761838921, "grad_norm": 0.6472403855345576, "learning_rate": 5.524646367923656e-06, "loss": 0.2991, "step": 13961 }, { "epoch": 0.9652264085724162, "grad_norm": 0.6069072455382969, "learning_rate": 5.524089695461965e-06, "loss": 0.3621, "step": 13962 }, { "epoch": 0.9652955409609402, "grad_norm": 0.620697070815371, "learning_rate": 5.5235330164317454e-06, "loss": 0.2644, "step": 13963 }, { "epoch": 0.9653646733494642, "grad_norm": 0.7459370226341653, "learning_rate": 5.522976330839974e-06, "loss": 0.2945, "step": 13964 }, { "epoch": 0.9654338057379882, "grad_norm": 0.7337986653131205, "learning_rate": 5.522419638693627e-06, "loss": 0.4025, "step": 13965 }, { "epoch": 0.9655029381265122, "grad_norm": 0.6698285123343536, "learning_rate": 5.521862939999683e-06, "loss": 0.3415, "step": 13966 }, { "epoch": 0.9655720705150362, "grad_norm": 0.6690108140318498, "learning_rate": 5.5213062347651184e-06, "loss": 0.2962, "step": 13967 }, { "epoch": 0.9656412029035604, "grad_norm": 0.6916864671218648, "learning_rate": 5.520749522996909e-06, "loss": 0.3455, "step": 13968 }, { "epoch": 0.9657103352920844, "grad_norm": 0.6886244045837492, "learning_rate": 5.5201928047020334e-06, "loss": 0.3878, "step": 13969 }, { "epoch": 0.9657794676806084, "grad_norm": 0.670200052046065, "learning_rate": 5.519636079887472e-06, "loss": 0.2745, "step": 13970 }, { "epoch": 0.9658486000691324, "grad_norm": 0.6186139123036541, "learning_rate": 5.519079348560198e-06, "loss": 0.2563, "step": 13971 }, { "epoch": 0.9659177324576564, "grad_norm": 0.7030813940560344, "learning_rate": 5.518522610727192e-06, "loss": 0.3572, "step": 13972 }, { "epoch": 0.9659868648461805, "grad_norm": 0.6081391247825384, "learning_rate": 5.517965866395428e-06, "loss": 0.3224, "step": 13973 }, { "epoch": 0.9660559972347045, "grad_norm": 0.6425089552216617, "learning_rate": 5.517409115571888e-06, "loss": 0.3295, "step": 13974 }, { "epoch": 0.9661251296232285, "grad_norm": 0.7735099447211847, "learning_rate": 5.516852358263548e-06, "loss": 0.3584, "step": 13975 }, { "epoch": 0.9661942620117525, "grad_norm": 0.7032658139603029, "learning_rate": 5.516295594477384e-06, "loss": 0.4948, "step": 13976 }, { "epoch": 0.9662633944002765, "grad_norm": 0.6144717674368184, "learning_rate": 5.515738824220378e-06, "loss": 0.2988, "step": 13977 }, { "epoch": 0.9663325267888005, "grad_norm": 0.8119232560865939, "learning_rate": 5.515182047499506e-06, "loss": 0.26, "step": 13978 }, { "epoch": 0.9664016591773246, "grad_norm": 0.7133961497971903, "learning_rate": 5.514625264321748e-06, "loss": 0.3615, "step": 13979 }, { "epoch": 0.9664707915658486, "grad_norm": 0.6712174132681269, "learning_rate": 5.5140684746940775e-06, "loss": 0.3552, "step": 13980 }, { "epoch": 0.9665399239543726, "grad_norm": 0.6460601328968393, "learning_rate": 5.513511678623477e-06, "loss": 0.2852, "step": 13981 }, { "epoch": 0.9666090563428966, "grad_norm": 0.7884530367293349, "learning_rate": 5.512954876116926e-06, "loss": 0.3021, "step": 13982 }, { "epoch": 0.9666781887314206, "grad_norm": 0.9156680960602557, "learning_rate": 5.5123980671813995e-06, "loss": 0.4417, "step": 13983 }, { "epoch": 0.9667473211199447, "grad_norm": 0.6918377028354858, "learning_rate": 5.511841251823877e-06, "loss": 0.2544, "step": 13984 }, { "epoch": 0.9668164535084687, "grad_norm": 0.7874016365263866, "learning_rate": 5.51128443005134e-06, "loss": 0.3034, "step": 13985 }, { "epoch": 0.9668855858969927, "grad_norm": 1.970333514846087, "learning_rate": 5.5107276018707635e-06, "loss": 0.3009, "step": 13986 }, { "epoch": 0.9669547182855167, "grad_norm": 0.6720219818571824, "learning_rate": 5.510170767289128e-06, "loss": 0.3155, "step": 13987 }, { "epoch": 0.9670238506740408, "grad_norm": 0.6000558824425541, "learning_rate": 5.509613926313413e-06, "loss": 0.2694, "step": 13988 }, { "epoch": 0.9670929830625649, "grad_norm": 0.7832819429730815, "learning_rate": 5.509057078950597e-06, "loss": 0.2744, "step": 13989 }, { "epoch": 0.9671621154510889, "grad_norm": 0.7016691537289049, "learning_rate": 5.508500225207658e-06, "loss": 0.3644, "step": 13990 }, { "epoch": 0.9672312478396129, "grad_norm": 0.7013663775553318, "learning_rate": 5.507943365091576e-06, "loss": 0.33, "step": 13991 }, { "epoch": 0.9673003802281369, "grad_norm": 0.6766975813123678, "learning_rate": 5.507386498609332e-06, "loss": 0.2579, "step": 13992 }, { "epoch": 0.9673695126166609, "grad_norm": 0.722082666714868, "learning_rate": 5.506829625767902e-06, "loss": 0.3092, "step": 13993 }, { "epoch": 0.967438645005185, "grad_norm": 0.7834345374073493, "learning_rate": 5.506272746574267e-06, "loss": 0.3974, "step": 13994 }, { "epoch": 0.967507777393709, "grad_norm": 0.7129221980854324, "learning_rate": 5.505715861035408e-06, "loss": 0.2607, "step": 13995 }, { "epoch": 0.967576909782233, "grad_norm": 0.725076142969156, "learning_rate": 5.505158969158303e-06, "loss": 0.2496, "step": 13996 }, { "epoch": 0.967646042170757, "grad_norm": 0.6912346119038139, "learning_rate": 5.504602070949931e-06, "loss": 0.3495, "step": 13997 }, { "epoch": 0.967715174559281, "grad_norm": 0.7012630159826885, "learning_rate": 5.504045166417272e-06, "loss": 0.3755, "step": 13998 }, { "epoch": 0.967784306947805, "grad_norm": 0.6482501133771831, "learning_rate": 5.503488255567309e-06, "loss": 0.2813, "step": 13999 }, { "epoch": 0.9678534393363291, "grad_norm": 0.8096231632900924, "learning_rate": 5.502931338407017e-06, "loss": 0.3068, "step": 14000 }, { "epoch": 0.9679225717248531, "grad_norm": 0.7088973843485668, "learning_rate": 5.502374414943378e-06, "loss": 0.4031, "step": 14001 }, { "epoch": 0.9679917041133771, "grad_norm": 0.6694347792637414, "learning_rate": 5.5018174851833735e-06, "loss": 0.2806, "step": 14002 }, { "epoch": 0.9680608365019011, "grad_norm": 0.6740306264310615, "learning_rate": 5.501260549133983e-06, "loss": 0.3484, "step": 14003 }, { "epoch": 0.9681299688904251, "grad_norm": 0.7718593087904669, "learning_rate": 5.5007036068021845e-06, "loss": 0.346, "step": 14004 }, { "epoch": 0.9681991012789491, "grad_norm": 1.0399720558475924, "learning_rate": 5.5001466581949604e-06, "loss": 0.3969, "step": 14005 }, { "epoch": 0.9682682336674732, "grad_norm": 0.8658633649410706, "learning_rate": 5.49958970331929e-06, "loss": 0.2408, "step": 14006 }, { "epoch": 0.9683373660559972, "grad_norm": 0.8554125130433736, "learning_rate": 5.499032742182155e-06, "loss": 0.3263, "step": 14007 }, { "epoch": 0.9684064984445213, "grad_norm": 0.6716678002668989, "learning_rate": 5.498475774790534e-06, "loss": 0.355, "step": 14008 }, { "epoch": 0.9684756308330453, "grad_norm": 4.429257582497999, "learning_rate": 5.49791880115141e-06, "loss": 0.2431, "step": 14009 }, { "epoch": 0.9685447632215693, "grad_norm": 0.7308254671585692, "learning_rate": 5.497361821271762e-06, "loss": 0.2431, "step": 14010 }, { "epoch": 0.9686138956100934, "grad_norm": 0.6981495925431853, "learning_rate": 5.49680483515857e-06, "loss": 0.3242, "step": 14011 }, { "epoch": 0.9686830279986174, "grad_norm": 0.651834945035525, "learning_rate": 5.496247842818817e-06, "loss": 0.3695, "step": 14012 }, { "epoch": 0.9687521603871414, "grad_norm": 0.6345400437035861, "learning_rate": 5.4956908442594835e-06, "loss": 0.2877, "step": 14013 }, { "epoch": 0.9688212927756654, "grad_norm": 0.6094053596555564, "learning_rate": 5.495133839487549e-06, "loss": 0.2886, "step": 14014 }, { "epoch": 0.9688904251641894, "grad_norm": 0.6632973548594107, "learning_rate": 5.494576828509995e-06, "loss": 0.3721, "step": 14015 }, { "epoch": 0.9689595575527135, "grad_norm": 0.6290823619789939, "learning_rate": 5.494019811333804e-06, "loss": 0.2677, "step": 14016 }, { "epoch": 0.9690286899412375, "grad_norm": 0.6204120063042663, "learning_rate": 5.493462787965956e-06, "loss": 0.3235, "step": 14017 }, { "epoch": 0.9690978223297615, "grad_norm": 0.6459337595387302, "learning_rate": 5.492905758413433e-06, "loss": 0.2899, "step": 14018 }, { "epoch": 0.9691669547182855, "grad_norm": 1.2758502925358641, "learning_rate": 5.492348722683215e-06, "loss": 0.4245, "step": 14019 }, { "epoch": 0.9692360871068095, "grad_norm": 0.7080549741541551, "learning_rate": 5.491791680782285e-06, "loss": 0.3202, "step": 14020 }, { "epoch": 0.9693052194953335, "grad_norm": 1.3172484359397711, "learning_rate": 5.491234632717624e-06, "loss": 0.2484, "step": 14021 }, { "epoch": 0.9693743518838576, "grad_norm": 0.6755895275459781, "learning_rate": 5.490677578496213e-06, "loss": 0.397, "step": 14022 }, { "epoch": 0.9694434842723816, "grad_norm": 0.6623919045249805, "learning_rate": 5.490120518125035e-06, "loss": 0.3612, "step": 14023 }, { "epoch": 0.9695126166609056, "grad_norm": 0.7066667378686873, "learning_rate": 5.48956345161107e-06, "loss": 0.2533, "step": 14024 }, { "epoch": 0.9695817490494296, "grad_norm": 0.6698739509339854, "learning_rate": 5.489006378961302e-06, "loss": 0.2935, "step": 14025 }, { "epoch": 0.9696508814379536, "grad_norm": 0.7254614192053009, "learning_rate": 5.48844930018271e-06, "loss": 0.4045, "step": 14026 }, { "epoch": 0.9697200138264777, "grad_norm": 0.7296137125823601, "learning_rate": 5.487892215282279e-06, "loss": 0.2624, "step": 14027 }, { "epoch": 0.9697891462150017, "grad_norm": 0.6183693588404302, "learning_rate": 5.487335124266991e-06, "loss": 0.2411, "step": 14028 }, { "epoch": 0.9698582786035258, "grad_norm": 0.6779418171283061, "learning_rate": 5.486778027143823e-06, "loss": 0.3752, "step": 14029 }, { "epoch": 0.9699274109920498, "grad_norm": 0.8129037923124094, "learning_rate": 5.486220923919764e-06, "loss": 0.3902, "step": 14030 }, { "epoch": 0.9699965433805738, "grad_norm": 0.6051076749749102, "learning_rate": 5.485663814601794e-06, "loss": 0.268, "step": 14031 }, { "epoch": 0.9700656757690979, "grad_norm": 0.6596153227840662, "learning_rate": 5.485106699196894e-06, "loss": 0.3093, "step": 14032 }, { "epoch": 0.9701348081576219, "grad_norm": 0.6445759253341088, "learning_rate": 5.484549577712046e-06, "loss": 0.3635, "step": 14033 }, { "epoch": 0.9702039405461459, "grad_norm": 0.6105478567660306, "learning_rate": 5.483992450154235e-06, "loss": 0.3103, "step": 14034 }, { "epoch": 0.9702730729346699, "grad_norm": 0.6267078487484575, "learning_rate": 5.483435316530442e-06, "loss": 0.2302, "step": 14035 }, { "epoch": 0.9703422053231939, "grad_norm": 0.6930004500043778, "learning_rate": 5.482878176847649e-06, "loss": 0.3108, "step": 14036 }, { "epoch": 0.970411337711718, "grad_norm": 0.7933995153043796, "learning_rate": 5.4823210311128415e-06, "loss": 0.4449, "step": 14037 }, { "epoch": 0.970480470100242, "grad_norm": 0.6641669062072499, "learning_rate": 5.4817638793330005e-06, "loss": 0.2694, "step": 14038 }, { "epoch": 0.970549602488766, "grad_norm": 0.5988690877667511, "learning_rate": 5.481206721515109e-06, "loss": 0.294, "step": 14039 }, { "epoch": 0.97061873487729, "grad_norm": 0.7267143556370308, "learning_rate": 5.480649557666149e-06, "loss": 0.3873, "step": 14040 }, { "epoch": 0.970687867265814, "grad_norm": 0.6454480251891359, "learning_rate": 5.480092387793105e-06, "loss": 0.2315, "step": 14041 }, { "epoch": 0.970756999654338, "grad_norm": 0.7492492143009757, "learning_rate": 5.47953521190296e-06, "loss": 0.2131, "step": 14042 }, { "epoch": 0.9708261320428621, "grad_norm": 0.6842445648925515, "learning_rate": 5.4789780300026975e-06, "loss": 0.346, "step": 14043 }, { "epoch": 0.9708952644313861, "grad_norm": 0.6863125651518576, "learning_rate": 5.478420842099298e-06, "loss": 0.3824, "step": 14044 }, { "epoch": 0.9709643968199101, "grad_norm": 0.5928661034394291, "learning_rate": 5.477863648199748e-06, "loss": 0.2578, "step": 14045 }, { "epoch": 0.9710335292084341, "grad_norm": 0.6290618865475239, "learning_rate": 5.477306448311032e-06, "loss": 0.233, "step": 14046 }, { "epoch": 0.9711026615969581, "grad_norm": 0.6724123459316289, "learning_rate": 5.476749242440128e-06, "loss": 0.4065, "step": 14047 }, { "epoch": 0.9711717939854821, "grad_norm": 0.6288493924344987, "learning_rate": 5.476192030594025e-06, "loss": 0.3244, "step": 14048 }, { "epoch": 0.9712409263740063, "grad_norm": 0.8662436110356997, "learning_rate": 5.475634812779705e-06, "loss": 0.3084, "step": 14049 }, { "epoch": 0.9713100587625303, "grad_norm": 0.7506020533740667, "learning_rate": 5.4750775890041515e-06, "loss": 0.3159, "step": 14050 }, { "epoch": 0.9713791911510543, "grad_norm": 0.6265145166286128, "learning_rate": 5.474520359274347e-06, "loss": 0.4091, "step": 14051 }, { "epoch": 0.9714483235395783, "grad_norm": 0.6328215441634071, "learning_rate": 5.4739631235972766e-06, "loss": 0.2456, "step": 14052 }, { "epoch": 0.9715174559281023, "grad_norm": 0.6172776742306743, "learning_rate": 5.473405881979926e-06, "loss": 0.2574, "step": 14053 }, { "epoch": 0.9715865883166264, "grad_norm": 0.6693605073478812, "learning_rate": 5.472848634429275e-06, "loss": 0.3554, "step": 14054 }, { "epoch": 0.9716557207051504, "grad_norm": 0.9199321515061527, "learning_rate": 5.472291380952313e-06, "loss": 0.4178, "step": 14055 }, { "epoch": 0.9717248530936744, "grad_norm": 0.6827042984013593, "learning_rate": 5.4717341215560204e-06, "loss": 0.238, "step": 14056 }, { "epoch": 0.9717939854821984, "grad_norm": 0.6966356037549094, "learning_rate": 5.471176856247382e-06, "loss": 0.3358, "step": 14057 }, { "epoch": 0.9718631178707224, "grad_norm": 0.7551644085905933, "learning_rate": 5.470619585033382e-06, "loss": 0.3668, "step": 14058 }, { "epoch": 0.9719322502592465, "grad_norm": 0.6833796940758398, "learning_rate": 5.470062307921006e-06, "loss": 0.3056, "step": 14059 }, { "epoch": 0.9720013826477705, "grad_norm": 0.7522973522558144, "learning_rate": 5.469505024917239e-06, "loss": 0.3379, "step": 14060 }, { "epoch": 0.9720705150362945, "grad_norm": 0.8082487724237444, "learning_rate": 5.468947736029063e-06, "loss": 0.3331, "step": 14061 }, { "epoch": 0.9721396474248185, "grad_norm": 0.7885905183241501, "learning_rate": 5.468390441263463e-06, "loss": 0.3764, "step": 14062 }, { "epoch": 0.9722087798133425, "grad_norm": 0.6438533382680914, "learning_rate": 5.467833140627427e-06, "loss": 0.2868, "step": 14063 }, { "epoch": 0.9722779122018665, "grad_norm": 0.783000353528076, "learning_rate": 5.467275834127936e-06, "loss": 0.2822, "step": 14064 }, { "epoch": 0.9723470445903906, "grad_norm": 0.7655846625181263, "learning_rate": 5.466718521771977e-06, "loss": 0.3596, "step": 14065 }, { "epoch": 0.9724161769789146, "grad_norm": 0.6314814320092421, "learning_rate": 5.466161203566533e-06, "loss": 0.3145, "step": 14066 }, { "epoch": 0.9724853093674386, "grad_norm": 0.6859234440645797, "learning_rate": 5.465603879518592e-06, "loss": 0.3047, "step": 14067 }, { "epoch": 0.9725544417559626, "grad_norm": 0.6655695562979796, "learning_rate": 5.465046549635137e-06, "loss": 0.2946, "step": 14068 }, { "epoch": 0.9726235741444867, "grad_norm": 0.7021694657175165, "learning_rate": 5.464489213923152e-06, "loss": 0.3673, "step": 14069 }, { "epoch": 0.9726927065330108, "grad_norm": 0.9265233920654846, "learning_rate": 5.463931872389625e-06, "loss": 0.2809, "step": 14070 }, { "epoch": 0.9727618389215348, "grad_norm": 0.6334404023507376, "learning_rate": 5.46337452504154e-06, "loss": 0.3004, "step": 14071 }, { "epoch": 0.9728309713100588, "grad_norm": 0.7182105017399872, "learning_rate": 5.462817171885881e-06, "loss": 0.4382, "step": 14072 }, { "epoch": 0.9729001036985828, "grad_norm": 0.7478299046601783, "learning_rate": 5.462259812929635e-06, "loss": 0.3219, "step": 14073 }, { "epoch": 0.9729692360871068, "grad_norm": 0.6609598578155852, "learning_rate": 5.461702448179788e-06, "loss": 0.2624, "step": 14074 }, { "epoch": 0.9730383684756309, "grad_norm": 0.6759919473859155, "learning_rate": 5.461145077643324e-06, "loss": 0.2656, "step": 14075 }, { "epoch": 0.9731075008641549, "grad_norm": 0.782687331495926, "learning_rate": 5.460587701327229e-06, "loss": 0.375, "step": 14076 }, { "epoch": 0.9731766332526789, "grad_norm": 0.6657552678755903, "learning_rate": 5.46003031923849e-06, "loss": 0.3072, "step": 14077 }, { "epoch": 0.9732457656412029, "grad_norm": 0.6338177927836403, "learning_rate": 5.459472931384092e-06, "loss": 0.2971, "step": 14078 }, { "epoch": 0.9733148980297269, "grad_norm": 0.6888434573539938, "learning_rate": 5.4589155377710205e-06, "loss": 0.3765, "step": 14079 }, { "epoch": 0.973384030418251, "grad_norm": 0.7259874648273095, "learning_rate": 5.458358138406261e-06, "loss": 0.5018, "step": 14080 }, { "epoch": 0.973453162806775, "grad_norm": 0.6739684877706472, "learning_rate": 5.457800733296802e-06, "loss": 0.2881, "step": 14081 }, { "epoch": 0.973522295195299, "grad_norm": 0.682894911808198, "learning_rate": 5.457243322449626e-06, "loss": 0.2673, "step": 14082 }, { "epoch": 0.973591427583823, "grad_norm": 0.7172913161557258, "learning_rate": 5.456685905871721e-06, "loss": 0.409, "step": 14083 }, { "epoch": 0.973660559972347, "grad_norm": 0.6994175113002795, "learning_rate": 5.456128483570075e-06, "loss": 0.3251, "step": 14084 }, { "epoch": 0.973729692360871, "grad_norm": 0.6169369583790327, "learning_rate": 5.4555710555516705e-06, "loss": 0.2957, "step": 14085 }, { "epoch": 0.973798824749395, "grad_norm": 0.6484051882450824, "learning_rate": 5.455013621823496e-06, "loss": 0.2988, "step": 14086 }, { "epoch": 0.9738679571379191, "grad_norm": 0.6376109274108366, "learning_rate": 5.454456182392539e-06, "loss": 0.3913, "step": 14087 }, { "epoch": 0.9739370895264431, "grad_norm": 1.1679711000500608, "learning_rate": 5.453898737265784e-06, "loss": 0.2802, "step": 14088 }, { "epoch": 0.9740062219149671, "grad_norm": 0.6730628094372059, "learning_rate": 5.453341286450219e-06, "loss": 0.2836, "step": 14089 }, { "epoch": 0.9740753543034912, "grad_norm": 0.7397335582528777, "learning_rate": 5.452783829952828e-06, "loss": 0.4295, "step": 14090 }, { "epoch": 0.9741444866920153, "grad_norm": 0.6600284461713392, "learning_rate": 5.452226367780602e-06, "loss": 0.3003, "step": 14091 }, { "epoch": 0.9742136190805393, "grad_norm": 0.7171993568657072, "learning_rate": 5.451668899940526e-06, "loss": 0.242, "step": 14092 }, { "epoch": 0.9742827514690633, "grad_norm": 0.6435639938069557, "learning_rate": 5.451111426439583e-06, "loss": 0.3142, "step": 14093 }, { "epoch": 0.9743518838575873, "grad_norm": 0.8114865432485758, "learning_rate": 5.4505539472847656e-06, "loss": 0.381, "step": 14094 }, { "epoch": 0.9744210162461113, "grad_norm": 0.7301037271606349, "learning_rate": 5.449996462483058e-06, "loss": 0.2798, "step": 14095 }, { "epoch": 0.9744901486346353, "grad_norm": 0.7182109845392283, "learning_rate": 5.449438972041449e-06, "loss": 0.3262, "step": 14096 }, { "epoch": 0.9745592810231594, "grad_norm": 0.7888380195759959, "learning_rate": 5.448881475966923e-06, "loss": 0.3742, "step": 14097 }, { "epoch": 0.9746284134116834, "grad_norm": 0.7704512177264637, "learning_rate": 5.44832397426647e-06, "loss": 0.3471, "step": 14098 }, { "epoch": 0.9746975458002074, "grad_norm": 0.6493611690858172, "learning_rate": 5.447766466947075e-06, "loss": 0.2675, "step": 14099 }, { "epoch": 0.9747666781887314, "grad_norm": 0.6376047050579882, "learning_rate": 5.447208954015726e-06, "loss": 0.3378, "step": 14100 }, { "epoch": 0.9748358105772554, "grad_norm": 0.6917776641674506, "learning_rate": 5.446651435479412e-06, "loss": 0.4095, "step": 14101 }, { "epoch": 0.9749049429657795, "grad_norm": 0.628365529508303, "learning_rate": 5.446093911345119e-06, "loss": 0.2258, "step": 14102 }, { "epoch": 0.9749740753543035, "grad_norm": 0.6697835493382691, "learning_rate": 5.445536381619836e-06, "loss": 0.2515, "step": 14103 }, { "epoch": 0.9750432077428275, "grad_norm": 0.8756133459112915, "learning_rate": 5.444978846310547e-06, "loss": 0.4343, "step": 14104 }, { "epoch": 0.9751123401313515, "grad_norm": 0.7389077835638008, "learning_rate": 5.4444213054242445e-06, "loss": 0.4177, "step": 14105 }, { "epoch": 0.9751814725198755, "grad_norm": 0.7137074559711151, "learning_rate": 5.4438637589679136e-06, "loss": 0.2624, "step": 14106 }, { "epoch": 0.9752506049083995, "grad_norm": 0.70365408450432, "learning_rate": 5.443306206948541e-06, "loss": 0.2857, "step": 14107 }, { "epoch": 0.9753197372969236, "grad_norm": 0.6611667156819264, "learning_rate": 5.442748649373118e-06, "loss": 0.3357, "step": 14108 }, { "epoch": 0.9753888696854476, "grad_norm": 0.6729161586902265, "learning_rate": 5.442191086248631e-06, "loss": 0.3068, "step": 14109 }, { "epoch": 0.9754580020739717, "grad_norm": 0.6662556714360044, "learning_rate": 5.441633517582067e-06, "loss": 0.3032, "step": 14110 }, { "epoch": 0.9755271344624957, "grad_norm": 0.6745694221455552, "learning_rate": 5.4410759433804135e-06, "loss": 0.3313, "step": 14111 }, { "epoch": 0.9755962668510197, "grad_norm": 0.7307438088631363, "learning_rate": 5.440518363650663e-06, "loss": 0.3421, "step": 14112 }, { "epoch": 0.9756653992395438, "grad_norm": 0.5836917438985333, "learning_rate": 5.4399607783998e-06, "loss": 0.2358, "step": 14113 }, { "epoch": 0.9757345316280678, "grad_norm": 0.6585982910729499, "learning_rate": 5.4394031876348145e-06, "loss": 0.3054, "step": 14114 }, { "epoch": 0.9758036640165918, "grad_norm": 0.7285232539317829, "learning_rate": 5.438845591362691e-06, "loss": 0.373, "step": 14115 }, { "epoch": 0.9758727964051158, "grad_norm": 0.894001516111619, "learning_rate": 5.438287989590425e-06, "loss": 0.2788, "step": 14116 }, { "epoch": 0.9759419287936398, "grad_norm": 0.595707647402545, "learning_rate": 5.437730382325e-06, "loss": 0.2723, "step": 14117 }, { "epoch": 0.9760110611821639, "grad_norm": 0.6490205054968649, "learning_rate": 5.437172769573404e-06, "loss": 0.3447, "step": 14118 }, { "epoch": 0.9760801935706879, "grad_norm": 0.659829420484087, "learning_rate": 5.436615151342629e-06, "loss": 0.3767, "step": 14119 }, { "epoch": 0.9761493259592119, "grad_norm": 0.6606652929199662, "learning_rate": 5.436057527639663e-06, "loss": 0.2242, "step": 14120 }, { "epoch": 0.9762184583477359, "grad_norm": 0.6756521725165848, "learning_rate": 5.435499898471493e-06, "loss": 0.3158, "step": 14121 }, { "epoch": 0.9762875907362599, "grad_norm": 0.6354627405441258, "learning_rate": 5.4349422638451084e-06, "loss": 0.3908, "step": 14122 }, { "epoch": 0.9763567231247839, "grad_norm": 0.6728244006435247, "learning_rate": 5.434384623767501e-06, "loss": 0.3356, "step": 14123 }, { "epoch": 0.976425855513308, "grad_norm": 0.6967161965575572, "learning_rate": 5.4338269782456554e-06, "loss": 0.2797, "step": 14124 }, { "epoch": 0.976494987901832, "grad_norm": 0.7651948266272675, "learning_rate": 5.4332693272865635e-06, "loss": 0.3226, "step": 14125 }, { "epoch": 0.976564120290356, "grad_norm": 0.6934143626234076, "learning_rate": 5.432711670897214e-06, "loss": 0.4128, "step": 14126 }, { "epoch": 0.97663325267888, "grad_norm": 0.6201390245669233, "learning_rate": 5.4321540090845955e-06, "loss": 0.2557, "step": 14127 }, { "epoch": 0.976702385067404, "grad_norm": 0.6730973110688984, "learning_rate": 5.431596341855698e-06, "loss": 0.3468, "step": 14128 }, { "epoch": 0.976771517455928, "grad_norm": 0.9365692484604561, "learning_rate": 5.43103866921751e-06, "loss": 0.3601, "step": 14129 }, { "epoch": 0.9768406498444522, "grad_norm": 0.6803629084520448, "learning_rate": 5.4304809911770215e-06, "loss": 0.3973, "step": 14130 }, { "epoch": 0.9769097822329762, "grad_norm": 0.6880674574305128, "learning_rate": 5.4299233077412235e-06, "loss": 0.3292, "step": 14131 }, { "epoch": 0.9769789146215002, "grad_norm": 0.8148895119055218, "learning_rate": 5.4293656189171025e-06, "loss": 0.3079, "step": 14132 }, { "epoch": 0.9770480470100242, "grad_norm": 0.6445410631981879, "learning_rate": 5.42880792471165e-06, "loss": 0.3203, "step": 14133 }, { "epoch": 0.9771171793985483, "grad_norm": 0.7043053353691247, "learning_rate": 5.4282502251318554e-06, "loss": 0.3171, "step": 14134 }, { "epoch": 0.9771863117870723, "grad_norm": 0.5874402305069353, "learning_rate": 5.42769252018471e-06, "loss": 0.2244, "step": 14135 }, { "epoch": 0.9772554441755963, "grad_norm": 0.7406423483297637, "learning_rate": 5.427134809877201e-06, "loss": 0.3371, "step": 14136 }, { "epoch": 0.9773245765641203, "grad_norm": 0.7013922642105723, "learning_rate": 5.4265770942163195e-06, "loss": 0.4103, "step": 14137 }, { "epoch": 0.9773937089526443, "grad_norm": 0.7403083652285589, "learning_rate": 5.426019373209056e-06, "loss": 0.2526, "step": 14138 }, { "epoch": 0.9774628413411683, "grad_norm": 0.7156435676810096, "learning_rate": 5.425461646862398e-06, "loss": 0.2957, "step": 14139 }, { "epoch": 0.9775319737296924, "grad_norm": 0.6519305492873957, "learning_rate": 5.424903915183338e-06, "loss": 0.3806, "step": 14140 }, { "epoch": 0.9776011061182164, "grad_norm": 0.657509172715458, "learning_rate": 5.424346178178867e-06, "loss": 0.2974, "step": 14141 }, { "epoch": 0.9776702385067404, "grad_norm": 0.6583544390720991, "learning_rate": 5.4237884358559735e-06, "loss": 0.2067, "step": 14142 }, { "epoch": 0.9777393708952644, "grad_norm": 0.7145200795941814, "learning_rate": 5.423230688221648e-06, "loss": 0.3432, "step": 14143 }, { "epoch": 0.9778085032837884, "grad_norm": 0.6846126649673441, "learning_rate": 5.4226729352828814e-06, "loss": 0.3598, "step": 14144 }, { "epoch": 0.9778776356723125, "grad_norm": 0.6492500120722688, "learning_rate": 5.422115177046663e-06, "loss": 0.2416, "step": 14145 }, { "epoch": 0.9779467680608365, "grad_norm": 0.5930142974030732, "learning_rate": 5.4215574135199845e-06, "loss": 0.2518, "step": 14146 }, { "epoch": 0.9780159004493605, "grad_norm": 0.6610348387894313, "learning_rate": 5.420999644709837e-06, "loss": 0.3465, "step": 14147 }, { "epoch": 0.9780850328378845, "grad_norm": 0.5814205226444751, "learning_rate": 5.42044187062321e-06, "loss": 0.2734, "step": 14148 }, { "epoch": 0.9781541652264085, "grad_norm": 0.5648602059653568, "learning_rate": 5.419884091267094e-06, "loss": 0.2594, "step": 14149 }, { "epoch": 0.9782232976149325, "grad_norm": 0.6782671113502367, "learning_rate": 5.419326306648479e-06, "loss": 0.2764, "step": 14150 }, { "epoch": 0.9782924300034567, "grad_norm": 0.5829123302846028, "learning_rate": 5.418768516774359e-06, "loss": 0.3648, "step": 14151 }, { "epoch": 0.9783615623919807, "grad_norm": 0.6789104041526491, "learning_rate": 5.418210721651722e-06, "loss": 0.3085, "step": 14152 }, { "epoch": 0.9784306947805047, "grad_norm": 0.6381461534293846, "learning_rate": 5.41765292128756e-06, "loss": 0.2511, "step": 14153 }, { "epoch": 0.9784998271690287, "grad_norm": 0.6738236049841682, "learning_rate": 5.417095115688864e-06, "loss": 0.3699, "step": 14154 }, { "epoch": 0.9785689595575527, "grad_norm": 0.6326525728557382, "learning_rate": 5.4165373048626255e-06, "loss": 0.3487, "step": 14155 }, { "epoch": 0.9786380919460768, "grad_norm": 0.645474960105585, "learning_rate": 5.4159794888158355e-06, "loss": 0.3066, "step": 14156 }, { "epoch": 0.9787072243346008, "grad_norm": 0.652621576454869, "learning_rate": 5.415421667555484e-06, "loss": 0.2656, "step": 14157 }, { "epoch": 0.9787763567231248, "grad_norm": 1.5402020029277448, "learning_rate": 5.414863841088563e-06, "loss": 0.3289, "step": 14158 }, { "epoch": 0.9788454891116488, "grad_norm": 0.9768771227954178, "learning_rate": 5.414306009422066e-06, "loss": 0.2977, "step": 14159 }, { "epoch": 0.9789146215001728, "grad_norm": 0.6491564599905979, "learning_rate": 5.4137481725629815e-06, "loss": 0.2197, "step": 14160 }, { "epoch": 0.9789837538886969, "grad_norm": 0.6527477733383381, "learning_rate": 5.413190330518301e-06, "loss": 0.3233, "step": 14161 }, { "epoch": 0.9790528862772209, "grad_norm": 0.6587969504992826, "learning_rate": 5.412632483295018e-06, "loss": 0.3819, "step": 14162 }, { "epoch": 0.9791220186657449, "grad_norm": 0.691997452125518, "learning_rate": 5.412074630900124e-06, "loss": 0.2805, "step": 14163 }, { "epoch": 0.9791911510542689, "grad_norm": 0.6289426659786177, "learning_rate": 5.411516773340608e-06, "loss": 0.2621, "step": 14164 }, { "epoch": 0.9792602834427929, "grad_norm": 0.662073556923559, "learning_rate": 5.410958910623465e-06, "loss": 0.4196, "step": 14165 }, { "epoch": 0.9793294158313169, "grad_norm": 0.6354616802565913, "learning_rate": 5.410401042755686e-06, "loss": 0.3488, "step": 14166 }, { "epoch": 0.979398548219841, "grad_norm": 0.5810787257524529, "learning_rate": 5.409843169744261e-06, "loss": 0.291, "step": 14167 }, { "epoch": 0.979467680608365, "grad_norm": 0.6852091794567746, "learning_rate": 5.4092852915961834e-06, "loss": 0.3058, "step": 14168 }, { "epoch": 0.979536812996889, "grad_norm": 0.6608868702639452, "learning_rate": 5.408727408318446e-06, "loss": 0.3535, "step": 14169 }, { "epoch": 0.979605945385413, "grad_norm": 0.6901118705575657, "learning_rate": 5.408169519918039e-06, "loss": 0.2676, "step": 14170 }, { "epoch": 0.9796750777739371, "grad_norm": 0.6161828322964139, "learning_rate": 5.407611626401955e-06, "loss": 0.2997, "step": 14171 }, { "epoch": 0.9797442101624612, "grad_norm": 0.6813717990414548, "learning_rate": 5.407053727777188e-06, "loss": 0.3794, "step": 14172 }, { "epoch": 0.9798133425509852, "grad_norm": 0.6891724790373319, "learning_rate": 5.406495824050728e-06, "loss": 0.3319, "step": 14173 }, { "epoch": 0.9798824749395092, "grad_norm": 0.6312380220665593, "learning_rate": 5.40593791522957e-06, "loss": 0.2073, "step": 14174 }, { "epoch": 0.9799516073280332, "grad_norm": 0.6939109639262373, "learning_rate": 5.4053800013207025e-06, "loss": 0.3579, "step": 14175 }, { "epoch": 0.9800207397165572, "grad_norm": 0.5996481154625476, "learning_rate": 5.40482208233112e-06, "loss": 0.3349, "step": 14176 }, { "epoch": 0.9800898721050813, "grad_norm": 0.6489745597734538, "learning_rate": 5.4042641582678164e-06, "loss": 0.2692, "step": 14177 }, { "epoch": 0.9801590044936053, "grad_norm": 2.5705682200204234, "learning_rate": 5.403706229137783e-06, "loss": 0.2714, "step": 14178 }, { "epoch": 0.9802281368821293, "grad_norm": 0.6992134825641045, "learning_rate": 5.403148294948012e-06, "loss": 0.4045, "step": 14179 }, { "epoch": 0.9802972692706533, "grad_norm": 0.6670187828682422, "learning_rate": 5.402590355705497e-06, "loss": 0.3692, "step": 14180 }, { "epoch": 0.9803664016591773, "grad_norm": 0.6807603026535066, "learning_rate": 5.402032411417231e-06, "loss": 0.2867, "step": 14181 }, { "epoch": 0.9804355340477013, "grad_norm": 0.6651868042849377, "learning_rate": 5.401474462090204e-06, "loss": 0.2482, "step": 14182 }, { "epoch": 0.9805046664362254, "grad_norm": 0.6675104235270373, "learning_rate": 5.400916507731413e-06, "loss": 0.3754, "step": 14183 }, { "epoch": 0.9805737988247494, "grad_norm": 0.6338189222181227, "learning_rate": 5.4003585483478485e-06, "loss": 0.2912, "step": 14184 }, { "epoch": 0.9806429312132734, "grad_norm": 0.6302355699319266, "learning_rate": 5.399800583946504e-06, "loss": 0.2874, "step": 14185 }, { "epoch": 0.9807120636017974, "grad_norm": 0.6676712062105118, "learning_rate": 5.3992426145343725e-06, "loss": 0.3241, "step": 14186 }, { "epoch": 0.9807811959903214, "grad_norm": 0.6998567400390509, "learning_rate": 5.398684640118448e-06, "loss": 0.413, "step": 14187 }, { "epoch": 0.9808503283788454, "grad_norm": 0.6748581013882652, "learning_rate": 5.398126660705722e-06, "loss": 0.3151, "step": 14188 }, { "epoch": 0.9809194607673695, "grad_norm": 0.6979753524165686, "learning_rate": 5.397568676303189e-06, "loss": 0.3424, "step": 14189 }, { "epoch": 0.9809885931558935, "grad_norm": 1.116109551825577, "learning_rate": 5.3970106869178425e-06, "loss": 0.3847, "step": 14190 }, { "epoch": 0.9810577255444176, "grad_norm": 0.6579288588873872, "learning_rate": 5.396452692556675e-06, "loss": 0.2779, "step": 14191 }, { "epoch": 0.9811268579329416, "grad_norm": 0.6999705041520425, "learning_rate": 5.395894693226682e-06, "loss": 0.3225, "step": 14192 }, { "epoch": 0.9811959903214656, "grad_norm": 0.6902173649266475, "learning_rate": 5.395336688934853e-06, "loss": 0.3583, "step": 14193 }, { "epoch": 0.9812651227099897, "grad_norm": 0.6817917495065794, "learning_rate": 5.394778679688186e-06, "loss": 0.391, "step": 14194 }, { "epoch": 0.9813342550985137, "grad_norm": 0.7427318126613173, "learning_rate": 5.3942206654936735e-06, "loss": 0.3304, "step": 14195 }, { "epoch": 0.9814033874870377, "grad_norm": 0.7494950705611818, "learning_rate": 5.393662646358307e-06, "loss": 0.2749, "step": 14196 }, { "epoch": 0.9814725198755617, "grad_norm": 0.6516825758886816, "learning_rate": 5.3931046222890815e-06, "loss": 0.3611, "step": 14197 }, { "epoch": 0.9815416522640857, "grad_norm": 0.6614745866571697, "learning_rate": 5.3925465932929925e-06, "loss": 0.3597, "step": 14198 }, { "epoch": 0.9816107846526098, "grad_norm": 0.6480283121613715, "learning_rate": 5.391988559377031e-06, "loss": 0.3139, "step": 14199 }, { "epoch": 0.9816799170411338, "grad_norm": 0.6814105298776268, "learning_rate": 5.3914305205481924e-06, "loss": 0.3804, "step": 14200 }, { "epoch": 0.9817490494296578, "grad_norm": 0.6390612977629849, "learning_rate": 5.390872476813473e-06, "loss": 0.413, "step": 14201 }, { "epoch": 0.9818181818181818, "grad_norm": 0.6803000481526325, "learning_rate": 5.390314428179863e-06, "loss": 0.2668, "step": 14202 }, { "epoch": 0.9818873142067058, "grad_norm": 0.6185635940078481, "learning_rate": 5.389756374654358e-06, "loss": 0.273, "step": 14203 }, { "epoch": 0.9819564465952298, "grad_norm": 0.6857839597797314, "learning_rate": 5.389198316243953e-06, "loss": 0.3694, "step": 14204 }, { "epoch": 0.9820255789837539, "grad_norm": 0.6319134460383998, "learning_rate": 5.3886402529556415e-06, "loss": 0.3326, "step": 14205 }, { "epoch": 0.9820947113722779, "grad_norm": 0.6302183538332855, "learning_rate": 5.388082184796417e-06, "loss": 0.2873, "step": 14206 }, { "epoch": 0.9821638437608019, "grad_norm": 0.678544229093378, "learning_rate": 5.387524111773275e-06, "loss": 0.3448, "step": 14207 }, { "epoch": 0.9822329761493259, "grad_norm": 0.6630940754443011, "learning_rate": 5.386966033893212e-06, "loss": 0.3993, "step": 14208 }, { "epoch": 0.9823021085378499, "grad_norm": 0.6725795883065672, "learning_rate": 5.386407951163219e-06, "loss": 0.2873, "step": 14209 }, { "epoch": 0.982371240926374, "grad_norm": 0.6593003150347575, "learning_rate": 5.385849863590291e-06, "loss": 0.34, "step": 14210 }, { "epoch": 0.982440373314898, "grad_norm": 0.6644439184125658, "learning_rate": 5.385291771181425e-06, "loss": 0.3147, "step": 14211 }, { "epoch": 0.9825095057034221, "grad_norm": 0.6911419000421747, "learning_rate": 5.384733673943613e-06, "loss": 0.3828, "step": 14212 }, { "epoch": 0.9825786380919461, "grad_norm": 0.6004837463387803, "learning_rate": 5.384175571883851e-06, "loss": 0.2672, "step": 14213 }, { "epoch": 0.9826477704804701, "grad_norm": 0.6766313379461797, "learning_rate": 5.383617465009134e-06, "loss": 0.2855, "step": 14214 }, { "epoch": 0.9827169028689942, "grad_norm": 0.6521639636889877, "learning_rate": 5.383059353326459e-06, "loss": 0.3391, "step": 14215 }, { "epoch": 0.9827860352575182, "grad_norm": 0.7106244022495511, "learning_rate": 5.382501236842817e-06, "loss": 0.3014, "step": 14216 }, { "epoch": 0.9828551676460422, "grad_norm": 0.5949308834913799, "learning_rate": 5.381943115565202e-06, "loss": 0.2189, "step": 14217 }, { "epoch": 0.9829243000345662, "grad_norm": 0.7996159450140675, "learning_rate": 5.381384989500615e-06, "loss": 0.3183, "step": 14218 }, { "epoch": 0.9829934324230902, "grad_norm": 0.6789551383778896, "learning_rate": 5.3808268586560465e-06, "loss": 0.3891, "step": 14219 }, { "epoch": 0.9830625648116142, "grad_norm": 0.7480050602013572, "learning_rate": 5.3802687230384935e-06, "loss": 0.3244, "step": 14220 }, { "epoch": 0.9831316972001383, "grad_norm": 0.605652089117344, "learning_rate": 5.379710582654949e-06, "loss": 0.2893, "step": 14221 }, { "epoch": 0.9832008295886623, "grad_norm": 0.6544952403346792, "learning_rate": 5.379152437512411e-06, "loss": 0.3681, "step": 14222 }, { "epoch": 0.9832699619771863, "grad_norm": 0.6855321708455824, "learning_rate": 5.3785942876178756e-06, "loss": 0.3712, "step": 14223 }, { "epoch": 0.9833390943657103, "grad_norm": 0.6209157631819181, "learning_rate": 5.3780361329783334e-06, "loss": 0.2347, "step": 14224 }, { "epoch": 0.9834082267542343, "grad_norm": 0.8649782593506704, "learning_rate": 5.3774779736007854e-06, "loss": 0.3382, "step": 14225 }, { "epoch": 0.9834773591427584, "grad_norm": 0.63554432744663, "learning_rate": 5.376919809492224e-06, "loss": 0.3725, "step": 14226 }, { "epoch": 0.9835464915312824, "grad_norm": 0.6494244798465744, "learning_rate": 5.376361640659645e-06, "loss": 0.2885, "step": 14227 }, { "epoch": 0.9836156239198064, "grad_norm": 0.627586584698512, "learning_rate": 5.375803467110043e-06, "loss": 0.275, "step": 14228 }, { "epoch": 0.9836847563083304, "grad_norm": 0.6895392911781767, "learning_rate": 5.375245288850418e-06, "loss": 0.4044, "step": 14229 }, { "epoch": 0.9837538886968544, "grad_norm": 0.9355652879383366, "learning_rate": 5.374687105887761e-06, "loss": 0.3469, "step": 14230 }, { "epoch": 0.9838230210853784, "grad_norm": 0.9774103552001543, "learning_rate": 5.374128918229072e-06, "loss": 0.2819, "step": 14231 }, { "epoch": 0.9838921534739026, "grad_norm": 0.7150293318599702, "learning_rate": 5.373570725881342e-06, "loss": 0.2998, "step": 14232 }, { "epoch": 0.9839612858624266, "grad_norm": 0.8523573407852874, "learning_rate": 5.373012528851571e-06, "loss": 0.3368, "step": 14233 }, { "epoch": 0.9840304182509506, "grad_norm": 0.6277615585387175, "learning_rate": 5.372454327146753e-06, "loss": 0.329, "step": 14234 }, { "epoch": 0.9840995506394746, "grad_norm": 0.540607688672933, "learning_rate": 5.371896120773885e-06, "loss": 0.272, "step": 14235 }, { "epoch": 0.9841686830279986, "grad_norm": 0.681031622393758, "learning_rate": 5.371337909739963e-06, "loss": 0.3271, "step": 14236 }, { "epoch": 0.9842378154165227, "grad_norm": 0.6471465090098304, "learning_rate": 5.370779694051983e-06, "loss": 0.3497, "step": 14237 }, { "epoch": 0.9843069478050467, "grad_norm": 0.631303265035346, "learning_rate": 5.3702214737169415e-06, "loss": 0.2672, "step": 14238 }, { "epoch": 0.9843760801935707, "grad_norm": 0.8647040439366126, "learning_rate": 5.369663248741833e-06, "loss": 0.2535, "step": 14239 }, { "epoch": 0.9844452125820947, "grad_norm": 0.6369873675513733, "learning_rate": 5.369105019133657e-06, "loss": 0.3859, "step": 14240 }, { "epoch": 0.9845143449706187, "grad_norm": 0.6773984412543033, "learning_rate": 5.368546784899407e-06, "loss": 0.3089, "step": 14241 }, { "epoch": 0.9845834773591428, "grad_norm": 0.6445134193453053, "learning_rate": 5.367988546046081e-06, "loss": 0.2559, "step": 14242 }, { "epoch": 0.9846526097476668, "grad_norm": 0.9430513821846145, "learning_rate": 5.367430302580677e-06, "loss": 0.3401, "step": 14243 }, { "epoch": 0.9847217421361908, "grad_norm": 0.6202035776104265, "learning_rate": 5.366872054510189e-06, "loss": 0.3763, "step": 14244 }, { "epoch": 0.9847908745247148, "grad_norm": 0.6128832937819312, "learning_rate": 5.366313801841614e-06, "loss": 0.3561, "step": 14245 }, { "epoch": 0.9848600069132388, "grad_norm": 0.7382020884713204, "learning_rate": 5.365755544581949e-06, "loss": 0.2932, "step": 14246 }, { "epoch": 0.9849291393017628, "grad_norm": 0.7040265769432694, "learning_rate": 5.365197282738192e-06, "loss": 0.4, "step": 14247 }, { "epoch": 0.9849982716902869, "grad_norm": 0.7179175560132407, "learning_rate": 5.364639016317338e-06, "loss": 0.2993, "step": 14248 }, { "epoch": 0.9850674040788109, "grad_norm": 0.659500836425933, "learning_rate": 5.364080745326385e-06, "loss": 0.3542, "step": 14249 }, { "epoch": 0.9851365364673349, "grad_norm": 0.6361964929900086, "learning_rate": 5.3635224697723285e-06, "loss": 0.2686, "step": 14250 }, { "epoch": 0.9852056688558589, "grad_norm": 0.6308124626544652, "learning_rate": 5.362964189662168e-06, "loss": 0.3334, "step": 14251 }, { "epoch": 0.985274801244383, "grad_norm": 0.6302796959237087, "learning_rate": 5.362405905002899e-06, "loss": 0.317, "step": 14252 }, { "epoch": 0.9853439336329071, "grad_norm": 0.8683930331089749, "learning_rate": 5.361847615801517e-06, "loss": 0.3192, "step": 14253 }, { "epoch": 0.9854130660214311, "grad_norm": 0.6793982523675136, "learning_rate": 5.361289322065023e-06, "loss": 0.3921, "step": 14254 }, { "epoch": 0.9854821984099551, "grad_norm": 0.7608501113340109, "learning_rate": 5.360731023800412e-06, "loss": 0.4137, "step": 14255 }, { "epoch": 0.9855513307984791, "grad_norm": 0.613432885384647, "learning_rate": 5.36017272101468e-06, "loss": 0.2958, "step": 14256 }, { "epoch": 0.9856204631870031, "grad_norm": 0.6778448756798141, "learning_rate": 5.3596144137148255e-06, "loss": 0.3234, "step": 14257 }, { "epoch": 0.9856895955755272, "grad_norm": 0.6589790812171692, "learning_rate": 5.359056101907848e-06, "loss": 0.4157, "step": 14258 }, { "epoch": 0.9857587279640512, "grad_norm": 0.7347832871136659, "learning_rate": 5.3584977856007415e-06, "loss": 0.2899, "step": 14259 }, { "epoch": 0.9858278603525752, "grad_norm": 0.7280936772722678, "learning_rate": 5.3579394648005046e-06, "loss": 0.3235, "step": 14260 }, { "epoch": 0.9858969927410992, "grad_norm": 0.6923412963841029, "learning_rate": 5.357381139514137e-06, "loss": 0.3562, "step": 14261 }, { "epoch": 0.9859661251296232, "grad_norm": 0.6400278948979894, "learning_rate": 5.356822809748634e-06, "loss": 0.3553, "step": 14262 }, { "epoch": 0.9860352575181472, "grad_norm": 0.6397905510635241, "learning_rate": 5.356264475510994e-06, "loss": 0.2992, "step": 14263 }, { "epoch": 0.9861043899066713, "grad_norm": 1.0769197703754376, "learning_rate": 5.355706136808213e-06, "loss": 0.2679, "step": 14264 }, { "epoch": 0.9861735222951953, "grad_norm": 0.6804850560525634, "learning_rate": 5.355147793647291e-06, "loss": 0.3545, "step": 14265 }, { "epoch": 0.9862426546837193, "grad_norm": 0.7076119260936489, "learning_rate": 5.354589446035227e-06, "loss": 0.2612, "step": 14266 }, { "epoch": 0.9863117870722433, "grad_norm": 0.7323021384666929, "learning_rate": 5.354031093979015e-06, "loss": 0.2665, "step": 14267 }, { "epoch": 0.9863809194607673, "grad_norm": 0.7372900845573389, "learning_rate": 5.353472737485657e-06, "loss": 0.3126, "step": 14268 }, { "epoch": 0.9864500518492914, "grad_norm": 0.6393848158588425, "learning_rate": 5.352914376562148e-06, "loss": 0.3677, "step": 14269 }, { "epoch": 0.9865191842378154, "grad_norm": 0.6184255278594231, "learning_rate": 5.352356011215487e-06, "loss": 0.2421, "step": 14270 }, { "epoch": 0.9865883166263394, "grad_norm": 0.8196430893122693, "learning_rate": 5.351797641452672e-06, "loss": 0.2429, "step": 14271 }, { "epoch": 0.9866574490148635, "grad_norm": 0.733454335130929, "learning_rate": 5.351239267280702e-06, "loss": 0.3318, "step": 14272 }, { "epoch": 0.9867265814033875, "grad_norm": 1.04364483176108, "learning_rate": 5.350680888706574e-06, "loss": 0.2836, "step": 14273 }, { "epoch": 0.9867957137919116, "grad_norm": 0.6008575943655406, "learning_rate": 5.350122505737288e-06, "loss": 0.2934, "step": 14274 }, { "epoch": 0.9868648461804356, "grad_norm": 0.6548405730187873, "learning_rate": 5.349564118379839e-06, "loss": 0.2919, "step": 14275 }, { "epoch": 0.9869339785689596, "grad_norm": 0.7304928877815902, "learning_rate": 5.34900572664123e-06, "loss": 0.4299, "step": 14276 }, { "epoch": 0.9870031109574836, "grad_norm": 0.6802753463627257, "learning_rate": 5.348447330528454e-06, "loss": 0.2477, "step": 14277 }, { "epoch": 0.9870722433460076, "grad_norm": 0.9692156671604846, "learning_rate": 5.347888930048516e-06, "loss": 0.2958, "step": 14278 }, { "epoch": 0.9871413757345316, "grad_norm": 0.7473726425853696, "learning_rate": 5.347330525208409e-06, "loss": 0.4007, "step": 14279 }, { "epoch": 0.9872105081230557, "grad_norm": 0.6497562628652616, "learning_rate": 5.346772116015135e-06, "loss": 0.384, "step": 14280 }, { "epoch": 0.9872796405115797, "grad_norm": 0.5897811022283993, "learning_rate": 5.34621370247569e-06, "loss": 0.2941, "step": 14281 }, { "epoch": 0.9873487729001037, "grad_norm": 0.6913427828554389, "learning_rate": 5.345655284597075e-06, "loss": 0.2904, "step": 14282 }, { "epoch": 0.9874179052886277, "grad_norm": 0.7010297290340718, "learning_rate": 5.345096862386287e-06, "loss": 0.4349, "step": 14283 }, { "epoch": 0.9874870376771517, "grad_norm": 0.6068165500310259, "learning_rate": 5.344538435850327e-06, "loss": 0.2679, "step": 14284 }, { "epoch": 0.9875561700656758, "grad_norm": 0.6786503908267425, "learning_rate": 5.343980004996191e-06, "loss": 0.3053, "step": 14285 }, { "epoch": 0.9876253024541998, "grad_norm": 0.6325741201405002, "learning_rate": 5.343421569830881e-06, "loss": 0.2913, "step": 14286 }, { "epoch": 0.9876944348427238, "grad_norm": 0.6690935311155781, "learning_rate": 5.342863130361394e-06, "loss": 0.4187, "step": 14287 }, { "epoch": 0.9877635672312478, "grad_norm": 0.620636983994479, "learning_rate": 5.342304686594728e-06, "loss": 0.3045, "step": 14288 }, { "epoch": 0.9878326996197718, "grad_norm": 0.6092310605323175, "learning_rate": 5.341746238537886e-06, "loss": 0.2834, "step": 14289 }, { "epoch": 0.9879018320082958, "grad_norm": 0.6658202430182362, "learning_rate": 5.341187786197864e-06, "loss": 0.3786, "step": 14290 }, { "epoch": 0.9879709643968199, "grad_norm": 0.6295646168424517, "learning_rate": 5.340629329581661e-06, "loss": 0.3345, "step": 14291 }, { "epoch": 0.9880400967853439, "grad_norm": 0.6674107576896786, "learning_rate": 5.340070868696278e-06, "loss": 0.2694, "step": 14292 }, { "epoch": 0.988109229173868, "grad_norm": 0.7678411528176536, "learning_rate": 5.3395124035487135e-06, "loss": 0.3185, "step": 14293 }, { "epoch": 0.988178361562392, "grad_norm": 0.6095918337748806, "learning_rate": 5.3389539341459675e-06, "loss": 0.3967, "step": 14294 }, { "epoch": 0.988247493950916, "grad_norm": 1.192956653836797, "learning_rate": 5.338395460495037e-06, "loss": 0.27, "step": 14295 }, { "epoch": 0.9883166263394401, "grad_norm": 1.4887377957615815, "learning_rate": 5.337836982602925e-06, "loss": 0.2558, "step": 14296 }, { "epoch": 0.9883857587279641, "grad_norm": 0.681345183661424, "learning_rate": 5.337278500476629e-06, "loss": 0.3573, "step": 14297 }, { "epoch": 0.9884548911164881, "grad_norm": 0.7118165857323732, "learning_rate": 5.336720014123149e-06, "loss": 0.3706, "step": 14298 }, { "epoch": 0.9885240235050121, "grad_norm": 0.5738881055834628, "learning_rate": 5.336161523549483e-06, "loss": 0.2979, "step": 14299 }, { "epoch": 0.9885931558935361, "grad_norm": 0.6371682557435971, "learning_rate": 5.335603028762635e-06, "loss": 0.3081, "step": 14300 }, { "epoch": 0.9886622882820602, "grad_norm": 0.5969197116685792, "learning_rate": 5.3350445297696e-06, "loss": 0.3949, "step": 14301 }, { "epoch": 0.9887314206705842, "grad_norm": 0.6158215388423104, "learning_rate": 5.334486026577381e-06, "loss": 0.2924, "step": 14302 }, { "epoch": 0.9888005530591082, "grad_norm": 0.5887796047872531, "learning_rate": 5.3339275191929756e-06, "loss": 0.2284, "step": 14303 }, { "epoch": 0.9888696854476322, "grad_norm": 0.746769324315512, "learning_rate": 5.3333690076233844e-06, "loss": 0.3845, "step": 14304 }, { "epoch": 0.9889388178361562, "grad_norm": 0.7524199913133377, "learning_rate": 5.3328104918756085e-06, "loss": 0.3857, "step": 14305 }, { "epoch": 0.9890079502246802, "grad_norm": 0.6623209845080582, "learning_rate": 5.332251971956646e-06, "loss": 0.2852, "step": 14306 }, { "epoch": 0.9890770826132043, "grad_norm": 0.6919863178993342, "learning_rate": 5.331693447873498e-06, "loss": 0.3692, "step": 14307 }, { "epoch": 0.9891462150017283, "grad_norm": 0.633589283381586, "learning_rate": 5.331134919633165e-06, "loss": 0.4116, "step": 14308 }, { "epoch": 0.9892153473902523, "grad_norm": 0.6257877434944145, "learning_rate": 5.330576387242647e-06, "loss": 0.285, "step": 14309 }, { "epoch": 0.9892844797787763, "grad_norm": 0.6323556247094771, "learning_rate": 5.330017850708944e-06, "loss": 0.2207, "step": 14310 }, { "epoch": 0.9893536121673003, "grad_norm": 0.7214240815978623, "learning_rate": 5.329459310039055e-06, "loss": 0.3874, "step": 14311 }, { "epoch": 0.9894227445558244, "grad_norm": 0.6638294576830109, "learning_rate": 5.328900765239984e-06, "loss": 0.3662, "step": 14312 }, { "epoch": 0.9894918769443485, "grad_norm": 0.6692292921428347, "learning_rate": 5.328342216318726e-06, "loss": 0.2286, "step": 14313 }, { "epoch": 0.9895610093328725, "grad_norm": 0.8141098047514641, "learning_rate": 5.327783663282285e-06, "loss": 0.2727, "step": 14314 }, { "epoch": 0.9896301417213965, "grad_norm": 0.6649841390847917, "learning_rate": 5.327225106137662e-06, "loss": 0.4146, "step": 14315 }, { "epoch": 0.9896992741099205, "grad_norm": 0.7010976943962905, "learning_rate": 5.326666544891856e-06, "loss": 0.3548, "step": 14316 }, { "epoch": 0.9897684064984446, "grad_norm": 0.6487433327235577, "learning_rate": 5.3261079795518665e-06, "loss": 0.2477, "step": 14317 }, { "epoch": 0.9898375388869686, "grad_norm": 0.5896819272634466, "learning_rate": 5.325549410124696e-06, "loss": 0.2313, "step": 14318 }, { "epoch": 0.9899066712754926, "grad_norm": 0.6496286753153587, "learning_rate": 5.324990836617346e-06, "loss": 0.4072, "step": 14319 }, { "epoch": 0.9899758036640166, "grad_norm": 0.6836531146750083, "learning_rate": 5.324432259036815e-06, "loss": 0.2951, "step": 14320 }, { "epoch": 0.9900449360525406, "grad_norm": 0.6662936054455272, "learning_rate": 5.323873677390104e-06, "loss": 0.3195, "step": 14321 }, { "epoch": 0.9901140684410646, "grad_norm": 0.6346185921089115, "learning_rate": 5.323315091684215e-06, "loss": 0.318, "step": 14322 }, { "epoch": 0.9901832008295887, "grad_norm": 0.6289658301665693, "learning_rate": 5.322756501926147e-06, "loss": 0.3783, "step": 14323 }, { "epoch": 0.9902523332181127, "grad_norm": 0.7743793413752198, "learning_rate": 5.322197908122903e-06, "loss": 0.2929, "step": 14324 }, { "epoch": 0.9903214656066367, "grad_norm": 0.6740537478186472, "learning_rate": 5.321639310281483e-06, "loss": 0.3114, "step": 14325 }, { "epoch": 0.9903905979951607, "grad_norm": 0.6298501520760098, "learning_rate": 5.32108070840889e-06, "loss": 0.3945, "step": 14326 }, { "epoch": 0.9904597303836847, "grad_norm": 0.6755163421828152, "learning_rate": 5.320522102512122e-06, "loss": 0.3117, "step": 14327 }, { "epoch": 0.9905288627722088, "grad_norm": 0.738378479723266, "learning_rate": 5.3199634925981815e-06, "loss": 0.2034, "step": 14328 }, { "epoch": 0.9905979951607328, "grad_norm": 0.913858739766888, "learning_rate": 5.319404878674069e-06, "loss": 0.3653, "step": 14329 }, { "epoch": 0.9906671275492568, "grad_norm": 0.6666206570270321, "learning_rate": 5.318846260746787e-06, "loss": 0.3292, "step": 14330 }, { "epoch": 0.9907362599377808, "grad_norm": 0.6281586911712868, "learning_rate": 5.318287638823336e-06, "loss": 0.2087, "step": 14331 }, { "epoch": 0.9908053923263048, "grad_norm": 0.7936457085336379, "learning_rate": 5.317729012910718e-06, "loss": 0.2493, "step": 14332 }, { "epoch": 0.990874524714829, "grad_norm": 0.7110334203307466, "learning_rate": 5.317170383015934e-06, "loss": 0.3839, "step": 14333 }, { "epoch": 0.990943657103353, "grad_norm": 0.7080475964391739, "learning_rate": 5.316611749145985e-06, "loss": 0.3166, "step": 14334 }, { "epoch": 0.991012789491877, "grad_norm": 0.677856363891003, "learning_rate": 5.3160531113078716e-06, "loss": 0.269, "step": 14335 }, { "epoch": 0.991081921880401, "grad_norm": 0.8463629206867593, "learning_rate": 5.315494469508598e-06, "loss": 0.3288, "step": 14336 }, { "epoch": 0.991151054268925, "grad_norm": 1.568372135119115, "learning_rate": 5.314935823755164e-06, "loss": 0.3155, "step": 14337 }, { "epoch": 0.991220186657449, "grad_norm": 0.86659108788944, "learning_rate": 5.31437717405457e-06, "loss": 0.2861, "step": 14338 }, { "epoch": 0.9912893190459731, "grad_norm": 0.6793822519444754, "learning_rate": 5.313818520413821e-06, "loss": 0.2743, "step": 14339 }, { "epoch": 0.9913584514344971, "grad_norm": 0.6802289843308419, "learning_rate": 5.313259862839916e-06, "loss": 0.4427, "step": 14340 }, { "epoch": 0.9914275838230211, "grad_norm": 0.7391315738029466, "learning_rate": 5.312701201339856e-06, "loss": 0.283, "step": 14341 }, { "epoch": 0.9914967162115451, "grad_norm": 0.5484673676370175, "learning_rate": 5.312142535920647e-06, "loss": 0.2479, "step": 14342 }, { "epoch": 0.9915658486000691, "grad_norm": 0.6191100565510609, "learning_rate": 5.311583866589288e-06, "loss": 0.3486, "step": 14343 }, { "epoch": 0.9916349809885932, "grad_norm": 0.7035170807961164, "learning_rate": 5.3110251933527815e-06, "loss": 0.4194, "step": 14344 }, { "epoch": 0.9917041133771172, "grad_norm": 0.6484588918112797, "learning_rate": 5.3104665162181265e-06, "loss": 0.2918, "step": 14345 }, { "epoch": 0.9917732457656412, "grad_norm": 0.6482641899469482, "learning_rate": 5.30990783519233e-06, "loss": 0.329, "step": 14346 }, { "epoch": 0.9918423781541652, "grad_norm": 0.743460211396864, "learning_rate": 5.309349150282392e-06, "loss": 0.3879, "step": 14347 }, { "epoch": 0.9919115105426892, "grad_norm": 0.6541383740864546, "learning_rate": 5.308790461495314e-06, "loss": 0.3577, "step": 14348 }, { "epoch": 0.9919806429312132, "grad_norm": 0.5652890151357894, "learning_rate": 5.308231768838099e-06, "loss": 0.2952, "step": 14349 }, { "epoch": 0.9920497753197373, "grad_norm": 0.6936000576325047, "learning_rate": 5.30767307231775e-06, "loss": 0.3583, "step": 14350 }, { "epoch": 0.9921189077082613, "grad_norm": 1.5642909021277842, "learning_rate": 5.3071143719412655e-06, "loss": 0.3587, "step": 14351 }, { "epoch": 0.9921880400967853, "grad_norm": 0.7131469755697386, "learning_rate": 5.306555667715652e-06, "loss": 0.2594, "step": 14352 }, { "epoch": 0.9922571724853093, "grad_norm": 0.641170117630263, "learning_rate": 5.30599695964791e-06, "loss": 0.2684, "step": 14353 }, { "epoch": 0.9923263048738334, "grad_norm": 0.6985941102162156, "learning_rate": 5.305438247745041e-06, "loss": 0.3969, "step": 14354 }, { "epoch": 0.9923954372623575, "grad_norm": 0.701375129480276, "learning_rate": 5.304879532014051e-06, "loss": 0.2839, "step": 14355 }, { "epoch": 0.9924645696508815, "grad_norm": 0.7876253207287908, "learning_rate": 5.304320812461939e-06, "loss": 0.3379, "step": 14356 }, { "epoch": 0.9925337020394055, "grad_norm": 0.9106980604325551, "learning_rate": 5.303762089095708e-06, "loss": 0.3103, "step": 14357 }, { "epoch": 0.9926028344279295, "grad_norm": 0.6418808854771152, "learning_rate": 5.303203361922362e-06, "loss": 0.3954, "step": 14358 }, { "epoch": 0.9926719668164535, "grad_norm": 0.6702649950318512, "learning_rate": 5.302644630948902e-06, "loss": 0.2991, "step": 14359 }, { "epoch": 0.9927410992049776, "grad_norm": 0.6195085058833445, "learning_rate": 5.302085896182333e-06, "loss": 0.2623, "step": 14360 }, { "epoch": 0.9928102315935016, "grad_norm": 0.6403080520056863, "learning_rate": 5.301527157629656e-06, "loss": 0.3063, "step": 14361 }, { "epoch": 0.9928793639820256, "grad_norm": 0.6610828529608207, "learning_rate": 5.300968415297874e-06, "loss": 0.4011, "step": 14362 }, { "epoch": 0.9929484963705496, "grad_norm": 0.6119231492716013, "learning_rate": 5.3004096691939895e-06, "loss": 0.3364, "step": 14363 }, { "epoch": 0.9930176287590736, "grad_norm": 0.6076180703159, "learning_rate": 5.299850919325007e-06, "loss": 0.3127, "step": 14364 }, { "epoch": 0.9930867611475976, "grad_norm": 0.6531303032990394, "learning_rate": 5.299292165697928e-06, "loss": 0.4413, "step": 14365 }, { "epoch": 0.9931558935361217, "grad_norm": 2.5765068169773135, "learning_rate": 5.298733408319755e-06, "loss": 0.2798, "step": 14366 }, { "epoch": 0.9932250259246457, "grad_norm": 0.6778002315995295, "learning_rate": 5.298174647197493e-06, "loss": 0.2808, "step": 14367 }, { "epoch": 0.9932941583131697, "grad_norm": 0.6991832960161986, "learning_rate": 5.297615882338143e-06, "loss": 0.3106, "step": 14368 }, { "epoch": 0.9933632907016937, "grad_norm": 0.6168335707209599, "learning_rate": 5.29705711374871e-06, "loss": 0.3885, "step": 14369 }, { "epoch": 0.9934324230902177, "grad_norm": 0.6312233110034305, "learning_rate": 5.296498341436195e-06, "loss": 0.2443, "step": 14370 }, { "epoch": 0.9935015554787417, "grad_norm": 0.7477055316799861, "learning_rate": 5.295939565407603e-06, "loss": 0.2553, "step": 14371 }, { "epoch": 0.9935706878672658, "grad_norm": 0.8980392100074898, "learning_rate": 5.295380785669937e-06, "loss": 0.3874, "step": 14372 }, { "epoch": 0.9936398202557898, "grad_norm": 0.6449962543080122, "learning_rate": 5.2948220022302e-06, "loss": 0.3077, "step": 14373 }, { "epoch": 0.9937089526443139, "grad_norm": 0.5654715251366103, "learning_rate": 5.294263215095394e-06, "loss": 0.2226, "step": 14374 }, { "epoch": 0.9937780850328379, "grad_norm": 0.6669562881847653, "learning_rate": 5.293704424272524e-06, "loss": 0.2748, "step": 14375 }, { "epoch": 0.993847217421362, "grad_norm": 0.6856137692947801, "learning_rate": 5.293145629768595e-06, "loss": 0.4183, "step": 14376 }, { "epoch": 0.993916349809886, "grad_norm": 0.6243398601179163, "learning_rate": 5.292586831590607e-06, "loss": 0.2508, "step": 14377 }, { "epoch": 0.99398548219841, "grad_norm": 0.7941721290434582, "learning_rate": 5.292028029745566e-06, "loss": 0.3012, "step": 14378 }, { "epoch": 0.994054614586934, "grad_norm": 0.7127100296559457, "learning_rate": 5.2914692242404754e-06, "loss": 0.3641, "step": 14379 }, { "epoch": 0.994123746975458, "grad_norm": 0.6857425836989564, "learning_rate": 5.290910415082337e-06, "loss": 0.3489, "step": 14380 }, { "epoch": 0.994192879363982, "grad_norm": 0.684034668577876, "learning_rate": 5.290351602278155e-06, "loss": 0.2904, "step": 14381 }, { "epoch": 0.9942620117525061, "grad_norm": 0.9372497603664424, "learning_rate": 5.289792785834935e-06, "loss": 0.308, "step": 14382 }, { "epoch": 0.9943311441410301, "grad_norm": 0.6625162799132949, "learning_rate": 5.2892339657596805e-06, "loss": 0.3831, "step": 14383 }, { "epoch": 0.9944002765295541, "grad_norm": 0.9499276581262718, "learning_rate": 5.288675142059393e-06, "loss": 0.3052, "step": 14384 }, { "epoch": 0.9944694089180781, "grad_norm": 0.6680870192213788, "learning_rate": 5.288116314741078e-06, "loss": 0.3051, "step": 14385 }, { "epoch": 0.9945385413066021, "grad_norm": 0.7275795999290967, "learning_rate": 5.28755748381174e-06, "loss": 0.3704, "step": 14386 }, { "epoch": 0.9946076736951261, "grad_norm": 1.1701831878582327, "learning_rate": 5.286998649278381e-06, "loss": 0.364, "step": 14387 }, { "epoch": 0.9946768060836502, "grad_norm": 1.8596771170245594, "learning_rate": 5.286439811148007e-06, "loss": 0.2708, "step": 14388 }, { "epoch": 0.9947459384721742, "grad_norm": 0.7760255218574812, "learning_rate": 5.28588096942762e-06, "loss": 0.284, "step": 14389 }, { "epoch": 0.9948150708606982, "grad_norm": 0.7388244240284144, "learning_rate": 5.285322124124227e-06, "loss": 0.3796, "step": 14390 }, { "epoch": 0.9948842032492222, "grad_norm": 0.681605524579326, "learning_rate": 5.284763275244829e-06, "loss": 0.3132, "step": 14391 }, { "epoch": 0.9949533356377462, "grad_norm": 0.6437205630327788, "learning_rate": 5.284204422796431e-06, "loss": 0.2577, "step": 14392 }, { "epoch": 0.9950224680262703, "grad_norm": 0.7743321522440775, "learning_rate": 5.283645566786039e-06, "loss": 0.3204, "step": 14393 }, { "epoch": 0.9950916004147944, "grad_norm": 0.7428954035089504, "learning_rate": 5.283086707220656e-06, "loss": 0.4234, "step": 14394 }, { "epoch": 0.9951607328033184, "grad_norm": 0.7170807606366219, "learning_rate": 5.282527844107285e-06, "loss": 0.2991, "step": 14395 }, { "epoch": 0.9952298651918424, "grad_norm": 0.7270880043611531, "learning_rate": 5.281968977452932e-06, "loss": 0.2897, "step": 14396 }, { "epoch": 0.9952989975803664, "grad_norm": 0.6600462527239224, "learning_rate": 5.281410107264603e-06, "loss": 0.3662, "step": 14397 }, { "epoch": 0.9953681299688905, "grad_norm": 0.7883807636370321, "learning_rate": 5.280851233549299e-06, "loss": 0.2872, "step": 14398 }, { "epoch": 0.9954372623574145, "grad_norm": 0.6701420225979668, "learning_rate": 5.2802923563140255e-06, "loss": 0.3036, "step": 14399 }, { "epoch": 0.9955063947459385, "grad_norm": 0.6839175203713177, "learning_rate": 5.279733475565788e-06, "loss": 0.3217, "step": 14400 }, { "epoch": 0.9955755271344625, "grad_norm": 0.9912386092504026, "learning_rate": 5.279174591311591e-06, "loss": 0.3977, "step": 14401 }, { "epoch": 0.9956446595229865, "grad_norm": 4.1731920908599935, "learning_rate": 5.278615703558438e-06, "loss": 0.3223, "step": 14402 }, { "epoch": 0.9957137919115105, "grad_norm": 0.6376252417150658, "learning_rate": 5.278056812313336e-06, "loss": 0.2566, "step": 14403 }, { "epoch": 0.9957829243000346, "grad_norm": 0.6627885436685259, "learning_rate": 5.277497917583287e-06, "loss": 0.2847, "step": 14404 }, { "epoch": 0.9958520566885586, "grad_norm": 0.6464065393807761, "learning_rate": 5.276939019375295e-06, "loss": 0.3177, "step": 14405 }, { "epoch": 0.9959211890770826, "grad_norm": 0.6501701431105038, "learning_rate": 5.276380117696369e-06, "loss": 0.3409, "step": 14406 }, { "epoch": 0.9959903214656066, "grad_norm": 0.7143115157086145, "learning_rate": 5.275821212553511e-06, "loss": 0.2799, "step": 14407 }, { "epoch": 0.9960594538541306, "grad_norm": 0.6214765626465887, "learning_rate": 5.275262303953726e-06, "loss": 0.3482, "step": 14408 }, { "epoch": 0.9961285862426547, "grad_norm": 0.6578337336800039, "learning_rate": 5.274703391904019e-06, "loss": 0.2535, "step": 14409 }, { "epoch": 0.9961977186311787, "grad_norm": 0.6632007073064304, "learning_rate": 5.2741444764113954e-06, "loss": 0.2786, "step": 14410 }, { "epoch": 0.9962668510197027, "grad_norm": 0.677384840824622, "learning_rate": 5.273585557482861e-06, "loss": 0.379, "step": 14411 }, { "epoch": 0.9963359834082267, "grad_norm": 1.022972378839448, "learning_rate": 5.273026635125419e-06, "loss": 0.3697, "step": 14412 }, { "epoch": 0.9964051157967507, "grad_norm": 0.6284125039578786, "learning_rate": 5.272467709346075e-06, "loss": 0.3172, "step": 14413 }, { "epoch": 0.9964742481852747, "grad_norm": 0.7994341528576835, "learning_rate": 5.271908780151835e-06, "loss": 0.3076, "step": 14414 }, { "epoch": 0.9965433805737989, "grad_norm": 0.6287274745977193, "learning_rate": 5.271349847549704e-06, "loss": 0.3845, "step": 14415 }, { "epoch": 0.9966125129623229, "grad_norm": 0.6661780318916931, "learning_rate": 5.270790911546686e-06, "loss": 0.3327, "step": 14416 }, { "epoch": 0.9966816453508469, "grad_norm": 0.6150499408341265, "learning_rate": 5.270231972149787e-06, "loss": 0.2808, "step": 14417 }, { "epoch": 0.9967507777393709, "grad_norm": 0.618397789900607, "learning_rate": 5.269673029366015e-06, "loss": 0.2994, "step": 14418 }, { "epoch": 0.996819910127895, "grad_norm": 0.6693731176127048, "learning_rate": 5.269114083202369e-06, "loss": 0.3743, "step": 14419 }, { "epoch": 0.996889042516419, "grad_norm": 0.7037524963041182, "learning_rate": 5.2685551336658615e-06, "loss": 0.3025, "step": 14420 }, { "epoch": 0.996958174904943, "grad_norm": 0.5961518403367165, "learning_rate": 5.267996180763493e-06, "loss": 0.266, "step": 14421 }, { "epoch": 0.997027307293467, "grad_norm": 0.6508180427087696, "learning_rate": 5.267437224502271e-06, "loss": 0.365, "step": 14422 }, { "epoch": 0.997096439681991, "grad_norm": 0.6341916266374474, "learning_rate": 5.2668782648892e-06, "loss": 0.2663, "step": 14423 }, { "epoch": 0.997165572070515, "grad_norm": 0.6381895248994698, "learning_rate": 5.266319301931288e-06, "loss": 0.2901, "step": 14424 }, { "epoch": 0.9972347044590391, "grad_norm": 0.6342989541378916, "learning_rate": 5.265760335635537e-06, "loss": 0.2763, "step": 14425 }, { "epoch": 0.9973038368475631, "grad_norm": 0.6655130975384064, "learning_rate": 5.265201366008957e-06, "loss": 0.3786, "step": 14426 }, { "epoch": 0.9973729692360871, "grad_norm": 0.6119722466400641, "learning_rate": 5.264642393058546e-06, "loss": 0.3078, "step": 14427 }, { "epoch": 0.9974421016246111, "grad_norm": 0.6783831384462573, "learning_rate": 5.2640834167913195e-06, "loss": 0.2549, "step": 14428 }, { "epoch": 0.9975112340131351, "grad_norm": 0.7003392125839542, "learning_rate": 5.263524437214278e-06, "loss": 0.3799, "step": 14429 }, { "epoch": 0.9975803664016591, "grad_norm": 0.7142179241598245, "learning_rate": 5.262965454334427e-06, "loss": 0.3401, "step": 14430 }, { "epoch": 0.9976494987901832, "grad_norm": 0.7363924691873741, "learning_rate": 5.262406468158773e-06, "loss": 0.3214, "step": 14431 }, { "epoch": 0.9977186311787072, "grad_norm": 0.6402671529614472, "learning_rate": 5.261847478694323e-06, "loss": 0.275, "step": 14432 }, { "epoch": 0.9977877635672312, "grad_norm": 0.6437622479909475, "learning_rate": 5.261288485948083e-06, "loss": 0.3414, "step": 14433 }, { "epoch": 0.9978568959557552, "grad_norm": 0.6418369151086694, "learning_rate": 5.2607294899270556e-06, "loss": 0.2919, "step": 14434 }, { "epoch": 0.9979260283442793, "grad_norm": 0.6230268057559655, "learning_rate": 5.2601704906382516e-06, "loss": 0.2387, "step": 14435 }, { "epoch": 0.9979951607328034, "grad_norm": 0.6521184767830509, "learning_rate": 5.259611488088674e-06, "loss": 0.3808, "step": 14436 }, { "epoch": 0.9980642931213274, "grad_norm": 0.6273082923793248, "learning_rate": 5.2590524822853285e-06, "loss": 0.3325, "step": 14437 }, { "epoch": 0.9981334255098514, "grad_norm": 0.6619516754704975, "learning_rate": 5.2584934732352245e-06, "loss": 0.2573, "step": 14438 }, { "epoch": 0.9982025578983754, "grad_norm": 0.6649194145791212, "learning_rate": 5.257934460945365e-06, "loss": 0.3227, "step": 14439 }, { "epoch": 0.9982716902868994, "grad_norm": 0.8126592484979472, "learning_rate": 5.257375445422758e-06, "loss": 0.3572, "step": 14440 }, { "epoch": 0.9983408226754235, "grad_norm": 1.1175165336333246, "learning_rate": 5.256816426674408e-06, "loss": 0.361, "step": 14441 }, { "epoch": 0.9984099550639475, "grad_norm": 0.6424961545462812, "learning_rate": 5.256257404707323e-06, "loss": 0.2647, "step": 14442 }, { "epoch": 0.9984790874524715, "grad_norm": 0.6467602482236366, "learning_rate": 5.255698379528508e-06, "loss": 0.2854, "step": 14443 }, { "epoch": 0.9985482198409955, "grad_norm": 0.685947946206499, "learning_rate": 5.255139351144972e-06, "loss": 0.4163, "step": 14444 }, { "epoch": 0.9986173522295195, "grad_norm": 0.5993353406639836, "learning_rate": 5.254580319563718e-06, "loss": 0.2597, "step": 14445 }, { "epoch": 0.9986864846180435, "grad_norm": 0.6025847870893938, "learning_rate": 5.254021284791755e-06, "loss": 0.2628, "step": 14446 }, { "epoch": 0.9987556170065676, "grad_norm": 0.6368734282089602, "learning_rate": 5.253462246836089e-06, "loss": 0.3684, "step": 14447 }, { "epoch": 0.9988247493950916, "grad_norm": 0.675378523826554, "learning_rate": 5.252903205703723e-06, "loss": 0.329, "step": 14448 }, { "epoch": 0.9988938817836156, "grad_norm": 0.6367693602302669, "learning_rate": 5.252344161401669e-06, "loss": 0.2375, "step": 14449 }, { "epoch": 0.9989630141721396, "grad_norm": 0.6376353965593491, "learning_rate": 5.251785113936931e-06, "loss": 0.2866, "step": 14450 }, { "epoch": 0.9990321465606636, "grad_norm": 0.6679234325367182, "learning_rate": 5.251226063316517e-06, "loss": 0.3578, "step": 14451 }, { "epoch": 0.9991012789491877, "grad_norm": 0.6914603187328627, "learning_rate": 5.250667009547431e-06, "loss": 0.2582, "step": 14452 }, { "epoch": 0.9991704113377117, "grad_norm": 0.6363869909159984, "learning_rate": 5.250107952636682e-06, "loss": 0.2809, "step": 14453 }, { "epoch": 0.9992395437262357, "grad_norm": 0.7298609901601012, "learning_rate": 5.249548892591277e-06, "loss": 0.3881, "step": 14454 }, { "epoch": 0.9993086761147598, "grad_norm": 0.6191172466543102, "learning_rate": 5.2489898294182195e-06, "loss": 0.3061, "step": 14455 }, { "epoch": 0.9993778085032838, "grad_norm": 0.602440888055873, "learning_rate": 5.24843076312452e-06, "loss": 0.2739, "step": 14456 }, { "epoch": 0.9994469408918079, "grad_norm": 0.7724267627631906, "learning_rate": 5.247871693717186e-06, "loss": 0.3291, "step": 14457 }, { "epoch": 0.9995160732803319, "grad_norm": 0.6777419197097566, "learning_rate": 5.247312621203221e-06, "loss": 0.4042, "step": 14458 }, { "epoch": 0.9995852056688559, "grad_norm": 0.7323398163929749, "learning_rate": 5.246753545589634e-06, "loss": 0.2953, "step": 14459 }, { "epoch": 0.9996543380573799, "grad_norm": 1.033443135131651, "learning_rate": 5.246194466883431e-06, "loss": 0.2475, "step": 14460 }, { "epoch": 0.9997234704459039, "grad_norm": 0.6105675691430491, "learning_rate": 5.245635385091621e-06, "loss": 0.3387, "step": 14461 }, { "epoch": 0.999792602834428, "grad_norm": 0.8227887283342632, "learning_rate": 5.245076300221209e-06, "loss": 0.4307, "step": 14462 }, { "epoch": 0.999861735222952, "grad_norm": 0.6863413100747847, "learning_rate": 5.244517212279202e-06, "loss": 0.2805, "step": 14463 }, { "epoch": 0.999930867611476, "grad_norm": 0.6119454636410213, "learning_rate": 5.24395812127261e-06, "loss": 0.2801, "step": 14464 }, { "epoch": 1.0, "grad_norm": 0.7688963193971012, "learning_rate": 5.243399027208437e-06, "loss": 0.3715, "step": 14465 }, { "epoch": 1.000069132388524, "grad_norm": 0.6343126153278393, "learning_rate": 5.242839930093692e-06, "loss": 0.363, "step": 14466 }, { "epoch": 1.000138264777048, "grad_norm": 0.6457493452596862, "learning_rate": 5.2422808299353814e-06, "loss": 0.2617, "step": 14467 }, { "epoch": 1.000207397165572, "grad_norm": 0.6478034793210616, "learning_rate": 5.2417217267405156e-06, "loss": 0.282, "step": 14468 }, { "epoch": 1.000276529554096, "grad_norm": 0.615433046528056, "learning_rate": 5.241162620516096e-06, "loss": 0.326, "step": 14469 }, { "epoch": 1.00034566194262, "grad_norm": 0.6055858690007683, "learning_rate": 5.2406035112691345e-06, "loss": 0.3244, "step": 14470 }, { "epoch": 1.000414794331144, "grad_norm": 0.56711764552049, "learning_rate": 5.240044399006638e-06, "loss": 0.237, "step": 14471 }, { "epoch": 1.0004839267196681, "grad_norm": 0.6981001977261373, "learning_rate": 5.239485283735612e-06, "loss": 0.2996, "step": 14472 }, { "epoch": 1.0005530591081921, "grad_norm": 0.7613077089522826, "learning_rate": 5.238926165463065e-06, "loss": 0.3414, "step": 14473 }, { "epoch": 1.0006221914967162, "grad_norm": 0.6153928574010532, "learning_rate": 5.238367044196007e-06, "loss": 0.2767, "step": 14474 }, { "epoch": 1.0006913238852402, "grad_norm": 0.6331710728448076, "learning_rate": 5.237807919941442e-06, "loss": 0.2434, "step": 14475 }, { "epoch": 1.0007604562737642, "grad_norm": 0.665112578864907, "learning_rate": 5.2372487927063784e-06, "loss": 0.3362, "step": 14476 }, { "epoch": 1.0008295886622882, "grad_norm": 0.6907637430440711, "learning_rate": 5.236689662497827e-06, "loss": 0.3693, "step": 14477 }, { "epoch": 1.0008987210508122, "grad_norm": 0.6250325792452497, "learning_rate": 5.236130529322791e-06, "loss": 0.2442, "step": 14478 }, { "epoch": 1.0009678534393363, "grad_norm": 0.6812033974651329, "learning_rate": 5.235571393188281e-06, "loss": 0.3122, "step": 14479 }, { "epoch": 1.0010369858278603, "grad_norm": 0.7015145608731642, "learning_rate": 5.235012254101304e-06, "loss": 0.3871, "step": 14480 }, { "epoch": 1.0011061182163843, "grad_norm": 0.7482049797541103, "learning_rate": 5.234453112068868e-06, "loss": 0.3016, "step": 14481 }, { "epoch": 1.0011752506049083, "grad_norm": 0.6478655754971234, "learning_rate": 5.23389396709798e-06, "loss": 0.2473, "step": 14482 }, { "epoch": 1.0012443829934323, "grad_norm": 0.7203464982856133, "learning_rate": 5.233334819195649e-06, "loss": 0.3523, "step": 14483 }, { "epoch": 1.0013135153819563, "grad_norm": 0.6453751729126078, "learning_rate": 5.232775668368882e-06, "loss": 0.2886, "step": 14484 }, { "epoch": 1.0013826477704804, "grad_norm": 0.7657115162575516, "learning_rate": 5.2322165146246885e-06, "loss": 0.2575, "step": 14485 }, { "epoch": 1.0014517801590046, "grad_norm": 0.6875580698048047, "learning_rate": 5.231657357970075e-06, "loss": 0.2864, "step": 14486 }, { "epoch": 1.0015209125475286, "grad_norm": 0.7283889117511452, "learning_rate": 5.231098198412048e-06, "loss": 0.3287, "step": 14487 }, { "epoch": 1.0015900449360526, "grad_norm": 0.6662732702709819, "learning_rate": 5.23053903595762e-06, "loss": 0.2782, "step": 14488 }, { "epoch": 1.0016591773245767, "grad_norm": 0.7294522725582278, "learning_rate": 5.2299798706137964e-06, "loss": 0.2567, "step": 14489 }, { "epoch": 1.0017283097131007, "grad_norm": 0.674933028468614, "learning_rate": 5.229420702387584e-06, "loss": 0.2877, "step": 14490 }, { "epoch": 1.0017974421016247, "grad_norm": 0.5804959444047394, "learning_rate": 5.2288615312859926e-06, "loss": 0.3403, "step": 14491 }, { "epoch": 1.0018665744901487, "grad_norm": 0.6243461432281615, "learning_rate": 5.228302357316032e-06, "loss": 0.2132, "step": 14492 }, { "epoch": 1.0019357068786727, "grad_norm": 0.8334410744868522, "learning_rate": 5.227743180484708e-06, "loss": 0.2666, "step": 14493 }, { "epoch": 1.0020048392671967, "grad_norm": 0.6761439467797414, "learning_rate": 5.227184000799028e-06, "loss": 0.3024, "step": 14494 }, { "epoch": 1.0020739716557208, "grad_norm": 0.66870566422748, "learning_rate": 5.226624818266004e-06, "loss": 0.3725, "step": 14495 }, { "epoch": 1.0021431040442448, "grad_norm": 0.6439549920079543, "learning_rate": 5.226065632892642e-06, "loss": 0.2168, "step": 14496 }, { "epoch": 1.0022122364327688, "grad_norm": 0.6912680194241844, "learning_rate": 5.2255064446859504e-06, "loss": 0.292, "step": 14497 }, { "epoch": 1.0022813688212928, "grad_norm": 0.7627589207225848, "learning_rate": 5.224947253652938e-06, "loss": 0.3697, "step": 14498 }, { "epoch": 1.0023505012098168, "grad_norm": 0.6322861583443661, "learning_rate": 5.224388059800613e-06, "loss": 0.2742, "step": 14499 }, { "epoch": 1.0024196335983409, "grad_norm": 0.7674698310252587, "learning_rate": 5.223828863135984e-06, "loss": 0.2254, "step": 14500 }, { "epoch": 1.0024887659868649, "grad_norm": 0.7171911433636864, "learning_rate": 5.22326966366606e-06, "loss": 0.3206, "step": 14501 }, { "epoch": 1.002557898375389, "grad_norm": 0.652340309127005, "learning_rate": 5.222710461397849e-06, "loss": 0.3103, "step": 14502 }, { "epoch": 1.002627030763913, "grad_norm": 0.9162370973039577, "learning_rate": 5.222151256338362e-06, "loss": 0.2466, "step": 14503 }, { "epoch": 1.002696163152437, "grad_norm": 0.6151479607022354, "learning_rate": 5.221592048494603e-06, "loss": 0.2342, "step": 14504 }, { "epoch": 1.002765295540961, "grad_norm": 0.6839004984013309, "learning_rate": 5.221032837873583e-06, "loss": 0.357, "step": 14505 }, { "epoch": 1.002834427929485, "grad_norm": 0.6839114278995037, "learning_rate": 5.220473624482312e-06, "loss": 0.2994, "step": 14506 }, { "epoch": 1.002903560318009, "grad_norm": 0.6001318192625426, "learning_rate": 5.219914408327796e-06, "loss": 0.25, "step": 14507 }, { "epoch": 1.002972692706533, "grad_norm": 0.6629655753630643, "learning_rate": 5.2193551894170455e-06, "loss": 0.2654, "step": 14508 }, { "epoch": 1.003041825095057, "grad_norm": 0.7031490432266053, "learning_rate": 5.218795967757071e-06, "loss": 0.4134, "step": 14509 }, { "epoch": 1.003110957483581, "grad_norm": 0.6822957201149827, "learning_rate": 5.218236743354878e-06, "loss": 0.2556, "step": 14510 }, { "epoch": 1.003180089872105, "grad_norm": 0.656991661334845, "learning_rate": 5.2176775162174785e-06, "loss": 0.225, "step": 14511 }, { "epoch": 1.003249222260629, "grad_norm": 0.98166065984014, "learning_rate": 5.217118286351878e-06, "loss": 0.3563, "step": 14512 }, { "epoch": 1.003318354649153, "grad_norm": 0.6213934359493177, "learning_rate": 5.216559053765088e-06, "loss": 0.2203, "step": 14513 }, { "epoch": 1.003387487037677, "grad_norm": 0.546556494939633, "learning_rate": 5.215999818464117e-06, "loss": 0.244, "step": 14514 }, { "epoch": 1.0034566194262011, "grad_norm": 0.6230783614527988, "learning_rate": 5.215440580455973e-06, "loss": 0.2738, "step": 14515 }, { "epoch": 1.0035257518147251, "grad_norm": 0.5783296066598226, "learning_rate": 5.214881339747666e-06, "loss": 0.364, "step": 14516 }, { "epoch": 1.0035948842032492, "grad_norm": 0.5787679639217284, "learning_rate": 5.214322096346206e-06, "loss": 0.217, "step": 14517 }, { "epoch": 1.0036640165917732, "grad_norm": 0.6333608739665086, "learning_rate": 5.2137628502586005e-06, "loss": 0.2826, "step": 14518 }, { "epoch": 1.0037331489802972, "grad_norm": 0.677023174764167, "learning_rate": 5.213203601491858e-06, "loss": 0.3646, "step": 14519 }, { "epoch": 1.0038022813688212, "grad_norm": 0.6377537145597396, "learning_rate": 5.212644350052989e-06, "loss": 0.295, "step": 14520 }, { "epoch": 1.0038714137573452, "grad_norm": 0.6010511199714988, "learning_rate": 5.212085095949003e-06, "loss": 0.2558, "step": 14521 }, { "epoch": 1.0039405461458693, "grad_norm": 0.6294404503511177, "learning_rate": 5.211525839186909e-06, "loss": 0.2652, "step": 14522 }, { "epoch": 1.0040096785343933, "grad_norm": 0.6488239511380183, "learning_rate": 5.210966579773716e-06, "loss": 0.3438, "step": 14523 }, { "epoch": 1.0040788109229173, "grad_norm": 0.6260450345745567, "learning_rate": 5.2104073177164335e-06, "loss": 0.2518, "step": 14524 }, { "epoch": 1.0041479433114413, "grad_norm": 0.6178395954054068, "learning_rate": 5.2098480530220705e-06, "loss": 0.265, "step": 14525 }, { "epoch": 1.0042170756999655, "grad_norm": 0.7898128347221469, "learning_rate": 5.209288785697637e-06, "loss": 0.2842, "step": 14526 }, { "epoch": 1.0042862080884896, "grad_norm": 0.6533793175411937, "learning_rate": 5.208729515750142e-06, "loss": 0.3466, "step": 14527 }, { "epoch": 1.0043553404770136, "grad_norm": 0.8090275669917272, "learning_rate": 5.2081702431865955e-06, "loss": 0.3284, "step": 14528 }, { "epoch": 1.0044244728655376, "grad_norm": 0.6258453047939144, "learning_rate": 5.2076109680140064e-06, "loss": 0.2243, "step": 14529 }, { "epoch": 1.0044936052540616, "grad_norm": 0.630651256096278, "learning_rate": 5.207051690239382e-06, "loss": 0.372, "step": 14530 }, { "epoch": 1.0045627376425856, "grad_norm": 0.6715784060193223, "learning_rate": 5.206492409869737e-06, "loss": 0.2523, "step": 14531 }, { "epoch": 1.0046318700311097, "grad_norm": 0.646713529890098, "learning_rate": 5.205933126912077e-06, "loss": 0.2431, "step": 14532 }, { "epoch": 1.0047010024196337, "grad_norm": 0.6655426644038474, "learning_rate": 5.205373841373413e-06, "loss": 0.2523, "step": 14533 }, { "epoch": 1.0047701348081577, "grad_norm": 0.7091121663668633, "learning_rate": 5.204814553260754e-06, "loss": 0.3601, "step": 14534 }, { "epoch": 1.0048392671966817, "grad_norm": 0.6360884067036217, "learning_rate": 5.204255262581111e-06, "loss": 0.2876, "step": 14535 }, { "epoch": 1.0049083995852057, "grad_norm": 0.6547451941486314, "learning_rate": 5.203695969341492e-06, "loss": 0.2103, "step": 14536 }, { "epoch": 1.0049775319737297, "grad_norm": 0.6423886518995451, "learning_rate": 5.2031366735489076e-06, "loss": 0.3387, "step": 14537 }, { "epoch": 1.0050466643622538, "grad_norm": 0.7309743084171215, "learning_rate": 5.202577375210368e-06, "loss": 0.2637, "step": 14538 }, { "epoch": 1.0051157967507778, "grad_norm": 0.5716303225039463, "learning_rate": 5.202018074332883e-06, "loss": 0.2062, "step": 14539 }, { "epoch": 1.0051849291393018, "grad_norm": 0.7441313261261366, "learning_rate": 5.20145877092346e-06, "loss": 0.2024, "step": 14540 }, { "epoch": 1.0052540615278258, "grad_norm": 0.7310061746979398, "learning_rate": 5.200899464989112e-06, "loss": 0.3537, "step": 14541 }, { "epoch": 1.0053231939163498, "grad_norm": 0.6191721665898727, "learning_rate": 5.200340156536848e-06, "loss": 0.2575, "step": 14542 }, { "epoch": 1.0053923263048739, "grad_norm": 0.6656021215987247, "learning_rate": 5.199780845573677e-06, "loss": 0.2176, "step": 14543 }, { "epoch": 1.0054614586933979, "grad_norm": 0.6552204127712816, "learning_rate": 5.19922153210661e-06, "loss": 0.3261, "step": 14544 }, { "epoch": 1.0055305910819219, "grad_norm": 0.6868879539141391, "learning_rate": 5.198662216142657e-06, "loss": 0.2515, "step": 14545 }, { "epoch": 1.005599723470446, "grad_norm": 0.6308728712506653, "learning_rate": 5.198102897688827e-06, "loss": 0.22, "step": 14546 }, { "epoch": 1.00566885585897, "grad_norm": 0.6441804141162072, "learning_rate": 5.19754357675213e-06, "loss": 0.2362, "step": 14547 }, { "epoch": 1.005737988247494, "grad_norm": 0.7111058946472594, "learning_rate": 5.196984253339579e-06, "loss": 0.4624, "step": 14548 }, { "epoch": 1.005807120636018, "grad_norm": 0.8234258518749638, "learning_rate": 5.19642492745818e-06, "loss": 0.2898, "step": 14549 }, { "epoch": 1.005876253024542, "grad_norm": 0.8445216127006593, "learning_rate": 5.1958655991149465e-06, "loss": 0.2121, "step": 14550 }, { "epoch": 1.005945385413066, "grad_norm": 0.6559364044498441, "learning_rate": 5.1953062683168855e-06, "loss": 0.3514, "step": 14551 }, { "epoch": 1.00601451780159, "grad_norm": 0.6258463024885703, "learning_rate": 5.19474693507101e-06, "loss": 0.3745, "step": 14552 }, { "epoch": 1.006083650190114, "grad_norm": 0.7048814883741431, "learning_rate": 5.194187599384329e-06, "loss": 0.3222, "step": 14553 }, { "epoch": 1.006152782578638, "grad_norm": 0.6913364756824276, "learning_rate": 5.193628261263852e-06, "loss": 0.2838, "step": 14554 }, { "epoch": 1.006221914967162, "grad_norm": 0.6788152286109389, "learning_rate": 5.193068920716592e-06, "loss": 0.3418, "step": 14555 }, { "epoch": 1.006291047355686, "grad_norm": 0.6948479053962823, "learning_rate": 5.192509577749558e-06, "loss": 0.2707, "step": 14556 }, { "epoch": 1.00636017974421, "grad_norm": 0.6484420395842093, "learning_rate": 5.191950232369759e-06, "loss": 0.2321, "step": 14557 }, { "epoch": 1.0064293121327341, "grad_norm": 0.6805083335129873, "learning_rate": 5.191390884584206e-06, "loss": 0.3086, "step": 14558 }, { "epoch": 1.0064984445212581, "grad_norm": 0.6796908592948584, "learning_rate": 5.19083153439991e-06, "loss": 0.3528, "step": 14559 }, { "epoch": 1.0065675769097822, "grad_norm": 0.5827558894568716, "learning_rate": 5.1902721818238826e-06, "loss": 0.2458, "step": 14560 }, { "epoch": 1.0066367092983062, "grad_norm": 0.6578158890120862, "learning_rate": 5.189712826863131e-06, "loss": 0.2337, "step": 14561 }, { "epoch": 1.0067058416868302, "grad_norm": 0.6941124275571856, "learning_rate": 5.189153469524668e-06, "loss": 0.401, "step": 14562 }, { "epoch": 1.0067749740753542, "grad_norm": 0.6402176057088083, "learning_rate": 5.188594109815505e-06, "loss": 0.2868, "step": 14563 }, { "epoch": 1.0068441064638782, "grad_norm": 0.6618790533193951, "learning_rate": 5.188034747742651e-06, "loss": 0.2866, "step": 14564 }, { "epoch": 1.0069132388524022, "grad_norm": 0.8149409307498433, "learning_rate": 5.187475383313115e-06, "loss": 0.2687, "step": 14565 }, { "epoch": 1.0069823712409263, "grad_norm": 0.7674794661802424, "learning_rate": 5.186916016533913e-06, "loss": 0.3631, "step": 14566 }, { "epoch": 1.0070515036294505, "grad_norm": 0.6114898012654435, "learning_rate": 5.186356647412051e-06, "loss": 0.22, "step": 14567 }, { "epoch": 1.0071206360179745, "grad_norm": 0.6217166707012207, "learning_rate": 5.185797275954541e-06, "loss": 0.205, "step": 14568 }, { "epoch": 1.0071897684064985, "grad_norm": 0.6972885742935309, "learning_rate": 5.185237902168392e-06, "loss": 0.3517, "step": 14569 }, { "epoch": 1.0072589007950226, "grad_norm": 0.6683595802756536, "learning_rate": 5.184678526060618e-06, "loss": 0.3165, "step": 14570 }, { "epoch": 1.0073280331835466, "grad_norm": 0.6316339048036025, "learning_rate": 5.184119147638228e-06, "loss": 0.2657, "step": 14571 }, { "epoch": 1.0073971655720706, "grad_norm": 0.6533114662016996, "learning_rate": 5.1835597669082325e-06, "loss": 0.3464, "step": 14572 }, { "epoch": 1.0074662979605946, "grad_norm": 0.6368296678876376, "learning_rate": 5.183000383877644e-06, "loss": 0.3306, "step": 14573 }, { "epoch": 1.0075354303491186, "grad_norm": 0.7284173814126574, "learning_rate": 5.182440998553472e-06, "loss": 0.3112, "step": 14574 }, { "epoch": 1.0076045627376427, "grad_norm": 0.6486789713273382, "learning_rate": 5.1818816109427285e-06, "loss": 0.2399, "step": 14575 }, { "epoch": 1.0076736951261667, "grad_norm": 0.6654721507167299, "learning_rate": 5.181322221052422e-06, "loss": 0.3233, "step": 14576 }, { "epoch": 1.0077428275146907, "grad_norm": 0.690205174802552, "learning_rate": 5.180762828889565e-06, "loss": 0.3552, "step": 14577 }, { "epoch": 1.0078119599032147, "grad_norm": 0.7165911033811178, "learning_rate": 5.18020343446117e-06, "loss": 0.234, "step": 14578 }, { "epoch": 1.0078810922917387, "grad_norm": 0.6287420360154119, "learning_rate": 5.179644037774244e-06, "loss": 0.2331, "step": 14579 }, { "epoch": 1.0079502246802627, "grad_norm": 0.6804628956522523, "learning_rate": 5.179084638835803e-06, "loss": 0.3603, "step": 14580 }, { "epoch": 1.0080193570687868, "grad_norm": 0.6900780617584628, "learning_rate": 5.178525237652855e-06, "loss": 0.3018, "step": 14581 }, { "epoch": 1.0080884894573108, "grad_norm": 3.2399283885238965, "learning_rate": 5.177965834232412e-06, "loss": 0.247, "step": 14582 }, { "epoch": 1.0081576218458348, "grad_norm": 0.7458055275149168, "learning_rate": 5.177406428581484e-06, "loss": 0.3006, "step": 14583 }, { "epoch": 1.0082267542343588, "grad_norm": 0.7036564201949518, "learning_rate": 5.176847020707083e-06, "loss": 0.3584, "step": 14584 }, { "epoch": 1.0082958866228828, "grad_norm": 0.6804756332575568, "learning_rate": 5.176287610616222e-06, "loss": 0.231, "step": 14585 }, { "epoch": 1.0083650190114068, "grad_norm": 0.7073652865886674, "learning_rate": 5.175728198315909e-06, "loss": 0.2882, "step": 14586 }, { "epoch": 1.0084341513999309, "grad_norm": 0.7017589459076972, "learning_rate": 5.175168783813156e-06, "loss": 0.3491, "step": 14587 }, { "epoch": 1.0085032837884549, "grad_norm": 0.6641229857595865, "learning_rate": 5.174609367114976e-06, "loss": 0.3341, "step": 14588 }, { "epoch": 1.008572416176979, "grad_norm": 0.7471063807302932, "learning_rate": 5.174049948228379e-06, "loss": 0.3271, "step": 14589 }, { "epoch": 1.008641548565503, "grad_norm": 0.6904841850475812, "learning_rate": 5.173490527160375e-06, "loss": 0.2911, "step": 14590 }, { "epoch": 1.008710680954027, "grad_norm": 0.667055177549593, "learning_rate": 5.172931103917979e-06, "loss": 0.3629, "step": 14591 }, { "epoch": 1.008779813342551, "grad_norm": 0.8600853066011959, "learning_rate": 5.1723716785082e-06, "loss": 0.2504, "step": 14592 }, { "epoch": 1.008848945731075, "grad_norm": 0.5994041643964271, "learning_rate": 5.171812250938049e-06, "loss": 0.2038, "step": 14593 }, { "epoch": 1.008918078119599, "grad_norm": 0.6709660428490414, "learning_rate": 5.171252821214537e-06, "loss": 0.3144, "step": 14594 }, { "epoch": 1.008987210508123, "grad_norm": 0.6867321247071919, "learning_rate": 5.170693389344678e-06, "loss": 0.3205, "step": 14595 }, { "epoch": 1.009056342896647, "grad_norm": 0.8359911752816329, "learning_rate": 5.170133955335482e-06, "loss": 0.2573, "step": 14596 }, { "epoch": 1.009125475285171, "grad_norm": 0.7131308242294337, "learning_rate": 5.1695745191939596e-06, "loss": 0.2892, "step": 14597 }, { "epoch": 1.009194607673695, "grad_norm": 0.7425170140996161, "learning_rate": 5.169015080927123e-06, "loss": 0.3613, "step": 14598 }, { "epoch": 1.009263740062219, "grad_norm": 0.6989156951557024, "learning_rate": 5.168455640541985e-06, "loss": 0.2662, "step": 14599 }, { "epoch": 1.009332872450743, "grad_norm": 0.6050463136147347, "learning_rate": 5.167896198045557e-06, "loss": 0.2409, "step": 14600 }, { "epoch": 1.0094020048392671, "grad_norm": 0.6876398935201076, "learning_rate": 5.1673367534448484e-06, "loss": 0.2949, "step": 14601 }, { "epoch": 1.0094711372277911, "grad_norm": 0.636358300573906, "learning_rate": 5.166777306746873e-06, "loss": 0.3394, "step": 14602 }, { "epoch": 1.0095402696163152, "grad_norm": 0.7397569967302928, "learning_rate": 5.166217857958641e-06, "loss": 0.278, "step": 14603 }, { "epoch": 1.0096094020048392, "grad_norm": 0.6624645715914406, "learning_rate": 5.165658407087164e-06, "loss": 0.2353, "step": 14604 }, { "epoch": 1.0096785343933632, "grad_norm": 0.6603358162554965, "learning_rate": 5.165098954139456e-06, "loss": 0.3657, "step": 14605 }, { "epoch": 1.0097476667818872, "grad_norm": 0.7300139320007188, "learning_rate": 5.164539499122526e-06, "loss": 0.2955, "step": 14606 }, { "epoch": 1.0098167991704112, "grad_norm": 0.6428360378336258, "learning_rate": 5.163980042043387e-06, "loss": 0.2512, "step": 14607 }, { "epoch": 1.0098859315589355, "grad_norm": 0.6325124739644482, "learning_rate": 5.163420582909053e-06, "loss": 0.2913, "step": 14608 }, { "epoch": 1.0099550639474595, "grad_norm": 0.7722365871962525, "learning_rate": 5.1628611217265315e-06, "loss": 0.3105, "step": 14609 }, { "epoch": 1.0100241963359835, "grad_norm": 0.7461691399556649, "learning_rate": 5.162301658502837e-06, "loss": 0.2319, "step": 14610 }, { "epoch": 1.0100933287245075, "grad_norm": 0.9075901435776528, "learning_rate": 5.161742193244981e-06, "loss": 0.3093, "step": 14611 }, { "epoch": 1.0101624611130315, "grad_norm": 0.6777943277290684, "learning_rate": 5.161182725959974e-06, "loss": 0.3261, "step": 14612 }, { "epoch": 1.0102315935015556, "grad_norm": 0.66400336431793, "learning_rate": 5.160623256654832e-06, "loss": 0.3198, "step": 14613 }, { "epoch": 1.0103007258900796, "grad_norm": 0.6136053024803171, "learning_rate": 5.160063785336562e-06, "loss": 0.227, "step": 14614 }, { "epoch": 1.0103698582786036, "grad_norm": 0.7197360038164397, "learning_rate": 5.159504312012178e-06, "loss": 0.3401, "step": 14615 }, { "epoch": 1.0104389906671276, "grad_norm": 0.6815909929675245, "learning_rate": 5.158944836688693e-06, "loss": 0.355, "step": 14616 }, { "epoch": 1.0105081230556516, "grad_norm": 0.6759166619820224, "learning_rate": 5.158385359373118e-06, "loss": 0.27, "step": 14617 }, { "epoch": 1.0105772554441756, "grad_norm": 0.6434370340480957, "learning_rate": 5.157825880072464e-06, "loss": 0.2953, "step": 14618 }, { "epoch": 1.0106463878326997, "grad_norm": 0.6964205935565444, "learning_rate": 5.157266398793745e-06, "loss": 0.3336, "step": 14619 }, { "epoch": 1.0107155202212237, "grad_norm": 0.6248575726729559, "learning_rate": 5.156706915543973e-06, "loss": 0.281, "step": 14620 }, { "epoch": 1.0107846526097477, "grad_norm": 0.6705859475266285, "learning_rate": 5.15614743033016e-06, "loss": 0.2664, "step": 14621 }, { "epoch": 1.0108537849982717, "grad_norm": 0.6382354519389889, "learning_rate": 5.155587943159316e-06, "loss": 0.2855, "step": 14622 }, { "epoch": 1.0109229173867957, "grad_norm": 0.6583218971610741, "learning_rate": 5.155028454038455e-06, "loss": 0.3613, "step": 14623 }, { "epoch": 1.0109920497753198, "grad_norm": 0.6475441549386481, "learning_rate": 5.15446896297459e-06, "loss": 0.2958, "step": 14624 }, { "epoch": 1.0110611821638438, "grad_norm": 0.6556294186301791, "learning_rate": 5.15390946997473e-06, "loss": 0.2309, "step": 14625 }, { "epoch": 1.0111303145523678, "grad_norm": 0.7782219711964105, "learning_rate": 5.153349975045891e-06, "loss": 0.3334, "step": 14626 }, { "epoch": 1.0111994469408918, "grad_norm": 0.6855785480065016, "learning_rate": 5.152790478195086e-06, "loss": 0.3142, "step": 14627 }, { "epoch": 1.0112685793294158, "grad_norm": 0.695636022703009, "learning_rate": 5.152230979429322e-06, "loss": 0.2536, "step": 14628 }, { "epoch": 1.0113377117179398, "grad_norm": 0.6409779660761847, "learning_rate": 5.151671478755615e-06, "loss": 0.2469, "step": 14629 }, { "epoch": 1.0114068441064639, "grad_norm": 0.6409688855100726, "learning_rate": 5.151111976180977e-06, "loss": 0.3076, "step": 14630 }, { "epoch": 1.0114759764949879, "grad_norm": 0.6266781868382997, "learning_rate": 5.15055247171242e-06, "loss": 0.3216, "step": 14631 }, { "epoch": 1.011545108883512, "grad_norm": 0.6549072160497411, "learning_rate": 5.149992965356957e-06, "loss": 0.2225, "step": 14632 }, { "epoch": 1.011614241272036, "grad_norm": 0.7849423393944623, "learning_rate": 5.149433457121597e-06, "loss": 0.2862, "step": 14633 }, { "epoch": 1.01168337366056, "grad_norm": 0.6969715596617543, "learning_rate": 5.148873947013359e-06, "loss": 0.3963, "step": 14634 }, { "epoch": 1.011752506049084, "grad_norm": 0.6857614581282232, "learning_rate": 5.14831443503925e-06, "loss": 0.3017, "step": 14635 }, { "epoch": 1.011821638437608, "grad_norm": 0.6781685426486939, "learning_rate": 5.147754921206284e-06, "loss": 0.295, "step": 14636 }, { "epoch": 1.011890770826132, "grad_norm": 0.6714676020422183, "learning_rate": 5.147195405521475e-06, "loss": 0.361, "step": 14637 }, { "epoch": 1.011959903214656, "grad_norm": 0.7162054388293599, "learning_rate": 5.1466358879918335e-06, "loss": 0.3572, "step": 14638 }, { "epoch": 1.01202903560318, "grad_norm": 0.68269162857491, "learning_rate": 5.146076368624373e-06, "loss": 0.2565, "step": 14639 }, { "epoch": 1.012098167991704, "grad_norm": 0.7097653579401392, "learning_rate": 5.145516847426104e-06, "loss": 0.2812, "step": 14640 }, { "epoch": 1.012167300380228, "grad_norm": 0.6743451529823663, "learning_rate": 5.144957324404044e-06, "loss": 0.3468, "step": 14641 }, { "epoch": 1.012236432768752, "grad_norm": 0.6587607497051017, "learning_rate": 5.144397799565202e-06, "loss": 0.2276, "step": 14642 }, { "epoch": 1.012305565157276, "grad_norm": 0.6166307916992437, "learning_rate": 5.143838272916589e-06, "loss": 0.2376, "step": 14643 }, { "epoch": 1.0123746975458001, "grad_norm": 0.7698096573307308, "learning_rate": 5.143278744465222e-06, "loss": 0.3183, "step": 14644 }, { "epoch": 1.0124438299343241, "grad_norm": 0.7140405039295408, "learning_rate": 5.142719214218111e-06, "loss": 0.2782, "step": 14645 }, { "epoch": 1.0125129623228482, "grad_norm": 0.575214185731388, "learning_rate": 5.14215968218227e-06, "loss": 0.1989, "step": 14646 }, { "epoch": 1.0125820947113722, "grad_norm": 0.6679307014002308, "learning_rate": 5.1416001483647105e-06, "loss": 0.3064, "step": 14647 }, { "epoch": 1.0126512270998962, "grad_norm": 0.6391809733996595, "learning_rate": 5.141040612772446e-06, "loss": 0.3025, "step": 14648 }, { "epoch": 1.0127203594884204, "grad_norm": 0.6455308125089451, "learning_rate": 5.140481075412489e-06, "loss": 0.2651, "step": 14649 }, { "epoch": 1.0127894918769444, "grad_norm": 0.6292904290113153, "learning_rate": 5.139921536291853e-06, "loss": 0.2723, "step": 14650 }, { "epoch": 1.0128586242654685, "grad_norm": 1.2768989487749072, "learning_rate": 5.139361995417549e-06, "loss": 0.3523, "step": 14651 }, { "epoch": 1.0129277566539925, "grad_norm": 0.6992514119331397, "learning_rate": 5.138802452796592e-06, "loss": 0.3567, "step": 14652 }, { "epoch": 1.0129968890425165, "grad_norm": 0.6650795636893776, "learning_rate": 5.1382429084359945e-06, "loss": 0.3289, "step": 14653 }, { "epoch": 1.0130660214310405, "grad_norm": 0.6592754555941754, "learning_rate": 5.137683362342767e-06, "loss": 0.2315, "step": 14654 }, { "epoch": 1.0131351538195645, "grad_norm": 0.8406222359155034, "learning_rate": 5.137123814523925e-06, "loss": 0.3888, "step": 14655 }, { "epoch": 1.0132042862080886, "grad_norm": 0.6603196878241784, "learning_rate": 5.1365642649864826e-06, "loss": 0.2763, "step": 14656 }, { "epoch": 1.0132734185966126, "grad_norm": 0.5654182006671216, "learning_rate": 5.136004713737449e-06, "loss": 0.1986, "step": 14657 }, { "epoch": 1.0133425509851366, "grad_norm": 0.8220161578498281, "learning_rate": 5.135445160783839e-06, "loss": 0.3111, "step": 14658 }, { "epoch": 1.0134116833736606, "grad_norm": 0.6882128269837785, "learning_rate": 5.134885606132666e-06, "loss": 0.3856, "step": 14659 }, { "epoch": 1.0134808157621846, "grad_norm": 0.6693872395778313, "learning_rate": 5.134326049790942e-06, "loss": 0.2718, "step": 14660 }, { "epoch": 1.0135499481507086, "grad_norm": 0.8063727825000061, "learning_rate": 5.13376649176568e-06, "loss": 0.2868, "step": 14661 }, { "epoch": 1.0136190805392327, "grad_norm": 0.7454365002297716, "learning_rate": 5.133206932063896e-06, "loss": 0.3611, "step": 14662 }, { "epoch": 1.0136882129277567, "grad_norm": 0.8433768107072842, "learning_rate": 5.1326473706926005e-06, "loss": 0.3365, "step": 14663 }, { "epoch": 1.0137573453162807, "grad_norm": 0.5881971021213519, "learning_rate": 5.132087807658807e-06, "loss": 0.2226, "step": 14664 }, { "epoch": 1.0138264777048047, "grad_norm": 0.7478775013109101, "learning_rate": 5.131528242969527e-06, "loss": 0.3109, "step": 14665 }, { "epoch": 1.0138956100933287, "grad_norm": 0.6987993517540252, "learning_rate": 5.130968676631776e-06, "loss": 0.3751, "step": 14666 }, { "epoch": 1.0139647424818528, "grad_norm": 0.6304176755416642, "learning_rate": 5.1304091086525665e-06, "loss": 0.2307, "step": 14667 }, { "epoch": 1.0140338748703768, "grad_norm": 0.7116539886051982, "learning_rate": 5.129849539038912e-06, "loss": 0.2501, "step": 14668 }, { "epoch": 1.0141030072589008, "grad_norm": 0.8352173947495058, "learning_rate": 5.1292899677978244e-06, "loss": 0.2718, "step": 14669 }, { "epoch": 1.0141721396474248, "grad_norm": 0.6837702003903608, "learning_rate": 5.128730394936319e-06, "loss": 0.3658, "step": 14670 }, { "epoch": 1.0142412720359488, "grad_norm": 0.7299482003083755, "learning_rate": 5.128170820461406e-06, "loss": 0.269, "step": 14671 }, { "epoch": 1.0143104044244728, "grad_norm": 0.7301907706690072, "learning_rate": 5.127611244380102e-06, "loss": 0.2877, "step": 14672 }, { "epoch": 1.0143795368129969, "grad_norm": 0.7485819215920636, "learning_rate": 5.127051666699418e-06, "loss": 0.3577, "step": 14673 }, { "epoch": 1.0144486692015209, "grad_norm": 0.6586931177374311, "learning_rate": 5.126492087426369e-06, "loss": 0.2618, "step": 14674 }, { "epoch": 1.014517801590045, "grad_norm": 0.6151559745036576, "learning_rate": 5.1259325065679655e-06, "loss": 0.2093, "step": 14675 }, { "epoch": 1.014586933978569, "grad_norm": 0.6630455967668071, "learning_rate": 5.125372924131225e-06, "loss": 0.3083, "step": 14676 }, { "epoch": 1.014656066367093, "grad_norm": 0.6447854491039422, "learning_rate": 5.124813340123157e-06, "loss": 0.3494, "step": 14677 }, { "epoch": 1.014725198755617, "grad_norm": 0.6257572714322868, "learning_rate": 5.124253754550776e-06, "loss": 0.2405, "step": 14678 }, { "epoch": 1.014794331144141, "grad_norm": 0.5841570053497502, "learning_rate": 5.1236941674210985e-06, "loss": 0.2336, "step": 14679 }, { "epoch": 1.014863463532665, "grad_norm": 0.6580961812954215, "learning_rate": 5.1231345787411335e-06, "loss": 0.3483, "step": 14680 }, { "epoch": 1.014932595921189, "grad_norm": 0.7379819712241094, "learning_rate": 5.1225749885178965e-06, "loss": 0.2668, "step": 14681 }, { "epoch": 1.015001728309713, "grad_norm": 0.7446133564422164, "learning_rate": 5.122015396758401e-06, "loss": 0.2747, "step": 14682 }, { "epoch": 1.015070860698237, "grad_norm": 0.6444652860085662, "learning_rate": 5.12145580346966e-06, "loss": 0.2848, "step": 14683 }, { "epoch": 1.015139993086761, "grad_norm": 0.6999185130429273, "learning_rate": 5.120896208658688e-06, "loss": 0.4498, "step": 14684 }, { "epoch": 1.015209125475285, "grad_norm": 0.6347813636749847, "learning_rate": 5.120336612332497e-06, "loss": 0.2378, "step": 14685 }, { "epoch": 1.015278257863809, "grad_norm": 0.6109108877797511, "learning_rate": 5.119777014498101e-06, "loss": 0.2432, "step": 14686 }, { "epoch": 1.0153473902523331, "grad_norm": 0.655080932624445, "learning_rate": 5.119217415162515e-06, "loss": 0.345, "step": 14687 }, { "epoch": 1.0154165226408571, "grad_norm": 0.7323287536243333, "learning_rate": 5.118657814332751e-06, "loss": 0.2818, "step": 14688 }, { "epoch": 1.0154856550293814, "grad_norm": 0.6477067763696032, "learning_rate": 5.118098212015822e-06, "loss": 0.2717, "step": 14689 }, { "epoch": 1.0155547874179054, "grad_norm": 1.3783063891461833, "learning_rate": 5.117538608218744e-06, "loss": 0.2822, "step": 14690 }, { "epoch": 1.0156239198064294, "grad_norm": 0.9650351211498457, "learning_rate": 5.116979002948529e-06, "loss": 0.4145, "step": 14691 }, { "epoch": 1.0156930521949534, "grad_norm": 0.6187568028988313, "learning_rate": 5.116419396212191e-06, "loss": 0.2315, "step": 14692 }, { "epoch": 1.0157621845834774, "grad_norm": 0.7625385631419251, "learning_rate": 5.115859788016744e-06, "loss": 0.2445, "step": 14693 }, { "epoch": 1.0158313169720015, "grad_norm": 0.652087949538832, "learning_rate": 5.1153001783692e-06, "loss": 0.2971, "step": 14694 }, { "epoch": 1.0159004493605255, "grad_norm": 0.6471806882768649, "learning_rate": 5.114740567276576e-06, "loss": 0.3234, "step": 14695 }, { "epoch": 1.0159695817490495, "grad_norm": 0.671434075352476, "learning_rate": 5.1141809547458805e-06, "loss": 0.245, "step": 14696 }, { "epoch": 1.0160387141375735, "grad_norm": 0.7472685090998815, "learning_rate": 5.113621340784134e-06, "loss": 0.3927, "step": 14697 }, { "epoch": 1.0161078465260975, "grad_norm": 0.6479325333977859, "learning_rate": 5.113061725398345e-06, "loss": 0.4016, "step": 14698 }, { "epoch": 1.0161769789146216, "grad_norm": 0.6913270371140157, "learning_rate": 5.11250210859553e-06, "loss": 0.2533, "step": 14699 }, { "epoch": 1.0162461113031456, "grad_norm": 0.6439433120483163, "learning_rate": 5.111942490382701e-06, "loss": 0.2465, "step": 14700 }, { "epoch": 1.0163152436916696, "grad_norm": 0.7188718580053339, "learning_rate": 5.111382870766872e-06, "loss": 0.3681, "step": 14701 }, { "epoch": 1.0163843760801936, "grad_norm": 2.9452537199770767, "learning_rate": 5.110823249755059e-06, "loss": 0.2828, "step": 14702 }, { "epoch": 1.0164535084687176, "grad_norm": 0.6661029971604099, "learning_rate": 5.1102636273542735e-06, "loss": 0.2895, "step": 14703 }, { "epoch": 1.0165226408572416, "grad_norm": 0.7794485571740234, "learning_rate": 5.109704003571531e-06, "loss": 0.2349, "step": 14704 }, { "epoch": 1.0165917732457657, "grad_norm": 0.7125171773265144, "learning_rate": 5.109144378413844e-06, "loss": 0.4249, "step": 14705 }, { "epoch": 1.0166609056342897, "grad_norm": 0.6266081299123262, "learning_rate": 5.108584751888227e-06, "loss": 0.232, "step": 14706 }, { "epoch": 1.0167300380228137, "grad_norm": 0.7278834760380734, "learning_rate": 5.108025124001693e-06, "loss": 0.2953, "step": 14707 }, { "epoch": 1.0167991704113377, "grad_norm": 0.6527946019012212, "learning_rate": 5.1074654947612575e-06, "loss": 0.324, "step": 14708 }, { "epoch": 1.0168683027998617, "grad_norm": 0.6705673277482412, "learning_rate": 5.106905864173934e-06, "loss": 0.3403, "step": 14709 }, { "epoch": 1.0169374351883858, "grad_norm": 0.712145153923055, "learning_rate": 5.106346232246737e-06, "loss": 0.2173, "step": 14710 }, { "epoch": 1.0170065675769098, "grad_norm": 0.6015938797715412, "learning_rate": 5.105786598986678e-06, "loss": 0.2925, "step": 14711 }, { "epoch": 1.0170756999654338, "grad_norm": 0.661566121866132, "learning_rate": 5.105226964400774e-06, "loss": 0.3826, "step": 14712 }, { "epoch": 1.0171448323539578, "grad_norm": 0.7006273287180655, "learning_rate": 5.104667328496039e-06, "loss": 0.3253, "step": 14713 }, { "epoch": 1.0172139647424818, "grad_norm": 0.6244058705249489, "learning_rate": 5.104107691279484e-06, "loss": 0.2239, "step": 14714 }, { "epoch": 1.0172830971310058, "grad_norm": 0.663224460722399, "learning_rate": 5.103548052758125e-06, "loss": 0.26, "step": 14715 }, { "epoch": 1.0173522295195299, "grad_norm": 0.6047130261558153, "learning_rate": 5.1029884129389775e-06, "loss": 0.3286, "step": 14716 }, { "epoch": 1.0174213619080539, "grad_norm": 0.6198188951454099, "learning_rate": 5.102428771829052e-06, "loss": 0.2519, "step": 14717 }, { "epoch": 1.017490494296578, "grad_norm": 0.6575536807694692, "learning_rate": 5.101869129435365e-06, "loss": 0.2731, "step": 14718 }, { "epoch": 1.017559626685102, "grad_norm": 0.6812962848652577, "learning_rate": 5.101309485764931e-06, "loss": 0.3127, "step": 14719 }, { "epoch": 1.017628759073626, "grad_norm": 0.7313738944578584, "learning_rate": 5.100749840824763e-06, "loss": 0.2599, "step": 14720 }, { "epoch": 1.01769789146215, "grad_norm": 0.6327918818248077, "learning_rate": 5.100190194621877e-06, "loss": 0.2313, "step": 14721 }, { "epoch": 1.017767023850674, "grad_norm": 0.8189225295351535, "learning_rate": 5.099630547163284e-06, "loss": 0.2507, "step": 14722 }, { "epoch": 1.017836156239198, "grad_norm": 0.6928140542599545, "learning_rate": 5.099070898456e-06, "loss": 0.3836, "step": 14723 }, { "epoch": 1.017905288627722, "grad_norm": 0.7634786994914551, "learning_rate": 5.098511248507039e-06, "loss": 0.318, "step": 14724 }, { "epoch": 1.017974421016246, "grad_norm": 0.8210797992393016, "learning_rate": 5.097951597323415e-06, "loss": 0.2696, "step": 14725 }, { "epoch": 1.01804355340477, "grad_norm": 0.7057525940012743, "learning_rate": 5.097391944912144e-06, "loss": 0.2576, "step": 14726 }, { "epoch": 1.018112685793294, "grad_norm": 0.6796995117500472, "learning_rate": 5.096832291280238e-06, "loss": 0.3214, "step": 14727 }, { "epoch": 1.018181818181818, "grad_norm": 0.8531208846364187, "learning_rate": 5.096272636434713e-06, "loss": 0.2373, "step": 14728 }, { "epoch": 1.018250950570342, "grad_norm": 0.6959385050643487, "learning_rate": 5.09571298038258e-06, "loss": 0.2259, "step": 14729 }, { "epoch": 1.0183200829588663, "grad_norm": 0.6279623367862321, "learning_rate": 5.0951533231308566e-06, "loss": 0.3327, "step": 14730 }, { "epoch": 1.0183892153473904, "grad_norm": 0.6767715001517532, "learning_rate": 5.094593664686557e-06, "loss": 0.2877, "step": 14731 }, { "epoch": 1.0184583477359144, "grad_norm": 0.6096614281974456, "learning_rate": 5.094034005056693e-06, "loss": 0.2504, "step": 14732 }, { "epoch": 1.0185274801244384, "grad_norm": 0.6920026633062152, "learning_rate": 5.093474344248281e-06, "loss": 0.2945, "step": 14733 }, { "epoch": 1.0185966125129624, "grad_norm": 0.7964322279379794, "learning_rate": 5.092914682268337e-06, "loss": 0.3945, "step": 14734 }, { "epoch": 1.0186657449014864, "grad_norm": 0.6234400322287073, "learning_rate": 5.092355019123871e-06, "loss": 0.2557, "step": 14735 }, { "epoch": 1.0187348772900104, "grad_norm": 0.6449163101778019, "learning_rate": 5.0917953548219015e-06, "loss": 0.2219, "step": 14736 }, { "epoch": 1.0188040096785345, "grad_norm": 0.6706463105135513, "learning_rate": 5.09123568936944e-06, "loss": 0.3101, "step": 14737 }, { "epoch": 1.0188731420670585, "grad_norm": 0.7553223301618056, "learning_rate": 5.090676022773502e-06, "loss": 0.318, "step": 14738 }, { "epoch": 1.0189422744555825, "grad_norm": 0.6972648085849833, "learning_rate": 5.090116355041101e-06, "loss": 0.2722, "step": 14739 }, { "epoch": 1.0190114068441065, "grad_norm": 0.6472006897330527, "learning_rate": 5.0895566861792535e-06, "loss": 0.2616, "step": 14740 }, { "epoch": 1.0190805392326305, "grad_norm": 0.715190751637292, "learning_rate": 5.0889970161949735e-06, "loss": 0.3875, "step": 14741 }, { "epoch": 1.0191496716211546, "grad_norm": 0.7248384861756877, "learning_rate": 5.088437345095271e-06, "loss": 0.2546, "step": 14742 }, { "epoch": 1.0192188040096786, "grad_norm": 0.6112720801982735, "learning_rate": 5.087877672887168e-06, "loss": 0.2262, "step": 14743 }, { "epoch": 1.0192879363982026, "grad_norm": 0.7506852630508869, "learning_rate": 5.087317999577674e-06, "loss": 0.3568, "step": 14744 }, { "epoch": 1.0193570687867266, "grad_norm": 0.6897218614557776, "learning_rate": 5.086758325173805e-06, "loss": 0.2765, "step": 14745 }, { "epoch": 1.0194262011752506, "grad_norm": 0.7032203619227331, "learning_rate": 5.086198649682574e-06, "loss": 0.2632, "step": 14746 }, { "epoch": 1.0194953335637746, "grad_norm": 0.700849041772275, "learning_rate": 5.085638973110999e-06, "loss": 0.2656, "step": 14747 }, { "epoch": 1.0195644659522987, "grad_norm": 0.6535838172908732, "learning_rate": 5.085079295466091e-06, "loss": 0.33, "step": 14748 }, { "epoch": 1.0196335983408227, "grad_norm": 0.6134689054856519, "learning_rate": 5.084519616754865e-06, "loss": 0.2848, "step": 14749 }, { "epoch": 1.0197027307293467, "grad_norm": 0.6410642905736607, "learning_rate": 5.083959936984337e-06, "loss": 0.2521, "step": 14750 }, { "epoch": 1.0197718631178707, "grad_norm": 0.7175568403115378, "learning_rate": 5.083400256161521e-06, "loss": 0.3018, "step": 14751 }, { "epoch": 1.0198409955063947, "grad_norm": 0.6904227342668345, "learning_rate": 5.082840574293432e-06, "loss": 0.2986, "step": 14752 }, { "epoch": 1.0199101278949187, "grad_norm": 0.6459685799303096, "learning_rate": 5.082280891387085e-06, "loss": 0.2306, "step": 14753 }, { "epoch": 1.0199792602834428, "grad_norm": 0.6453861970113056, "learning_rate": 5.081721207449493e-06, "loss": 0.2919, "step": 14754 }, { "epoch": 1.0200483926719668, "grad_norm": 0.6892346527019908, "learning_rate": 5.081161522487671e-06, "loss": 0.35, "step": 14755 }, { "epoch": 1.0201175250604908, "grad_norm": 0.6416174100517744, "learning_rate": 5.080601836508635e-06, "loss": 0.2721, "step": 14756 }, { "epoch": 1.0201866574490148, "grad_norm": 0.9258223112163074, "learning_rate": 5.080042149519397e-06, "loss": 0.2258, "step": 14757 }, { "epoch": 1.0202557898375388, "grad_norm": 0.7467405682587127, "learning_rate": 5.0794824615269765e-06, "loss": 0.2822, "step": 14758 }, { "epoch": 1.0203249222260629, "grad_norm": 0.7021390231687918, "learning_rate": 5.078922772538384e-06, "loss": 0.3047, "step": 14759 }, { "epoch": 1.0203940546145869, "grad_norm": 0.7648087416559693, "learning_rate": 5.078363082560633e-06, "loss": 0.2597, "step": 14760 }, { "epoch": 1.020463187003111, "grad_norm": 0.7399046767849146, "learning_rate": 5.077803391600744e-06, "loss": 0.279, "step": 14761 }, { "epoch": 1.020532319391635, "grad_norm": 0.6707808914060351, "learning_rate": 5.077243699665727e-06, "loss": 0.3064, "step": 14762 }, { "epoch": 1.020601451780159, "grad_norm": 0.9255613132455155, "learning_rate": 5.076684006762599e-06, "loss": 0.2926, "step": 14763 }, { "epoch": 1.020670584168683, "grad_norm": 0.6331304472326567, "learning_rate": 5.076124312898372e-06, "loss": 0.2229, "step": 14764 }, { "epoch": 1.020739716557207, "grad_norm": 0.6476912745613355, "learning_rate": 5.075564618080064e-06, "loss": 0.2664, "step": 14765 }, { "epoch": 1.020808848945731, "grad_norm": 0.7050179554368307, "learning_rate": 5.075004922314687e-06, "loss": 0.3503, "step": 14766 }, { "epoch": 1.020877981334255, "grad_norm": 0.6101562236575749, "learning_rate": 5.074445225609258e-06, "loss": 0.2542, "step": 14767 }, { "epoch": 1.020947113722779, "grad_norm": 0.609385253127425, "learning_rate": 5.073885527970791e-06, "loss": 0.2699, "step": 14768 }, { "epoch": 1.021016246111303, "grad_norm": 0.6826342076074607, "learning_rate": 5.0733258294063e-06, "loss": 0.3034, "step": 14769 }, { "epoch": 1.0210853784998273, "grad_norm": 0.7456648293449426, "learning_rate": 5.072766129922803e-06, "loss": 0.3923, "step": 14770 }, { "epoch": 1.0211545108883513, "grad_norm": 0.6692685047932903, "learning_rate": 5.072206429527309e-06, "loss": 0.2817, "step": 14771 }, { "epoch": 1.0212236432768753, "grad_norm": 0.6257485879990252, "learning_rate": 5.071646728226839e-06, "loss": 0.242, "step": 14772 }, { "epoch": 1.0212927756653993, "grad_norm": 0.7331547791897439, "learning_rate": 5.071087026028404e-06, "loss": 0.3737, "step": 14773 }, { "epoch": 1.0213619080539234, "grad_norm": 0.6538953667767713, "learning_rate": 5.0705273229390205e-06, "loss": 0.2612, "step": 14774 }, { "epoch": 1.0214310404424474, "grad_norm": 0.6351758294476434, "learning_rate": 5.069967618965701e-06, "loss": 0.2537, "step": 14775 }, { "epoch": 1.0215001728309714, "grad_norm": 0.6212338072291259, "learning_rate": 5.069407914115464e-06, "loss": 0.3083, "step": 14776 }, { "epoch": 1.0215693052194954, "grad_norm": 0.6674942031924533, "learning_rate": 5.068848208395323e-06, "loss": 0.319, "step": 14777 }, { "epoch": 1.0216384376080194, "grad_norm": 0.5599466885105092, "learning_rate": 5.068288501812291e-06, "loss": 0.2383, "step": 14778 }, { "epoch": 1.0217075699965434, "grad_norm": 0.6310058724920247, "learning_rate": 5.0677287943733855e-06, "loss": 0.2531, "step": 14779 }, { "epoch": 1.0217767023850675, "grad_norm": 1.166030406709893, "learning_rate": 5.067169086085622e-06, "loss": 0.4105, "step": 14780 }, { "epoch": 1.0218458347735915, "grad_norm": 0.6652577816127236, "learning_rate": 5.066609376956012e-06, "loss": 0.2891, "step": 14781 }, { "epoch": 1.0219149671621155, "grad_norm": 1.326842843343465, "learning_rate": 5.066049666991571e-06, "loss": 0.2469, "step": 14782 }, { "epoch": 1.0219840995506395, "grad_norm": 0.7148734202566147, "learning_rate": 5.0654899561993175e-06, "loss": 0.3494, "step": 14783 }, { "epoch": 1.0220532319391635, "grad_norm": 0.6307877266022047, "learning_rate": 5.064930244586263e-06, "loss": 0.3773, "step": 14784 }, { "epoch": 1.0221223643276875, "grad_norm": 0.6901027060599519, "learning_rate": 5.064370532159424e-06, "loss": 0.2458, "step": 14785 }, { "epoch": 1.0221914967162116, "grad_norm": 0.6473658507781629, "learning_rate": 5.063810818925816e-06, "loss": 0.2798, "step": 14786 }, { "epoch": 1.0222606291047356, "grad_norm": 0.6723011126247734, "learning_rate": 5.063251104892453e-06, "loss": 0.3084, "step": 14787 }, { "epoch": 1.0223297614932596, "grad_norm": 0.6545604128860949, "learning_rate": 5.062691390066351e-06, "loss": 0.3129, "step": 14788 }, { "epoch": 1.0223988938817836, "grad_norm": 0.5969771526086953, "learning_rate": 5.062131674454523e-06, "loss": 0.261, "step": 14789 }, { "epoch": 1.0224680262703076, "grad_norm": 0.6759160301596254, "learning_rate": 5.061571958063986e-06, "loss": 0.2693, "step": 14790 }, { "epoch": 1.0225371586588317, "grad_norm": 0.7146121711198716, "learning_rate": 5.0610122409017545e-06, "loss": 0.3864, "step": 14791 }, { "epoch": 1.0226062910473557, "grad_norm": 0.6875474202374457, "learning_rate": 5.060452522974844e-06, "loss": 0.2534, "step": 14792 }, { "epoch": 1.0226754234358797, "grad_norm": 1.3603383534854183, "learning_rate": 5.0598928042902665e-06, "loss": 0.2442, "step": 14793 }, { "epoch": 1.0227445558244037, "grad_norm": 0.7803642306281697, "learning_rate": 5.059333084855042e-06, "loss": 0.306, "step": 14794 }, { "epoch": 1.0228136882129277, "grad_norm": 0.6809726049636345, "learning_rate": 5.058773364676183e-06, "loss": 0.3825, "step": 14795 }, { "epoch": 1.0228828206014517, "grad_norm": 0.7712728625327301, "learning_rate": 5.058213643760702e-06, "loss": 0.2266, "step": 14796 }, { "epoch": 1.0229519529899758, "grad_norm": 0.6914512958147451, "learning_rate": 5.057653922115619e-06, "loss": 0.2468, "step": 14797 }, { "epoch": 1.0230210853784998, "grad_norm": 0.7666457491139734, "learning_rate": 5.057094199747948e-06, "loss": 0.3556, "step": 14798 }, { "epoch": 1.0230902177670238, "grad_norm": 0.8099903957305742, "learning_rate": 5.056534476664703e-06, "loss": 0.2936, "step": 14799 }, { "epoch": 1.0231593501555478, "grad_norm": 0.7085154543969183, "learning_rate": 5.055974752872897e-06, "loss": 0.2274, "step": 14800 }, { "epoch": 1.0232284825440718, "grad_norm": 0.6172432418029644, "learning_rate": 5.05541502837955e-06, "loss": 0.2469, "step": 14801 }, { "epoch": 1.0232976149325959, "grad_norm": 0.7457169725306342, "learning_rate": 5.054855303191673e-06, "loss": 0.3645, "step": 14802 }, { "epoch": 1.0233667473211199, "grad_norm": 0.681243580414356, "learning_rate": 5.0542955773162825e-06, "loss": 0.2578, "step": 14803 }, { "epoch": 1.023435879709644, "grad_norm": 0.6607565727326147, "learning_rate": 5.053735850760395e-06, "loss": 0.2596, "step": 14804 }, { "epoch": 1.023505012098168, "grad_norm": 0.7410226317645094, "learning_rate": 5.053176123531024e-06, "loss": 0.359, "step": 14805 }, { "epoch": 1.023574144486692, "grad_norm": 0.6517957212381604, "learning_rate": 5.052616395635184e-06, "loss": 0.2834, "step": 14806 }, { "epoch": 1.023643276875216, "grad_norm": 0.633008357076515, "learning_rate": 5.052056667079892e-06, "loss": 0.2302, "step": 14807 }, { "epoch": 1.02371240926374, "grad_norm": 0.6656748620555317, "learning_rate": 5.051496937872164e-06, "loss": 0.3006, "step": 14808 }, { "epoch": 1.023781541652264, "grad_norm": 0.664965268925551, "learning_rate": 5.050937208019012e-06, "loss": 0.3486, "step": 14809 }, { "epoch": 1.023850674040788, "grad_norm": 1.083884937070767, "learning_rate": 5.050377477527453e-06, "loss": 0.2263, "step": 14810 }, { "epoch": 1.0239198064293122, "grad_norm": 0.7101833190921529, "learning_rate": 5.049817746404504e-06, "loss": 0.2601, "step": 14811 }, { "epoch": 1.0239889388178363, "grad_norm": 0.6480289248178837, "learning_rate": 5.049258014657177e-06, "loss": 0.395, "step": 14812 }, { "epoch": 1.0240580712063603, "grad_norm": 0.6976308994222415, "learning_rate": 5.048698282292488e-06, "loss": 0.2355, "step": 14813 }, { "epoch": 1.0241272035948843, "grad_norm": 1.7329579113195528, "learning_rate": 5.048138549317454e-06, "loss": 0.2927, "step": 14814 }, { "epoch": 1.0241963359834083, "grad_norm": 0.701175068397676, "learning_rate": 5.04757881573909e-06, "loss": 0.2909, "step": 14815 }, { "epoch": 1.0242654683719323, "grad_norm": 0.6829179161574621, "learning_rate": 5.04701908156441e-06, "loss": 0.3814, "step": 14816 }, { "epoch": 1.0243346007604563, "grad_norm": 0.6368746632460701, "learning_rate": 5.046459346800429e-06, "loss": 0.242, "step": 14817 }, { "epoch": 1.0244037331489804, "grad_norm": 0.6146762563205453, "learning_rate": 5.045899611454163e-06, "loss": 0.2555, "step": 14818 }, { "epoch": 1.0244728655375044, "grad_norm": 0.6880259165616501, "learning_rate": 5.0453398755326276e-06, "loss": 0.3605, "step": 14819 }, { "epoch": 1.0245419979260284, "grad_norm": 0.6244825615076218, "learning_rate": 5.044780139042838e-06, "loss": 0.2881, "step": 14820 }, { "epoch": 1.0246111303145524, "grad_norm": 0.6534574341976945, "learning_rate": 5.044220401991808e-06, "loss": 0.3119, "step": 14821 }, { "epoch": 1.0246802627030764, "grad_norm": 0.6659969025387931, "learning_rate": 5.043660664386556e-06, "loss": 0.3094, "step": 14822 }, { "epoch": 1.0247493950916005, "grad_norm": 0.7335193072104947, "learning_rate": 5.043100926234095e-06, "loss": 0.3458, "step": 14823 }, { "epoch": 1.0248185274801245, "grad_norm": 0.6536843360934548, "learning_rate": 5.0425411875414396e-06, "loss": 0.275, "step": 14824 }, { "epoch": 1.0248876598686485, "grad_norm": 0.6356432650815438, "learning_rate": 5.041981448315607e-06, "loss": 0.2469, "step": 14825 }, { "epoch": 1.0249567922571725, "grad_norm": 0.6566962167819953, "learning_rate": 5.041421708563613e-06, "loss": 0.3049, "step": 14826 }, { "epoch": 1.0250259246456965, "grad_norm": 0.6016150600762447, "learning_rate": 5.040861968292471e-06, "loss": 0.3331, "step": 14827 }, { "epoch": 1.0250950570342205, "grad_norm": 0.6323790981672908, "learning_rate": 5.040302227509196e-06, "loss": 0.2325, "step": 14828 }, { "epoch": 1.0251641894227446, "grad_norm": 0.7491785488405599, "learning_rate": 5.039742486220806e-06, "loss": 0.3097, "step": 14829 }, { "epoch": 1.0252333218112686, "grad_norm": 0.6308312417885418, "learning_rate": 5.0391827444343145e-06, "loss": 0.3438, "step": 14830 }, { "epoch": 1.0253024541997926, "grad_norm": 1.5392915257480306, "learning_rate": 5.038623002156736e-06, "loss": 0.2884, "step": 14831 }, { "epoch": 1.0253715865883166, "grad_norm": 0.6407365886685611, "learning_rate": 5.03806325939509e-06, "loss": 0.252, "step": 14832 }, { "epoch": 1.0254407189768406, "grad_norm": 0.6878818111620364, "learning_rate": 5.037503516156387e-06, "loss": 0.2927, "step": 14833 }, { "epoch": 1.0255098513653647, "grad_norm": 0.6386594654362971, "learning_rate": 5.036943772447644e-06, "loss": 0.3408, "step": 14834 }, { "epoch": 1.0255789837538887, "grad_norm": 0.7650610417603041, "learning_rate": 5.036384028275877e-06, "loss": 0.2521, "step": 14835 }, { "epoch": 1.0256481161424127, "grad_norm": 0.6589837258441906, "learning_rate": 5.0358242836481016e-06, "loss": 0.2492, "step": 14836 }, { "epoch": 1.0257172485309367, "grad_norm": 0.6694719236398274, "learning_rate": 5.035264538571333e-06, "loss": 0.3056, "step": 14837 }, { "epoch": 1.0257863809194607, "grad_norm": 0.7242444434518702, "learning_rate": 5.034704793052585e-06, "loss": 0.3525, "step": 14838 }, { "epoch": 1.0258555133079847, "grad_norm": 0.6483156230258742, "learning_rate": 5.034145047098875e-06, "loss": 0.2831, "step": 14839 }, { "epoch": 1.0259246456965088, "grad_norm": 0.6729618119671567, "learning_rate": 5.033585300717218e-06, "loss": 0.2896, "step": 14840 }, { "epoch": 1.0259937780850328, "grad_norm": 0.7076641922259721, "learning_rate": 5.03302555391463e-06, "loss": 0.4334, "step": 14841 }, { "epoch": 1.0260629104735568, "grad_norm": 0.6606993631453691, "learning_rate": 5.032465806698125e-06, "loss": 0.279, "step": 14842 }, { "epoch": 1.0261320428620808, "grad_norm": 1.0342622364566552, "learning_rate": 5.031906059074719e-06, "loss": 0.2276, "step": 14843 }, { "epoch": 1.0262011752506048, "grad_norm": 0.6608417218545399, "learning_rate": 5.031346311051427e-06, "loss": 0.3259, "step": 14844 }, { "epoch": 1.0262703076391289, "grad_norm": 0.6208604341282725, "learning_rate": 5.030786562635266e-06, "loss": 0.3306, "step": 14845 }, { "epoch": 1.0263394400276529, "grad_norm": 0.7095516622975784, "learning_rate": 5.030226813833249e-06, "loss": 0.2783, "step": 14846 }, { "epoch": 1.026408572416177, "grad_norm": 0.6581120806161065, "learning_rate": 5.0296670646523935e-06, "loss": 0.3141, "step": 14847 }, { "epoch": 1.026477704804701, "grad_norm": 0.6583708641642482, "learning_rate": 5.029107315099715e-06, "loss": 0.4175, "step": 14848 }, { "epoch": 1.026546837193225, "grad_norm": 1.2677604994067306, "learning_rate": 5.0285475651822254e-06, "loss": 0.2595, "step": 14849 }, { "epoch": 1.026615969581749, "grad_norm": 0.6632175481473502, "learning_rate": 5.027987814906946e-06, "loss": 0.2761, "step": 14850 }, { "epoch": 1.0266851019702732, "grad_norm": 0.8628793233490464, "learning_rate": 5.027428064280889e-06, "loss": 0.2674, "step": 14851 }, { "epoch": 1.0267542343587972, "grad_norm": 0.8094591303865476, "learning_rate": 5.026868313311069e-06, "loss": 0.3072, "step": 14852 }, { "epoch": 1.0268233667473212, "grad_norm": 0.6998662868163891, "learning_rate": 5.026308562004504e-06, "loss": 0.2345, "step": 14853 }, { "epoch": 1.0268924991358452, "grad_norm": 0.7776733403660502, "learning_rate": 5.025748810368208e-06, "loss": 0.2544, "step": 14854 }, { "epoch": 1.0269616315243693, "grad_norm": 0.6623617853755802, "learning_rate": 5.025189058409195e-06, "loss": 0.3574, "step": 14855 }, { "epoch": 1.0270307639128933, "grad_norm": 0.7109425869285583, "learning_rate": 5.024629306134482e-06, "loss": 0.2479, "step": 14856 }, { "epoch": 1.0270998963014173, "grad_norm": 0.6786465040478598, "learning_rate": 5.024069553551086e-06, "loss": 0.2531, "step": 14857 }, { "epoch": 1.0271690286899413, "grad_norm": 0.6264568946290872, "learning_rate": 5.023509800666022e-06, "loss": 0.2762, "step": 14858 }, { "epoch": 1.0272381610784653, "grad_norm": 1.3766532433300982, "learning_rate": 5.022950047486303e-06, "loss": 0.3106, "step": 14859 }, { "epoch": 1.0273072934669893, "grad_norm": 1.2410334934121185, "learning_rate": 5.0223902940189466e-06, "loss": 0.2602, "step": 14860 }, { "epoch": 1.0273764258555134, "grad_norm": 0.6974410889161048, "learning_rate": 5.021830540270968e-06, "loss": 0.2785, "step": 14861 }, { "epoch": 1.0274455582440374, "grad_norm": 0.9504813841631627, "learning_rate": 5.021270786249381e-06, "loss": 0.3994, "step": 14862 }, { "epoch": 1.0275146906325614, "grad_norm": 0.6552895673212719, "learning_rate": 5.020711031961205e-06, "loss": 0.3348, "step": 14863 }, { "epoch": 1.0275838230210854, "grad_norm": 0.6510986419963758, "learning_rate": 5.020151277413452e-06, "loss": 0.2199, "step": 14864 }, { "epoch": 1.0276529554096094, "grad_norm": 0.6335435591271198, "learning_rate": 5.019591522613139e-06, "loss": 0.2603, "step": 14865 }, { "epoch": 1.0277220877981335, "grad_norm": 0.6529448114473568, "learning_rate": 5.01903176756728e-06, "loss": 0.3532, "step": 14866 }, { "epoch": 1.0277912201866575, "grad_norm": 0.6668255871771286, "learning_rate": 5.018472012282894e-06, "loss": 0.2211, "step": 14867 }, { "epoch": 1.0278603525751815, "grad_norm": 0.571449838455986, "learning_rate": 5.017912256766994e-06, "loss": 0.224, "step": 14868 }, { "epoch": 1.0279294849637055, "grad_norm": 0.7035497618838394, "learning_rate": 5.017352501026594e-06, "loss": 0.2724, "step": 14869 }, { "epoch": 1.0279986173522295, "grad_norm": 0.7599640362417842, "learning_rate": 5.0167927450687115e-06, "loss": 0.3079, "step": 14870 }, { "epoch": 1.0280677497407535, "grad_norm": 0.6844394786262041, "learning_rate": 5.016232988900364e-06, "loss": 0.2518, "step": 14871 }, { "epoch": 1.0281368821292776, "grad_norm": 0.7007044889770784, "learning_rate": 5.0156732325285625e-06, "loss": 0.2751, "step": 14872 }, { "epoch": 1.0282060145178016, "grad_norm": 0.699772715628296, "learning_rate": 5.015113475960327e-06, "loss": 0.3212, "step": 14873 }, { "epoch": 1.0282751469063256, "grad_norm": 0.6926816384334241, "learning_rate": 5.014553719202669e-06, "loss": 0.2845, "step": 14874 }, { "epoch": 1.0283442792948496, "grad_norm": 0.6314671181049045, "learning_rate": 5.013993962262608e-06, "loss": 0.2364, "step": 14875 }, { "epoch": 1.0284134116833736, "grad_norm": 0.8507804040664788, "learning_rate": 5.013434205147157e-06, "loss": 0.2736, "step": 14876 }, { "epoch": 1.0284825440718977, "grad_norm": 0.7121023853016897, "learning_rate": 5.012874447863332e-06, "loss": 0.3127, "step": 14877 }, { "epoch": 1.0285516764604217, "grad_norm": 0.6391749766702295, "learning_rate": 5.012314690418149e-06, "loss": 0.2132, "step": 14878 }, { "epoch": 1.0286208088489457, "grad_norm": 0.7305465957977482, "learning_rate": 5.011754932818623e-06, "loss": 0.2552, "step": 14879 }, { "epoch": 1.0286899412374697, "grad_norm": 0.7201108402141926, "learning_rate": 5.011195175071771e-06, "loss": 0.3748, "step": 14880 }, { "epoch": 1.0287590736259937, "grad_norm": 0.7409230095991689, "learning_rate": 5.010635417184606e-06, "loss": 0.2898, "step": 14881 }, { "epoch": 1.0288282060145177, "grad_norm": 0.6502963884977545, "learning_rate": 5.010075659164145e-06, "loss": 0.2165, "step": 14882 }, { "epoch": 1.0288973384030418, "grad_norm": 0.7691026847552423, "learning_rate": 5.009515901017406e-06, "loss": 0.2892, "step": 14883 }, { "epoch": 1.0289664707915658, "grad_norm": 0.6790275835075413, "learning_rate": 5.008956142751399e-06, "loss": 0.3187, "step": 14884 }, { "epoch": 1.0290356031800898, "grad_norm": 0.7309575012197238, "learning_rate": 5.008396384373144e-06, "loss": 0.3188, "step": 14885 }, { "epoch": 1.0291047355686138, "grad_norm": 0.6416261218692481, "learning_rate": 5.007836625889656e-06, "loss": 0.3083, "step": 14886 }, { "epoch": 1.0291738679571378, "grad_norm": 0.6612476597687659, "learning_rate": 5.007276867307949e-06, "loss": 0.335, "step": 14887 }, { "epoch": 1.0292430003456619, "grad_norm": 0.7239270976596657, "learning_rate": 5.00671710863504e-06, "loss": 0.3135, "step": 14888 }, { "epoch": 1.0293121327341859, "grad_norm": 0.7077605204788597, "learning_rate": 5.0061573498779435e-06, "loss": 0.2663, "step": 14889 }, { "epoch": 1.0293812651227099, "grad_norm": 0.6602385376882344, "learning_rate": 5.005597591043676e-06, "loss": 0.262, "step": 14890 }, { "epoch": 1.029450397511234, "grad_norm": 0.6736278236856054, "learning_rate": 5.0050378321392534e-06, "loss": 0.3686, "step": 14891 }, { "epoch": 1.0295195298997581, "grad_norm": 0.6181729253873536, "learning_rate": 5.004478073171689e-06, "loss": 0.2469, "step": 14892 }, { "epoch": 1.0295886622882822, "grad_norm": 0.8249640311197032, "learning_rate": 5.003918314148001e-06, "loss": 0.2727, "step": 14893 }, { "epoch": 1.0296577946768062, "grad_norm": 0.7047636531721211, "learning_rate": 5.003358555075204e-06, "loss": 0.3292, "step": 14894 }, { "epoch": 1.0297269270653302, "grad_norm": 0.6818285703153951, "learning_rate": 5.002798795960312e-06, "loss": 0.3765, "step": 14895 }, { "epoch": 1.0297960594538542, "grad_norm": 0.6156759936987507, "learning_rate": 5.002239036810344e-06, "loss": 0.2534, "step": 14896 }, { "epoch": 1.0298651918423782, "grad_norm": 0.7318474392755231, "learning_rate": 5.001679277632312e-06, "loss": 0.2853, "step": 14897 }, { "epoch": 1.0299343242309023, "grad_norm": 0.8107707940590738, "learning_rate": 5.0011195184332355e-06, "loss": 0.3985, "step": 14898 }, { "epoch": 1.0300034566194263, "grad_norm": 0.6738718544732883, "learning_rate": 5.000559759220125e-06, "loss": 0.2426, "step": 14899 }, { "epoch": 1.0300725890079503, "grad_norm": 0.5913929421708951, "learning_rate": 5e-06, "loss": 0.2373, "step": 14900 }, { "epoch": 1.0301417213964743, "grad_norm": 0.7926926798029609, "learning_rate": 4.999440240779878e-06, "loss": 0.2883, "step": 14901 }, { "epoch": 1.0302108537849983, "grad_norm": 0.6421835477274797, "learning_rate": 4.998880481566766e-06, "loss": 0.2962, "step": 14902 }, { "epoch": 1.0302799861735223, "grad_norm": 0.6734431666469222, "learning_rate": 4.99832072236769e-06, "loss": 0.2707, "step": 14903 }, { "epoch": 1.0303491185620464, "grad_norm": 0.6159884262125661, "learning_rate": 4.997760963189658e-06, "loss": 0.2468, "step": 14904 }, { "epoch": 1.0304182509505704, "grad_norm": 0.6747136460143942, "learning_rate": 4.997201204039689e-06, "loss": 0.3574, "step": 14905 }, { "epoch": 1.0304873833390944, "grad_norm": 0.635158440771943, "learning_rate": 4.996641444924798e-06, "loss": 0.2785, "step": 14906 }, { "epoch": 1.0305565157276184, "grad_norm": 0.6154686228145023, "learning_rate": 4.9960816858520015e-06, "loss": 0.2466, "step": 14907 }, { "epoch": 1.0306256481161424, "grad_norm": 0.6711709571572952, "learning_rate": 4.995521926828313e-06, "loss": 0.3151, "step": 14908 }, { "epoch": 1.0306947805046665, "grad_norm": 0.6670371087452646, "learning_rate": 4.99496216786075e-06, "loss": 0.3994, "step": 14909 }, { "epoch": 1.0307639128931905, "grad_norm": 1.0449348921014063, "learning_rate": 4.994402408956325e-06, "loss": 0.2609, "step": 14910 }, { "epoch": 1.0308330452817145, "grad_norm": 0.6497005326005391, "learning_rate": 4.993842650122059e-06, "loss": 0.2126, "step": 14911 }, { "epoch": 1.0309021776702385, "grad_norm": 0.6664961753354476, "learning_rate": 4.993282891364962e-06, "loss": 0.3249, "step": 14912 }, { "epoch": 1.0309713100587625, "grad_norm": 0.919692399120053, "learning_rate": 4.992723132692053e-06, "loss": 0.2811, "step": 14913 }, { "epoch": 1.0310404424472865, "grad_norm": 0.6467101757738892, "learning_rate": 4.992163374110345e-06, "loss": 0.2553, "step": 14914 }, { "epoch": 1.0311095748358106, "grad_norm": 0.6997543583949042, "learning_rate": 4.9916036156268585e-06, "loss": 0.2579, "step": 14915 }, { "epoch": 1.0311787072243346, "grad_norm": 0.7126661261562118, "learning_rate": 4.991043857248603e-06, "loss": 0.3261, "step": 14916 }, { "epoch": 1.0312478396128586, "grad_norm": 0.6984493521869319, "learning_rate": 4.990484098982598e-06, "loss": 0.2381, "step": 14917 }, { "epoch": 1.0313169720013826, "grad_norm": 0.5710715954113047, "learning_rate": 4.9899243408358556e-06, "loss": 0.2179, "step": 14918 }, { "epoch": 1.0313861043899066, "grad_norm": 0.6698079831778846, "learning_rate": 4.989364582815396e-06, "loss": 0.3653, "step": 14919 }, { "epoch": 1.0314552367784307, "grad_norm": 0.7606765393958088, "learning_rate": 4.9888048249282305e-06, "loss": 0.3746, "step": 14920 }, { "epoch": 1.0315243691669547, "grad_norm": 0.6373389813535468, "learning_rate": 4.9882450671813795e-06, "loss": 0.2471, "step": 14921 }, { "epoch": 1.0315935015554787, "grad_norm": 0.8183133955519792, "learning_rate": 4.987685309581852e-06, "loss": 0.2404, "step": 14922 }, { "epoch": 1.0316626339440027, "grad_norm": 0.6966610201618733, "learning_rate": 4.987125552136669e-06, "loss": 0.3584, "step": 14923 }, { "epoch": 1.0317317663325267, "grad_norm": 0.6365377310136264, "learning_rate": 4.986565794852844e-06, "loss": 0.2898, "step": 14924 }, { "epoch": 1.0318008987210507, "grad_norm": 0.6854252537667919, "learning_rate": 4.986006037737394e-06, "loss": 0.2488, "step": 14925 }, { "epoch": 1.0318700311095748, "grad_norm": 0.6636695631996379, "learning_rate": 4.985446280797332e-06, "loss": 0.3563, "step": 14926 }, { "epoch": 1.0319391634980988, "grad_norm": 0.7573577809394126, "learning_rate": 4.984886524039675e-06, "loss": 0.4022, "step": 14927 }, { "epoch": 1.0320082958866228, "grad_norm": 0.6486454554184315, "learning_rate": 4.984326767471438e-06, "loss": 0.2702, "step": 14928 }, { "epoch": 1.0320774282751468, "grad_norm": 0.6669525266391882, "learning_rate": 4.9837670110996385e-06, "loss": 0.2866, "step": 14929 }, { "epoch": 1.0321465606636708, "grad_norm": 0.6374484154293596, "learning_rate": 4.983207254931289e-06, "loss": 0.3742, "step": 14930 }, { "epoch": 1.0322156930521948, "grad_norm": 0.6715702339266103, "learning_rate": 4.982647498973408e-06, "loss": 0.2851, "step": 14931 }, { "epoch": 1.032284825440719, "grad_norm": 0.7129279244350317, "learning_rate": 4.982087743233008e-06, "loss": 0.2513, "step": 14932 }, { "epoch": 1.032353957829243, "grad_norm": 0.9196896376573335, "learning_rate": 4.981527987717109e-06, "loss": 0.2635, "step": 14933 }, { "epoch": 1.0324230902177671, "grad_norm": 0.7487800458086, "learning_rate": 4.9809682324327205e-06, "loss": 0.3402, "step": 14934 }, { "epoch": 1.0324922226062911, "grad_norm": 0.6655350086848415, "learning_rate": 4.980408477386863e-06, "loss": 0.3347, "step": 14935 }, { "epoch": 1.0325613549948152, "grad_norm": 0.617243847537564, "learning_rate": 4.979848722586549e-06, "loss": 0.2766, "step": 14936 }, { "epoch": 1.0326304873833392, "grad_norm": 0.736335619789627, "learning_rate": 4.979288968038796e-06, "loss": 0.329, "step": 14937 }, { "epoch": 1.0326996197718632, "grad_norm": 0.7348848735937358, "learning_rate": 4.978729213750618e-06, "loss": 0.3041, "step": 14938 }, { "epoch": 1.0327687521603872, "grad_norm": 0.8610986942715063, "learning_rate": 4.978169459729034e-06, "loss": 0.2388, "step": 14939 }, { "epoch": 1.0328378845489112, "grad_norm": 0.6452213613140395, "learning_rate": 4.977609705981054e-06, "loss": 0.282, "step": 14940 }, { "epoch": 1.0329070169374353, "grad_norm": 0.6460379625932252, "learning_rate": 4.977049952513698e-06, "loss": 0.3131, "step": 14941 }, { "epoch": 1.0329761493259593, "grad_norm": 0.680942389909635, "learning_rate": 4.9764901993339786e-06, "loss": 0.2599, "step": 14942 }, { "epoch": 1.0330452817144833, "grad_norm": 0.7286995612060142, "learning_rate": 4.975930446448915e-06, "loss": 0.2127, "step": 14943 }, { "epoch": 1.0331144141030073, "grad_norm": 0.686032678318654, "learning_rate": 4.975370693865518e-06, "loss": 0.3394, "step": 14944 }, { "epoch": 1.0331835464915313, "grad_norm": 0.5805880927342353, "learning_rate": 4.974810941590806e-06, "loss": 0.2785, "step": 14945 }, { "epoch": 1.0332526788800553, "grad_norm": 0.6827957219297422, "learning_rate": 4.974251189631793e-06, "loss": 0.2519, "step": 14946 }, { "epoch": 1.0333218112685794, "grad_norm": 0.6123575552354547, "learning_rate": 4.973691437995498e-06, "loss": 0.2676, "step": 14947 }, { "epoch": 1.0333909436571034, "grad_norm": 0.9541565548347067, "learning_rate": 4.9731316866889315e-06, "loss": 0.3375, "step": 14948 }, { "epoch": 1.0334600760456274, "grad_norm": 0.6446656968560373, "learning_rate": 4.972571935719112e-06, "loss": 0.309, "step": 14949 }, { "epoch": 1.0335292084341514, "grad_norm": 0.6244284726662062, "learning_rate": 4.972012185093054e-06, "loss": 0.2148, "step": 14950 }, { "epoch": 1.0335983408226754, "grad_norm": 0.6667918022863548, "learning_rate": 4.971452434817775e-06, "loss": 0.2817, "step": 14951 }, { "epoch": 1.0336674732111995, "grad_norm": 0.6834397873248146, "learning_rate": 4.970892684900287e-06, "loss": 0.3537, "step": 14952 }, { "epoch": 1.0337366055997235, "grad_norm": 0.66839209245257, "learning_rate": 4.970332935347608e-06, "loss": 0.2388, "step": 14953 }, { "epoch": 1.0338057379882475, "grad_norm": 0.6567378039153975, "learning_rate": 4.969773186166752e-06, "loss": 0.2892, "step": 14954 }, { "epoch": 1.0338748703767715, "grad_norm": 0.6643410177187308, "learning_rate": 4.969213437364736e-06, "loss": 0.3207, "step": 14955 }, { "epoch": 1.0339440027652955, "grad_norm": 0.7391299651643038, "learning_rate": 4.968653688948573e-06, "loss": 0.3025, "step": 14956 }, { "epoch": 1.0340131351538195, "grad_norm": 0.5872771900443773, "learning_rate": 4.968093940925283e-06, "loss": 0.2337, "step": 14957 }, { "epoch": 1.0340822675423436, "grad_norm": 0.7037347871087419, "learning_rate": 4.967534193301876e-06, "loss": 0.3382, "step": 14958 }, { "epoch": 1.0341513999308676, "grad_norm": 0.746038550155939, "learning_rate": 4.9669744460853715e-06, "loss": 0.3367, "step": 14959 }, { "epoch": 1.0342205323193916, "grad_norm": 0.8603713297051591, "learning_rate": 4.966414699282782e-06, "loss": 0.2704, "step": 14960 }, { "epoch": 1.0342896647079156, "grad_norm": 0.728593477336165, "learning_rate": 4.965854952901126e-06, "loss": 0.3257, "step": 14961 }, { "epoch": 1.0343587970964396, "grad_norm": 0.6871289721279867, "learning_rate": 4.965295206947415e-06, "loss": 0.3684, "step": 14962 }, { "epoch": 1.0344279294849636, "grad_norm": 0.6693404195800404, "learning_rate": 4.964735461428669e-06, "loss": 0.2734, "step": 14963 }, { "epoch": 1.0344970618734877, "grad_norm": 0.6511184333297645, "learning_rate": 4.964175716351899e-06, "loss": 0.2533, "step": 14964 }, { "epoch": 1.0345661942620117, "grad_norm": 0.6931193544921287, "learning_rate": 4.963615971724124e-06, "loss": 0.2778, "step": 14965 }, { "epoch": 1.0346353266505357, "grad_norm": 0.6734459485591564, "learning_rate": 4.963056227552357e-06, "loss": 0.3836, "step": 14966 }, { "epoch": 1.0347044590390597, "grad_norm": 0.8523147736146621, "learning_rate": 4.9624964838436156e-06, "loss": 0.2604, "step": 14967 }, { "epoch": 1.0347735914275837, "grad_norm": 0.7147512327811776, "learning_rate": 4.961936740604913e-06, "loss": 0.3105, "step": 14968 }, { "epoch": 1.0348427238161078, "grad_norm": 0.6536712242879132, "learning_rate": 4.9613769978432644e-06, "loss": 0.3195, "step": 14969 }, { "epoch": 1.0349118562046318, "grad_norm": 0.7598786108517219, "learning_rate": 4.960817255565688e-06, "loss": 0.3303, "step": 14970 }, { "epoch": 1.0349809885931558, "grad_norm": 1.0096902797367913, "learning_rate": 4.960257513779195e-06, "loss": 0.2387, "step": 14971 }, { "epoch": 1.0350501209816798, "grad_norm": 0.6054092576765033, "learning_rate": 4.9596977724908066e-06, "loss": 0.2354, "step": 14972 }, { "epoch": 1.0351192533702038, "grad_norm": 0.7012178052232756, "learning_rate": 4.95913803170753e-06, "loss": 0.3138, "step": 14973 }, { "epoch": 1.035188385758728, "grad_norm": 0.7156358110890523, "learning_rate": 4.95857829143639e-06, "loss": 0.3185, "step": 14974 }, { "epoch": 1.035257518147252, "grad_norm": 0.688382735626771, "learning_rate": 4.958018551684394e-06, "loss": 0.2786, "step": 14975 }, { "epoch": 1.035326650535776, "grad_norm": 0.6249547200347896, "learning_rate": 4.957458812458562e-06, "loss": 0.308, "step": 14976 }, { "epoch": 1.0353957829243001, "grad_norm": 0.6923431630432799, "learning_rate": 4.956899073765907e-06, "loss": 0.3563, "step": 14977 }, { "epoch": 1.0354649153128241, "grad_norm": 0.6053962645089791, "learning_rate": 4.956339335613446e-06, "loss": 0.2326, "step": 14978 }, { "epoch": 1.0355340477013482, "grad_norm": 0.6951490114178901, "learning_rate": 4.955779598008193e-06, "loss": 0.2771, "step": 14979 }, { "epoch": 1.0356031800898722, "grad_norm": 0.6562474412526152, "learning_rate": 4.955219860957164e-06, "loss": 0.3293, "step": 14980 }, { "epoch": 1.0356723124783962, "grad_norm": 0.5998241282487667, "learning_rate": 4.954660124467373e-06, "loss": 0.2962, "step": 14981 }, { "epoch": 1.0357414448669202, "grad_norm": 0.6762048631738857, "learning_rate": 4.95410038854584e-06, "loss": 0.2419, "step": 14982 }, { "epoch": 1.0358105772554442, "grad_norm": 0.7017130229265953, "learning_rate": 4.953540653199574e-06, "loss": 0.2657, "step": 14983 }, { "epoch": 1.0358797096439682, "grad_norm": 0.6428688604976514, "learning_rate": 4.952980918435593e-06, "loss": 0.3872, "step": 14984 }, { "epoch": 1.0359488420324923, "grad_norm": 0.7282334506853506, "learning_rate": 4.9524211842609115e-06, "loss": 0.3162, "step": 14985 }, { "epoch": 1.0360179744210163, "grad_norm": 0.73326111695901, "learning_rate": 4.951861450682548e-06, "loss": 0.2245, "step": 14986 }, { "epoch": 1.0360871068095403, "grad_norm": 0.7330424311864533, "learning_rate": 4.951301717707514e-06, "loss": 0.3193, "step": 14987 }, { "epoch": 1.0361562391980643, "grad_norm": 0.645017665123307, "learning_rate": 4.950741985342826e-06, "loss": 0.2946, "step": 14988 }, { "epoch": 1.0362253715865883, "grad_norm": 0.6212965842938993, "learning_rate": 4.9501822535954976e-06, "loss": 0.2373, "step": 14989 }, { "epoch": 1.0362945039751124, "grad_norm": 0.6812065238438092, "learning_rate": 4.949622522472548e-06, "loss": 0.2729, "step": 14990 }, { "epoch": 1.0363636363636364, "grad_norm": 0.673539665320353, "learning_rate": 4.9490627919809895e-06, "loss": 0.3526, "step": 14991 }, { "epoch": 1.0364327687521604, "grad_norm": 0.7363090635682336, "learning_rate": 4.948503062127839e-06, "loss": 0.2486, "step": 14992 }, { "epoch": 1.0365019011406844, "grad_norm": 0.6001667410886399, "learning_rate": 4.94794333292011e-06, "loss": 0.2675, "step": 14993 }, { "epoch": 1.0365710335292084, "grad_norm": 0.7222637988345282, "learning_rate": 4.947383604364818e-06, "loss": 0.3217, "step": 14994 }, { "epoch": 1.0366401659177324, "grad_norm": 0.6687330002285382, "learning_rate": 4.946823876468977e-06, "loss": 0.2849, "step": 14995 }, { "epoch": 1.0367092983062565, "grad_norm": 0.6262719751087813, "learning_rate": 4.946264149239608e-06, "loss": 0.2717, "step": 14996 }, { "epoch": 1.0367784306947805, "grad_norm": 0.7114373155467105, "learning_rate": 4.945704422683718e-06, "loss": 0.2804, "step": 14997 }, { "epoch": 1.0368475630833045, "grad_norm": 0.6272935425081553, "learning_rate": 4.945144696808329e-06, "loss": 0.3005, "step": 14998 }, { "epoch": 1.0369166954718285, "grad_norm": 0.6876404493615844, "learning_rate": 4.9445849716204515e-06, "loss": 0.254, "step": 14999 }, { "epoch": 1.0369858278603525, "grad_norm": 0.7146962049571447, "learning_rate": 4.944025247127104e-06, "loss": 0.2521, "step": 15000 }, { "epoch": 1.0370549602488766, "grad_norm": 0.7305062389681258, "learning_rate": 4.9434655233352994e-06, "loss": 0.3031, "step": 15001 }, { "epoch": 1.0371240926374006, "grad_norm": 0.7201562344859407, "learning_rate": 4.942905800252053e-06, "loss": 0.3872, "step": 15002 }, { "epoch": 1.0371932250259246, "grad_norm": 0.7985267806800269, "learning_rate": 4.94234607788438e-06, "loss": 0.2532, "step": 15003 }, { "epoch": 1.0372623574144486, "grad_norm": 0.7474243085967262, "learning_rate": 4.9417863562392986e-06, "loss": 0.2914, "step": 15004 }, { "epoch": 1.0373314898029726, "grad_norm": 0.7072559742297738, "learning_rate": 4.941226635323819e-06, "loss": 0.3819, "step": 15005 }, { "epoch": 1.0374006221914966, "grad_norm": 0.6261104465888877, "learning_rate": 4.94066691514496e-06, "loss": 0.223, "step": 15006 }, { "epoch": 1.0374697545800207, "grad_norm": 0.656879865017942, "learning_rate": 4.9401071957097334e-06, "loss": 0.2462, "step": 15007 }, { "epoch": 1.0375388869685447, "grad_norm": 0.7622224890536532, "learning_rate": 4.939547477025159e-06, "loss": 0.3311, "step": 15008 }, { "epoch": 1.0376080193570687, "grad_norm": 0.6792800648673655, "learning_rate": 4.938987759098246e-06, "loss": 0.3717, "step": 15009 }, { "epoch": 1.0376771517455927, "grad_norm": 0.9025447108173759, "learning_rate": 4.938428041936016e-06, "loss": 0.2291, "step": 15010 }, { "epoch": 1.0377462841341167, "grad_norm": 0.6927199286157417, "learning_rate": 4.937868325545478e-06, "loss": 0.2342, "step": 15011 }, { "epoch": 1.0378154165226408, "grad_norm": 0.6542524340989245, "learning_rate": 4.93730860993365e-06, "loss": 0.3664, "step": 15012 }, { "epoch": 1.037884548911165, "grad_norm": 0.6848642861599382, "learning_rate": 4.936748895107547e-06, "loss": 0.2787, "step": 15013 }, { "epoch": 1.037953681299689, "grad_norm": 0.6169361050602351, "learning_rate": 4.9361891810741856e-06, "loss": 0.2458, "step": 15014 }, { "epoch": 1.038022813688213, "grad_norm": 0.6935318392592296, "learning_rate": 4.935629467840576e-06, "loss": 0.3104, "step": 15015 }, { "epoch": 1.038091946076737, "grad_norm": 0.6690918816648849, "learning_rate": 4.9350697554137375e-06, "loss": 0.4111, "step": 15016 }, { "epoch": 1.038161078465261, "grad_norm": 0.6392688869878232, "learning_rate": 4.9345100438006825e-06, "loss": 0.2398, "step": 15017 }, { "epoch": 1.038230210853785, "grad_norm": 0.6065060781779698, "learning_rate": 4.93395033300843e-06, "loss": 0.2043, "step": 15018 }, { "epoch": 1.038299343242309, "grad_norm": 0.6463629347008915, "learning_rate": 4.93339062304399e-06, "loss": 0.2927, "step": 15019 }, { "epoch": 1.0383684756308331, "grad_norm": 0.6198840335818144, "learning_rate": 4.932830913914381e-06, "loss": 0.3081, "step": 15020 }, { "epoch": 1.0384376080193571, "grad_norm": 0.7560643943274702, "learning_rate": 4.9322712056266145e-06, "loss": 0.2986, "step": 15021 }, { "epoch": 1.0385067404078812, "grad_norm": 0.6667110655835097, "learning_rate": 4.93171149818771e-06, "loss": 0.319, "step": 15022 }, { "epoch": 1.0385758727964052, "grad_norm": 0.6484413655199628, "learning_rate": 4.931151791604679e-06, "loss": 0.3117, "step": 15023 }, { "epoch": 1.0386450051849292, "grad_norm": 0.7514673810613339, "learning_rate": 4.930592085884537e-06, "loss": 0.2668, "step": 15024 }, { "epoch": 1.0387141375734532, "grad_norm": 0.7145247848534537, "learning_rate": 4.930032381034299e-06, "loss": 0.2463, "step": 15025 }, { "epoch": 1.0387832699619772, "grad_norm": 0.6385228489978976, "learning_rate": 4.929472677060981e-06, "loss": 0.3663, "step": 15026 }, { "epoch": 1.0388524023505012, "grad_norm": 0.675352965302485, "learning_rate": 4.928912973971597e-06, "loss": 0.3861, "step": 15027 }, { "epoch": 1.0389215347390253, "grad_norm": 0.6176770094606766, "learning_rate": 4.928353271773163e-06, "loss": 0.2551, "step": 15028 }, { "epoch": 1.0389906671275493, "grad_norm": 0.6272772054711747, "learning_rate": 4.927793570472691e-06, "loss": 0.2766, "step": 15029 }, { "epoch": 1.0390597995160733, "grad_norm": 0.6858660377836527, "learning_rate": 4.927233870077199e-06, "loss": 0.3568, "step": 15030 }, { "epoch": 1.0391289319045973, "grad_norm": 0.6719859301628571, "learning_rate": 4.9266741705937e-06, "loss": 0.2433, "step": 15031 }, { "epoch": 1.0391980642931213, "grad_norm": 0.7206080744055638, "learning_rate": 4.92611447202921e-06, "loss": 0.2302, "step": 15032 }, { "epoch": 1.0392671966816454, "grad_norm": 0.8471739826742138, "learning_rate": 4.925554774390744e-06, "loss": 0.2337, "step": 15033 }, { "epoch": 1.0393363290701694, "grad_norm": 0.6742863154139461, "learning_rate": 4.924995077685314e-06, "loss": 0.3218, "step": 15034 }, { "epoch": 1.0394054614586934, "grad_norm": 0.6735815782244374, "learning_rate": 4.924435381919939e-06, "loss": 0.2705, "step": 15035 }, { "epoch": 1.0394745938472174, "grad_norm": 0.7077813446566683, "learning_rate": 4.9238756871016295e-06, "loss": 0.2645, "step": 15036 }, { "epoch": 1.0395437262357414, "grad_norm": 0.8329616314739792, "learning_rate": 4.9233159932374045e-06, "loss": 0.3466, "step": 15037 }, { "epoch": 1.0396128586242654, "grad_norm": 0.6537392944671071, "learning_rate": 4.922756300334274e-06, "loss": 0.2644, "step": 15038 }, { "epoch": 1.0396819910127895, "grad_norm": 0.6440711346291648, "learning_rate": 4.922196608399259e-06, "loss": 0.2658, "step": 15039 }, { "epoch": 1.0397511234013135, "grad_norm": 0.7306058795476811, "learning_rate": 4.9216369174393674e-06, "loss": 0.2409, "step": 15040 }, { "epoch": 1.0398202557898375, "grad_norm": 0.6536618121704415, "learning_rate": 4.921077227461618e-06, "loss": 0.3409, "step": 15041 }, { "epoch": 1.0398893881783615, "grad_norm": 0.5986531012815972, "learning_rate": 4.920517538473025e-06, "loss": 0.2274, "step": 15042 }, { "epoch": 1.0399585205668855, "grad_norm": 0.7087329198213839, "learning_rate": 4.9199578504806045e-06, "loss": 0.2197, "step": 15043 }, { "epoch": 1.0400276529554096, "grad_norm": 0.7051853597835539, "learning_rate": 4.9193981634913664e-06, "loss": 0.3551, "step": 15044 }, { "epoch": 1.0400967853439336, "grad_norm": 0.6597928991942252, "learning_rate": 4.918838477512331e-06, "loss": 0.2511, "step": 15045 }, { "epoch": 1.0401659177324576, "grad_norm": 0.9145193026882029, "learning_rate": 4.918278792550509e-06, "loss": 0.2925, "step": 15046 }, { "epoch": 1.0402350501209816, "grad_norm": 0.6421617015345356, "learning_rate": 4.917719108612917e-06, "loss": 0.209, "step": 15047 }, { "epoch": 1.0403041825095056, "grad_norm": 0.7540493857049265, "learning_rate": 4.917159425706568e-06, "loss": 0.3238, "step": 15048 }, { "epoch": 1.0403733148980296, "grad_norm": 0.6611058323580989, "learning_rate": 4.9165997438384805e-06, "loss": 0.3195, "step": 15049 }, { "epoch": 1.0404424472865537, "grad_norm": 0.5509107386128856, "learning_rate": 4.916040063015664e-06, "loss": 0.2275, "step": 15050 }, { "epoch": 1.0405115796750777, "grad_norm": 0.6641911174394577, "learning_rate": 4.915480383245137e-06, "loss": 0.3421, "step": 15051 }, { "epoch": 1.0405807120636017, "grad_norm": 0.6652600227448214, "learning_rate": 4.914920704533911e-06, "loss": 0.3103, "step": 15052 }, { "epoch": 1.0406498444521257, "grad_norm": 0.652766422105029, "learning_rate": 4.914361026889004e-06, "loss": 0.2628, "step": 15053 }, { "epoch": 1.0407189768406497, "grad_norm": 0.6246355222171087, "learning_rate": 4.913801350317427e-06, "loss": 0.3113, "step": 15054 }, { "epoch": 1.040788109229174, "grad_norm": 0.6010293957369901, "learning_rate": 4.913241674826197e-06, "loss": 0.2829, "step": 15055 }, { "epoch": 1.040857241617698, "grad_norm": 0.6663703525933907, "learning_rate": 4.912682000422327e-06, "loss": 0.3154, "step": 15056 }, { "epoch": 1.040926374006222, "grad_norm": 0.6122738326853268, "learning_rate": 4.912122327112834e-06, "loss": 0.2372, "step": 15057 }, { "epoch": 1.040995506394746, "grad_norm": 0.6937357327962012, "learning_rate": 4.911562654904729e-06, "loss": 0.3186, "step": 15058 }, { "epoch": 1.04106463878327, "grad_norm": 0.7741293537767107, "learning_rate": 4.911002983805029e-06, "loss": 0.3666, "step": 15059 }, { "epoch": 1.041133771171794, "grad_norm": 0.6342440433497958, "learning_rate": 4.910443313820747e-06, "loss": 0.2458, "step": 15060 }, { "epoch": 1.041202903560318, "grad_norm": 0.6197461742017943, "learning_rate": 4.909883644958901e-06, "loss": 0.2596, "step": 15061 }, { "epoch": 1.041272035948842, "grad_norm": 0.6148240323996452, "learning_rate": 4.909323977226499e-06, "loss": 0.3, "step": 15062 }, { "epoch": 1.0413411683373661, "grad_norm": 0.6422542513284376, "learning_rate": 4.9087643106305625e-06, "loss": 0.343, "step": 15063 }, { "epoch": 1.0414103007258901, "grad_norm": 0.6430379359471713, "learning_rate": 4.9082046451781e-06, "loss": 0.2232, "step": 15064 }, { "epoch": 1.0414794331144142, "grad_norm": 0.7243317583578671, "learning_rate": 4.90764498087613e-06, "loss": 0.3403, "step": 15065 }, { "epoch": 1.0415485655029382, "grad_norm": 0.6487005360660058, "learning_rate": 4.907085317731664e-06, "loss": 0.3588, "step": 15066 }, { "epoch": 1.0416176978914622, "grad_norm": 0.7049086365965215, "learning_rate": 4.90652565575172e-06, "loss": 0.2539, "step": 15067 }, { "epoch": 1.0416868302799862, "grad_norm": 0.6803720231152323, "learning_rate": 4.905965994943308e-06, "loss": 0.2729, "step": 15068 }, { "epoch": 1.0417559626685102, "grad_norm": 0.6981891502564215, "learning_rate": 4.905406335313445e-06, "loss": 0.3997, "step": 15069 }, { "epoch": 1.0418250950570342, "grad_norm": 0.7066919747222686, "learning_rate": 4.904846676869143e-06, "loss": 0.3711, "step": 15070 }, { "epoch": 1.0418942274455583, "grad_norm": 0.623716388054509, "learning_rate": 4.904287019617422e-06, "loss": 0.2171, "step": 15071 }, { "epoch": 1.0419633598340823, "grad_norm": 0.7898551041142314, "learning_rate": 4.903727363565289e-06, "loss": 0.2666, "step": 15072 }, { "epoch": 1.0420324922226063, "grad_norm": 0.6438437092388388, "learning_rate": 4.903167708719764e-06, "loss": 0.3547, "step": 15073 }, { "epoch": 1.0421016246111303, "grad_norm": 0.6943941134267535, "learning_rate": 4.902608055087857e-06, "loss": 0.3458, "step": 15074 }, { "epoch": 1.0421707569996543, "grad_norm": 0.6004781848045209, "learning_rate": 4.902048402676586e-06, "loss": 0.1994, "step": 15075 }, { "epoch": 1.0422398893881784, "grad_norm": 0.6654734161460477, "learning_rate": 4.901488751492962e-06, "loss": 0.3277, "step": 15076 }, { "epoch": 1.0423090217767024, "grad_norm": 0.6561559537690357, "learning_rate": 4.900929101544002e-06, "loss": 0.3192, "step": 15077 }, { "epoch": 1.0423781541652264, "grad_norm": 0.7973106122156058, "learning_rate": 4.900369452836717e-06, "loss": 0.2955, "step": 15078 }, { "epoch": 1.0424472865537504, "grad_norm": 0.7336708824808192, "learning_rate": 4.899809805378125e-06, "loss": 0.2939, "step": 15079 }, { "epoch": 1.0425164189422744, "grad_norm": 0.7501435991674541, "learning_rate": 4.899250159175237e-06, "loss": 0.3297, "step": 15080 }, { "epoch": 1.0425855513307984, "grad_norm": 0.921041353381343, "learning_rate": 4.898690514235071e-06, "loss": 0.2928, "step": 15081 }, { "epoch": 1.0426546837193225, "grad_norm": 0.556628846510354, "learning_rate": 4.898130870564635e-06, "loss": 0.2398, "step": 15082 }, { "epoch": 1.0427238161078465, "grad_norm": 1.4547048838503853, "learning_rate": 4.89757122817095e-06, "loss": 0.2815, "step": 15083 }, { "epoch": 1.0427929484963705, "grad_norm": 0.6973097771683275, "learning_rate": 4.897011587061023e-06, "loss": 0.3194, "step": 15084 }, { "epoch": 1.0428620808848945, "grad_norm": 0.7541156879206092, "learning_rate": 4.896451947241876e-06, "loss": 0.2592, "step": 15085 }, { "epoch": 1.0429312132734185, "grad_norm": 0.6726204799589963, "learning_rate": 4.895892308720517e-06, "loss": 0.3086, "step": 15086 }, { "epoch": 1.0430003456619426, "grad_norm": 0.7354451475981653, "learning_rate": 4.8953326715039626e-06, "loss": 0.3689, "step": 15087 }, { "epoch": 1.0430694780504666, "grad_norm": 0.7238719169147207, "learning_rate": 4.894773035599225e-06, "loss": 0.2713, "step": 15088 }, { "epoch": 1.0431386104389906, "grad_norm": 0.6956186230916648, "learning_rate": 4.8942134010133224e-06, "loss": 0.2619, "step": 15089 }, { "epoch": 1.0432077428275146, "grad_norm": 0.7101348904951663, "learning_rate": 4.893653767753264e-06, "loss": 0.2874, "step": 15090 }, { "epoch": 1.0432768752160386, "grad_norm": 0.6472941234875808, "learning_rate": 4.8930941358260666e-06, "loss": 0.3538, "step": 15091 }, { "epoch": 1.0433460076045626, "grad_norm": 0.6558618105685525, "learning_rate": 4.8925345052387425e-06, "loss": 0.2839, "step": 15092 }, { "epoch": 1.0434151399930867, "grad_norm": 0.6939071746948717, "learning_rate": 4.891974875998308e-06, "loss": 0.2171, "step": 15093 }, { "epoch": 1.0434842723816107, "grad_norm": 0.7155424928180903, "learning_rate": 4.891415248111774e-06, "loss": 0.3603, "step": 15094 }, { "epoch": 1.043553404770135, "grad_norm": 0.9926854133566262, "learning_rate": 4.890855621586158e-06, "loss": 0.353, "step": 15095 }, { "epoch": 1.043622537158659, "grad_norm": 0.6762088827391984, "learning_rate": 4.89029599642847e-06, "loss": 0.2688, "step": 15096 }, { "epoch": 1.043691669547183, "grad_norm": 0.6871611329471646, "learning_rate": 4.889736372645727e-06, "loss": 0.2608, "step": 15097 }, { "epoch": 1.043760801935707, "grad_norm": 0.6661837771671819, "learning_rate": 4.889176750244941e-06, "loss": 0.3704, "step": 15098 }, { "epoch": 1.043829934324231, "grad_norm": 1.8375562112211057, "learning_rate": 4.888617129233129e-06, "loss": 0.2412, "step": 15099 }, { "epoch": 1.043899066712755, "grad_norm": 0.6156413120195605, "learning_rate": 4.888057509617301e-06, "loss": 0.2435, "step": 15100 }, { "epoch": 1.043968199101279, "grad_norm": 0.6764479009322169, "learning_rate": 4.887497891404471e-06, "loss": 0.3793, "step": 15101 }, { "epoch": 1.044037331489803, "grad_norm": 0.7131887335976069, "learning_rate": 4.886938274601657e-06, "loss": 0.3549, "step": 15102 }, { "epoch": 1.044106463878327, "grad_norm": 0.6481176803324473, "learning_rate": 4.886378659215868e-06, "loss": 0.2574, "step": 15103 }, { "epoch": 1.044175596266851, "grad_norm": 0.6022728083401022, "learning_rate": 4.88581904525412e-06, "loss": 0.2565, "step": 15104 }, { "epoch": 1.044244728655375, "grad_norm": 0.6725392556571979, "learning_rate": 4.885259432723426e-06, "loss": 0.3032, "step": 15105 }, { "epoch": 1.0443138610438991, "grad_norm": 0.6845192940841862, "learning_rate": 4.8846998216308025e-06, "loss": 0.2641, "step": 15106 }, { "epoch": 1.0443829934324231, "grad_norm": 0.6705413875471087, "learning_rate": 4.8841402119832585e-06, "loss": 0.2827, "step": 15107 }, { "epoch": 1.0444521258209472, "grad_norm": 0.7244148414471268, "learning_rate": 4.883580603787811e-06, "loss": 0.3008, "step": 15108 }, { "epoch": 1.0445212582094712, "grad_norm": 0.6353422290664336, "learning_rate": 4.883020997051472e-06, "loss": 0.3783, "step": 15109 }, { "epoch": 1.0445903905979952, "grad_norm": 0.7978031452256055, "learning_rate": 4.882461391781259e-06, "loss": 0.2183, "step": 15110 }, { "epoch": 1.0446595229865192, "grad_norm": 0.7159817766195193, "learning_rate": 4.881901787984179e-06, "loss": 0.218, "step": 15111 }, { "epoch": 1.0447286553750432, "grad_norm": 2.969469537395377, "learning_rate": 4.881342185667251e-06, "loss": 0.3655, "step": 15112 }, { "epoch": 1.0447977877635672, "grad_norm": 0.8838355918029419, "learning_rate": 4.880782584837486e-06, "loss": 0.2717, "step": 15113 }, { "epoch": 1.0448669201520913, "grad_norm": 0.6484877746523939, "learning_rate": 4.880222985501901e-06, "loss": 0.2549, "step": 15114 }, { "epoch": 1.0449360525406153, "grad_norm": 0.6378705493618108, "learning_rate": 4.879663387667504e-06, "loss": 0.2468, "step": 15115 }, { "epoch": 1.0450051849291393, "grad_norm": 0.7169938394483447, "learning_rate": 4.879103791341315e-06, "loss": 0.4548, "step": 15116 }, { "epoch": 1.0450743173176633, "grad_norm": 0.7176623150568717, "learning_rate": 4.878544196530341e-06, "loss": 0.3167, "step": 15117 }, { "epoch": 1.0451434497061873, "grad_norm": 0.6708701091672171, "learning_rate": 4.877984603241601e-06, "loss": 0.2366, "step": 15118 }, { "epoch": 1.0452125820947114, "grad_norm": 0.738698270707258, "learning_rate": 4.877425011482104e-06, "loss": 0.3781, "step": 15119 }, { "epoch": 1.0452817144832354, "grad_norm": 0.8102532416714457, "learning_rate": 4.876865421258869e-06, "loss": 0.3043, "step": 15120 }, { "epoch": 1.0453508468717594, "grad_norm": 0.654265576377795, "learning_rate": 4.876305832578904e-06, "loss": 0.2315, "step": 15121 }, { "epoch": 1.0454199792602834, "grad_norm": 0.7553041275287319, "learning_rate": 4.875746245449225e-06, "loss": 0.2868, "step": 15122 }, { "epoch": 1.0454891116488074, "grad_norm": 0.5931551277069144, "learning_rate": 4.875186659876845e-06, "loss": 0.3787, "step": 15123 }, { "epoch": 1.0455582440373314, "grad_norm": 0.6271828033002134, "learning_rate": 4.874627075868778e-06, "loss": 0.2819, "step": 15124 }, { "epoch": 1.0456273764258555, "grad_norm": 0.648410012598476, "learning_rate": 4.874067493432036e-06, "loss": 0.2519, "step": 15125 }, { "epoch": 1.0456965088143795, "grad_norm": 0.625904734745484, "learning_rate": 4.873507912573634e-06, "loss": 0.3225, "step": 15126 }, { "epoch": 1.0457656412029035, "grad_norm": 1.3221178630451456, "learning_rate": 4.8729483333005835e-06, "loss": 0.3503, "step": 15127 }, { "epoch": 1.0458347735914275, "grad_norm": 0.6882294495110718, "learning_rate": 4.872388755619901e-06, "loss": 0.212, "step": 15128 }, { "epoch": 1.0459039059799515, "grad_norm": 0.8374541771515329, "learning_rate": 4.871829179538596e-06, "loss": 0.2914, "step": 15129 }, { "epoch": 1.0459730383684755, "grad_norm": 0.7049154013776806, "learning_rate": 4.871269605063683e-06, "loss": 0.4192, "step": 15130 }, { "epoch": 1.0460421707569996, "grad_norm": 0.6639077530239468, "learning_rate": 4.870710032202176e-06, "loss": 0.331, "step": 15131 }, { "epoch": 1.0461113031455236, "grad_norm": 0.6319741190066056, "learning_rate": 4.870150460961091e-06, "loss": 0.2138, "step": 15132 }, { "epoch": 1.0461804355340476, "grad_norm": 0.7064385774434814, "learning_rate": 4.8695908913474335e-06, "loss": 0.3537, "step": 15133 }, { "epoch": 1.0462495679225716, "grad_norm": 0.6541201997417765, "learning_rate": 4.869031323368226e-06, "loss": 0.3929, "step": 15134 }, { "epoch": 1.0463187003110956, "grad_norm": 0.6426546829612664, "learning_rate": 4.868471757030474e-06, "loss": 0.2757, "step": 15135 }, { "epoch": 1.0463878326996197, "grad_norm": 0.71174334275, "learning_rate": 4.867912192341195e-06, "loss": 0.2888, "step": 15136 }, { "epoch": 1.046456965088144, "grad_norm": 0.6763196269939561, "learning_rate": 4.8673526293073995e-06, "loss": 0.3086, "step": 15137 }, { "epoch": 1.046526097476668, "grad_norm": 0.6895481717354854, "learning_rate": 4.866793067936105e-06, "loss": 0.2909, "step": 15138 }, { "epoch": 1.046595229865192, "grad_norm": 0.6225831979909028, "learning_rate": 4.866233508234319e-06, "loss": 0.2715, "step": 15139 }, { "epoch": 1.046664362253716, "grad_norm": 0.7359145466104621, "learning_rate": 4.865673950209059e-06, "loss": 0.229, "step": 15140 }, { "epoch": 1.04673349464224, "grad_norm": 0.7205255046147188, "learning_rate": 4.865114393867335e-06, "loss": 0.3844, "step": 15141 }, { "epoch": 1.046802627030764, "grad_norm": 0.6896852607669606, "learning_rate": 4.864554839216163e-06, "loss": 0.2585, "step": 15142 }, { "epoch": 1.046871759419288, "grad_norm": 0.6049792432389658, "learning_rate": 4.863995286262553e-06, "loss": 0.2834, "step": 15143 }, { "epoch": 1.046940891807812, "grad_norm": 0.6977607387892762, "learning_rate": 4.86343573501352e-06, "loss": 0.3464, "step": 15144 }, { "epoch": 1.047010024196336, "grad_norm": 0.7121770440970072, "learning_rate": 4.862876185476075e-06, "loss": 0.2797, "step": 15145 }, { "epoch": 1.04707915658486, "grad_norm": 0.5922187263140791, "learning_rate": 4.862316637657234e-06, "loss": 0.2145, "step": 15146 }, { "epoch": 1.047148288973384, "grad_norm": 0.6463523741092024, "learning_rate": 4.861757091564008e-06, "loss": 0.3096, "step": 15147 }, { "epoch": 1.047217421361908, "grad_norm": 0.7070889814692056, "learning_rate": 4.86119754720341e-06, "loss": 0.3983, "step": 15148 }, { "epoch": 1.0472865537504321, "grad_norm": 0.689450450460216, "learning_rate": 4.860638004582451e-06, "loss": 0.2728, "step": 15149 }, { "epoch": 1.0473556861389561, "grad_norm": 1.28346750028829, "learning_rate": 4.860078463708149e-06, "loss": 0.2596, "step": 15150 }, { "epoch": 1.0474248185274802, "grad_norm": 0.7082133473337926, "learning_rate": 4.859518924587512e-06, "loss": 0.3065, "step": 15151 }, { "epoch": 1.0474939509160042, "grad_norm": 0.6557514065965523, "learning_rate": 4.858959387227556e-06, "loss": 0.3486, "step": 15152 }, { "epoch": 1.0475630833045282, "grad_norm": 0.5479512552911042, "learning_rate": 4.85839985163529e-06, "loss": 0.2499, "step": 15153 }, { "epoch": 1.0476322156930522, "grad_norm": 0.601863419398347, "learning_rate": 4.8578403178177315e-06, "loss": 0.2334, "step": 15154 }, { "epoch": 1.0477013480815762, "grad_norm": 0.6898639251702688, "learning_rate": 4.857280785781889e-06, "loss": 0.4109, "step": 15155 }, { "epoch": 1.0477704804701002, "grad_norm": 0.6378998094640133, "learning_rate": 4.8567212555347795e-06, "loss": 0.2588, "step": 15156 }, { "epoch": 1.0478396128586243, "grad_norm": 0.6502831782371884, "learning_rate": 4.856161727083411e-06, "loss": 0.2519, "step": 15157 }, { "epoch": 1.0479087452471483, "grad_norm": 0.6851161511898095, "learning_rate": 4.8556022004348e-06, "loss": 0.2896, "step": 15158 }, { "epoch": 1.0479778776356723, "grad_norm": 0.7185387372355799, "learning_rate": 4.855042675595957e-06, "loss": 0.3921, "step": 15159 }, { "epoch": 1.0480470100241963, "grad_norm": 0.583892938221076, "learning_rate": 4.854483152573897e-06, "loss": 0.2804, "step": 15160 }, { "epoch": 1.0481161424127203, "grad_norm": 0.6247926894504033, "learning_rate": 4.853923631375629e-06, "loss": 0.2316, "step": 15161 }, { "epoch": 1.0481852748012443, "grad_norm": 0.6935494074118177, "learning_rate": 4.853364112008168e-06, "loss": 0.3346, "step": 15162 }, { "epoch": 1.0482544071897684, "grad_norm": 0.6756970183753052, "learning_rate": 4.852804594478526e-06, "loss": 0.2526, "step": 15163 }, { "epoch": 1.0483235395782924, "grad_norm": 0.6017072079381014, "learning_rate": 4.8522450787937175e-06, "loss": 0.1777, "step": 15164 }, { "epoch": 1.0483926719668164, "grad_norm": 0.7028804637763585, "learning_rate": 4.8516855649607516e-06, "loss": 0.287, "step": 15165 }, { "epoch": 1.0484618043553404, "grad_norm": 0.6723755155524082, "learning_rate": 4.851126052986643e-06, "loss": 0.3298, "step": 15166 }, { "epoch": 1.0485309367438644, "grad_norm": 0.6046187636537411, "learning_rate": 4.850566542878405e-06, "loss": 0.229, "step": 15167 }, { "epoch": 1.0486000691323885, "grad_norm": 0.6655473250301764, "learning_rate": 4.850007034643045e-06, "loss": 0.2548, "step": 15168 }, { "epoch": 1.0486692015209125, "grad_norm": 0.8461281350848273, "learning_rate": 4.849447528287582e-06, "loss": 0.3358, "step": 15169 }, { "epoch": 1.0487383339094365, "grad_norm": 0.6700822381506286, "learning_rate": 4.848888023819025e-06, "loss": 0.3495, "step": 15170 }, { "epoch": 1.0488074662979605, "grad_norm": 0.5919402362733627, "learning_rate": 4.848328521244387e-06, "loss": 0.2053, "step": 15171 }, { "epoch": 1.0488765986864845, "grad_norm": 0.7277368784262337, "learning_rate": 4.847769020570679e-06, "loss": 0.3102, "step": 15172 }, { "epoch": 1.0489457310750085, "grad_norm": 0.6632960643154329, "learning_rate": 4.847209521804918e-06, "loss": 0.3665, "step": 15173 }, { "epoch": 1.0490148634635326, "grad_norm": 0.695251865480741, "learning_rate": 4.8466500249541094e-06, "loss": 0.2443, "step": 15174 }, { "epoch": 1.0490839958520566, "grad_norm": 0.7025286873484645, "learning_rate": 4.846090530025271e-06, "loss": 0.2664, "step": 15175 }, { "epoch": 1.0491531282405808, "grad_norm": 0.6007487887786546, "learning_rate": 4.845531037025411e-06, "loss": 0.2394, "step": 15176 }, { "epoch": 1.0492222606291048, "grad_norm": 0.6450170448374664, "learning_rate": 4.844971545961547e-06, "loss": 0.3352, "step": 15177 }, { "epoch": 1.0492913930176289, "grad_norm": 0.6461219811852345, "learning_rate": 4.844412056840686e-06, "loss": 0.2146, "step": 15178 }, { "epoch": 1.0493605254061529, "grad_norm": 0.6036595480923067, "learning_rate": 4.843852569669843e-06, "loss": 0.2367, "step": 15179 }, { "epoch": 1.049429657794677, "grad_norm": 0.6624673371480096, "learning_rate": 4.843293084456028e-06, "loss": 0.3187, "step": 15180 }, { "epoch": 1.049498790183201, "grad_norm": 0.603029495415665, "learning_rate": 4.842733601206257e-06, "loss": 0.2553, "step": 15181 }, { "epoch": 1.049567922571725, "grad_norm": 0.7073584605109051, "learning_rate": 4.842174119927537e-06, "loss": 0.1979, "step": 15182 }, { "epoch": 1.049637054960249, "grad_norm": 0.6179990780828507, "learning_rate": 4.841614640626884e-06, "loss": 0.3114, "step": 15183 }, { "epoch": 1.049706187348773, "grad_norm": 0.6766704265109424, "learning_rate": 4.8410551633113085e-06, "loss": 0.3403, "step": 15184 }, { "epoch": 1.049775319737297, "grad_norm": 0.7495454821143067, "learning_rate": 4.840495687987825e-06, "loss": 0.2784, "step": 15185 }, { "epoch": 1.049844452125821, "grad_norm": 0.6846436685271869, "learning_rate": 4.83993621466344e-06, "loss": 0.2294, "step": 15186 }, { "epoch": 1.049913584514345, "grad_norm": 0.643182227563531, "learning_rate": 4.839376743345171e-06, "loss": 0.3708, "step": 15187 }, { "epoch": 1.049982716902869, "grad_norm": 0.6646880263294422, "learning_rate": 4.8388172740400265e-06, "loss": 0.3026, "step": 15188 }, { "epoch": 1.050051849291393, "grad_norm": 0.6428972239237819, "learning_rate": 4.838257806755021e-06, "loss": 0.2247, "step": 15189 }, { "epoch": 1.050120981679917, "grad_norm": 0.8282456407681982, "learning_rate": 4.837698341497164e-06, "loss": 0.3028, "step": 15190 }, { "epoch": 1.050190114068441, "grad_norm": 1.3809955807104108, "learning_rate": 4.837138878273471e-06, "loss": 0.3623, "step": 15191 }, { "epoch": 1.0502592464569651, "grad_norm": 0.691563623689886, "learning_rate": 4.836579417090949e-06, "loss": 0.2729, "step": 15192 }, { "epoch": 1.0503283788454891, "grad_norm": 0.6634204315499721, "learning_rate": 4.836019957956614e-06, "loss": 0.2187, "step": 15193 }, { "epoch": 1.0503975112340131, "grad_norm": 0.7083856549896856, "learning_rate": 4.8354605008774744e-06, "loss": 0.3417, "step": 15194 }, { "epoch": 1.0504666436225372, "grad_norm": 0.6189736297347331, "learning_rate": 4.834901045860547e-06, "loss": 0.2732, "step": 15195 }, { "epoch": 1.0505357760110612, "grad_norm": 0.6573718120505944, "learning_rate": 4.834341592912837e-06, "loss": 0.2397, "step": 15196 }, { "epoch": 1.0506049083995852, "grad_norm": 0.6744980962274557, "learning_rate": 4.8337821420413615e-06, "loss": 0.2358, "step": 15197 }, { "epoch": 1.0506740407881092, "grad_norm": 0.6367120344009829, "learning_rate": 4.833222693253129e-06, "loss": 0.3238, "step": 15198 }, { "epoch": 1.0507431731766332, "grad_norm": 0.6986991676095605, "learning_rate": 4.832663246555154e-06, "loss": 0.3166, "step": 15199 }, { "epoch": 1.0508123055651573, "grad_norm": 0.6208985820387841, "learning_rate": 4.832103801954445e-06, "loss": 0.2223, "step": 15200 }, { "epoch": 1.0508814379536813, "grad_norm": 0.7220664899458069, "learning_rate": 4.831544359458016e-06, "loss": 0.3124, "step": 15201 }, { "epoch": 1.0509505703422053, "grad_norm": 0.7243421976187484, "learning_rate": 4.830984919072877e-06, "loss": 0.3854, "step": 15202 }, { "epoch": 1.0510197027307293, "grad_norm": 0.5669240335851383, "learning_rate": 4.830425480806042e-06, "loss": 0.2618, "step": 15203 }, { "epoch": 1.0510888351192533, "grad_norm": 0.646845248441143, "learning_rate": 4.829866044664519e-06, "loss": 0.2586, "step": 15204 }, { "epoch": 1.0511579675077773, "grad_norm": 0.7210916049861686, "learning_rate": 4.829306610655323e-06, "loss": 0.3824, "step": 15205 }, { "epoch": 1.0512270998963014, "grad_norm": 0.6720098801417737, "learning_rate": 4.828747178785464e-06, "loss": 0.2471, "step": 15206 }, { "epoch": 1.0512962322848254, "grad_norm": 0.7387842459198388, "learning_rate": 4.828187749061953e-06, "loss": 0.2281, "step": 15207 }, { "epoch": 1.0513653646733494, "grad_norm": 0.8180289417494662, "learning_rate": 4.827628321491801e-06, "loss": 0.334, "step": 15208 }, { "epoch": 1.0514344970618734, "grad_norm": 0.6926609241557582, "learning_rate": 4.827068896082023e-06, "loss": 0.3481, "step": 15209 }, { "epoch": 1.0515036294503974, "grad_norm": 0.6946611817653829, "learning_rate": 4.826509472839625e-06, "loss": 0.2689, "step": 15210 }, { "epoch": 1.0515727618389215, "grad_norm": 0.6727269594082972, "learning_rate": 4.825950051771623e-06, "loss": 0.3363, "step": 15211 }, { "epoch": 1.0516418942274455, "grad_norm": 0.6392832535640699, "learning_rate": 4.825390632885024e-06, "loss": 0.3621, "step": 15212 }, { "epoch": 1.0517110266159695, "grad_norm": 0.6807829506443303, "learning_rate": 4.824831216186846e-06, "loss": 0.3248, "step": 15213 }, { "epoch": 1.0517801590044935, "grad_norm": 0.6137169410175479, "learning_rate": 4.824271801684092e-06, "loss": 0.3128, "step": 15214 }, { "epoch": 1.0518492913930175, "grad_norm": 0.6803233388127945, "learning_rate": 4.82371238938378e-06, "loss": 0.2882, "step": 15215 }, { "epoch": 1.0519184237815415, "grad_norm": 0.6645137582802269, "learning_rate": 4.823152979292916e-06, "loss": 0.3501, "step": 15216 }, { "epoch": 1.0519875561700656, "grad_norm": 0.620455549593677, "learning_rate": 4.822593571418518e-06, "loss": 0.2603, "step": 15217 }, { "epoch": 1.0520566885585898, "grad_norm": 0.6697725877293926, "learning_rate": 4.82203416576759e-06, "loss": 0.2869, "step": 15218 }, { "epoch": 1.0521258209471138, "grad_norm": 0.6302660950948443, "learning_rate": 4.821474762347146e-06, "loss": 0.3104, "step": 15219 }, { "epoch": 1.0521949533356378, "grad_norm": 0.6409605669477128, "learning_rate": 4.820915361164197e-06, "loss": 0.3217, "step": 15220 }, { "epoch": 1.0522640857241619, "grad_norm": 0.6325630278127504, "learning_rate": 4.820355962225757e-06, "loss": 0.2135, "step": 15221 }, { "epoch": 1.0523332181126859, "grad_norm": 0.6748491309064826, "learning_rate": 4.819796565538831e-06, "loss": 0.3159, "step": 15222 }, { "epoch": 1.05240235050121, "grad_norm": 0.6738571194998237, "learning_rate": 4.819237171110436e-06, "loss": 0.3558, "step": 15223 }, { "epoch": 1.052471482889734, "grad_norm": 0.813711807588141, "learning_rate": 4.818677778947579e-06, "loss": 0.2766, "step": 15224 }, { "epoch": 1.052540615278258, "grad_norm": 0.6153143676557739, "learning_rate": 4.818118389057273e-06, "loss": 0.2559, "step": 15225 }, { "epoch": 1.052609747666782, "grad_norm": 0.7153300179059472, "learning_rate": 4.817559001446528e-06, "loss": 0.3162, "step": 15226 }, { "epoch": 1.052678880055306, "grad_norm": 0.6507631721197236, "learning_rate": 4.816999616122357e-06, "loss": 0.3284, "step": 15227 }, { "epoch": 1.05274801244383, "grad_norm": 0.6373778602671761, "learning_rate": 4.8164402330917674e-06, "loss": 0.2372, "step": 15228 }, { "epoch": 1.052817144832354, "grad_norm": 0.767531885713893, "learning_rate": 4.815880852361773e-06, "loss": 0.2336, "step": 15229 }, { "epoch": 1.052886277220878, "grad_norm": 0.6513005547823745, "learning_rate": 4.815321473939382e-06, "loss": 0.3817, "step": 15230 }, { "epoch": 1.052955409609402, "grad_norm": 2.369806768444723, "learning_rate": 4.814762097831609e-06, "loss": 0.2882, "step": 15231 }, { "epoch": 1.053024541997926, "grad_norm": 0.6159426051959614, "learning_rate": 4.814202724045462e-06, "loss": 0.2686, "step": 15232 }, { "epoch": 1.05309367438645, "grad_norm": 0.6681248076895574, "learning_rate": 4.8136433525879515e-06, "loss": 0.3424, "step": 15233 }, { "epoch": 1.053162806774974, "grad_norm": 0.6504010760732611, "learning_rate": 4.813083983466091e-06, "loss": 0.3608, "step": 15234 }, { "epoch": 1.053231939163498, "grad_norm": 0.6141793939173991, "learning_rate": 4.812524616686886e-06, "loss": 0.2436, "step": 15235 }, { "epoch": 1.0533010715520221, "grad_norm": 0.5952013246645649, "learning_rate": 4.811965252257352e-06, "loss": 0.2619, "step": 15236 }, { "epoch": 1.0533702039405461, "grad_norm": 0.708405599474026, "learning_rate": 4.811405890184497e-06, "loss": 0.3509, "step": 15237 }, { "epoch": 1.0534393363290702, "grad_norm": 0.793234981891013, "learning_rate": 4.810846530475334e-06, "loss": 0.3119, "step": 15238 }, { "epoch": 1.0535084687175942, "grad_norm": 0.6232160193314489, "learning_rate": 4.81028717313687e-06, "loss": 0.2157, "step": 15239 }, { "epoch": 1.0535776011061182, "grad_norm": 0.8305694474477575, "learning_rate": 4.809727818176121e-06, "loss": 0.3201, "step": 15240 }, { "epoch": 1.0536467334946422, "grad_norm": 0.6964543286715646, "learning_rate": 4.809168465600091e-06, "loss": 0.3935, "step": 15241 }, { "epoch": 1.0537158658831662, "grad_norm": 0.6670326397461169, "learning_rate": 4.808609115415796e-06, "loss": 0.2769, "step": 15242 }, { "epoch": 1.0537849982716903, "grad_norm": 0.6532866449044746, "learning_rate": 4.808049767630243e-06, "loss": 0.2831, "step": 15243 }, { "epoch": 1.0538541306602143, "grad_norm": 0.707641790759198, "learning_rate": 4.807490422250445e-06, "loss": 0.3013, "step": 15244 }, { "epoch": 1.0539232630487383, "grad_norm": 0.6468979032938901, "learning_rate": 4.806931079283409e-06, "loss": 0.3433, "step": 15245 }, { "epoch": 1.0539923954372623, "grad_norm": 0.5882201963927621, "learning_rate": 4.806371738736149e-06, "loss": 0.2436, "step": 15246 }, { "epoch": 1.0540615278257863, "grad_norm": 0.6644750612866432, "learning_rate": 4.805812400615672e-06, "loss": 0.3216, "step": 15247 }, { "epoch": 1.0541306602143103, "grad_norm": 0.7065415510621833, "learning_rate": 4.805253064928992e-06, "loss": 0.3729, "step": 15248 }, { "epoch": 1.0541997926028344, "grad_norm": 0.6023326989562341, "learning_rate": 4.804693731683116e-06, "loss": 0.2389, "step": 15249 }, { "epoch": 1.0542689249913584, "grad_norm": 0.6593840805604744, "learning_rate": 4.804134400885057e-06, "loss": 0.2612, "step": 15250 }, { "epoch": 1.0543380573798824, "grad_norm": 0.6400758695404414, "learning_rate": 4.803575072541821e-06, "loss": 0.2999, "step": 15251 }, { "epoch": 1.0544071897684064, "grad_norm": 1.0381893731585703, "learning_rate": 4.8030157466604246e-06, "loss": 0.3053, "step": 15252 }, { "epoch": 1.0544763221569304, "grad_norm": 0.6555080007657232, "learning_rate": 4.802456423247871e-06, "loss": 0.262, "step": 15253 }, { "epoch": 1.0545454545454545, "grad_norm": 0.6565518611426586, "learning_rate": 4.801897102311176e-06, "loss": 0.2744, "step": 15254 }, { "epoch": 1.0546145869339785, "grad_norm": 0.7274221214817229, "learning_rate": 4.801337783857345e-06, "loss": 0.3847, "step": 15255 }, { "epoch": 1.0546837193225025, "grad_norm": 2.382173190400789, "learning_rate": 4.8007784678933925e-06, "loss": 0.3113, "step": 15256 }, { "epoch": 1.0547528517110267, "grad_norm": 0.5598888329185627, "learning_rate": 4.8002191544263236e-06, "loss": 0.1878, "step": 15257 }, { "epoch": 1.0548219840995507, "grad_norm": 0.6196674828784386, "learning_rate": 4.799659843463154e-06, "loss": 0.255, "step": 15258 }, { "epoch": 1.0548911164880748, "grad_norm": 0.6969184397739927, "learning_rate": 4.79910053501089e-06, "loss": 0.3388, "step": 15259 }, { "epoch": 1.0549602488765988, "grad_norm": 3.114859527035632, "learning_rate": 4.7985412290765415e-06, "loss": 0.2648, "step": 15260 }, { "epoch": 1.0550293812651228, "grad_norm": 0.7346768584389857, "learning_rate": 4.797981925667119e-06, "loss": 0.243, "step": 15261 }, { "epoch": 1.0550985136536468, "grad_norm": 0.7253536890663111, "learning_rate": 4.7974226247896345e-06, "loss": 0.3582, "step": 15262 }, { "epoch": 1.0551676460421708, "grad_norm": 0.742982533311201, "learning_rate": 4.796863326451093e-06, "loss": 0.3417, "step": 15263 }, { "epoch": 1.0552367784306949, "grad_norm": 0.7420060739045459, "learning_rate": 4.796304030658509e-06, "loss": 0.2478, "step": 15264 }, { "epoch": 1.0553059108192189, "grad_norm": 1.1330329587589736, "learning_rate": 4.79574473741889e-06, "loss": 0.2392, "step": 15265 }, { "epoch": 1.055375043207743, "grad_norm": 0.6230374589924333, "learning_rate": 4.795185446739248e-06, "loss": 0.3404, "step": 15266 }, { "epoch": 1.055444175596267, "grad_norm": 0.6374086953512612, "learning_rate": 4.7946261586265876e-06, "loss": 0.2791, "step": 15267 }, { "epoch": 1.055513307984791, "grad_norm": 0.6501704917974194, "learning_rate": 4.7940668730879245e-06, "loss": 0.2624, "step": 15268 }, { "epoch": 1.055582440373315, "grad_norm": 0.6695465033080642, "learning_rate": 4.793507590130263e-06, "loss": 0.3327, "step": 15269 }, { "epoch": 1.055651572761839, "grad_norm": 0.7160021743363426, "learning_rate": 4.792948309760619e-06, "loss": 0.3543, "step": 15270 }, { "epoch": 1.055720705150363, "grad_norm": 0.7191676343293736, "learning_rate": 4.792389031985995e-06, "loss": 0.2498, "step": 15271 }, { "epoch": 1.055789837538887, "grad_norm": 0.6813992514901253, "learning_rate": 4.791829756813406e-06, "loss": 0.2847, "step": 15272 }, { "epoch": 1.055858969927411, "grad_norm": 0.6861844921391791, "learning_rate": 4.791270484249858e-06, "loss": 0.3532, "step": 15273 }, { "epoch": 1.055928102315935, "grad_norm": 0.6558407655562628, "learning_rate": 4.7907112143023645e-06, "loss": 0.2586, "step": 15274 }, { "epoch": 1.055997234704459, "grad_norm": 0.8572406444810703, "learning_rate": 4.790151946977929e-06, "loss": 0.1991, "step": 15275 }, { "epoch": 1.056066367092983, "grad_norm": 0.6721113351825387, "learning_rate": 4.789592682283567e-06, "loss": 0.3198, "step": 15276 }, { "epoch": 1.056135499481507, "grad_norm": 0.7836256010388818, "learning_rate": 4.789033420226285e-06, "loss": 0.3819, "step": 15277 }, { "epoch": 1.056204631870031, "grad_norm": 0.637396181158113, "learning_rate": 4.788474160813092e-06, "loss": 0.2589, "step": 15278 }, { "epoch": 1.0562737642585551, "grad_norm": 0.6963489038955157, "learning_rate": 4.787914904050997e-06, "loss": 0.2874, "step": 15279 }, { "epoch": 1.0563428966470791, "grad_norm": 0.6625779523732119, "learning_rate": 4.787355649947012e-06, "loss": 0.3341, "step": 15280 }, { "epoch": 1.0564120290356032, "grad_norm": 0.8025447094605834, "learning_rate": 4.786796398508143e-06, "loss": 0.3321, "step": 15281 }, { "epoch": 1.0564811614241272, "grad_norm": 0.6778408180948333, "learning_rate": 4.786237149741402e-06, "loss": 0.235, "step": 15282 }, { "epoch": 1.0565502938126512, "grad_norm": 0.6630545000107991, "learning_rate": 4.785677903653795e-06, "loss": 0.3271, "step": 15283 }, { "epoch": 1.0566194262011752, "grad_norm": 0.7173224496692996, "learning_rate": 4.785118660252335e-06, "loss": 0.3064, "step": 15284 }, { "epoch": 1.0566885585896992, "grad_norm": 0.6228295063320589, "learning_rate": 4.784559419544027e-06, "loss": 0.2615, "step": 15285 }, { "epoch": 1.0567576909782233, "grad_norm": 0.652651182621187, "learning_rate": 4.784000181535885e-06, "loss": 0.2522, "step": 15286 }, { "epoch": 1.0568268233667473, "grad_norm": 0.6289142482681375, "learning_rate": 4.783440946234913e-06, "loss": 0.315, "step": 15287 }, { "epoch": 1.0568959557552713, "grad_norm": 0.7578935893604021, "learning_rate": 4.782881713648124e-06, "loss": 0.3995, "step": 15288 }, { "epoch": 1.0569650881437953, "grad_norm": 0.6565468912113664, "learning_rate": 4.782322483782522e-06, "loss": 0.2468, "step": 15289 }, { "epoch": 1.0570342205323193, "grad_norm": 0.7188750449227571, "learning_rate": 4.7817632566451225e-06, "loss": 0.2679, "step": 15290 }, { "epoch": 1.0571033529208433, "grad_norm": 0.6242532212290228, "learning_rate": 4.781204032242929e-06, "loss": 0.3377, "step": 15291 }, { "epoch": 1.0571724853093674, "grad_norm": 0.6559889342113614, "learning_rate": 4.780644810582955e-06, "loss": 0.2816, "step": 15292 }, { "epoch": 1.0572416176978914, "grad_norm": 0.6398200421835585, "learning_rate": 4.780085591672203e-06, "loss": 0.2547, "step": 15293 }, { "epoch": 1.0573107500864154, "grad_norm": 0.6932767974105605, "learning_rate": 4.77952637551769e-06, "loss": 0.3092, "step": 15294 }, { "epoch": 1.0573798824749394, "grad_norm": 0.7334958139021708, "learning_rate": 4.7789671621264175e-06, "loss": 0.2896, "step": 15295 }, { "epoch": 1.0574490148634634, "grad_norm": 0.7626052771115841, "learning_rate": 4.778407951505398e-06, "loss": 0.261, "step": 15296 }, { "epoch": 1.0575181472519875, "grad_norm": 0.6731918600912672, "learning_rate": 4.777848743661639e-06, "loss": 0.2895, "step": 15297 }, { "epoch": 1.0575872796405115, "grad_norm": 0.6505210121583724, "learning_rate": 4.7772895386021514e-06, "loss": 0.3553, "step": 15298 }, { "epoch": 1.0576564120290357, "grad_norm": 0.6796094671321493, "learning_rate": 4.776730336333941e-06, "loss": 0.2722, "step": 15299 }, { "epoch": 1.0577255444175597, "grad_norm": 0.7069724533264989, "learning_rate": 4.776171136864016e-06, "loss": 0.2694, "step": 15300 }, { "epoch": 1.0577946768060837, "grad_norm": 0.713100132700057, "learning_rate": 4.775611940199389e-06, "loss": 0.3117, "step": 15301 }, { "epoch": 1.0578638091946078, "grad_norm": 0.6754542641405945, "learning_rate": 4.775052746347064e-06, "loss": 0.4181, "step": 15302 }, { "epoch": 1.0579329415831318, "grad_norm": 0.6294941062896728, "learning_rate": 4.774493555314051e-06, "loss": 0.2828, "step": 15303 }, { "epoch": 1.0580020739716558, "grad_norm": 0.7248251115194512, "learning_rate": 4.773934367107359e-06, "loss": 0.2698, "step": 15304 }, { "epoch": 1.0580712063601798, "grad_norm": 0.6574629058287688, "learning_rate": 4.7733751817339985e-06, "loss": 0.3829, "step": 15305 }, { "epoch": 1.0581403387487038, "grad_norm": 0.6403933002302828, "learning_rate": 4.772815999200973e-06, "loss": 0.2606, "step": 15306 }, { "epoch": 1.0582094711372279, "grad_norm": 1.2168111670828274, "learning_rate": 4.772256819515294e-06, "loss": 0.238, "step": 15307 }, { "epoch": 1.0582786035257519, "grad_norm": 0.651606025897152, "learning_rate": 4.77169764268397e-06, "loss": 0.2739, "step": 15308 }, { "epoch": 1.0583477359142759, "grad_norm": 0.6544480663097103, "learning_rate": 4.771138468714009e-06, "loss": 0.3464, "step": 15309 }, { "epoch": 1.0584168683028, "grad_norm": 0.6205052597578065, "learning_rate": 4.770579297612417e-06, "loss": 0.2454, "step": 15310 }, { "epoch": 1.058486000691324, "grad_norm": 0.6427822655032756, "learning_rate": 4.770020129386207e-06, "loss": 0.2168, "step": 15311 }, { "epoch": 1.058555133079848, "grad_norm": 0.6534401646941511, "learning_rate": 4.769460964042381e-06, "loss": 0.3157, "step": 15312 }, { "epoch": 1.058624265468372, "grad_norm": 0.6792990888630785, "learning_rate": 4.768901801587953e-06, "loss": 0.3659, "step": 15313 }, { "epoch": 1.058693397856896, "grad_norm": 0.6074534861737431, "learning_rate": 4.7683426420299266e-06, "loss": 0.2333, "step": 15314 }, { "epoch": 1.05876253024542, "grad_norm": 0.6742013750824682, "learning_rate": 4.767783485375314e-06, "loss": 0.2744, "step": 15315 }, { "epoch": 1.058831662633944, "grad_norm": 0.6455448930364142, "learning_rate": 4.767224331631119e-06, "loss": 0.3529, "step": 15316 }, { "epoch": 1.058900795022468, "grad_norm": 0.7527802973844534, "learning_rate": 4.766665180804353e-06, "loss": 0.244, "step": 15317 }, { "epoch": 1.058969927410992, "grad_norm": 0.6599757577242507, "learning_rate": 4.7661060329020205e-06, "loss": 0.2025, "step": 15318 }, { "epoch": 1.059039059799516, "grad_norm": 0.6684118312233845, "learning_rate": 4.765546887931135e-06, "loss": 0.3674, "step": 15319 }, { "epoch": 1.05910819218804, "grad_norm": 0.6995034617017134, "learning_rate": 4.764987745898698e-06, "loss": 0.3537, "step": 15320 }, { "epoch": 1.059177324576564, "grad_norm": 0.5947970988276027, "learning_rate": 4.7644286068117204e-06, "loss": 0.2435, "step": 15321 }, { "epoch": 1.0592464569650881, "grad_norm": 0.6436914416588901, "learning_rate": 4.7638694706772094e-06, "loss": 0.3032, "step": 15322 }, { "epoch": 1.0593155893536121, "grad_norm": 0.6848700493571398, "learning_rate": 4.763310337502176e-06, "loss": 0.3527, "step": 15323 }, { "epoch": 1.0593847217421362, "grad_norm": 0.66568640874725, "learning_rate": 4.762751207293622e-06, "loss": 0.3105, "step": 15324 }, { "epoch": 1.0594538541306602, "grad_norm": 0.7925698311578763, "learning_rate": 4.7621920800585604e-06, "loss": 0.2593, "step": 15325 }, { "epoch": 1.0595229865191842, "grad_norm": 0.8310599023665421, "learning_rate": 4.761632955803995e-06, "loss": 0.2996, "step": 15326 }, { "epoch": 1.0595921189077082, "grad_norm": 0.7097198339012684, "learning_rate": 4.761073834536937e-06, "loss": 0.3692, "step": 15327 }, { "epoch": 1.0596612512962322, "grad_norm": 0.646746385439333, "learning_rate": 4.7605147162643895e-06, "loss": 0.2981, "step": 15328 }, { "epoch": 1.0597303836847562, "grad_norm": 0.5928162438418879, "learning_rate": 4.759955600993365e-06, "loss": 0.2946, "step": 15329 }, { "epoch": 1.0597995160732803, "grad_norm": 0.6932005425819334, "learning_rate": 4.759396488730867e-06, "loss": 0.4133, "step": 15330 }, { "epoch": 1.0598686484618043, "grad_norm": 0.9974480356659459, "learning_rate": 4.758837379483906e-06, "loss": 0.2839, "step": 15331 }, { "epoch": 1.0599377808503283, "grad_norm": 0.6551453599259168, "learning_rate": 4.758278273259487e-06, "loss": 0.2385, "step": 15332 }, { "epoch": 1.0600069132388523, "grad_norm": 0.705387693942178, "learning_rate": 4.75771917006462e-06, "loss": 0.3111, "step": 15333 }, { "epoch": 1.0600760456273763, "grad_norm": 0.7419196514978977, "learning_rate": 4.757160069906309e-06, "loss": 0.3571, "step": 15334 }, { "epoch": 1.0601451780159004, "grad_norm": 0.6122511863808205, "learning_rate": 4.756600972791564e-06, "loss": 0.2336, "step": 15335 }, { "epoch": 1.0602143104044244, "grad_norm": 0.6969010105152313, "learning_rate": 4.756041878727391e-06, "loss": 0.289, "step": 15336 }, { "epoch": 1.0602834427929484, "grad_norm": 0.6893299055142965, "learning_rate": 4.7554827877208e-06, "loss": 0.3717, "step": 15337 }, { "epoch": 1.0603525751814726, "grad_norm": 0.6949834614471679, "learning_rate": 4.754923699778793e-06, "loss": 0.2579, "step": 15338 }, { "epoch": 1.0604217075699967, "grad_norm": 0.5838838109449143, "learning_rate": 4.754364614908381e-06, "loss": 0.2775, "step": 15339 }, { "epoch": 1.0604908399585207, "grad_norm": 0.6396593651917497, "learning_rate": 4.753805533116569e-06, "loss": 0.2668, "step": 15340 }, { "epoch": 1.0605599723470447, "grad_norm": 0.6834962854751537, "learning_rate": 4.753246454410369e-06, "loss": 0.3556, "step": 15341 }, { "epoch": 1.0606291047355687, "grad_norm": 0.643836793341215, "learning_rate": 4.75268737879678e-06, "loss": 0.2787, "step": 15342 }, { "epoch": 1.0606982371240927, "grad_norm": 0.7904133269751863, "learning_rate": 4.7521283062828164e-06, "loss": 0.2745, "step": 15343 }, { "epoch": 1.0607673695126167, "grad_norm": 0.661488729063257, "learning_rate": 4.75156923687548e-06, "loss": 0.3098, "step": 15344 }, { "epoch": 1.0608365019011408, "grad_norm": 0.6876477853486673, "learning_rate": 4.751010170581782e-06, "loss": 0.329, "step": 15345 }, { "epoch": 1.0609056342896648, "grad_norm": 0.6670694038400417, "learning_rate": 4.750451107408724e-06, "loss": 0.2701, "step": 15346 }, { "epoch": 1.0609747666781888, "grad_norm": 0.7081404828060547, "learning_rate": 4.74989204736332e-06, "loss": 0.2864, "step": 15347 }, { "epoch": 1.0610438990667128, "grad_norm": 0.6947183272609191, "learning_rate": 4.74933299045257e-06, "loss": 0.3375, "step": 15348 }, { "epoch": 1.0611130314552368, "grad_norm": 0.6190360945512527, "learning_rate": 4.748773936683485e-06, "loss": 0.2491, "step": 15349 }, { "epoch": 1.0611821638437609, "grad_norm": 0.600586888086986, "learning_rate": 4.748214886063069e-06, "loss": 0.207, "step": 15350 }, { "epoch": 1.0612512962322849, "grad_norm": 0.7777676382283276, "learning_rate": 4.747655838598332e-06, "loss": 0.3576, "step": 15351 }, { "epoch": 1.0613204286208089, "grad_norm": 0.645312891737086, "learning_rate": 4.747096794296277e-06, "loss": 0.3054, "step": 15352 }, { "epoch": 1.061389561009333, "grad_norm": 0.7916313042572317, "learning_rate": 4.746537753163913e-06, "loss": 0.2797, "step": 15353 }, { "epoch": 1.061458693397857, "grad_norm": 0.8723102710473788, "learning_rate": 4.7459787152082455e-06, "loss": 0.2747, "step": 15354 }, { "epoch": 1.061527825786381, "grad_norm": 0.7082448701300824, "learning_rate": 4.7454196804362835e-06, "loss": 0.3952, "step": 15355 }, { "epoch": 1.061596958174905, "grad_norm": 0.6990016256052906, "learning_rate": 4.7448606488550294e-06, "loss": 0.2658, "step": 15356 }, { "epoch": 1.061666090563429, "grad_norm": 0.6691329782808608, "learning_rate": 4.744301620471492e-06, "loss": 0.2392, "step": 15357 }, { "epoch": 1.061735222951953, "grad_norm": 0.6673184524049433, "learning_rate": 4.7437425952926774e-06, "loss": 0.2744, "step": 15358 }, { "epoch": 1.061804355340477, "grad_norm": 0.7709268669641901, "learning_rate": 4.743183573325594e-06, "loss": 0.3888, "step": 15359 }, { "epoch": 1.061873487729001, "grad_norm": 0.6918281118857773, "learning_rate": 4.742624554577244e-06, "loss": 0.329, "step": 15360 }, { "epoch": 1.061942620117525, "grad_norm": 0.6336825543527737, "learning_rate": 4.7420655390546365e-06, "loss": 0.2451, "step": 15361 }, { "epoch": 1.062011752506049, "grad_norm": 0.7642234580092842, "learning_rate": 4.741506526764776e-06, "loss": 0.3803, "step": 15362 }, { "epoch": 1.062080884894573, "grad_norm": 0.9922392504443912, "learning_rate": 4.740947517714673e-06, "loss": 0.3121, "step": 15363 }, { "epoch": 1.062150017283097, "grad_norm": 0.6080906228505873, "learning_rate": 4.740388511911327e-06, "loss": 0.2342, "step": 15364 }, { "epoch": 1.0622191496716211, "grad_norm": 0.6906322299240375, "learning_rate": 4.73982950936175e-06, "loss": 0.2919, "step": 15365 }, { "epoch": 1.0622882820601451, "grad_norm": 0.7290012783824088, "learning_rate": 4.739270510072946e-06, "loss": 0.3401, "step": 15366 }, { "epoch": 1.0623574144486692, "grad_norm": 0.6370555699349366, "learning_rate": 4.738711514051919e-06, "loss": 0.2795, "step": 15367 }, { "epoch": 1.0624265468371932, "grad_norm": 0.6273630135962225, "learning_rate": 4.73815252130568e-06, "loss": 0.2678, "step": 15368 }, { "epoch": 1.0624956792257172, "grad_norm": 0.7443489528509605, "learning_rate": 4.737593531841229e-06, "loss": 0.3055, "step": 15369 }, { "epoch": 1.0625648116142412, "grad_norm": 0.7656039235500436, "learning_rate": 4.737034545665576e-06, "loss": 0.3314, "step": 15370 }, { "epoch": 1.0626339440027652, "grad_norm": 0.6638516788449788, "learning_rate": 4.736475562785724e-06, "loss": 0.252, "step": 15371 }, { "epoch": 1.0627030763912892, "grad_norm": 0.7521770281247732, "learning_rate": 4.735916583208683e-06, "loss": 0.2858, "step": 15372 }, { "epoch": 1.0627722087798133, "grad_norm": 0.6844160264221576, "learning_rate": 4.7353576069414544e-06, "loss": 0.3831, "step": 15373 }, { "epoch": 1.0628413411683373, "grad_norm": 0.6415622328315838, "learning_rate": 4.734798633991047e-06, "loss": 0.3072, "step": 15374 }, { "epoch": 1.0629104735568613, "grad_norm": 0.6169564066067375, "learning_rate": 4.734239664364464e-06, "loss": 0.2558, "step": 15375 }, { "epoch": 1.0629796059453853, "grad_norm": 0.6982079755879674, "learning_rate": 4.733680698068715e-06, "loss": 0.2973, "step": 15376 }, { "epoch": 1.0630487383339093, "grad_norm": 0.8172027769225303, "learning_rate": 4.7331217351108015e-06, "loss": 0.3875, "step": 15377 }, { "epoch": 1.0631178707224334, "grad_norm": 0.6180793550505377, "learning_rate": 4.732562775497731e-06, "loss": 0.2383, "step": 15378 }, { "epoch": 1.0631870031109574, "grad_norm": 0.8518104234674218, "learning_rate": 4.732003819236508e-06, "loss": 0.2696, "step": 15379 }, { "epoch": 1.0632561354994814, "grad_norm": 0.6815703090011539, "learning_rate": 4.731444866334142e-06, "loss": 0.3172, "step": 15380 }, { "epoch": 1.0633252678880056, "grad_norm": 0.670805234294675, "learning_rate": 4.7308859167976315e-06, "loss": 0.2934, "step": 15381 }, { "epoch": 1.0633944002765296, "grad_norm": 0.6423931457076658, "learning_rate": 4.7303269706339885e-06, "loss": 0.2472, "step": 15382 }, { "epoch": 1.0634635326650537, "grad_norm": 0.8787161360054158, "learning_rate": 4.729768027850214e-06, "loss": 0.2977, "step": 15383 }, { "epoch": 1.0635326650535777, "grad_norm": 0.7291417276659523, "learning_rate": 4.729209088453315e-06, "loss": 0.3644, "step": 15384 }, { "epoch": 1.0636017974421017, "grad_norm": 0.7574776315473822, "learning_rate": 4.728650152450297e-06, "loss": 0.305, "step": 15385 }, { "epoch": 1.0636709298306257, "grad_norm": 0.6452170338569867, "learning_rate": 4.728091219848167e-06, "loss": 0.2639, "step": 15386 }, { "epoch": 1.0637400622191497, "grad_norm": 0.7833962238616343, "learning_rate": 4.7275322906539265e-06, "loss": 0.362, "step": 15387 }, { "epoch": 1.0638091946076738, "grad_norm": 1.030256884367429, "learning_rate": 4.726973364874583e-06, "loss": 0.2998, "step": 15388 }, { "epoch": 1.0638783269961978, "grad_norm": 1.232789175621129, "learning_rate": 4.72641444251714e-06, "loss": 0.2609, "step": 15389 }, { "epoch": 1.0639474593847218, "grad_norm": 0.67828229195356, "learning_rate": 4.725855523588606e-06, "loss": 0.2747, "step": 15390 }, { "epoch": 1.0640165917732458, "grad_norm": 0.60285976649281, "learning_rate": 4.725296608095981e-06, "loss": 0.3345, "step": 15391 }, { "epoch": 1.0640857241617698, "grad_norm": 0.7561552830950174, "learning_rate": 4.724737696046275e-06, "loss": 0.2667, "step": 15392 }, { "epoch": 1.0641548565502938, "grad_norm": 0.653933327132362, "learning_rate": 4.72417878744649e-06, "loss": 0.2379, "step": 15393 }, { "epoch": 1.0642239889388179, "grad_norm": 0.6668081488585204, "learning_rate": 4.723619882303632e-06, "loss": 0.3246, "step": 15394 }, { "epoch": 1.0642931213273419, "grad_norm": 0.6526601834921623, "learning_rate": 4.7230609806247055e-06, "loss": 0.308, "step": 15395 }, { "epoch": 1.064362253715866, "grad_norm": 0.6024325919061266, "learning_rate": 4.7225020824167155e-06, "loss": 0.2036, "step": 15396 }, { "epoch": 1.06443138610439, "grad_norm": 0.6380781699650279, "learning_rate": 4.721943187686666e-06, "loss": 0.2871, "step": 15397 }, { "epoch": 1.064500518492914, "grad_norm": 0.6957709518012368, "learning_rate": 4.721384296441563e-06, "loss": 0.3999, "step": 15398 }, { "epoch": 1.064569650881438, "grad_norm": 0.7561215611565318, "learning_rate": 4.7208254086884095e-06, "loss": 0.2976, "step": 15399 }, { "epoch": 1.064638783269962, "grad_norm": 0.9067924803740623, "learning_rate": 4.720266524434213e-06, "loss": 0.2437, "step": 15400 }, { "epoch": 1.064707915658486, "grad_norm": 0.6968564673377964, "learning_rate": 4.719707643685975e-06, "loss": 0.3281, "step": 15401 }, { "epoch": 1.06477704804701, "grad_norm": 0.6806318390025266, "learning_rate": 4.719148766450702e-06, "loss": 0.3461, "step": 15402 }, { "epoch": 1.064846180435534, "grad_norm": 0.9101671862103561, "learning_rate": 4.718589892735398e-06, "loss": 0.2289, "step": 15403 }, { "epoch": 1.064915312824058, "grad_norm": 0.7011891877299227, "learning_rate": 4.7180310225470686e-06, "loss": 0.3424, "step": 15404 }, { "epoch": 1.064984445212582, "grad_norm": 0.665253997833009, "learning_rate": 4.717472155892715e-06, "loss": 0.346, "step": 15405 }, { "epoch": 1.065053577601106, "grad_norm": 0.6128671771886082, "learning_rate": 4.716913292779346e-06, "loss": 0.2817, "step": 15406 }, { "epoch": 1.06512270998963, "grad_norm": 0.7748507749740735, "learning_rate": 4.716354433213962e-06, "loss": 0.2414, "step": 15407 }, { "epoch": 1.0651918423781541, "grad_norm": 0.6757977898124927, "learning_rate": 4.71579557720357e-06, "loss": 0.2698, "step": 15408 }, { "epoch": 1.0652609747666781, "grad_norm": 0.7119244446166441, "learning_rate": 4.715236724755172e-06, "loss": 0.3833, "step": 15409 }, { "epoch": 1.0653301071552022, "grad_norm": 0.6690699836825793, "learning_rate": 4.714677875875775e-06, "loss": 0.2402, "step": 15410 }, { "epoch": 1.0653992395437262, "grad_norm": 0.6356374526544943, "learning_rate": 4.71411903057238e-06, "loss": 0.2363, "step": 15411 }, { "epoch": 1.0654683719322502, "grad_norm": 0.7757394005651785, "learning_rate": 4.713560188851995e-06, "loss": 0.3406, "step": 15412 }, { "epoch": 1.0655375043207742, "grad_norm": 0.9480152954411962, "learning_rate": 4.71300135072162e-06, "loss": 0.296, "step": 15413 }, { "epoch": 1.0656066367092982, "grad_norm": 0.6690823905183871, "learning_rate": 4.712442516188261e-06, "loss": 0.3042, "step": 15414 }, { "epoch": 1.0656757690978222, "grad_norm": 0.7436190170905271, "learning_rate": 4.7118836852589225e-06, "loss": 0.2735, "step": 15415 }, { "epoch": 1.0657449014863463, "grad_norm": 0.807220892015817, "learning_rate": 4.711324857940609e-06, "loss": 0.3411, "step": 15416 }, { "epoch": 1.0658140338748703, "grad_norm": 0.6300336313441204, "learning_rate": 4.71076603424032e-06, "loss": 0.2419, "step": 15417 }, { "epoch": 1.0658831662633943, "grad_norm": 0.761336313182257, "learning_rate": 4.710207214165066e-06, "loss": 0.2323, "step": 15418 }, { "epoch": 1.0659522986519185, "grad_norm": 0.7010107242588876, "learning_rate": 4.709648397721845e-06, "loss": 0.3434, "step": 15419 }, { "epoch": 1.0660214310404426, "grad_norm": 0.6616339946755225, "learning_rate": 4.709089584917664e-06, "loss": 0.323, "step": 15420 }, { "epoch": 1.0660905634289666, "grad_norm": 0.7662174568533242, "learning_rate": 4.708530775759525e-06, "loss": 0.2349, "step": 15421 }, { "epoch": 1.0661596958174906, "grad_norm": 0.6694715731221258, "learning_rate": 4.707971970254435e-06, "loss": 0.3105, "step": 15422 }, { "epoch": 1.0662288282060146, "grad_norm": 0.739230304847578, "learning_rate": 4.707413168409393e-06, "loss": 0.3455, "step": 15423 }, { "epoch": 1.0662979605945386, "grad_norm": 0.6667229113156284, "learning_rate": 4.706854370231406e-06, "loss": 0.2491, "step": 15424 }, { "epoch": 1.0663670929830626, "grad_norm": 0.6232902097213165, "learning_rate": 4.706295575727475e-06, "loss": 0.2642, "step": 15425 }, { "epoch": 1.0664362253715867, "grad_norm": 0.8340372258429063, "learning_rate": 4.705736784904608e-06, "loss": 0.3306, "step": 15426 }, { "epoch": 1.0665053577601107, "grad_norm": 0.7746011076657615, "learning_rate": 4.705177997769802e-06, "loss": 0.2969, "step": 15427 }, { "epoch": 1.0665744901486347, "grad_norm": 0.7518016696172576, "learning_rate": 4.7046192143300644e-06, "loss": 0.2419, "step": 15428 }, { "epoch": 1.0666436225371587, "grad_norm": 0.7296290839435842, "learning_rate": 4.704060434592398e-06, "loss": 0.2438, "step": 15429 }, { "epoch": 1.0667127549256827, "grad_norm": 0.6533502000623024, "learning_rate": 4.703501658563807e-06, "loss": 0.3056, "step": 15430 }, { "epoch": 1.0667818873142068, "grad_norm": 0.6511109110525363, "learning_rate": 4.702942886251291e-06, "loss": 0.2546, "step": 15431 }, { "epoch": 1.0668510197027308, "grad_norm": 0.613068624346517, "learning_rate": 4.702384117661859e-06, "loss": 0.1861, "step": 15432 }, { "epoch": 1.0669201520912548, "grad_norm": 0.697081015220183, "learning_rate": 4.7018253528025104e-06, "loss": 0.323, "step": 15433 }, { "epoch": 1.0669892844797788, "grad_norm": 0.6503889557273355, "learning_rate": 4.701266591680247e-06, "loss": 0.3632, "step": 15434 }, { "epoch": 1.0670584168683028, "grad_norm": 0.6599807328060955, "learning_rate": 4.700707834302075e-06, "loss": 0.2477, "step": 15435 }, { "epoch": 1.0671275492568268, "grad_norm": 0.6682831263684879, "learning_rate": 4.700149080674995e-06, "loss": 0.2107, "step": 15436 }, { "epoch": 1.0671966816453509, "grad_norm": 0.6365746304141553, "learning_rate": 4.699590330806012e-06, "loss": 0.334, "step": 15437 }, { "epoch": 1.0672658140338749, "grad_norm": 0.6205193214730458, "learning_rate": 4.699031584702127e-06, "loss": 0.32, "step": 15438 }, { "epoch": 1.067334946422399, "grad_norm": 0.6125741363302344, "learning_rate": 4.698472842370347e-06, "loss": 0.2909, "step": 15439 }, { "epoch": 1.067404078810923, "grad_norm": 0.7100401376788956, "learning_rate": 4.697914103817669e-06, "loss": 0.2948, "step": 15440 }, { "epoch": 1.067473211199447, "grad_norm": 0.7086909507765813, "learning_rate": 4.697355369051099e-06, "loss": 0.3682, "step": 15441 }, { "epoch": 1.067542343587971, "grad_norm": 0.7302923837019368, "learning_rate": 4.696796638077639e-06, "loss": 0.3298, "step": 15442 }, { "epoch": 1.067611475976495, "grad_norm": 0.6630839453907721, "learning_rate": 4.696237910904295e-06, "loss": 0.2148, "step": 15443 }, { "epoch": 1.067680608365019, "grad_norm": 0.694789535856682, "learning_rate": 4.695679187538064e-06, "loss": 0.3049, "step": 15444 }, { "epoch": 1.067749740753543, "grad_norm": 0.6651136163531786, "learning_rate": 4.695120467985952e-06, "loss": 0.305, "step": 15445 }, { "epoch": 1.067818873142067, "grad_norm": 0.6657012253531843, "learning_rate": 4.6945617522549595e-06, "loss": 0.2818, "step": 15446 }, { "epoch": 1.067888005530591, "grad_norm": 0.6814179599454847, "learning_rate": 4.694003040352093e-06, "loss": 0.2432, "step": 15447 }, { "epoch": 1.067957137919115, "grad_norm": 0.723501795170021, "learning_rate": 4.693444332284351e-06, "loss": 0.3544, "step": 15448 }, { "epoch": 1.068026270307639, "grad_norm": 0.6353472747815355, "learning_rate": 4.692885628058736e-06, "loss": 0.2584, "step": 15449 }, { "epoch": 1.068095402696163, "grad_norm": 0.6352206199298949, "learning_rate": 4.692326927682253e-06, "loss": 0.2618, "step": 15450 }, { "epoch": 1.0681645350846871, "grad_norm": 0.6611256398188532, "learning_rate": 4.691768231161904e-06, "loss": 0.2954, "step": 15451 }, { "epoch": 1.0682336674732111, "grad_norm": 0.6927876111888646, "learning_rate": 4.691209538504687e-06, "loss": 0.3238, "step": 15452 }, { "epoch": 1.0683027998617352, "grad_norm": 0.661266791448418, "learning_rate": 4.6906508497176094e-06, "loss": 0.2767, "step": 15453 }, { "epoch": 1.0683719322502592, "grad_norm": 0.6964622153590057, "learning_rate": 4.690092164807671e-06, "loss": 0.2829, "step": 15454 }, { "epoch": 1.0684410646387832, "grad_norm": 0.8240982113311229, "learning_rate": 4.689533483781874e-06, "loss": 0.3352, "step": 15455 }, { "epoch": 1.0685101970273072, "grad_norm": 0.6465162022085329, "learning_rate": 4.688974806647221e-06, "loss": 0.2673, "step": 15456 }, { "epoch": 1.0685793294158312, "grad_norm": 0.64849771996342, "learning_rate": 4.688416133410715e-06, "loss": 0.2218, "step": 15457 }, { "epoch": 1.0686484618043552, "grad_norm": 0.659676825366472, "learning_rate": 4.687857464079355e-06, "loss": 0.2955, "step": 15458 }, { "epoch": 1.0687175941928793, "grad_norm": 0.6800619334266206, "learning_rate": 4.6872987986601445e-06, "loss": 0.3105, "step": 15459 }, { "epoch": 1.0687867265814033, "grad_norm": 0.6914761546338309, "learning_rate": 4.686740137160086e-06, "loss": 0.3164, "step": 15460 }, { "epoch": 1.0688558589699273, "grad_norm": 0.6353395208495904, "learning_rate": 4.686181479586182e-06, "loss": 0.2376, "step": 15461 }, { "epoch": 1.0689249913584515, "grad_norm": 0.8638630491465351, "learning_rate": 4.685622825945432e-06, "loss": 0.3363, "step": 15462 }, { "epoch": 1.0689941237469756, "grad_norm": 1.757450911422929, "learning_rate": 4.685064176244839e-06, "loss": 0.304, "step": 15463 }, { "epoch": 1.0690632561354996, "grad_norm": 0.6185916671174705, "learning_rate": 4.684505530491403e-06, "loss": 0.2577, "step": 15464 }, { "epoch": 1.0691323885240236, "grad_norm": 0.7004127094626685, "learning_rate": 4.683946888692131e-06, "loss": 0.2353, "step": 15465 }, { "epoch": 1.0692015209125476, "grad_norm": 0.6541569218917618, "learning_rate": 4.683388250854017e-06, "loss": 0.3411, "step": 15466 }, { "epoch": 1.0692706533010716, "grad_norm": 0.7643916686952915, "learning_rate": 4.682829616984068e-06, "loss": 0.2763, "step": 15467 }, { "epoch": 1.0693397856895956, "grad_norm": 0.9020362229965257, "learning_rate": 4.682270987089282e-06, "loss": 0.1922, "step": 15468 }, { "epoch": 1.0694089180781197, "grad_norm": 0.7021908757067299, "learning_rate": 4.6817123611766655e-06, "loss": 0.3797, "step": 15469 }, { "epoch": 1.0694780504666437, "grad_norm": 0.6533221758093406, "learning_rate": 4.681153739253213e-06, "loss": 0.3171, "step": 15470 }, { "epoch": 1.0695471828551677, "grad_norm": 0.6284903121144646, "learning_rate": 4.6805951213259325e-06, "loss": 0.282, "step": 15471 }, { "epoch": 1.0696163152436917, "grad_norm": 0.8483446222399301, "learning_rate": 4.680036507401819e-06, "loss": 0.2429, "step": 15472 }, { "epoch": 1.0696854476322157, "grad_norm": 0.664508220782894, "learning_rate": 4.679477897487879e-06, "loss": 0.3311, "step": 15473 }, { "epoch": 1.0697545800207398, "grad_norm": 0.7043545865949827, "learning_rate": 4.67891929159111e-06, "loss": 0.276, "step": 15474 }, { "epoch": 1.0698237124092638, "grad_norm": 0.7845861734567583, "learning_rate": 4.678360689718518e-06, "loss": 0.2663, "step": 15475 }, { "epoch": 1.0698928447977878, "grad_norm": 0.643393232739875, "learning_rate": 4.677802091877097e-06, "loss": 0.3096, "step": 15476 }, { "epoch": 1.0699619771863118, "grad_norm": 0.6578984120528389, "learning_rate": 4.677243498073854e-06, "loss": 0.3331, "step": 15477 }, { "epoch": 1.0700311095748358, "grad_norm": 0.722404024031678, "learning_rate": 4.676684908315786e-06, "loss": 0.2671, "step": 15478 }, { "epoch": 1.0701002419633598, "grad_norm": 0.652681748821725, "learning_rate": 4.676126322609899e-06, "loss": 0.2935, "step": 15479 }, { "epoch": 1.0701693743518839, "grad_norm": 0.6612897901791563, "learning_rate": 4.675567740963187e-06, "loss": 0.3556, "step": 15480 }, { "epoch": 1.0702385067404079, "grad_norm": 0.863923677109557, "learning_rate": 4.675009163382655e-06, "loss": 0.2871, "step": 15481 }, { "epoch": 1.070307639128932, "grad_norm": 0.6950632853878645, "learning_rate": 4.674450589875304e-06, "loss": 0.2423, "step": 15482 }, { "epoch": 1.070376771517456, "grad_norm": 0.6441308130007359, "learning_rate": 4.673892020448135e-06, "loss": 0.3245, "step": 15483 }, { "epoch": 1.07044590390598, "grad_norm": 0.6801956925868915, "learning_rate": 4.6733334551081455e-06, "loss": 0.3945, "step": 15484 }, { "epoch": 1.070515036294504, "grad_norm": 0.7237082556734034, "learning_rate": 4.6727748938623395e-06, "loss": 0.3111, "step": 15485 }, { "epoch": 1.070584168683028, "grad_norm": 0.6783671105947315, "learning_rate": 4.672216336717715e-06, "loss": 0.2379, "step": 15486 }, { "epoch": 1.070653301071552, "grad_norm": 0.6328619708017005, "learning_rate": 4.671657783681276e-06, "loss": 0.3618, "step": 15487 }, { "epoch": 1.070722433460076, "grad_norm": 0.6791048070000285, "learning_rate": 4.671099234760017e-06, "loss": 0.3522, "step": 15488 }, { "epoch": 1.0707915658486, "grad_norm": 0.8245076710022325, "learning_rate": 4.6705406899609454e-06, "loss": 0.2837, "step": 15489 }, { "epoch": 1.070860698237124, "grad_norm": 0.6518356258740066, "learning_rate": 4.669982149291057e-06, "loss": 0.3185, "step": 15490 }, { "epoch": 1.070929830625648, "grad_norm": 0.6669629112907132, "learning_rate": 4.6694236127573536e-06, "loss": 0.3106, "step": 15491 }, { "epoch": 1.070998963014172, "grad_norm": 0.6825418333013548, "learning_rate": 4.6688650803668346e-06, "loss": 0.3104, "step": 15492 }, { "epoch": 1.071068095402696, "grad_norm": 0.8956673348077048, "learning_rate": 4.668306552126503e-06, "loss": 0.2327, "step": 15493 }, { "epoch": 1.0711372277912201, "grad_norm": 0.6501377193621857, "learning_rate": 4.667748028043355e-06, "loss": 0.3067, "step": 15494 }, { "epoch": 1.0712063601797441, "grad_norm": 0.6484686803798078, "learning_rate": 4.667189508124393e-06, "loss": 0.338, "step": 15495 }, { "epoch": 1.0712754925682682, "grad_norm": 0.6124396436625287, "learning_rate": 4.666630992376616e-06, "loss": 0.267, "step": 15496 }, { "epoch": 1.0713446249567922, "grad_norm": 0.6428950362204255, "learning_rate": 4.666072480807027e-06, "loss": 0.2727, "step": 15497 }, { "epoch": 1.0714137573453162, "grad_norm": 0.680339128801648, "learning_rate": 4.6655139734226205e-06, "loss": 0.3824, "step": 15498 }, { "epoch": 1.0714828897338402, "grad_norm": 0.6715886585685933, "learning_rate": 4.664955470230401e-06, "loss": 0.2367, "step": 15499 }, { "epoch": 1.0715520221223644, "grad_norm": 0.5971061278247046, "learning_rate": 4.664396971237367e-06, "loss": 0.2424, "step": 15500 }, { "epoch": 1.0716211545108885, "grad_norm": 0.683090287363435, "learning_rate": 4.663838476450518e-06, "loss": 0.2801, "step": 15501 }, { "epoch": 1.0716902868994125, "grad_norm": 0.654804795472504, "learning_rate": 4.663279985876853e-06, "loss": 0.3267, "step": 15502 }, { "epoch": 1.0717594192879365, "grad_norm": 0.6157214574172796, "learning_rate": 4.662721499523372e-06, "loss": 0.252, "step": 15503 }, { "epoch": 1.0718285516764605, "grad_norm": 0.8424425178113812, "learning_rate": 4.662163017397077e-06, "loss": 0.2934, "step": 15504 }, { "epoch": 1.0718976840649845, "grad_norm": 0.6965397016316193, "learning_rate": 4.661604539504964e-06, "loss": 0.3743, "step": 15505 }, { "epoch": 1.0719668164535086, "grad_norm": 0.7674498048493861, "learning_rate": 4.661046065854035e-06, "loss": 0.2752, "step": 15506 }, { "epoch": 1.0720359488420326, "grad_norm": 0.597486135716958, "learning_rate": 4.660487596451287e-06, "loss": 0.2468, "step": 15507 }, { "epoch": 1.0721050812305566, "grad_norm": 0.6514763991485939, "learning_rate": 4.659929131303723e-06, "loss": 0.2789, "step": 15508 }, { "epoch": 1.0721742136190806, "grad_norm": 0.6264656376370324, "learning_rate": 4.65937067041834e-06, "loss": 0.3223, "step": 15509 }, { "epoch": 1.0722433460076046, "grad_norm": 0.6810484069002144, "learning_rate": 4.658812213802139e-06, "loss": 0.2597, "step": 15510 }, { "epoch": 1.0723124783961286, "grad_norm": 0.8435467593574776, "learning_rate": 4.658253761462115e-06, "loss": 0.23, "step": 15511 }, { "epoch": 1.0723816107846527, "grad_norm": 0.7289608607427872, "learning_rate": 4.657695313405273e-06, "loss": 0.331, "step": 15512 }, { "epoch": 1.0724507431731767, "grad_norm": 0.6944133858632706, "learning_rate": 4.6571368696386074e-06, "loss": 0.2684, "step": 15513 }, { "epoch": 1.0725198755617007, "grad_norm": 0.5975044028187133, "learning_rate": 4.6565784301691215e-06, "loss": 0.2253, "step": 15514 }, { "epoch": 1.0725890079502247, "grad_norm": 0.8257222238225246, "learning_rate": 4.65601999500381e-06, "loss": 0.241, "step": 15515 }, { "epoch": 1.0726581403387487, "grad_norm": 0.6441857767667887, "learning_rate": 4.6554615641496754e-06, "loss": 0.3656, "step": 15516 }, { "epoch": 1.0727272727272728, "grad_norm": 0.6185268540824136, "learning_rate": 4.6549031376137134e-06, "loss": 0.2249, "step": 15517 }, { "epoch": 1.0727964051157968, "grad_norm": 0.5980277705310294, "learning_rate": 4.654344715402928e-06, "loss": 0.2313, "step": 15518 }, { "epoch": 1.0728655375043208, "grad_norm": 0.612880614814955, "learning_rate": 4.653786297524312e-06, "loss": 0.3219, "step": 15519 }, { "epoch": 1.0729346698928448, "grad_norm": 0.6833844254333723, "learning_rate": 4.6532278839848675e-06, "loss": 0.3849, "step": 15520 }, { "epoch": 1.0730038022813688, "grad_norm": 0.5875611477603617, "learning_rate": 4.652669474791592e-06, "loss": 0.2384, "step": 15521 }, { "epoch": 1.0730729346698928, "grad_norm": 0.7332808791492099, "learning_rate": 4.6521110699514875e-06, "loss": 0.3183, "step": 15522 }, { "epoch": 1.0731420670584169, "grad_norm": 0.6594705147346944, "learning_rate": 4.6515526694715464e-06, "loss": 0.3455, "step": 15523 }, { "epoch": 1.0732111994469409, "grad_norm": 0.6476162596991345, "learning_rate": 4.6509942733587726e-06, "loss": 0.3023, "step": 15524 }, { "epoch": 1.073280331835465, "grad_norm": 0.8421442780055808, "learning_rate": 4.6504358816201615e-06, "loss": 0.2366, "step": 15525 }, { "epoch": 1.073349464223989, "grad_norm": 0.751192264497744, "learning_rate": 4.649877494262715e-06, "loss": 0.3274, "step": 15526 }, { "epoch": 1.073418596612513, "grad_norm": 0.5981121453977517, "learning_rate": 4.649319111293427e-06, "loss": 0.3061, "step": 15527 }, { "epoch": 1.073487729001037, "grad_norm": 0.649335286579548, "learning_rate": 4.6487607327193e-06, "loss": 0.2503, "step": 15528 }, { "epoch": 1.073556861389561, "grad_norm": 0.6520628873970492, "learning_rate": 4.648202358547329e-06, "loss": 0.2497, "step": 15529 }, { "epoch": 1.073625993778085, "grad_norm": 0.6740443942535179, "learning_rate": 4.647643988784515e-06, "loss": 0.3725, "step": 15530 }, { "epoch": 1.073695126166609, "grad_norm": 0.6209164688821062, "learning_rate": 4.6470856234378525e-06, "loss": 0.2685, "step": 15531 }, { "epoch": 1.073764258555133, "grad_norm": 0.6624706208500734, "learning_rate": 4.646527262514345e-06, "loss": 0.2728, "step": 15532 }, { "epoch": 1.073833390943657, "grad_norm": 0.6544358350946178, "learning_rate": 4.645968906020985e-06, "loss": 0.2286, "step": 15533 }, { "epoch": 1.073902523332181, "grad_norm": 0.6420384800902573, "learning_rate": 4.645410553964774e-06, "loss": 0.3775, "step": 15534 }, { "epoch": 1.073971655720705, "grad_norm": 0.9319775903022521, "learning_rate": 4.644852206352709e-06, "loss": 0.3095, "step": 15535 }, { "epoch": 1.074040788109229, "grad_norm": 0.6313093009595612, "learning_rate": 4.644293863191789e-06, "loss": 0.231, "step": 15536 }, { "epoch": 1.0741099204977531, "grad_norm": 0.6475803321195936, "learning_rate": 4.643735524489007e-06, "loss": 0.3446, "step": 15537 }, { "epoch": 1.0741790528862771, "grad_norm": 0.7088920215666822, "learning_rate": 4.6431771902513675e-06, "loss": 0.3584, "step": 15538 }, { "epoch": 1.0742481852748011, "grad_norm": 0.626066081768302, "learning_rate": 4.642618860485864e-06, "loss": 0.2872, "step": 15539 }, { "epoch": 1.0743173176633252, "grad_norm": 0.7157152426913843, "learning_rate": 4.642060535199496e-06, "loss": 0.2904, "step": 15540 }, { "epoch": 1.0743864500518492, "grad_norm": 0.6435029025904275, "learning_rate": 4.641502214399259e-06, "loss": 0.348, "step": 15541 }, { "epoch": 1.0744555824403732, "grad_norm": 0.6388816879710162, "learning_rate": 4.640943898092154e-06, "loss": 0.2828, "step": 15542 }, { "epoch": 1.0745247148288972, "grad_norm": 0.8440175732877533, "learning_rate": 4.6403855862851745e-06, "loss": 0.2386, "step": 15543 }, { "epoch": 1.0745938472174215, "grad_norm": 0.7062563731443064, "learning_rate": 4.639827278985321e-06, "loss": 0.341, "step": 15544 }, { "epoch": 1.0746629796059455, "grad_norm": 0.6754653331396954, "learning_rate": 4.639268976199589e-06, "loss": 0.3275, "step": 15545 }, { "epoch": 1.0747321119944695, "grad_norm": 0.6666284412608361, "learning_rate": 4.6387106779349785e-06, "loss": 0.2745, "step": 15546 }, { "epoch": 1.0748012443829935, "grad_norm": 0.630207449060072, "learning_rate": 4.6381523841984835e-06, "loss": 0.2642, "step": 15547 }, { "epoch": 1.0748703767715175, "grad_norm": 0.6981436740259194, "learning_rate": 4.637594094997103e-06, "loss": 0.3427, "step": 15548 }, { "epoch": 1.0749395091600416, "grad_norm": 0.5968637134608413, "learning_rate": 4.637035810337833e-06, "loss": 0.2421, "step": 15549 }, { "epoch": 1.0750086415485656, "grad_norm": 0.6263818217950016, "learning_rate": 4.636477530227672e-06, "loss": 0.2148, "step": 15550 }, { "epoch": 1.0750777739370896, "grad_norm": 0.6600205635800794, "learning_rate": 4.635919254673616e-06, "loss": 0.2668, "step": 15551 }, { "epoch": 1.0751469063256136, "grad_norm": 0.6703261543755836, "learning_rate": 4.635360983682664e-06, "loss": 0.3402, "step": 15552 }, { "epoch": 1.0752160387141376, "grad_norm": 0.6174990833173295, "learning_rate": 4.634802717261809e-06, "loss": 0.2533, "step": 15553 }, { "epoch": 1.0752851711026616, "grad_norm": 0.6932104127900118, "learning_rate": 4.634244455418053e-06, "loss": 0.3004, "step": 15554 }, { "epoch": 1.0753543034911857, "grad_norm": 0.7066511512309971, "learning_rate": 4.633686198158387e-06, "loss": 0.3771, "step": 15555 }, { "epoch": 1.0754234358797097, "grad_norm": 0.697774110703863, "learning_rate": 4.633127945489813e-06, "loss": 0.3294, "step": 15556 }, { "epoch": 1.0754925682682337, "grad_norm": 0.6217770519196534, "learning_rate": 4.632569697419324e-06, "loss": 0.2263, "step": 15557 }, { "epoch": 1.0755617006567577, "grad_norm": 0.6638822528236733, "learning_rate": 4.63201145395392e-06, "loss": 0.2596, "step": 15558 }, { "epoch": 1.0756308330452817, "grad_norm": 0.7581228726277626, "learning_rate": 4.631453215100592e-06, "loss": 0.3772, "step": 15559 }, { "epoch": 1.0756999654338057, "grad_norm": 0.7186741123445378, "learning_rate": 4.6308949808663444e-06, "loss": 0.249, "step": 15560 }, { "epoch": 1.0757690978223298, "grad_norm": 0.6077465054051526, "learning_rate": 4.630336751258167e-06, "loss": 0.2354, "step": 15561 }, { "epoch": 1.0758382302108538, "grad_norm": 0.6959768797347882, "learning_rate": 4.629778526283059e-06, "loss": 0.3634, "step": 15562 }, { "epoch": 1.0759073625993778, "grad_norm": 0.589398226013944, "learning_rate": 4.629220305948017e-06, "loss": 0.2456, "step": 15563 }, { "epoch": 1.0759764949879018, "grad_norm": 0.6529348971988044, "learning_rate": 4.628662090260039e-06, "loss": 0.2526, "step": 15564 }, { "epoch": 1.0760456273764258, "grad_norm": 0.7791739120062083, "learning_rate": 4.628103879226117e-06, "loss": 0.2812, "step": 15565 }, { "epoch": 1.0761147597649499, "grad_norm": 0.7273262585639819, "learning_rate": 4.627545672853248e-06, "loss": 0.3397, "step": 15566 }, { "epoch": 1.0761838921534739, "grad_norm": 0.6160383879756007, "learning_rate": 4.626987471148432e-06, "loss": 0.2726, "step": 15567 }, { "epoch": 1.076253024541998, "grad_norm": 0.7294191260798918, "learning_rate": 4.6264292741186595e-06, "loss": 0.2601, "step": 15568 }, { "epoch": 1.076322156930522, "grad_norm": 0.6708353858850292, "learning_rate": 4.625871081770931e-06, "loss": 0.2905, "step": 15569 }, { "epoch": 1.076391289319046, "grad_norm": 0.8258052453761792, "learning_rate": 4.62531289411224e-06, "loss": 0.3043, "step": 15570 }, { "epoch": 1.07646042170757, "grad_norm": 0.6279999548326353, "learning_rate": 4.624754711149585e-06, "loss": 0.2886, "step": 15571 }, { "epoch": 1.076529554096094, "grad_norm": 0.6447584104603175, "learning_rate": 4.624196532889958e-06, "loss": 0.2182, "step": 15572 }, { "epoch": 1.076598686484618, "grad_norm": 0.6295674364123599, "learning_rate": 4.623638359340358e-06, "loss": 0.3533, "step": 15573 }, { "epoch": 1.076667818873142, "grad_norm": 0.9429811228640039, "learning_rate": 4.623080190507779e-06, "loss": 0.235, "step": 15574 }, { "epoch": 1.076736951261666, "grad_norm": 0.6521746158599039, "learning_rate": 4.622522026399218e-06, "loss": 0.2239, "step": 15575 }, { "epoch": 1.07680608365019, "grad_norm": 0.7717715617304313, "learning_rate": 4.621963867021667e-06, "loss": 0.2528, "step": 15576 }, { "epoch": 1.076875216038714, "grad_norm": 0.6368042704506431, "learning_rate": 4.621405712382128e-06, "loss": 0.3384, "step": 15577 }, { "epoch": 1.076944348427238, "grad_norm": 0.6934503646424156, "learning_rate": 4.62084756248759e-06, "loss": 0.3031, "step": 15578 }, { "epoch": 1.077013480815762, "grad_norm": 0.6252211054082156, "learning_rate": 4.620289417345052e-06, "loss": 0.2642, "step": 15579 }, { "epoch": 1.077082613204286, "grad_norm": 0.6857053931746931, "learning_rate": 4.619731276961508e-06, "loss": 0.3311, "step": 15580 }, { "epoch": 1.0771517455928101, "grad_norm": 0.6515477889171791, "learning_rate": 4.619173141343956e-06, "loss": 0.3103, "step": 15581 }, { "epoch": 1.0772208779813344, "grad_norm": 0.5721737843667037, "learning_rate": 4.618615010499387e-06, "loss": 0.2445, "step": 15582 }, { "epoch": 1.0772900103698584, "grad_norm": 0.6648265164962603, "learning_rate": 4.618056884434799e-06, "loss": 0.2796, "step": 15583 }, { "epoch": 1.0773591427583824, "grad_norm": 0.6296749674477948, "learning_rate": 4.617498763157186e-06, "loss": 0.3656, "step": 15584 }, { "epoch": 1.0774282751469064, "grad_norm": 0.879199751715606, "learning_rate": 4.6169406466735446e-06, "loss": 0.2292, "step": 15585 }, { "epoch": 1.0774974075354304, "grad_norm": 0.6750811507541459, "learning_rate": 4.616382534990867e-06, "loss": 0.2679, "step": 15586 }, { "epoch": 1.0775665399239545, "grad_norm": 0.6085007388611995, "learning_rate": 4.61582442811615e-06, "loss": 0.3256, "step": 15587 }, { "epoch": 1.0776356723124785, "grad_norm": 0.6781833882464108, "learning_rate": 4.6152663260563875e-06, "loss": 0.2966, "step": 15588 }, { "epoch": 1.0777048047010025, "grad_norm": 0.6463574934825795, "learning_rate": 4.6147082288185775e-06, "loss": 0.2728, "step": 15589 }, { "epoch": 1.0777739370895265, "grad_norm": 0.64824188637837, "learning_rate": 4.614150136409711e-06, "loss": 0.3287, "step": 15590 }, { "epoch": 1.0778430694780505, "grad_norm": 0.689157722467964, "learning_rate": 4.613592048836784e-06, "loss": 0.3887, "step": 15591 }, { "epoch": 1.0779122018665745, "grad_norm": 0.6648980905563293, "learning_rate": 4.61303396610679e-06, "loss": 0.2842, "step": 15592 }, { "epoch": 1.0779813342550986, "grad_norm": 0.5768850492130805, "learning_rate": 4.612475888226727e-06, "loss": 0.223, "step": 15593 }, { "epoch": 1.0780504666436226, "grad_norm": 0.664260685234016, "learning_rate": 4.611917815203584e-06, "loss": 0.3078, "step": 15594 }, { "epoch": 1.0781195990321466, "grad_norm": 0.7257344395870143, "learning_rate": 4.611359747044361e-06, "loss": 0.3085, "step": 15595 }, { "epoch": 1.0781887314206706, "grad_norm": 0.7082434696437713, "learning_rate": 4.610801683756049e-06, "loss": 0.2501, "step": 15596 }, { "epoch": 1.0782578638091946, "grad_norm": 0.6330342946264633, "learning_rate": 4.610243625345643e-06, "loss": 0.2372, "step": 15597 }, { "epoch": 1.0783269961977187, "grad_norm": 0.6549797509665587, "learning_rate": 4.609685571820138e-06, "loss": 0.3564, "step": 15598 }, { "epoch": 1.0783961285862427, "grad_norm": 0.7375262786405605, "learning_rate": 4.60912752318653e-06, "loss": 0.3034, "step": 15599 }, { "epoch": 1.0784652609747667, "grad_norm": 0.5914372565650898, "learning_rate": 4.6085694794518075e-06, "loss": 0.2223, "step": 15600 }, { "epoch": 1.0785343933632907, "grad_norm": 0.6836317658393511, "learning_rate": 4.6080114406229705e-06, "loss": 0.3138, "step": 15601 }, { "epoch": 1.0786035257518147, "grad_norm": 0.7007855332959539, "learning_rate": 4.607453406707008e-06, "loss": 0.3746, "step": 15602 }, { "epoch": 1.0786726581403387, "grad_norm": 0.6345424184719418, "learning_rate": 4.60689537771092e-06, "loss": 0.2656, "step": 15603 }, { "epoch": 1.0787417905288628, "grad_norm": 0.6573692569063428, "learning_rate": 4.606337353641694e-06, "loss": 0.2559, "step": 15604 }, { "epoch": 1.0788109229173868, "grad_norm": 1.9119538900259319, "learning_rate": 4.605779334506328e-06, "loss": 0.3403, "step": 15605 }, { "epoch": 1.0788800553059108, "grad_norm": 0.6177382708241577, "learning_rate": 4.605221320311814e-06, "loss": 0.2535, "step": 15606 }, { "epoch": 1.0789491876944348, "grad_norm": 0.601548241904861, "learning_rate": 4.6046633110651476e-06, "loss": 0.2158, "step": 15607 }, { "epoch": 1.0790183200829588, "grad_norm": 0.6774986106887667, "learning_rate": 4.604105306773319e-06, "loss": 0.3019, "step": 15608 }, { "epoch": 1.0790874524714829, "grad_norm": 0.696684771155726, "learning_rate": 4.603547307443326e-06, "loss": 0.3949, "step": 15609 }, { "epoch": 1.0791565848600069, "grad_norm": 0.6231224185010665, "learning_rate": 4.602989313082158e-06, "loss": 0.2588, "step": 15610 }, { "epoch": 1.079225717248531, "grad_norm": 0.6974280357108761, "learning_rate": 4.602431323696813e-06, "loss": 0.3059, "step": 15611 }, { "epoch": 1.079294849637055, "grad_norm": 0.6381859119969522, "learning_rate": 4.6018733392942795e-06, "loss": 0.2981, "step": 15612 }, { "epoch": 1.079363982025579, "grad_norm": 0.6520171900224898, "learning_rate": 4.6013153598815544e-06, "loss": 0.3173, "step": 15613 }, { "epoch": 1.079433114414103, "grad_norm": 0.6344681849502899, "learning_rate": 4.6007573854656275e-06, "loss": 0.3108, "step": 15614 }, { "epoch": 1.079502246802627, "grad_norm": 0.6534538163125175, "learning_rate": 4.600199416053497e-06, "loss": 0.255, "step": 15615 }, { "epoch": 1.079571379191151, "grad_norm": 0.64734591717864, "learning_rate": 4.5996414516521515e-06, "loss": 0.386, "step": 15616 }, { "epoch": 1.079640511579675, "grad_norm": 0.9170040974822719, "learning_rate": 4.599083492268589e-06, "loss": 0.2591, "step": 15617 }, { "epoch": 1.079709643968199, "grad_norm": 0.6018000812625242, "learning_rate": 4.598525537909796e-06, "loss": 0.2319, "step": 15618 }, { "epoch": 1.079778776356723, "grad_norm": 0.6627963743093189, "learning_rate": 4.597967588582771e-06, "loss": 0.3279, "step": 15619 }, { "epoch": 1.079847908745247, "grad_norm": 0.6447404931597135, "learning_rate": 4.597409644294503e-06, "loss": 0.3348, "step": 15620 }, { "epoch": 1.079917041133771, "grad_norm": 0.6869230596389281, "learning_rate": 4.596851705051989e-06, "loss": 0.2524, "step": 15621 }, { "epoch": 1.079986173522295, "grad_norm": 0.6692104329738779, "learning_rate": 4.596293770862217e-06, "loss": 0.3107, "step": 15622 }, { "epoch": 1.080055305910819, "grad_norm": 0.7891967397728813, "learning_rate": 4.595735841732184e-06, "loss": 0.363, "step": 15623 }, { "epoch": 1.0801244382993431, "grad_norm": 0.6648855305524911, "learning_rate": 4.595177917668879e-06, "loss": 0.2556, "step": 15624 }, { "epoch": 1.0801935706878674, "grad_norm": 0.7553210456590336, "learning_rate": 4.594619998679299e-06, "loss": 0.2326, "step": 15625 }, { "epoch": 1.0802627030763914, "grad_norm": 0.644726159708292, "learning_rate": 4.594062084770432e-06, "loss": 0.2762, "step": 15626 }, { "epoch": 1.0803318354649154, "grad_norm": 0.6489356068729949, "learning_rate": 4.593504175949273e-06, "loss": 0.322, "step": 15627 }, { "epoch": 1.0804009678534394, "grad_norm": 0.6671550765807558, "learning_rate": 4.592946272222813e-06, "loss": 0.2926, "step": 15628 }, { "epoch": 1.0804701002419634, "grad_norm": 0.6489665643658947, "learning_rate": 4.592388373598046e-06, "loss": 0.2616, "step": 15629 }, { "epoch": 1.0805392326304875, "grad_norm": 0.7094417939045313, "learning_rate": 4.591830480081962e-06, "loss": 0.3383, "step": 15630 }, { "epoch": 1.0806083650190115, "grad_norm": 0.6565155272594025, "learning_rate": 4.5912725916815555e-06, "loss": 0.3258, "step": 15631 }, { "epoch": 1.0806774974075355, "grad_norm": 0.6330304736960585, "learning_rate": 4.590714708403819e-06, "loss": 0.2786, "step": 15632 }, { "epoch": 1.0807466297960595, "grad_norm": 0.6876653999623562, "learning_rate": 4.5901568302557404e-06, "loss": 0.2396, "step": 15633 }, { "epoch": 1.0808157621845835, "grad_norm": 0.6565241664142113, "learning_rate": 4.589598957244317e-06, "loss": 0.3392, "step": 15634 }, { "epoch": 1.0808848945731075, "grad_norm": 0.651621705235523, "learning_rate": 4.5890410893765365e-06, "loss": 0.23, "step": 15635 }, { "epoch": 1.0809540269616316, "grad_norm": 0.5934170321723897, "learning_rate": 4.588483226659393e-06, "loss": 0.2189, "step": 15636 }, { "epoch": 1.0810231593501556, "grad_norm": 2.1697945331310815, "learning_rate": 4.587925369099878e-06, "loss": 0.3444, "step": 15637 }, { "epoch": 1.0810922917386796, "grad_norm": 0.7050442178406295, "learning_rate": 4.5873675167049845e-06, "loss": 0.2885, "step": 15638 }, { "epoch": 1.0811614241272036, "grad_norm": 0.6228572780894136, "learning_rate": 4.5868096694817e-06, "loss": 0.2024, "step": 15639 }, { "epoch": 1.0812305565157276, "grad_norm": 0.7200412522163192, "learning_rate": 4.586251827437021e-06, "loss": 0.2699, "step": 15640 }, { "epoch": 1.0812996889042517, "grad_norm": 0.7381302849047978, "learning_rate": 4.585693990577935e-06, "loss": 0.3592, "step": 15641 }, { "epoch": 1.0813688212927757, "grad_norm": 0.5899542314986398, "learning_rate": 4.585136158911438e-06, "loss": 0.2014, "step": 15642 }, { "epoch": 1.0814379536812997, "grad_norm": 0.7205807142480729, "learning_rate": 4.584578332444518e-06, "loss": 0.2892, "step": 15643 }, { "epoch": 1.0815070860698237, "grad_norm": 0.6114474035992511, "learning_rate": 4.584020511184167e-06, "loss": 0.3083, "step": 15644 }, { "epoch": 1.0815762184583477, "grad_norm": 0.6610349432562211, "learning_rate": 4.583462695137375e-06, "loss": 0.3046, "step": 15645 }, { "epoch": 1.0816453508468717, "grad_norm": 0.7563762430086843, "learning_rate": 4.582904884311138e-06, "loss": 0.2236, "step": 15646 }, { "epoch": 1.0817144832353958, "grad_norm": 0.7914954493486897, "learning_rate": 4.582347078712441e-06, "loss": 0.2339, "step": 15647 }, { "epoch": 1.0817836156239198, "grad_norm": 0.6685253551887728, "learning_rate": 4.581789278348281e-06, "loss": 0.311, "step": 15648 }, { "epoch": 1.0818527480124438, "grad_norm": 2.38408520147208, "learning_rate": 4.581231483225643e-06, "loss": 0.3132, "step": 15649 }, { "epoch": 1.0819218804009678, "grad_norm": 0.6838430935974027, "learning_rate": 4.580673693351522e-06, "loss": 0.2624, "step": 15650 }, { "epoch": 1.0819910127894918, "grad_norm": 0.6699779557540578, "learning_rate": 4.580115908732908e-06, "loss": 0.3151, "step": 15651 }, { "epoch": 1.0820601451780159, "grad_norm": 0.8056129401269431, "learning_rate": 4.5795581293767935e-06, "loss": 0.3949, "step": 15652 }, { "epoch": 1.0821292775665399, "grad_norm": 0.6244013292816576, "learning_rate": 4.579000355290165e-06, "loss": 0.2338, "step": 15653 }, { "epoch": 1.082198409955064, "grad_norm": 0.7451734510921716, "learning_rate": 4.578442586480017e-06, "loss": 0.2989, "step": 15654 }, { "epoch": 1.082267542343588, "grad_norm": 0.6721880134817395, "learning_rate": 4.577884822953338e-06, "loss": 0.2993, "step": 15655 }, { "epoch": 1.082336674732112, "grad_norm": 0.612884831393226, "learning_rate": 4.577327064717121e-06, "loss": 0.2402, "step": 15656 }, { "epoch": 1.082405807120636, "grad_norm": 0.6948673902361424, "learning_rate": 4.576769311778354e-06, "loss": 0.2673, "step": 15657 }, { "epoch": 1.08247493950916, "grad_norm": 0.6775604973103397, "learning_rate": 4.576211564144028e-06, "loss": 0.2974, "step": 15658 }, { "epoch": 1.082544071897684, "grad_norm": 0.6920032831011876, "learning_rate": 4.575653821821134e-06, "loss": 0.3549, "step": 15659 }, { "epoch": 1.082613204286208, "grad_norm": 0.7067444413403646, "learning_rate": 4.575096084816664e-06, "loss": 0.3147, "step": 15660 }, { "epoch": 1.082682336674732, "grad_norm": 0.624776181846177, "learning_rate": 4.5745383531376036e-06, "loss": 0.2382, "step": 15661 }, { "epoch": 1.082751469063256, "grad_norm": 0.7207797698472443, "learning_rate": 4.573980626790947e-06, "loss": 0.3347, "step": 15662 }, { "epoch": 1.0828206014517803, "grad_norm": 0.7555597656155733, "learning_rate": 4.573422905783682e-06, "loss": 0.3024, "step": 15663 }, { "epoch": 1.0828897338403043, "grad_norm": 0.5881087595077127, "learning_rate": 4.572865190122802e-06, "loss": 0.2106, "step": 15664 }, { "epoch": 1.0829588662288283, "grad_norm": 0.72777406688802, "learning_rate": 4.572307479815292e-06, "loss": 0.2997, "step": 15665 }, { "epoch": 1.0830279986173523, "grad_norm": 0.6745372674781583, "learning_rate": 4.571749774868145e-06, "loss": 0.4358, "step": 15666 }, { "epoch": 1.0830971310058763, "grad_norm": 0.6715270522554699, "learning_rate": 4.57119207528835e-06, "loss": 0.2807, "step": 15667 }, { "epoch": 1.0831662633944004, "grad_norm": 1.542204732889847, "learning_rate": 4.570634381082898e-06, "loss": 0.2971, "step": 15668 }, { "epoch": 1.0832353957829244, "grad_norm": 1.6171950040414018, "learning_rate": 4.570076692258777e-06, "loss": 0.3446, "step": 15669 }, { "epoch": 1.0833045281714484, "grad_norm": 0.7217562911495603, "learning_rate": 4.569519008822979e-06, "loss": 0.3183, "step": 15670 }, { "epoch": 1.0833736605599724, "grad_norm": 1.030111207805797, "learning_rate": 4.568961330782491e-06, "loss": 0.2821, "step": 15671 }, { "epoch": 1.0834427929484964, "grad_norm": 0.7324357961378696, "learning_rate": 4.568403658144303e-06, "loss": 0.3682, "step": 15672 }, { "epoch": 1.0835119253370205, "grad_norm": 0.6521020849744193, "learning_rate": 4.567845990915405e-06, "loss": 0.3706, "step": 15673 }, { "epoch": 1.0835810577255445, "grad_norm": 0.773628918352605, "learning_rate": 4.567288329102788e-06, "loss": 0.2127, "step": 15674 }, { "epoch": 1.0836501901140685, "grad_norm": 0.6501240502984496, "learning_rate": 4.566730672713437e-06, "loss": 0.2568, "step": 15675 }, { "epoch": 1.0837193225025925, "grad_norm": 0.6975640500330378, "learning_rate": 4.566173021754346e-06, "loss": 0.3477, "step": 15676 }, { "epoch": 1.0837884548911165, "grad_norm": 1.095989606484068, "learning_rate": 4.5656153762325e-06, "loss": 0.355, "step": 15677 }, { "epoch": 1.0838575872796405, "grad_norm": 0.6480140357395611, "learning_rate": 4.565057736154892e-06, "loss": 0.2511, "step": 15678 }, { "epoch": 1.0839267196681646, "grad_norm": 0.711314463547162, "learning_rate": 4.564500101528508e-06, "loss": 0.2859, "step": 15679 }, { "epoch": 1.0839958520566886, "grad_norm": 0.798901038666867, "learning_rate": 4.563942472360338e-06, "loss": 0.3592, "step": 15680 }, { "epoch": 1.0840649844452126, "grad_norm": 0.6169396985664986, "learning_rate": 4.5633848486573716e-06, "loss": 0.2504, "step": 15681 }, { "epoch": 1.0841341168337366, "grad_norm": 1.034318738703226, "learning_rate": 4.562827230426598e-06, "loss": 0.2724, "step": 15682 }, { "epoch": 1.0842032492222606, "grad_norm": 3.1187455085004614, "learning_rate": 4.562269617675002e-06, "loss": 0.2914, "step": 15683 }, { "epoch": 1.0842723816107847, "grad_norm": 0.8669796908082463, "learning_rate": 4.561712010409577e-06, "loss": 0.3382, "step": 15684 }, { "epoch": 1.0843415139993087, "grad_norm": 0.6749324661273663, "learning_rate": 4.561154408637309e-06, "loss": 0.2532, "step": 15685 }, { "epoch": 1.0844106463878327, "grad_norm": 0.6932729473166845, "learning_rate": 4.560596812365188e-06, "loss": 0.2684, "step": 15686 }, { "epoch": 1.0844797787763567, "grad_norm": 0.7219452502620288, "learning_rate": 4.5600392216002e-06, "loss": 0.3423, "step": 15687 }, { "epoch": 1.0845489111648807, "grad_norm": 0.7549525234357074, "learning_rate": 4.5594816363493384e-06, "loss": 0.3255, "step": 15688 }, { "epoch": 1.0846180435534047, "grad_norm": 0.6573551429545047, "learning_rate": 4.558924056619586e-06, "loss": 0.2759, "step": 15689 }, { "epoch": 1.0846871759419288, "grad_norm": 0.7215913165284839, "learning_rate": 4.558366482417934e-06, "loss": 0.2793, "step": 15690 }, { "epoch": 1.0847563083304528, "grad_norm": 0.7436427920801189, "learning_rate": 4.55780891375137e-06, "loss": 0.3449, "step": 15691 }, { "epoch": 1.0848254407189768, "grad_norm": 0.6187139166873012, "learning_rate": 4.5572513506268835e-06, "loss": 0.2673, "step": 15692 }, { "epoch": 1.0848945731075008, "grad_norm": 0.6228183275133492, "learning_rate": 4.556693793051459e-06, "loss": 0.2527, "step": 15693 }, { "epoch": 1.0849637054960248, "grad_norm": 0.8042109159788284, "learning_rate": 4.556136241032088e-06, "loss": 0.3521, "step": 15694 }, { "epoch": 1.0850328378845489, "grad_norm": 0.7085064305281958, "learning_rate": 4.555578694575756e-06, "loss": 0.2878, "step": 15695 }, { "epoch": 1.0851019702730729, "grad_norm": 0.7454618980405425, "learning_rate": 4.555021153689454e-06, "loss": 0.3083, "step": 15696 }, { "epoch": 1.0851711026615969, "grad_norm": 0.6948963365073528, "learning_rate": 4.554463618380166e-06, "loss": 0.2825, "step": 15697 }, { "epoch": 1.085240235050121, "grad_norm": 0.7508343646071021, "learning_rate": 4.553906088654882e-06, "loss": 0.3911, "step": 15698 }, { "epoch": 1.085309367438645, "grad_norm": 0.9444789084252236, "learning_rate": 4.55334856452059e-06, "loss": 0.251, "step": 15699 }, { "epoch": 1.085378499827169, "grad_norm": 0.8297290102969566, "learning_rate": 4.552791045984275e-06, "loss": 0.2497, "step": 15700 }, { "epoch": 1.085447632215693, "grad_norm": 0.700236159064756, "learning_rate": 4.5522335330529275e-06, "loss": 0.3453, "step": 15701 }, { "epoch": 1.085516764604217, "grad_norm": 0.7566476997249088, "learning_rate": 4.551676025733532e-06, "loss": 0.3336, "step": 15702 }, { "epoch": 1.085585896992741, "grad_norm": 0.6283107596907884, "learning_rate": 4.551118524033079e-06, "loss": 0.2343, "step": 15703 }, { "epoch": 1.085655029381265, "grad_norm": 0.6734675440638904, "learning_rate": 4.550561027958552e-06, "loss": 0.2442, "step": 15704 }, { "epoch": 1.085724161769789, "grad_norm": 0.7372533925557032, "learning_rate": 4.5500035375169434e-06, "loss": 0.4082, "step": 15705 }, { "epoch": 1.0857932941583133, "grad_norm": 0.6694662594423103, "learning_rate": 4.549446052715235e-06, "loss": 0.3239, "step": 15706 }, { "epoch": 1.0858624265468373, "grad_norm": 0.637050703098634, "learning_rate": 4.548888573560418e-06, "loss": 0.2273, "step": 15707 }, { "epoch": 1.0859315589353613, "grad_norm": 0.6504008345454034, "learning_rate": 4.548331100059477e-06, "loss": 0.2953, "step": 15708 }, { "epoch": 1.0860006913238853, "grad_norm": 0.6188197151399677, "learning_rate": 4.5477736322194e-06, "loss": 0.3391, "step": 15709 }, { "epoch": 1.0860698237124093, "grad_norm": 0.6683275695173562, "learning_rate": 4.547216170047173e-06, "loss": 0.244, "step": 15710 }, { "epoch": 1.0861389561009334, "grad_norm": 0.6604251769814511, "learning_rate": 4.546658713549783e-06, "loss": 0.2331, "step": 15711 }, { "epoch": 1.0862080884894574, "grad_norm": 0.6654464029206054, "learning_rate": 4.546101262734217e-06, "loss": 0.2999, "step": 15712 }, { "epoch": 1.0862772208779814, "grad_norm": 0.6941891014278355, "learning_rate": 4.545543817607464e-06, "loss": 0.2989, "step": 15713 }, { "epoch": 1.0863463532665054, "grad_norm": 0.7061643654626976, "learning_rate": 4.544986378176506e-06, "loss": 0.2421, "step": 15714 }, { "epoch": 1.0864154856550294, "grad_norm": 0.6801326713356856, "learning_rate": 4.544428944448331e-06, "loss": 0.3465, "step": 15715 }, { "epoch": 1.0864846180435535, "grad_norm": 0.6477100703794391, "learning_rate": 4.543871516429927e-06, "loss": 0.3714, "step": 15716 }, { "epoch": 1.0865537504320775, "grad_norm": 0.669709671010017, "learning_rate": 4.543314094128281e-06, "loss": 0.2654, "step": 15717 }, { "epoch": 1.0866228828206015, "grad_norm": 0.5665979778994031, "learning_rate": 4.542756677550376e-06, "loss": 0.2161, "step": 15718 }, { "epoch": 1.0866920152091255, "grad_norm": 0.711753403086786, "learning_rate": 4.542199266703201e-06, "loss": 0.3093, "step": 15719 }, { "epoch": 1.0867611475976495, "grad_norm": 0.7484716642127607, "learning_rate": 4.54164186159374e-06, "loss": 0.3301, "step": 15720 }, { "epoch": 1.0868302799861735, "grad_norm": 0.7178726615498717, "learning_rate": 4.541084462228981e-06, "loss": 0.2557, "step": 15721 }, { "epoch": 1.0868994123746976, "grad_norm": 0.6201689575168144, "learning_rate": 4.5405270686159085e-06, "loss": 0.2689, "step": 15722 }, { "epoch": 1.0869685447632216, "grad_norm": 0.7554878823856386, "learning_rate": 4.539969680761512e-06, "loss": 0.3718, "step": 15723 }, { "epoch": 1.0870376771517456, "grad_norm": 0.6364466098012598, "learning_rate": 4.539412298672772e-06, "loss": 0.2664, "step": 15724 }, { "epoch": 1.0871068095402696, "grad_norm": 0.6143501501021751, "learning_rate": 4.538854922356678e-06, "loss": 0.2084, "step": 15725 }, { "epoch": 1.0871759419287936, "grad_norm": 0.7085444256936952, "learning_rate": 4.538297551820213e-06, "loss": 0.3309, "step": 15726 }, { "epoch": 1.0872450743173177, "grad_norm": 0.7727695675700714, "learning_rate": 4.537740187070367e-06, "loss": 0.296, "step": 15727 }, { "epoch": 1.0873142067058417, "grad_norm": 0.6642033104586972, "learning_rate": 4.53718282811412e-06, "loss": 0.2413, "step": 15728 }, { "epoch": 1.0873833390943657, "grad_norm": 0.667452776193322, "learning_rate": 4.5366254749584624e-06, "loss": 0.2899, "step": 15729 }, { "epoch": 1.0874524714828897, "grad_norm": 0.740389629432948, "learning_rate": 4.536068127610376e-06, "loss": 0.3676, "step": 15730 }, { "epoch": 1.0875216038714137, "grad_norm": 0.6589686612913126, "learning_rate": 4.53551078607685e-06, "loss": 0.2854, "step": 15731 }, { "epoch": 1.0875907362599377, "grad_norm": 0.6413626597282703, "learning_rate": 4.5349534503648655e-06, "loss": 0.2795, "step": 15732 }, { "epoch": 1.0876598686484618, "grad_norm": 0.6368841327608542, "learning_rate": 4.53439612048141e-06, "loss": 0.2721, "step": 15733 }, { "epoch": 1.0877290010369858, "grad_norm": 0.6793965495841852, "learning_rate": 4.533838796433468e-06, "loss": 0.4017, "step": 15734 }, { "epoch": 1.0877981334255098, "grad_norm": 0.664447700897425, "learning_rate": 4.533281478228026e-06, "loss": 0.2777, "step": 15735 }, { "epoch": 1.0878672658140338, "grad_norm": 0.6804112713891727, "learning_rate": 4.5327241658720654e-06, "loss": 0.2495, "step": 15736 }, { "epoch": 1.0879363982025578, "grad_norm": 0.6245520573411897, "learning_rate": 4.532166859372575e-06, "loss": 0.3265, "step": 15737 }, { "epoch": 1.0880055305910818, "grad_norm": 0.6404469228995563, "learning_rate": 4.531609558736538e-06, "loss": 0.332, "step": 15738 }, { "epoch": 1.0880746629796059, "grad_norm": 0.5915343670672842, "learning_rate": 4.5310522639709385e-06, "loss": 0.2384, "step": 15739 }, { "epoch": 1.0881437953681299, "grad_norm": 0.6672608983942662, "learning_rate": 4.530494975082762e-06, "loss": 0.3116, "step": 15740 }, { "epoch": 1.088212927756654, "grad_norm": 0.6966381317614266, "learning_rate": 4.529937692078995e-06, "loss": 0.3987, "step": 15741 }, { "epoch": 1.088282060145178, "grad_norm": 1.3114429364709497, "learning_rate": 4.5293804149666184e-06, "loss": 0.3195, "step": 15742 }, { "epoch": 1.088351192533702, "grad_norm": 0.647019337634446, "learning_rate": 4.52882314375262e-06, "loss": 0.2457, "step": 15743 }, { "epoch": 1.0884203249222262, "grad_norm": 0.6961022588311258, "learning_rate": 4.52826587844398e-06, "loss": 0.3144, "step": 15744 }, { "epoch": 1.0884894573107502, "grad_norm": 0.6317380102211492, "learning_rate": 4.527708619047689e-06, "loss": 0.2817, "step": 15745 }, { "epoch": 1.0885585896992742, "grad_norm": 0.7184407780378749, "learning_rate": 4.527151365570724e-06, "loss": 0.3108, "step": 15746 }, { "epoch": 1.0886277220877982, "grad_norm": 0.6965714802755723, "learning_rate": 4.5265941180200756e-06, "loss": 0.2639, "step": 15747 }, { "epoch": 1.0886968544763223, "grad_norm": 1.35475553279864, "learning_rate": 4.5260368764027225e-06, "loss": 0.3448, "step": 15748 }, { "epoch": 1.0887659868648463, "grad_norm": 0.7064529446077217, "learning_rate": 4.525479640725655e-06, "loss": 0.2538, "step": 15749 }, { "epoch": 1.0888351192533703, "grad_norm": 0.6965556832016551, "learning_rate": 4.52492241099585e-06, "loss": 0.2742, "step": 15750 }, { "epoch": 1.0889042516418943, "grad_norm": 0.7227248904311817, "learning_rate": 4.524365187220297e-06, "loss": 0.3558, "step": 15751 }, { "epoch": 1.0889733840304183, "grad_norm": 0.6984092375296752, "learning_rate": 4.523807969405975e-06, "loss": 0.3527, "step": 15752 }, { "epoch": 1.0890425164189423, "grad_norm": 0.8193177370757723, "learning_rate": 4.523250757559873e-06, "loss": 0.2767, "step": 15753 }, { "epoch": 1.0891116488074664, "grad_norm": 0.6295710149087345, "learning_rate": 4.522693551688971e-06, "loss": 0.2333, "step": 15754 }, { "epoch": 1.0891807811959904, "grad_norm": 0.6771032688559605, "learning_rate": 4.522136351800253e-06, "loss": 0.3286, "step": 15755 }, { "epoch": 1.0892499135845144, "grad_norm": 0.7075480737559761, "learning_rate": 4.521579157900702e-06, "loss": 0.3014, "step": 15756 }, { "epoch": 1.0893190459730384, "grad_norm": 0.6683075860094467, "learning_rate": 4.521021969997304e-06, "loss": 0.2436, "step": 15757 }, { "epoch": 1.0893881783615624, "grad_norm": 0.7218218927424561, "learning_rate": 4.52046478809704e-06, "loss": 0.2987, "step": 15758 }, { "epoch": 1.0894573107500864, "grad_norm": 0.7051313526698849, "learning_rate": 4.519907612206896e-06, "loss": 0.3617, "step": 15759 }, { "epoch": 1.0895264431386105, "grad_norm": 0.6640553064553877, "learning_rate": 4.519350442333852e-06, "loss": 0.234, "step": 15760 }, { "epoch": 1.0895955755271345, "grad_norm": 0.6740545745693698, "learning_rate": 4.518793278484893e-06, "loss": 0.3107, "step": 15761 }, { "epoch": 1.0896647079156585, "grad_norm": 0.6914822047033062, "learning_rate": 4.518236120667e-06, "loss": 0.3516, "step": 15762 }, { "epoch": 1.0897338403041825, "grad_norm": 0.6797118028149238, "learning_rate": 4.517678968887159e-06, "loss": 0.3001, "step": 15763 }, { "epoch": 1.0898029726927065, "grad_norm": 0.7041890647700613, "learning_rate": 4.51712182315235e-06, "loss": 0.2629, "step": 15764 }, { "epoch": 1.0898721050812306, "grad_norm": 0.6342298961457756, "learning_rate": 4.5165646834695595e-06, "loss": 0.2146, "step": 15765 }, { "epoch": 1.0899412374697546, "grad_norm": 0.6406161467566245, "learning_rate": 4.516007549845768e-06, "loss": 0.367, "step": 15766 }, { "epoch": 1.0900103698582786, "grad_norm": 0.6234157964171451, "learning_rate": 4.515450422287956e-06, "loss": 0.2369, "step": 15767 }, { "epoch": 1.0900795022468026, "grad_norm": 0.6564831460987443, "learning_rate": 4.514893300803109e-06, "loss": 0.2614, "step": 15768 }, { "epoch": 1.0901486346353266, "grad_norm": 0.6625008547453605, "learning_rate": 4.5143361853982075e-06, "loss": 0.3274, "step": 15769 }, { "epoch": 1.0902177670238506, "grad_norm": 0.6447868472800958, "learning_rate": 4.513779076080238e-06, "loss": 0.2961, "step": 15770 }, { "epoch": 1.0902868994123747, "grad_norm": 0.6439141593352609, "learning_rate": 4.5132219728561774e-06, "loss": 0.2461, "step": 15771 }, { "epoch": 1.0903560318008987, "grad_norm": 0.6668158741517742, "learning_rate": 4.512664875733013e-06, "loss": 0.2776, "step": 15772 }, { "epoch": 1.0904251641894227, "grad_norm": 0.677892821517621, "learning_rate": 4.512107784717722e-06, "loss": 0.4176, "step": 15773 }, { "epoch": 1.0904942965779467, "grad_norm": 0.7688346658044539, "learning_rate": 4.511550699817293e-06, "loss": 0.3227, "step": 15774 }, { "epoch": 1.0905634289664707, "grad_norm": 0.6010276894134139, "learning_rate": 4.510993621038699e-06, "loss": 0.239, "step": 15775 }, { "epoch": 1.0906325613549948, "grad_norm": 0.699538359571843, "learning_rate": 4.510436548388932e-06, "loss": 0.3594, "step": 15776 }, { "epoch": 1.0907016937435188, "grad_norm": 0.6647289829878352, "learning_rate": 4.509879481874967e-06, "loss": 0.2941, "step": 15777 }, { "epoch": 1.0907708261320428, "grad_norm": 0.7052877632993043, "learning_rate": 4.509322421503789e-06, "loss": 0.2204, "step": 15778 }, { "epoch": 1.0908399585205668, "grad_norm": 0.6461427886716087, "learning_rate": 4.508765367282377e-06, "loss": 0.2668, "step": 15779 }, { "epoch": 1.0909090909090908, "grad_norm": 0.6496270981566732, "learning_rate": 4.508208319217718e-06, "loss": 0.3032, "step": 15780 }, { "epoch": 1.0909782232976148, "grad_norm": 0.6734850011532605, "learning_rate": 4.507651277316786e-06, "loss": 0.2653, "step": 15781 }, { "epoch": 1.0910473556861389, "grad_norm": 0.6613324985922869, "learning_rate": 4.50709424158657e-06, "loss": 0.2219, "step": 15782 }, { "epoch": 1.0911164880746629, "grad_norm": 0.7240500651844765, "learning_rate": 4.506537212034045e-06, "loss": 0.281, "step": 15783 }, { "epoch": 1.091185620463187, "grad_norm": 0.6728179059956407, "learning_rate": 4.5059801886661985e-06, "loss": 0.3423, "step": 15784 }, { "epoch": 1.091254752851711, "grad_norm": 2.0080656886634176, "learning_rate": 4.505423171490006e-06, "loss": 0.2491, "step": 15785 }, { "epoch": 1.091323885240235, "grad_norm": 0.642427055507309, "learning_rate": 4.504866160512453e-06, "loss": 0.2482, "step": 15786 }, { "epoch": 1.0913930176287592, "grad_norm": 0.7083153419591877, "learning_rate": 4.504309155740518e-06, "loss": 0.3421, "step": 15787 }, { "epoch": 1.0914621500172832, "grad_norm": 0.8040016301767402, "learning_rate": 4.503752157181185e-06, "loss": 0.3478, "step": 15788 }, { "epoch": 1.0915312824058072, "grad_norm": 0.5972888721171232, "learning_rate": 4.503195164841431e-06, "loss": 0.1985, "step": 15789 }, { "epoch": 1.0916004147943312, "grad_norm": 0.8052700936134322, "learning_rate": 4.502638178728241e-06, "loss": 0.2999, "step": 15790 }, { "epoch": 1.0916695471828552, "grad_norm": 0.6647031301391728, "learning_rate": 4.502081198848592e-06, "loss": 0.3667, "step": 15791 }, { "epoch": 1.0917386795713793, "grad_norm": 0.6563139008278552, "learning_rate": 4.501524225209468e-06, "loss": 0.2556, "step": 15792 }, { "epoch": 1.0918078119599033, "grad_norm": 0.7539673043315196, "learning_rate": 4.500967257817847e-06, "loss": 0.2762, "step": 15793 }, { "epoch": 1.0918769443484273, "grad_norm": 0.681031716538267, "learning_rate": 4.500410296680712e-06, "loss": 0.4186, "step": 15794 }, { "epoch": 1.0919460767369513, "grad_norm": 0.6373187520114197, "learning_rate": 4.499853341805041e-06, "loss": 0.2629, "step": 15795 }, { "epoch": 1.0920152091254753, "grad_norm": 0.6810244294228469, "learning_rate": 4.499296393197817e-06, "loss": 0.2499, "step": 15796 }, { "epoch": 1.0920843415139994, "grad_norm": 0.6831460304164553, "learning_rate": 4.498739450866018e-06, "loss": 0.2536, "step": 15797 }, { "epoch": 1.0921534739025234, "grad_norm": 0.5979249481958534, "learning_rate": 4.498182514816628e-06, "loss": 0.3328, "step": 15798 }, { "epoch": 1.0922226062910474, "grad_norm": 0.6315326701915015, "learning_rate": 4.497625585056622e-06, "loss": 0.2989, "step": 15799 }, { "epoch": 1.0922917386795714, "grad_norm": 0.6069032102258969, "learning_rate": 4.497068661592984e-06, "loss": 0.1963, "step": 15800 }, { "epoch": 1.0923608710680954, "grad_norm": 0.7102681230595927, "learning_rate": 4.496511744432692e-06, "loss": 0.3069, "step": 15801 }, { "epoch": 1.0924300034566194, "grad_norm": 0.7138703901100105, "learning_rate": 4.495954833582729e-06, "loss": 0.338, "step": 15802 }, { "epoch": 1.0924991358451435, "grad_norm": 0.6436691157000151, "learning_rate": 4.49539792905007e-06, "loss": 0.2797, "step": 15803 }, { "epoch": 1.0925682682336675, "grad_norm": 1.4215468851206576, "learning_rate": 4.494841030841698e-06, "loss": 0.2624, "step": 15804 }, { "epoch": 1.0926374006221915, "grad_norm": 0.6813861787958143, "learning_rate": 4.494284138964593e-06, "loss": 0.3416, "step": 15805 }, { "epoch": 1.0927065330107155, "grad_norm": 0.624080457075192, "learning_rate": 4.493727253425734e-06, "loss": 0.26, "step": 15806 }, { "epoch": 1.0927756653992395, "grad_norm": 0.618079685139155, "learning_rate": 4.4931703742321e-06, "loss": 0.2724, "step": 15807 }, { "epoch": 1.0928447977877636, "grad_norm": 0.6716392111224383, "learning_rate": 4.49261350139067e-06, "loss": 0.3339, "step": 15808 }, { "epoch": 1.0929139301762876, "grad_norm": 0.7274986959068718, "learning_rate": 4.492056634908424e-06, "loss": 0.323, "step": 15809 }, { "epoch": 1.0929830625648116, "grad_norm": 0.670597796621995, "learning_rate": 4.491499774792343e-06, "loss": 0.2262, "step": 15810 }, { "epoch": 1.0930521949533356, "grad_norm": 0.698131233892974, "learning_rate": 4.490942921049404e-06, "loss": 0.2613, "step": 15811 }, { "epoch": 1.0931213273418596, "grad_norm": 0.69373251130035, "learning_rate": 4.490386073686589e-06, "loss": 0.3673, "step": 15812 }, { "epoch": 1.0931904597303836, "grad_norm": 0.6454285822600475, "learning_rate": 4.489829232710873e-06, "loss": 0.2778, "step": 15813 }, { "epoch": 1.0932595921189077, "grad_norm": 0.6591627957425428, "learning_rate": 4.489272398129237e-06, "loss": 0.2379, "step": 15814 }, { "epoch": 1.0933287245074317, "grad_norm": 0.6761990124105356, "learning_rate": 4.488715569948661e-06, "loss": 0.2355, "step": 15815 }, { "epoch": 1.0933978568959557, "grad_norm": 0.6708976117677704, "learning_rate": 4.488158748176124e-06, "loss": 0.3409, "step": 15816 }, { "epoch": 1.0934669892844797, "grad_norm": 0.8021596152112842, "learning_rate": 4.487601932818601e-06, "loss": 0.2764, "step": 15817 }, { "epoch": 1.0935361216730037, "grad_norm": 0.6552616692891011, "learning_rate": 4.487045123883075e-06, "loss": 0.249, "step": 15818 }, { "epoch": 1.0936052540615278, "grad_norm": 0.7506995236420644, "learning_rate": 4.486488321376522e-06, "loss": 0.3088, "step": 15819 }, { "epoch": 1.0936743864500518, "grad_norm": 0.7032181451723477, "learning_rate": 4.485931525305924e-06, "loss": 0.3238, "step": 15820 }, { "epoch": 1.0937435188385758, "grad_norm": 0.6098164914764004, "learning_rate": 4.485374735678254e-06, "loss": 0.2502, "step": 15821 }, { "epoch": 1.0938126512270998, "grad_norm": 0.6870002266342561, "learning_rate": 4.4848179525004945e-06, "loss": 0.2459, "step": 15822 }, { "epoch": 1.0938817836156238, "grad_norm": 0.7135446436968409, "learning_rate": 4.484261175779622e-06, "loss": 0.3408, "step": 15823 }, { "epoch": 1.0939509160041478, "grad_norm": 0.6823737751416301, "learning_rate": 4.483704405522617e-06, "loss": 0.2461, "step": 15824 }, { "epoch": 1.094020048392672, "grad_norm": 0.7092513621127934, "learning_rate": 4.483147641736454e-06, "loss": 0.2733, "step": 15825 }, { "epoch": 1.094089180781196, "grad_norm": 0.6547634250069434, "learning_rate": 4.482590884428114e-06, "loss": 0.2563, "step": 15826 }, { "epoch": 1.0941583131697201, "grad_norm": 0.6365223722944485, "learning_rate": 4.482034133604573e-06, "loss": 0.3459, "step": 15827 }, { "epoch": 1.0942274455582441, "grad_norm": 0.568502840974669, "learning_rate": 4.48147738927281e-06, "loss": 0.206, "step": 15828 }, { "epoch": 1.0942965779467682, "grad_norm": 0.6822588939627312, "learning_rate": 4.480920651439803e-06, "loss": 0.2811, "step": 15829 }, { "epoch": 1.0943657103352922, "grad_norm": 0.6632825643519794, "learning_rate": 4.48036392011253e-06, "loss": 0.3543, "step": 15830 }, { "epoch": 1.0944348427238162, "grad_norm": 0.7427994486723669, "learning_rate": 4.479807195297967e-06, "loss": 0.2679, "step": 15831 }, { "epoch": 1.0945039751123402, "grad_norm": 0.6055090656041885, "learning_rate": 4.479250477003092e-06, "loss": 0.2526, "step": 15832 }, { "epoch": 1.0945731075008642, "grad_norm": 0.6605801601158389, "learning_rate": 4.478693765234885e-06, "loss": 0.2883, "step": 15833 }, { "epoch": 1.0946422398893882, "grad_norm": 0.7622438760516899, "learning_rate": 4.478137060000319e-06, "loss": 0.3601, "step": 15834 }, { "epoch": 1.0947113722779123, "grad_norm": 0.6559309175838542, "learning_rate": 4.477580361306375e-06, "loss": 0.2699, "step": 15835 }, { "epoch": 1.0947805046664363, "grad_norm": 0.6462901011917027, "learning_rate": 4.477023669160028e-06, "loss": 0.2486, "step": 15836 }, { "epoch": 1.0948496370549603, "grad_norm": 0.7878708159584832, "learning_rate": 4.476466983568257e-06, "loss": 0.3891, "step": 15837 }, { "epoch": 1.0949187694434843, "grad_norm": 0.7042307226283359, "learning_rate": 4.475910304538037e-06, "loss": 0.2632, "step": 15838 }, { "epoch": 1.0949879018320083, "grad_norm": 0.6326939704249336, "learning_rate": 4.475353632076347e-06, "loss": 0.2461, "step": 15839 }, { "epoch": 1.0950570342205324, "grad_norm": 0.6837297947672896, "learning_rate": 4.4747969661901616e-06, "loss": 0.2453, "step": 15840 }, { "epoch": 1.0951261666090564, "grad_norm": 0.6736824756227906, "learning_rate": 4.474240306886461e-06, "loss": 0.348, "step": 15841 }, { "epoch": 1.0951952989975804, "grad_norm": 0.7026047266324297, "learning_rate": 4.473683654172218e-06, "loss": 0.2145, "step": 15842 }, { "epoch": 1.0952644313861044, "grad_norm": 0.8149793873687092, "learning_rate": 4.473127008054413e-06, "loss": 0.2375, "step": 15843 }, { "epoch": 1.0953335637746284, "grad_norm": 0.7660002399052179, "learning_rate": 4.472570368540018e-06, "loss": 0.3518, "step": 15844 }, { "epoch": 1.0954026961631524, "grad_norm": 0.6688622154535286, "learning_rate": 4.472013735636015e-06, "loss": 0.318, "step": 15845 }, { "epoch": 1.0954718285516765, "grad_norm": 0.5931333106807944, "learning_rate": 4.4714571093493744e-06, "loss": 0.2651, "step": 15846 }, { "epoch": 1.0955409609402005, "grad_norm": 0.653172686998725, "learning_rate": 4.4709004896870795e-06, "loss": 0.2847, "step": 15847 }, { "epoch": 1.0956100933287245, "grad_norm": 0.6414350790226919, "learning_rate": 4.4703438766561e-06, "loss": 0.3554, "step": 15848 }, { "epoch": 1.0956792257172485, "grad_norm": 0.6101968277525057, "learning_rate": 4.469787270263416e-06, "loss": 0.2337, "step": 15849 }, { "epoch": 1.0957483581057725, "grad_norm": 0.606833600000794, "learning_rate": 4.4692306705160015e-06, "loss": 0.3077, "step": 15850 }, { "epoch": 1.0958174904942966, "grad_norm": 0.6689995148205976, "learning_rate": 4.468674077420836e-06, "loss": 0.2947, "step": 15851 }, { "epoch": 1.0958866228828206, "grad_norm": 0.6818395144104518, "learning_rate": 4.468117490984889e-06, "loss": 0.3085, "step": 15852 }, { "epoch": 1.0959557552713446, "grad_norm": 0.6719733941024787, "learning_rate": 4.467560911215142e-06, "loss": 0.2468, "step": 15853 }, { "epoch": 1.0960248876598686, "grad_norm": 0.6166969109602937, "learning_rate": 4.467004338118568e-06, "loss": 0.2594, "step": 15854 }, { "epoch": 1.0960940200483926, "grad_norm": 0.6692102638188272, "learning_rate": 4.466447771702145e-06, "loss": 0.3216, "step": 15855 }, { "epoch": 1.0961631524369166, "grad_norm": 0.6389864937876047, "learning_rate": 4.465891211972846e-06, "loss": 0.2527, "step": 15856 }, { "epoch": 1.0962322848254407, "grad_norm": 0.7350095820424686, "learning_rate": 4.465334658937647e-06, "loss": 0.2594, "step": 15857 }, { "epoch": 1.0963014172139647, "grad_norm": 0.7060469689777504, "learning_rate": 4.464778112603524e-06, "loss": 0.3114, "step": 15858 }, { "epoch": 1.0963705496024887, "grad_norm": 0.6626437889590434, "learning_rate": 4.464221572977454e-06, "loss": 0.3585, "step": 15859 }, { "epoch": 1.0964396819910127, "grad_norm": 0.6343999588843952, "learning_rate": 4.463665040066409e-06, "loss": 0.2607, "step": 15860 }, { "epoch": 1.0965088143795367, "grad_norm": 0.5782107675627637, "learning_rate": 4.463108513877365e-06, "loss": 0.178, "step": 15861 }, { "epoch": 1.0965779467680608, "grad_norm": 0.7064224340008792, "learning_rate": 4.462551994417297e-06, "loss": 0.3251, "step": 15862 }, { "epoch": 1.0966470791565848, "grad_norm": 0.6847883723154247, "learning_rate": 4.461995481693184e-06, "loss": 0.3547, "step": 15863 }, { "epoch": 1.0967162115451088, "grad_norm": 0.7057130805431043, "learning_rate": 4.461438975711994e-06, "loss": 0.2273, "step": 15864 }, { "epoch": 1.0967853439336328, "grad_norm": 0.6742854171127786, "learning_rate": 4.460882476480708e-06, "loss": 0.2926, "step": 15865 }, { "epoch": 1.0968544763221568, "grad_norm": 0.6892711559923065, "learning_rate": 4.460325984006296e-06, "loss": 0.3788, "step": 15866 }, { "epoch": 1.0969236087106808, "grad_norm": 0.6668140613836848, "learning_rate": 4.459769498295735e-06, "loss": 0.2637, "step": 15867 }, { "epoch": 1.0969927410992049, "grad_norm": 0.6480483224950282, "learning_rate": 4.459213019356e-06, "loss": 0.2595, "step": 15868 }, { "epoch": 1.097061873487729, "grad_norm": 0.7105975202831571, "learning_rate": 4.458656547194065e-06, "loss": 0.3121, "step": 15869 }, { "epoch": 1.0971310058762531, "grad_norm": 0.6967903302894423, "learning_rate": 4.458100081816903e-06, "loss": 0.3828, "step": 15870 }, { "epoch": 1.0972001382647771, "grad_norm": 0.6644644245389798, "learning_rate": 4.45754362323149e-06, "loss": 0.2872, "step": 15871 }, { "epoch": 1.0972692706533012, "grad_norm": 0.6923725760005983, "learning_rate": 4.456987171444799e-06, "loss": 0.2501, "step": 15872 }, { "epoch": 1.0973384030418252, "grad_norm": 0.685828259473472, "learning_rate": 4.456430726463807e-06, "loss": 0.3954, "step": 15873 }, { "epoch": 1.0974075354303492, "grad_norm": 1.0464573218972044, "learning_rate": 4.455874288295484e-06, "loss": 0.2275, "step": 15874 }, { "epoch": 1.0974766678188732, "grad_norm": 0.6437817464267388, "learning_rate": 4.455317856946806e-06, "loss": 0.2131, "step": 15875 }, { "epoch": 1.0975458002073972, "grad_norm": 0.6328007404698709, "learning_rate": 4.4547614324247455e-06, "loss": 0.3214, "step": 15876 }, { "epoch": 1.0976149325959212, "grad_norm": 0.6782807574121614, "learning_rate": 4.45420501473628e-06, "loss": 0.3061, "step": 15877 }, { "epoch": 1.0976840649844453, "grad_norm": 0.6577702869536747, "learning_rate": 4.4536486038883786e-06, "loss": 0.2835, "step": 15878 }, { "epoch": 1.0977531973729693, "grad_norm": 0.6378497258895445, "learning_rate": 4.4530921998880174e-06, "loss": 0.2292, "step": 15879 }, { "epoch": 1.0978223297614933, "grad_norm": 0.7381695532756533, "learning_rate": 4.452535802742169e-06, "loss": 0.3262, "step": 15880 }, { "epoch": 1.0978914621500173, "grad_norm": 0.8016794390927121, "learning_rate": 4.451979412457808e-06, "loss": 0.2838, "step": 15881 }, { "epoch": 1.0979605945385413, "grad_norm": 0.6983456535161949, "learning_rate": 4.451423029041906e-06, "loss": 0.236, "step": 15882 }, { "epoch": 1.0980297269270654, "grad_norm": 1.2366658572491886, "learning_rate": 4.45086665250144e-06, "loss": 0.2788, "step": 15883 }, { "epoch": 1.0980988593155894, "grad_norm": 0.6917193794664318, "learning_rate": 4.450310282843378e-06, "loss": 0.3524, "step": 15884 }, { "epoch": 1.0981679917041134, "grad_norm": 0.6370802194198751, "learning_rate": 4.449753920074697e-06, "loss": 0.2094, "step": 15885 }, { "epoch": 1.0982371240926374, "grad_norm": 0.6366761712386485, "learning_rate": 4.449197564202367e-06, "loss": 0.2368, "step": 15886 }, { "epoch": 1.0983062564811614, "grad_norm": 0.6485012448079303, "learning_rate": 4.448641215233365e-06, "loss": 0.4079, "step": 15887 }, { "epoch": 1.0983753888696854, "grad_norm": 0.7267111973646679, "learning_rate": 4.448084873174659e-06, "loss": 0.3508, "step": 15888 }, { "epoch": 1.0984445212582095, "grad_norm": 0.6156380041508875, "learning_rate": 4.447528538033227e-06, "loss": 0.2196, "step": 15889 }, { "epoch": 1.0985136536467335, "grad_norm": 0.8286153480131271, "learning_rate": 4.446972209816036e-06, "loss": 0.2721, "step": 15890 }, { "epoch": 1.0985827860352575, "grad_norm": 0.6654673156002439, "learning_rate": 4.446415888530064e-06, "loss": 0.403, "step": 15891 }, { "epoch": 1.0986519184237815, "grad_norm": 0.9818970028725879, "learning_rate": 4.445859574182278e-06, "loss": 0.2581, "step": 15892 }, { "epoch": 1.0987210508123055, "grad_norm": 0.5803325264518295, "learning_rate": 4.445303266779656e-06, "loss": 0.253, "step": 15893 }, { "epoch": 1.0987901832008296, "grad_norm": 0.6596663736408498, "learning_rate": 4.444746966329165e-06, "loss": 0.3187, "step": 15894 }, { "epoch": 1.0988593155893536, "grad_norm": 0.6684479739234882, "learning_rate": 4.444190672837783e-06, "loss": 0.3307, "step": 15895 }, { "epoch": 1.0989284479778776, "grad_norm": 0.6684985579754101, "learning_rate": 4.443634386312477e-06, "loss": 0.2439, "step": 15896 }, { "epoch": 1.0989975803664016, "grad_norm": 0.6718658249782827, "learning_rate": 4.443078106760222e-06, "loss": 0.2513, "step": 15897 }, { "epoch": 1.0990667127549256, "grad_norm": 0.6674073227955774, "learning_rate": 4.442521834187989e-06, "loss": 0.4236, "step": 15898 }, { "epoch": 1.0991358451434496, "grad_norm": 0.6497715229079075, "learning_rate": 4.441965568602748e-06, "loss": 0.2601, "step": 15899 }, { "epoch": 1.0992049775319737, "grad_norm": 0.6070806540012764, "learning_rate": 4.441409310011476e-06, "loss": 0.2608, "step": 15900 }, { "epoch": 1.0992741099204977, "grad_norm": 0.6881125779410703, "learning_rate": 4.440853058421139e-06, "loss": 0.2985, "step": 15901 }, { "epoch": 1.0993432423090217, "grad_norm": 0.7198137058259622, "learning_rate": 4.440296813838712e-06, "loss": 0.3912, "step": 15902 }, { "epoch": 1.0994123746975457, "grad_norm": 0.7514160510859766, "learning_rate": 4.439740576271164e-06, "loss": 0.2722, "step": 15903 }, { "epoch": 1.0994815070860697, "grad_norm": 0.7772170044325697, "learning_rate": 4.439184345725471e-06, "loss": 0.2462, "step": 15904 }, { "epoch": 1.0995506394745937, "grad_norm": 0.6594479376816943, "learning_rate": 4.438628122208599e-06, "loss": 0.3435, "step": 15905 }, { "epoch": 1.0996197718631178, "grad_norm": 0.726880962339796, "learning_rate": 4.438071905727523e-06, "loss": 0.2548, "step": 15906 }, { "epoch": 1.099688904251642, "grad_norm": 0.6405377150884048, "learning_rate": 4.437515696289212e-06, "loss": 0.3038, "step": 15907 }, { "epoch": 1.099758036640166, "grad_norm": 0.6885408775038397, "learning_rate": 4.4369594939006395e-06, "loss": 0.3011, "step": 15908 }, { "epoch": 1.09982716902869, "grad_norm": 0.6654774257530011, "learning_rate": 4.436403298568774e-06, "loss": 0.3471, "step": 15909 }, { "epoch": 1.099896301417214, "grad_norm": 0.6234802222541029, "learning_rate": 4.4358471103005865e-06, "loss": 0.2253, "step": 15910 }, { "epoch": 1.099965433805738, "grad_norm": 0.715017658860112, "learning_rate": 4.4352909291030485e-06, "loss": 0.2386, "step": 15911 }, { "epoch": 1.100034566194262, "grad_norm": 0.7094582268042237, "learning_rate": 4.434734754983133e-06, "loss": 0.3186, "step": 15912 }, { "epoch": 1.1001036985827861, "grad_norm": 0.6734677097890597, "learning_rate": 4.434178587947806e-06, "loss": 0.309, "step": 15913 }, { "epoch": 1.1001728309713101, "grad_norm": 0.7394440528283731, "learning_rate": 4.433622428004043e-06, "loss": 0.2209, "step": 15914 }, { "epoch": 1.1002419633598342, "grad_norm": 0.6518900601994544, "learning_rate": 4.433066275158811e-06, "loss": 0.2408, "step": 15915 }, { "epoch": 1.1003110957483582, "grad_norm": 0.6895255297677086, "learning_rate": 4.432510129419082e-06, "loss": 0.3652, "step": 15916 }, { "epoch": 1.1003802281368822, "grad_norm": 0.6524631330123057, "learning_rate": 4.4319539907918245e-06, "loss": 0.2283, "step": 15917 }, { "epoch": 1.1004493605254062, "grad_norm": 0.6172152170366692, "learning_rate": 4.431397859284011e-06, "loss": 0.2485, "step": 15918 }, { "epoch": 1.1005184929139302, "grad_norm": 0.6647144986358016, "learning_rate": 4.430841734902611e-06, "loss": 0.3278, "step": 15919 }, { "epoch": 1.1005876253024542, "grad_norm": 1.1018158942738794, "learning_rate": 4.430285617654593e-06, "loss": 0.3285, "step": 15920 }, { "epoch": 1.1006567576909783, "grad_norm": 0.6748154415993233, "learning_rate": 4.429729507546928e-06, "loss": 0.2187, "step": 15921 }, { "epoch": 1.1007258900795023, "grad_norm": 0.6401829364768216, "learning_rate": 4.4291734045865885e-06, "loss": 0.2655, "step": 15922 }, { "epoch": 1.1007950224680263, "grad_norm": 0.6784710541384311, "learning_rate": 4.428617308780539e-06, "loss": 0.3679, "step": 15923 }, { "epoch": 1.1008641548565503, "grad_norm": 0.68910806426627, "learning_rate": 4.4280612201357535e-06, "loss": 0.2552, "step": 15924 }, { "epoch": 1.1009332872450743, "grad_norm": 0.6271698984387688, "learning_rate": 4.427505138659198e-06, "loss": 0.2233, "step": 15925 }, { "epoch": 1.1010024196335984, "grad_norm": 0.6960664280951466, "learning_rate": 4.426949064357847e-06, "loss": 0.3296, "step": 15926 }, { "epoch": 1.1010715520221224, "grad_norm": 0.6783627902097972, "learning_rate": 4.426392997238664e-06, "loss": 0.3379, "step": 15927 }, { "epoch": 1.1011406844106464, "grad_norm": 0.5703309514546313, "learning_rate": 4.425836937308622e-06, "loss": 0.2392, "step": 15928 }, { "epoch": 1.1012098167991704, "grad_norm": 1.2356603904431436, "learning_rate": 4.4252808845746885e-06, "loss": 0.283, "step": 15929 }, { "epoch": 1.1012789491876944, "grad_norm": 0.6499153672215436, "learning_rate": 4.424724839043836e-06, "loss": 0.3398, "step": 15930 }, { "epoch": 1.1013480815762184, "grad_norm": 0.6468921100135792, "learning_rate": 4.424168800723028e-06, "loss": 0.2626, "step": 15931 }, { "epoch": 1.1014172139647425, "grad_norm": 0.6659860937584308, "learning_rate": 4.423612769619239e-06, "loss": 0.2501, "step": 15932 }, { "epoch": 1.1014863463532665, "grad_norm": 0.63897254195076, "learning_rate": 4.423056745739433e-06, "loss": 0.2414, "step": 15933 }, { "epoch": 1.1015554787417905, "grad_norm": 0.6811985543083264, "learning_rate": 4.422500729090583e-06, "loss": 0.3363, "step": 15934 }, { "epoch": 1.1016246111303145, "grad_norm": 0.5859367458769318, "learning_rate": 4.421944719679654e-06, "loss": 0.2404, "step": 15935 }, { "epoch": 1.1016937435188385, "grad_norm": 0.6555975511746862, "learning_rate": 4.421388717513618e-06, "loss": 0.2299, "step": 15936 }, { "epoch": 1.1017628759073625, "grad_norm": 0.6870453591678314, "learning_rate": 4.420832722599441e-06, "loss": 0.3313, "step": 15937 }, { "epoch": 1.1018320082958866, "grad_norm": 0.6732413971315985, "learning_rate": 4.420276734944093e-06, "loss": 0.2968, "step": 15938 }, { "epoch": 1.1019011406844106, "grad_norm": 0.6073307545533769, "learning_rate": 4.41972075455454e-06, "loss": 0.2251, "step": 15939 }, { "epoch": 1.1019702730729346, "grad_norm": 0.6579149261421966, "learning_rate": 4.419164781437754e-06, "loss": 0.2697, "step": 15940 }, { "epoch": 1.1020394054614586, "grad_norm": 0.7008713800032168, "learning_rate": 4.4186088156006995e-06, "loss": 0.4417, "step": 15941 }, { "epoch": 1.1021085378499826, "grad_norm": 0.6516600486047331, "learning_rate": 4.4180528570503465e-06, "loss": 0.2156, "step": 15942 }, { "epoch": 1.1021776702385067, "grad_norm": 0.655289637799338, "learning_rate": 4.417496905793661e-06, "loss": 0.2035, "step": 15943 }, { "epoch": 1.1022468026270307, "grad_norm": 0.7371937018804182, "learning_rate": 4.416940961837615e-06, "loss": 0.399, "step": 15944 }, { "epoch": 1.1023159350155547, "grad_norm": 0.7291967030863538, "learning_rate": 4.416385025189171e-06, "loss": 0.334, "step": 15945 }, { "epoch": 1.1023850674040787, "grad_norm": 0.6950021034468354, "learning_rate": 4.4158290958553e-06, "loss": 0.2559, "step": 15946 }, { "epoch": 1.1024541997926027, "grad_norm": 0.6199578489136269, "learning_rate": 4.415273173842969e-06, "loss": 0.2416, "step": 15947 }, { "epoch": 1.1025233321811267, "grad_norm": 0.7961496731964439, "learning_rate": 4.414717259159147e-06, "loss": 0.4349, "step": 15948 }, { "epoch": 1.1025924645696508, "grad_norm": 0.6971819115478834, "learning_rate": 4.414161351810796e-06, "loss": 0.2664, "step": 15949 }, { "epoch": 1.102661596958175, "grad_norm": 0.6942715453797736, "learning_rate": 4.41360545180489e-06, "loss": 0.2912, "step": 15950 }, { "epoch": 1.102730729346699, "grad_norm": 0.6286374671121739, "learning_rate": 4.413049559148391e-06, "loss": 0.2559, "step": 15951 }, { "epoch": 1.102799861735223, "grad_norm": 0.6298941380486449, "learning_rate": 4.41249367384827e-06, "loss": 0.3365, "step": 15952 }, { "epoch": 1.102868994123747, "grad_norm": 0.5944521883707129, "learning_rate": 4.41193779591149e-06, "loss": 0.2449, "step": 15953 }, { "epoch": 1.102938126512271, "grad_norm": 0.6771811544108827, "learning_rate": 4.411381925345024e-06, "loss": 0.2733, "step": 15954 }, { "epoch": 1.103007258900795, "grad_norm": 0.7323919194636741, "learning_rate": 4.410826062155832e-06, "loss": 0.4222, "step": 15955 }, { "epoch": 1.1030763912893191, "grad_norm": 0.6929197511840653, "learning_rate": 4.410270206350885e-06, "loss": 0.28, "step": 15956 }, { "epoch": 1.1031455236778431, "grad_norm": 0.6214554546107709, "learning_rate": 4.409714357937148e-06, "loss": 0.2043, "step": 15957 }, { "epoch": 1.1032146560663671, "grad_norm": 1.1792584143637819, "learning_rate": 4.40915851692159e-06, "loss": 0.2969, "step": 15958 }, { "epoch": 1.1032837884548912, "grad_norm": 0.6574759639529605, "learning_rate": 4.408602683311174e-06, "loss": 0.3436, "step": 15959 }, { "epoch": 1.1033529208434152, "grad_norm": 0.6327794166081581, "learning_rate": 4.408046857112868e-06, "loss": 0.2215, "step": 15960 }, { "epoch": 1.1034220532319392, "grad_norm": 0.5672247045180628, "learning_rate": 4.4074910383336374e-06, "loss": 0.2297, "step": 15961 }, { "epoch": 1.1034911856204632, "grad_norm": 0.6681915505498517, "learning_rate": 4.4069352269804514e-06, "loss": 0.3303, "step": 15962 }, { "epoch": 1.1035603180089872, "grad_norm": 0.7265972971235853, "learning_rate": 4.406379423060273e-06, "loss": 0.2965, "step": 15963 }, { "epoch": 1.1036294503975113, "grad_norm": 0.6528424307486103, "learning_rate": 4.405823626580069e-06, "loss": 0.2981, "step": 15964 }, { "epoch": 1.1036985827860353, "grad_norm": 0.7533855539851699, "learning_rate": 4.4052678375468065e-06, "loss": 0.3005, "step": 15965 }, { "epoch": 1.1037677151745593, "grad_norm": 0.6471219758425771, "learning_rate": 4.404712055967449e-06, "loss": 0.3519, "step": 15966 }, { "epoch": 1.1038368475630833, "grad_norm": 0.7755316443937798, "learning_rate": 4.4041562818489636e-06, "loss": 0.3281, "step": 15967 }, { "epoch": 1.1039059799516073, "grad_norm": 0.647424813203527, "learning_rate": 4.403600515198316e-06, "loss": 0.2542, "step": 15968 }, { "epoch": 1.1039751123401313, "grad_norm": 0.6367825123096267, "learning_rate": 4.403044756022473e-06, "loss": 0.3077, "step": 15969 }, { "epoch": 1.1040442447286554, "grad_norm": 0.650641492158966, "learning_rate": 4.402489004328395e-06, "loss": 0.2784, "step": 15970 }, { "epoch": 1.1041133771171794, "grad_norm": 0.6784822276856987, "learning_rate": 4.401933260123055e-06, "loss": 0.2708, "step": 15971 }, { "epoch": 1.1041825095057034, "grad_norm": 0.6893740082137569, "learning_rate": 4.401377523413412e-06, "loss": 0.2381, "step": 15972 }, { "epoch": 1.1042516418942274, "grad_norm": 0.67277895431051, "learning_rate": 4.400821794206435e-06, "loss": 0.3928, "step": 15973 }, { "epoch": 1.1043207742827514, "grad_norm": 0.6396884494947229, "learning_rate": 4.400266072509086e-06, "loss": 0.2779, "step": 15974 }, { "epoch": 1.1043899066712755, "grad_norm": 0.6622433669822564, "learning_rate": 4.399710358328333e-06, "loss": 0.2625, "step": 15975 }, { "epoch": 1.1044590390597995, "grad_norm": 0.7316585372918498, "learning_rate": 4.399154651671138e-06, "loss": 0.3713, "step": 15976 }, { "epoch": 1.1045281714483235, "grad_norm": 0.6877376953802521, "learning_rate": 4.398598952544469e-06, "loss": 0.3032, "step": 15977 }, { "epoch": 1.1045973038368475, "grad_norm": 0.6994646737581091, "learning_rate": 4.398043260955286e-06, "loss": 0.2543, "step": 15978 }, { "epoch": 1.1046664362253715, "grad_norm": 0.6252815123490156, "learning_rate": 4.397487576910559e-06, "loss": 0.22, "step": 15979 }, { "epoch": 1.1047355686138955, "grad_norm": 0.6319553934914531, "learning_rate": 4.396931900417249e-06, "loss": 0.3476, "step": 15980 }, { "epoch": 1.1048047010024196, "grad_norm": 0.8136798977923023, "learning_rate": 4.3963762314823215e-06, "loss": 0.2663, "step": 15981 }, { "epoch": 1.1048738333909436, "grad_norm": 0.7028218575736046, "learning_rate": 4.39582057011274e-06, "loss": 0.2641, "step": 15982 }, { "epoch": 1.1049429657794676, "grad_norm": 0.6977972675809397, "learning_rate": 4.395264916315472e-06, "loss": 0.3689, "step": 15983 }, { "epoch": 1.1050120981679916, "grad_norm": 0.6745553282684251, "learning_rate": 4.394709270097476e-06, "loss": 0.3252, "step": 15984 }, { "epoch": 1.1050812305565156, "grad_norm": 0.6737690165397294, "learning_rate": 4.394153631465721e-06, "loss": 0.317, "step": 15985 }, { "epoch": 1.1051503629450397, "grad_norm": 0.6261395012252268, "learning_rate": 4.393598000427168e-06, "loss": 0.2812, "step": 15986 }, { "epoch": 1.1052194953335637, "grad_norm": 0.7402986909170204, "learning_rate": 4.393042376988784e-06, "loss": 0.3107, "step": 15987 }, { "epoch": 1.105288627722088, "grad_norm": 0.6599051351525882, "learning_rate": 4.3924867611575266e-06, "loss": 0.2794, "step": 15988 }, { "epoch": 1.105357760110612, "grad_norm": 0.7491820234198702, "learning_rate": 4.391931152940367e-06, "loss": 0.2476, "step": 15989 }, { "epoch": 1.105426892499136, "grad_norm": 0.6536239402205208, "learning_rate": 4.391375552344263e-06, "loss": 0.2928, "step": 15990 }, { "epoch": 1.10549602488766, "grad_norm": 2.2136085474412344, "learning_rate": 4.390819959376182e-06, "loss": 0.3117, "step": 15991 }, { "epoch": 1.105565157276184, "grad_norm": 0.7897097571984959, "learning_rate": 4.390264374043085e-06, "loss": 0.2936, "step": 15992 }, { "epoch": 1.105634289664708, "grad_norm": 0.6781659457523591, "learning_rate": 4.389708796351938e-06, "loss": 0.233, "step": 15993 }, { "epoch": 1.105703422053232, "grad_norm": 0.6314302573738035, "learning_rate": 4.389153226309699e-06, "loss": 0.2818, "step": 15994 }, { "epoch": 1.105772554441756, "grad_norm": 0.6295322692183636, "learning_rate": 4.388597663923336e-06, "loss": 0.3183, "step": 15995 }, { "epoch": 1.10584168683028, "grad_norm": 0.5852163550049934, "learning_rate": 4.388042109199808e-06, "loss": 0.2619, "step": 15996 }, { "epoch": 1.105910819218804, "grad_norm": 0.62110023987994, "learning_rate": 4.387486562146084e-06, "loss": 0.2836, "step": 15997 }, { "epoch": 1.105979951607328, "grad_norm": 0.6396346583715756, "learning_rate": 4.386931022769119e-06, "loss": 0.3229, "step": 15998 }, { "epoch": 1.1060490839958521, "grad_norm": 0.7344102201186985, "learning_rate": 4.386375491075882e-06, "loss": 0.2575, "step": 15999 }, { "epoch": 1.1061182163843761, "grad_norm": 0.6644364923084924, "learning_rate": 4.385819967073331e-06, "loss": 0.2684, "step": 16000 }, { "epoch": 1.1061873487729001, "grad_norm": 0.6824006019476269, "learning_rate": 4.385264450768433e-06, "loss": 0.3495, "step": 16001 }, { "epoch": 1.1062564811614242, "grad_norm": 0.7054670916714643, "learning_rate": 4.384708942168146e-06, "loss": 0.3035, "step": 16002 }, { "epoch": 1.1063256135499482, "grad_norm": 0.6441250841759186, "learning_rate": 4.3841534412794365e-06, "loss": 0.2148, "step": 16003 }, { "epoch": 1.1063947459384722, "grad_norm": 0.5974068404066105, "learning_rate": 4.383597948109263e-06, "loss": 0.2391, "step": 16004 }, { "epoch": 1.1064638783269962, "grad_norm": 0.7075863501305607, "learning_rate": 4.383042462664591e-06, "loss": 0.3767, "step": 16005 }, { "epoch": 1.1065330107155202, "grad_norm": 0.6107833218883565, "learning_rate": 4.382486984952377e-06, "loss": 0.2618, "step": 16006 }, { "epoch": 1.1066021431040443, "grad_norm": 0.6107694956175989, "learning_rate": 4.38193151497959e-06, "loss": 0.2459, "step": 16007 }, { "epoch": 1.1066712754925683, "grad_norm": 0.6664258597680628, "learning_rate": 4.381376052753187e-06, "loss": 0.3352, "step": 16008 }, { "epoch": 1.1067404078810923, "grad_norm": 0.6559016144245555, "learning_rate": 4.3808205982801325e-06, "loss": 0.3537, "step": 16009 }, { "epoch": 1.1068095402696163, "grad_norm": 0.6962336439507776, "learning_rate": 4.380265151567385e-06, "loss": 0.2553, "step": 16010 }, { "epoch": 1.1068786726581403, "grad_norm": 0.6895712684309971, "learning_rate": 4.3797097126219105e-06, "loss": 0.248, "step": 16011 }, { "epoch": 1.1069478050466643, "grad_norm": 0.6958875114902796, "learning_rate": 4.379154281450665e-06, "loss": 0.3292, "step": 16012 }, { "epoch": 1.1070169374351884, "grad_norm": 0.7025480889254022, "learning_rate": 4.3785988580606134e-06, "loss": 0.3099, "step": 16013 }, { "epoch": 1.1070860698237124, "grad_norm": 0.630342993493831, "learning_rate": 4.378043442458716e-06, "loss": 0.243, "step": 16014 }, { "epoch": 1.1071552022122364, "grad_norm": 0.688564412307655, "learning_rate": 4.377488034651936e-06, "loss": 0.2965, "step": 16015 }, { "epoch": 1.1072243346007604, "grad_norm": 0.7040706377402449, "learning_rate": 4.3769326346472306e-06, "loss": 0.3663, "step": 16016 }, { "epoch": 1.1072934669892844, "grad_norm": 0.6511093580291627, "learning_rate": 4.376377242451564e-06, "loss": 0.2494, "step": 16017 }, { "epoch": 1.1073625993778085, "grad_norm": 0.9763203183537971, "learning_rate": 4.3758218580718944e-06, "loss": 0.2408, "step": 16018 }, { "epoch": 1.1074317317663325, "grad_norm": 0.6587247809080684, "learning_rate": 4.375266481515186e-06, "loss": 0.3135, "step": 16019 }, { "epoch": 1.1075008641548565, "grad_norm": 0.6518933203404548, "learning_rate": 4.374711112788395e-06, "loss": 0.2873, "step": 16020 }, { "epoch": 1.1075699965433805, "grad_norm": 0.6960304615489408, "learning_rate": 4.374155751898486e-06, "loss": 0.2764, "step": 16021 }, { "epoch": 1.1076391289319045, "grad_norm": 0.6833166098779656, "learning_rate": 4.373600398852416e-06, "loss": 0.2604, "step": 16022 }, { "epoch": 1.1077082613204285, "grad_norm": 0.6582635803906074, "learning_rate": 4.3730450536571495e-06, "loss": 0.3347, "step": 16023 }, { "epoch": 1.1077773937089526, "grad_norm": 0.6737288929519429, "learning_rate": 4.372489716319641e-06, "loss": 0.2477, "step": 16024 }, { "epoch": 1.1078465260974766, "grad_norm": 0.6196793755813809, "learning_rate": 4.371934386846858e-06, "loss": 0.2327, "step": 16025 }, { "epoch": 1.1079156584860006, "grad_norm": 0.6971008662687533, "learning_rate": 4.371379065245754e-06, "loss": 0.2864, "step": 16026 }, { "epoch": 1.1079847908745246, "grad_norm": 0.6715178050469707, "learning_rate": 4.370823751523293e-06, "loss": 0.3587, "step": 16027 }, { "epoch": 1.1080539232630486, "grad_norm": 0.6659026643258975, "learning_rate": 4.370268445686431e-06, "loss": 0.2484, "step": 16028 }, { "epoch": 1.1081230556515727, "grad_norm": 0.6132846875406774, "learning_rate": 4.369713147742134e-06, "loss": 0.2842, "step": 16029 }, { "epoch": 1.1081921880400967, "grad_norm": 0.7216105718438801, "learning_rate": 4.369157857697355e-06, "loss": 0.3273, "step": 16030 }, { "epoch": 1.108261320428621, "grad_norm": 0.6931272274270625, "learning_rate": 4.3686025755590575e-06, "loss": 0.2593, "step": 16031 }, { "epoch": 1.108330452817145, "grad_norm": 1.0258792529628074, "learning_rate": 4.368047301334202e-06, "loss": 0.2412, "step": 16032 }, { "epoch": 1.108399585205669, "grad_norm": 0.6319038400435445, "learning_rate": 4.367492035029743e-06, "loss": 0.2148, "step": 16033 }, { "epoch": 1.108468717594193, "grad_norm": 0.6863453235018427, "learning_rate": 4.366936776652643e-06, "loss": 0.3258, "step": 16034 }, { "epoch": 1.108537849982717, "grad_norm": 0.7632766153411444, "learning_rate": 4.36638152620986e-06, "loss": 0.2519, "step": 16035 }, { "epoch": 1.108606982371241, "grad_norm": 0.6835403613781856, "learning_rate": 4.365826283708356e-06, "loss": 0.26, "step": 16036 }, { "epoch": 1.108676114759765, "grad_norm": 0.7087737402382829, "learning_rate": 4.3652710491550865e-06, "loss": 0.3549, "step": 16037 }, { "epoch": 1.108745247148289, "grad_norm": 0.6593173564653745, "learning_rate": 4.364715822557012e-06, "loss": 0.244, "step": 16038 }, { "epoch": 1.108814379536813, "grad_norm": 0.6521277447750878, "learning_rate": 4.364160603921091e-06, "loss": 0.2716, "step": 16039 }, { "epoch": 1.108883511925337, "grad_norm": 0.6358139121193753, "learning_rate": 4.363605393254282e-06, "loss": 0.2401, "step": 16040 }, { "epoch": 1.108952644313861, "grad_norm": 0.6678076850348522, "learning_rate": 4.363050190563543e-06, "loss": 0.3347, "step": 16041 }, { "epoch": 1.109021776702385, "grad_norm": 0.6236666190381129, "learning_rate": 4.362494995855834e-06, "loss": 0.23, "step": 16042 }, { "epoch": 1.1090909090909091, "grad_norm": 0.8927097718195345, "learning_rate": 4.361939809138112e-06, "loss": 0.3074, "step": 16043 }, { "epoch": 1.1091600414794331, "grad_norm": 0.6918610779009459, "learning_rate": 4.361384630417336e-06, "loss": 0.2984, "step": 16044 }, { "epoch": 1.1092291738679572, "grad_norm": 0.6996730228802142, "learning_rate": 4.360829459700464e-06, "loss": 0.3743, "step": 16045 }, { "epoch": 1.1092983062564812, "grad_norm": 0.7365845658662765, "learning_rate": 4.360274296994456e-06, "loss": 0.2943, "step": 16046 }, { "epoch": 1.1093674386450052, "grad_norm": 0.6869332486019961, "learning_rate": 4.359719142306264e-06, "loss": 0.2907, "step": 16047 }, { "epoch": 1.1094365710335292, "grad_norm": 0.6928512845492044, "learning_rate": 4.359163995642854e-06, "loss": 0.4079, "step": 16048 }, { "epoch": 1.1095057034220532, "grad_norm": 0.6391309104881807, "learning_rate": 4.358608857011176e-06, "loss": 0.2601, "step": 16049 }, { "epoch": 1.1095748358105773, "grad_norm": 0.6877336749606352, "learning_rate": 4.358053726418195e-06, "loss": 0.3107, "step": 16050 }, { "epoch": 1.1096439681991013, "grad_norm": 0.6559780071867768, "learning_rate": 4.357498603870862e-06, "loss": 0.286, "step": 16051 }, { "epoch": 1.1097131005876253, "grad_norm": 0.6454491873065358, "learning_rate": 4.356943489376139e-06, "loss": 0.3452, "step": 16052 }, { "epoch": 1.1097822329761493, "grad_norm": 0.7330727490085289, "learning_rate": 4.356388382940981e-06, "loss": 0.2453, "step": 16053 }, { "epoch": 1.1098513653646733, "grad_norm": 0.6891724621658678, "learning_rate": 4.355833284572348e-06, "loss": 0.275, "step": 16054 }, { "epoch": 1.1099204977531973, "grad_norm": 0.7389934262520504, "learning_rate": 4.355278194277192e-06, "loss": 0.3483, "step": 16055 }, { "epoch": 1.1099896301417214, "grad_norm": 0.7363026922856095, "learning_rate": 4.3547231120624754e-06, "loss": 0.3347, "step": 16056 }, { "epoch": 1.1100587625302454, "grad_norm": 0.6000281226896115, "learning_rate": 4.354168037935152e-06, "loss": 0.2218, "step": 16057 }, { "epoch": 1.1101278949187694, "grad_norm": 0.6589976859905406, "learning_rate": 4.353612971902182e-06, "loss": 0.3001, "step": 16058 }, { "epoch": 1.1101970273072934, "grad_norm": 0.6614699722174604, "learning_rate": 4.353057913970516e-06, "loss": 0.4042, "step": 16059 }, { "epoch": 1.1102661596958174, "grad_norm": 0.6209273402173241, "learning_rate": 4.352502864147118e-06, "loss": 0.2822, "step": 16060 }, { "epoch": 1.1103352920843415, "grad_norm": 0.688950947124255, "learning_rate": 4.351947822438939e-06, "loss": 0.2647, "step": 16061 }, { "epoch": 1.1104044244728655, "grad_norm": 0.7032872313843201, "learning_rate": 4.3513927888529385e-06, "loss": 0.3478, "step": 16062 }, { "epoch": 1.1104735568613895, "grad_norm": 0.7252275227733344, "learning_rate": 4.350837763396071e-06, "loss": 0.3001, "step": 16063 }, { "epoch": 1.1105426892499135, "grad_norm": 0.6964153909653931, "learning_rate": 4.350282746075296e-06, "loss": 0.2531, "step": 16064 }, { "epoch": 1.1106118216384375, "grad_norm": 0.8237580883929428, "learning_rate": 4.349727736897564e-06, "loss": 0.2684, "step": 16065 }, { "epoch": 1.1106809540269615, "grad_norm": 0.7227519043440848, "learning_rate": 4.3491727358698364e-06, "loss": 0.405, "step": 16066 }, { "epoch": 1.1107500864154856, "grad_norm": 0.6745644729904603, "learning_rate": 4.348617742999066e-06, "loss": 0.2699, "step": 16067 }, { "epoch": 1.1108192188040096, "grad_norm": 0.6579492582947003, "learning_rate": 4.348062758292212e-06, "loss": 0.1999, "step": 16068 }, { "epoch": 1.1108883511925338, "grad_norm": 0.7133519834948202, "learning_rate": 4.3475077817562256e-06, "loss": 0.3433, "step": 16069 }, { "epoch": 1.1109574835810578, "grad_norm": 0.6249278753072187, "learning_rate": 4.3469528133980656e-06, "loss": 0.2931, "step": 16070 }, { "epoch": 1.1110266159695819, "grad_norm": 0.686098632181645, "learning_rate": 4.346397853224686e-06, "loss": 0.2852, "step": 16071 }, { "epoch": 1.1110957483581059, "grad_norm": 0.6241900779973021, "learning_rate": 4.345842901243045e-06, "loss": 0.2541, "step": 16072 }, { "epoch": 1.11116488074663, "grad_norm": 0.6262524032259948, "learning_rate": 4.345287957460093e-06, "loss": 0.3238, "step": 16073 }, { "epoch": 1.111234013135154, "grad_norm": 0.6336097947123844, "learning_rate": 4.34473302188279e-06, "loss": 0.2621, "step": 16074 }, { "epoch": 1.111303145523678, "grad_norm": 0.6697123452328155, "learning_rate": 4.344178094518087e-06, "loss": 0.2868, "step": 16075 }, { "epoch": 1.111372277912202, "grad_norm": 0.6405051053163526, "learning_rate": 4.3436231753729445e-06, "loss": 0.2823, "step": 16076 }, { "epoch": 1.111441410300726, "grad_norm": 0.669970854380823, "learning_rate": 4.34306826445431e-06, "loss": 0.3292, "step": 16077 }, { "epoch": 1.11151054268925, "grad_norm": 0.6690543694676858, "learning_rate": 4.342513361769147e-06, "loss": 0.2537, "step": 16078 }, { "epoch": 1.111579675077774, "grad_norm": 0.696372439392461, "learning_rate": 4.341958467324403e-06, "loss": 0.2747, "step": 16079 }, { "epoch": 1.111648807466298, "grad_norm": 0.7020117004122451, "learning_rate": 4.341403581127035e-06, "loss": 0.3443, "step": 16080 }, { "epoch": 1.111717939854822, "grad_norm": 0.8404088012544036, "learning_rate": 4.340848703183999e-06, "loss": 0.3223, "step": 16081 }, { "epoch": 1.111787072243346, "grad_norm": 0.6717087191136742, "learning_rate": 4.340293833502248e-06, "loss": 0.2706, "step": 16082 }, { "epoch": 1.11185620463187, "grad_norm": 0.6759103053157294, "learning_rate": 4.3397389720887365e-06, "loss": 0.285, "step": 16083 }, { "epoch": 1.111925337020394, "grad_norm": 0.6833398115434371, "learning_rate": 4.339184118950418e-06, "loss": 0.3459, "step": 16084 }, { "epoch": 1.111994469408918, "grad_norm": 0.6648727969112169, "learning_rate": 4.338629274094247e-06, "loss": 0.2611, "step": 16085 }, { "epoch": 1.1120636017974421, "grad_norm": 0.616142908205256, "learning_rate": 4.33807443752718e-06, "loss": 0.2289, "step": 16086 }, { "epoch": 1.1121327341859661, "grad_norm": 0.6239002995721428, "learning_rate": 4.337519609256166e-06, "loss": 0.3298, "step": 16087 }, { "epoch": 1.1122018665744902, "grad_norm": 0.6963771463042674, "learning_rate": 4.336964789288164e-06, "loss": 0.3127, "step": 16088 }, { "epoch": 1.1122709989630142, "grad_norm": 0.6465023007916507, "learning_rate": 4.336409977630123e-06, "loss": 0.2564, "step": 16089 }, { "epoch": 1.1123401313515382, "grad_norm": 0.7716434958149603, "learning_rate": 4.3358551742890005e-06, "loss": 0.2944, "step": 16090 }, { "epoch": 1.1124092637400622, "grad_norm": 0.6923296508632711, "learning_rate": 4.335300379271746e-06, "loss": 0.4279, "step": 16091 }, { "epoch": 1.1124783961285862, "grad_norm": 0.6440469246188796, "learning_rate": 4.334745592585316e-06, "loss": 0.2593, "step": 16092 }, { "epoch": 1.1125475285171103, "grad_norm": 0.626509851583485, "learning_rate": 4.3341908142366626e-06, "loss": 0.2254, "step": 16093 }, { "epoch": 1.1126166609056343, "grad_norm": 0.6734533630352413, "learning_rate": 4.333636044232741e-06, "loss": 0.2969, "step": 16094 }, { "epoch": 1.1126857932941583, "grad_norm": 0.6656509583545374, "learning_rate": 4.333081282580499e-06, "loss": 0.3292, "step": 16095 }, { "epoch": 1.1127549256826823, "grad_norm": 0.6910834001988289, "learning_rate": 4.3325265292868964e-06, "loss": 0.238, "step": 16096 }, { "epoch": 1.1128240580712063, "grad_norm": 0.6337804740112596, "learning_rate": 4.331971784358881e-06, "loss": 0.2684, "step": 16097 }, { "epoch": 1.1128931904597303, "grad_norm": 0.7075610772083837, "learning_rate": 4.331417047803408e-06, "loss": 0.3351, "step": 16098 }, { "epoch": 1.1129623228482544, "grad_norm": 0.6809663191820545, "learning_rate": 4.330862319627429e-06, "loss": 0.2347, "step": 16099 }, { "epoch": 1.1130314552367784, "grad_norm": 0.7055654431240861, "learning_rate": 4.330307599837895e-06, "loss": 0.2529, "step": 16100 }, { "epoch": 1.1131005876253024, "grad_norm": 0.6320079030784723, "learning_rate": 4.329752888441761e-06, "loss": 0.2733, "step": 16101 }, { "epoch": 1.1131697200138264, "grad_norm": 0.6604646248605359, "learning_rate": 4.329198185445979e-06, "loss": 0.3263, "step": 16102 }, { "epoch": 1.1132388524023504, "grad_norm": 0.627418069337169, "learning_rate": 4.328643490857502e-06, "loss": 0.256, "step": 16103 }, { "epoch": 1.1133079847908744, "grad_norm": 0.6654546373559599, "learning_rate": 4.3280888046832775e-06, "loss": 0.2751, "step": 16104 }, { "epoch": 1.1133771171793985, "grad_norm": 0.6901482529800802, "learning_rate": 4.327534126930263e-06, "loss": 0.4051, "step": 16105 }, { "epoch": 1.1134462495679225, "grad_norm": 0.6890213825299565, "learning_rate": 4.326979457605406e-06, "loss": 0.3348, "step": 16106 }, { "epoch": 1.1135153819564465, "grad_norm": 0.626562201388975, "learning_rate": 4.326424796715664e-06, "loss": 0.2098, "step": 16107 }, { "epoch": 1.1135845143449705, "grad_norm": 0.7778767127188991, "learning_rate": 4.325870144267981e-06, "loss": 0.3542, "step": 16108 }, { "epoch": 1.1136536467334945, "grad_norm": 0.6477853547061913, "learning_rate": 4.325315500269316e-06, "loss": 0.3377, "step": 16109 }, { "epoch": 1.1137227791220186, "grad_norm": 0.7236282090961472, "learning_rate": 4.3247608647266145e-06, "loss": 0.3003, "step": 16110 }, { "epoch": 1.1137919115105426, "grad_norm": 0.6647573024332822, "learning_rate": 4.324206237646834e-06, "loss": 0.1699, "step": 16111 }, { "epoch": 1.1138610438990668, "grad_norm": 0.6310022769612769, "learning_rate": 4.323651619036919e-06, "loss": 0.32, "step": 16112 }, { "epoch": 1.1139301762875908, "grad_norm": 0.7899805205329102, "learning_rate": 4.323097008903826e-06, "loss": 0.2998, "step": 16113 }, { "epoch": 1.1139993086761149, "grad_norm": 0.6292127802534158, "learning_rate": 4.322542407254502e-06, "loss": 0.1939, "step": 16114 }, { "epoch": 1.1140684410646389, "grad_norm": 0.6975210532023467, "learning_rate": 4.321987814095901e-06, "loss": 0.2789, "step": 16115 }, { "epoch": 1.1141375734531629, "grad_norm": 0.6635509748688888, "learning_rate": 4.3214332294349726e-06, "loss": 0.3933, "step": 16116 }, { "epoch": 1.114206705841687, "grad_norm": 0.6188378095171616, "learning_rate": 4.3208786532786696e-06, "loss": 0.2006, "step": 16117 }, { "epoch": 1.114275838230211, "grad_norm": 0.8939855723115749, "learning_rate": 4.320324085633938e-06, "loss": 0.2227, "step": 16118 }, { "epoch": 1.114344970618735, "grad_norm": 0.7162406027419834, "learning_rate": 4.319769526507733e-06, "loss": 0.3858, "step": 16119 }, { "epoch": 1.114414103007259, "grad_norm": 0.6876513442878476, "learning_rate": 4.319214975907001e-06, "loss": 0.3174, "step": 16120 }, { "epoch": 1.114483235395783, "grad_norm": 0.6778599029063834, "learning_rate": 4.318660433838697e-06, "loss": 0.2672, "step": 16121 }, { "epoch": 1.114552367784307, "grad_norm": 0.7657567278597376, "learning_rate": 4.318105900309767e-06, "loss": 0.2476, "step": 16122 }, { "epoch": 1.114621500172831, "grad_norm": 0.7127903332359633, "learning_rate": 4.317551375327163e-06, "loss": 0.3316, "step": 16123 }, { "epoch": 1.114690632561355, "grad_norm": 0.782012600023942, "learning_rate": 4.3169968588978335e-06, "loss": 0.2875, "step": 16124 }, { "epoch": 1.114759764949879, "grad_norm": 0.6253180477934188, "learning_rate": 4.316442351028732e-06, "loss": 0.2173, "step": 16125 }, { "epoch": 1.114828897338403, "grad_norm": 0.750952963148714, "learning_rate": 4.315887851726803e-06, "loss": 0.3167, "step": 16126 }, { "epoch": 1.114898029726927, "grad_norm": 0.9018424183569891, "learning_rate": 4.315333360999001e-06, "loss": 0.3178, "step": 16127 }, { "epoch": 1.114967162115451, "grad_norm": 0.5915558975100483, "learning_rate": 4.314778878852271e-06, "loss": 0.2278, "step": 16128 }, { "epoch": 1.1150362945039751, "grad_norm": 0.6467151205239284, "learning_rate": 4.3142244052935685e-06, "loss": 0.2267, "step": 16129 }, { "epoch": 1.1151054268924991, "grad_norm": 0.6446610789518095, "learning_rate": 4.313669940329835e-06, "loss": 0.3594, "step": 16130 }, { "epoch": 1.1151745592810232, "grad_norm": 0.7349532288171011, "learning_rate": 4.313115483968028e-06, "loss": 0.3545, "step": 16131 }, { "epoch": 1.1152436916695472, "grad_norm": 0.6302041099118906, "learning_rate": 4.31256103621509e-06, "loss": 0.2608, "step": 16132 }, { "epoch": 1.1153128240580712, "grad_norm": 0.6877252989862263, "learning_rate": 4.312006597077974e-06, "loss": 0.24, "step": 16133 }, { "epoch": 1.1153819564465952, "grad_norm": 0.7164770428734186, "learning_rate": 4.311452166563626e-06, "loss": 0.3635, "step": 16134 }, { "epoch": 1.1154510888351192, "grad_norm": 0.7608063200709692, "learning_rate": 4.310897744678999e-06, "loss": 0.2594, "step": 16135 }, { "epoch": 1.1155202212236432, "grad_norm": 0.6547419678511557, "learning_rate": 4.310343331431037e-06, "loss": 0.2266, "step": 16136 }, { "epoch": 1.1155893536121673, "grad_norm": 0.7285080590675383, "learning_rate": 4.309788926826691e-06, "loss": 0.3973, "step": 16137 }, { "epoch": 1.1156584860006913, "grad_norm": 0.6421050359239181, "learning_rate": 4.309234530872908e-06, "loss": 0.2714, "step": 16138 }, { "epoch": 1.1157276183892153, "grad_norm": 0.8146621696093823, "learning_rate": 4.30868014357664e-06, "loss": 0.2487, "step": 16139 }, { "epoch": 1.1157967507777393, "grad_norm": 0.7048759004134824, "learning_rate": 4.308125764944831e-06, "loss": 0.2619, "step": 16140 }, { "epoch": 1.1158658831662633, "grad_norm": 0.6649061880081196, "learning_rate": 4.307571394984431e-06, "loss": 0.375, "step": 16141 }, { "epoch": 1.1159350155547874, "grad_norm": 0.6965569637531394, "learning_rate": 4.307017033702387e-06, "loss": 0.2694, "step": 16142 }, { "epoch": 1.1160041479433114, "grad_norm": 0.641844748529677, "learning_rate": 4.30646268110565e-06, "loss": 0.237, "step": 16143 }, { "epoch": 1.1160732803318354, "grad_norm": 0.67272214022953, "learning_rate": 4.305908337201163e-06, "loss": 0.3359, "step": 16144 }, { "epoch": 1.1161424127203594, "grad_norm": 0.6616817766665892, "learning_rate": 4.3053540019958785e-06, "loss": 0.316, "step": 16145 }, { "epoch": 1.1162115451088834, "grad_norm": 0.6199742023190993, "learning_rate": 4.30479967549674e-06, "loss": 0.2589, "step": 16146 }, { "epoch": 1.1162806774974074, "grad_norm": 0.6581759203883546, "learning_rate": 4.3042453577107e-06, "loss": 0.2297, "step": 16147 }, { "epoch": 1.1163498098859315, "grad_norm": 0.7327355089781432, "learning_rate": 4.3036910486446985e-06, "loss": 0.3447, "step": 16148 }, { "epoch": 1.1164189422744555, "grad_norm": 0.7207479446104432, "learning_rate": 4.303136748305691e-06, "loss": 0.2671, "step": 16149 }, { "epoch": 1.1164880746629797, "grad_norm": 0.6224696872350403, "learning_rate": 4.302582456700618e-06, "loss": 0.2552, "step": 16150 }, { "epoch": 1.1165572070515037, "grad_norm": 0.9051398556531496, "learning_rate": 4.302028173836431e-06, "loss": 0.2922, "step": 16151 }, { "epoch": 1.1166263394400278, "grad_norm": 0.6271002717124694, "learning_rate": 4.301473899720074e-06, "loss": 0.2828, "step": 16152 }, { "epoch": 1.1166954718285518, "grad_norm": 0.5913915123270683, "learning_rate": 4.300919634358497e-06, "loss": 0.2379, "step": 16153 }, { "epoch": 1.1167646042170758, "grad_norm": 0.8927415772899052, "learning_rate": 4.300365377758643e-06, "loss": 0.2854, "step": 16154 }, { "epoch": 1.1168337366055998, "grad_norm": 0.8073478214704174, "learning_rate": 4.299811129927461e-06, "loss": 0.3657, "step": 16155 }, { "epoch": 1.1169028689941238, "grad_norm": 0.6517671850042287, "learning_rate": 4.2992568908718965e-06, "loss": 0.2394, "step": 16156 }, { "epoch": 1.1169720013826478, "grad_norm": 0.6822920231771057, "learning_rate": 4.298702660598898e-06, "loss": 0.2987, "step": 16157 }, { "epoch": 1.1170411337711719, "grad_norm": 0.6476910882865988, "learning_rate": 4.298148439115408e-06, "loss": 0.2925, "step": 16158 }, { "epoch": 1.1171102661596959, "grad_norm": 0.6725746658727167, "learning_rate": 4.297594226428377e-06, "loss": 0.3858, "step": 16159 }, { "epoch": 1.11717939854822, "grad_norm": 0.6272886018173318, "learning_rate": 4.297040022544746e-06, "loss": 0.2534, "step": 16160 }, { "epoch": 1.117248530936744, "grad_norm": 0.9927564007460971, "learning_rate": 4.296485827471467e-06, "loss": 0.2648, "step": 16161 }, { "epoch": 1.117317663325268, "grad_norm": 0.6333215739494201, "learning_rate": 4.295931641215481e-06, "loss": 0.3332, "step": 16162 }, { "epoch": 1.117386795713792, "grad_norm": 0.700839022469538, "learning_rate": 4.2953774637837355e-06, "loss": 0.263, "step": 16163 }, { "epoch": 1.117455928102316, "grad_norm": 0.6410765955564484, "learning_rate": 4.294823295183177e-06, "loss": 0.2523, "step": 16164 }, { "epoch": 1.11752506049084, "grad_norm": 0.6622898333244719, "learning_rate": 4.2942691354207484e-06, "loss": 0.3128, "step": 16165 }, { "epoch": 1.117594192879364, "grad_norm": 0.6424065223058574, "learning_rate": 4.293714984503398e-06, "loss": 0.3562, "step": 16166 }, { "epoch": 1.117663325267888, "grad_norm": 0.7081679114639717, "learning_rate": 4.293160842438068e-06, "loss": 0.2412, "step": 16167 }, { "epoch": 1.117732457656412, "grad_norm": 0.6829516559601972, "learning_rate": 4.2926067092317074e-06, "loss": 0.2268, "step": 16168 }, { "epoch": 1.117801590044936, "grad_norm": 0.7387410269238425, "learning_rate": 4.292052584891258e-06, "loss": 0.3091, "step": 16169 }, { "epoch": 1.11787072243346, "grad_norm": 0.6121506052542628, "learning_rate": 4.291498469423669e-06, "loss": 0.321, "step": 16170 }, { "epoch": 1.117939854821984, "grad_norm": 0.7720620655243466, "learning_rate": 4.290944362835878e-06, "loss": 0.2467, "step": 16171 }, { "epoch": 1.1180089872105081, "grad_norm": 0.670473775410533, "learning_rate": 4.290390265134837e-06, "loss": 0.2778, "step": 16172 }, { "epoch": 1.1180781195990321, "grad_norm": 0.7036287625440807, "learning_rate": 4.289836176327486e-06, "loss": 0.3667, "step": 16173 }, { "epoch": 1.1181472519875562, "grad_norm": 0.663478842554246, "learning_rate": 4.289282096420773e-06, "loss": 0.2867, "step": 16174 }, { "epoch": 1.1182163843760802, "grad_norm": 0.6299882815502765, "learning_rate": 4.288728025421639e-06, "loss": 0.2457, "step": 16175 }, { "epoch": 1.1182855167646042, "grad_norm": 0.7193111089467026, "learning_rate": 4.2881739633370306e-06, "loss": 0.2524, "step": 16176 }, { "epoch": 1.1183546491531282, "grad_norm": 0.7382040599977784, "learning_rate": 4.287619910173891e-06, "loss": 0.3668, "step": 16177 }, { "epoch": 1.1184237815416522, "grad_norm": 0.6870736991725731, "learning_rate": 4.287065865939165e-06, "loss": 0.3345, "step": 16178 }, { "epoch": 1.1184929139301762, "grad_norm": 0.7616158099158015, "learning_rate": 4.286511830639795e-06, "loss": 0.2616, "step": 16179 }, { "epoch": 1.1185620463187003, "grad_norm": 0.6866069487990968, "learning_rate": 4.285957804282727e-06, "loss": 0.3459, "step": 16180 }, { "epoch": 1.1186311787072243, "grad_norm": 0.6516395677674529, "learning_rate": 4.2854037868749025e-06, "loss": 0.272, "step": 16181 }, { "epoch": 1.1187003110957483, "grad_norm": 0.6258397826781105, "learning_rate": 4.284849778423268e-06, "loss": 0.2537, "step": 16182 }, { "epoch": 1.1187694434842723, "grad_norm": 0.6254234326506813, "learning_rate": 4.284295778934764e-06, "loss": 0.2817, "step": 16183 }, { "epoch": 1.1188385758727963, "grad_norm": 0.6974739656822966, "learning_rate": 4.283741788416336e-06, "loss": 0.342, "step": 16184 }, { "epoch": 1.1189077082613204, "grad_norm": 0.6497996298362801, "learning_rate": 4.283187806874924e-06, "loss": 0.2911, "step": 16185 }, { "epoch": 1.1189768406498444, "grad_norm": 0.6297320397207277, "learning_rate": 4.282633834317476e-06, "loss": 0.2312, "step": 16186 }, { "epoch": 1.1190459730383684, "grad_norm": 0.6389721675232132, "learning_rate": 4.282079870750932e-06, "loss": 0.3372, "step": 16187 }, { "epoch": 1.1191151054268924, "grad_norm": 0.6426239542438748, "learning_rate": 4.2815259161822374e-06, "loss": 0.2689, "step": 16188 }, { "epoch": 1.1191842378154164, "grad_norm": 1.1122440404667215, "learning_rate": 4.2809719706183314e-06, "loss": 0.2082, "step": 16189 }, { "epoch": 1.1192533702039404, "grad_norm": 0.6745296343904849, "learning_rate": 4.280418034066159e-06, "loss": 0.3095, "step": 16190 }, { "epoch": 1.1193225025924645, "grad_norm": 0.7144303797243471, "learning_rate": 4.279864106532661e-06, "loss": 0.3566, "step": 16191 }, { "epoch": 1.1193916349809885, "grad_norm": 2.465878755301934, "learning_rate": 4.279310188024784e-06, "loss": 0.3099, "step": 16192 }, { "epoch": 1.1194607673695125, "grad_norm": 0.9490202510210168, "learning_rate": 4.278756278549466e-06, "loss": 0.2574, "step": 16193 }, { "epoch": 1.1195298997580367, "grad_norm": 0.7435091061701674, "learning_rate": 4.278202378113651e-06, "loss": 0.3464, "step": 16194 }, { "epoch": 1.1195990321465608, "grad_norm": 0.6470225626800054, "learning_rate": 4.27764848672428e-06, "loss": 0.3228, "step": 16195 }, { "epoch": 1.1196681645350848, "grad_norm": 0.6524118837747006, "learning_rate": 4.2770946043883e-06, "loss": 0.2642, "step": 16196 }, { "epoch": 1.1197372969236088, "grad_norm": 0.6832341501362837, "learning_rate": 4.276540731112645e-06, "loss": 0.2457, "step": 16197 }, { "epoch": 1.1198064293121328, "grad_norm": 0.7218901525891194, "learning_rate": 4.275986866904263e-06, "loss": 0.3268, "step": 16198 }, { "epoch": 1.1198755617006568, "grad_norm": 0.7881091910869996, "learning_rate": 4.275433011770093e-06, "loss": 0.232, "step": 16199 }, { "epoch": 1.1199446940891808, "grad_norm": 0.6219674202591101, "learning_rate": 4.274879165717078e-06, "loss": 0.2214, "step": 16200 }, { "epoch": 1.1200138264777049, "grad_norm": 0.6552815207277801, "learning_rate": 4.274325328752155e-06, "loss": 0.3115, "step": 16201 }, { "epoch": 1.1200829588662289, "grad_norm": 0.7455815315611255, "learning_rate": 4.273771500882273e-06, "loss": 0.3928, "step": 16202 }, { "epoch": 1.120152091254753, "grad_norm": 0.7025478850724551, "learning_rate": 4.273217682114368e-06, "loss": 0.2617, "step": 16203 }, { "epoch": 1.120221223643277, "grad_norm": 0.664886535196787, "learning_rate": 4.272663872455383e-06, "loss": 0.2383, "step": 16204 }, { "epoch": 1.120290356031801, "grad_norm": 0.6403444993683488, "learning_rate": 4.272110071912256e-06, "loss": 0.3539, "step": 16205 }, { "epoch": 1.120359488420325, "grad_norm": 0.7388788864605081, "learning_rate": 4.271556280491933e-06, "loss": 0.2823, "step": 16206 }, { "epoch": 1.120428620808849, "grad_norm": 0.7096480295440363, "learning_rate": 4.2710024982013505e-06, "loss": 0.2117, "step": 16207 }, { "epoch": 1.120497753197373, "grad_norm": 0.7484228666050634, "learning_rate": 4.270448725047451e-06, "loss": 0.3322, "step": 16208 }, { "epoch": 1.120566885585897, "grad_norm": 0.6522775291421881, "learning_rate": 4.269894961037175e-06, "loss": 0.3188, "step": 16209 }, { "epoch": 1.120636017974421, "grad_norm": 0.6505888201493093, "learning_rate": 4.269341206177463e-06, "loss": 0.2724, "step": 16210 }, { "epoch": 1.120705150362945, "grad_norm": 0.6705376437195433, "learning_rate": 4.268787460475255e-06, "loss": 0.2344, "step": 16211 }, { "epoch": 1.120774282751469, "grad_norm": 0.7039852646408167, "learning_rate": 4.268233723937491e-06, "loss": 0.3212, "step": 16212 }, { "epoch": 1.120843415139993, "grad_norm": 0.6943999202854414, "learning_rate": 4.267679996571111e-06, "loss": 0.3365, "step": 16213 }, { "epoch": 1.120912547528517, "grad_norm": 0.6196448033794333, "learning_rate": 4.267126278383057e-06, "loss": 0.2103, "step": 16214 }, { "epoch": 1.1209816799170411, "grad_norm": 0.6727744795481975, "learning_rate": 4.266572569380265e-06, "loss": 0.2788, "step": 16215 }, { "epoch": 1.1210508123055651, "grad_norm": 0.8559720101083477, "learning_rate": 4.266018869569678e-06, "loss": 0.3631, "step": 16216 }, { "epoch": 1.1211199446940892, "grad_norm": 0.7969421860623161, "learning_rate": 4.265465178958234e-06, "loss": 0.2388, "step": 16217 }, { "epoch": 1.1211890770826132, "grad_norm": 0.5713358774705865, "learning_rate": 4.2649114975528755e-06, "loss": 0.2221, "step": 16218 }, { "epoch": 1.1212582094711372, "grad_norm": 0.6427446009504869, "learning_rate": 4.264357825360535e-06, "loss": 0.3181, "step": 16219 }, { "epoch": 1.1213273418596612, "grad_norm": 0.7095789729953156, "learning_rate": 4.263804162388161e-06, "loss": 0.2953, "step": 16220 }, { "epoch": 1.1213964742481852, "grad_norm": 0.748074110898956, "learning_rate": 4.263250508642685e-06, "loss": 0.2278, "step": 16221 }, { "epoch": 1.1214656066367092, "grad_norm": 0.7275448841826031, "learning_rate": 4.262696864131052e-06, "loss": 0.2831, "step": 16222 }, { "epoch": 1.1215347390252333, "grad_norm": 0.7126160428428197, "learning_rate": 4.262143228860195e-06, "loss": 0.3888, "step": 16223 }, { "epoch": 1.1216038714137573, "grad_norm": 0.6959684012115025, "learning_rate": 4.2615896028370585e-06, "loss": 0.26, "step": 16224 }, { "epoch": 1.1216730038022813, "grad_norm": 0.7539425412278061, "learning_rate": 4.261035986068576e-06, "loss": 0.2418, "step": 16225 }, { "epoch": 1.1217421361908053, "grad_norm": 0.7897940141840954, "learning_rate": 4.260482378561691e-06, "loss": 0.3678, "step": 16226 }, { "epoch": 1.1218112685793293, "grad_norm": 0.6784079842424213, "learning_rate": 4.2599287803233365e-06, "loss": 0.2801, "step": 16227 }, { "epoch": 1.1218804009678534, "grad_norm": 0.6791986620340891, "learning_rate": 4.259375191360458e-06, "loss": 0.2649, "step": 16228 }, { "epoch": 1.1219495333563774, "grad_norm": 0.660741654409771, "learning_rate": 4.258821611679987e-06, "loss": 0.2413, "step": 16229 }, { "epoch": 1.1220186657449014, "grad_norm": 0.6700184030890644, "learning_rate": 4.258268041288865e-06, "loss": 0.3087, "step": 16230 }, { "epoch": 1.1220877981334254, "grad_norm": 0.6183175671019003, "learning_rate": 4.2577144801940315e-06, "loss": 0.2421, "step": 16231 }, { "epoch": 1.1221569305219496, "grad_norm": 0.6274566999493608, "learning_rate": 4.2571609284024205e-06, "loss": 0.2365, "step": 16232 }, { "epoch": 1.1222260629104737, "grad_norm": 0.656034612290569, "learning_rate": 4.256607385920971e-06, "loss": 0.2222, "step": 16233 }, { "epoch": 1.1222951952989977, "grad_norm": 0.6377706995505793, "learning_rate": 4.256053852756621e-06, "loss": 0.3342, "step": 16234 }, { "epoch": 1.1223643276875217, "grad_norm": 0.6665294332046147, "learning_rate": 4.2555003289163114e-06, "loss": 0.2671, "step": 16235 }, { "epoch": 1.1224334600760457, "grad_norm": 0.7559657108783593, "learning_rate": 4.254946814406973e-06, "loss": 0.2498, "step": 16236 }, { "epoch": 1.1225025924645697, "grad_norm": 0.6822323672974515, "learning_rate": 4.2543933092355485e-06, "loss": 0.3421, "step": 16237 }, { "epoch": 1.1225717248530938, "grad_norm": 0.6339312896238497, "learning_rate": 4.253839813408972e-06, "loss": 0.2813, "step": 16238 }, { "epoch": 1.1226408572416178, "grad_norm": 0.6432405283064871, "learning_rate": 4.253286326934183e-06, "loss": 0.2525, "step": 16239 }, { "epoch": 1.1227099896301418, "grad_norm": 0.6912326351055536, "learning_rate": 4.252732849818116e-06, "loss": 0.267, "step": 16240 }, { "epoch": 1.1227791220186658, "grad_norm": 0.9932917067613019, "learning_rate": 4.2521793820677115e-06, "loss": 0.342, "step": 16241 }, { "epoch": 1.1228482544071898, "grad_norm": 0.901760540167211, "learning_rate": 4.251625923689902e-06, "loss": 0.2911, "step": 16242 }, { "epoch": 1.1229173867957138, "grad_norm": 0.6354542460101571, "learning_rate": 4.251072474691627e-06, "loss": 0.2269, "step": 16243 }, { "epoch": 1.1229865191842379, "grad_norm": 0.9795023394283748, "learning_rate": 4.250519035079821e-06, "loss": 0.3167, "step": 16244 }, { "epoch": 1.1230556515727619, "grad_norm": 0.7854146194743269, "learning_rate": 4.249965604861423e-06, "loss": 0.3379, "step": 16245 }, { "epoch": 1.123124783961286, "grad_norm": 0.6215235716345087, "learning_rate": 4.249412184043366e-06, "loss": 0.2605, "step": 16246 }, { "epoch": 1.12319391634981, "grad_norm": 0.6680480453965179, "learning_rate": 4.248858772632589e-06, "loss": 0.1918, "step": 16247 }, { "epoch": 1.123263048738334, "grad_norm": 0.6642749192925242, "learning_rate": 4.248305370636025e-06, "loss": 0.3008, "step": 16248 }, { "epoch": 1.123332181126858, "grad_norm": 0.7665054327936845, "learning_rate": 4.247751978060615e-06, "loss": 0.2672, "step": 16249 }, { "epoch": 1.123401313515382, "grad_norm": 0.9338253043431661, "learning_rate": 4.247198594913288e-06, "loss": 0.2414, "step": 16250 }, { "epoch": 1.123470445903906, "grad_norm": 0.6098936575908422, "learning_rate": 4.246645221200985e-06, "loss": 0.2756, "step": 16251 }, { "epoch": 1.12353957829243, "grad_norm": 0.6720998342017588, "learning_rate": 4.2460918569306385e-06, "loss": 0.3368, "step": 16252 }, { "epoch": 1.123608710680954, "grad_norm": 0.8105781205088928, "learning_rate": 4.245538502109188e-06, "loss": 0.245, "step": 16253 }, { "epoch": 1.123677843069478, "grad_norm": 0.7516348635710679, "learning_rate": 4.244985156743563e-06, "loss": 0.2518, "step": 16254 }, { "epoch": 1.123746975458002, "grad_norm": 0.7076222598680555, "learning_rate": 4.244431820840702e-06, "loss": 0.3383, "step": 16255 }, { "epoch": 1.123816107846526, "grad_norm": 0.6835311453616115, "learning_rate": 4.24387849440754e-06, "loss": 0.2118, "step": 16256 }, { "epoch": 1.12388524023505, "grad_norm": 0.6588185504343547, "learning_rate": 4.243325177451013e-06, "loss": 0.2488, "step": 16257 }, { "epoch": 1.1239543726235741, "grad_norm": 0.6612561459886487, "learning_rate": 4.242771869978053e-06, "loss": 0.2861, "step": 16258 }, { "epoch": 1.1240235050120981, "grad_norm": 0.6406809818308719, "learning_rate": 4.242218571995598e-06, "loss": 0.3498, "step": 16259 }, { "epoch": 1.1240926374006222, "grad_norm": 0.6697683318589057, "learning_rate": 4.241665283510579e-06, "loss": 0.2521, "step": 16260 }, { "epoch": 1.1241617697891462, "grad_norm": 0.6412719827028889, "learning_rate": 4.241112004529933e-06, "loss": 0.2593, "step": 16261 }, { "epoch": 1.1242309021776702, "grad_norm": 0.7103832972216111, "learning_rate": 4.240558735060594e-06, "loss": 0.4099, "step": 16262 }, { "epoch": 1.1243000345661942, "grad_norm": 0.7948425982641876, "learning_rate": 4.240005475109497e-06, "loss": 0.2846, "step": 16263 }, { "epoch": 1.1243691669547182, "grad_norm": 0.6832253117233972, "learning_rate": 4.2394522246835735e-06, "loss": 0.2452, "step": 16264 }, { "epoch": 1.1244382993432422, "grad_norm": 0.6864877713522058, "learning_rate": 4.2388989837897595e-06, "loss": 0.2406, "step": 16265 }, { "epoch": 1.1245074317317663, "grad_norm": 0.6256338648521897, "learning_rate": 4.238345752434988e-06, "loss": 0.3463, "step": 16266 }, { "epoch": 1.1245765641202903, "grad_norm": 1.5118970999602133, "learning_rate": 4.2377925306261955e-06, "loss": 0.2898, "step": 16267 }, { "epoch": 1.1246456965088143, "grad_norm": 1.6999242416433942, "learning_rate": 4.2372393183703105e-06, "loss": 0.2005, "step": 16268 }, { "epoch": 1.1247148288973383, "grad_norm": 0.7100276427381376, "learning_rate": 4.236686115674271e-06, "loss": 0.4057, "step": 16269 }, { "epoch": 1.1247839612858623, "grad_norm": 0.6898592080415207, "learning_rate": 4.236132922545008e-06, "loss": 0.3105, "step": 16270 }, { "epoch": 1.1248530936743864, "grad_norm": 0.6446778646274325, "learning_rate": 4.235579738989458e-06, "loss": 0.2171, "step": 16271 }, { "epoch": 1.1249222260629104, "grad_norm": 0.6685597684521305, "learning_rate": 4.235026565014548e-06, "loss": 0.2574, "step": 16272 }, { "epoch": 1.1249913584514344, "grad_norm": 1.0133825675269603, "learning_rate": 4.2344734006272185e-06, "loss": 0.3497, "step": 16273 }, { "epoch": 1.1250604908399584, "grad_norm": 0.625239311594134, "learning_rate": 4.233920245834395e-06, "loss": 0.2471, "step": 16274 }, { "epoch": 1.1251296232284824, "grad_norm": 0.6501711234568082, "learning_rate": 4.233367100643017e-06, "loss": 0.2265, "step": 16275 }, { "epoch": 1.1251987556170067, "grad_norm": 0.7384494721282613, "learning_rate": 4.232813965060013e-06, "loss": 0.3122, "step": 16276 }, { "epoch": 1.1252678880055307, "grad_norm": 0.6483342126104631, "learning_rate": 4.232260839092319e-06, "loss": 0.3679, "step": 16277 }, { "epoch": 1.1253370203940547, "grad_norm": 0.6722404903482962, "learning_rate": 4.231707722746862e-06, "loss": 0.2142, "step": 16278 }, { "epoch": 1.1254061527825787, "grad_norm": 0.7744098530053241, "learning_rate": 4.23115461603058e-06, "loss": 0.3023, "step": 16279 }, { "epoch": 1.1254752851711027, "grad_norm": 0.6874619577898404, "learning_rate": 4.230601518950401e-06, "loss": 0.3297, "step": 16280 }, { "epoch": 1.1255444175596268, "grad_norm": 0.6976936124314826, "learning_rate": 4.2300484315132615e-06, "loss": 0.2719, "step": 16281 }, { "epoch": 1.1256135499481508, "grad_norm": 0.6320916310610085, "learning_rate": 4.229495353726089e-06, "loss": 0.2141, "step": 16282 }, { "epoch": 1.1256826823366748, "grad_norm": 0.7411133609520817, "learning_rate": 4.228942285595818e-06, "loss": 0.2858, "step": 16283 }, { "epoch": 1.1257518147251988, "grad_norm": 0.8859119795276741, "learning_rate": 4.2283892271293775e-06, "loss": 0.3701, "step": 16284 }, { "epoch": 1.1258209471137228, "grad_norm": 0.8044326570606058, "learning_rate": 4.227836178333704e-06, "loss": 0.2444, "step": 16285 }, { "epoch": 1.1258900795022468, "grad_norm": 0.6855114062754468, "learning_rate": 4.227283139215724e-06, "loss": 0.2659, "step": 16286 }, { "epoch": 1.1259592118907709, "grad_norm": 0.6994711688457307, "learning_rate": 4.226730109782372e-06, "loss": 0.3698, "step": 16287 }, { "epoch": 1.1260283442792949, "grad_norm": 0.6699433701367924, "learning_rate": 4.226177090040576e-06, "loss": 0.3041, "step": 16288 }, { "epoch": 1.126097476667819, "grad_norm": 0.7019380501565093, "learning_rate": 4.2256240799972725e-06, "loss": 0.2358, "step": 16289 }, { "epoch": 1.126166609056343, "grad_norm": 0.684952269015765, "learning_rate": 4.225071079659385e-06, "loss": 0.3677, "step": 16290 }, { "epoch": 1.126235741444867, "grad_norm": 0.6817638838299835, "learning_rate": 4.224518089033852e-06, "loss": 0.3366, "step": 16291 }, { "epoch": 1.126304873833391, "grad_norm": 0.6785777504105764, "learning_rate": 4.2239651081276e-06, "loss": 0.2881, "step": 16292 }, { "epoch": 1.126374006221915, "grad_norm": 0.6665788299233402, "learning_rate": 4.2234121369475606e-06, "loss": 0.2705, "step": 16293 }, { "epoch": 1.126443138610439, "grad_norm": 0.677685240105017, "learning_rate": 4.222859175500663e-06, "loss": 0.2806, "step": 16294 }, { "epoch": 1.126512270998963, "grad_norm": 0.6774209618094332, "learning_rate": 4.222306223793841e-06, "loss": 0.3379, "step": 16295 }, { "epoch": 1.126581403387487, "grad_norm": 1.5852763120679032, "learning_rate": 4.221753281834021e-06, "loss": 0.2639, "step": 16296 }, { "epoch": 1.126650535776011, "grad_norm": 0.8014378067503262, "learning_rate": 4.221200349628135e-06, "loss": 0.2541, "step": 16297 }, { "epoch": 1.126719668164535, "grad_norm": 0.8691622999442855, "learning_rate": 4.220647427183114e-06, "loss": 0.3252, "step": 16298 }, { "epoch": 1.126788800553059, "grad_norm": 0.6595746934266933, "learning_rate": 4.220094514505885e-06, "loss": 0.2566, "step": 16299 }, { "epoch": 1.126857932941583, "grad_norm": 0.6267614680967297, "learning_rate": 4.219541611603381e-06, "loss": 0.2266, "step": 16300 }, { "epoch": 1.1269270653301071, "grad_norm": 0.6604143322550146, "learning_rate": 4.218988718482528e-06, "loss": 0.256, "step": 16301 }, { "epoch": 1.1269961977186311, "grad_norm": 0.8217275782289705, "learning_rate": 4.21843583515026e-06, "loss": 0.3266, "step": 16302 }, { "epoch": 1.1270653301071551, "grad_norm": 0.6502504581344285, "learning_rate": 4.217882961613503e-06, "loss": 0.2318, "step": 16303 }, { "epoch": 1.1271344624956792, "grad_norm": 0.6504293965118448, "learning_rate": 4.217330097879187e-06, "loss": 0.2622, "step": 16304 }, { "epoch": 1.1272035948842032, "grad_norm": 0.6584033413445778, "learning_rate": 4.216777243954241e-06, "loss": 0.326, "step": 16305 }, { "epoch": 1.1272727272727272, "grad_norm": 0.7218441158810499, "learning_rate": 4.216224399845597e-06, "loss": 0.2972, "step": 16306 }, { "epoch": 1.1273418596612512, "grad_norm": 0.6633619315849525, "learning_rate": 4.215671565560178e-06, "loss": 0.265, "step": 16307 }, { "epoch": 1.1274109920497752, "grad_norm": 0.6319995302988314, "learning_rate": 4.215118741104918e-06, "loss": 0.3087, "step": 16308 }, { "epoch": 1.1274801244382993, "grad_norm": 0.6969218291677763, "learning_rate": 4.214565926486743e-06, "loss": 0.3136, "step": 16309 }, { "epoch": 1.1275492568268233, "grad_norm": 0.6028965220060587, "learning_rate": 4.214013121712584e-06, "loss": 0.2342, "step": 16310 }, { "epoch": 1.1276183892153473, "grad_norm": 0.665225942627936, "learning_rate": 4.213460326789366e-06, "loss": 0.2316, "step": 16311 }, { "epoch": 1.1276875216038715, "grad_norm": 0.6547848306974539, "learning_rate": 4.212907541724022e-06, "loss": 0.341, "step": 16312 }, { "epoch": 1.1277566539923956, "grad_norm": 0.7752676081105365, "learning_rate": 4.212354766523474e-06, "loss": 0.3435, "step": 16313 }, { "epoch": 1.1278257863809196, "grad_norm": 0.6576630761207116, "learning_rate": 4.211802001194655e-06, "loss": 0.256, "step": 16314 }, { "epoch": 1.1278949187694436, "grad_norm": 0.6545995578045789, "learning_rate": 4.2112492457444905e-06, "loss": 0.2607, "step": 16315 }, { "epoch": 1.1279640511579676, "grad_norm": 0.6182628793828853, "learning_rate": 4.210696500179911e-06, "loss": 0.344, "step": 16316 }, { "epoch": 1.1280331835464916, "grad_norm": 0.6740735689800366, "learning_rate": 4.210143764507841e-06, "loss": 0.2668, "step": 16317 }, { "epoch": 1.1281023159350156, "grad_norm": 0.6474293702710844, "learning_rate": 4.2095910387352096e-06, "loss": 0.2456, "step": 16318 }, { "epoch": 1.1281714483235397, "grad_norm": 0.9654641807193518, "learning_rate": 4.209038322868944e-06, "loss": 0.3137, "step": 16319 }, { "epoch": 1.1282405807120637, "grad_norm": 0.6648797134005895, "learning_rate": 4.208485616915972e-06, "loss": 0.3324, "step": 16320 }, { "epoch": 1.1283097131005877, "grad_norm": 0.6017547754840664, "learning_rate": 4.207932920883219e-06, "loss": 0.2808, "step": 16321 }, { "epoch": 1.1283788454891117, "grad_norm": 0.6994619015475879, "learning_rate": 4.207380234777614e-06, "loss": 0.3032, "step": 16322 }, { "epoch": 1.1284479778776357, "grad_norm": 0.5912484590259485, "learning_rate": 4.206827558606083e-06, "loss": 0.327, "step": 16323 }, { "epoch": 1.1285171102661598, "grad_norm": 0.5993421854266381, "learning_rate": 4.206274892375554e-06, "loss": 0.2266, "step": 16324 }, { "epoch": 1.1285862426546838, "grad_norm": 0.7016444872289312, "learning_rate": 4.205722236092952e-06, "loss": 0.2178, "step": 16325 }, { "epoch": 1.1286553750432078, "grad_norm": 0.7128866879425815, "learning_rate": 4.205169589765204e-06, "loss": 0.326, "step": 16326 }, { "epoch": 1.1287245074317318, "grad_norm": 0.7405691808645625, "learning_rate": 4.204616953399237e-06, "loss": 0.37, "step": 16327 }, { "epoch": 1.1287936398202558, "grad_norm": 0.6324583401237719, "learning_rate": 4.204064327001977e-06, "loss": 0.2664, "step": 16328 }, { "epoch": 1.1288627722087798, "grad_norm": 0.7548512312027631, "learning_rate": 4.20351171058035e-06, "loss": 0.2254, "step": 16329 }, { "epoch": 1.1289319045973039, "grad_norm": 0.6856060531896392, "learning_rate": 4.2029591041412835e-06, "loss": 0.3587, "step": 16330 }, { "epoch": 1.1290010369858279, "grad_norm": 0.6392225191719539, "learning_rate": 4.202406507691701e-06, "loss": 0.276, "step": 16331 }, { "epoch": 1.129070169374352, "grad_norm": 0.6594858998364477, "learning_rate": 4.20185392123853e-06, "loss": 0.2922, "step": 16332 }, { "epoch": 1.129139301762876, "grad_norm": 0.6904497460382875, "learning_rate": 4.201301344788695e-06, "loss": 0.2733, "step": 16333 }, { "epoch": 1.1292084341514, "grad_norm": 0.7233425897472192, "learning_rate": 4.200748778349125e-06, "loss": 0.3625, "step": 16334 }, { "epoch": 1.129277566539924, "grad_norm": 0.6837520790224841, "learning_rate": 4.20019622192674e-06, "loss": 0.2613, "step": 16335 }, { "epoch": 1.129346698928448, "grad_norm": 0.6554003227900065, "learning_rate": 4.1996436755284695e-06, "loss": 0.2249, "step": 16336 }, { "epoch": 1.129415831316972, "grad_norm": 0.7586778334612472, "learning_rate": 4.199091139161236e-06, "loss": 0.3486, "step": 16337 }, { "epoch": 1.129484963705496, "grad_norm": 0.6765870825776921, "learning_rate": 4.198538612831967e-06, "loss": 0.2968, "step": 16338 }, { "epoch": 1.12955409609402, "grad_norm": 0.6512721717704527, "learning_rate": 4.197986096547585e-06, "loss": 0.2301, "step": 16339 }, { "epoch": 1.129623228482544, "grad_norm": 0.6566253167153837, "learning_rate": 4.197433590315017e-06, "loss": 0.2862, "step": 16340 }, { "epoch": 1.129692360871068, "grad_norm": 0.6512043045723283, "learning_rate": 4.196881094141185e-06, "loss": 0.3226, "step": 16341 }, { "epoch": 1.129761493259592, "grad_norm": 0.7059498539204407, "learning_rate": 4.196328608033019e-06, "loss": 0.2295, "step": 16342 }, { "epoch": 1.129830625648116, "grad_norm": 0.6314643691400293, "learning_rate": 4.195776131997436e-06, "loss": 0.2721, "step": 16343 }, { "epoch": 1.1298997580366401, "grad_norm": 0.6295088740525848, "learning_rate": 4.195223666041365e-06, "loss": 0.2857, "step": 16344 }, { "epoch": 1.1299688904251641, "grad_norm": 0.6600436970542131, "learning_rate": 4.194671210171729e-06, "loss": 0.2991, "step": 16345 }, { "epoch": 1.1300380228136881, "grad_norm": 0.6323040389449517, "learning_rate": 4.194118764395453e-06, "loss": 0.2686, "step": 16346 }, { "epoch": 1.1301071552022122, "grad_norm": 0.6814022856168256, "learning_rate": 4.1935663287194586e-06, "loss": 0.2614, "step": 16347 }, { "epoch": 1.1301762875907362, "grad_norm": 0.6976216149988046, "learning_rate": 4.193013903150674e-06, "loss": 0.3304, "step": 16348 }, { "epoch": 1.1302454199792602, "grad_norm": 0.6785079597932477, "learning_rate": 4.192461487696017e-06, "loss": 0.2708, "step": 16349 }, { "epoch": 1.1303145523677842, "grad_norm": 0.6043144339361219, "learning_rate": 4.191909082362417e-06, "loss": 0.2232, "step": 16350 }, { "epoch": 1.1303836847563082, "grad_norm": 0.6392118732497211, "learning_rate": 4.191356687156791e-06, "loss": 0.2863, "step": 16351 }, { "epoch": 1.1304528171448323, "grad_norm": 0.6861159574250547, "learning_rate": 4.19080430208607e-06, "loss": 0.305, "step": 16352 }, { "epoch": 1.1305219495333563, "grad_norm": 0.6258766348062259, "learning_rate": 4.19025192715717e-06, "loss": 0.2214, "step": 16353 }, { "epoch": 1.1305910819218803, "grad_norm": 0.6909559112477546, "learning_rate": 4.1896995623770175e-06, "loss": 0.2796, "step": 16354 }, { "epoch": 1.1306602143104043, "grad_norm": 0.6920339577179965, "learning_rate": 4.189147207752535e-06, "loss": 0.3867, "step": 16355 }, { "epoch": 1.1307293466989283, "grad_norm": 0.6619229519392463, "learning_rate": 4.188594863290646e-06, "loss": 0.3115, "step": 16356 }, { "epoch": 1.1307984790874526, "grad_norm": 0.7019106887785327, "learning_rate": 4.188042528998272e-06, "loss": 0.2015, "step": 16357 }, { "epoch": 1.1308676114759766, "grad_norm": 0.6648990475860738, "learning_rate": 4.187490204882335e-06, "loss": 0.2816, "step": 16358 }, { "epoch": 1.1309367438645006, "grad_norm": 0.6868468019197279, "learning_rate": 4.186937890949759e-06, "loss": 0.3865, "step": 16359 }, { "epoch": 1.1310058762530246, "grad_norm": 0.6483038507008395, "learning_rate": 4.186385587207467e-06, "loss": 0.2378, "step": 16360 }, { "epoch": 1.1310750086415486, "grad_norm": 0.6363078494296299, "learning_rate": 4.1858332936623766e-06, "loss": 0.2247, "step": 16361 }, { "epoch": 1.1311441410300727, "grad_norm": 0.6615291909784059, "learning_rate": 4.185281010321416e-06, "loss": 0.2986, "step": 16362 }, { "epoch": 1.1312132734185967, "grad_norm": 0.6570120171238514, "learning_rate": 4.1847287371915034e-06, "loss": 0.3282, "step": 16363 }, { "epoch": 1.1312824058071207, "grad_norm": 0.6841141264602131, "learning_rate": 4.184176474279559e-06, "loss": 0.2718, "step": 16364 }, { "epoch": 1.1313515381956447, "grad_norm": 0.6842710684273601, "learning_rate": 4.1836242215925094e-06, "loss": 0.2467, "step": 16365 }, { "epoch": 1.1314206705841687, "grad_norm": 0.6622057901826849, "learning_rate": 4.183071979137271e-06, "loss": 0.2515, "step": 16366 }, { "epoch": 1.1314898029726927, "grad_norm": 0.7507244449841026, "learning_rate": 4.1825197469207685e-06, "loss": 0.2975, "step": 16367 }, { "epoch": 1.1315589353612168, "grad_norm": 0.7130727755309456, "learning_rate": 4.1819675249499215e-06, "loss": 0.2743, "step": 16368 }, { "epoch": 1.1316280677497408, "grad_norm": 0.8981889264829209, "learning_rate": 4.181415313231653e-06, "loss": 0.344, "step": 16369 }, { "epoch": 1.1316972001382648, "grad_norm": 0.6357149160753035, "learning_rate": 4.180863111772882e-06, "loss": 0.3236, "step": 16370 }, { "epoch": 1.1317663325267888, "grad_norm": 0.8593358442097148, "learning_rate": 4.1803109205805295e-06, "loss": 0.2705, "step": 16371 }, { "epoch": 1.1318354649153128, "grad_norm": 0.7252672180882608, "learning_rate": 4.179758739661517e-06, "loss": 0.2749, "step": 16372 }, { "epoch": 1.1319045973038369, "grad_norm": 0.6627494852276705, "learning_rate": 4.179206569022766e-06, "loss": 0.3446, "step": 16373 }, { "epoch": 1.1319737296923609, "grad_norm": 0.6792222787784845, "learning_rate": 4.178654408671194e-06, "loss": 0.3198, "step": 16374 }, { "epoch": 1.132042862080885, "grad_norm": 0.663157807754755, "learning_rate": 4.1781022586137256e-06, "loss": 0.2528, "step": 16375 }, { "epoch": 1.132111994469409, "grad_norm": 0.9596901819000372, "learning_rate": 4.177550118857276e-06, "loss": 0.3174, "step": 16376 }, { "epoch": 1.132181126857933, "grad_norm": 0.6289789443348145, "learning_rate": 4.176997989408771e-06, "loss": 0.3697, "step": 16377 }, { "epoch": 1.132250259246457, "grad_norm": 0.665135888311544, "learning_rate": 4.176445870275125e-06, "loss": 0.2765, "step": 16378 }, { "epoch": 1.132319391634981, "grad_norm": 0.5993838503292465, "learning_rate": 4.175893761463261e-06, "loss": 0.2469, "step": 16379 }, { "epoch": 1.132388524023505, "grad_norm": 0.6880887483472051, "learning_rate": 4.175341662980098e-06, "loss": 0.3849, "step": 16380 }, { "epoch": 1.132457656412029, "grad_norm": 0.6520212417527118, "learning_rate": 4.174789574832555e-06, "loss": 0.2873, "step": 16381 }, { "epoch": 1.132526788800553, "grad_norm": 0.6237361041801137, "learning_rate": 4.174237497027552e-06, "loss": 0.2094, "step": 16382 }, { "epoch": 1.132595921189077, "grad_norm": 0.7353794841394697, "learning_rate": 4.173685429572011e-06, "loss": 0.3021, "step": 16383 }, { "epoch": 1.132665053577601, "grad_norm": 1.0218567687402464, "learning_rate": 4.173133372472845e-06, "loss": 0.4285, "step": 16384 }, { "epoch": 1.132734185966125, "grad_norm": 0.7790458211728652, "learning_rate": 4.172581325736979e-06, "loss": 0.2751, "step": 16385 }, { "epoch": 1.132803318354649, "grad_norm": 0.5954595807552104, "learning_rate": 4.172029289371327e-06, "loss": 0.2593, "step": 16386 }, { "epoch": 1.132872450743173, "grad_norm": 0.7321406154768103, "learning_rate": 4.171477263382814e-06, "loss": 0.3754, "step": 16387 }, { "epoch": 1.1329415831316971, "grad_norm": 0.7029922683582106, "learning_rate": 4.170925247778352e-06, "loss": 0.3438, "step": 16388 }, { "epoch": 1.1330107155202211, "grad_norm": 0.6755986038576131, "learning_rate": 4.170373242564864e-06, "loss": 0.2459, "step": 16389 }, { "epoch": 1.1330798479087452, "grad_norm": 0.6645098852412648, "learning_rate": 4.169821247749265e-06, "loss": 0.2701, "step": 16390 }, { "epoch": 1.1331489802972692, "grad_norm": 0.8083381811587272, "learning_rate": 4.169269263338478e-06, "loss": 0.3289, "step": 16391 }, { "epoch": 1.1332181126857932, "grad_norm": 0.6335927088736967, "learning_rate": 4.168717289339415e-06, "loss": 0.2598, "step": 16392 }, { "epoch": 1.1332872450743174, "grad_norm": 0.6745411582340817, "learning_rate": 4.168165325759e-06, "loss": 0.2955, "step": 16393 }, { "epoch": 1.1333563774628415, "grad_norm": 0.6531902757836486, "learning_rate": 4.167613372604147e-06, "loss": 0.3509, "step": 16394 }, { "epoch": 1.1334255098513655, "grad_norm": 0.6741065000590155, "learning_rate": 4.167061429881776e-06, "loss": 0.291, "step": 16395 }, { "epoch": 1.1334946422398895, "grad_norm": 0.6386782546790016, "learning_rate": 4.166509497598801e-06, "loss": 0.2378, "step": 16396 }, { "epoch": 1.1335637746284135, "grad_norm": 0.8203692626840321, "learning_rate": 4.165957575762144e-06, "loss": 0.3158, "step": 16397 }, { "epoch": 1.1336329070169375, "grad_norm": 0.6895301354530732, "learning_rate": 4.16540566437872e-06, "loss": 0.3515, "step": 16398 }, { "epoch": 1.1337020394054615, "grad_norm": 0.627436844662926, "learning_rate": 4.164853763455446e-06, "loss": 0.2587, "step": 16399 }, { "epoch": 1.1337711717939856, "grad_norm": 0.9483427417252742, "learning_rate": 4.16430187299924e-06, "loss": 0.1939, "step": 16400 }, { "epoch": 1.1338403041825096, "grad_norm": 0.6778879472528458, "learning_rate": 4.16374999301702e-06, "loss": 0.3042, "step": 16401 }, { "epoch": 1.1339094365710336, "grad_norm": 0.6502677080906744, "learning_rate": 4.163198123515699e-06, "loss": 0.3485, "step": 16402 }, { "epoch": 1.1339785689595576, "grad_norm": 0.6144949433303178, "learning_rate": 4.162646264502198e-06, "loss": 0.2575, "step": 16403 }, { "epoch": 1.1340477013480816, "grad_norm": 0.6543287293217342, "learning_rate": 4.16209441598343e-06, "loss": 0.2469, "step": 16404 }, { "epoch": 1.1341168337366057, "grad_norm": 0.6992991669623524, "learning_rate": 4.161542577966317e-06, "loss": 0.3709, "step": 16405 }, { "epoch": 1.1341859661251297, "grad_norm": 1.1822373776745916, "learning_rate": 4.160990750457767e-06, "loss": 0.2827, "step": 16406 }, { "epoch": 1.1342550985136537, "grad_norm": 0.689129530946059, "learning_rate": 4.160438933464703e-06, "loss": 0.2906, "step": 16407 }, { "epoch": 1.1343242309021777, "grad_norm": 0.6421801834173229, "learning_rate": 4.159887126994038e-06, "loss": 0.2744, "step": 16408 }, { "epoch": 1.1343933632907017, "grad_norm": 0.7004999366728077, "learning_rate": 4.15933533105269e-06, "loss": 0.371, "step": 16409 }, { "epoch": 1.1344624956792257, "grad_norm": 0.6190975371666858, "learning_rate": 4.158783545647572e-06, "loss": 0.2331, "step": 16410 }, { "epoch": 1.1345316280677498, "grad_norm": 0.6253147350860916, "learning_rate": 4.158231770785601e-06, "loss": 0.2546, "step": 16411 }, { "epoch": 1.1346007604562738, "grad_norm": 0.6824675564030294, "learning_rate": 4.157680006473692e-06, "loss": 0.3299, "step": 16412 }, { "epoch": 1.1346698928447978, "grad_norm": 0.7014347831633663, "learning_rate": 4.157128252718763e-06, "loss": 0.2441, "step": 16413 }, { "epoch": 1.1347390252333218, "grad_norm": 0.7326884765009637, "learning_rate": 4.156576509527726e-06, "loss": 0.1809, "step": 16414 }, { "epoch": 1.1348081576218458, "grad_norm": 0.7122851879441596, "learning_rate": 4.156024776907497e-06, "loss": 0.3324, "step": 16415 }, { "epoch": 1.1348772900103699, "grad_norm": 0.6920393718924942, "learning_rate": 4.155473054864991e-06, "loss": 0.3673, "step": 16416 }, { "epoch": 1.1349464223988939, "grad_norm": 0.6871062945646126, "learning_rate": 4.1549213434071245e-06, "loss": 0.293, "step": 16417 }, { "epoch": 1.135015554787418, "grad_norm": 0.6606455278342626, "learning_rate": 4.154369642540809e-06, "loss": 0.2722, "step": 16418 }, { "epoch": 1.135084687175942, "grad_norm": 0.648848507125257, "learning_rate": 4.153817952272963e-06, "loss": 0.2755, "step": 16419 }, { "epoch": 1.135153819564466, "grad_norm": 0.7486972669419101, "learning_rate": 4.1532662726104975e-06, "loss": 0.3473, "step": 16420 }, { "epoch": 1.13522295195299, "grad_norm": 0.7053144613331506, "learning_rate": 4.152714603560329e-06, "loss": 0.246, "step": 16421 }, { "epoch": 1.135292084341514, "grad_norm": 0.6450574437580724, "learning_rate": 4.1521629451293694e-06, "loss": 0.2869, "step": 16422 }, { "epoch": 1.135361216730038, "grad_norm": 0.6752557099034655, "learning_rate": 4.151611297324537e-06, "loss": 0.3863, "step": 16423 }, { "epoch": 1.135430349118562, "grad_norm": 0.7164065516653195, "learning_rate": 4.15105966015274e-06, "loss": 0.2521, "step": 16424 }, { "epoch": 1.135499481507086, "grad_norm": 0.6013681335996525, "learning_rate": 4.150508033620897e-06, "loss": 0.2737, "step": 16425 }, { "epoch": 1.13556861389561, "grad_norm": 0.6950497541552453, "learning_rate": 4.149956417735919e-06, "loss": 0.3515, "step": 16426 }, { "epoch": 1.135637746284134, "grad_norm": 0.6447968153004067, "learning_rate": 4.149404812504721e-06, "loss": 0.3201, "step": 16427 }, { "epoch": 1.135706878672658, "grad_norm": 0.6082196856143712, "learning_rate": 4.148853217934215e-06, "loss": 0.2457, "step": 16428 }, { "epoch": 1.135776011061182, "grad_norm": 0.630702846154142, "learning_rate": 4.148301634031315e-06, "loss": 0.2963, "step": 16429 }, { "epoch": 1.135845143449706, "grad_norm": 0.7091130767633055, "learning_rate": 4.1477500608029364e-06, "loss": 0.394, "step": 16430 }, { "epoch": 1.1359142758382301, "grad_norm": 0.6174588114853105, "learning_rate": 4.147198498255987e-06, "loss": 0.2472, "step": 16431 }, { "epoch": 1.1359834082267541, "grad_norm": 0.6820383052507621, "learning_rate": 4.146646946397384e-06, "loss": 0.1819, "step": 16432 }, { "epoch": 1.1360525406152782, "grad_norm": 0.6707696927622069, "learning_rate": 4.146095405234037e-06, "loss": 0.2816, "step": 16433 }, { "epoch": 1.1361216730038022, "grad_norm": 0.7568263745441176, "learning_rate": 4.145543874772863e-06, "loss": 0.3182, "step": 16434 }, { "epoch": 1.1361908053923262, "grad_norm": 0.6424526410728129, "learning_rate": 4.144992355020768e-06, "loss": 0.2768, "step": 16435 }, { "epoch": 1.1362599377808502, "grad_norm": 0.6608382267200588, "learning_rate": 4.144440845984671e-06, "loss": 0.287, "step": 16436 }, { "epoch": 1.1363290701693742, "grad_norm": 0.7116517602082041, "learning_rate": 4.14388934767148e-06, "loss": 0.3128, "step": 16437 }, { "epoch": 1.1363982025578983, "grad_norm": 0.7187810589897435, "learning_rate": 4.143337860088108e-06, "loss": 0.3276, "step": 16438 }, { "epoch": 1.1364673349464225, "grad_norm": 0.7021270379544227, "learning_rate": 4.142786383241467e-06, "loss": 0.2653, "step": 16439 }, { "epoch": 1.1365364673349465, "grad_norm": 0.7272439607091442, "learning_rate": 4.142234917138471e-06, "loss": 0.2584, "step": 16440 }, { "epoch": 1.1366055997234705, "grad_norm": 0.6687466450042092, "learning_rate": 4.141683461786027e-06, "loss": 0.3746, "step": 16441 }, { "epoch": 1.1366747321119945, "grad_norm": 0.6743055175638312, "learning_rate": 4.1411320171910504e-06, "loss": 0.2857, "step": 16442 }, { "epoch": 1.1367438645005186, "grad_norm": 0.6216616503335057, "learning_rate": 4.14058058336045e-06, "loss": 0.2334, "step": 16443 }, { "epoch": 1.1368129968890426, "grad_norm": 0.6662571365476994, "learning_rate": 4.1400291603011415e-06, "loss": 0.3257, "step": 16444 }, { "epoch": 1.1368821292775666, "grad_norm": 0.7620509390811047, "learning_rate": 4.139477748020029e-06, "loss": 0.3016, "step": 16445 }, { "epoch": 1.1369512616660906, "grad_norm": 0.6409668132099658, "learning_rate": 4.13892634652403e-06, "loss": 0.2583, "step": 16446 }, { "epoch": 1.1370203940546146, "grad_norm": 0.7230991275847987, "learning_rate": 4.13837495582005e-06, "loss": 0.2454, "step": 16447 }, { "epoch": 1.1370895264431387, "grad_norm": 0.6317776992777824, "learning_rate": 4.137823575915005e-06, "loss": 0.3313, "step": 16448 }, { "epoch": 1.1371586588316627, "grad_norm": 0.627072346431927, "learning_rate": 4.1372722068158e-06, "loss": 0.2673, "step": 16449 }, { "epoch": 1.1372277912201867, "grad_norm": 0.6366559324757983, "learning_rate": 4.136720848529351e-06, "loss": 0.2316, "step": 16450 }, { "epoch": 1.1372969236087107, "grad_norm": 0.6571059447734852, "learning_rate": 4.136169501062563e-06, "loss": 0.2956, "step": 16451 }, { "epoch": 1.1373660559972347, "grad_norm": 0.6975774848410797, "learning_rate": 4.13561816442235e-06, "loss": 0.347, "step": 16452 }, { "epoch": 1.1374351883857587, "grad_norm": 0.6178023141686558, "learning_rate": 4.13506683861562e-06, "loss": 0.2405, "step": 16453 }, { "epoch": 1.1375043207742828, "grad_norm": 0.7130201670382601, "learning_rate": 4.134515523649286e-06, "loss": 0.3119, "step": 16454 }, { "epoch": 1.1375734531628068, "grad_norm": 0.6921057520519476, "learning_rate": 4.133964219530253e-06, "loss": 0.3973, "step": 16455 }, { "epoch": 1.1376425855513308, "grad_norm": 0.6546011519504097, "learning_rate": 4.1334129262654345e-06, "loss": 0.2371, "step": 16456 }, { "epoch": 1.1377117179398548, "grad_norm": 0.7015239504334876, "learning_rate": 4.132861643861737e-06, "loss": 0.2692, "step": 16457 }, { "epoch": 1.1377808503283788, "grad_norm": 0.7688132999451599, "learning_rate": 4.132310372326074e-06, "loss": 0.2914, "step": 16458 }, { "epoch": 1.1378499827169029, "grad_norm": 0.8043279401376987, "learning_rate": 4.131759111665349e-06, "loss": 0.4131, "step": 16459 }, { "epoch": 1.1379191151054269, "grad_norm": 0.6908783376502907, "learning_rate": 4.131207861886475e-06, "loss": 0.2687, "step": 16460 }, { "epoch": 1.1379882474939509, "grad_norm": 0.6793251975759313, "learning_rate": 4.1306566229963594e-06, "loss": 0.3004, "step": 16461 }, { "epoch": 1.138057379882475, "grad_norm": 0.6912002568741613, "learning_rate": 4.130105395001914e-06, "loss": 0.3399, "step": 16462 }, { "epoch": 1.138126512270999, "grad_norm": 0.805572118097339, "learning_rate": 4.129554177910043e-06, "loss": 0.27, "step": 16463 }, { "epoch": 1.138195644659523, "grad_norm": 0.6533855053758412, "learning_rate": 4.129002971727658e-06, "loss": 0.2912, "step": 16464 }, { "epoch": 1.138264777048047, "grad_norm": 0.7915378507401456, "learning_rate": 4.128451776461665e-06, "loss": 0.3411, "step": 16465 }, { "epoch": 1.138333909436571, "grad_norm": 0.6441705931344942, "learning_rate": 4.127900592118976e-06, "loss": 0.3526, "step": 16466 }, { "epoch": 1.138403041825095, "grad_norm": 0.6506908943714237, "learning_rate": 4.127349418706495e-06, "loss": 0.2929, "step": 16467 }, { "epoch": 1.138472174213619, "grad_norm": 0.5822731326259121, "learning_rate": 4.126798256231132e-06, "loss": 0.2073, "step": 16468 }, { "epoch": 1.138541306602143, "grad_norm": 0.7019559234046582, "learning_rate": 4.126247104699793e-06, "loss": 0.2934, "step": 16469 }, { "epoch": 1.138610438990667, "grad_norm": 0.6916114168956018, "learning_rate": 4.12569596411939e-06, "loss": 0.3171, "step": 16470 }, { "epoch": 1.138679571379191, "grad_norm": 0.7259824335739359, "learning_rate": 4.125144834496826e-06, "loss": 0.2451, "step": 16471 }, { "epoch": 1.138748703767715, "grad_norm": 0.6363046562585547, "learning_rate": 4.124593715839012e-06, "loss": 0.2536, "step": 16472 }, { "epoch": 1.138817836156239, "grad_norm": 0.6950613698783565, "learning_rate": 4.1240426081528515e-06, "loss": 0.385, "step": 16473 }, { "epoch": 1.1388869685447633, "grad_norm": 0.7188965280225111, "learning_rate": 4.123491511445255e-06, "loss": 0.2717, "step": 16474 }, { "epoch": 1.1389561009332874, "grad_norm": 0.6067542476008065, "learning_rate": 4.122940425723128e-06, "loss": 0.2326, "step": 16475 }, { "epoch": 1.1390252333218114, "grad_norm": 0.7024282156837311, "learning_rate": 4.1223893509933785e-06, "loss": 0.2928, "step": 16476 }, { "epoch": 1.1390943657103354, "grad_norm": 0.7019329696046265, "learning_rate": 4.12183828726291e-06, "loss": 0.3194, "step": 16477 }, { "epoch": 1.1391634980988594, "grad_norm": 0.7320143555581153, "learning_rate": 4.121287234538633e-06, "loss": 0.313, "step": 16478 }, { "epoch": 1.1392326304873834, "grad_norm": 0.7044166355757472, "learning_rate": 4.1207361928274514e-06, "loss": 0.2305, "step": 16479 }, { "epoch": 1.1393017628759075, "grad_norm": 0.6933409125507283, "learning_rate": 4.120185162136275e-06, "loss": 0.35, "step": 16480 }, { "epoch": 1.1393708952644315, "grad_norm": 0.651742954917116, "learning_rate": 4.119634142472004e-06, "loss": 0.2824, "step": 16481 }, { "epoch": 1.1394400276529555, "grad_norm": 0.6291478410804042, "learning_rate": 4.119083133841551e-06, "loss": 0.2399, "step": 16482 }, { "epoch": 1.1395091600414795, "grad_norm": 0.7093072721244421, "learning_rate": 4.118532136251816e-06, "loss": 0.313, "step": 16483 }, { "epoch": 1.1395782924300035, "grad_norm": 0.6380453893055243, "learning_rate": 4.117981149709711e-06, "loss": 0.2979, "step": 16484 }, { "epoch": 1.1396474248185275, "grad_norm": 0.6753607746853044, "learning_rate": 4.117430174222135e-06, "loss": 0.2628, "step": 16485 }, { "epoch": 1.1397165572070516, "grad_norm": 0.6087063065107177, "learning_rate": 4.116879209795999e-06, "loss": 0.2277, "step": 16486 }, { "epoch": 1.1397856895955756, "grad_norm": 0.6298309025534695, "learning_rate": 4.116328256438204e-06, "loss": 0.3294, "step": 16487 }, { "epoch": 1.1398548219840996, "grad_norm": 0.9898410278783933, "learning_rate": 4.115777314155659e-06, "loss": 0.2893, "step": 16488 }, { "epoch": 1.1399239543726236, "grad_norm": 0.7533499802960646, "learning_rate": 4.115226382955266e-06, "loss": 0.2568, "step": 16489 }, { "epoch": 1.1399930867611476, "grad_norm": 0.7186460340411903, "learning_rate": 4.1146754628439335e-06, "loss": 0.246, "step": 16490 }, { "epoch": 1.1400622191496717, "grad_norm": 0.7446570624903711, "learning_rate": 4.114124553828562e-06, "loss": 0.3468, "step": 16491 }, { "epoch": 1.1401313515381957, "grad_norm": 0.6049608795455159, "learning_rate": 4.11357365591606e-06, "loss": 0.275, "step": 16492 }, { "epoch": 1.1402004839267197, "grad_norm": 0.6521492402363137, "learning_rate": 4.113022769113328e-06, "loss": 0.2278, "step": 16493 }, { "epoch": 1.1402696163152437, "grad_norm": 0.6437614172642153, "learning_rate": 4.112471893427276e-06, "loss": 0.3124, "step": 16494 }, { "epoch": 1.1403387487037677, "grad_norm": 0.6602797657670896, "learning_rate": 4.111921028864803e-06, "loss": 0.2783, "step": 16495 }, { "epoch": 1.1404078810922917, "grad_norm": 0.7026404211037768, "learning_rate": 4.111370175432815e-06, "loss": 0.246, "step": 16496 }, { "epoch": 1.1404770134808158, "grad_norm": 0.7060998748725167, "learning_rate": 4.110819333138217e-06, "loss": 0.2863, "step": 16497 }, { "epoch": 1.1405461458693398, "grad_norm": 0.6940344342407154, "learning_rate": 4.110268501987911e-06, "loss": 0.3537, "step": 16498 }, { "epoch": 1.1406152782578638, "grad_norm": 0.6665004804066955, "learning_rate": 4.109717681988802e-06, "loss": 0.2656, "step": 16499 }, { "epoch": 1.1406844106463878, "grad_norm": 0.8743244984470925, "learning_rate": 4.109166873147792e-06, "loss": 0.2159, "step": 16500 }, { "epoch": 1.1407535430349118, "grad_norm": 0.7048340495712467, "learning_rate": 4.108616075471787e-06, "loss": 0.2754, "step": 16501 }, { "epoch": 1.1408226754234359, "grad_norm": 0.6816683917330606, "learning_rate": 4.108065288967688e-06, "loss": 0.3087, "step": 16502 }, { "epoch": 1.1408918078119599, "grad_norm": 0.7722767614128908, "learning_rate": 4.107514513642399e-06, "loss": 0.2166, "step": 16503 }, { "epoch": 1.1409609402004839, "grad_norm": 0.6345789817435329, "learning_rate": 4.106963749502821e-06, "loss": 0.2568, "step": 16504 }, { "epoch": 1.141030072589008, "grad_norm": 0.6756965655403544, "learning_rate": 4.106412996555862e-06, "loss": 0.3354, "step": 16505 }, { "epoch": 1.141099204977532, "grad_norm": 0.5976189864288801, "learning_rate": 4.105862254808418e-06, "loss": 0.2504, "step": 16506 }, { "epoch": 1.141168337366056, "grad_norm": 0.695120377757965, "learning_rate": 4.105311524267398e-06, "loss": 0.2858, "step": 16507 }, { "epoch": 1.14123746975458, "grad_norm": 0.6450910394306836, "learning_rate": 4.1047608049397e-06, "loss": 0.2728, "step": 16508 }, { "epoch": 1.141306602143104, "grad_norm": 0.6750711965217818, "learning_rate": 4.104210096832228e-06, "loss": 0.3575, "step": 16509 }, { "epoch": 1.141375734531628, "grad_norm": 0.6560619713913315, "learning_rate": 4.103659399951883e-06, "loss": 0.2503, "step": 16510 }, { "epoch": 1.141444866920152, "grad_norm": 0.6955730112845857, "learning_rate": 4.1031087143055695e-06, "loss": 0.2484, "step": 16511 }, { "epoch": 1.141513999308676, "grad_norm": 0.7451518860895375, "learning_rate": 4.102558039900186e-06, "loss": 0.3658, "step": 16512 }, { "epoch": 1.1415831316972, "grad_norm": 0.6475709016205505, "learning_rate": 4.102007376742637e-06, "loss": 0.3019, "step": 16513 }, { "epoch": 1.141652264085724, "grad_norm": 0.6345564328164218, "learning_rate": 4.101456724839822e-06, "loss": 0.2569, "step": 16514 }, { "epoch": 1.141721396474248, "grad_norm": 0.6388146970567133, "learning_rate": 4.100906084198645e-06, "loss": 0.2424, "step": 16515 }, { "epoch": 1.141790528862772, "grad_norm": 0.6680468278586884, "learning_rate": 4.100355454826005e-06, "loss": 0.3393, "step": 16516 }, { "epoch": 1.1418596612512961, "grad_norm": 0.6792832802990402, "learning_rate": 4.0998048367288035e-06, "loss": 0.2473, "step": 16517 }, { "epoch": 1.1419287936398201, "grad_norm": 0.6237494971644136, "learning_rate": 4.099254229913941e-06, "loss": 0.2321, "step": 16518 }, { "epoch": 1.1419979260283442, "grad_norm": 0.6414351301892844, "learning_rate": 4.098703634388322e-06, "loss": 0.3266, "step": 16519 }, { "epoch": 1.1420670584168684, "grad_norm": 0.635259670505287, "learning_rate": 4.098153050158842e-06, "loss": 0.2824, "step": 16520 }, { "epoch": 1.1421361908053924, "grad_norm": 0.6525176696677933, "learning_rate": 4.0976024772324054e-06, "loss": 0.2607, "step": 16521 }, { "epoch": 1.1422053231939164, "grad_norm": 0.6443155485616049, "learning_rate": 4.097051915615909e-06, "loss": 0.2355, "step": 16522 }, { "epoch": 1.1422744555824405, "grad_norm": 0.7819472092258105, "learning_rate": 4.0965013653162595e-06, "loss": 0.3261, "step": 16523 }, { "epoch": 1.1423435879709645, "grad_norm": 0.6658995493886191, "learning_rate": 4.095950826340349e-06, "loss": 0.2547, "step": 16524 }, { "epoch": 1.1424127203594885, "grad_norm": 0.5898818843045981, "learning_rate": 4.095400298695084e-06, "loss": 0.2579, "step": 16525 }, { "epoch": 1.1424818527480125, "grad_norm": 0.7569796844703705, "learning_rate": 4.094849782387361e-06, "loss": 0.2767, "step": 16526 }, { "epoch": 1.1425509851365365, "grad_norm": 0.6987616159389555, "learning_rate": 4.094299277424081e-06, "loss": 0.3656, "step": 16527 }, { "epoch": 1.1426201175250605, "grad_norm": 0.6160463656048333, "learning_rate": 4.093748783812143e-06, "loss": 0.2652, "step": 16528 }, { "epoch": 1.1426892499135846, "grad_norm": 0.675596185654225, "learning_rate": 4.0931983015584485e-06, "loss": 0.2647, "step": 16529 }, { "epoch": 1.1427583823021086, "grad_norm": 0.7358743099669367, "learning_rate": 4.092647830669893e-06, "loss": 0.3631, "step": 16530 }, { "epoch": 1.1428275146906326, "grad_norm": 0.7178999440142132, "learning_rate": 4.092097371153378e-06, "loss": 0.2816, "step": 16531 }, { "epoch": 1.1428966470791566, "grad_norm": 0.6287507174546402, "learning_rate": 4.091546923015802e-06, "loss": 0.2329, "step": 16532 }, { "epoch": 1.1429657794676806, "grad_norm": 0.7456497762743949, "learning_rate": 4.0909964862640655e-06, "loss": 0.3324, "step": 16533 }, { "epoch": 1.1430349118562046, "grad_norm": 0.7055614533401601, "learning_rate": 4.0904460609050645e-06, "loss": 0.351, "step": 16534 }, { "epoch": 1.1431040442447287, "grad_norm": 0.646185093764363, "learning_rate": 4.0898956469457e-06, "loss": 0.2908, "step": 16535 }, { "epoch": 1.1431731766332527, "grad_norm": 0.7034424549543098, "learning_rate": 4.089345244392867e-06, "loss": 0.2864, "step": 16536 }, { "epoch": 1.1432423090217767, "grad_norm": 0.8069240426433961, "learning_rate": 4.088794853253469e-06, "loss": 0.3461, "step": 16537 }, { "epoch": 1.1433114414103007, "grad_norm": 0.6457512637328569, "learning_rate": 4.0882444735344e-06, "loss": 0.2896, "step": 16538 }, { "epoch": 1.1433805737988247, "grad_norm": 0.6198392386225492, "learning_rate": 4.08769410524256e-06, "loss": 0.2321, "step": 16539 }, { "epoch": 1.1434497061873488, "grad_norm": 0.66243969393072, "learning_rate": 4.087143748384846e-06, "loss": 0.2452, "step": 16540 }, { "epoch": 1.1435188385758728, "grad_norm": 0.6816083464364854, "learning_rate": 4.0865934029681565e-06, "loss": 0.3416, "step": 16541 }, { "epoch": 1.1435879709643968, "grad_norm": 0.6427975827380081, "learning_rate": 4.0860430689993865e-06, "loss": 0.2591, "step": 16542 }, { "epoch": 1.1436571033529208, "grad_norm": 0.6092422419189424, "learning_rate": 4.0854927464854395e-06, "loss": 0.2058, "step": 16543 }, { "epoch": 1.1437262357414448, "grad_norm": 0.6644856122323921, "learning_rate": 4.084942435433206e-06, "loss": 0.332, "step": 16544 }, { "epoch": 1.1437953681299688, "grad_norm": 0.6743223319500562, "learning_rate": 4.084392135849588e-06, "loss": 0.3073, "step": 16545 }, { "epoch": 1.1438645005184929, "grad_norm": 0.6299000222720559, "learning_rate": 4.083841847741478e-06, "loss": 0.2198, "step": 16546 }, { "epoch": 1.1439336329070169, "grad_norm": 0.6250982636501686, "learning_rate": 4.083291571115778e-06, "loss": 0.2444, "step": 16547 }, { "epoch": 1.144002765295541, "grad_norm": 0.6439427392896669, "learning_rate": 4.082741305979381e-06, "loss": 0.2969, "step": 16548 }, { "epoch": 1.144071897684065, "grad_norm": 0.6920086285073893, "learning_rate": 4.082191052339185e-06, "loss": 0.3078, "step": 16549 }, { "epoch": 1.144141030072589, "grad_norm": 0.6375260456037188, "learning_rate": 4.081640810202086e-06, "loss": 0.2396, "step": 16550 }, { "epoch": 1.144210162461113, "grad_norm": 0.6558680352784557, "learning_rate": 4.081090579574982e-06, "loss": 0.3184, "step": 16551 }, { "epoch": 1.144279294849637, "grad_norm": 0.6483642595429338, "learning_rate": 4.080540360464765e-06, "loss": 0.3939, "step": 16552 }, { "epoch": 1.144348427238161, "grad_norm": 0.7131268420818901, "learning_rate": 4.079990152878336e-06, "loss": 0.2557, "step": 16553 }, { "epoch": 1.144417559626685, "grad_norm": 0.7285036494565335, "learning_rate": 4.0794399568225875e-06, "loss": 0.2908, "step": 16554 }, { "epoch": 1.144486692015209, "grad_norm": 0.6970622850868207, "learning_rate": 4.0788897723044175e-06, "loss": 0.3686, "step": 16555 }, { "epoch": 1.1445558244037333, "grad_norm": 0.6860451394263201, "learning_rate": 4.078339599330719e-06, "loss": 0.2487, "step": 16556 }, { "epoch": 1.1446249567922573, "grad_norm": 0.6571963116941488, "learning_rate": 4.07778943790839e-06, "loss": 0.253, "step": 16557 }, { "epoch": 1.1446940891807813, "grad_norm": 0.7064839036389449, "learning_rate": 4.077239288044323e-06, "loss": 0.2598, "step": 16558 }, { "epoch": 1.1447632215693053, "grad_norm": 0.742525515516228, "learning_rate": 4.076689149745417e-06, "loss": 0.3138, "step": 16559 }, { "epoch": 1.1448323539578293, "grad_norm": 0.7720553087098034, "learning_rate": 4.076139023018562e-06, "loss": 0.2558, "step": 16560 }, { "epoch": 1.1449014863463534, "grad_norm": 0.6220034662042817, "learning_rate": 4.075588907870659e-06, "loss": 0.2349, "step": 16561 }, { "epoch": 1.1449706187348774, "grad_norm": 0.661639493253408, "learning_rate": 4.075038804308597e-06, "loss": 0.3471, "step": 16562 }, { "epoch": 1.1450397511234014, "grad_norm": 0.7221254901950631, "learning_rate": 4.074488712339273e-06, "loss": 0.3, "step": 16563 }, { "epoch": 1.1451088835119254, "grad_norm": 0.6491291543550886, "learning_rate": 4.073938631969584e-06, "loss": 0.2553, "step": 16564 }, { "epoch": 1.1451780159004494, "grad_norm": 0.6901491562056657, "learning_rate": 4.0733885632064185e-06, "loss": 0.2616, "step": 16565 }, { "epoch": 1.1452471482889734, "grad_norm": 0.6765916470008366, "learning_rate": 4.072838506056675e-06, "loss": 0.3841, "step": 16566 }, { "epoch": 1.1453162806774975, "grad_norm": 0.6094343841059945, "learning_rate": 4.072288460527246e-06, "loss": 0.2039, "step": 16567 }, { "epoch": 1.1453854130660215, "grad_norm": 0.6621694688240359, "learning_rate": 4.071738426625026e-06, "loss": 0.275, "step": 16568 }, { "epoch": 1.1454545454545455, "grad_norm": 0.6526436533146439, "learning_rate": 4.0711884043569074e-06, "loss": 0.3035, "step": 16569 }, { "epoch": 1.1455236778430695, "grad_norm": 0.6903457532041476, "learning_rate": 4.070638393729785e-06, "loss": 0.2814, "step": 16570 }, { "epoch": 1.1455928102315935, "grad_norm": 0.6329364785997356, "learning_rate": 4.070088394750552e-06, "loss": 0.2158, "step": 16571 }, { "epoch": 1.1456619426201176, "grad_norm": 0.6596959625282721, "learning_rate": 4.069538407426102e-06, "loss": 0.2704, "step": 16572 }, { "epoch": 1.1457310750086416, "grad_norm": 0.7540494960337031, "learning_rate": 4.068988431763326e-06, "loss": 0.4468, "step": 16573 }, { "epoch": 1.1458002073971656, "grad_norm": 1.3537688900482594, "learning_rate": 4.06843846776912e-06, "loss": 0.2936, "step": 16574 }, { "epoch": 1.1458693397856896, "grad_norm": 0.6351638278027822, "learning_rate": 4.067888515450375e-06, "loss": 0.2599, "step": 16575 }, { "epoch": 1.1459384721742136, "grad_norm": 0.5950438048851163, "learning_rate": 4.067338574813985e-06, "loss": 0.2895, "step": 16576 }, { "epoch": 1.1460076045627376, "grad_norm": 0.7168360352699513, "learning_rate": 4.066788645866838e-06, "loss": 0.3258, "step": 16577 }, { "epoch": 1.1460767369512617, "grad_norm": 0.9482158121402671, "learning_rate": 4.066238728615834e-06, "loss": 0.2689, "step": 16578 }, { "epoch": 1.1461458693397857, "grad_norm": 0.654622044113772, "learning_rate": 4.065688823067858e-06, "loss": 0.2797, "step": 16579 }, { "epoch": 1.1462150017283097, "grad_norm": 0.6709103364263715, "learning_rate": 4.065138929229808e-06, "loss": 0.3379, "step": 16580 }, { "epoch": 1.1462841341168337, "grad_norm": 0.7476133387773622, "learning_rate": 4.064589047108571e-06, "loss": 0.2683, "step": 16581 }, { "epoch": 1.1463532665053577, "grad_norm": 0.6346568986923199, "learning_rate": 4.064039176711043e-06, "loss": 0.2694, "step": 16582 }, { "epoch": 1.1464223988938818, "grad_norm": 0.6199095870411556, "learning_rate": 4.0634893180441115e-06, "loss": 0.2372, "step": 16583 }, { "epoch": 1.1464915312824058, "grad_norm": 0.5999575079162419, "learning_rate": 4.0629394711146716e-06, "loss": 0.2937, "step": 16584 }, { "epoch": 1.1465606636709298, "grad_norm": 0.5974500975398386, "learning_rate": 4.062389635929612e-06, "loss": 0.2165, "step": 16585 }, { "epoch": 1.1466297960594538, "grad_norm": 0.6274098107606181, "learning_rate": 4.0618398124958275e-06, "loss": 0.1965, "step": 16586 }, { "epoch": 1.1466989284479778, "grad_norm": 0.6363922320212577, "learning_rate": 4.061290000820204e-06, "loss": 0.3352, "step": 16587 }, { "epoch": 1.1467680608365018, "grad_norm": 0.6350523201287503, "learning_rate": 4.060740200909636e-06, "loss": 0.287, "step": 16588 }, { "epoch": 1.1468371932250259, "grad_norm": 0.5938709568576063, "learning_rate": 4.0601904127710135e-06, "loss": 0.2351, "step": 16589 }, { "epoch": 1.1469063256135499, "grad_norm": 0.7180843510511328, "learning_rate": 4.059640636411228e-06, "loss": 0.342, "step": 16590 }, { "epoch": 1.146975458002074, "grad_norm": 0.6506299715389754, "learning_rate": 4.059090871837168e-06, "loss": 0.3841, "step": 16591 }, { "epoch": 1.147044590390598, "grad_norm": 0.675181891808083, "learning_rate": 4.0585411190557246e-06, "loss": 0.2929, "step": 16592 }, { "epoch": 1.147113722779122, "grad_norm": 0.662554290803987, "learning_rate": 4.057991378073788e-06, "loss": 0.2049, "step": 16593 }, { "epoch": 1.147182855167646, "grad_norm": 0.6504575146663307, "learning_rate": 4.05744164889825e-06, "loss": 0.2882, "step": 16594 }, { "epoch": 1.14725198755617, "grad_norm": 0.672491055719213, "learning_rate": 4.056891931535995e-06, "loss": 0.2961, "step": 16595 }, { "epoch": 1.147321119944694, "grad_norm": 0.7883268178604834, "learning_rate": 4.056342225993921e-06, "loss": 0.2327, "step": 16596 }, { "epoch": 1.147390252333218, "grad_norm": 1.0977565713073305, "learning_rate": 4.055792532278911e-06, "loss": 0.3051, "step": 16597 }, { "epoch": 1.147459384721742, "grad_norm": 0.7329088975600512, "learning_rate": 4.055242850397857e-06, "loss": 0.3717, "step": 16598 }, { "epoch": 1.147528517110266, "grad_norm": 0.721631163283515, "learning_rate": 4.0546931803576475e-06, "loss": 0.3046, "step": 16599 }, { "epoch": 1.14759764949879, "grad_norm": 0.7364336578093179, "learning_rate": 4.054143522165174e-06, "loss": 0.2286, "step": 16600 }, { "epoch": 1.1476667818873143, "grad_norm": 0.6932750380046625, "learning_rate": 4.053593875827321e-06, "loss": 0.3343, "step": 16601 }, { "epoch": 1.1477359142758383, "grad_norm": 0.7806382266913389, "learning_rate": 4.053044241350981e-06, "loss": 0.4023, "step": 16602 }, { "epoch": 1.1478050466643623, "grad_norm": 0.6267651553328049, "learning_rate": 4.0524946187430405e-06, "loss": 0.2506, "step": 16603 }, { "epoch": 1.1478741790528864, "grad_norm": 0.6388933368970564, "learning_rate": 4.051945008010391e-06, "loss": 0.2576, "step": 16604 }, { "epoch": 1.1479433114414104, "grad_norm": 0.6596187431459423, "learning_rate": 4.051395409159917e-06, "loss": 0.344, "step": 16605 }, { "epoch": 1.1480124438299344, "grad_norm": 0.7090261120403965, "learning_rate": 4.05084582219851e-06, "loss": 0.3227, "step": 16606 }, { "epoch": 1.1480815762184584, "grad_norm": 0.6222844201921842, "learning_rate": 4.050296247133055e-06, "loss": 0.2433, "step": 16607 }, { "epoch": 1.1481507086069824, "grad_norm": 0.7996084430650519, "learning_rate": 4.049746683970444e-06, "loss": 0.2678, "step": 16608 }, { "epoch": 1.1482198409955064, "grad_norm": 0.6447911301278226, "learning_rate": 4.049197132717561e-06, "loss": 0.3499, "step": 16609 }, { "epoch": 1.1482889733840305, "grad_norm": 0.647267362953496, "learning_rate": 4.048647593381296e-06, "loss": 0.231, "step": 16610 }, { "epoch": 1.1483581057725545, "grad_norm": 0.6919650882299054, "learning_rate": 4.048098065968534e-06, "loss": 0.2672, "step": 16611 }, { "epoch": 1.1484272381610785, "grad_norm": 0.6744025221289099, "learning_rate": 4.047548550486166e-06, "loss": 0.3808, "step": 16612 }, { "epoch": 1.1484963705496025, "grad_norm": 0.6775345775587912, "learning_rate": 4.046999046941076e-06, "loss": 0.2689, "step": 16613 }, { "epoch": 1.1485655029381265, "grad_norm": 0.8156946762757166, "learning_rate": 4.046449555340154e-06, "loss": 0.2718, "step": 16614 }, { "epoch": 1.1486346353266506, "grad_norm": 0.6953369244090369, "learning_rate": 4.045900075690283e-06, "loss": 0.3124, "step": 16615 }, { "epoch": 1.1487037677151746, "grad_norm": 0.6345173086791833, "learning_rate": 4.045350607998353e-06, "loss": 0.3912, "step": 16616 }, { "epoch": 1.1487729001036986, "grad_norm": 0.6620144004931127, "learning_rate": 4.0448011522712485e-06, "loss": 0.2536, "step": 16617 }, { "epoch": 1.1488420324922226, "grad_norm": 0.6046831851866913, "learning_rate": 4.0442517085158584e-06, "loss": 0.2116, "step": 16618 }, { "epoch": 1.1489111648807466, "grad_norm": 0.6475542666366445, "learning_rate": 4.043702276739066e-06, "loss": 0.3222, "step": 16619 }, { "epoch": 1.1489802972692706, "grad_norm": 0.6781164575309929, "learning_rate": 4.04315285694776e-06, "loss": 0.3438, "step": 16620 }, { "epoch": 1.1490494296577947, "grad_norm": 0.7019184561218341, "learning_rate": 4.042603449148824e-06, "loss": 0.2712, "step": 16621 }, { "epoch": 1.1491185620463187, "grad_norm": 0.6804131480891707, "learning_rate": 4.042054053349148e-06, "loss": 0.2409, "step": 16622 }, { "epoch": 1.1491876944348427, "grad_norm": 0.6793207077017109, "learning_rate": 4.041504669555612e-06, "loss": 0.3604, "step": 16623 }, { "epoch": 1.1492568268233667, "grad_norm": 0.6659948340631386, "learning_rate": 4.040955297775106e-06, "loss": 0.3016, "step": 16624 }, { "epoch": 1.1493259592118907, "grad_norm": 1.6552549310095754, "learning_rate": 4.0404059380145135e-06, "loss": 0.2543, "step": 16625 }, { "epoch": 1.1493950916004148, "grad_norm": 0.7274819297215154, "learning_rate": 4.039856590280721e-06, "loss": 0.3481, "step": 16626 }, { "epoch": 1.1494642239889388, "grad_norm": 0.7092142898506058, "learning_rate": 4.039307254580611e-06, "loss": 0.3489, "step": 16627 }, { "epoch": 1.1495333563774628, "grad_norm": 0.7012112717112582, "learning_rate": 4.038757930921071e-06, "loss": 0.2491, "step": 16628 }, { "epoch": 1.1496024887659868, "grad_norm": 0.7132646208708773, "learning_rate": 4.038208619308986e-06, "loss": 0.3217, "step": 16629 }, { "epoch": 1.1496716211545108, "grad_norm": 0.8480185645339982, "learning_rate": 4.037659319751237e-06, "loss": 0.3751, "step": 16630 }, { "epoch": 1.1497407535430348, "grad_norm": 0.7172469473918731, "learning_rate": 4.0371100322547145e-06, "loss": 0.2647, "step": 16631 }, { "epoch": 1.1498098859315589, "grad_norm": 0.6445663140725165, "learning_rate": 4.036560756826296e-06, "loss": 0.2136, "step": 16632 }, { "epoch": 1.1498790183200829, "grad_norm": 0.6749659778361423, "learning_rate": 4.036011493472871e-06, "loss": 0.24, "step": 16633 }, { "epoch": 1.149948150708607, "grad_norm": 0.661527657654194, "learning_rate": 4.03546224220132e-06, "loss": 0.3288, "step": 16634 }, { "epoch": 1.150017283097131, "grad_norm": 0.6824591298122611, "learning_rate": 4.0349130030185305e-06, "loss": 0.2808, "step": 16635 }, { "epoch": 1.150086415485655, "grad_norm": 0.7294960837856003, "learning_rate": 4.0343637759313825e-06, "loss": 0.3003, "step": 16636 }, { "epoch": 1.1501555478741792, "grad_norm": 0.6775457563741684, "learning_rate": 4.033814560946762e-06, "loss": 0.3535, "step": 16637 }, { "epoch": 1.1502246802627032, "grad_norm": 0.7370183643117133, "learning_rate": 4.033265358071552e-06, "loss": 0.3127, "step": 16638 }, { "epoch": 1.1502938126512272, "grad_norm": 0.6435578829505311, "learning_rate": 4.032716167312637e-06, "loss": 0.2461, "step": 16639 }, { "epoch": 1.1503629450397512, "grad_norm": 0.7651113086714024, "learning_rate": 4.032166988676896e-06, "loss": 0.318, "step": 16640 }, { "epoch": 1.1504320774282752, "grad_norm": 1.1058653655752075, "learning_rate": 4.031617822171215e-06, "loss": 0.3851, "step": 16641 }, { "epoch": 1.1505012098167993, "grad_norm": 0.7164235647601318, "learning_rate": 4.031068667802477e-06, "loss": 0.3077, "step": 16642 }, { "epoch": 1.1505703422053233, "grad_norm": 0.7652588229254877, "learning_rate": 4.030519525577566e-06, "loss": 0.1959, "step": 16643 }, { "epoch": 1.1506394745938473, "grad_norm": 0.649535246874636, "learning_rate": 4.029970395503359e-06, "loss": 0.3076, "step": 16644 }, { "epoch": 1.1507086069823713, "grad_norm": 0.7368521002234439, "learning_rate": 4.029421277586745e-06, "loss": 0.3374, "step": 16645 }, { "epoch": 1.1507777393708953, "grad_norm": 0.6592877484022199, "learning_rate": 4.028872171834601e-06, "loss": 0.2559, "step": 16646 }, { "epoch": 1.1508468717594194, "grad_norm": 0.6711675322835662, "learning_rate": 4.028323078253813e-06, "loss": 0.2942, "step": 16647 }, { "epoch": 1.1509160041479434, "grad_norm": 0.6678175955852027, "learning_rate": 4.027773996851259e-06, "loss": 0.3716, "step": 16648 }, { "epoch": 1.1509851365364674, "grad_norm": 0.6456292074773063, "learning_rate": 4.027224927633825e-06, "loss": 0.2612, "step": 16649 }, { "epoch": 1.1510542689249914, "grad_norm": 0.6565679228921633, "learning_rate": 4.02667587060839e-06, "loss": 0.2655, "step": 16650 }, { "epoch": 1.1511234013135154, "grad_norm": 0.6873435293140023, "learning_rate": 4.026126825781836e-06, "loss": 0.3281, "step": 16651 }, { "epoch": 1.1511925337020394, "grad_norm": 0.7689352457072787, "learning_rate": 4.025577793161043e-06, "loss": 0.3997, "step": 16652 }, { "epoch": 1.1512616660905635, "grad_norm": 0.5922455843532373, "learning_rate": 4.0250287727528964e-06, "loss": 0.2835, "step": 16653 }, { "epoch": 1.1513307984790875, "grad_norm": 0.8136685969539522, "learning_rate": 4.024479764564272e-06, "loss": 0.2676, "step": 16654 }, { "epoch": 1.1513999308676115, "grad_norm": 0.6657912991404699, "learning_rate": 4.023930768602054e-06, "loss": 0.3698, "step": 16655 }, { "epoch": 1.1514690632561355, "grad_norm": 0.668160329583784, "learning_rate": 4.023381784873119e-06, "loss": 0.2287, "step": 16656 }, { "epoch": 1.1515381956446595, "grad_norm": 0.6786671973271104, "learning_rate": 4.022832813384355e-06, "loss": 0.2777, "step": 16657 }, { "epoch": 1.1516073280331836, "grad_norm": 0.8441117342251714, "learning_rate": 4.022283854142635e-06, "loss": 0.2951, "step": 16658 }, { "epoch": 1.1516764604217076, "grad_norm": 0.7554502296097034, "learning_rate": 4.021734907154843e-06, "loss": 0.376, "step": 16659 }, { "epoch": 1.1517455928102316, "grad_norm": 0.7007744058032184, "learning_rate": 4.021185972427857e-06, "loss": 0.2881, "step": 16660 }, { "epoch": 1.1518147251987556, "grad_norm": 0.7868883168341579, "learning_rate": 4.0206370499685594e-06, "loss": 0.2052, "step": 16661 }, { "epoch": 1.1518838575872796, "grad_norm": 0.7415663332225758, "learning_rate": 4.0200881397838275e-06, "loss": 0.387, "step": 16662 }, { "epoch": 1.1519529899758036, "grad_norm": 0.7054584795510744, "learning_rate": 4.019539241880543e-06, "loss": 0.3643, "step": 16663 }, { "epoch": 1.1520221223643277, "grad_norm": 0.6422003224395565, "learning_rate": 4.018990356265583e-06, "loss": 0.2524, "step": 16664 }, { "epoch": 1.1520912547528517, "grad_norm": 0.6155778041863675, "learning_rate": 4.018441482945831e-06, "loss": 0.2217, "step": 16665 }, { "epoch": 1.1521603871413757, "grad_norm": 0.6447227338349988, "learning_rate": 4.0178926219281596e-06, "loss": 0.3547, "step": 16666 }, { "epoch": 1.1522295195298997, "grad_norm": 0.6451297812445799, "learning_rate": 4.017343773219455e-06, "loss": 0.2866, "step": 16667 }, { "epoch": 1.1522986519184237, "grad_norm": 0.6162935248818173, "learning_rate": 4.01679493682659e-06, "loss": 0.2241, "step": 16668 }, { "epoch": 1.1523677843069478, "grad_norm": 0.680218785605905, "learning_rate": 4.016246112756447e-06, "loss": 0.2859, "step": 16669 }, { "epoch": 1.1524369166954718, "grad_norm": 0.6968764648643122, "learning_rate": 4.015697301015903e-06, "loss": 0.3252, "step": 16670 }, { "epoch": 1.1525060490839958, "grad_norm": 0.6434990185488814, "learning_rate": 4.0151485016118385e-06, "loss": 0.2486, "step": 16671 }, { "epoch": 1.1525751814725198, "grad_norm": 0.6182037156720283, "learning_rate": 4.014599714551127e-06, "loss": 0.2671, "step": 16672 }, { "epoch": 1.1526443138610438, "grad_norm": 0.6327082748874784, "learning_rate": 4.014050939840652e-06, "loss": 0.3161, "step": 16673 }, { "epoch": 1.1527134462495678, "grad_norm": 0.6505680582084469, "learning_rate": 4.013502177487287e-06, "loss": 0.2694, "step": 16674 }, { "epoch": 1.1527825786380919, "grad_norm": 0.5846408735244696, "learning_rate": 4.012953427497914e-06, "loss": 0.2732, "step": 16675 }, { "epoch": 1.1528517110266159, "grad_norm": 0.6902605944638754, "learning_rate": 4.012404689879407e-06, "loss": 0.3049, "step": 16676 }, { "epoch": 1.15292084341514, "grad_norm": 0.6177953516474717, "learning_rate": 4.011855964638645e-06, "loss": 0.2946, "step": 16677 }, { "epoch": 1.152989975803664, "grad_norm": 0.64452165920117, "learning_rate": 4.0113072517825045e-06, "loss": 0.278, "step": 16678 }, { "epoch": 1.153059108192188, "grad_norm": 0.6724018657959944, "learning_rate": 4.010758551317865e-06, "loss": 0.2326, "step": 16679 }, { "epoch": 1.153128240580712, "grad_norm": 1.323905901459125, "learning_rate": 4.010209863251599e-06, "loss": 0.3563, "step": 16680 }, { "epoch": 1.153197372969236, "grad_norm": 0.7261218808513102, "learning_rate": 4.009661187590588e-06, "loss": 0.254, "step": 16681 }, { "epoch": 1.1532665053577602, "grad_norm": 0.8190263320664998, "learning_rate": 4.009112524341705e-06, "loss": 0.244, "step": 16682 }, { "epoch": 1.1533356377462842, "grad_norm": 0.6890128306660874, "learning_rate": 4.00856387351183e-06, "loss": 0.2615, "step": 16683 }, { "epoch": 1.1534047701348082, "grad_norm": 0.6955648900007951, "learning_rate": 4.008015235107835e-06, "loss": 0.4123, "step": 16684 }, { "epoch": 1.1534739025233323, "grad_norm": 0.7005378844347704, "learning_rate": 4.007466609136602e-06, "loss": 0.2228, "step": 16685 }, { "epoch": 1.1535430349118563, "grad_norm": 0.6406192319010037, "learning_rate": 4.006917995605002e-06, "loss": 0.21, "step": 16686 }, { "epoch": 1.1536121673003803, "grad_norm": 0.7331277107581831, "learning_rate": 4.006369394519913e-06, "loss": 0.3811, "step": 16687 }, { "epoch": 1.1536812996889043, "grad_norm": 0.6303264050151448, "learning_rate": 4.0058208058882096e-06, "loss": 0.2835, "step": 16688 }, { "epoch": 1.1537504320774283, "grad_norm": 0.7030842217574396, "learning_rate": 4.00527222971677e-06, "loss": 0.2442, "step": 16689 }, { "epoch": 1.1538195644659524, "grad_norm": 0.6958777010770526, "learning_rate": 4.004723666012467e-06, "loss": 0.2814, "step": 16690 }, { "epoch": 1.1538886968544764, "grad_norm": 0.6500380910778146, "learning_rate": 4.004175114782176e-06, "loss": 0.3818, "step": 16691 }, { "epoch": 1.1539578292430004, "grad_norm": 0.6546878814622172, "learning_rate": 4.003626576032773e-06, "loss": 0.2667, "step": 16692 }, { "epoch": 1.1540269616315244, "grad_norm": 0.6727345712660967, "learning_rate": 4.003078049771134e-06, "loss": 0.2685, "step": 16693 }, { "epoch": 1.1540960940200484, "grad_norm": 0.6545056776066597, "learning_rate": 4.002529536004131e-06, "loss": 0.3446, "step": 16694 }, { "epoch": 1.1541652264085724, "grad_norm": 0.6731037475570966, "learning_rate": 4.001981034738641e-06, "loss": 0.2918, "step": 16695 }, { "epoch": 1.1542343587970965, "grad_norm": 0.6749145541448607, "learning_rate": 4.001432545981539e-06, "loss": 0.2727, "step": 16696 }, { "epoch": 1.1543034911856205, "grad_norm": 0.6365568567514852, "learning_rate": 4.000884069739696e-06, "loss": 0.248, "step": 16697 }, { "epoch": 1.1543726235741445, "grad_norm": 0.6674544019615392, "learning_rate": 4.0003356060199885e-06, "loss": 0.4302, "step": 16698 }, { "epoch": 1.1544417559626685, "grad_norm": 0.6564461548292444, "learning_rate": 3.99978715482929e-06, "loss": 0.2644, "step": 16699 }, { "epoch": 1.1545108883511925, "grad_norm": 0.7457652621017188, "learning_rate": 3.999238716174477e-06, "loss": 0.2071, "step": 16700 }, { "epoch": 1.1545800207397166, "grad_norm": 0.8425419978023869, "learning_rate": 3.998690290062417e-06, "loss": 0.278, "step": 16701 }, { "epoch": 1.1546491531282406, "grad_norm": 0.6717057696319519, "learning_rate": 3.99814187649999e-06, "loss": 0.3617, "step": 16702 }, { "epoch": 1.1547182855167646, "grad_norm": 0.6642902636470462, "learning_rate": 3.997593475494065e-06, "loss": 0.2517, "step": 16703 }, { "epoch": 1.1547874179052886, "grad_norm": 0.6048551912452335, "learning_rate": 3.997045087051518e-06, "loss": 0.2192, "step": 16704 }, { "epoch": 1.1548565502938126, "grad_norm": 0.7267083378787439, "learning_rate": 3.99649671117922e-06, "loss": 0.3551, "step": 16705 }, { "epoch": 1.1549256826823366, "grad_norm": 0.6215002549425407, "learning_rate": 3.995948347884047e-06, "loss": 0.2801, "step": 16706 }, { "epoch": 1.1549948150708607, "grad_norm": 0.6282694578434758, "learning_rate": 3.9953999971728665e-06, "loss": 0.2107, "step": 16707 }, { "epoch": 1.1550639474593847, "grad_norm": 0.6628272412220608, "learning_rate": 3.9948516590525564e-06, "loss": 0.3523, "step": 16708 }, { "epoch": 1.1551330798479087, "grad_norm": 0.690878295575353, "learning_rate": 3.994303333529986e-06, "loss": 0.3466, "step": 16709 }, { "epoch": 1.1552022122364327, "grad_norm": 0.6243182240337276, "learning_rate": 3.99375502061203e-06, "loss": 0.2654, "step": 16710 }, { "epoch": 1.1552713446249567, "grad_norm": 0.6380264435228581, "learning_rate": 3.993206720305558e-06, "loss": 0.3069, "step": 16711 }, { "epoch": 1.1553404770134807, "grad_norm": 0.6696156469499858, "learning_rate": 3.992658432617444e-06, "loss": 0.3345, "step": 16712 }, { "epoch": 1.1554096094020048, "grad_norm": 0.6967890386171565, "learning_rate": 3.9921101575545565e-06, "loss": 0.2984, "step": 16713 }, { "epoch": 1.1554787417905288, "grad_norm": 0.7199943008607351, "learning_rate": 3.991561895123772e-06, "loss": 0.2751, "step": 16714 }, { "epoch": 1.1555478741790528, "grad_norm": 0.6375999438818916, "learning_rate": 3.991013645331958e-06, "loss": 0.2743, "step": 16715 }, { "epoch": 1.1556170065675768, "grad_norm": 0.6609718810872303, "learning_rate": 3.990465408185988e-06, "loss": 0.3588, "step": 16716 }, { "epoch": 1.1556861389561008, "grad_norm": 0.6261984414336809, "learning_rate": 3.989917183692732e-06, "loss": 0.2712, "step": 16717 }, { "epoch": 1.155755271344625, "grad_norm": 0.5575846457742107, "learning_rate": 3.989368971859063e-06, "loss": 0.2546, "step": 16718 }, { "epoch": 1.155824403733149, "grad_norm": 0.6941320321677472, "learning_rate": 3.988820772691848e-06, "loss": 0.2951, "step": 16719 }, { "epoch": 1.1558935361216731, "grad_norm": 0.696184243832229, "learning_rate": 3.9882725861979625e-06, "loss": 0.3426, "step": 16720 }, { "epoch": 1.1559626685101971, "grad_norm": 0.6459637910926848, "learning_rate": 3.987724412384273e-06, "loss": 0.2394, "step": 16721 }, { "epoch": 1.1560318008987212, "grad_norm": 0.6302705901525592, "learning_rate": 3.987176251257652e-06, "loss": 0.2468, "step": 16722 }, { "epoch": 1.1561009332872452, "grad_norm": 0.7557378636334826, "learning_rate": 3.986628102824968e-06, "loss": 0.3799, "step": 16723 }, { "epoch": 1.1561700656757692, "grad_norm": 0.6926789226972018, "learning_rate": 3.986079967093096e-06, "loss": 0.2608, "step": 16724 }, { "epoch": 1.1562391980642932, "grad_norm": 0.6791019042879503, "learning_rate": 3.985531844068899e-06, "loss": 0.2983, "step": 16725 }, { "epoch": 1.1563083304528172, "grad_norm": 0.6604573205430997, "learning_rate": 3.984983733759251e-06, "loss": 0.3377, "step": 16726 }, { "epoch": 1.1563774628413412, "grad_norm": 0.6849935045571253, "learning_rate": 3.984435636171019e-06, "loss": 0.3427, "step": 16727 }, { "epoch": 1.1564465952298653, "grad_norm": 0.8362875339364888, "learning_rate": 3.983887551311077e-06, "loss": 0.2462, "step": 16728 }, { "epoch": 1.1565157276183893, "grad_norm": 0.6734499828949859, "learning_rate": 3.983339479186288e-06, "loss": 0.2842, "step": 16729 }, { "epoch": 1.1565848600069133, "grad_norm": 0.8944774576231422, "learning_rate": 3.9827914198035265e-06, "loss": 0.343, "step": 16730 }, { "epoch": 1.1566539923954373, "grad_norm": 0.6299607993456376, "learning_rate": 3.982243373169658e-06, "loss": 0.2338, "step": 16731 }, { "epoch": 1.1567231247839613, "grad_norm": 0.6942569907806939, "learning_rate": 3.9816953392915535e-06, "loss": 0.2657, "step": 16732 }, { "epoch": 1.1567922571724853, "grad_norm": 0.664216627799737, "learning_rate": 3.981147318176079e-06, "loss": 0.3091, "step": 16733 }, { "epoch": 1.1568613895610094, "grad_norm": 0.6622064017574428, "learning_rate": 3.980599309830107e-06, "loss": 0.3156, "step": 16734 }, { "epoch": 1.1569305219495334, "grad_norm": 0.6287547733407116, "learning_rate": 3.980051314260501e-06, "loss": 0.2396, "step": 16735 }, { "epoch": 1.1569996543380574, "grad_norm": 1.0491005249960434, "learning_rate": 3.979503331474135e-06, "loss": 0.2842, "step": 16736 }, { "epoch": 1.1570687867265814, "grad_norm": 0.6410839426298909, "learning_rate": 3.978955361477869e-06, "loss": 0.276, "step": 16737 }, { "epoch": 1.1571379191151054, "grad_norm": 0.8316165613284007, "learning_rate": 3.978407404278579e-06, "loss": 0.2796, "step": 16738 }, { "epoch": 1.1572070515036295, "grad_norm": 0.6885438799415106, "learning_rate": 3.977859459883128e-06, "loss": 0.2616, "step": 16739 }, { "epoch": 1.1572761838921535, "grad_norm": 0.6903846425054886, "learning_rate": 3.977311528298385e-06, "loss": 0.3302, "step": 16740 }, { "epoch": 1.1573453162806775, "grad_norm": 0.6505720627216537, "learning_rate": 3.976763609531215e-06, "loss": 0.3317, "step": 16741 }, { "epoch": 1.1574144486692015, "grad_norm": 0.6353860712064946, "learning_rate": 3.976215703588491e-06, "loss": 0.2152, "step": 16742 }, { "epoch": 1.1574835810577255, "grad_norm": 0.671166872755106, "learning_rate": 3.975667810477072e-06, "loss": 0.2776, "step": 16743 }, { "epoch": 1.1575527134462495, "grad_norm": 0.7002069088665928, "learning_rate": 3.975119930203831e-06, "loss": 0.3821, "step": 16744 }, { "epoch": 1.1576218458347736, "grad_norm": 0.8381016847479834, "learning_rate": 3.974572062775631e-06, "loss": 0.318, "step": 16745 }, { "epoch": 1.1576909782232976, "grad_norm": 0.59974223687824, "learning_rate": 3.974024208199343e-06, "loss": 0.2533, "step": 16746 }, { "epoch": 1.1577601106118216, "grad_norm": 0.6370085698335286, "learning_rate": 3.973476366481829e-06, "loss": 0.2498, "step": 16747 }, { "epoch": 1.1578292430003456, "grad_norm": 1.1921826158743083, "learning_rate": 3.972928537629957e-06, "loss": 0.3343, "step": 16748 }, { "epoch": 1.1578983753888696, "grad_norm": 0.6534079453666706, "learning_rate": 3.972380721650593e-06, "loss": 0.2757, "step": 16749 }, { "epoch": 1.1579675077773937, "grad_norm": 0.6227540961231606, "learning_rate": 3.971832918550604e-06, "loss": 0.2613, "step": 16750 }, { "epoch": 1.1580366401659177, "grad_norm": 0.7983858073696917, "learning_rate": 3.971285128336852e-06, "loss": 0.2925, "step": 16751 }, { "epoch": 1.1581057725544417, "grad_norm": 0.6604469648536695, "learning_rate": 3.970737351016207e-06, "loss": 0.3604, "step": 16752 }, { "epoch": 1.1581749049429657, "grad_norm": 0.6964992630686782, "learning_rate": 3.970189586595531e-06, "loss": 0.2612, "step": 16753 }, { "epoch": 1.1582440373314897, "grad_norm": 0.7004670606667153, "learning_rate": 3.969641835081692e-06, "loss": 0.2989, "step": 16754 }, { "epoch": 1.1583131697200137, "grad_norm": 0.6518984698256782, "learning_rate": 3.969094096481552e-06, "loss": 0.3076, "step": 16755 }, { "epoch": 1.1583823021085378, "grad_norm": 0.6386707989089784, "learning_rate": 3.968546370801981e-06, "loss": 0.3137, "step": 16756 }, { "epoch": 1.1584514344970618, "grad_norm": 0.6288518031235172, "learning_rate": 3.967998658049838e-06, "loss": 0.2933, "step": 16757 }, { "epoch": 1.1585205668855858, "grad_norm": 0.6917146306121285, "learning_rate": 3.967450958231991e-06, "loss": 0.3142, "step": 16758 }, { "epoch": 1.1585896992741098, "grad_norm": 0.639178538325616, "learning_rate": 3.966903271355302e-06, "loss": 0.3189, "step": 16759 }, { "epoch": 1.1586588316626338, "grad_norm": 0.6586708068159827, "learning_rate": 3.96635559742664e-06, "loss": 0.2223, "step": 16760 }, { "epoch": 1.1587279640511579, "grad_norm": 0.6112584673452911, "learning_rate": 3.965807936452864e-06, "loss": 0.2592, "step": 16761 }, { "epoch": 1.1587970964396819, "grad_norm": 0.6568878446296372, "learning_rate": 3.96526028844084e-06, "loss": 0.29, "step": 16762 }, { "epoch": 1.158866228828206, "grad_norm": 0.5914359011206112, "learning_rate": 3.964712653397433e-06, "loss": 0.2464, "step": 16763 }, { "epoch": 1.1589353612167301, "grad_norm": 0.6587434803254286, "learning_rate": 3.964165031329503e-06, "loss": 0.2214, "step": 16764 }, { "epoch": 1.1590044936052541, "grad_norm": 0.643847500574784, "learning_rate": 3.963617422243919e-06, "loss": 0.2768, "step": 16765 }, { "epoch": 1.1590736259937782, "grad_norm": 1.1708692111925103, "learning_rate": 3.963069826147538e-06, "loss": 0.319, "step": 16766 }, { "epoch": 1.1591427583823022, "grad_norm": 0.6834883865408681, "learning_rate": 3.962522243047228e-06, "loss": 0.2734, "step": 16767 }, { "epoch": 1.1592118907708262, "grad_norm": 0.6505224644181626, "learning_rate": 3.961974672949848e-06, "loss": 0.2422, "step": 16768 }, { "epoch": 1.1592810231593502, "grad_norm": 0.7352041801223611, "learning_rate": 3.961427115862266e-06, "loss": 0.3527, "step": 16769 }, { "epoch": 1.1593501555478742, "grad_norm": 0.6755924560927411, "learning_rate": 3.96087957179134e-06, "loss": 0.3373, "step": 16770 }, { "epoch": 1.1594192879363983, "grad_norm": 0.6849979242804196, "learning_rate": 3.960332040743935e-06, "loss": 0.2762, "step": 16771 }, { "epoch": 1.1594884203249223, "grad_norm": 0.6606659904681889, "learning_rate": 3.95978452272691e-06, "loss": 0.2428, "step": 16772 }, { "epoch": 1.1595575527134463, "grad_norm": 0.6411919099535475, "learning_rate": 3.959237017747133e-06, "loss": 0.3325, "step": 16773 }, { "epoch": 1.1596266851019703, "grad_norm": 0.6779753267408076, "learning_rate": 3.958689525811461e-06, "loss": 0.2851, "step": 16774 }, { "epoch": 1.1596958174904943, "grad_norm": 0.6468495753350335, "learning_rate": 3.958142046926758e-06, "loss": 0.2364, "step": 16775 }, { "epoch": 1.1597649498790183, "grad_norm": 0.6676526749732087, "learning_rate": 3.9575945810998836e-06, "loss": 0.3343, "step": 16776 }, { "epoch": 1.1598340822675424, "grad_norm": 0.7008810810305673, "learning_rate": 3.957047128337704e-06, "loss": 0.3823, "step": 16777 }, { "epoch": 1.1599032146560664, "grad_norm": 0.6426732966309706, "learning_rate": 3.956499688647075e-06, "loss": 0.258, "step": 16778 }, { "epoch": 1.1599723470445904, "grad_norm": 0.6562625943089332, "learning_rate": 3.955952262034861e-06, "loss": 0.2422, "step": 16779 }, { "epoch": 1.1600414794331144, "grad_norm": 0.709242973660765, "learning_rate": 3.955404848507921e-06, "loss": 0.3644, "step": 16780 }, { "epoch": 1.1601106118216384, "grad_norm": 0.6351582028388426, "learning_rate": 3.954857448073119e-06, "loss": 0.2563, "step": 16781 }, { "epoch": 1.1601797442101625, "grad_norm": 0.6168582744308057, "learning_rate": 3.954310060737313e-06, "loss": 0.2398, "step": 16782 }, { "epoch": 1.1602488765986865, "grad_norm": 0.6659816235602898, "learning_rate": 3.953762686507364e-06, "loss": 0.3053, "step": 16783 }, { "epoch": 1.1603180089872105, "grad_norm": 0.6617053134780094, "learning_rate": 3.953215325390132e-06, "loss": 0.3641, "step": 16784 }, { "epoch": 1.1603871413757345, "grad_norm": 0.5954368383217288, "learning_rate": 3.95266797739248e-06, "loss": 0.2512, "step": 16785 }, { "epoch": 1.1604562737642585, "grad_norm": 0.6620449199720687, "learning_rate": 3.952120642521263e-06, "loss": 0.2533, "step": 16786 }, { "epoch": 1.1605254061527825, "grad_norm": 0.6707387932798523, "learning_rate": 3.9515733207833464e-06, "loss": 0.3465, "step": 16787 }, { "epoch": 1.1605945385413066, "grad_norm": 0.6606928387266697, "learning_rate": 3.951026012185585e-06, "loss": 0.274, "step": 16788 }, { "epoch": 1.1606636709298306, "grad_norm": 0.5829510862328813, "learning_rate": 3.950478716734843e-06, "loss": 0.2816, "step": 16789 }, { "epoch": 1.1607328033183546, "grad_norm": 0.6688760107866049, "learning_rate": 3.949931434437973e-06, "loss": 0.2652, "step": 16790 }, { "epoch": 1.1608019357068786, "grad_norm": 0.6926795319190762, "learning_rate": 3.9493841653018435e-06, "loss": 0.3826, "step": 16791 }, { "epoch": 1.1608710680954026, "grad_norm": 0.6315137121207274, "learning_rate": 3.948836909333306e-06, "loss": 0.2953, "step": 16792 }, { "epoch": 1.1609402004839267, "grad_norm": 0.6762042136868732, "learning_rate": 3.948289666539222e-06, "loss": 0.21, "step": 16793 }, { "epoch": 1.1610093328724507, "grad_norm": 0.7030884272866206, "learning_rate": 3.94774243692645e-06, "loss": 0.315, "step": 16794 }, { "epoch": 1.1610784652609747, "grad_norm": 0.6994079688703112, "learning_rate": 3.947195220501851e-06, "loss": 0.2539, "step": 16795 }, { "epoch": 1.1611475976494987, "grad_norm": 0.7501877868778072, "learning_rate": 3.946648017272278e-06, "loss": 0.3131, "step": 16796 }, { "epoch": 1.1612167300380227, "grad_norm": 0.6861344561868377, "learning_rate": 3.946100827244594e-06, "loss": 0.2444, "step": 16797 }, { "epoch": 1.1612858624265467, "grad_norm": 0.7074308539005859, "learning_rate": 3.9455536504256546e-06, "loss": 0.344, "step": 16798 }, { "epoch": 1.161354994815071, "grad_norm": 0.6557673529407181, "learning_rate": 3.94500648682232e-06, "loss": 0.23, "step": 16799 }, { "epoch": 1.161424127203595, "grad_norm": 0.6256371861322876, "learning_rate": 3.944459336441446e-06, "loss": 0.258, "step": 16800 }, { "epoch": 1.161493259592119, "grad_norm": 0.6573366766504644, "learning_rate": 3.943912199289889e-06, "loss": 0.2799, "step": 16801 }, { "epoch": 1.161562391980643, "grad_norm": 0.7785104918764746, "learning_rate": 3.943365075374508e-06, "loss": 0.3241, "step": 16802 }, { "epoch": 1.161631524369167, "grad_norm": 0.662741726195267, "learning_rate": 3.9428179647021625e-06, "loss": 0.2456, "step": 16803 }, { "epoch": 1.161700656757691, "grad_norm": 0.6825538447501481, "learning_rate": 3.9422708672797045e-06, "loss": 0.2475, "step": 16804 }, { "epoch": 1.161769789146215, "grad_norm": 0.7270844341478707, "learning_rate": 3.941723783113995e-06, "loss": 0.3959, "step": 16805 }, { "epoch": 1.161838921534739, "grad_norm": 0.7419889652774102, "learning_rate": 3.9411767122118885e-06, "loss": 0.2692, "step": 16806 }, { "epoch": 1.1619080539232631, "grad_norm": 0.6616187065888286, "learning_rate": 3.940629654580243e-06, "loss": 0.264, "step": 16807 }, { "epoch": 1.1619771863117871, "grad_norm": 0.6250817145798591, "learning_rate": 3.940082610225913e-06, "loss": 0.2598, "step": 16808 }, { "epoch": 1.1620463187003112, "grad_norm": 0.6644491245324139, "learning_rate": 3.939535579155758e-06, "loss": 0.3545, "step": 16809 }, { "epoch": 1.1621154510888352, "grad_norm": 0.650592312109872, "learning_rate": 3.93898856137663e-06, "loss": 0.2119, "step": 16810 }, { "epoch": 1.1621845834773592, "grad_norm": 0.6734309469616505, "learning_rate": 3.938441556895389e-06, "loss": 0.3027, "step": 16811 }, { "epoch": 1.1622537158658832, "grad_norm": 0.6625557937067041, "learning_rate": 3.937894565718887e-06, "loss": 0.35, "step": 16812 }, { "epoch": 1.1623228482544072, "grad_norm": 0.6366587918643023, "learning_rate": 3.937347587853982e-06, "loss": 0.27, "step": 16813 }, { "epoch": 1.1623919806429313, "grad_norm": 0.815503317027142, "learning_rate": 3.936800623307528e-06, "loss": 0.226, "step": 16814 }, { "epoch": 1.1624611130314553, "grad_norm": 0.6874311678140844, "learning_rate": 3.936253672086382e-06, "loss": 0.3206, "step": 16815 }, { "epoch": 1.1625302454199793, "grad_norm": 0.6543436740398392, "learning_rate": 3.935706734197396e-06, "loss": 0.3637, "step": 16816 }, { "epoch": 1.1625993778085033, "grad_norm": 0.5860819648594112, "learning_rate": 3.93515980964743e-06, "loss": 0.2578, "step": 16817 }, { "epoch": 1.1626685101970273, "grad_norm": 0.666730597157339, "learning_rate": 3.9346128984433315e-06, "loss": 0.2472, "step": 16818 }, { "epoch": 1.1627376425855513, "grad_norm": 0.7590268082684244, "learning_rate": 3.934066000591962e-06, "loss": 0.2847, "step": 16819 }, { "epoch": 1.1628067749740754, "grad_norm": 0.7278545472190303, "learning_rate": 3.933519116100171e-06, "loss": 0.3032, "step": 16820 }, { "epoch": 1.1628759073625994, "grad_norm": 0.7665950422687416, "learning_rate": 3.932972244974816e-06, "loss": 0.2296, "step": 16821 }, { "epoch": 1.1629450397511234, "grad_norm": 0.6657345433469838, "learning_rate": 3.932425387222749e-06, "loss": 0.2385, "step": 16822 }, { "epoch": 1.1630141721396474, "grad_norm": 0.6426417103245626, "learning_rate": 3.931878542850826e-06, "loss": 0.3569, "step": 16823 }, { "epoch": 1.1630833045281714, "grad_norm": 0.6317106813563448, "learning_rate": 3.931331711865897e-06, "loss": 0.2416, "step": 16824 }, { "epoch": 1.1631524369166955, "grad_norm": 0.6972515327794312, "learning_rate": 3.93078489427482e-06, "loss": 0.2602, "step": 16825 }, { "epoch": 1.1632215693052195, "grad_norm": 0.7064350757180194, "learning_rate": 3.930238090084443e-06, "loss": 0.2976, "step": 16826 }, { "epoch": 1.1632907016937435, "grad_norm": 0.669772163221557, "learning_rate": 3.9296912993016265e-06, "loss": 0.3326, "step": 16827 }, { "epoch": 1.1633598340822675, "grad_norm": 0.6577173332300443, "learning_rate": 3.929144521933217e-06, "loss": 0.2524, "step": 16828 }, { "epoch": 1.1634289664707915, "grad_norm": 0.6866957192850095, "learning_rate": 3.928597757986071e-06, "loss": 0.2633, "step": 16829 }, { "epoch": 1.1634980988593155, "grad_norm": 0.7324718976360374, "learning_rate": 3.928051007467041e-06, "loss": 0.3396, "step": 16830 }, { "epoch": 1.1635672312478396, "grad_norm": 0.9507434381743318, "learning_rate": 3.927504270382977e-06, "loss": 0.2732, "step": 16831 }, { "epoch": 1.1636363636363636, "grad_norm": 0.5866641339557667, "learning_rate": 3.9269575467407335e-06, "loss": 0.2198, "step": 16832 }, { "epoch": 1.1637054960248876, "grad_norm": 0.8355418675953824, "learning_rate": 3.9264108365471606e-06, "loss": 0.2728, "step": 16833 }, { "epoch": 1.1637746284134116, "grad_norm": 0.6809188812601952, "learning_rate": 3.925864139809115e-06, "loss": 0.3806, "step": 16834 }, { "epoch": 1.1638437608019356, "grad_norm": 0.6654672106155604, "learning_rate": 3.925317456533444e-06, "loss": 0.2362, "step": 16835 }, { "epoch": 1.1639128931904597, "grad_norm": 0.6213756410526229, "learning_rate": 3.9247707867270005e-06, "loss": 0.1864, "step": 16836 }, { "epoch": 1.1639820255789837, "grad_norm": 0.6734726133382277, "learning_rate": 3.9242241303966364e-06, "loss": 0.3452, "step": 16837 }, { "epoch": 1.1640511579675077, "grad_norm": 0.6894972195942398, "learning_rate": 3.923677487549205e-06, "loss": 0.3227, "step": 16838 }, { "epoch": 1.1641202903560317, "grad_norm": 0.6026492241255738, "learning_rate": 3.923130858191552e-06, "loss": 0.253, "step": 16839 }, { "epoch": 1.1641894227445557, "grad_norm": 0.727632791518887, "learning_rate": 3.922584242330534e-06, "loss": 0.2756, "step": 16840 }, { "epoch": 1.1642585551330797, "grad_norm": 0.6599694497106429, "learning_rate": 3.922037639972999e-06, "loss": 0.3508, "step": 16841 }, { "epoch": 1.1643276875216038, "grad_norm": 0.6424138628854554, "learning_rate": 3.9214910511258e-06, "loss": 0.2516, "step": 16842 }, { "epoch": 1.1643968199101278, "grad_norm": 0.6065172344613363, "learning_rate": 3.920944475795785e-06, "loss": 0.2132, "step": 16843 }, { "epoch": 1.1644659522986518, "grad_norm": 0.6930326817579394, "learning_rate": 3.9203979139898056e-06, "loss": 0.307, "step": 16844 }, { "epoch": 1.164535084687176, "grad_norm": 0.8084506293241412, "learning_rate": 3.91985136571471e-06, "loss": 0.313, "step": 16845 }, { "epoch": 1.1646042170757, "grad_norm": 0.7120096556825978, "learning_rate": 3.919304830977352e-06, "loss": 0.2141, "step": 16846 }, { "epoch": 1.164673349464224, "grad_norm": 0.7050181404552299, "learning_rate": 3.918758309784578e-06, "loss": 0.2438, "step": 16847 }, { "epoch": 1.164742481852748, "grad_norm": 0.8261898023278172, "learning_rate": 3.918211802143241e-06, "loss": 0.3744, "step": 16848 }, { "epoch": 1.164811614241272, "grad_norm": 0.6441896558759048, "learning_rate": 3.9176653080601865e-06, "loss": 0.2463, "step": 16849 }, { "epoch": 1.1648807466297961, "grad_norm": 0.6793687345664131, "learning_rate": 3.917118827542267e-06, "loss": 0.2738, "step": 16850 }, { "epoch": 1.1649498790183201, "grad_norm": 0.6465919184480442, "learning_rate": 3.916572360596329e-06, "loss": 0.2985, "step": 16851 }, { "epoch": 1.1650190114068442, "grad_norm": 0.6747137271076119, "learning_rate": 3.916025907229225e-06, "loss": 0.3458, "step": 16852 }, { "epoch": 1.1650881437953682, "grad_norm": 0.6310786383146031, "learning_rate": 3.915479467447801e-06, "loss": 0.2608, "step": 16853 }, { "epoch": 1.1651572761838922, "grad_norm": 0.6649113745539372, "learning_rate": 3.914933041258906e-06, "loss": 0.2764, "step": 16854 }, { "epoch": 1.1652264085724162, "grad_norm": 0.7000278911464884, "learning_rate": 3.9143866286693885e-06, "loss": 0.2986, "step": 16855 }, { "epoch": 1.1652955409609402, "grad_norm": 0.7120397628888709, "learning_rate": 3.9138402296861e-06, "loss": 0.3074, "step": 16856 }, { "epoch": 1.1653646733494643, "grad_norm": 0.6439614078366352, "learning_rate": 3.913293844315883e-06, "loss": 0.2844, "step": 16857 }, { "epoch": 1.1654338057379883, "grad_norm": 0.6830703040427946, "learning_rate": 3.91274747256559e-06, "loss": 0.3153, "step": 16858 }, { "epoch": 1.1655029381265123, "grad_norm": 0.9545740627778732, "learning_rate": 3.912201114442066e-06, "loss": 0.3333, "step": 16859 }, { "epoch": 1.1655720705150363, "grad_norm": 0.6590752018925611, "learning_rate": 3.9116547699521615e-06, "loss": 0.2362, "step": 16860 }, { "epoch": 1.1656412029035603, "grad_norm": 0.7835681685278476, "learning_rate": 3.911108439102718e-06, "loss": 0.2491, "step": 16861 }, { "epoch": 1.1657103352920843, "grad_norm": 0.6296195429125202, "learning_rate": 3.910562121900592e-06, "loss": 0.2955, "step": 16862 }, { "epoch": 1.1657794676806084, "grad_norm": 0.7341264761197305, "learning_rate": 3.910015818352623e-06, "loss": 0.2782, "step": 16863 }, { "epoch": 1.1658486000691324, "grad_norm": 0.6465365113391038, "learning_rate": 3.909469528465661e-06, "loss": 0.2439, "step": 16864 }, { "epoch": 1.1659177324576564, "grad_norm": 0.6718876826510636, "learning_rate": 3.908923252246552e-06, "loss": 0.2787, "step": 16865 }, { "epoch": 1.1659868648461804, "grad_norm": 0.6636897827555844, "learning_rate": 3.9083769897021446e-06, "loss": 0.354, "step": 16866 }, { "epoch": 1.1660559972347044, "grad_norm": 0.8650641274101804, "learning_rate": 3.907830740839281e-06, "loss": 0.2379, "step": 16867 }, { "epoch": 1.1661251296232285, "grad_norm": 0.652053007045859, "learning_rate": 3.9072845056648115e-06, "loss": 0.255, "step": 16868 }, { "epoch": 1.1661942620117525, "grad_norm": 0.8568358455258983, "learning_rate": 3.90673828418558e-06, "loss": 0.303, "step": 16869 }, { "epoch": 1.1662633944002765, "grad_norm": 0.6756967965999603, "learning_rate": 3.906192076408435e-06, "loss": 0.3276, "step": 16870 }, { "epoch": 1.1663325267888005, "grad_norm": 0.6273181555248637, "learning_rate": 3.905645882340217e-06, "loss": 0.2961, "step": 16871 }, { "epoch": 1.1664016591773245, "grad_norm": 0.7116171436995059, "learning_rate": 3.905099701987778e-06, "loss": 0.2738, "step": 16872 }, { "epoch": 1.1664707915658485, "grad_norm": 0.6635216059932846, "learning_rate": 3.904553535357958e-06, "loss": 0.341, "step": 16873 }, { "epoch": 1.1665399239543726, "grad_norm": 0.6355171269062432, "learning_rate": 3.904007382457607e-06, "loss": 0.2756, "step": 16874 }, { "epoch": 1.1666090563428966, "grad_norm": 0.6550347445746829, "learning_rate": 3.903461243293566e-06, "loss": 0.2774, "step": 16875 }, { "epoch": 1.1666781887314206, "grad_norm": 0.6405677063820306, "learning_rate": 3.902915117872682e-06, "loss": 0.2904, "step": 16876 }, { "epoch": 1.1667473211199446, "grad_norm": 0.6605005335525018, "learning_rate": 3.902369006201798e-06, "loss": 0.3224, "step": 16877 }, { "epoch": 1.1668164535084686, "grad_norm": 0.6100609169813496, "learning_rate": 3.901822908287762e-06, "loss": 0.2122, "step": 16878 }, { "epoch": 1.1668855858969926, "grad_norm": 0.728510000923944, "learning_rate": 3.9012768241374125e-06, "loss": 0.2566, "step": 16879 }, { "epoch": 1.1669547182855167, "grad_norm": 0.6552601139541718, "learning_rate": 3.9007307537576005e-06, "loss": 0.3553, "step": 16880 }, { "epoch": 1.167023850674041, "grad_norm": 0.678322904965257, "learning_rate": 3.900184697155165e-06, "loss": 0.2789, "step": 16881 }, { "epoch": 1.167092983062565, "grad_norm": 0.6336583320035679, "learning_rate": 3.899638654336953e-06, "loss": 0.2488, "step": 16882 }, { "epoch": 1.167162115451089, "grad_norm": 0.677779613355452, "learning_rate": 3.899092625309804e-06, "loss": 0.2737, "step": 16883 }, { "epoch": 1.167231247839613, "grad_norm": 0.6381422022144192, "learning_rate": 3.898546610080568e-06, "loss": 0.3109, "step": 16884 }, { "epoch": 1.167300380228137, "grad_norm": 0.6149770079413699, "learning_rate": 3.898000608656082e-06, "loss": 0.2361, "step": 16885 }, { "epoch": 1.167369512616661, "grad_norm": 0.6208468921714045, "learning_rate": 3.897454621043193e-06, "loss": 0.2109, "step": 16886 }, { "epoch": 1.167438645005185, "grad_norm": 0.704574465068422, "learning_rate": 3.896908647248742e-06, "loss": 0.3258, "step": 16887 }, { "epoch": 1.167507777393709, "grad_norm": 0.5929245850442738, "learning_rate": 3.896362687279574e-06, "loss": 0.2595, "step": 16888 }, { "epoch": 1.167576909782233, "grad_norm": 0.6680176058201738, "learning_rate": 3.895816741142528e-06, "loss": 0.2541, "step": 16889 }, { "epoch": 1.167646042170757, "grad_norm": 0.6802159983749759, "learning_rate": 3.89527080884445e-06, "loss": 0.309, "step": 16890 }, { "epoch": 1.167715174559281, "grad_norm": 0.6876594703578974, "learning_rate": 3.89472489039218e-06, "loss": 0.3563, "step": 16891 }, { "epoch": 1.167784306947805, "grad_norm": 0.646118357698877, "learning_rate": 3.894178985792563e-06, "loss": 0.2621, "step": 16892 }, { "epoch": 1.1678534393363291, "grad_norm": 0.6545420153471351, "learning_rate": 3.893633095052438e-06, "loss": 0.2342, "step": 16893 }, { "epoch": 1.1679225717248531, "grad_norm": 0.6704356710906485, "learning_rate": 3.893087218178647e-06, "loss": 0.3248, "step": 16894 }, { "epoch": 1.1679917041133772, "grad_norm": 0.6580543746850294, "learning_rate": 3.892541355178034e-06, "loss": 0.3585, "step": 16895 }, { "epoch": 1.1680608365019012, "grad_norm": 0.8447253418710443, "learning_rate": 3.891995506057437e-06, "loss": 0.3211, "step": 16896 }, { "epoch": 1.1681299688904252, "grad_norm": 0.65448512815145, "learning_rate": 3.8914496708237e-06, "loss": 0.2343, "step": 16897 }, { "epoch": 1.1681991012789492, "grad_norm": 0.6706922627469497, "learning_rate": 3.890903849483662e-06, "loss": 0.3517, "step": 16898 }, { "epoch": 1.1682682336674732, "grad_norm": 0.7102295226432592, "learning_rate": 3.890358042044166e-06, "loss": 0.3679, "step": 16899 }, { "epoch": 1.1683373660559973, "grad_norm": 0.6140758370033798, "learning_rate": 3.88981224851205e-06, "loss": 0.2526, "step": 16900 }, { "epoch": 1.1684064984445213, "grad_norm": 1.0585939184687279, "learning_rate": 3.889266468894159e-06, "loss": 0.3201, "step": 16901 }, { "epoch": 1.1684756308330453, "grad_norm": 2.346208878163392, "learning_rate": 3.888720703197329e-06, "loss": 0.4116, "step": 16902 }, { "epoch": 1.1685447632215693, "grad_norm": 0.6240730128642147, "learning_rate": 3.888174951428402e-06, "loss": 0.2727, "step": 16903 }, { "epoch": 1.1686138956100933, "grad_norm": 0.6365182301625777, "learning_rate": 3.887629213594217e-06, "loss": 0.2163, "step": 16904 }, { "epoch": 1.1686830279986173, "grad_norm": 0.6388153238476979, "learning_rate": 3.887083489701617e-06, "loss": 0.3333, "step": 16905 }, { "epoch": 1.1687521603871414, "grad_norm": 0.792073352376672, "learning_rate": 3.886537779757438e-06, "loss": 0.3164, "step": 16906 }, { "epoch": 1.1688212927756654, "grad_norm": 0.6336143786494041, "learning_rate": 3.8859920837685205e-06, "loss": 0.2285, "step": 16907 }, { "epoch": 1.1688904251641894, "grad_norm": 0.7905820051390554, "learning_rate": 3.885446401741704e-06, "loss": 0.3196, "step": 16908 }, { "epoch": 1.1689595575527134, "grad_norm": 0.6549221902745977, "learning_rate": 3.88490073368383e-06, "loss": 0.3503, "step": 16909 }, { "epoch": 1.1690286899412374, "grad_norm": 0.7632831050818473, "learning_rate": 3.884355079601733e-06, "loss": 0.2399, "step": 16910 }, { "epoch": 1.1690978223297614, "grad_norm": 0.7736684408235691, "learning_rate": 3.883809439502255e-06, "loss": 0.2638, "step": 16911 }, { "epoch": 1.1691669547182855, "grad_norm": 0.6786844584040699, "learning_rate": 3.883263813392233e-06, "loss": 0.3711, "step": 16912 }, { "epoch": 1.1692360871068095, "grad_norm": 1.307034648844112, "learning_rate": 3.882718201278507e-06, "loss": 0.3152, "step": 16913 }, { "epoch": 1.1693052194953335, "grad_norm": 0.5963438555868076, "learning_rate": 3.882172603167914e-06, "loss": 0.236, "step": 16914 }, { "epoch": 1.1693743518838575, "grad_norm": 0.8156082633813543, "learning_rate": 3.881627019067293e-06, "loss": 0.2694, "step": 16915 }, { "epoch": 1.1694434842723815, "grad_norm": 0.6296580113498359, "learning_rate": 3.88108144898348e-06, "loss": 0.3411, "step": 16916 }, { "epoch": 1.1695126166609056, "grad_norm": 0.6743571667631061, "learning_rate": 3.8805358929233164e-06, "loss": 0.2378, "step": 16917 }, { "epoch": 1.1695817490494296, "grad_norm": 0.6074297075318835, "learning_rate": 3.879990350893635e-06, "loss": 0.2671, "step": 16918 }, { "epoch": 1.1696508814379536, "grad_norm": 0.6908376945545387, "learning_rate": 3.879444822901279e-06, "loss": 0.3613, "step": 16919 }, { "epoch": 1.1697200138264776, "grad_norm": 0.6793491279616659, "learning_rate": 3.8788993089530805e-06, "loss": 0.3732, "step": 16920 }, { "epoch": 1.1697891462150016, "grad_norm": 0.67273151879374, "learning_rate": 3.878353809055879e-06, "loss": 0.2421, "step": 16921 }, { "epoch": 1.1698582786035256, "grad_norm": 0.705335682629038, "learning_rate": 3.87780832321651e-06, "loss": 0.2965, "step": 16922 }, { "epoch": 1.1699274109920497, "grad_norm": 0.629027878526235, "learning_rate": 3.877262851441813e-06, "loss": 0.3278, "step": 16923 }, { "epoch": 1.1699965433805737, "grad_norm": 0.6598114300595384, "learning_rate": 3.876717393738621e-06, "loss": 0.2825, "step": 16924 }, { "epoch": 1.1700656757690977, "grad_norm": 0.6474887073748711, "learning_rate": 3.876171950113772e-06, "loss": 0.279, "step": 16925 }, { "epoch": 1.170134808157622, "grad_norm": 0.7122871828999969, "learning_rate": 3.875626520574101e-06, "loss": 0.285, "step": 16926 }, { "epoch": 1.170203940546146, "grad_norm": 0.6426077241103637, "learning_rate": 3.875081105126448e-06, "loss": 0.2758, "step": 16927 }, { "epoch": 1.17027307293467, "grad_norm": 0.6267765618225292, "learning_rate": 3.8745357037776435e-06, "loss": 0.2256, "step": 16928 }, { "epoch": 1.170342205323194, "grad_norm": 0.6661175729153316, "learning_rate": 3.873990316534527e-06, "loss": 0.2724, "step": 16929 }, { "epoch": 1.170411337711718, "grad_norm": 0.6960777129760133, "learning_rate": 3.87344494340393e-06, "loss": 0.353, "step": 16930 }, { "epoch": 1.170480470100242, "grad_norm": 0.8384391649800618, "learning_rate": 3.872899584392693e-06, "loss": 0.264, "step": 16931 }, { "epoch": 1.170549602488766, "grad_norm": 0.6053633057149936, "learning_rate": 3.872354239507646e-06, "loss": 0.2193, "step": 16932 }, { "epoch": 1.17061873487729, "grad_norm": 0.6720523821090362, "learning_rate": 3.871808908755628e-06, "loss": 0.2431, "step": 16933 }, { "epoch": 1.170687867265814, "grad_norm": 0.6683176765689155, "learning_rate": 3.87126359214347e-06, "loss": 0.3652, "step": 16934 }, { "epoch": 1.170756999654338, "grad_norm": 0.6186689728516245, "learning_rate": 3.870718289678011e-06, "loss": 0.2063, "step": 16935 }, { "epoch": 1.1708261320428621, "grad_norm": 0.6364167506618953, "learning_rate": 3.8701730013660805e-06, "loss": 0.225, "step": 16936 }, { "epoch": 1.1708952644313861, "grad_norm": 0.7864450190703214, "learning_rate": 3.8696277272145175e-06, "loss": 0.3662, "step": 16937 }, { "epoch": 1.1709643968199102, "grad_norm": 0.682489179763756, "learning_rate": 3.8690824672301515e-06, "loss": 0.2673, "step": 16938 }, { "epoch": 1.1710335292084342, "grad_norm": 1.9068730835267098, "learning_rate": 3.8685372214198206e-06, "loss": 0.2256, "step": 16939 }, { "epoch": 1.1711026615969582, "grad_norm": 0.6963687905497916, "learning_rate": 3.867991989790354e-06, "loss": 0.2653, "step": 16940 }, { "epoch": 1.1711717939854822, "grad_norm": 0.6213452064356437, "learning_rate": 3.867446772348591e-06, "loss": 0.3206, "step": 16941 }, { "epoch": 1.1712409263740062, "grad_norm": 0.6638396883184966, "learning_rate": 3.866901569101359e-06, "loss": 0.2195, "step": 16942 }, { "epoch": 1.1713100587625302, "grad_norm": 0.7162833979310548, "learning_rate": 3.866356380055495e-06, "loss": 0.2662, "step": 16943 }, { "epoch": 1.1713791911510543, "grad_norm": 0.654389639379359, "learning_rate": 3.865811205217829e-06, "loss": 0.2911, "step": 16944 }, { "epoch": 1.1714483235395783, "grad_norm": 0.6290859398546913, "learning_rate": 3.865266044595198e-06, "loss": 0.3164, "step": 16945 }, { "epoch": 1.1715174559281023, "grad_norm": 0.6478752964845441, "learning_rate": 3.86472089819443e-06, "loss": 0.2354, "step": 16946 }, { "epoch": 1.1715865883166263, "grad_norm": 0.6794826650464575, "learning_rate": 3.8641757660223606e-06, "loss": 0.2951, "step": 16947 }, { "epoch": 1.1716557207051503, "grad_norm": 2.072385940504138, "learning_rate": 3.86363064808582e-06, "loss": 0.3407, "step": 16948 }, { "epoch": 1.1717248530936744, "grad_norm": 0.6829039258367023, "learning_rate": 3.863085544391643e-06, "loss": 0.3327, "step": 16949 }, { "epoch": 1.1717939854821984, "grad_norm": 0.6817095161495668, "learning_rate": 3.862540454946657e-06, "loss": 0.2179, "step": 16950 }, { "epoch": 1.1718631178707224, "grad_norm": 0.6290339819321645, "learning_rate": 3.8619953797577e-06, "loss": 0.2859, "step": 16951 }, { "epoch": 1.1719322502592464, "grad_norm": 0.6931034168515142, "learning_rate": 3.861450318831597e-06, "loss": 0.3671, "step": 16952 }, { "epoch": 1.1720013826477704, "grad_norm": 0.7313361639302866, "learning_rate": 3.8609052721751844e-06, "loss": 0.2252, "step": 16953 }, { "epoch": 1.1720705150362944, "grad_norm": 0.6054074330459569, "learning_rate": 3.86036023979529e-06, "loss": 0.2973, "step": 16954 }, { "epoch": 1.1721396474248185, "grad_norm": 0.6757464809628234, "learning_rate": 3.859815221698749e-06, "loss": 0.3528, "step": 16955 }, { "epoch": 1.1722087798133425, "grad_norm": 0.5917707658744844, "learning_rate": 3.859270217892386e-06, "loss": 0.2712, "step": 16956 }, { "epoch": 1.1722779122018665, "grad_norm": 0.7302192740305836, "learning_rate": 3.858725228383036e-06, "loss": 0.2386, "step": 16957 }, { "epoch": 1.1723470445903905, "grad_norm": 0.7255343124285033, "learning_rate": 3.858180253177529e-06, "loss": 0.3653, "step": 16958 }, { "epoch": 1.1724161769789145, "grad_norm": 0.672553422514432, "learning_rate": 3.857635292282696e-06, "loss": 0.3498, "step": 16959 }, { "epoch": 1.1724853093674386, "grad_norm": 0.6675808835958605, "learning_rate": 3.857090345705364e-06, "loss": 0.2582, "step": 16960 }, { "epoch": 1.1725544417559626, "grad_norm": 0.6097976289728614, "learning_rate": 3.856545413452366e-06, "loss": 0.2314, "step": 16961 }, { "epoch": 1.1726235741444868, "grad_norm": 0.66724382869345, "learning_rate": 3.856000495530532e-06, "loss": 0.3602, "step": 16962 }, { "epoch": 1.1726927065330108, "grad_norm": 0.694728126553514, "learning_rate": 3.855455591946688e-06, "loss": 0.2905, "step": 16963 }, { "epoch": 1.1727618389215348, "grad_norm": 0.7494448627984466, "learning_rate": 3.8549107027076674e-06, "loss": 0.2356, "step": 16964 }, { "epoch": 1.1728309713100589, "grad_norm": 0.6978051193784008, "learning_rate": 3.854365827820295e-06, "loss": 0.3155, "step": 16965 }, { "epoch": 1.1729001036985829, "grad_norm": 0.7214912771197578, "learning_rate": 3.853820967291405e-06, "loss": 0.349, "step": 16966 }, { "epoch": 1.172969236087107, "grad_norm": 0.6818993590912504, "learning_rate": 3.8532761211278225e-06, "loss": 0.3027, "step": 16967 }, { "epoch": 1.173038368475631, "grad_norm": 0.6027948036821312, "learning_rate": 3.852731289336378e-06, "loss": 0.2151, "step": 16968 }, { "epoch": 1.173107500864155, "grad_norm": 0.757267194361626, "learning_rate": 3.8521864719238985e-06, "loss": 0.3516, "step": 16969 }, { "epoch": 1.173176633252679, "grad_norm": 0.667114851154081, "learning_rate": 3.851641668897213e-06, "loss": 0.3462, "step": 16970 }, { "epoch": 1.173245765641203, "grad_norm": 0.6047486852672156, "learning_rate": 3.85109688026315e-06, "loss": 0.1938, "step": 16971 }, { "epoch": 1.173314898029727, "grad_norm": 0.6684276871433018, "learning_rate": 3.85055210602854e-06, "loss": 0.336, "step": 16972 }, { "epoch": 1.173384030418251, "grad_norm": 0.6158981584842764, "learning_rate": 3.850007346200205e-06, "loss": 0.3524, "step": 16973 }, { "epoch": 1.173453162806775, "grad_norm": 0.6559314630673749, "learning_rate": 3.849462600784977e-06, "loss": 0.2619, "step": 16974 }, { "epoch": 1.173522295195299, "grad_norm": 0.6237110738785804, "learning_rate": 3.84891786978968e-06, "loss": 0.2589, "step": 16975 }, { "epoch": 1.173591427583823, "grad_norm": 0.6596137150048289, "learning_rate": 3.848373153221147e-06, "loss": 0.2766, "step": 16976 }, { "epoch": 1.173660559972347, "grad_norm": 0.6321786804796171, "learning_rate": 3.847828451086198e-06, "loss": 0.372, "step": 16977 }, { "epoch": 1.173729692360871, "grad_norm": 0.6116531805236906, "learning_rate": 3.8472837633916656e-06, "loss": 0.2517, "step": 16978 }, { "epoch": 1.1737988247493951, "grad_norm": 0.6367362839506313, "learning_rate": 3.846739090144373e-06, "loss": 0.2186, "step": 16979 }, { "epoch": 1.1738679571379191, "grad_norm": 0.614099524606555, "learning_rate": 3.846194431351148e-06, "loss": 0.2991, "step": 16980 }, { "epoch": 1.1739370895264432, "grad_norm": 0.6558054090750857, "learning_rate": 3.845649787018816e-06, "loss": 0.2353, "step": 16981 }, { "epoch": 1.1740062219149672, "grad_norm": 0.6157157918396162, "learning_rate": 3.8451051571542055e-06, "loss": 0.2259, "step": 16982 }, { "epoch": 1.1740753543034912, "grad_norm": 0.7262437274796094, "learning_rate": 3.844560541764138e-06, "loss": 0.2861, "step": 16983 }, { "epoch": 1.1741444866920152, "grad_norm": 0.6935833161403566, "learning_rate": 3.844015940855446e-06, "loss": 0.328, "step": 16984 }, { "epoch": 1.1742136190805392, "grad_norm": 0.7060501788858875, "learning_rate": 3.843471354434949e-06, "loss": 0.3121, "step": 16985 }, { "epoch": 1.1742827514690632, "grad_norm": 0.6123864096657556, "learning_rate": 3.842926782509475e-06, "loss": 0.2357, "step": 16986 }, { "epoch": 1.1743518838575873, "grad_norm": 0.6708154951150354, "learning_rate": 3.842382225085848e-06, "loss": 0.3334, "step": 16987 }, { "epoch": 1.1744210162461113, "grad_norm": 0.6845898074371596, "learning_rate": 3.841837682170894e-06, "loss": 0.3064, "step": 16988 }, { "epoch": 1.1744901486346353, "grad_norm": 0.8166695138164253, "learning_rate": 3.841293153771439e-06, "loss": 0.2669, "step": 16989 }, { "epoch": 1.1745592810231593, "grad_norm": 0.6975941533296627, "learning_rate": 3.840748639894307e-06, "loss": 0.2838, "step": 16990 }, { "epoch": 1.1746284134116833, "grad_norm": 0.679642131190552, "learning_rate": 3.8402041405463204e-06, "loss": 0.3322, "step": 16991 }, { "epoch": 1.1746975458002074, "grad_norm": 0.6050349075956672, "learning_rate": 3.839659655734306e-06, "loss": 0.2368, "step": 16992 }, { "epoch": 1.1747666781887314, "grad_norm": 0.7828393481565241, "learning_rate": 3.839115185465087e-06, "loss": 0.2376, "step": 16993 }, { "epoch": 1.1748358105772554, "grad_norm": 0.6543310875937446, "learning_rate": 3.838570729745488e-06, "loss": 0.3217, "step": 16994 }, { "epoch": 1.1749049429657794, "grad_norm": 0.6794215511814018, "learning_rate": 3.838026288582331e-06, "loss": 0.3454, "step": 16995 }, { "epoch": 1.1749740753543034, "grad_norm": 0.6214060772020397, "learning_rate": 3.837481861982442e-06, "loss": 0.2624, "step": 16996 }, { "epoch": 1.1750432077428274, "grad_norm": 0.6873318869039645, "learning_rate": 3.836937449952642e-06, "loss": 0.2683, "step": 16997 }, { "epoch": 1.1751123401313515, "grad_norm": 0.6915378888048073, "learning_rate": 3.836393052499757e-06, "loss": 0.3322, "step": 16998 }, { "epoch": 1.1751814725198755, "grad_norm": 0.6984194659180952, "learning_rate": 3.835848669630608e-06, "loss": 0.3034, "step": 16999 }, { "epoch": 1.1752506049083995, "grad_norm": 0.6278084316188305, "learning_rate": 3.835304301352017e-06, "loss": 0.2413, "step": 17000 }, { "epoch": 1.1753197372969235, "grad_norm": 0.6896554492145166, "learning_rate": 3.834759947670809e-06, "loss": 0.339, "step": 17001 }, { "epoch": 1.1753888696854475, "grad_norm": 0.8169461852754611, "learning_rate": 3.834215608593808e-06, "loss": 0.3831, "step": 17002 }, { "epoch": 1.1754580020739716, "grad_norm": 0.7044511898762674, "learning_rate": 3.83367128412783e-06, "loss": 0.3073, "step": 17003 }, { "epoch": 1.1755271344624956, "grad_norm": 0.7942536747962474, "learning_rate": 3.833126974279703e-06, "loss": 0.2376, "step": 17004 }, { "epoch": 1.1755962668510196, "grad_norm": 0.9156449990652608, "learning_rate": 3.832582679056246e-06, "loss": 0.3681, "step": 17005 }, { "epoch": 1.1756653992395436, "grad_norm": 0.6905417363412247, "learning_rate": 3.832038398464283e-06, "loss": 0.297, "step": 17006 }, { "epoch": 1.1757345316280676, "grad_norm": 0.5901364563937059, "learning_rate": 3.831494132510634e-06, "loss": 0.1757, "step": 17007 }, { "epoch": 1.1758036640165919, "grad_norm": 0.7055695911981557, "learning_rate": 3.8309498812021216e-06, "loss": 0.2904, "step": 17008 }, { "epoch": 1.1758727964051159, "grad_norm": 0.6682873864432216, "learning_rate": 3.830405644545564e-06, "loss": 0.3229, "step": 17009 }, { "epoch": 1.17594192879364, "grad_norm": 0.6876141680838651, "learning_rate": 3.829861422547787e-06, "loss": 0.2908, "step": 17010 }, { "epoch": 1.176011061182164, "grad_norm": 0.6429295683898709, "learning_rate": 3.829317215215606e-06, "loss": 0.2686, "step": 17011 }, { "epoch": 1.176080193570688, "grad_norm": 0.6705295108543592, "learning_rate": 3.828773022555848e-06, "loss": 0.346, "step": 17012 }, { "epoch": 1.176149325959212, "grad_norm": 0.6784126490343069, "learning_rate": 3.828228844575327e-06, "loss": 0.2935, "step": 17013 }, { "epoch": 1.176218458347736, "grad_norm": 0.5949368104884515, "learning_rate": 3.827684681280868e-06, "loss": 0.2225, "step": 17014 }, { "epoch": 1.17628759073626, "grad_norm": 0.7065605170235301, "learning_rate": 3.827140532679288e-06, "loss": 0.2959, "step": 17015 }, { "epoch": 1.176356723124784, "grad_norm": 0.726224954555251, "learning_rate": 3.82659639877741e-06, "loss": 0.3429, "step": 17016 }, { "epoch": 1.176425855513308, "grad_norm": 0.842589153596338, "learning_rate": 3.826052279582051e-06, "loss": 0.2399, "step": 17017 }, { "epoch": 1.176494987901832, "grad_norm": 0.5718524466692008, "learning_rate": 3.825508175100033e-06, "loss": 0.2472, "step": 17018 }, { "epoch": 1.176564120290356, "grad_norm": 0.6758073425054114, "learning_rate": 3.824964085338171e-06, "loss": 0.3389, "step": 17019 }, { "epoch": 1.17663325267888, "grad_norm": 0.6901636272744032, "learning_rate": 3.824420010303291e-06, "loss": 0.3178, "step": 17020 }, { "epoch": 1.176702385067404, "grad_norm": 0.5760504419454574, "learning_rate": 3.823875950002204e-06, "loss": 0.2042, "step": 17021 }, { "epoch": 1.1767715174559281, "grad_norm": 0.8617929826054445, "learning_rate": 3.823331904441737e-06, "loss": 0.2534, "step": 17022 }, { "epoch": 1.1768406498444521, "grad_norm": 0.6826889718489981, "learning_rate": 3.822787873628703e-06, "loss": 0.3882, "step": 17023 }, { "epoch": 1.1769097822329762, "grad_norm": 0.6720777752857129, "learning_rate": 3.8222438575699225e-06, "loss": 0.2792, "step": 17024 }, { "epoch": 1.1769789146215002, "grad_norm": 0.7162450906894742, "learning_rate": 3.821699856272213e-06, "loss": 0.2485, "step": 17025 }, { "epoch": 1.1770480470100242, "grad_norm": 0.6376385053931315, "learning_rate": 3.8211558697423946e-06, "loss": 0.2577, "step": 17026 }, { "epoch": 1.1771171793985482, "grad_norm": 0.6883449617340592, "learning_rate": 3.820611897987282e-06, "loss": 0.2816, "step": 17027 }, { "epoch": 1.1771863117870722, "grad_norm": 0.788303765644677, "learning_rate": 3.820067941013695e-06, "loss": 0.2842, "step": 17028 }, { "epoch": 1.1772554441755962, "grad_norm": 0.6880379091320287, "learning_rate": 3.819523998828452e-06, "loss": 0.2859, "step": 17029 }, { "epoch": 1.1773245765641203, "grad_norm": 0.6700234295854945, "learning_rate": 3.818980071438367e-06, "loss": 0.347, "step": 17030 }, { "epoch": 1.1773937089526443, "grad_norm": 0.7539977724717845, "learning_rate": 3.818436158850261e-06, "loss": 0.2642, "step": 17031 }, { "epoch": 1.1774628413411683, "grad_norm": 0.6642941019564049, "learning_rate": 3.817892261070947e-06, "loss": 0.2391, "step": 17032 }, { "epoch": 1.1775319737296923, "grad_norm": 0.6285810370693196, "learning_rate": 3.817348378107246e-06, "loss": 0.2507, "step": 17033 }, { "epoch": 1.1776011061182163, "grad_norm": 0.6405508441433156, "learning_rate": 3.8168045099659705e-06, "loss": 0.2952, "step": 17034 }, { "epoch": 1.1776702385067404, "grad_norm": 0.737389750915878, "learning_rate": 3.8162606566539406e-06, "loss": 0.3005, "step": 17035 }, { "epoch": 1.1777393708952644, "grad_norm": 0.5814460109514855, "learning_rate": 3.815716818177969e-06, "loss": 0.2283, "step": 17036 }, { "epoch": 1.1778085032837884, "grad_norm": 0.7490321372613868, "learning_rate": 3.815172994544876e-06, "loss": 0.3724, "step": 17037 }, { "epoch": 1.1778776356723124, "grad_norm": 0.7107073609509889, "learning_rate": 3.8146291857614735e-06, "loss": 0.3342, "step": 17038 }, { "epoch": 1.1779467680608364, "grad_norm": 0.5827365358093473, "learning_rate": 3.81408539183458e-06, "loss": 0.2718, "step": 17039 }, { "epoch": 1.1780159004493604, "grad_norm": 0.6594843636551087, "learning_rate": 3.8135416127710085e-06, "loss": 0.2182, "step": 17040 }, { "epoch": 1.1780850328378845, "grad_norm": 0.6970341726074567, "learning_rate": 3.8129978485775765e-06, "loss": 0.3877, "step": 17041 }, { "epoch": 1.1781541652264085, "grad_norm": 0.631525864646413, "learning_rate": 3.812454099261097e-06, "loss": 0.2958, "step": 17042 }, { "epoch": 1.1782232976149327, "grad_norm": 0.6467370910043606, "learning_rate": 3.811910364828388e-06, "loss": 0.2292, "step": 17043 }, { "epoch": 1.1782924300034567, "grad_norm": 0.6701296459502939, "learning_rate": 3.8113666452862604e-06, "loss": 0.324, "step": 17044 }, { "epoch": 1.1783615623919808, "grad_norm": 0.694102604514823, "learning_rate": 3.810822940641532e-06, "loss": 0.2902, "step": 17045 }, { "epoch": 1.1784306947805048, "grad_norm": 0.6636728416112275, "learning_rate": 3.8102792509010144e-06, "loss": 0.2871, "step": 17046 }, { "epoch": 1.1784998271690288, "grad_norm": 0.692061057253432, "learning_rate": 3.8097355760715256e-06, "loss": 0.325, "step": 17047 }, { "epoch": 1.1785689595575528, "grad_norm": 0.7504637118960068, "learning_rate": 3.809191916159875e-06, "loss": 0.3624, "step": 17048 }, { "epoch": 1.1786380919460768, "grad_norm": 0.8162059264378625, "learning_rate": 3.8086482711728796e-06, "loss": 0.2999, "step": 17049 }, { "epoch": 1.1787072243346008, "grad_norm": 0.6913214483909521, "learning_rate": 3.808104641117351e-06, "loss": 0.283, "step": 17050 }, { "epoch": 1.1787763567231249, "grad_norm": 0.7062693002528434, "learning_rate": 3.807561026000106e-06, "loss": 0.288, "step": 17051 }, { "epoch": 1.1788454891116489, "grad_norm": 0.6862365351343412, "learning_rate": 3.8070174258279537e-06, "loss": 0.3367, "step": 17052 }, { "epoch": 1.178914621500173, "grad_norm": 0.6175544848019332, "learning_rate": 3.8064738406077096e-06, "loss": 0.2741, "step": 17053 }, { "epoch": 1.178983753888697, "grad_norm": 0.6351840718727556, "learning_rate": 3.8059302703461855e-06, "loss": 0.2304, "step": 17054 }, { "epoch": 1.179052886277221, "grad_norm": 0.6654447291622703, "learning_rate": 3.8053867150501954e-06, "loss": 0.3558, "step": 17055 }, { "epoch": 1.179122018665745, "grad_norm": 0.7218440102997145, "learning_rate": 3.8048431747265502e-06, "loss": 0.3294, "step": 17056 }, { "epoch": 1.179191151054269, "grad_norm": 0.6847490502370899, "learning_rate": 3.804299649382063e-06, "loss": 0.2137, "step": 17057 }, { "epoch": 1.179260283442793, "grad_norm": 0.6914509638529409, "learning_rate": 3.8037561390235455e-06, "loss": 0.321, "step": 17058 }, { "epoch": 1.179329415831317, "grad_norm": 0.6525494876167159, "learning_rate": 3.803212643657811e-06, "loss": 0.3465, "step": 17059 }, { "epoch": 1.179398548219841, "grad_norm": 0.9229906173327252, "learning_rate": 3.802669163291669e-06, "loss": 0.2774, "step": 17060 }, { "epoch": 1.179467680608365, "grad_norm": 0.6582230986312736, "learning_rate": 3.8021256979319344e-06, "loss": 0.2449, "step": 17061 }, { "epoch": 1.179536812996889, "grad_norm": 0.7222335366153853, "learning_rate": 3.8015822475854144e-06, "loss": 0.392, "step": 17062 }, { "epoch": 1.179605945385413, "grad_norm": 0.6377379003618536, "learning_rate": 3.801038812258923e-06, "loss": 0.266, "step": 17063 }, { "epoch": 1.179675077773937, "grad_norm": 0.6992228480413449, "learning_rate": 3.8004953919592697e-06, "loss": 0.2708, "step": 17064 }, { "epoch": 1.1797442101624611, "grad_norm": 0.7081547542323431, "learning_rate": 3.7999519866932682e-06, "loss": 0.2857, "step": 17065 }, { "epoch": 1.1798133425509851, "grad_norm": 0.671525075708465, "learning_rate": 3.7994085964677247e-06, "loss": 0.4382, "step": 17066 }, { "epoch": 1.1798824749395092, "grad_norm": 0.8619675202576594, "learning_rate": 3.798865221289453e-06, "loss": 0.2769, "step": 17067 }, { "epoch": 1.1799516073280332, "grad_norm": 0.7037060254721857, "learning_rate": 3.7983218611652613e-06, "loss": 0.2501, "step": 17068 }, { "epoch": 1.1800207397165572, "grad_norm": 0.6847423750839384, "learning_rate": 3.7977785161019627e-06, "loss": 0.3389, "step": 17069 }, { "epoch": 1.1800898721050812, "grad_norm": 0.6198228374834646, "learning_rate": 3.797235186106363e-06, "loss": 0.2957, "step": 17070 }, { "epoch": 1.1801590044936052, "grad_norm": 0.6117038419837979, "learning_rate": 3.796691871185274e-06, "loss": 0.253, "step": 17071 }, { "epoch": 1.1802281368821292, "grad_norm": 0.6736345467775372, "learning_rate": 3.796148571345505e-06, "loss": 0.2595, "step": 17072 }, { "epoch": 1.1802972692706533, "grad_norm": 0.7299242648351209, "learning_rate": 3.795605286593867e-06, "loss": 0.386, "step": 17073 }, { "epoch": 1.1803664016591773, "grad_norm": 0.6694494985725683, "learning_rate": 3.795062016937165e-06, "loss": 0.2863, "step": 17074 }, { "epoch": 1.1804355340477013, "grad_norm": 0.759219516700184, "learning_rate": 3.7945187623822117e-06, "loss": 0.273, "step": 17075 }, { "epoch": 1.1805046664362253, "grad_norm": 0.6977380915940323, "learning_rate": 3.7939755229358134e-06, "loss": 0.3445, "step": 17076 }, { "epoch": 1.1805737988247493, "grad_norm": 0.648298117086246, "learning_rate": 3.79343229860478e-06, "loss": 0.3557, "step": 17077 }, { "epoch": 1.1806429312132734, "grad_norm": 0.6493939539754371, "learning_rate": 3.792889089395919e-06, "loss": 0.2724, "step": 17078 }, { "epoch": 1.1807120636017974, "grad_norm": 0.6341158071066595, "learning_rate": 3.792345895316042e-06, "loss": 0.2275, "step": 17079 }, { "epoch": 1.1807811959903214, "grad_norm": 0.69470058142157, "learning_rate": 3.7918027163719508e-06, "loss": 0.3729, "step": 17080 }, { "epoch": 1.1808503283788454, "grad_norm": 0.7410870280553478, "learning_rate": 3.7912595525704575e-06, "loss": 0.2961, "step": 17081 }, { "epoch": 1.1809194607673694, "grad_norm": 0.6592036197350576, "learning_rate": 3.7907164039183687e-06, "loss": 0.216, "step": 17082 }, { "epoch": 1.1809885931558934, "grad_norm": 0.8856891803468355, "learning_rate": 3.7901732704224924e-06, "loss": 0.2823, "step": 17083 }, { "epoch": 1.1810577255444175, "grad_norm": 0.7152071979291653, "learning_rate": 3.789630152089634e-06, "loss": 0.3522, "step": 17084 }, { "epoch": 1.1811268579329415, "grad_norm": 0.658947517282059, "learning_rate": 3.7890870489266023e-06, "loss": 0.2479, "step": 17085 }, { "epoch": 1.1811959903214655, "grad_norm": 0.7311731951740786, "learning_rate": 3.7885439609402024e-06, "loss": 0.2499, "step": 17086 }, { "epoch": 1.1812651227099895, "grad_norm": 0.6799256218335173, "learning_rate": 3.788000888137244e-06, "loss": 0.3434, "step": 17087 }, { "epoch": 1.1813342550985135, "grad_norm": 0.6511246859301828, "learning_rate": 3.78745783052453e-06, "loss": 0.3197, "step": 17088 }, { "epoch": 1.1814033874870378, "grad_norm": 0.6327943849983042, "learning_rate": 3.7869147881088687e-06, "loss": 0.2102, "step": 17089 }, { "epoch": 1.1814725198755618, "grad_norm": 0.6840409911088036, "learning_rate": 3.786371760897065e-06, "loss": 0.3021, "step": 17090 }, { "epoch": 1.1815416522640858, "grad_norm": 0.6647701593274538, "learning_rate": 3.785828748895928e-06, "loss": 0.3657, "step": 17091 }, { "epoch": 1.1816107846526098, "grad_norm": 0.8202192143706092, "learning_rate": 3.7852857521122564e-06, "loss": 0.2401, "step": 17092 }, { "epoch": 1.1816799170411338, "grad_norm": 0.573266843285201, "learning_rate": 3.784742770552864e-06, "loss": 0.2026, "step": 17093 }, { "epoch": 1.1817490494296579, "grad_norm": 0.6516551198921411, "learning_rate": 3.7841998042245504e-06, "loss": 0.3205, "step": 17094 }, { "epoch": 1.1818181818181819, "grad_norm": 0.6498805101985794, "learning_rate": 3.7836568531341234e-06, "loss": 0.3117, "step": 17095 }, { "epoch": 1.181887314206706, "grad_norm": 0.6713364650715423, "learning_rate": 3.783113917288388e-06, "loss": 0.2211, "step": 17096 }, { "epoch": 1.18195644659523, "grad_norm": 0.6228847910122924, "learning_rate": 3.782570996694146e-06, "loss": 0.2009, "step": 17097 }, { "epoch": 1.182025578983754, "grad_norm": 0.6853554885251298, "learning_rate": 3.782028091358205e-06, "loss": 0.3582, "step": 17098 }, { "epoch": 1.182094711372278, "grad_norm": 0.6942722651992053, "learning_rate": 3.781485201287367e-06, "loss": 0.2342, "step": 17099 }, { "epoch": 1.182163843760802, "grad_norm": 0.6785184528810444, "learning_rate": 3.7809423264884404e-06, "loss": 0.2186, "step": 17100 }, { "epoch": 1.182232976149326, "grad_norm": 0.664783931360648, "learning_rate": 3.7803994669682233e-06, "loss": 0.335, "step": 17101 }, { "epoch": 1.18230210853785, "grad_norm": 0.660047870117665, "learning_rate": 3.7798566227335238e-06, "loss": 0.3566, "step": 17102 }, { "epoch": 1.182371240926374, "grad_norm": 0.7193107435828419, "learning_rate": 3.7793137937911427e-06, "loss": 0.2755, "step": 17103 }, { "epoch": 1.182440373314898, "grad_norm": 0.6490096887553086, "learning_rate": 3.778770980147887e-06, "loss": 0.3159, "step": 17104 }, { "epoch": 1.182509505703422, "grad_norm": 0.6517724532973782, "learning_rate": 3.778228181810556e-06, "loss": 0.363, "step": 17105 }, { "epoch": 1.182578638091946, "grad_norm": 0.7038413739439833, "learning_rate": 3.777685398785955e-06, "loss": 0.2983, "step": 17106 }, { "epoch": 1.18264777048047, "grad_norm": 0.6077214610610074, "learning_rate": 3.7771426310808857e-06, "loss": 0.2682, "step": 17107 }, { "epoch": 1.1827169028689941, "grad_norm": 0.6868048666875669, "learning_rate": 3.776599878702153e-06, "loss": 0.2853, "step": 17108 }, { "epoch": 1.1827860352575181, "grad_norm": 0.652071850386472, "learning_rate": 3.776057141656555e-06, "loss": 0.3887, "step": 17109 }, { "epoch": 1.1828551676460421, "grad_norm": 0.6366844160105742, "learning_rate": 3.7755144199508982e-06, "loss": 0.2432, "step": 17110 }, { "epoch": 1.1829243000345662, "grad_norm": 0.6953807789763662, "learning_rate": 3.7749717135919816e-06, "loss": 0.2749, "step": 17111 }, { "epoch": 1.1829934324230902, "grad_norm": 0.7870067761016846, "learning_rate": 3.77442902258661e-06, "loss": 0.3744, "step": 17112 }, { "epoch": 1.1830625648116142, "grad_norm": 0.6989446046449972, "learning_rate": 3.7738863469415815e-06, "loss": 0.3407, "step": 17113 }, { "epoch": 1.1831316972001382, "grad_norm": 0.8846864845059945, "learning_rate": 3.7733436866637018e-06, "loss": 0.2426, "step": 17114 }, { "epoch": 1.1832008295886622, "grad_norm": 0.7639199675489499, "learning_rate": 3.772801041759768e-06, "loss": 0.3259, "step": 17115 }, { "epoch": 1.1832699619771863, "grad_norm": 0.7289563460608154, "learning_rate": 3.7722584122365845e-06, "loss": 0.3913, "step": 17116 }, { "epoch": 1.1833390943657103, "grad_norm": 0.6605797635292833, "learning_rate": 3.77171579810095e-06, "loss": 0.2663, "step": 17117 }, { "epoch": 1.1834082267542343, "grad_norm": 0.6229424408609907, "learning_rate": 3.7711731993596674e-06, "loss": 0.2375, "step": 17118 }, { "epoch": 1.1834773591427583, "grad_norm": 0.6497472932895844, "learning_rate": 3.770630616019534e-06, "loss": 0.2934, "step": 17119 }, { "epoch": 1.1835464915312823, "grad_norm": 0.7883089406522775, "learning_rate": 3.7700880480873524e-06, "loss": 0.3154, "step": 17120 }, { "epoch": 1.1836156239198063, "grad_norm": 0.6931245754074897, "learning_rate": 3.769545495569922e-06, "loss": 0.2929, "step": 17121 }, { "epoch": 1.1836847563083304, "grad_norm": 0.6593356767811371, "learning_rate": 3.769002958474045e-06, "loss": 0.2837, "step": 17122 }, { "epoch": 1.1837538886968544, "grad_norm": 0.6707033259889745, "learning_rate": 3.768460436806517e-06, "loss": 0.3853, "step": 17123 }, { "epoch": 1.1838230210853786, "grad_norm": 0.6769818378508167, "learning_rate": 3.7679179305741397e-06, "loss": 0.3439, "step": 17124 }, { "epoch": 1.1838921534739026, "grad_norm": 0.7274762966165295, "learning_rate": 3.7673754397837127e-06, "loss": 0.2509, "step": 17125 }, { "epoch": 1.1839612858624267, "grad_norm": 0.6951841837284641, "learning_rate": 3.7668329644420367e-06, "loss": 0.3203, "step": 17126 }, { "epoch": 1.1840304182509507, "grad_norm": 0.6624953180439196, "learning_rate": 3.7662905045559057e-06, "loss": 0.3753, "step": 17127 }, { "epoch": 1.1840995506394747, "grad_norm": 0.6774595742139612, "learning_rate": 3.7657480601321232e-06, "loss": 0.2744, "step": 17128 }, { "epoch": 1.1841686830279987, "grad_norm": 0.6609691967715731, "learning_rate": 3.7652056311774853e-06, "loss": 0.2425, "step": 17129 }, { "epoch": 1.1842378154165227, "grad_norm": 0.6380215293278829, "learning_rate": 3.7646632176987925e-06, "loss": 0.3465, "step": 17130 }, { "epoch": 1.1843069478050468, "grad_norm": 0.6925232835862435, "learning_rate": 3.76412081970284e-06, "loss": 0.2449, "step": 17131 }, { "epoch": 1.1843760801935708, "grad_norm": 0.6865743747724631, "learning_rate": 3.76357843719643e-06, "loss": 0.2821, "step": 17132 }, { "epoch": 1.1844452125820948, "grad_norm": 0.6746598831441933, "learning_rate": 3.7630360701863557e-06, "loss": 0.2962, "step": 17133 }, { "epoch": 1.1845143449706188, "grad_norm": 0.6968887932438371, "learning_rate": 3.7624937186794177e-06, "loss": 0.3492, "step": 17134 }, { "epoch": 1.1845834773591428, "grad_norm": 0.6519114102872745, "learning_rate": 3.761951382682412e-06, "loss": 0.2436, "step": 17135 }, { "epoch": 1.1846526097476668, "grad_norm": 0.653642117251989, "learning_rate": 3.761409062202137e-06, "loss": 0.209, "step": 17136 }, { "epoch": 1.1847217421361909, "grad_norm": 0.6790276397195274, "learning_rate": 3.7608667572453883e-06, "loss": 0.3443, "step": 17137 }, { "epoch": 1.1847908745247149, "grad_norm": 0.7485410951803464, "learning_rate": 3.7603244678189637e-06, "loss": 0.3144, "step": 17138 }, { "epoch": 1.184860006913239, "grad_norm": 0.626925646723683, "learning_rate": 3.759782193929659e-06, "loss": 0.1991, "step": 17139 }, { "epoch": 1.184929139301763, "grad_norm": 0.7086649940353077, "learning_rate": 3.7592399355842737e-06, "loss": 0.2685, "step": 17140 }, { "epoch": 1.184998271690287, "grad_norm": 0.6326306730123865, "learning_rate": 3.7586976927895992e-06, "loss": 0.3587, "step": 17141 }, { "epoch": 1.185067404078811, "grad_norm": 0.8135400833520562, "learning_rate": 3.7581554655524345e-06, "loss": 0.2318, "step": 17142 }, { "epoch": 1.185136536467335, "grad_norm": 0.627844822609545, "learning_rate": 3.757613253879574e-06, "loss": 0.2215, "step": 17143 }, { "epoch": 1.185205668855859, "grad_norm": 0.698508669235612, "learning_rate": 3.7570710577778163e-06, "loss": 0.3809, "step": 17144 }, { "epoch": 1.185274801244383, "grad_norm": 0.6957817412713002, "learning_rate": 3.756528877253953e-06, "loss": 0.3555, "step": 17145 }, { "epoch": 1.185343933632907, "grad_norm": 0.6759514176900651, "learning_rate": 3.7559867123147818e-06, "loss": 0.2507, "step": 17146 }, { "epoch": 1.185413066021431, "grad_norm": 0.706214467182873, "learning_rate": 3.7554445629670966e-06, "loss": 0.3113, "step": 17147 }, { "epoch": 1.185482198409955, "grad_norm": 0.6628827730369206, "learning_rate": 3.7549024292176935e-06, "loss": 0.3667, "step": 17148 }, { "epoch": 1.185551330798479, "grad_norm": 0.7591725588718887, "learning_rate": 3.754360311073365e-06, "loss": 0.2592, "step": 17149 }, { "epoch": 1.185620463187003, "grad_norm": 0.6287811259268431, "learning_rate": 3.75381820854091e-06, "loss": 0.2493, "step": 17150 }, { "epoch": 1.1856895955755271, "grad_norm": 0.6428993270976162, "learning_rate": 3.7532761216271176e-06, "loss": 0.2853, "step": 17151 }, { "epoch": 1.1857587279640511, "grad_norm": 0.695129804300364, "learning_rate": 3.7527340503387845e-06, "loss": 0.3466, "step": 17152 }, { "epoch": 1.1858278603525751, "grad_norm": 0.7079739007851636, "learning_rate": 3.7521919946827033e-06, "loss": 0.2721, "step": 17153 }, { "epoch": 1.1858969927410992, "grad_norm": 0.9384520962564056, "learning_rate": 3.7516499546656717e-06, "loss": 0.2788, "step": 17154 }, { "epoch": 1.1859661251296232, "grad_norm": 0.7075813537406073, "learning_rate": 3.7511079302944775e-06, "loss": 0.4021, "step": 17155 }, { "epoch": 1.1860352575181472, "grad_norm": 0.6483236661338911, "learning_rate": 3.7505659215759173e-06, "loss": 0.2948, "step": 17156 }, { "epoch": 1.1861043899066712, "grad_norm": 0.6704893485897928, "learning_rate": 3.750023928516783e-06, "loss": 0.2694, "step": 17157 }, { "epoch": 1.1861735222951952, "grad_norm": 0.6735737069141914, "learning_rate": 3.7494819511238702e-06, "loss": 0.2936, "step": 17158 }, { "epoch": 1.1862426546837193, "grad_norm": 0.655498538010184, "learning_rate": 3.748939989403967e-06, "loss": 0.4262, "step": 17159 }, { "epoch": 1.1863117870722433, "grad_norm": 0.6385467846604622, "learning_rate": 3.7483980433638694e-06, "loss": 0.2314, "step": 17160 }, { "epoch": 1.1863809194607673, "grad_norm": 0.6755373923929189, "learning_rate": 3.7478561130103703e-06, "loss": 0.2843, "step": 17161 }, { "epoch": 1.1864500518492913, "grad_norm": 0.7015306314335944, "learning_rate": 3.747314198350258e-06, "loss": 0.3307, "step": 17162 }, { "epoch": 1.1865191842378153, "grad_norm": 0.8328617596893952, "learning_rate": 3.746772299390328e-06, "loss": 0.2862, "step": 17163 }, { "epoch": 1.1865883166263393, "grad_norm": 0.7454946587531589, "learning_rate": 3.7462304161373684e-06, "loss": 0.2538, "step": 17164 }, { "epoch": 1.1866574490148634, "grad_norm": 0.6474496235461967, "learning_rate": 3.7456885485981763e-06, "loss": 0.2251, "step": 17165 }, { "epoch": 1.1867265814033874, "grad_norm": 0.6520854593843369, "learning_rate": 3.745146696779537e-06, "loss": 0.3733, "step": 17166 }, { "epoch": 1.1867957137919114, "grad_norm": 0.7443375661015842, "learning_rate": 3.744604860688247e-06, "loss": 0.2196, "step": 17167 }, { "epoch": 1.1868648461804354, "grad_norm": 0.6731458276992033, "learning_rate": 3.744063040331093e-06, "loss": 0.2706, "step": 17168 }, { "epoch": 1.1869339785689594, "grad_norm": 0.6851235310686752, "learning_rate": 3.7435212357148677e-06, "loss": 0.3213, "step": 17169 }, { "epoch": 1.1870031109574837, "grad_norm": 0.6644410935796371, "learning_rate": 3.742979446846361e-06, "loss": 0.3276, "step": 17170 }, { "epoch": 1.1870722433460077, "grad_norm": 0.6302161512214222, "learning_rate": 3.742437673732366e-06, "loss": 0.2276, "step": 17171 }, { "epoch": 1.1871413757345317, "grad_norm": 0.6864999421120356, "learning_rate": 3.7418959163796674e-06, "loss": 0.2439, "step": 17172 }, { "epoch": 1.1872105081230557, "grad_norm": 0.6642247061844612, "learning_rate": 3.741354174795061e-06, "loss": 0.347, "step": 17173 }, { "epoch": 1.1872796405115797, "grad_norm": 0.8903027930948383, "learning_rate": 3.7408124489853315e-06, "loss": 0.2718, "step": 17174 }, { "epoch": 1.1873487729001038, "grad_norm": 0.6680334589881144, "learning_rate": 3.7402707389572734e-06, "loss": 0.2074, "step": 17175 }, { "epoch": 1.1874179052886278, "grad_norm": 0.7356063547831035, "learning_rate": 3.7397290447176714e-06, "loss": 0.2301, "step": 17176 }, { "epoch": 1.1874870376771518, "grad_norm": 0.6969517372685237, "learning_rate": 3.739187366273318e-06, "loss": 0.3478, "step": 17177 }, { "epoch": 1.1875561700656758, "grad_norm": 1.266989191371729, "learning_rate": 3.7386457036309997e-06, "loss": 0.29, "step": 17178 }, { "epoch": 1.1876253024541998, "grad_norm": 0.7438909745230379, "learning_rate": 3.738104056797508e-06, "loss": 0.2722, "step": 17179 }, { "epoch": 1.1876944348427239, "grad_norm": 0.676047756670149, "learning_rate": 3.737562425779628e-06, "loss": 0.3428, "step": 17180 }, { "epoch": 1.1877635672312479, "grad_norm": 0.6563793955735773, "learning_rate": 3.7370208105841523e-06, "loss": 0.2594, "step": 17181 }, { "epoch": 1.187832699619772, "grad_norm": 0.6919422826629549, "learning_rate": 3.736479211217865e-06, "loss": 0.2998, "step": 17182 }, { "epoch": 1.187901832008296, "grad_norm": 0.6336443639790664, "learning_rate": 3.735937627687557e-06, "loss": 0.258, "step": 17183 }, { "epoch": 1.18797096439682, "grad_norm": 0.7039289811057219, "learning_rate": 3.735396060000014e-06, "loss": 0.3071, "step": 17184 }, { "epoch": 1.188040096785344, "grad_norm": 0.629920463970204, "learning_rate": 3.734854508162027e-06, "loss": 0.204, "step": 17185 }, { "epoch": 1.188109229173868, "grad_norm": 0.71164493906668, "learning_rate": 3.7343129721803782e-06, "loss": 0.2166, "step": 17186 }, { "epoch": 1.188178361562392, "grad_norm": 0.645910387103889, "learning_rate": 3.7337714520618596e-06, "loss": 0.3057, "step": 17187 }, { "epoch": 1.188247493950916, "grad_norm": 0.6317479944258028, "learning_rate": 3.7332299478132545e-06, "loss": 0.2922, "step": 17188 }, { "epoch": 1.18831662633944, "grad_norm": 0.6336319039482233, "learning_rate": 3.7326884594413537e-06, "loss": 0.2744, "step": 17189 }, { "epoch": 1.188385758727964, "grad_norm": 0.6445702977797304, "learning_rate": 3.7321469869529394e-06, "loss": 0.2767, "step": 17190 }, { "epoch": 1.188454891116488, "grad_norm": 0.6574689551352696, "learning_rate": 3.7316055303548004e-06, "loss": 0.3599, "step": 17191 }, { "epoch": 1.188524023505012, "grad_norm": 0.725388069783656, "learning_rate": 3.731064089653722e-06, "loss": 0.2446, "step": 17192 }, { "epoch": 1.188593155893536, "grad_norm": 0.7378320975620162, "learning_rate": 3.7305226648564928e-06, "loss": 0.2451, "step": 17193 }, { "epoch": 1.18866228828206, "grad_norm": 0.637061819599922, "learning_rate": 3.7299812559698944e-06, "loss": 0.2845, "step": 17194 }, { "epoch": 1.1887314206705841, "grad_norm": 0.7224593317019425, "learning_rate": 3.7294398630007154e-06, "loss": 0.3864, "step": 17195 }, { "epoch": 1.1888005530591081, "grad_norm": 0.6733018343475698, "learning_rate": 3.7288984859557394e-06, "loss": 0.2817, "step": 17196 }, { "epoch": 1.1888696854476322, "grad_norm": 0.674575212051261, "learning_rate": 3.7283571248417543e-06, "loss": 0.2479, "step": 17197 }, { "epoch": 1.1889388178361562, "grad_norm": 0.7185421570806694, "learning_rate": 3.7278157796655416e-06, "loss": 0.3292, "step": 17198 }, { "epoch": 1.1890079502246802, "grad_norm": 0.6332335703846554, "learning_rate": 3.7272744504338885e-06, "loss": 0.2754, "step": 17199 }, { "epoch": 1.1890770826132042, "grad_norm": 0.7573774398763162, "learning_rate": 3.726733137153578e-06, "loss": 0.2164, "step": 17200 }, { "epoch": 1.1891462150017282, "grad_norm": 0.667098087230424, "learning_rate": 3.726191839831397e-06, "loss": 0.3268, "step": 17201 }, { "epoch": 1.1892153473902523, "grad_norm": 0.7170749025169589, "learning_rate": 3.7256505584741265e-06, "loss": 0.3614, "step": 17202 }, { "epoch": 1.1892844797787763, "grad_norm": 0.6400315246172162, "learning_rate": 3.7251092930885533e-06, "loss": 0.2436, "step": 17203 }, { "epoch": 1.1893536121673003, "grad_norm": 0.6896212358957109, "learning_rate": 3.7245680436814593e-06, "loss": 0.3045, "step": 17204 }, { "epoch": 1.1894227445558243, "grad_norm": 0.6732347061869517, "learning_rate": 3.724026810259629e-06, "loss": 0.3661, "step": 17205 }, { "epoch": 1.1894918769443485, "grad_norm": 0.7558848545347576, "learning_rate": 3.7234855928298452e-06, "loss": 0.2292, "step": 17206 }, { "epoch": 1.1895610093328726, "grad_norm": 0.6009168913296467, "learning_rate": 3.7229443913988925e-06, "loss": 0.2202, "step": 17207 }, { "epoch": 1.1896301417213966, "grad_norm": 0.7169417633117697, "learning_rate": 3.7224032059735515e-06, "loss": 0.3333, "step": 17208 }, { "epoch": 1.1896992741099206, "grad_norm": 0.6994332635794213, "learning_rate": 3.7218620365606073e-06, "loss": 0.3488, "step": 17209 }, { "epoch": 1.1897684064984446, "grad_norm": 0.707934760743333, "learning_rate": 3.72132088316684e-06, "loss": 0.2936, "step": 17210 }, { "epoch": 1.1898375388869686, "grad_norm": 0.7980579808274818, "learning_rate": 3.7207797457990364e-06, "loss": 0.2546, "step": 17211 }, { "epoch": 1.1899066712754927, "grad_norm": 0.705741661123846, "learning_rate": 3.720238624463973e-06, "loss": 0.334, "step": 17212 }, { "epoch": 1.1899758036640167, "grad_norm": 0.6599975729462244, "learning_rate": 3.719697519168436e-06, "loss": 0.3337, "step": 17213 }, { "epoch": 1.1900449360525407, "grad_norm": 0.630020902984334, "learning_rate": 3.719156429919205e-06, "loss": 0.2414, "step": 17214 }, { "epoch": 1.1901140684410647, "grad_norm": 0.760026493843471, "learning_rate": 3.7186153567230642e-06, "loss": 0.2648, "step": 17215 }, { "epoch": 1.1901832008295887, "grad_norm": 0.6553525153397097, "learning_rate": 3.7180742995867915e-06, "loss": 0.3578, "step": 17216 }, { "epoch": 1.1902523332181127, "grad_norm": 0.6174184416927027, "learning_rate": 3.7175332585171697e-06, "loss": 0.2146, "step": 17217 }, { "epoch": 1.1903214656066368, "grad_norm": 0.6861599547117586, "learning_rate": 3.7169922335209797e-06, "loss": 0.2722, "step": 17218 }, { "epoch": 1.1903905979951608, "grad_norm": 0.719154463858991, "learning_rate": 3.7164512246050034e-06, "loss": 0.3061, "step": 17219 }, { "epoch": 1.1904597303836848, "grad_norm": 0.7870316291072292, "learning_rate": 3.715910231776019e-06, "loss": 0.2262, "step": 17220 }, { "epoch": 1.1905288627722088, "grad_norm": 0.6886303643047105, "learning_rate": 3.7153692550408105e-06, "loss": 0.2531, "step": 17221 }, { "epoch": 1.1905979951607328, "grad_norm": 0.6409079848520531, "learning_rate": 3.7148282944061543e-06, "loss": 0.2854, "step": 17222 }, { "epoch": 1.1906671275492569, "grad_norm": 0.6742529092312078, "learning_rate": 3.714287349878832e-06, "loss": 0.3347, "step": 17223 }, { "epoch": 1.1907362599377809, "grad_norm": 0.64369047244482, "learning_rate": 3.713746421465623e-06, "loss": 0.2638, "step": 17224 }, { "epoch": 1.190805392326305, "grad_norm": 0.615569208723071, "learning_rate": 3.7132055091733096e-06, "loss": 0.2323, "step": 17225 }, { "epoch": 1.190874524714829, "grad_norm": 0.8620825372981606, "learning_rate": 3.712664613008666e-06, "loss": 0.305, "step": 17226 }, { "epoch": 1.190943657103353, "grad_norm": 0.7199601659759989, "learning_rate": 3.712123732978475e-06, "loss": 0.3399, "step": 17227 }, { "epoch": 1.191012789491877, "grad_norm": 0.6264835325329228, "learning_rate": 3.7115828690895163e-06, "loss": 0.2482, "step": 17228 }, { "epoch": 1.191081921880401, "grad_norm": 0.6467332021050605, "learning_rate": 3.711042021348564e-06, "loss": 0.2589, "step": 17229 }, { "epoch": 1.191151054268925, "grad_norm": 0.7293556268503965, "learning_rate": 3.7105011897624023e-06, "loss": 0.3432, "step": 17230 }, { "epoch": 1.191220186657449, "grad_norm": 0.6820295355374293, "learning_rate": 3.709960374337805e-06, "loss": 0.2851, "step": 17231 }, { "epoch": 1.191289319045973, "grad_norm": 0.6011248563052788, "learning_rate": 3.7094195750815544e-06, "loss": 0.2318, "step": 17232 }, { "epoch": 1.191358451434497, "grad_norm": 0.6961681063727192, "learning_rate": 3.7088787920004243e-06, "loss": 0.2914, "step": 17233 }, { "epoch": 1.191427583823021, "grad_norm": 0.7071630474532478, "learning_rate": 3.7083380251011954e-06, "loss": 0.3261, "step": 17234 }, { "epoch": 1.191496716211545, "grad_norm": 0.6258798525533517, "learning_rate": 3.707797274390643e-06, "loss": 0.215, "step": 17235 }, { "epoch": 1.191565848600069, "grad_norm": 0.6648832846065633, "learning_rate": 3.707256539875548e-06, "loss": 0.2414, "step": 17236 }, { "epoch": 1.191634980988593, "grad_norm": 0.6768283232165296, "learning_rate": 3.706715821562682e-06, "loss": 0.3574, "step": 17237 }, { "epoch": 1.1917041133771171, "grad_norm": 0.7578089087154807, "learning_rate": 3.706175119458829e-06, "loss": 0.3313, "step": 17238 }, { "epoch": 1.1917732457656411, "grad_norm": 0.6825319438503711, "learning_rate": 3.7056344335707596e-06, "loss": 0.1984, "step": 17239 }, { "epoch": 1.1918423781541652, "grad_norm": 0.6733106380379116, "learning_rate": 3.705093763905253e-06, "loss": 0.3264, "step": 17240 }, { "epoch": 1.1919115105426892, "grad_norm": 0.7112618935199034, "learning_rate": 3.7045531104690847e-06, "loss": 0.3975, "step": 17241 }, { "epoch": 1.1919806429312132, "grad_norm": 0.6500471165593347, "learning_rate": 3.704012473269033e-06, "loss": 0.26, "step": 17242 }, { "epoch": 1.1920497753197372, "grad_norm": 0.8078787188235895, "learning_rate": 3.7034718523118705e-06, "loss": 0.2737, "step": 17243 }, { "epoch": 1.1921189077082612, "grad_norm": 0.6764625485931758, "learning_rate": 3.7029312476043755e-06, "loss": 0.3375, "step": 17244 }, { "epoch": 1.1921880400967853, "grad_norm": 0.6564714058350976, "learning_rate": 3.702390659153322e-06, "loss": 0.2833, "step": 17245 }, { "epoch": 1.1922571724853093, "grad_norm": 0.6038147453587414, "learning_rate": 3.701850086965487e-06, "loss": 0.2063, "step": 17246 }, { "epoch": 1.1923263048738333, "grad_norm": 0.6500941099398562, "learning_rate": 3.701309531047643e-06, "loss": 0.2719, "step": 17247 }, { "epoch": 1.1923954372623573, "grad_norm": 0.7141681991585858, "learning_rate": 3.7007689914065683e-06, "loss": 0.4343, "step": 17248 }, { "epoch": 1.1924645696508813, "grad_norm": 0.9790033612347923, "learning_rate": 3.7002284680490336e-06, "loss": 0.3229, "step": 17249 }, { "epoch": 1.1925337020394053, "grad_norm": 0.5747026894396792, "learning_rate": 3.699687960981818e-06, "loss": 0.2222, "step": 17250 }, { "epoch": 1.1926028344279296, "grad_norm": 0.7074929383580216, "learning_rate": 3.699147470211691e-06, "loss": 0.3277, "step": 17251 }, { "epoch": 1.1926719668164536, "grad_norm": 0.6992586354693564, "learning_rate": 3.6986069957454298e-06, "loss": 0.3228, "step": 17252 }, { "epoch": 1.1927410992049776, "grad_norm": 0.6024229329474026, "learning_rate": 3.6980665375898067e-06, "loss": 0.2498, "step": 17253 }, { "epoch": 1.1928102315935016, "grad_norm": 0.7130283430094064, "learning_rate": 3.697526095751598e-06, "loss": 0.2263, "step": 17254 }, { "epoch": 1.1928793639820257, "grad_norm": 1.1817794345471118, "learning_rate": 3.6969856702375727e-06, "loss": 0.3195, "step": 17255 }, { "epoch": 1.1929484963705497, "grad_norm": 0.6764814426058974, "learning_rate": 3.69644526105451e-06, "loss": 0.2917, "step": 17256 }, { "epoch": 1.1930176287590737, "grad_norm": 0.5896080834485051, "learning_rate": 3.695904868209177e-06, "loss": 0.2231, "step": 17257 }, { "epoch": 1.1930867611475977, "grad_norm": 0.6594998646562936, "learning_rate": 3.6953644917083498e-06, "loss": 0.2818, "step": 17258 }, { "epoch": 1.1931558935361217, "grad_norm": 0.6155286740311909, "learning_rate": 3.6948241315588e-06, "loss": 0.3268, "step": 17259 }, { "epoch": 1.1932250259246457, "grad_norm": 0.6925956205483877, "learning_rate": 3.6942837877673027e-06, "loss": 0.2702, "step": 17260 }, { "epoch": 1.1932941583131698, "grad_norm": 0.6342997285966312, "learning_rate": 3.693743460340626e-06, "loss": 0.251, "step": 17261 }, { "epoch": 1.1933632907016938, "grad_norm": 0.6479028724637831, "learning_rate": 3.6932031492855445e-06, "loss": 0.3042, "step": 17262 }, { "epoch": 1.1934324230902178, "grad_norm": 0.6984808330139898, "learning_rate": 3.692662854608828e-06, "loss": 0.2911, "step": 17263 }, { "epoch": 1.1935015554787418, "grad_norm": 0.6886779320994981, "learning_rate": 3.692122576317253e-06, "loss": 0.2777, "step": 17264 }, { "epoch": 1.1935706878672658, "grad_norm": 0.7232955904475793, "learning_rate": 3.6915823144175842e-06, "loss": 0.2714, "step": 17265 }, { "epoch": 1.1936398202557899, "grad_norm": 0.7302424950950944, "learning_rate": 3.691042068916598e-06, "loss": 0.3899, "step": 17266 }, { "epoch": 1.1937089526443139, "grad_norm": 0.7245253976934133, "learning_rate": 3.6905018398210626e-06, "loss": 0.2695, "step": 17267 }, { "epoch": 1.1937780850328379, "grad_norm": 0.6345242048857012, "learning_rate": 3.689961627137751e-06, "loss": 0.2407, "step": 17268 }, { "epoch": 1.193847217421362, "grad_norm": 0.6567247056671357, "learning_rate": 3.6894214308734305e-06, "loss": 0.3069, "step": 17269 }, { "epoch": 1.193916349809886, "grad_norm": 0.6124246787523984, "learning_rate": 3.6888812510348747e-06, "loss": 0.2503, "step": 17270 }, { "epoch": 1.19398548219841, "grad_norm": 0.6693971032056082, "learning_rate": 3.6883410876288516e-06, "loss": 0.2225, "step": 17271 }, { "epoch": 1.194054614586934, "grad_norm": 0.9152120789352961, "learning_rate": 3.687800940662134e-06, "loss": 0.2985, "step": 17272 }, { "epoch": 1.194123746975458, "grad_norm": 0.6642498713757821, "learning_rate": 3.687260810141488e-06, "loss": 0.3705, "step": 17273 }, { "epoch": 1.194192879363982, "grad_norm": 0.7298157915261817, "learning_rate": 3.6867206960736875e-06, "loss": 0.2711, "step": 17274 }, { "epoch": 1.194262011752506, "grad_norm": 0.7221965075024828, "learning_rate": 3.6861805984654976e-06, "loss": 0.2402, "step": 17275 }, { "epoch": 1.19433114414103, "grad_norm": 0.6159484501083894, "learning_rate": 3.6856405173236897e-06, "loss": 0.2534, "step": 17276 }, { "epoch": 1.194400276529554, "grad_norm": 0.6573792824272214, "learning_rate": 3.685100452655032e-06, "loss": 0.4055, "step": 17277 }, { "epoch": 1.194469408918078, "grad_norm": 0.6170319091978446, "learning_rate": 3.684560404466294e-06, "loss": 0.257, "step": 17278 }, { "epoch": 1.194538541306602, "grad_norm": 0.6581469787369068, "learning_rate": 3.684020372764243e-06, "loss": 0.2182, "step": 17279 }, { "epoch": 1.194607673695126, "grad_norm": 0.6641989895313455, "learning_rate": 3.6834803575556496e-06, "loss": 0.3135, "step": 17280 }, { "epoch": 1.1946768060836501, "grad_norm": 0.6727103240284596, "learning_rate": 3.682940358847279e-06, "loss": 0.2391, "step": 17281 }, { "epoch": 1.1947459384721741, "grad_norm": 0.6824123184333053, "learning_rate": 3.6824003766459026e-06, "loss": 0.261, "step": 17282 }, { "epoch": 1.1948150708606982, "grad_norm": 0.6908084735310612, "learning_rate": 3.681860410958284e-06, "loss": 0.3043, "step": 17283 }, { "epoch": 1.1948842032492222, "grad_norm": 0.6689899191753738, "learning_rate": 3.6813204617911942e-06, "loss": 0.3618, "step": 17284 }, { "epoch": 1.1949533356377462, "grad_norm": 0.664509528767142, "learning_rate": 3.680780529151398e-06, "loss": 0.2469, "step": 17285 }, { "epoch": 1.1950224680262702, "grad_norm": 0.6096009374978082, "learning_rate": 3.6802406130456654e-06, "loss": 0.2619, "step": 17286 }, { "epoch": 1.1950916004147945, "grad_norm": 0.6397054789137661, "learning_rate": 3.6797007134807594e-06, "loss": 0.3625, "step": 17287 }, { "epoch": 1.1951607328033185, "grad_norm": 0.6228346970078397, "learning_rate": 3.6791608304634497e-06, "loss": 0.296, "step": 17288 }, { "epoch": 1.1952298651918425, "grad_norm": 0.6731658665547682, "learning_rate": 3.678620964000501e-06, "loss": 0.2377, "step": 17289 }, { "epoch": 1.1952989975803665, "grad_norm": 0.5867190264113518, "learning_rate": 3.6780811140986818e-06, "loss": 0.2125, "step": 17290 }, { "epoch": 1.1953681299688905, "grad_norm": 0.7544191467944226, "learning_rate": 3.6775412807647553e-06, "loss": 0.3375, "step": 17291 }, { "epoch": 1.1954372623574145, "grad_norm": 0.6739252713139459, "learning_rate": 3.677001464005491e-06, "loss": 0.2609, "step": 17292 }, { "epoch": 1.1955063947459386, "grad_norm": 0.629810990122222, "learning_rate": 3.6764616638276505e-06, "loss": 0.2091, "step": 17293 }, { "epoch": 1.1955755271344626, "grad_norm": 0.71170898246386, "learning_rate": 3.6759218802380014e-06, "loss": 0.3552, "step": 17294 }, { "epoch": 1.1956446595229866, "grad_norm": 0.6379277588047413, "learning_rate": 3.675382113243311e-06, "loss": 0.2607, "step": 17295 }, { "epoch": 1.1957137919115106, "grad_norm": 0.8160381819609873, "learning_rate": 3.6748423628503386e-06, "loss": 0.2281, "step": 17296 }, { "epoch": 1.1957829243000346, "grad_norm": 0.6278330564520956, "learning_rate": 3.674302629065854e-06, "loss": 0.2453, "step": 17297 }, { "epoch": 1.1958520566885587, "grad_norm": 0.6814249190924287, "learning_rate": 3.673762911896619e-06, "loss": 0.3336, "step": 17298 }, { "epoch": 1.1959211890770827, "grad_norm": 0.6761349215584909, "learning_rate": 3.673223211349401e-06, "loss": 0.2668, "step": 17299 }, { "epoch": 1.1959903214656067, "grad_norm": 0.6476742124507291, "learning_rate": 3.6726835274309603e-06, "loss": 0.2405, "step": 17300 }, { "epoch": 1.1960594538541307, "grad_norm": 0.6895292888549499, "learning_rate": 3.672143860148064e-06, "loss": 0.3, "step": 17301 }, { "epoch": 1.1961285862426547, "grad_norm": 0.7001044195223995, "learning_rate": 3.671604209507473e-06, "loss": 0.3011, "step": 17302 }, { "epoch": 1.1961977186311787, "grad_norm": 0.80197689020979, "learning_rate": 3.6710645755159555e-06, "loss": 0.2653, "step": 17303 }, { "epoch": 1.1962668510197028, "grad_norm": 0.7270247226873351, "learning_rate": 3.6705249581802697e-06, "loss": 0.2409, "step": 17304 }, { "epoch": 1.1963359834082268, "grad_norm": 0.6949751249444568, "learning_rate": 3.669985357507182e-06, "loss": 0.3212, "step": 17305 }, { "epoch": 1.1964051157967508, "grad_norm": 0.6325526826915868, "learning_rate": 3.669445773503454e-06, "loss": 0.2438, "step": 17306 }, { "epoch": 1.1964742481852748, "grad_norm": 0.8692467550891645, "learning_rate": 3.6689062061758495e-06, "loss": 0.2365, "step": 17307 }, { "epoch": 1.1965433805737988, "grad_norm": 0.654260331921467, "learning_rate": 3.668366655531128e-06, "loss": 0.3151, "step": 17308 }, { "epoch": 1.1966125129623228, "grad_norm": 0.7334911161312091, "learning_rate": 3.6678271215760577e-06, "loss": 0.3292, "step": 17309 }, { "epoch": 1.1966816453508469, "grad_norm": 0.7458508219667607, "learning_rate": 3.6672876043173934e-06, "loss": 0.255, "step": 17310 }, { "epoch": 1.1967507777393709, "grad_norm": 0.6794594305694969, "learning_rate": 3.666748103761903e-06, "loss": 0.2725, "step": 17311 }, { "epoch": 1.196819910127895, "grad_norm": 0.6572404646958563, "learning_rate": 3.6662086199163437e-06, "loss": 0.3256, "step": 17312 }, { "epoch": 1.196889042516419, "grad_norm": 0.6416538810869453, "learning_rate": 3.665669152787482e-06, "loss": 0.2617, "step": 17313 }, { "epoch": 1.196958174904943, "grad_norm": 0.5914844723009195, "learning_rate": 3.6651297023820744e-06, "loss": 0.2151, "step": 17314 }, { "epoch": 1.197027307293467, "grad_norm": 0.6829518577618063, "learning_rate": 3.6645902687068845e-06, "loss": 0.2566, "step": 17315 }, { "epoch": 1.197096439681991, "grad_norm": 0.6837238021108052, "learning_rate": 3.6640508517686713e-06, "loss": 0.3411, "step": 17316 }, { "epoch": 1.197165572070515, "grad_norm": 0.765578723898554, "learning_rate": 3.663511451574199e-06, "loss": 0.2499, "step": 17317 }, { "epoch": 1.197234704459039, "grad_norm": 0.6444175868998444, "learning_rate": 3.6629720681302227e-06, "loss": 0.2669, "step": 17318 }, { "epoch": 1.197303836847563, "grad_norm": 0.7172899189646844, "learning_rate": 3.6624327014435067e-06, "loss": 0.3415, "step": 17319 }, { "epoch": 1.197372969236087, "grad_norm": 1.258842800838833, "learning_rate": 3.661893351520809e-06, "loss": 0.2789, "step": 17320 }, { "epoch": 1.197442101624611, "grad_norm": 0.6069906335939014, "learning_rate": 3.6613540183688922e-06, "loss": 0.2048, "step": 17321 }, { "epoch": 1.197511234013135, "grad_norm": 1.169026718566088, "learning_rate": 3.6608147019945124e-06, "loss": 0.2737, "step": 17322 }, { "epoch": 1.197580366401659, "grad_norm": 0.6413900951469415, "learning_rate": 3.6602754024044305e-06, "loss": 0.3231, "step": 17323 }, { "epoch": 1.1976494987901831, "grad_norm": 0.7010380631478388, "learning_rate": 3.659736119605405e-06, "loss": 0.2371, "step": 17324 }, { "epoch": 1.1977186311787071, "grad_norm": 0.5945870588400656, "learning_rate": 3.659196853604198e-06, "loss": 0.2468, "step": 17325 }, { "epoch": 1.1977877635672312, "grad_norm": 0.6366230412043657, "learning_rate": 3.6586576044075615e-06, "loss": 0.2602, "step": 17326 }, { "epoch": 1.1978568959557552, "grad_norm": 0.6318350925831495, "learning_rate": 3.6581183720222625e-06, "loss": 0.3392, "step": 17327 }, { "epoch": 1.1979260283442792, "grad_norm": 0.9043698962437002, "learning_rate": 3.657579156455052e-06, "loss": 0.2628, "step": 17328 }, { "epoch": 1.1979951607328032, "grad_norm": 0.7213293391130465, "learning_rate": 3.6570399577126926e-06, "loss": 0.2967, "step": 17329 }, { "epoch": 1.1980642931213272, "grad_norm": 0.7420612440496344, "learning_rate": 3.6565007758019395e-06, "loss": 0.3274, "step": 17330 }, { "epoch": 1.1981334255098512, "grad_norm": 0.6947961930573491, "learning_rate": 3.655961610729554e-06, "loss": 0.2904, "step": 17331 }, { "epoch": 1.1982025578983753, "grad_norm": 0.7522733314833729, "learning_rate": 3.655422462502288e-06, "loss": 0.2028, "step": 17332 }, { "epoch": 1.1982716902868995, "grad_norm": 0.613849017838223, "learning_rate": 3.654883331126904e-06, "loss": 0.216, "step": 17333 }, { "epoch": 1.1983408226754235, "grad_norm": 1.0877985762208635, "learning_rate": 3.654344216610156e-06, "loss": 0.3441, "step": 17334 }, { "epoch": 1.1984099550639475, "grad_norm": 0.6366539873558079, "learning_rate": 3.653805118958803e-06, "loss": 0.2564, "step": 17335 }, { "epoch": 1.1984790874524716, "grad_norm": 0.647628429319465, "learning_rate": 3.6532660381795994e-06, "loss": 0.2397, "step": 17336 }, { "epoch": 1.1985482198409956, "grad_norm": 0.6772059952424268, "learning_rate": 3.652726974279303e-06, "loss": 0.3403, "step": 17337 }, { "epoch": 1.1986173522295196, "grad_norm": 0.7155584472548823, "learning_rate": 3.652187927264669e-06, "loss": 0.301, "step": 17338 }, { "epoch": 1.1986864846180436, "grad_norm": 0.6777988294537166, "learning_rate": 3.651648897142456e-06, "loss": 0.2964, "step": 17339 }, { "epoch": 1.1987556170065676, "grad_norm": 0.6981916297109829, "learning_rate": 3.6511098839194158e-06, "loss": 0.2807, "step": 17340 }, { "epoch": 1.1988247493950916, "grad_norm": 0.6384385034182406, "learning_rate": 3.650570887602307e-06, "loss": 0.3908, "step": 17341 }, { "epoch": 1.1988938817836157, "grad_norm": 0.6836341404977394, "learning_rate": 3.650031908197883e-06, "loss": 0.2508, "step": 17342 }, { "epoch": 1.1989630141721397, "grad_norm": 0.7667458025864955, "learning_rate": 3.6494929457129007e-06, "loss": 0.2272, "step": 17343 }, { "epoch": 1.1990321465606637, "grad_norm": 0.7217621082501431, "learning_rate": 3.6489540001541135e-06, "loss": 0.3186, "step": 17344 }, { "epoch": 1.1991012789491877, "grad_norm": 0.7674422352479624, "learning_rate": 3.6484150715282786e-06, "loss": 0.323, "step": 17345 }, { "epoch": 1.1991704113377117, "grad_norm": 0.6329727087097854, "learning_rate": 3.647876159842147e-06, "loss": 0.2352, "step": 17346 }, { "epoch": 1.1992395437262358, "grad_norm": 0.7556833015263256, "learning_rate": 3.647337265102475e-06, "loss": 0.233, "step": 17347 }, { "epoch": 1.1993086761147598, "grad_norm": 0.6643171363810839, "learning_rate": 3.6467983873160167e-06, "loss": 0.3018, "step": 17348 }, { "epoch": 1.1993778085032838, "grad_norm": 0.7527865443425162, "learning_rate": 3.646259526489527e-06, "loss": 0.3227, "step": 17349 }, { "epoch": 1.1994469408918078, "grad_norm": 0.6364454486703394, "learning_rate": 3.6457206826297562e-06, "loss": 0.2521, "step": 17350 }, { "epoch": 1.1995160732803318, "grad_norm": 0.8736697024923078, "learning_rate": 3.6451818557434613e-06, "loss": 0.2966, "step": 17351 }, { "epoch": 1.1995852056688558, "grad_norm": 0.7319774497242809, "learning_rate": 3.644643045837393e-06, "loss": 0.3784, "step": 17352 }, { "epoch": 1.1996543380573799, "grad_norm": 0.6676516893167224, "learning_rate": 3.6441042529183067e-06, "loss": 0.3234, "step": 17353 }, { "epoch": 1.1997234704459039, "grad_norm": 0.6074658562350591, "learning_rate": 3.643565476992953e-06, "loss": 0.2247, "step": 17354 }, { "epoch": 1.199792602834428, "grad_norm": 0.6756620213923351, "learning_rate": 3.6430267180680856e-06, "loss": 0.3368, "step": 17355 }, { "epoch": 1.199861735222952, "grad_norm": 0.8775402376321668, "learning_rate": 3.6424879761504558e-06, "loss": 0.3274, "step": 17356 }, { "epoch": 1.199930867611476, "grad_norm": 0.6123941571724857, "learning_rate": 3.6419492512468185e-06, "loss": 0.1963, "step": 17357 }, { "epoch": 1.2, "grad_norm": 0.7069892748783071, "learning_rate": 3.641410543363922e-06, "loss": 0.233, "step": 17358 }, { "epoch": 1.200069132388524, "grad_norm": 0.6603481312110504, "learning_rate": 3.6408718525085206e-06, "loss": 0.2728, "step": 17359 }, { "epoch": 1.200138264777048, "grad_norm": 0.6192203653817002, "learning_rate": 3.6403331786873642e-06, "loss": 0.2307, "step": 17360 }, { "epoch": 1.200207397165572, "grad_norm": 0.6878625307750498, "learning_rate": 3.6397945219072064e-06, "loss": 0.2533, "step": 17361 }, { "epoch": 1.200276529554096, "grad_norm": 0.7606017941454705, "learning_rate": 3.639255882174797e-06, "loss": 0.3519, "step": 17362 }, { "epoch": 1.20034566194262, "grad_norm": 0.8308824233032215, "learning_rate": 3.6387172594968856e-06, "loss": 0.3283, "step": 17363 }, { "epoch": 1.200414794331144, "grad_norm": 0.6739630168591321, "learning_rate": 3.6381786538802245e-06, "loss": 0.237, "step": 17364 }, { "epoch": 1.200483926719668, "grad_norm": 0.6400870009626402, "learning_rate": 3.6376400653315622e-06, "loss": 0.2429, "step": 17365 }, { "epoch": 1.200553059108192, "grad_norm": 0.6486731386652652, "learning_rate": 3.637101493857653e-06, "loss": 0.3789, "step": 17366 }, { "epoch": 1.2006221914967161, "grad_norm": 0.7414476478773702, "learning_rate": 3.636562939465243e-06, "loss": 0.238, "step": 17367 }, { "epoch": 1.2006913238852404, "grad_norm": 0.6390579967591351, "learning_rate": 3.6360244021610834e-06, "loss": 0.235, "step": 17368 }, { "epoch": 1.2007604562737644, "grad_norm": 0.6944890655132662, "learning_rate": 3.6354858819519233e-06, "loss": 0.3892, "step": 17369 }, { "epoch": 1.2008295886622884, "grad_norm": 0.719444820954407, "learning_rate": 3.634947378844515e-06, "loss": 0.3638, "step": 17370 }, { "epoch": 1.2008987210508124, "grad_norm": 0.6272279630950376, "learning_rate": 3.6344088928456023e-06, "loss": 0.2686, "step": 17371 }, { "epoch": 1.2009678534393364, "grad_norm": 0.6591580495386546, "learning_rate": 3.633870423961939e-06, "loss": 0.2729, "step": 17372 }, { "epoch": 1.2010369858278604, "grad_norm": 0.7022142434564096, "learning_rate": 3.6333319722002703e-06, "loss": 0.4275, "step": 17373 }, { "epoch": 1.2011061182163845, "grad_norm": 0.6545664139681112, "learning_rate": 3.6327935375673485e-06, "loss": 0.2283, "step": 17374 }, { "epoch": 1.2011752506049085, "grad_norm": 0.6700101823422563, "learning_rate": 3.6322551200699185e-06, "loss": 0.2375, "step": 17375 }, { "epoch": 1.2012443829934325, "grad_norm": 0.7929913959444927, "learning_rate": 3.6317167197147294e-06, "loss": 0.27, "step": 17376 }, { "epoch": 1.2013135153819565, "grad_norm": 0.6835185302836979, "learning_rate": 3.6311783365085296e-06, "loss": 0.3637, "step": 17377 }, { "epoch": 1.2013826477704805, "grad_norm": 0.6722331879655142, "learning_rate": 3.6306399704580686e-06, "loss": 0.2426, "step": 17378 }, { "epoch": 1.2014517801590046, "grad_norm": 0.7016608342902457, "learning_rate": 3.630101621570088e-06, "loss": 0.2763, "step": 17379 }, { "epoch": 1.2015209125475286, "grad_norm": 0.7084361932862676, "learning_rate": 3.6295632898513423e-06, "loss": 0.397, "step": 17380 }, { "epoch": 1.2015900449360526, "grad_norm": 0.6761653911355796, "learning_rate": 3.6290249753085732e-06, "loss": 0.295, "step": 17381 }, { "epoch": 1.2016591773245766, "grad_norm": 0.6417195554245722, "learning_rate": 3.6284866779485307e-06, "loss": 0.2355, "step": 17382 }, { "epoch": 1.2017283097131006, "grad_norm": 0.6742454791422768, "learning_rate": 3.6279483977779584e-06, "loss": 0.2551, "step": 17383 }, { "epoch": 1.2017974421016246, "grad_norm": 0.6365137956845469, "learning_rate": 3.6274101348036067e-06, "loss": 0.2993, "step": 17384 }, { "epoch": 1.2018665744901487, "grad_norm": 0.6970246476512766, "learning_rate": 3.6268718890322173e-06, "loss": 0.2739, "step": 17385 }, { "epoch": 1.2019357068786727, "grad_norm": 0.6700035516239323, "learning_rate": 3.62633366047054e-06, "loss": 0.2876, "step": 17386 }, { "epoch": 1.2020048392671967, "grad_norm": 0.6335256432400997, "learning_rate": 3.6257954491253172e-06, "loss": 0.3541, "step": 17387 }, { "epoch": 1.2020739716557207, "grad_norm": 0.6638866968028773, "learning_rate": 3.6252572550032982e-06, "loss": 0.3017, "step": 17388 }, { "epoch": 1.2021431040442447, "grad_norm": 0.5801783703908877, "learning_rate": 3.624719078111224e-06, "loss": 0.1984, "step": 17389 }, { "epoch": 1.2022122364327688, "grad_norm": 0.6858349316141907, "learning_rate": 3.624180918455843e-06, "loss": 0.2525, "step": 17390 }, { "epoch": 1.2022813688212928, "grad_norm": 0.6619305996331063, "learning_rate": 3.6236427760438984e-06, "loss": 0.2858, "step": 17391 }, { "epoch": 1.2023505012098168, "grad_norm": 0.6336740203032314, "learning_rate": 3.6231046508821368e-06, "loss": 0.306, "step": 17392 }, { "epoch": 1.2024196335983408, "grad_norm": 0.6774064101613892, "learning_rate": 3.6225665429772992e-06, "loss": 0.2861, "step": 17393 }, { "epoch": 1.2024887659868648, "grad_norm": 0.6792912318549824, "learning_rate": 3.6220284523361333e-06, "loss": 0.2831, "step": 17394 }, { "epoch": 1.2025578983753888, "grad_norm": 0.6818769123029129, "learning_rate": 3.62149037896538e-06, "loss": 0.3367, "step": 17395 }, { "epoch": 1.2026270307639129, "grad_norm": 0.6500737880026326, "learning_rate": 3.6209523228717873e-06, "loss": 0.2493, "step": 17396 }, { "epoch": 1.2026961631524369, "grad_norm": 0.6270206343761882, "learning_rate": 3.6204142840620933e-06, "loss": 0.2543, "step": 17397 }, { "epoch": 1.202765295540961, "grad_norm": 0.6704385203409196, "learning_rate": 3.619876262543047e-06, "loss": 0.3705, "step": 17398 }, { "epoch": 1.202834427929485, "grad_norm": 0.7133391015382861, "learning_rate": 3.6193382583213866e-06, "loss": 0.2637, "step": 17399 }, { "epoch": 1.202903560318009, "grad_norm": 0.6356987289151886, "learning_rate": 3.6188002714038583e-06, "loss": 0.2339, "step": 17400 }, { "epoch": 1.202972692706533, "grad_norm": 0.771775878664655, "learning_rate": 3.618262301797203e-06, "loss": 0.2817, "step": 17401 }, { "epoch": 1.203041825095057, "grad_norm": 0.717564798173197, "learning_rate": 3.617724349508166e-06, "loss": 0.3458, "step": 17402 }, { "epoch": 1.203110957483581, "grad_norm": 0.6223264209644438, "learning_rate": 3.6171864145434855e-06, "loss": 0.2554, "step": 17403 }, { "epoch": 1.203180089872105, "grad_norm": 0.6158340153060224, "learning_rate": 3.6166484969099056e-06, "loss": 0.2339, "step": 17404 }, { "epoch": 1.203249222260629, "grad_norm": 0.6713102773976904, "learning_rate": 3.6161105966141687e-06, "loss": 0.3521, "step": 17405 }, { "epoch": 1.203318354649153, "grad_norm": 0.6660343245385234, "learning_rate": 3.615572713663017e-06, "loss": 0.2506, "step": 17406 }, { "epoch": 1.203387487037677, "grad_norm": 0.6723377413807925, "learning_rate": 3.615034848063188e-06, "loss": 0.2297, "step": 17407 }, { "epoch": 1.203456619426201, "grad_norm": 0.7142064110815001, "learning_rate": 3.6144969998214272e-06, "loss": 0.2936, "step": 17408 }, { "epoch": 1.203525751814725, "grad_norm": 0.6435487297588204, "learning_rate": 3.6139591689444732e-06, "loss": 0.336, "step": 17409 }, { "epoch": 1.2035948842032491, "grad_norm": 0.7407369246250651, "learning_rate": 3.6134213554390698e-06, "loss": 0.246, "step": 17410 }, { "epoch": 1.2036640165917731, "grad_norm": 0.6453102023344552, "learning_rate": 3.612883559311953e-06, "loss": 0.2385, "step": 17411 }, { "epoch": 1.2037331489802972, "grad_norm": 0.6982466134845142, "learning_rate": 3.6123457805698657e-06, "loss": 0.3335, "step": 17412 }, { "epoch": 1.2038022813688212, "grad_norm": 0.691620356408139, "learning_rate": 3.611808019219547e-06, "loss": 0.324, "step": 17413 }, { "epoch": 1.2038714137573454, "grad_norm": 0.6408816017344015, "learning_rate": 3.61127027526774e-06, "loss": 0.204, "step": 17414 }, { "epoch": 1.2039405461458694, "grad_norm": 0.7223578326721266, "learning_rate": 3.610732548721178e-06, "loss": 0.2992, "step": 17415 }, { "epoch": 1.2040096785343934, "grad_norm": 0.6826921124971729, "learning_rate": 3.6101948395866078e-06, "loss": 0.3941, "step": 17416 }, { "epoch": 1.2040788109229175, "grad_norm": 0.6166771098154735, "learning_rate": 3.6096571478707626e-06, "loss": 0.2483, "step": 17417 }, { "epoch": 1.2041479433114415, "grad_norm": 0.6250555243328986, "learning_rate": 3.6091194735803854e-06, "loss": 0.281, "step": 17418 }, { "epoch": 1.2042170756999655, "grad_norm": 1.0145639867281637, "learning_rate": 3.608581816722212e-06, "loss": 0.312, "step": 17419 }, { "epoch": 1.2042862080884895, "grad_norm": 0.7571253337446575, "learning_rate": 3.608044177302985e-06, "loss": 0.3705, "step": 17420 }, { "epoch": 1.2043553404770135, "grad_norm": 0.6249122574568899, "learning_rate": 3.6075065553294373e-06, "loss": 0.2389, "step": 17421 }, { "epoch": 1.2044244728655376, "grad_norm": 0.6052286804463216, "learning_rate": 3.606968950808312e-06, "loss": 0.2423, "step": 17422 }, { "epoch": 1.2044936052540616, "grad_norm": 0.7223516300569647, "learning_rate": 3.6064313637463434e-06, "loss": 0.3719, "step": 17423 }, { "epoch": 1.2045627376425856, "grad_norm": 0.6665735043762215, "learning_rate": 3.6058937941502723e-06, "loss": 0.2842, "step": 17424 }, { "epoch": 1.2046318700311096, "grad_norm": 0.6677710855455623, "learning_rate": 3.605356242026833e-06, "loss": 0.2335, "step": 17425 }, { "epoch": 1.2047010024196336, "grad_norm": 3.1730517213917877, "learning_rate": 3.6048187073827653e-06, "loss": 0.3148, "step": 17426 }, { "epoch": 1.2047701348081576, "grad_norm": 0.6911535360878356, "learning_rate": 3.6042811902248065e-06, "loss": 0.3573, "step": 17427 }, { "epoch": 1.2048392671966817, "grad_norm": 0.7087517088676473, "learning_rate": 3.60374369055969e-06, "loss": 0.239, "step": 17428 }, { "epoch": 1.2049083995852057, "grad_norm": 0.6624417586442992, "learning_rate": 3.6032062083941555e-06, "loss": 0.2606, "step": 17429 }, { "epoch": 1.2049775319737297, "grad_norm": 0.6412994762947194, "learning_rate": 3.602668743734937e-06, "loss": 0.3133, "step": 17430 }, { "epoch": 1.2050466643622537, "grad_norm": 0.6637976077137994, "learning_rate": 3.6021312965887754e-06, "loss": 0.2913, "step": 17431 }, { "epoch": 1.2051157967507777, "grad_norm": 0.6396561315204196, "learning_rate": 3.6015938669623996e-06, "loss": 0.2982, "step": 17432 }, { "epoch": 1.2051849291393018, "grad_norm": 0.6865584289624543, "learning_rate": 3.6010564548625526e-06, "loss": 0.2844, "step": 17433 }, { "epoch": 1.2052540615278258, "grad_norm": 0.6625672117998385, "learning_rate": 3.6005190602959637e-06, "loss": 0.3511, "step": 17434 }, { "epoch": 1.2053231939163498, "grad_norm": 0.745844432615706, "learning_rate": 3.5999816832693726e-06, "loss": 0.2274, "step": 17435 }, { "epoch": 1.2053923263048738, "grad_norm": 0.9003209412590967, "learning_rate": 3.599444323789511e-06, "loss": 0.2788, "step": 17436 }, { "epoch": 1.2054614586933978, "grad_norm": 0.7205194060132522, "learning_rate": 3.598906981863118e-06, "loss": 0.3197, "step": 17437 }, { "epoch": 1.2055305910819218, "grad_norm": 0.6498239849590128, "learning_rate": 3.598369657496923e-06, "loss": 0.2552, "step": 17438 }, { "epoch": 1.2055997234704459, "grad_norm": 0.6600428492782257, "learning_rate": 3.597832350697664e-06, "loss": 0.2518, "step": 17439 }, { "epoch": 1.2056688558589699, "grad_norm": 0.6696194187835637, "learning_rate": 3.5972950614720735e-06, "loss": 0.2502, "step": 17440 }, { "epoch": 1.205737988247494, "grad_norm": 0.7529687545024114, "learning_rate": 3.596757789826889e-06, "loss": 0.3939, "step": 17441 }, { "epoch": 1.205807120636018, "grad_norm": 0.6659703903292188, "learning_rate": 3.5962205357688383e-06, "loss": 0.235, "step": 17442 }, { "epoch": 1.205876253024542, "grad_norm": 0.6155755689970696, "learning_rate": 3.5956832993046598e-06, "loss": 0.2183, "step": 17443 }, { "epoch": 1.205945385413066, "grad_norm": 0.6762438082833794, "learning_rate": 3.595146080441083e-06, "loss": 0.3842, "step": 17444 }, { "epoch": 1.20601451780159, "grad_norm": 0.6482556764027583, "learning_rate": 3.5946088791848454e-06, "loss": 0.3183, "step": 17445 }, { "epoch": 1.206083650190114, "grad_norm": 0.6035372096990209, "learning_rate": 3.5940716955426757e-06, "loss": 0.2344, "step": 17446 }, { "epoch": 1.206152782578638, "grad_norm": 0.6160756881909477, "learning_rate": 3.5935345295213096e-06, "loss": 0.2407, "step": 17447 }, { "epoch": 1.206221914967162, "grad_norm": 0.767778512420169, "learning_rate": 3.5929973811274766e-06, "loss": 0.3408, "step": 17448 }, { "epoch": 1.2062910473556863, "grad_norm": 0.6904357131312515, "learning_rate": 3.592460250367913e-06, "loss": 0.2662, "step": 17449 }, { "epoch": 1.2063601797442103, "grad_norm": 0.6550190668622372, "learning_rate": 3.591923137249345e-06, "loss": 0.2453, "step": 17450 }, { "epoch": 1.2064293121327343, "grad_norm": 0.6506032758154929, "learning_rate": 3.5913860417785106e-06, "loss": 0.2996, "step": 17451 }, { "epoch": 1.2064984445212583, "grad_norm": 0.7209380804163418, "learning_rate": 3.590848963962137e-06, "loss": 0.3519, "step": 17452 }, { "epoch": 1.2065675769097823, "grad_norm": 0.6713857308364505, "learning_rate": 3.590311903806957e-06, "loss": 0.263, "step": 17453 }, { "epoch": 1.2066367092983064, "grad_norm": 0.8139831175958551, "learning_rate": 3.5897748613197013e-06, "loss": 0.2633, "step": 17454 }, { "epoch": 1.2067058416868304, "grad_norm": 0.6567279841892578, "learning_rate": 3.589237836507103e-06, "loss": 0.322, "step": 17455 }, { "epoch": 1.2067749740753544, "grad_norm": 0.6665369172308242, "learning_rate": 3.5887008293758886e-06, "loss": 0.2675, "step": 17456 }, { "epoch": 1.2068441064638784, "grad_norm": 0.6383735523785018, "learning_rate": 3.5881638399327912e-06, "loss": 0.2614, "step": 17457 }, { "epoch": 1.2069132388524024, "grad_norm": 0.6508058572485346, "learning_rate": 3.5876268681845403e-06, "loss": 0.257, "step": 17458 }, { "epoch": 1.2069823712409264, "grad_norm": 0.7275825307382641, "learning_rate": 3.5870899141378677e-06, "loss": 0.3369, "step": 17459 }, { "epoch": 1.2070515036294505, "grad_norm": 0.660128691260438, "learning_rate": 3.5865529777995e-06, "loss": 0.2853, "step": 17460 }, { "epoch": 1.2071206360179745, "grad_norm": 0.6379515595894218, "learning_rate": 3.586016059176169e-06, "loss": 0.2926, "step": 17461 }, { "epoch": 1.2071897684064985, "grad_norm": 0.690425060253271, "learning_rate": 3.585479158274602e-06, "loss": 0.3457, "step": 17462 }, { "epoch": 1.2072589007950225, "grad_norm": 0.6892480634168717, "learning_rate": 3.5849422751015316e-06, "loss": 0.3039, "step": 17463 }, { "epoch": 1.2073280331835465, "grad_norm": 0.6237615777218992, "learning_rate": 3.5844054096636833e-06, "loss": 0.2088, "step": 17464 }, { "epoch": 1.2073971655720706, "grad_norm": 0.6886324376209667, "learning_rate": 3.583868561967787e-06, "loss": 0.2889, "step": 17465 }, { "epoch": 1.2074662979605946, "grad_norm": 0.6674534195218603, "learning_rate": 3.5833317320205706e-06, "loss": 0.3862, "step": 17466 }, { "epoch": 1.2075354303491186, "grad_norm": 0.6068800925768763, "learning_rate": 3.5827949198287647e-06, "loss": 0.2475, "step": 17467 }, { "epoch": 1.2076045627376426, "grad_norm": 0.6953571437484459, "learning_rate": 3.582258125399092e-06, "loss": 0.2851, "step": 17468 }, { "epoch": 1.2076736951261666, "grad_norm": 0.6483642392151473, "learning_rate": 3.5817213487382874e-06, "loss": 0.3341, "step": 17469 }, { "epoch": 1.2077428275146906, "grad_norm": 0.6763022317965716, "learning_rate": 3.581184589853072e-06, "loss": 0.3318, "step": 17470 }, { "epoch": 1.2078119599032147, "grad_norm": 0.783743104116102, "learning_rate": 3.580647848750177e-06, "loss": 0.2342, "step": 17471 }, { "epoch": 1.2078810922917387, "grad_norm": 0.7086253768985615, "learning_rate": 3.5801111254363264e-06, "loss": 0.2769, "step": 17472 }, { "epoch": 1.2079502246802627, "grad_norm": 0.6593825726732894, "learning_rate": 3.5795744199182524e-06, "loss": 0.3201, "step": 17473 }, { "epoch": 1.2080193570687867, "grad_norm": 1.6998651568040448, "learning_rate": 3.5790377322026747e-06, "loss": 0.3007, "step": 17474 }, { "epoch": 1.2080884894573107, "grad_norm": 0.6228705070069148, "learning_rate": 3.5785010622963255e-06, "loss": 0.2281, "step": 17475 }, { "epoch": 1.2081576218458348, "grad_norm": 0.6632008258549176, "learning_rate": 3.5779644102059274e-06, "loss": 0.285, "step": 17476 }, { "epoch": 1.2082267542343588, "grad_norm": 0.6933624397670577, "learning_rate": 3.577427775938209e-06, "loss": 0.3168, "step": 17477 }, { "epoch": 1.2082958866228828, "grad_norm": 0.602440791737441, "learning_rate": 3.5768911594998932e-06, "loss": 0.2743, "step": 17478 }, { "epoch": 1.2083650190114068, "grad_norm": 0.6801056161748767, "learning_rate": 3.5763545608977085e-06, "loss": 0.2824, "step": 17479 }, { "epoch": 1.2084341513999308, "grad_norm": 0.7331936410363545, "learning_rate": 3.5758179801383774e-06, "loss": 0.3655, "step": 17480 }, { "epoch": 1.2085032837884548, "grad_norm": 0.7556601141365592, "learning_rate": 3.5752814172286283e-06, "loss": 0.2382, "step": 17481 }, { "epoch": 1.2085724161769789, "grad_norm": 0.6590583890385756, "learning_rate": 3.574744872175182e-06, "loss": 0.2575, "step": 17482 }, { "epoch": 1.2086415485655029, "grad_norm": 0.6501510677733832, "learning_rate": 3.5742083449847662e-06, "loss": 0.3014, "step": 17483 }, { "epoch": 1.208710680954027, "grad_norm": 0.656555350105037, "learning_rate": 3.5736718356641037e-06, "loss": 0.2934, "step": 17484 }, { "epoch": 1.208779813342551, "grad_norm": 0.6272470715858118, "learning_rate": 3.573135344219921e-06, "loss": 0.2591, "step": 17485 }, { "epoch": 1.208848945731075, "grad_norm": 0.6564379695958288, "learning_rate": 3.5725988706589377e-06, "loss": 0.2138, "step": 17486 }, { "epoch": 1.208918078119599, "grad_norm": 0.6994503304841592, "learning_rate": 3.5720624149878834e-06, "loss": 0.3071, "step": 17487 }, { "epoch": 1.208987210508123, "grad_norm": 0.6547301942301623, "learning_rate": 3.5715259772134757e-06, "loss": 0.3382, "step": 17488 }, { "epoch": 1.209056342896647, "grad_norm": 0.6536218402406503, "learning_rate": 3.5709895573424426e-06, "loss": 0.2193, "step": 17489 }, { "epoch": 1.209125475285171, "grad_norm": 0.5970045316610891, "learning_rate": 3.5704531553815036e-06, "loss": 0.1925, "step": 17490 }, { "epoch": 1.209194607673695, "grad_norm": 0.796534255995206, "learning_rate": 3.569916771337385e-06, "loss": 0.3743, "step": 17491 }, { "epoch": 1.209263740062219, "grad_norm": 0.6399176583502246, "learning_rate": 3.5693804052168057e-06, "loss": 0.2586, "step": 17492 }, { "epoch": 1.209332872450743, "grad_norm": 0.5961346806091384, "learning_rate": 3.5688440570264915e-06, "loss": 0.2342, "step": 17493 }, { "epoch": 1.209402004839267, "grad_norm": 0.6491208438098425, "learning_rate": 3.5683077267731637e-06, "loss": 0.2983, "step": 17494 }, { "epoch": 1.2094711372277913, "grad_norm": 0.6652262083457466, "learning_rate": 3.567771414463541e-06, "loss": 0.2839, "step": 17495 }, { "epoch": 1.2095402696163153, "grad_norm": 0.6574836500619118, "learning_rate": 3.5672351201043496e-06, "loss": 0.2388, "step": 17496 }, { "epoch": 1.2096094020048394, "grad_norm": 0.6805033870294955, "learning_rate": 3.5666988437023074e-06, "loss": 0.3227, "step": 17497 }, { "epoch": 1.2096785343933634, "grad_norm": 0.6995946436197675, "learning_rate": 3.56616258526414e-06, "loss": 0.3577, "step": 17498 }, { "epoch": 1.2097476667818874, "grad_norm": 0.650362635924249, "learning_rate": 3.5656263447965634e-06, "loss": 0.3081, "step": 17499 }, { "epoch": 1.2098167991704114, "grad_norm": 0.5913563269588636, "learning_rate": 3.5650901223063023e-06, "loss": 0.1971, "step": 17500 }, { "epoch": 1.2098859315589354, "grad_norm": 0.7117852868459568, "learning_rate": 3.5645539178000744e-06, "loss": 0.3239, "step": 17501 }, { "epoch": 1.2099550639474594, "grad_norm": 0.738458336904526, "learning_rate": 3.5640177312846036e-06, "loss": 0.3968, "step": 17502 }, { "epoch": 1.2100241963359835, "grad_norm": 0.6685899637292763, "learning_rate": 3.5634815627666043e-06, "loss": 0.2409, "step": 17503 }, { "epoch": 1.2100933287245075, "grad_norm": 0.6415186472330023, "learning_rate": 3.562945412252804e-06, "loss": 0.2694, "step": 17504 }, { "epoch": 1.2101624611130315, "grad_norm": 0.6803172855079197, "learning_rate": 3.5624092797499154e-06, "loss": 0.324, "step": 17505 }, { "epoch": 1.2102315935015555, "grad_norm": 0.706331374734403, "learning_rate": 3.5618731652646628e-06, "loss": 0.2405, "step": 17506 }, { "epoch": 1.2103007258900795, "grad_norm": 0.8273285001926051, "learning_rate": 3.5613370688037617e-06, "loss": 0.2381, "step": 17507 }, { "epoch": 1.2103698582786035, "grad_norm": 0.6898143164352771, "learning_rate": 3.560800990373936e-06, "loss": 0.2807, "step": 17508 }, { "epoch": 1.2104389906671276, "grad_norm": 0.7909971645155645, "learning_rate": 3.5602649299818994e-06, "loss": 0.3791, "step": 17509 }, { "epoch": 1.2105081230556516, "grad_norm": 0.8026059471312859, "learning_rate": 3.5597288876343733e-06, "loss": 0.2674, "step": 17510 }, { "epoch": 1.2105772554441756, "grad_norm": 0.717348683064166, "learning_rate": 3.5591928633380748e-06, "loss": 0.2568, "step": 17511 }, { "epoch": 1.2106463878326996, "grad_norm": 0.7536316710448163, "learning_rate": 3.5586568570997237e-06, "loss": 0.3388, "step": 17512 }, { "epoch": 1.2107155202212236, "grad_norm": 0.6831686735723949, "learning_rate": 3.558120868926035e-06, "loss": 0.2662, "step": 17513 }, { "epoch": 1.2107846526097477, "grad_norm": 0.7870593446423706, "learning_rate": 3.5575848988237295e-06, "loss": 0.2744, "step": 17514 }, { "epoch": 1.2108537849982717, "grad_norm": 0.61295543108702, "learning_rate": 3.557048946799522e-06, "loss": 0.2456, "step": 17515 }, { "epoch": 1.2109229173867957, "grad_norm": 0.6742163062369794, "learning_rate": 3.5565130128601322e-06, "loss": 0.3836, "step": 17516 }, { "epoch": 1.2109920497753197, "grad_norm": 0.6280876367313736, "learning_rate": 3.555977097012274e-06, "loss": 0.2443, "step": 17517 }, { "epoch": 1.2110611821638437, "grad_norm": 0.5673341911487516, "learning_rate": 3.5554411992626682e-06, "loss": 0.1999, "step": 17518 }, { "epoch": 1.2111303145523677, "grad_norm": 0.6490588425481189, "learning_rate": 3.554905319618026e-06, "loss": 0.2475, "step": 17519 }, { "epoch": 1.2111994469408918, "grad_norm": 0.599368681997635, "learning_rate": 3.5543694580850707e-06, "loss": 0.3205, "step": 17520 }, { "epoch": 1.2112685793294158, "grad_norm": 0.7553743813722341, "learning_rate": 3.55383361467051e-06, "loss": 0.2934, "step": 17521 }, { "epoch": 1.2113377117179398, "grad_norm": 0.6807834762723667, "learning_rate": 3.5532977893810677e-06, "loss": 0.291, "step": 17522 }, { "epoch": 1.2114068441064638, "grad_norm": 0.7509949774033896, "learning_rate": 3.5527619822234537e-06, "loss": 0.2671, "step": 17523 }, { "epoch": 1.2114759764949878, "grad_norm": 0.6478825008746723, "learning_rate": 3.5522261932043866e-06, "loss": 0.2555, "step": 17524 }, { "epoch": 1.2115451088835119, "grad_norm": 0.6453754037875798, "learning_rate": 3.551690422330579e-06, "loss": 0.2712, "step": 17525 }, { "epoch": 1.2116142412720359, "grad_norm": 0.7221051840378229, "learning_rate": 3.5511546696087505e-06, "loss": 0.3, "step": 17526 }, { "epoch": 1.21168337366056, "grad_norm": 0.9081919195409912, "learning_rate": 3.55061893504561e-06, "loss": 0.3455, "step": 17527 }, { "epoch": 1.211752506049084, "grad_norm": 0.629091212678099, "learning_rate": 3.5500832186478756e-06, "loss": 0.2134, "step": 17528 }, { "epoch": 1.211821638437608, "grad_norm": 0.6624324775324554, "learning_rate": 3.5495475204222596e-06, "loss": 0.2399, "step": 17529 }, { "epoch": 1.211890770826132, "grad_norm": 0.6853564356994667, "learning_rate": 3.549011840375479e-06, "loss": 0.346, "step": 17530 }, { "epoch": 1.2119599032146562, "grad_norm": 0.7130322288401437, "learning_rate": 3.5484761785142428e-06, "loss": 0.3455, "step": 17531 }, { "epoch": 1.2120290356031802, "grad_norm": 0.6756700408496047, "learning_rate": 3.54794053484527e-06, "loss": 0.273, "step": 17532 }, { "epoch": 1.2120981679917042, "grad_norm": 0.6349527524423032, "learning_rate": 3.5474049093752687e-06, "loss": 0.2353, "step": 17533 }, { "epoch": 1.2121673003802282, "grad_norm": 0.668766327838741, "learning_rate": 3.546869302110957e-06, "loss": 0.3716, "step": 17534 }, { "epoch": 1.2122364327687523, "grad_norm": 0.5945303564866089, "learning_rate": 3.546333713059044e-06, "loss": 0.2886, "step": 17535 }, { "epoch": 1.2123055651572763, "grad_norm": 0.6907468436162009, "learning_rate": 3.5457981422262444e-06, "loss": 0.2477, "step": 17536 }, { "epoch": 1.2123746975458003, "grad_norm": 0.6754527725714008, "learning_rate": 3.5452625896192686e-06, "loss": 0.3265, "step": 17537 }, { "epoch": 1.2124438299343243, "grad_norm": 0.6777250704802958, "learning_rate": 3.5447270552448325e-06, "loss": 0.3143, "step": 17538 }, { "epoch": 1.2125129623228483, "grad_norm": 0.6009691757905945, "learning_rate": 3.5441915391096427e-06, "loss": 0.212, "step": 17539 }, { "epoch": 1.2125820947113723, "grad_norm": 0.7308555444117861, "learning_rate": 3.5436560412204168e-06, "loss": 0.3204, "step": 17540 }, { "epoch": 1.2126512270998964, "grad_norm": 0.6732476675683396, "learning_rate": 3.5431205615838615e-06, "loss": 0.3482, "step": 17541 }, { "epoch": 1.2127203594884204, "grad_norm": 0.7204199048733154, "learning_rate": 3.542585100206691e-06, "loss": 0.3033, "step": 17542 }, { "epoch": 1.2127894918769444, "grad_norm": 0.6412061287181043, "learning_rate": 3.542049657095615e-06, "loss": 0.2216, "step": 17543 }, { "epoch": 1.2128586242654684, "grad_norm": 0.6526983841181352, "learning_rate": 3.541514232257346e-06, "loss": 0.3064, "step": 17544 }, { "epoch": 1.2129277566539924, "grad_norm": 0.6751401804626748, "learning_rate": 3.540978825698591e-06, "loss": 0.3288, "step": 17545 }, { "epoch": 1.2129968890425165, "grad_norm": 0.683365539893335, "learning_rate": 3.5404434374260643e-06, "loss": 0.2856, "step": 17546 }, { "epoch": 1.2130660214310405, "grad_norm": 0.605492657768372, "learning_rate": 3.5399080674464735e-06, "loss": 0.2562, "step": 17547 }, { "epoch": 1.2131351538195645, "grad_norm": 0.686320379417386, "learning_rate": 3.5393727157665315e-06, "loss": 0.3435, "step": 17548 }, { "epoch": 1.2132042862080885, "grad_norm": 0.6140976946828244, "learning_rate": 3.5388373823929434e-06, "loss": 0.2804, "step": 17549 }, { "epoch": 1.2132734185966125, "grad_norm": 0.6170482792702737, "learning_rate": 3.5383020673324224e-06, "loss": 0.2104, "step": 17550 }, { "epoch": 1.2133425509851365, "grad_norm": 0.7491057546413791, "learning_rate": 3.537766770591675e-06, "loss": 0.2773, "step": 17551 }, { "epoch": 1.2134116833736606, "grad_norm": 0.7241442666634617, "learning_rate": 3.537231492177414e-06, "loss": 0.3327, "step": 17552 }, { "epoch": 1.2134808157621846, "grad_norm": 0.5902609177208344, "learning_rate": 3.5366962320963443e-06, "loss": 0.2435, "step": 17553 }, { "epoch": 1.2135499481507086, "grad_norm": 0.6618130520998633, "learning_rate": 3.536160990355177e-06, "loss": 0.2931, "step": 17554 }, { "epoch": 1.2136190805392326, "grad_norm": 0.6258982067642389, "learning_rate": 3.5356257669606177e-06, "loss": 0.3151, "step": 17555 }, { "epoch": 1.2136882129277566, "grad_norm": 0.6486706753729343, "learning_rate": 3.5350905619193787e-06, "loss": 0.2521, "step": 17556 }, { "epoch": 1.2137573453162807, "grad_norm": 0.6763015391492339, "learning_rate": 3.534555375238161e-06, "loss": 0.2561, "step": 17557 }, { "epoch": 1.2138264777048047, "grad_norm": 0.6876655112188078, "learning_rate": 3.5340202069236807e-06, "loss": 0.3214, "step": 17558 }, { "epoch": 1.2138956100933287, "grad_norm": 0.6592090455485107, "learning_rate": 3.5334850569826384e-06, "loss": 0.3415, "step": 17559 }, { "epoch": 1.2139647424818527, "grad_norm": 0.7288664470946666, "learning_rate": 3.532949925421745e-06, "loss": 0.234, "step": 17560 }, { "epoch": 1.2140338748703767, "grad_norm": 0.6027817342692773, "learning_rate": 3.5324148122477074e-06, "loss": 0.2444, "step": 17561 }, { "epoch": 1.2141030072589007, "grad_norm": 0.6864576395638305, "learning_rate": 3.5318797174672293e-06, "loss": 0.3318, "step": 17562 }, { "epoch": 1.2141721396474248, "grad_norm": 0.6930170243675835, "learning_rate": 3.5313446410870195e-06, "loss": 0.2764, "step": 17563 }, { "epoch": 1.2142412720359488, "grad_norm": 0.6813825133649427, "learning_rate": 3.5308095831137822e-06, "loss": 0.3141, "step": 17564 }, { "epoch": 1.2143104044244728, "grad_norm": 0.6886041832297124, "learning_rate": 3.530274543554228e-06, "loss": 0.2778, "step": 17565 }, { "epoch": 1.2143795368129968, "grad_norm": 0.7354512765024916, "learning_rate": 3.529739522415057e-06, "loss": 0.3417, "step": 17566 }, { "epoch": 1.2144486692015208, "grad_norm": 0.7603500852510874, "learning_rate": 3.5292045197029785e-06, "loss": 0.2435, "step": 17567 }, { "epoch": 1.2145178015900449, "grad_norm": 0.6571017666758017, "learning_rate": 3.5286695354246954e-06, "loss": 0.2364, "step": 17568 }, { "epoch": 1.2145869339785689, "grad_norm": 0.670589422099305, "learning_rate": 3.5281345695869164e-06, "loss": 0.3314, "step": 17569 }, { "epoch": 1.214656066367093, "grad_norm": 0.6634065075903082, "learning_rate": 3.527599622196341e-06, "loss": 0.2978, "step": 17570 }, { "epoch": 1.214725198755617, "grad_norm": 0.6289290509816562, "learning_rate": 3.527064693259678e-06, "loss": 0.2321, "step": 17571 }, { "epoch": 1.214794331144141, "grad_norm": 0.6454966606039427, "learning_rate": 3.52652978278363e-06, "loss": 0.2592, "step": 17572 }, { "epoch": 1.214863463532665, "grad_norm": 0.6835179751452536, "learning_rate": 3.525994890774903e-06, "loss": 0.3407, "step": 17573 }, { "epoch": 1.214932595921189, "grad_norm": 0.7345112758535453, "learning_rate": 3.5254600172401983e-06, "loss": 0.2685, "step": 17574 }, { "epoch": 1.215001728309713, "grad_norm": 0.7030562524037703, "learning_rate": 3.5249251621862214e-06, "loss": 0.2445, "step": 17575 }, { "epoch": 1.2150708606982372, "grad_norm": 0.6474785944785001, "learning_rate": 3.5243903256196745e-06, "loss": 0.3303, "step": 17576 }, { "epoch": 1.2151399930867612, "grad_norm": 0.6490022978263826, "learning_rate": 3.523855507547262e-06, "loss": 0.3115, "step": 17577 }, { "epoch": 1.2152091254752853, "grad_norm": 0.6040558350002663, "learning_rate": 3.5233207079756854e-06, "loss": 0.2364, "step": 17578 }, { "epoch": 1.2152782578638093, "grad_norm": 0.7874247799947418, "learning_rate": 3.522785926911651e-06, "loss": 0.2751, "step": 17579 }, { "epoch": 1.2153473902523333, "grad_norm": 0.7679040029531832, "learning_rate": 3.5222511643618563e-06, "loss": 0.3308, "step": 17580 }, { "epoch": 1.2154165226408573, "grad_norm": 1.0807685820737978, "learning_rate": 3.521716420333008e-06, "loss": 0.2919, "step": 17581 }, { "epoch": 1.2154856550293813, "grad_norm": 0.6384391110712212, "learning_rate": 3.5211816948318043e-06, "loss": 0.2538, "step": 17582 }, { "epoch": 1.2155547874179053, "grad_norm": 0.6411829152481033, "learning_rate": 3.520646987864951e-06, "loss": 0.281, "step": 17583 }, { "epoch": 1.2156239198064294, "grad_norm": 0.6226223156327351, "learning_rate": 3.520112299439146e-06, "loss": 0.3446, "step": 17584 }, { "epoch": 1.2156930521949534, "grad_norm": 0.6722554026390258, "learning_rate": 3.5195776295610935e-06, "loss": 0.2539, "step": 17585 }, { "epoch": 1.2157621845834774, "grad_norm": 0.6640882510718444, "learning_rate": 3.5190429782374923e-06, "loss": 0.1899, "step": 17586 }, { "epoch": 1.2158313169720014, "grad_norm": 0.677374099309749, "learning_rate": 3.518508345475047e-06, "loss": 0.33, "step": 17587 }, { "epoch": 1.2159004493605254, "grad_norm": 0.7192075856808197, "learning_rate": 3.5179737312804527e-06, "loss": 0.2947, "step": 17588 }, { "epoch": 1.2159695817490495, "grad_norm": 0.597009167374929, "learning_rate": 3.5174391356604144e-06, "loss": 0.263, "step": 17589 }, { "epoch": 1.2160387141375735, "grad_norm": 0.7066701269853, "learning_rate": 3.51690455862163e-06, "loss": 0.283, "step": 17590 }, { "epoch": 1.2161078465260975, "grad_norm": 0.6858660038466112, "learning_rate": 3.5163700001708024e-06, "loss": 0.3432, "step": 17591 }, { "epoch": 1.2161769789146215, "grad_norm": 0.6285549505059852, "learning_rate": 3.515835460314625e-06, "loss": 0.2609, "step": 17592 }, { "epoch": 1.2162461113031455, "grad_norm": 0.5871240714258126, "learning_rate": 3.5153009390598058e-06, "loss": 0.2308, "step": 17593 }, { "epoch": 1.2163152436916695, "grad_norm": 0.6746579452631303, "learning_rate": 3.5147664364130374e-06, "loss": 0.3171, "step": 17594 }, { "epoch": 1.2163843760801936, "grad_norm": 0.7163926545082798, "learning_rate": 3.514231952381022e-06, "loss": 0.2918, "step": 17595 }, { "epoch": 1.2164535084687176, "grad_norm": 1.1263912377172267, "learning_rate": 3.5136974869704566e-06, "loss": 0.2778, "step": 17596 }, { "epoch": 1.2165226408572416, "grad_norm": 0.6702521390782802, "learning_rate": 3.5131630401880433e-06, "loss": 0.2873, "step": 17597 }, { "epoch": 1.2165917732457656, "grad_norm": 0.6795901592749537, "learning_rate": 3.512628612040475e-06, "loss": 0.4139, "step": 17598 }, { "epoch": 1.2166609056342896, "grad_norm": 0.6327246555810573, "learning_rate": 3.5120942025344546e-06, "loss": 0.263, "step": 17599 }, { "epoch": 1.2167300380228137, "grad_norm": 0.6731425140843376, "learning_rate": 3.511559811676677e-06, "loss": 0.2792, "step": 17600 }, { "epoch": 1.2167991704113377, "grad_norm": 0.7034566474677056, "learning_rate": 3.511025439473842e-06, "loss": 0.2814, "step": 17601 }, { "epoch": 1.2168683027998617, "grad_norm": 0.6202721243136564, "learning_rate": 3.5104910859326445e-06, "loss": 0.2892, "step": 17602 }, { "epoch": 1.2169374351883857, "grad_norm": 0.5837261696475453, "learning_rate": 3.509956751059784e-06, "loss": 0.193, "step": 17603 }, { "epoch": 1.2170065675769097, "grad_norm": 0.7203280932355004, "learning_rate": 3.509422434861956e-06, "loss": 0.2639, "step": 17604 }, { "epoch": 1.2170756999654337, "grad_norm": 0.6389312666594145, "learning_rate": 3.5088881373458593e-06, "loss": 0.337, "step": 17605 }, { "epoch": 1.2171448323539578, "grad_norm": 0.6712566578905604, "learning_rate": 3.508353858518187e-06, "loss": 0.2494, "step": 17606 }, { "epoch": 1.2172139647424818, "grad_norm": 0.6906065058044096, "learning_rate": 3.507819598385638e-06, "loss": 0.2762, "step": 17607 }, { "epoch": 1.2172830971310058, "grad_norm": 0.670669857440186, "learning_rate": 3.5072853569549065e-06, "loss": 0.2512, "step": 17608 }, { "epoch": 1.2173522295195298, "grad_norm": 0.6512741552927502, "learning_rate": 3.506751134232691e-06, "loss": 0.2952, "step": 17609 }, { "epoch": 1.2174213619080538, "grad_norm": 0.662067134623918, "learning_rate": 3.506216930225682e-06, "loss": 0.2269, "step": 17610 }, { "epoch": 1.2174904942965779, "grad_norm": 0.6082202997262386, "learning_rate": 3.5056827449405816e-06, "loss": 0.2533, "step": 17611 }, { "epoch": 1.217559626685102, "grad_norm": 0.6623364314661948, "learning_rate": 3.505148578384079e-06, "loss": 0.2987, "step": 17612 }, { "epoch": 1.217628759073626, "grad_norm": 0.685690665152764, "learning_rate": 3.5046144305628728e-06, "loss": 0.3181, "step": 17613 }, { "epoch": 1.2176978914621501, "grad_norm": 0.7505309985109999, "learning_rate": 3.5040803014836546e-06, "loss": 0.2232, "step": 17614 }, { "epoch": 1.2177670238506741, "grad_norm": 0.6728832716804831, "learning_rate": 3.503546191153122e-06, "loss": 0.282, "step": 17615 }, { "epoch": 1.2178361562391982, "grad_norm": 0.6297171782722308, "learning_rate": 3.503012099577965e-06, "loss": 0.3435, "step": 17616 }, { "epoch": 1.2179052886277222, "grad_norm": 0.5992621397805933, "learning_rate": 3.502478026764882e-06, "loss": 0.2315, "step": 17617 }, { "epoch": 1.2179744210162462, "grad_norm": 0.6020676559101634, "learning_rate": 3.501943972720562e-06, "loss": 0.23, "step": 17618 }, { "epoch": 1.2180435534047702, "grad_norm": 0.6440382391907288, "learning_rate": 3.501409937451704e-06, "loss": 0.3184, "step": 17619 }, { "epoch": 1.2181126857932942, "grad_norm": 0.6470424223661077, "learning_rate": 3.5008759209649964e-06, "loss": 0.3271, "step": 17620 }, { "epoch": 1.2181818181818183, "grad_norm": 0.5766559471514825, "learning_rate": 3.500341923267134e-06, "loss": 0.245, "step": 17621 }, { "epoch": 1.2182509505703423, "grad_norm": 0.6389106036966513, "learning_rate": 3.499807944364808e-06, "loss": 0.2437, "step": 17622 }, { "epoch": 1.2183200829588663, "grad_norm": 0.7037808466213361, "learning_rate": 3.499273984264715e-06, "loss": 0.3767, "step": 17623 }, { "epoch": 1.2183892153473903, "grad_norm": 0.5955127079947826, "learning_rate": 3.4987400429735422e-06, "loss": 0.2474, "step": 17624 }, { "epoch": 1.2184583477359143, "grad_norm": 0.5951370252281268, "learning_rate": 3.4982061204979855e-06, "loss": 0.2412, "step": 17625 }, { "epoch": 1.2185274801244383, "grad_norm": 0.6357866986061645, "learning_rate": 3.497672216844733e-06, "loss": 0.2701, "step": 17626 }, { "epoch": 1.2185966125129624, "grad_norm": 0.6991930842322929, "learning_rate": 3.497138332020481e-06, "loss": 0.3179, "step": 17627 }, { "epoch": 1.2186657449014864, "grad_norm": 0.638228126097299, "learning_rate": 3.4966044660319166e-06, "loss": 0.2684, "step": 17628 }, { "epoch": 1.2187348772900104, "grad_norm": 0.6793444935259009, "learning_rate": 3.4960706188857307e-06, "loss": 0.287, "step": 17629 }, { "epoch": 1.2188040096785344, "grad_norm": 0.6618867022401836, "learning_rate": 3.4955367905886177e-06, "loss": 0.3352, "step": 17630 }, { "epoch": 1.2188731420670584, "grad_norm": 0.7376392096235008, "learning_rate": 3.4950029811472653e-06, "loss": 0.2804, "step": 17631 }, { "epoch": 1.2189422744555825, "grad_norm": 0.6463413597234974, "learning_rate": 3.494469190568366e-06, "loss": 0.2499, "step": 17632 }, { "epoch": 1.2190114068441065, "grad_norm": 0.6893577367788958, "learning_rate": 3.4939354188586073e-06, "loss": 0.2953, "step": 17633 }, { "epoch": 1.2190805392326305, "grad_norm": 0.6525980638034526, "learning_rate": 3.4934016660246806e-06, "loss": 0.3626, "step": 17634 }, { "epoch": 1.2191496716211545, "grad_norm": 0.6660030799521472, "learning_rate": 3.492867932073275e-06, "loss": 0.2224, "step": 17635 }, { "epoch": 1.2192188040096785, "grad_norm": 0.9047691331622735, "learning_rate": 3.492334217011082e-06, "loss": 0.2525, "step": 17636 }, { "epoch": 1.2192879363982025, "grad_norm": 0.6864949195144029, "learning_rate": 3.491800520844787e-06, "loss": 0.3168, "step": 17637 }, { "epoch": 1.2193570687867266, "grad_norm": 1.2231120260378965, "learning_rate": 3.4912668435810824e-06, "loss": 0.2741, "step": 17638 }, { "epoch": 1.2194262011752506, "grad_norm": 0.7144570356841355, "learning_rate": 3.490733185226654e-06, "loss": 0.3121, "step": 17639 }, { "epoch": 1.2194953335637746, "grad_norm": 0.6373433661459419, "learning_rate": 3.4901995457881942e-06, "loss": 0.2698, "step": 17640 }, { "epoch": 1.2195644659522986, "grad_norm": 0.7065614417067109, "learning_rate": 3.4896659252723865e-06, "loss": 0.3604, "step": 17641 }, { "epoch": 1.2196335983408226, "grad_norm": 0.695878662144616, "learning_rate": 3.489132323685922e-06, "loss": 0.296, "step": 17642 }, { "epoch": 1.2197027307293467, "grad_norm": 0.7030390325731967, "learning_rate": 3.4885987410354864e-06, "loss": 0.2373, "step": 17643 }, { "epoch": 1.2197718631178707, "grad_norm": 0.6636319954645468, "learning_rate": 3.4880651773277707e-06, "loss": 0.3267, "step": 17644 }, { "epoch": 1.2198409955063947, "grad_norm": 0.6711907232077876, "learning_rate": 3.4875316325694573e-06, "loss": 0.3077, "step": 17645 }, { "epoch": 1.2199101278949187, "grad_norm": 0.5962996620805697, "learning_rate": 3.4869981067672377e-06, "loss": 0.2068, "step": 17646 }, { "epoch": 1.2199792602834427, "grad_norm": 0.7539196511235094, "learning_rate": 3.4864645999277946e-06, "loss": 0.2392, "step": 17647 }, { "epoch": 1.2200483926719667, "grad_norm": 0.7069840874232693, "learning_rate": 3.4859311120578186e-06, "loss": 0.3744, "step": 17648 }, { "epoch": 1.2201175250604908, "grad_norm": 0.6231795089518556, "learning_rate": 3.485397643163993e-06, "loss": 0.2488, "step": 17649 }, { "epoch": 1.2201866574490148, "grad_norm": 0.6643008898064499, "learning_rate": 3.484864193253007e-06, "loss": 0.2261, "step": 17650 }, { "epoch": 1.2202557898375388, "grad_norm": 0.6204891345863423, "learning_rate": 3.4843307623315425e-06, "loss": 0.2998, "step": 17651 }, { "epoch": 1.2203249222260628, "grad_norm": 0.678040366318538, "learning_rate": 3.4837973504062877e-06, "loss": 0.334, "step": 17652 }, { "epoch": 1.2203940546145868, "grad_norm": 0.6073865840057485, "learning_rate": 3.4832639574839266e-06, "loss": 0.2182, "step": 17653 }, { "epoch": 1.2204631870031108, "grad_norm": 0.6609170753819898, "learning_rate": 3.4827305835711465e-06, "loss": 0.2808, "step": 17654 }, { "epoch": 1.2205323193916349, "grad_norm": 0.6538138688931664, "learning_rate": 3.4821972286746295e-06, "loss": 0.3204, "step": 17655 }, { "epoch": 1.2206014517801589, "grad_norm": 0.8235546176135077, "learning_rate": 3.4816638928010627e-06, "loss": 0.297, "step": 17656 }, { "epoch": 1.220670584168683, "grad_norm": 0.6686029002147188, "learning_rate": 3.481130575957128e-06, "loss": 0.1944, "step": 17657 }, { "epoch": 1.2207397165572071, "grad_norm": 0.7211234221306456, "learning_rate": 3.4805972781495134e-06, "loss": 0.3153, "step": 17658 }, { "epoch": 1.2208088489457312, "grad_norm": 0.6512211111984252, "learning_rate": 3.4800639993848984e-06, "loss": 0.2938, "step": 17659 }, { "epoch": 1.2208779813342552, "grad_norm": 0.6783837704382862, "learning_rate": 3.4795307396699697e-06, "loss": 0.2274, "step": 17660 }, { "epoch": 1.2209471137227792, "grad_norm": 0.8493359040136582, "learning_rate": 3.4789974990114083e-06, "loss": 0.2883, "step": 17661 }, { "epoch": 1.2210162461113032, "grad_norm": 0.6600116981592568, "learning_rate": 3.478464277415901e-06, "loss": 0.3927, "step": 17662 }, { "epoch": 1.2210853784998272, "grad_norm": 0.659145373744675, "learning_rate": 3.477931074890126e-06, "loss": 0.2891, "step": 17663 }, { "epoch": 1.2211545108883513, "grad_norm": 0.6075499743967523, "learning_rate": 3.4773978914407714e-06, "loss": 0.2404, "step": 17664 }, { "epoch": 1.2212236432768753, "grad_norm": 0.8717291225875687, "learning_rate": 3.476864727074515e-06, "loss": 0.3082, "step": 17665 }, { "epoch": 1.2212927756653993, "grad_norm": 0.7178987946257334, "learning_rate": 3.4763315817980424e-06, "loss": 0.3629, "step": 17666 }, { "epoch": 1.2213619080539233, "grad_norm": 0.6577154404595669, "learning_rate": 3.4757984556180336e-06, "loss": 0.266, "step": 17667 }, { "epoch": 1.2214310404424473, "grad_norm": 0.6785444843041976, "learning_rate": 3.4752653485411724e-06, "loss": 0.2073, "step": 17668 }, { "epoch": 1.2215001728309713, "grad_norm": 0.7763429188279198, "learning_rate": 3.4747322605741372e-06, "loss": 0.3073, "step": 17669 }, { "epoch": 1.2215693052194954, "grad_norm": 0.6655680294078594, "learning_rate": 3.4741991917236125e-06, "loss": 0.3203, "step": 17670 }, { "epoch": 1.2216384376080194, "grad_norm": 0.6264372042878779, "learning_rate": 3.473666141996277e-06, "loss": 0.2739, "step": 17671 }, { "epoch": 1.2217075699965434, "grad_norm": 0.6878284872004584, "learning_rate": 3.4731331113988142e-06, "loss": 0.2432, "step": 17672 }, { "epoch": 1.2217767023850674, "grad_norm": 0.7535518115010046, "learning_rate": 3.4726000999379006e-06, "loss": 0.3729, "step": 17673 }, { "epoch": 1.2218458347735914, "grad_norm": 0.6955092635661733, "learning_rate": 3.472067107620221e-06, "loss": 0.2849, "step": 17674 }, { "epoch": 1.2219149671621155, "grad_norm": 0.649206359993102, "learning_rate": 3.471534134452452e-06, "loss": 0.2342, "step": 17675 }, { "epoch": 1.2219840995506395, "grad_norm": 0.675516103384947, "learning_rate": 3.471001180441277e-06, "loss": 0.2877, "step": 17676 }, { "epoch": 1.2220532319391635, "grad_norm": 0.6927066607317702, "learning_rate": 3.4704682455933715e-06, "loss": 0.3181, "step": 17677 }, { "epoch": 1.2221223643276875, "grad_norm": 0.6105303896557419, "learning_rate": 3.4699353299154183e-06, "loss": 0.2319, "step": 17678 }, { "epoch": 1.2221914967162115, "grad_norm": 0.6835421076609787, "learning_rate": 3.469402433414094e-06, "loss": 0.3185, "step": 17679 }, { "epoch": 1.2222606291047355, "grad_norm": 0.7577949512073252, "learning_rate": 3.4688695560960807e-06, "loss": 0.3915, "step": 17680 }, { "epoch": 1.2223297614932596, "grad_norm": 0.8242926285376417, "learning_rate": 3.4683366979680514e-06, "loss": 0.2802, "step": 17681 }, { "epoch": 1.2223988938817836, "grad_norm": 0.6811796253611261, "learning_rate": 3.4678038590366924e-06, "loss": 0.2329, "step": 17682 }, { "epoch": 1.2224680262703076, "grad_norm": 0.7504375225814863, "learning_rate": 3.4672710393086752e-06, "loss": 0.3397, "step": 17683 }, { "epoch": 1.2225371586588316, "grad_norm": 0.6366524250609426, "learning_rate": 3.4667382387906813e-06, "loss": 0.3244, "step": 17684 }, { "epoch": 1.2226062910473556, "grad_norm": 0.6595197707600857, "learning_rate": 3.466205457489386e-06, "loss": 0.2635, "step": 17685 }, { "epoch": 1.2226754234358796, "grad_norm": 0.6799780466610316, "learning_rate": 3.4656726954114707e-06, "loss": 0.3356, "step": 17686 }, { "epoch": 1.2227445558244037, "grad_norm": 0.6136610467645707, "learning_rate": 3.4651399525636074e-06, "loss": 0.3169, "step": 17687 }, { "epoch": 1.2228136882129277, "grad_norm": 0.8574512042949042, "learning_rate": 3.464607228952477e-06, "loss": 0.2805, "step": 17688 }, { "epoch": 1.2228828206014517, "grad_norm": 1.7110119954662397, "learning_rate": 3.464074524584754e-06, "loss": 0.2703, "step": 17689 }, { "epoch": 1.2229519529899757, "grad_norm": 0.694487800190651, "learning_rate": 3.463541839467118e-06, "loss": 0.2912, "step": 17690 }, { "epoch": 1.2230210853784997, "grad_norm": 0.6876465270180457, "learning_rate": 3.463009173606241e-06, "loss": 0.3801, "step": 17691 }, { "epoch": 1.2230902177670238, "grad_norm": 0.696581486213248, "learning_rate": 3.4624765270088023e-06, "loss": 0.2706, "step": 17692 }, { "epoch": 1.223159350155548, "grad_norm": 0.6874728191463856, "learning_rate": 3.461943899681477e-06, "loss": 0.2493, "step": 17693 }, { "epoch": 1.223228482544072, "grad_norm": 0.7900851441216251, "learning_rate": 3.4614112916309383e-06, "loss": 0.3577, "step": 17694 }, { "epoch": 1.223297614932596, "grad_norm": 0.7474180276295889, "learning_rate": 3.4608787028638646e-06, "loss": 0.3204, "step": 17695 }, { "epoch": 1.22336674732112, "grad_norm": 0.6134220721649704, "learning_rate": 3.4603461333869286e-06, "loss": 0.2222, "step": 17696 }, { "epoch": 1.223435879709644, "grad_norm": 0.6221787298536895, "learning_rate": 3.459813583206808e-06, "loss": 0.2147, "step": 17697 }, { "epoch": 1.223505012098168, "grad_norm": 0.6374067953354468, "learning_rate": 3.4592810523301734e-06, "loss": 0.3481, "step": 17698 }, { "epoch": 1.223574144486692, "grad_norm": 0.6176065051666709, "learning_rate": 3.4587485407637026e-06, "loss": 0.2533, "step": 17699 }, { "epoch": 1.2236432768752161, "grad_norm": 0.7035092540925924, "learning_rate": 3.4582160485140676e-06, "loss": 0.2746, "step": 17700 }, { "epoch": 1.2237124092637401, "grad_norm": 0.7388245183488329, "learning_rate": 3.4576835755879435e-06, "loss": 0.3755, "step": 17701 }, { "epoch": 1.2237815416522642, "grad_norm": 0.769575169507323, "learning_rate": 3.457151121992003e-06, "loss": 0.3597, "step": 17702 }, { "epoch": 1.2238506740407882, "grad_norm": 0.680343932220371, "learning_rate": 3.456618687732921e-06, "loss": 0.2107, "step": 17703 }, { "epoch": 1.2239198064293122, "grad_norm": 0.715687610603872, "learning_rate": 3.456086272817368e-06, "loss": 0.332, "step": 17704 }, { "epoch": 1.2239889388178362, "grad_norm": 0.7729513005187472, "learning_rate": 3.4555538772520193e-06, "loss": 0.3558, "step": 17705 }, { "epoch": 1.2240580712063602, "grad_norm": 0.6807828614246858, "learning_rate": 3.455021501043545e-06, "loss": 0.2818, "step": 17706 }, { "epoch": 1.2241272035948843, "grad_norm": 0.8188523810070256, "learning_rate": 3.4544891441986217e-06, "loss": 0.2521, "step": 17707 }, { "epoch": 1.2241963359834083, "grad_norm": 0.6444058471170648, "learning_rate": 3.4539568067239172e-06, "loss": 0.2857, "step": 17708 }, { "epoch": 1.2242654683719323, "grad_norm": 0.6679719273866838, "learning_rate": 3.4534244886261053e-06, "loss": 0.3444, "step": 17709 }, { "epoch": 1.2243346007604563, "grad_norm": 0.6351527144116984, "learning_rate": 3.4528921899118567e-06, "loss": 0.225, "step": 17710 }, { "epoch": 1.2244037331489803, "grad_norm": 0.7143893116075832, "learning_rate": 3.4523599105878458e-06, "loss": 0.2168, "step": 17711 }, { "epoch": 1.2244728655375043, "grad_norm": 0.7814384670031319, "learning_rate": 3.4518276506607393e-06, "loss": 0.3386, "step": 17712 }, { "epoch": 1.2245419979260284, "grad_norm": 0.7175966924593669, "learning_rate": 3.451295410137211e-06, "loss": 0.3369, "step": 17713 }, { "epoch": 1.2246111303145524, "grad_norm": 0.6900463322722987, "learning_rate": 3.4507631890239306e-06, "loss": 0.2119, "step": 17714 }, { "epoch": 1.2246802627030764, "grad_norm": 0.6621820470138244, "learning_rate": 3.450230987327571e-06, "loss": 0.2727, "step": 17715 }, { "epoch": 1.2247493950916004, "grad_norm": 0.6545663483461314, "learning_rate": 3.4496988050547975e-06, "loss": 0.3396, "step": 17716 }, { "epoch": 1.2248185274801244, "grad_norm": 0.6292547866433116, "learning_rate": 3.4491666422122845e-06, "loss": 0.2257, "step": 17717 }, { "epoch": 1.2248876598686484, "grad_norm": 0.7026355461765499, "learning_rate": 3.448634498806699e-06, "loss": 0.2717, "step": 17718 }, { "epoch": 1.2249567922571725, "grad_norm": 0.705323843935358, "learning_rate": 3.4481023748447117e-06, "loss": 0.3226, "step": 17719 }, { "epoch": 1.2250259246456965, "grad_norm": 0.7002322743576432, "learning_rate": 3.4475702703329917e-06, "loss": 0.3573, "step": 17720 }, { "epoch": 1.2250950570342205, "grad_norm": 0.6463736785515803, "learning_rate": 3.4470381852782098e-06, "loss": 0.2453, "step": 17721 }, { "epoch": 1.2251641894227445, "grad_norm": 0.6028885939145218, "learning_rate": 3.44650611968703e-06, "loss": 0.2314, "step": 17722 }, { "epoch": 1.2252333218112685, "grad_norm": 0.6985930743048667, "learning_rate": 3.445974073566125e-06, "loss": 0.3344, "step": 17723 }, { "epoch": 1.2253024541997926, "grad_norm": 0.7448988934307282, "learning_rate": 3.4454420469221617e-06, "loss": 0.2389, "step": 17724 }, { "epoch": 1.2253715865883166, "grad_norm": 0.6510959234982019, "learning_rate": 3.444910039761809e-06, "loss": 0.2823, "step": 17725 }, { "epoch": 1.2254407189768406, "grad_norm": 0.712630793353038, "learning_rate": 3.4443780520917324e-06, "loss": 0.3985, "step": 17726 }, { "epoch": 1.2255098513653646, "grad_norm": 0.7966804640676938, "learning_rate": 3.4438460839186012e-06, "loss": 0.3784, "step": 17727 }, { "epoch": 1.2255789837538886, "grad_norm": 0.631193366930472, "learning_rate": 3.4433141352490817e-06, "loss": 0.2755, "step": 17728 }, { "epoch": 1.2256481161424126, "grad_norm": 0.6685114575165875, "learning_rate": 3.442782206089843e-06, "loss": 0.2857, "step": 17729 }, { "epoch": 1.2257172485309367, "grad_norm": 0.7344592346684372, "learning_rate": 3.442250296447549e-06, "loss": 0.3294, "step": 17730 }, { "epoch": 1.2257863809194607, "grad_norm": 0.7314002296966103, "learning_rate": 3.441718406328868e-06, "loss": 0.2803, "step": 17731 }, { "epoch": 1.2258555133079847, "grad_norm": 0.5953741360645741, "learning_rate": 3.441186535740465e-06, "loss": 0.209, "step": 17732 }, { "epoch": 1.2259246456965087, "grad_norm": 0.6256738053783012, "learning_rate": 3.44065468468901e-06, "loss": 0.2629, "step": 17733 }, { "epoch": 1.2259937780850327, "grad_norm": 0.6852608629100888, "learning_rate": 3.4401228531811614e-06, "loss": 0.3522, "step": 17734 }, { "epoch": 1.2260629104735568, "grad_norm": 0.6950239069791566, "learning_rate": 3.4395910412235926e-06, "loss": 0.2537, "step": 17735 }, { "epoch": 1.2261320428620808, "grad_norm": 0.693909051305641, "learning_rate": 3.4390592488229635e-06, "loss": 0.1887, "step": 17736 }, { "epoch": 1.2262011752506048, "grad_norm": 0.8003069657319223, "learning_rate": 3.438527475985942e-06, "loss": 0.3432, "step": 17737 }, { "epoch": 1.2262703076391288, "grad_norm": 0.7139599971101303, "learning_rate": 3.4379957227191906e-06, "loss": 0.3253, "step": 17738 }, { "epoch": 1.226339440027653, "grad_norm": 0.6850044970282838, "learning_rate": 3.4374639890293783e-06, "loss": 0.236, "step": 17739 }, { "epoch": 1.226408572416177, "grad_norm": 0.7015270652634793, "learning_rate": 3.436932274923164e-06, "loss": 0.3051, "step": 17740 }, { "epoch": 1.226477704804701, "grad_norm": 0.6439994733419501, "learning_rate": 3.4364005804072147e-06, "loss": 0.3912, "step": 17741 }, { "epoch": 1.226546837193225, "grad_norm": 0.648125069703032, "learning_rate": 3.435868905488193e-06, "loss": 0.2471, "step": 17742 }, { "epoch": 1.2266159695817491, "grad_norm": 0.6628425687026729, "learning_rate": 3.4353372501727657e-06, "loss": 0.2139, "step": 17743 }, { "epoch": 1.2266851019702731, "grad_norm": 0.7813760106513957, "learning_rate": 3.4348056144675914e-06, "loss": 0.3339, "step": 17744 }, { "epoch": 1.2267542343587972, "grad_norm": 0.7053885702930955, "learning_rate": 3.434273998379336e-06, "loss": 0.265, "step": 17745 }, { "epoch": 1.2268233667473212, "grad_norm": 0.6986485022057273, "learning_rate": 3.4337424019146616e-06, "loss": 0.3201, "step": 17746 }, { "epoch": 1.2268924991358452, "grad_norm": 0.6616369312810292, "learning_rate": 3.4332108250802333e-06, "loss": 0.28, "step": 17747 }, { "epoch": 1.2269616315243692, "grad_norm": 0.6756249542435242, "learning_rate": 3.432679267882708e-06, "loss": 0.3844, "step": 17748 }, { "epoch": 1.2270307639128932, "grad_norm": 0.6637187913607823, "learning_rate": 3.4321477303287532e-06, "loss": 0.3067, "step": 17749 }, { "epoch": 1.2270998963014172, "grad_norm": 0.7813504503183921, "learning_rate": 3.4316162124250274e-06, "loss": 0.2245, "step": 17750 }, { "epoch": 1.2271690286899413, "grad_norm": 0.6961490465733627, "learning_rate": 3.4310847141781957e-06, "loss": 0.3183, "step": 17751 }, { "epoch": 1.2272381610784653, "grad_norm": 1.4412217099210396, "learning_rate": 3.4305532355949133e-06, "loss": 0.3536, "step": 17752 }, { "epoch": 1.2273072934669893, "grad_norm": 0.6542068926635464, "learning_rate": 3.430021776681849e-06, "loss": 0.2358, "step": 17753 }, { "epoch": 1.2273764258555133, "grad_norm": 0.6292137155397972, "learning_rate": 3.429490337445658e-06, "loss": 0.2423, "step": 17754 }, { "epoch": 1.2274455582440373, "grad_norm": 0.6611453505306306, "learning_rate": 3.428958917893004e-06, "loss": 0.306, "step": 17755 }, { "epoch": 1.2275146906325614, "grad_norm": 0.664762091197658, "learning_rate": 3.428427518030545e-06, "loss": 0.2743, "step": 17756 }, { "epoch": 1.2275838230210854, "grad_norm": 0.7341451307871157, "learning_rate": 3.4278961378649444e-06, "loss": 0.2758, "step": 17757 }, { "epoch": 1.2276529554096094, "grad_norm": 1.0619906142550346, "learning_rate": 3.4273647774028586e-06, "loss": 0.2832, "step": 17758 }, { "epoch": 1.2277220877981334, "grad_norm": 0.6977209102174581, "learning_rate": 3.426833436650949e-06, "loss": 0.3535, "step": 17759 }, { "epoch": 1.2277912201866574, "grad_norm": 0.6940727660846081, "learning_rate": 3.426302115615877e-06, "loss": 0.2273, "step": 17760 }, { "epoch": 1.2278603525751814, "grad_norm": 0.814674948391898, "learning_rate": 3.4257708143042967e-06, "loss": 0.2702, "step": 17761 }, { "epoch": 1.2279294849637055, "grad_norm": 0.6664609096601701, "learning_rate": 3.4252395327228716e-06, "loss": 0.3881, "step": 17762 }, { "epoch": 1.2279986173522295, "grad_norm": 0.6951184985157375, "learning_rate": 3.4247082708782577e-06, "loss": 0.3118, "step": 17763 }, { "epoch": 1.2280677497407535, "grad_norm": 0.6508080281463301, "learning_rate": 3.4241770287771157e-06, "loss": 0.221, "step": 17764 }, { "epoch": 1.2281368821292775, "grad_norm": 0.7200962862147617, "learning_rate": 3.4236458064261004e-06, "loss": 0.3169, "step": 17765 }, { "epoch": 1.2282060145178015, "grad_norm": 0.6660897080883057, "learning_rate": 3.423114603831873e-06, "loss": 0.3495, "step": 17766 }, { "epoch": 1.2282751469063256, "grad_norm": 0.6236827207886743, "learning_rate": 3.422583421001089e-06, "loss": 0.2445, "step": 17767 }, { "epoch": 1.2283442792948496, "grad_norm": 0.7792549766877767, "learning_rate": 3.422052257940409e-06, "loss": 0.2406, "step": 17768 }, { "epoch": 1.2284134116833736, "grad_norm": 0.6901256159107312, "learning_rate": 3.4215211146564864e-06, "loss": 0.2795, "step": 17769 }, { "epoch": 1.2284825440718976, "grad_norm": 0.652682260458811, "learning_rate": 3.4209899911559796e-06, "loss": 0.3355, "step": 17770 }, { "epoch": 1.2285516764604216, "grad_norm": 0.6755680213770111, "learning_rate": 3.4204588874455446e-06, "loss": 0.2717, "step": 17771 }, { "epoch": 1.2286208088489456, "grad_norm": 0.6640061566916862, "learning_rate": 3.4199278035318408e-06, "loss": 0.2383, "step": 17772 }, { "epoch": 1.2286899412374697, "grad_norm": 0.7104539897176073, "learning_rate": 3.41939673942152e-06, "loss": 0.3333, "step": 17773 }, { "epoch": 1.228759073625994, "grad_norm": 0.6553701001912455, "learning_rate": 3.4188656951212428e-06, "loss": 0.2678, "step": 17774 }, { "epoch": 1.228828206014518, "grad_norm": 0.6390749147131958, "learning_rate": 3.41833467063766e-06, "loss": 0.2924, "step": 17775 }, { "epoch": 1.228897338403042, "grad_norm": 0.7127815513453906, "learning_rate": 3.417803665977431e-06, "loss": 0.2956, "step": 17776 }, { "epoch": 1.228966470791566, "grad_norm": 0.8929311512503856, "learning_rate": 3.4172726811472086e-06, "loss": 0.3385, "step": 17777 }, { "epoch": 1.22903560318009, "grad_norm": 0.6305101711570814, "learning_rate": 3.4167417161536505e-06, "loss": 0.2369, "step": 17778 }, { "epoch": 1.229104735568614, "grad_norm": 0.6158504593610424, "learning_rate": 3.4162107710034074e-06, "loss": 0.2396, "step": 17779 }, { "epoch": 1.229173867957138, "grad_norm": 0.7189801735424378, "learning_rate": 3.415679845703137e-06, "loss": 0.36, "step": 17780 }, { "epoch": 1.229243000345662, "grad_norm": 1.1536448306290794, "learning_rate": 3.4151489402594913e-06, "loss": 0.2929, "step": 17781 }, { "epoch": 1.229312132734186, "grad_norm": 0.6491910713257624, "learning_rate": 3.4146180546791272e-06, "loss": 0.2139, "step": 17782 }, { "epoch": 1.22938126512271, "grad_norm": 0.6317647681069548, "learning_rate": 3.4140871889686946e-06, "loss": 0.2856, "step": 17783 }, { "epoch": 1.229450397511234, "grad_norm": 0.6490428992327337, "learning_rate": 3.41355634313485e-06, "loss": 0.367, "step": 17784 }, { "epoch": 1.229519529899758, "grad_norm": 0.6242174777104821, "learning_rate": 3.413025517184245e-06, "loss": 0.2134, "step": 17785 }, { "epoch": 1.2295886622882821, "grad_norm": 0.6038484018308702, "learning_rate": 3.412494711123534e-06, "loss": 0.2601, "step": 17786 }, { "epoch": 1.2296577946768061, "grad_norm": 0.6726390282634412, "learning_rate": 3.4119639249593674e-06, "loss": 0.3181, "step": 17787 }, { "epoch": 1.2297269270653302, "grad_norm": 0.6954999614110876, "learning_rate": 3.4114331586984e-06, "loss": 0.3222, "step": 17788 }, { "epoch": 1.2297960594538542, "grad_norm": 0.6605175159611464, "learning_rate": 3.4109024123472823e-06, "loss": 0.2714, "step": 17789 }, { "epoch": 1.2298651918423782, "grad_norm": 0.7194281807821441, "learning_rate": 3.410371685912668e-06, "loss": 0.3149, "step": 17790 }, { "epoch": 1.2299343242309022, "grad_norm": 0.6884734549097338, "learning_rate": 3.4098409794012057e-06, "loss": 0.3947, "step": 17791 }, { "epoch": 1.2300034566194262, "grad_norm": 0.6840661582451365, "learning_rate": 3.409310292819552e-06, "loss": 0.2691, "step": 17792 }, { "epoch": 1.2300725890079502, "grad_norm": 0.6727306695371852, "learning_rate": 3.4087796261743527e-06, "loss": 0.2272, "step": 17793 }, { "epoch": 1.2301417213964743, "grad_norm": 0.6211934480510989, "learning_rate": 3.408248979472262e-06, "loss": 0.2894, "step": 17794 }, { "epoch": 1.2302108537849983, "grad_norm": 0.7044108442752006, "learning_rate": 3.4077183527199294e-06, "loss": 0.2881, "step": 17795 }, { "epoch": 1.2302799861735223, "grad_norm": 0.904953047110991, "learning_rate": 3.407187745924007e-06, "loss": 0.2633, "step": 17796 }, { "epoch": 1.2303491185620463, "grad_norm": 0.7141003189425793, "learning_rate": 3.4066571590911425e-06, "loss": 0.2565, "step": 17797 }, { "epoch": 1.2304182509505703, "grad_norm": 0.7137744501006422, "learning_rate": 3.406126592227987e-06, "loss": 0.3642, "step": 17798 }, { "epoch": 1.2304873833390944, "grad_norm": 0.6763418171247146, "learning_rate": 3.4055960453411895e-06, "loss": 0.2856, "step": 17799 }, { "epoch": 1.2305565157276184, "grad_norm": 0.6090008344643782, "learning_rate": 3.4050655184374026e-06, "loss": 0.1939, "step": 17800 }, { "epoch": 1.2306256481161424, "grad_norm": 0.6829533920018356, "learning_rate": 3.4045350115232714e-06, "loss": 0.297, "step": 17801 }, { "epoch": 1.2306947805046664, "grad_norm": 0.6783199551493507, "learning_rate": 3.4040045246054475e-06, "loss": 0.3445, "step": 17802 }, { "epoch": 1.2307639128931904, "grad_norm": 0.597772880878333, "learning_rate": 3.4034740576905774e-06, "loss": 0.2602, "step": 17803 }, { "epoch": 1.2308330452817144, "grad_norm": 1.069096369182669, "learning_rate": 3.402943610785313e-06, "loss": 0.2213, "step": 17804 }, { "epoch": 1.2309021776702385, "grad_norm": 0.653071858472393, "learning_rate": 3.402413183896298e-06, "loss": 0.3963, "step": 17805 }, { "epoch": 1.2309713100587625, "grad_norm": 0.7018801802465189, "learning_rate": 3.4018827770301845e-06, "loss": 0.3002, "step": 17806 }, { "epoch": 1.2310404424472865, "grad_norm": 0.6755702066956523, "learning_rate": 3.401352390193616e-06, "loss": 0.2396, "step": 17807 }, { "epoch": 1.2311095748358105, "grad_norm": 0.6740264882479596, "learning_rate": 3.4008220233932444e-06, "loss": 0.3013, "step": 17808 }, { "epoch": 1.2311787072243345, "grad_norm": 0.6821546180399441, "learning_rate": 3.400291676635714e-06, "loss": 0.3941, "step": 17809 }, { "epoch": 1.2312478396128586, "grad_norm": 0.7178098084139473, "learning_rate": 3.399761349927674e-06, "loss": 0.2385, "step": 17810 }, { "epoch": 1.2313169720013826, "grad_norm": 0.6598190640788268, "learning_rate": 3.3992310432757686e-06, "loss": 0.2417, "step": 17811 }, { "epoch": 1.2313861043899066, "grad_norm": 0.6378754323562051, "learning_rate": 3.398700756686646e-06, "loss": 0.2932, "step": 17812 }, { "epoch": 1.2314552367784306, "grad_norm": 0.6436828505770242, "learning_rate": 3.3981704901669504e-06, "loss": 0.2534, "step": 17813 }, { "epoch": 1.2315243691669546, "grad_norm": 0.6467453340076118, "learning_rate": 3.3976402437233313e-06, "loss": 0.1966, "step": 17814 }, { "epoch": 1.2315935015554786, "grad_norm": 0.7292138935139273, "learning_rate": 3.3971100173624304e-06, "loss": 0.2895, "step": 17815 }, { "epoch": 1.2316626339440027, "grad_norm": 0.7832917334219739, "learning_rate": 3.396579811090896e-06, "loss": 0.3817, "step": 17816 }, { "epoch": 1.2317317663325267, "grad_norm": 0.6000072095294179, "learning_rate": 3.396049624915371e-06, "loss": 0.2282, "step": 17817 }, { "epoch": 1.2318008987210507, "grad_norm": 1.2602576577358415, "learning_rate": 3.3955194588425034e-06, "loss": 0.2407, "step": 17818 }, { "epoch": 1.2318700311095747, "grad_norm": 0.6799966613980498, "learning_rate": 3.3949893128789346e-06, "loss": 0.3418, "step": 17819 }, { "epoch": 1.231939163498099, "grad_norm": 0.6385107523686911, "learning_rate": 3.394459187031311e-06, "loss": 0.2826, "step": 17820 }, { "epoch": 1.232008295886623, "grad_norm": 0.6626244399477736, "learning_rate": 3.3939290813062754e-06, "loss": 0.3148, "step": 17821 }, { "epoch": 1.232077428275147, "grad_norm": 0.6213985078611985, "learning_rate": 3.393398995710475e-06, "loss": 0.2385, "step": 17822 }, { "epoch": 1.232146560663671, "grad_norm": 0.6609897277668905, "learning_rate": 3.3928689302505476e-06, "loss": 0.3626, "step": 17823 }, { "epoch": 1.232215693052195, "grad_norm": 0.7331869453642567, "learning_rate": 3.392338884933143e-06, "loss": 0.2756, "step": 17824 }, { "epoch": 1.232284825440719, "grad_norm": 0.7100521411012051, "learning_rate": 3.3918088597649003e-06, "loss": 0.2563, "step": 17825 }, { "epoch": 1.232353957829243, "grad_norm": 0.7112557725195271, "learning_rate": 3.391278854752463e-06, "loss": 0.3038, "step": 17826 }, { "epoch": 1.232423090217767, "grad_norm": 0.6958578995392806, "learning_rate": 3.390748869902477e-06, "loss": 0.3514, "step": 17827 }, { "epoch": 1.232492222606291, "grad_norm": 0.6391588431940163, "learning_rate": 3.3902189052215796e-06, "loss": 0.2911, "step": 17828 }, { "epoch": 1.2325613549948151, "grad_norm": 0.7179409941057628, "learning_rate": 3.3896889607164163e-06, "loss": 0.2693, "step": 17829 }, { "epoch": 1.2326304873833391, "grad_norm": 0.7211068569834844, "learning_rate": 3.3891590363936265e-06, "loss": 0.3336, "step": 17830 }, { "epoch": 1.2326996197718632, "grad_norm": 0.6728832010188764, "learning_rate": 3.3886291322598564e-06, "loss": 0.2502, "step": 17831 }, { "epoch": 1.2327687521603872, "grad_norm": 0.7022199559388941, "learning_rate": 3.388099248321742e-06, "loss": 0.2455, "step": 17832 }, { "epoch": 1.2328378845489112, "grad_norm": 0.7109817395227787, "learning_rate": 3.3875693845859286e-06, "loss": 0.3192, "step": 17833 }, { "epoch": 1.2329070169374352, "grad_norm": 0.6690590933818505, "learning_rate": 3.387039541059054e-06, "loss": 0.3952, "step": 17834 }, { "epoch": 1.2329761493259592, "grad_norm": 0.599888311706295, "learning_rate": 3.3865097177477625e-06, "loss": 0.2336, "step": 17835 }, { "epoch": 1.2330452817144832, "grad_norm": 0.6511140525985727, "learning_rate": 3.38597991465869e-06, "loss": 0.2646, "step": 17836 }, { "epoch": 1.2331144141030073, "grad_norm": 0.7182707474431516, "learning_rate": 3.38545013179848e-06, "loss": 0.3561, "step": 17837 }, { "epoch": 1.2331835464915313, "grad_norm": 0.6941546442653492, "learning_rate": 3.384920369173771e-06, "loss": 0.3092, "step": 17838 }, { "epoch": 1.2332526788800553, "grad_norm": 0.6574865895829196, "learning_rate": 3.384390626791204e-06, "loss": 0.2314, "step": 17839 }, { "epoch": 1.2333218112685793, "grad_norm": 1.113398196057494, "learning_rate": 3.3838609046574157e-06, "loss": 0.2639, "step": 17840 }, { "epoch": 1.2333909436571033, "grad_norm": 0.6957807983738133, "learning_rate": 3.3833312027790477e-06, "loss": 0.412, "step": 17841 }, { "epoch": 1.2334600760456274, "grad_norm": 0.6599981584170078, "learning_rate": 3.382801521162737e-06, "loss": 0.2509, "step": 17842 }, { "epoch": 1.2335292084341514, "grad_norm": 0.624691307053714, "learning_rate": 3.382271859815124e-06, "loss": 0.232, "step": 17843 }, { "epoch": 1.2335983408226754, "grad_norm": 0.7167786298905742, "learning_rate": 3.381742218742846e-06, "loss": 0.3456, "step": 17844 }, { "epoch": 1.2336674732111994, "grad_norm": 0.6957033835178867, "learning_rate": 3.381212597952542e-06, "loss": 0.2984, "step": 17845 }, { "epoch": 1.2337366055997234, "grad_norm": 0.6864927952173934, "learning_rate": 3.3806829974508487e-06, "loss": 0.2567, "step": 17846 }, { "epoch": 1.2338057379882474, "grad_norm": 0.7082429802947469, "learning_rate": 3.380153417244405e-06, "loss": 0.3122, "step": 17847 }, { "epoch": 1.2338748703767715, "grad_norm": 0.6987116302937574, "learning_rate": 3.379623857339846e-06, "loss": 0.3647, "step": 17848 }, { "epoch": 1.2339440027652955, "grad_norm": 0.6468956855286159, "learning_rate": 3.3790943177438122e-06, "loss": 0.2931, "step": 17849 }, { "epoch": 1.2340131351538195, "grad_norm": 1.1318179873708154, "learning_rate": 3.3785647984629367e-06, "loss": 0.2123, "step": 17850 }, { "epoch": 1.2340822675423435, "grad_norm": 0.6921014713588304, "learning_rate": 3.378035299503859e-06, "loss": 0.2793, "step": 17851 }, { "epoch": 1.2341513999308675, "grad_norm": 0.6818129856334935, "learning_rate": 3.3775058208732133e-06, "loss": 0.3044, "step": 17852 }, { "epoch": 1.2342205323193916, "grad_norm": 0.6371748307697294, "learning_rate": 3.376976362577639e-06, "loss": 0.2455, "step": 17853 }, { "epoch": 1.2342896647079156, "grad_norm": 0.6096751549744853, "learning_rate": 3.3764469246237664e-06, "loss": 0.2408, "step": 17854 }, { "epoch": 1.2343587970964396, "grad_norm": 0.6797362308121655, "learning_rate": 3.375917507018236e-06, "loss": 0.3836, "step": 17855 }, { "epoch": 1.2344279294849638, "grad_norm": 0.6452415721861483, "learning_rate": 3.37538810976768e-06, "loss": 0.2831, "step": 17856 }, { "epoch": 1.2344970618734878, "grad_norm": 0.6407915338433465, "learning_rate": 3.3748587328787373e-06, "loss": 0.2558, "step": 17857 }, { "epoch": 1.2345661942620119, "grad_norm": 1.009913042138104, "learning_rate": 3.3743293763580377e-06, "loss": 0.3189, "step": 17858 }, { "epoch": 1.2346353266505359, "grad_norm": 0.7037854310405431, "learning_rate": 3.3738000402122194e-06, "loss": 0.3454, "step": 17859 }, { "epoch": 1.23470445903906, "grad_norm": 1.1763000615740102, "learning_rate": 3.3732707244479145e-06, "loss": 0.2764, "step": 17860 }, { "epoch": 1.234773591427584, "grad_norm": 0.6294874899890724, "learning_rate": 3.3727414290717595e-06, "loss": 0.2333, "step": 17861 }, { "epoch": 1.234842723816108, "grad_norm": 0.7274172592675323, "learning_rate": 3.372212154090385e-06, "loss": 0.3285, "step": 17862 }, { "epoch": 1.234911856204632, "grad_norm": 0.8886868393035771, "learning_rate": 3.371682899510429e-06, "loss": 0.3154, "step": 17863 }, { "epoch": 1.234980988593156, "grad_norm": 0.6167156201088436, "learning_rate": 3.3711536653385196e-06, "loss": 0.2605, "step": 17864 }, { "epoch": 1.23505012098168, "grad_norm": 0.6742895936465184, "learning_rate": 3.370624451581293e-06, "loss": 0.2565, "step": 17865 }, { "epoch": 1.235119253370204, "grad_norm": 0.6923208926422204, "learning_rate": 3.3700952582453805e-06, "loss": 0.367, "step": 17866 }, { "epoch": 1.235188385758728, "grad_norm": 0.7005288500576231, "learning_rate": 3.369566085337417e-06, "loss": 0.2801, "step": 17867 }, { "epoch": 1.235257518147252, "grad_norm": 0.6114550857020332, "learning_rate": 3.369036932864031e-06, "loss": 0.22, "step": 17868 }, { "epoch": 1.235326650535776, "grad_norm": 0.6517860291607473, "learning_rate": 3.3685078008318574e-06, "loss": 0.3289, "step": 17869 }, { "epoch": 1.2353957829243, "grad_norm": 0.7088114334510973, "learning_rate": 3.367978689247526e-06, "loss": 0.3038, "step": 17870 }, { "epoch": 1.235464915312824, "grad_norm": 0.6689485359442789, "learning_rate": 3.3674495981176714e-06, "loss": 0.253, "step": 17871 }, { "epoch": 1.2355340477013481, "grad_norm": 0.6879638592288643, "learning_rate": 3.366920527448921e-06, "loss": 0.2386, "step": 17872 }, { "epoch": 1.2356031800898721, "grad_norm": 0.6919138258021084, "learning_rate": 3.366391477247908e-06, "loss": 0.3655, "step": 17873 }, { "epoch": 1.2356723124783962, "grad_norm": 0.6709001513085816, "learning_rate": 3.3658624475212616e-06, "loss": 0.3002, "step": 17874 }, { "epoch": 1.2357414448669202, "grad_norm": 0.6501006664011293, "learning_rate": 3.3653334382756154e-06, "loss": 0.2446, "step": 17875 }, { "epoch": 1.2358105772554442, "grad_norm": 0.6499182899759592, "learning_rate": 3.3648044495175946e-06, "loss": 0.3367, "step": 17876 }, { "epoch": 1.2358797096439682, "grad_norm": 0.7221309742717509, "learning_rate": 3.3642754812538338e-06, "loss": 0.3763, "step": 17877 }, { "epoch": 1.2359488420324922, "grad_norm": 0.7106605006716739, "learning_rate": 3.363746533490959e-06, "loss": 0.2712, "step": 17878 }, { "epoch": 1.2360179744210162, "grad_norm": 0.6834243427761137, "learning_rate": 3.3632176062356026e-06, "loss": 0.2455, "step": 17879 }, { "epoch": 1.2360871068095403, "grad_norm": 0.6942507533365706, "learning_rate": 3.3626886994943917e-06, "loss": 0.3467, "step": 17880 }, { "epoch": 1.2361562391980643, "grad_norm": 0.6886203429032712, "learning_rate": 3.362159813273958e-06, "loss": 0.2617, "step": 17881 }, { "epoch": 1.2362253715865883, "grad_norm": 0.698879256257564, "learning_rate": 3.361630947580926e-06, "loss": 0.2999, "step": 17882 }, { "epoch": 1.2362945039751123, "grad_norm": 0.6554408251060863, "learning_rate": 3.3611021024219275e-06, "loss": 0.2988, "step": 17883 }, { "epoch": 1.2363636363636363, "grad_norm": 0.6770505294172583, "learning_rate": 3.360573277803588e-06, "loss": 0.406, "step": 17884 }, { "epoch": 1.2364327687521603, "grad_norm": 0.633357238909726, "learning_rate": 3.36004447373254e-06, "loss": 0.2302, "step": 17885 }, { "epoch": 1.2365019011406844, "grad_norm": 0.6588405730721019, "learning_rate": 3.359515690215405e-06, "loss": 0.216, "step": 17886 }, { "epoch": 1.2365710335292084, "grad_norm": 0.6913900412764534, "learning_rate": 3.358986927258814e-06, "loss": 0.3577, "step": 17887 }, { "epoch": 1.2366401659177324, "grad_norm": 0.7377289562659217, "learning_rate": 3.3584581848693925e-06, "loss": 0.3133, "step": 17888 }, { "epoch": 1.2367092983062564, "grad_norm": 0.69683717440621, "learning_rate": 3.357929463053771e-06, "loss": 0.2358, "step": 17889 }, { "epoch": 1.2367784306947804, "grad_norm": 0.6549604590926456, "learning_rate": 3.35740076181857e-06, "loss": 0.2598, "step": 17890 }, { "epoch": 1.2368475630833045, "grad_norm": 0.658014364055571, "learning_rate": 3.3568720811704203e-06, "loss": 0.3272, "step": 17891 }, { "epoch": 1.2369166954718285, "grad_norm": 0.7340052837817733, "learning_rate": 3.3563434211159463e-06, "loss": 0.284, "step": 17892 }, { "epoch": 1.2369858278603525, "grad_norm": 0.6770041861526158, "learning_rate": 3.3558147816617765e-06, "loss": 0.2639, "step": 17893 }, { "epoch": 1.2370549602488765, "grad_norm": 0.6088065217594326, "learning_rate": 3.3552861628145326e-06, "loss": 0.3111, "step": 17894 }, { "epoch": 1.2371240926374005, "grad_norm": 0.7250244596503608, "learning_rate": 3.3547575645808404e-06, "loss": 0.2994, "step": 17895 }, { "epoch": 1.2371932250259245, "grad_norm": 0.5967504246199508, "learning_rate": 3.354228986967329e-06, "loss": 0.2618, "step": 17896 }, { "epoch": 1.2372623574144486, "grad_norm": 0.7129683183584106, "learning_rate": 3.3537004299806163e-06, "loss": 0.3161, "step": 17897 }, { "epoch": 1.2373314898029726, "grad_norm": 0.6942101513481688, "learning_rate": 3.3531718936273346e-06, "loss": 0.3417, "step": 17898 }, { "epoch": 1.2374006221914966, "grad_norm": 0.6085906177127235, "learning_rate": 3.3526433779141023e-06, "loss": 0.2018, "step": 17899 }, { "epoch": 1.2374697545800206, "grad_norm": 0.6009819131652517, "learning_rate": 3.352114882847546e-06, "loss": 0.2185, "step": 17900 }, { "epoch": 1.2375388869685449, "grad_norm": 0.6487694795556711, "learning_rate": 3.3515864084342874e-06, "loss": 0.3172, "step": 17901 }, { "epoch": 1.2376080193570689, "grad_norm": 0.6576450490562644, "learning_rate": 3.351057954680954e-06, "loss": 0.2973, "step": 17902 }, { "epoch": 1.237677151745593, "grad_norm": 0.6358625277147217, "learning_rate": 3.350529521594164e-06, "loss": 0.2015, "step": 17903 }, { "epoch": 1.237746284134117, "grad_norm": 0.6316979499633567, "learning_rate": 3.350001109180544e-06, "loss": 0.264, "step": 17904 }, { "epoch": 1.237815416522641, "grad_norm": 0.7357525033585223, "learning_rate": 3.349472717446715e-06, "loss": 0.3429, "step": 17905 }, { "epoch": 1.237884548911165, "grad_norm": 0.6255526443693845, "learning_rate": 3.3489443463993023e-06, "loss": 0.2438, "step": 17906 }, { "epoch": 1.237953681299689, "grad_norm": 0.6467910780204604, "learning_rate": 3.3484159960449224e-06, "loss": 0.2234, "step": 17907 }, { "epoch": 1.238022813688213, "grad_norm": 0.6810334785585873, "learning_rate": 3.3478876663902028e-06, "loss": 0.2346, "step": 17908 }, { "epoch": 1.238091946076737, "grad_norm": 0.6805178790879214, "learning_rate": 3.3473593574417618e-06, "loss": 0.3576, "step": 17909 }, { "epoch": 1.238161078465261, "grad_norm": 0.5784045120750265, "learning_rate": 3.3468310692062235e-06, "loss": 0.2746, "step": 17910 }, { "epoch": 1.238230210853785, "grad_norm": 0.6728359105385425, "learning_rate": 3.3463028016902056e-06, "loss": 0.2497, "step": 17911 }, { "epoch": 1.238299343242309, "grad_norm": 0.6873696516414213, "learning_rate": 3.3457745549003325e-06, "loss": 0.3058, "step": 17912 }, { "epoch": 1.238368475630833, "grad_norm": 0.6606555571574827, "learning_rate": 3.3452463288432214e-06, "loss": 0.3016, "step": 17913 }, { "epoch": 1.238437608019357, "grad_norm": 0.637557596353202, "learning_rate": 3.344718123525496e-06, "loss": 0.2442, "step": 17914 }, { "epoch": 1.2385067404078811, "grad_norm": 0.7018646196198369, "learning_rate": 3.344189938953774e-06, "loss": 0.2421, "step": 17915 }, { "epoch": 1.2385758727964051, "grad_norm": 0.6619209864841363, "learning_rate": 3.3436617751346784e-06, "loss": 0.366, "step": 17916 }, { "epoch": 1.2386450051849291, "grad_norm": 0.8371679251903204, "learning_rate": 3.343133632074824e-06, "loss": 0.287, "step": 17917 }, { "epoch": 1.2387141375734532, "grad_norm": 0.6526163208386069, "learning_rate": 3.342605509780834e-06, "loss": 0.2888, "step": 17918 }, { "epoch": 1.2387832699619772, "grad_norm": 0.7215208511906567, "learning_rate": 3.3420774082593254e-06, "loss": 0.3151, "step": 17919 }, { "epoch": 1.2388524023505012, "grad_norm": 0.7051064071782345, "learning_rate": 3.3415493275169196e-06, "loss": 0.3182, "step": 17920 }, { "epoch": 1.2389215347390252, "grad_norm": 0.6309246660421175, "learning_rate": 3.3410212675602314e-06, "loss": 0.2247, "step": 17921 }, { "epoch": 1.2389906671275492, "grad_norm": 0.6771471641044514, "learning_rate": 3.340493228395882e-06, "loss": 0.2405, "step": 17922 }, { "epoch": 1.2390597995160733, "grad_norm": 0.6361742389798861, "learning_rate": 3.339965210030487e-06, "loss": 0.323, "step": 17923 }, { "epoch": 1.2391289319045973, "grad_norm": 0.6269573316128533, "learning_rate": 3.3394372124706687e-06, "loss": 0.2109, "step": 17924 }, { "epoch": 1.2391980642931213, "grad_norm": 0.597988529117976, "learning_rate": 3.3389092357230384e-06, "loss": 0.2103, "step": 17925 }, { "epoch": 1.2392671966816453, "grad_norm": 0.6540626640904962, "learning_rate": 3.3383812797942184e-06, "loss": 0.2945, "step": 17926 }, { "epoch": 1.2393363290701693, "grad_norm": 0.672056198833661, "learning_rate": 3.3378533446908223e-06, "loss": 0.3061, "step": 17927 }, { "epoch": 1.2394054614586933, "grad_norm": 0.6306134493299699, "learning_rate": 3.33732543041947e-06, "loss": 0.2671, "step": 17928 }, { "epoch": 1.2394745938472174, "grad_norm": 0.7783272814588333, "learning_rate": 3.3367975369867742e-06, "loss": 0.2463, "step": 17929 }, { "epoch": 1.2395437262357414, "grad_norm": 0.6751579473254459, "learning_rate": 3.3362696643993555e-06, "loss": 0.3938, "step": 17930 }, { "epoch": 1.2396128586242654, "grad_norm": 0.7362639601613825, "learning_rate": 3.3357418126638254e-06, "loss": 0.237, "step": 17931 }, { "epoch": 1.2396819910127894, "grad_norm": 0.6344190344618196, "learning_rate": 3.335213981786802e-06, "loss": 0.2002, "step": 17932 }, { "epoch": 1.2397511234013134, "grad_norm": 0.7762737445365363, "learning_rate": 3.3346861717749e-06, "loss": 0.2561, "step": 17933 }, { "epoch": 1.2398202557898375, "grad_norm": 0.6519718577479844, "learning_rate": 3.334158382634738e-06, "loss": 0.3519, "step": 17934 }, { "epoch": 1.2398893881783615, "grad_norm": 0.6460357320711959, "learning_rate": 3.3336306143729245e-06, "loss": 0.2225, "step": 17935 }, { "epoch": 1.2399585205668855, "grad_norm": 0.6201635694733615, "learning_rate": 3.333102866996079e-06, "loss": 0.2127, "step": 17936 }, { "epoch": 1.2400276529554097, "grad_norm": 0.7051177550132327, "learning_rate": 3.332575140510813e-06, "loss": 0.369, "step": 17937 }, { "epoch": 1.2400967853439337, "grad_norm": 0.8252819363857451, "learning_rate": 3.332047434923745e-06, "loss": 0.2817, "step": 17938 }, { "epoch": 1.2401659177324578, "grad_norm": 0.7228180054858303, "learning_rate": 3.3315197502414832e-06, "loss": 0.3201, "step": 17939 }, { "epoch": 1.2402350501209818, "grad_norm": 0.7012565517218923, "learning_rate": 3.330992086470645e-06, "loss": 0.2479, "step": 17940 }, { "epoch": 1.2403041825095058, "grad_norm": 0.5838495170063316, "learning_rate": 3.3304644436178407e-06, "loss": 0.3333, "step": 17941 }, { "epoch": 1.2403733148980298, "grad_norm": 0.6809160788363549, "learning_rate": 3.329936821689688e-06, "loss": 0.2596, "step": 17942 }, { "epoch": 1.2404424472865538, "grad_norm": 0.6670198743265479, "learning_rate": 3.329409220692794e-06, "loss": 0.2781, "step": 17943 }, { "epoch": 1.2405115796750779, "grad_norm": 1.150762180390569, "learning_rate": 3.3288816406337758e-06, "loss": 0.3323, "step": 17944 }, { "epoch": 1.2405807120636019, "grad_norm": 0.7097723122026705, "learning_rate": 3.3283540815192426e-06, "loss": 0.3011, "step": 17945 }, { "epoch": 1.240649844452126, "grad_norm": 0.7613427136149258, "learning_rate": 3.3278265433558106e-06, "loss": 0.2334, "step": 17946 }, { "epoch": 1.24071897684065, "grad_norm": 0.6755227051010106, "learning_rate": 3.3272990261500857e-06, "loss": 0.2674, "step": 17947 }, { "epoch": 1.240788109229174, "grad_norm": 0.6715133461476203, "learning_rate": 3.3267715299086843e-06, "loss": 0.3603, "step": 17948 }, { "epoch": 1.240857241617698, "grad_norm": 0.7610888863366169, "learning_rate": 3.3262440546382147e-06, "loss": 0.2671, "step": 17949 }, { "epoch": 1.240926374006222, "grad_norm": 0.746518447119647, "learning_rate": 3.325716600345289e-06, "loss": 0.2139, "step": 17950 }, { "epoch": 1.240995506394746, "grad_norm": 1.1051973319440962, "learning_rate": 3.325189167036518e-06, "loss": 0.2891, "step": 17951 }, { "epoch": 1.24106463878327, "grad_norm": 0.6348441744220918, "learning_rate": 3.3246617547185136e-06, "loss": 0.3351, "step": 17952 }, { "epoch": 1.241133771171794, "grad_norm": 0.7213674202529351, "learning_rate": 3.3241343633978818e-06, "loss": 0.259, "step": 17953 }, { "epoch": 1.241202903560318, "grad_norm": 0.6422197779426168, "learning_rate": 3.3236069930812365e-06, "loss": 0.2293, "step": 17954 }, { "epoch": 1.241272035948842, "grad_norm": 0.7375228262870825, "learning_rate": 3.3230796437751846e-06, "loss": 0.3939, "step": 17955 }, { "epoch": 1.241341168337366, "grad_norm": 0.7061607882768015, "learning_rate": 3.3225523154863393e-06, "loss": 0.2458, "step": 17956 }, { "epoch": 1.24141030072589, "grad_norm": 0.6355595576271198, "learning_rate": 3.3220250082213045e-06, "loss": 0.2323, "step": 17957 }, { "epoch": 1.241479433114414, "grad_norm": 0.6253485504184496, "learning_rate": 3.321497721986693e-06, "loss": 0.256, "step": 17958 }, { "epoch": 1.2415485655029381, "grad_norm": 0.7420425416291311, "learning_rate": 3.3209704567891133e-06, "loss": 0.3769, "step": 17959 }, { "epoch": 1.2416176978914621, "grad_norm": 0.7254084806022317, "learning_rate": 3.3204432126351706e-06, "loss": 0.2463, "step": 17960 }, { "epoch": 1.2416868302799862, "grad_norm": 0.6333218899972975, "learning_rate": 3.319915989531476e-06, "loss": 0.2547, "step": 17961 }, { "epoch": 1.2417559626685102, "grad_norm": 0.6364571541742607, "learning_rate": 3.3193887874846346e-06, "loss": 0.2977, "step": 17962 }, { "epoch": 1.2418250950570342, "grad_norm": 0.695858489856123, "learning_rate": 3.318861606501258e-06, "loss": 0.2801, "step": 17963 }, { "epoch": 1.2418942274455582, "grad_norm": 0.7423266424327102, "learning_rate": 3.318334446587949e-06, "loss": 0.2754, "step": 17964 }, { "epoch": 1.2419633598340822, "grad_norm": 0.6472855230513579, "learning_rate": 3.317807307751318e-06, "loss": 0.246, "step": 17965 }, { "epoch": 1.2420324922226063, "grad_norm": 0.9928610981910831, "learning_rate": 3.317280189997969e-06, "loss": 0.3525, "step": 17966 }, { "epoch": 1.2421016246111303, "grad_norm": 0.675425775403575, "learning_rate": 3.3167530933345117e-06, "loss": 0.259, "step": 17967 }, { "epoch": 1.2421707569996543, "grad_norm": 1.8563686032004043, "learning_rate": 3.3162260177675477e-06, "loss": 0.22, "step": 17968 }, { "epoch": 1.2422398893881783, "grad_norm": 0.6917056306691416, "learning_rate": 3.3156989633036884e-06, "loss": 0.3212, "step": 17969 }, { "epoch": 1.2423090217767023, "grad_norm": 0.6744367956251442, "learning_rate": 3.3151719299495355e-06, "loss": 0.3258, "step": 17970 }, { "epoch": 1.2423781541652263, "grad_norm": 0.655872801241432, "learning_rate": 3.3146449177116968e-06, "loss": 0.2412, "step": 17971 }, { "epoch": 1.2424472865537504, "grad_norm": 0.6395795863898432, "learning_rate": 3.3141179265967747e-06, "loss": 0.266, "step": 17972 }, { "epoch": 1.2425164189422744, "grad_norm": 0.7543201387370253, "learning_rate": 3.3135909566113784e-06, "loss": 0.3619, "step": 17973 }, { "epoch": 1.2425855513307984, "grad_norm": 0.6371395795977062, "learning_rate": 3.3130640077621074e-06, "loss": 0.2416, "step": 17974 }, { "epoch": 1.2426546837193224, "grad_norm": 0.630430146531939, "learning_rate": 3.31253708005557e-06, "loss": 0.2076, "step": 17975 }, { "epoch": 1.2427238161078464, "grad_norm": 0.6637329715424718, "learning_rate": 3.3120101734983677e-06, "loss": 0.3334, "step": 17976 }, { "epoch": 1.2427929484963705, "grad_norm": 0.6383502383566585, "learning_rate": 3.311483288097107e-06, "loss": 0.3277, "step": 17977 }, { "epoch": 1.2428620808848945, "grad_norm": 0.6097481493208822, "learning_rate": 3.3109564238583893e-06, "loss": 0.2664, "step": 17978 }, { "epoch": 1.2429312132734185, "grad_norm": 0.6649301936566778, "learning_rate": 3.3104295807888187e-06, "loss": 0.2475, "step": 17979 }, { "epoch": 1.2430003456619425, "grad_norm": 0.6883570023204286, "learning_rate": 3.3099027588949974e-06, "loss": 0.3428, "step": 17980 }, { "epoch": 1.2430694780504665, "grad_norm": 0.6295906015916898, "learning_rate": 3.3093759581835307e-06, "loss": 0.2436, "step": 17981 }, { "epoch": 1.2431386104389905, "grad_norm": 0.58541710614353, "learning_rate": 3.308849178661017e-06, "loss": 0.2278, "step": 17982 }, { "epoch": 1.2432077428275148, "grad_norm": 0.5935680481438603, "learning_rate": 3.3083224203340623e-06, "loss": 0.2454, "step": 17983 }, { "epoch": 1.2432768752160388, "grad_norm": 0.6971379148806361, "learning_rate": 3.3077956832092663e-06, "loss": 0.357, "step": 17984 }, { "epoch": 1.2433460076045628, "grad_norm": 0.6505330692088275, "learning_rate": 3.3072689672932333e-06, "loss": 0.2658, "step": 17985 }, { "epoch": 1.2434151399930868, "grad_norm": 0.6577089788449708, "learning_rate": 3.3067422725925598e-06, "loss": 0.2349, "step": 17986 }, { "epoch": 1.2434842723816109, "grad_norm": 0.8899952528429428, "learning_rate": 3.3062155991138532e-06, "loss": 0.3339, "step": 17987 }, { "epoch": 1.2435534047701349, "grad_norm": 0.6558900436810587, "learning_rate": 3.30568894686371e-06, "loss": 0.318, "step": 17988 }, { "epoch": 1.243622537158659, "grad_norm": 0.6945129401765984, "learning_rate": 3.3051623158487324e-06, "loss": 0.2428, "step": 17989 }, { "epoch": 1.243691669547183, "grad_norm": 0.6836933249625462, "learning_rate": 3.3046357060755196e-06, "loss": 0.3524, "step": 17990 }, { "epoch": 1.243760801935707, "grad_norm": 0.7068121830233979, "learning_rate": 3.304109117550676e-06, "loss": 0.3816, "step": 17991 }, { "epoch": 1.243829934324231, "grad_norm": 0.7167915689958744, "learning_rate": 3.3035825502807944e-06, "loss": 0.258, "step": 17992 }, { "epoch": 1.243899066712755, "grad_norm": 0.6606428147814076, "learning_rate": 3.30305600427248e-06, "loss": 0.2854, "step": 17993 }, { "epoch": 1.243968199101279, "grad_norm": 0.6738904029585411, "learning_rate": 3.3025294795323293e-06, "loss": 0.3774, "step": 17994 }, { "epoch": 1.244037331489803, "grad_norm": 0.668837445568804, "learning_rate": 3.302002976066944e-06, "loss": 0.2977, "step": 17995 }, { "epoch": 1.244106463878327, "grad_norm": 0.6179943245486361, "learning_rate": 3.301476493882919e-06, "loss": 0.2766, "step": 17996 }, { "epoch": 1.244175596266851, "grad_norm": 0.6280475382071344, "learning_rate": 3.300950032986857e-06, "loss": 0.1969, "step": 17997 }, { "epoch": 1.244244728655375, "grad_norm": 2.031170966711998, "learning_rate": 3.3004235933853523e-06, "loss": 0.3568, "step": 17998 }, { "epoch": 1.244313861043899, "grad_norm": 0.6020811702811815, "learning_rate": 3.2998971750850074e-06, "loss": 0.2791, "step": 17999 }, { "epoch": 1.244382993432423, "grad_norm": 0.6156831284716694, "learning_rate": 3.299370778092415e-06, "loss": 0.2119, "step": 18000 }, { "epoch": 1.244452125820947, "grad_norm": 0.602762152102698, "learning_rate": 3.2988444024141762e-06, "loss": 0.258, "step": 18001 }, { "epoch": 1.2445212582094711, "grad_norm": 0.655864353847921, "learning_rate": 3.2983180480568855e-06, "loss": 0.2967, "step": 18002 }, { "epoch": 1.2445903905979951, "grad_norm": 0.6604841143109567, "learning_rate": 3.2977917150271433e-06, "loss": 0.2522, "step": 18003 }, { "epoch": 1.2446595229865192, "grad_norm": 0.6445558237120764, "learning_rate": 3.297265403331542e-06, "loss": 0.2429, "step": 18004 }, { "epoch": 1.2447286553750432, "grad_norm": 0.6759635105463495, "learning_rate": 3.2967391129766833e-06, "loss": 0.374, "step": 18005 }, { "epoch": 1.2447977877635672, "grad_norm": 0.6555246202429674, "learning_rate": 3.2962128439691577e-06, "loss": 0.2437, "step": 18006 }, { "epoch": 1.2448669201520912, "grad_norm": 0.6629168223115616, "learning_rate": 3.295686596315565e-06, "loss": 0.2529, "step": 18007 }, { "epoch": 1.2449360525406152, "grad_norm": 0.7207127063645413, "learning_rate": 3.295160370022498e-06, "loss": 0.2855, "step": 18008 }, { "epoch": 1.2450051849291393, "grad_norm": 0.6865071014510016, "learning_rate": 3.2946341650965553e-06, "loss": 0.3346, "step": 18009 }, { "epoch": 1.2450743173176633, "grad_norm": 0.6778885254937606, "learning_rate": 3.294107981544328e-06, "loss": 0.2493, "step": 18010 }, { "epoch": 1.2451434497061873, "grad_norm": 0.614060744778995, "learning_rate": 3.293581819372413e-06, "loss": 0.2595, "step": 18011 }, { "epoch": 1.2452125820947113, "grad_norm": 0.6693228859071856, "learning_rate": 3.2930556785874045e-06, "loss": 0.3733, "step": 18012 }, { "epoch": 1.2452817144832353, "grad_norm": 4.864492803089126, "learning_rate": 3.2925295591958997e-06, "loss": 0.3322, "step": 18013 }, { "epoch": 1.2453508468717593, "grad_norm": 0.69917057166242, "learning_rate": 3.292003461204486e-06, "loss": 0.2644, "step": 18014 }, { "epoch": 1.2454199792602834, "grad_norm": 0.9099535632201845, "learning_rate": 3.2914773846197633e-06, "loss": 0.3019, "step": 18015 }, { "epoch": 1.2454891116488074, "grad_norm": 0.753711887044291, "learning_rate": 3.290951329448321e-06, "loss": 0.3636, "step": 18016 }, { "epoch": 1.2455582440373314, "grad_norm": 0.6645299841054777, "learning_rate": 3.290425295696756e-06, "loss": 0.272, "step": 18017 }, { "epoch": 1.2456273764258556, "grad_norm": 0.5903884962580191, "learning_rate": 3.289899283371657e-06, "loss": 0.1954, "step": 18018 }, { "epoch": 1.2456965088143797, "grad_norm": 0.7823001793121465, "learning_rate": 3.2893732924796196e-06, "loss": 0.3647, "step": 18019 }, { "epoch": 1.2457656412029037, "grad_norm": 0.6740483546794132, "learning_rate": 3.288847323027234e-06, "loss": 0.2929, "step": 18020 }, { "epoch": 1.2458347735914277, "grad_norm": 0.6031848065890028, "learning_rate": 3.288321375021095e-06, "loss": 0.2765, "step": 18021 }, { "epoch": 1.2459039059799517, "grad_norm": 0.6469543034711926, "learning_rate": 3.2877954484677914e-06, "loss": 0.2798, "step": 18022 }, { "epoch": 1.2459730383684757, "grad_norm": 0.6996377660675399, "learning_rate": 3.287269543373919e-06, "loss": 0.3141, "step": 18023 }, { "epoch": 1.2460421707569997, "grad_norm": 0.6903503313039024, "learning_rate": 3.2867436597460635e-06, "loss": 0.2453, "step": 18024 }, { "epoch": 1.2461113031455238, "grad_norm": 0.6551502528478801, "learning_rate": 3.2862177975908204e-06, "loss": 0.2602, "step": 18025 }, { "epoch": 1.2461804355340478, "grad_norm": 0.679754588639249, "learning_rate": 3.2856919569147793e-06, "loss": 0.3, "step": 18026 }, { "epoch": 1.2462495679225718, "grad_norm": 0.6795575558151309, "learning_rate": 3.2851661377245293e-06, "loss": 0.367, "step": 18027 }, { "epoch": 1.2463187003110958, "grad_norm": 0.6969471501213057, "learning_rate": 3.2846403400266613e-06, "loss": 0.2144, "step": 18028 }, { "epoch": 1.2463878326996198, "grad_norm": 0.6709987699903469, "learning_rate": 3.284114563827765e-06, "loss": 0.2089, "step": 18029 }, { "epoch": 1.2464569650881439, "grad_norm": 0.6102857588306048, "learning_rate": 3.2835888091344328e-06, "loss": 0.3229, "step": 18030 }, { "epoch": 1.2465260974766679, "grad_norm": 0.6799510737962386, "learning_rate": 3.2830630759532487e-06, "loss": 0.2519, "step": 18031 }, { "epoch": 1.246595229865192, "grad_norm": 0.7835859008177428, "learning_rate": 3.282537364290807e-06, "loss": 0.2247, "step": 18032 }, { "epoch": 1.246664362253716, "grad_norm": 0.6694958486585381, "learning_rate": 3.2820116741536936e-06, "loss": 0.2376, "step": 18033 }, { "epoch": 1.24673349464224, "grad_norm": 0.6724530205388471, "learning_rate": 3.2814860055484985e-06, "loss": 0.304, "step": 18034 }, { "epoch": 1.246802627030764, "grad_norm": 0.74954252914463, "learning_rate": 3.2809603584818097e-06, "loss": 0.3126, "step": 18035 }, { "epoch": 1.246871759419288, "grad_norm": 0.7224973051202298, "learning_rate": 3.2804347329602145e-06, "loss": 0.2262, "step": 18036 }, { "epoch": 1.246940891807812, "grad_norm": 0.7089602143740221, "learning_rate": 3.2799091289903007e-06, "loss": 0.342, "step": 18037 }, { "epoch": 1.247010024196336, "grad_norm": 0.628094528223992, "learning_rate": 3.279383546578658e-06, "loss": 0.2864, "step": 18038 }, { "epoch": 1.24707915658486, "grad_norm": 0.6170012220156438, "learning_rate": 3.278857985731869e-06, "loss": 0.2181, "step": 18039 }, { "epoch": 1.247148288973384, "grad_norm": 0.843482311171, "learning_rate": 3.278332446456527e-06, "loss": 0.2558, "step": 18040 }, { "epoch": 1.247217421361908, "grad_norm": 0.6596681195806081, "learning_rate": 3.277806928759213e-06, "loss": 0.3241, "step": 18041 }, { "epoch": 1.247286553750432, "grad_norm": 0.7209168483506945, "learning_rate": 3.277281432646517e-06, "loss": 0.2823, "step": 18042 }, { "epoch": 1.247355686138956, "grad_norm": 0.6620941728076952, "learning_rate": 3.276755958125023e-06, "loss": 0.227, "step": 18043 }, { "epoch": 1.24742481852748, "grad_norm": 0.6417053433375125, "learning_rate": 3.2762305052013195e-06, "loss": 0.2839, "step": 18044 }, { "epoch": 1.2474939509160041, "grad_norm": 0.6358493873995404, "learning_rate": 3.2757050738819886e-06, "loss": 0.3113, "step": 18045 }, { "epoch": 1.2475630833045281, "grad_norm": 0.623179171051731, "learning_rate": 3.275179664173618e-06, "loss": 0.2416, "step": 18046 }, { "epoch": 1.2476322156930522, "grad_norm": 0.6560011041134814, "learning_rate": 3.2746542760827914e-06, "loss": 0.2489, "step": 18047 }, { "epoch": 1.2477013480815762, "grad_norm": 0.7069950934794026, "learning_rate": 3.2741289096160965e-06, "loss": 0.3435, "step": 18048 }, { "epoch": 1.2477704804701002, "grad_norm": 0.6894786424906597, "learning_rate": 3.2736035647801145e-06, "loss": 0.2497, "step": 18049 }, { "epoch": 1.2478396128586242, "grad_norm": 0.6418325223186554, "learning_rate": 3.273078241581431e-06, "loss": 0.2269, "step": 18050 }, { "epoch": 1.2479087452471482, "grad_norm": 0.6624233878984107, "learning_rate": 3.2725529400266292e-06, "loss": 0.3063, "step": 18051 }, { "epoch": 1.2479778776356723, "grad_norm": 0.7200163609419955, "learning_rate": 3.2720276601222956e-06, "loss": 0.3655, "step": 18052 }, { "epoch": 1.2480470100241963, "grad_norm": 0.6213215774481138, "learning_rate": 3.2715024018750097e-06, "loss": 0.2615, "step": 18053 }, { "epoch": 1.2481161424127203, "grad_norm": 0.6414447620110328, "learning_rate": 3.270977165291357e-06, "loss": 0.284, "step": 18054 }, { "epoch": 1.2481852748012443, "grad_norm": 0.7515707351275057, "learning_rate": 3.2704519503779197e-06, "loss": 0.3188, "step": 18055 }, { "epoch": 1.2482544071897683, "grad_norm": 0.6950535335599959, "learning_rate": 3.2699267571412823e-06, "loss": 0.3178, "step": 18056 }, { "epoch": 1.2483235395782923, "grad_norm": 1.7325536721934571, "learning_rate": 3.2694015855880223e-06, "loss": 0.2256, "step": 18057 }, { "epoch": 1.2483926719668164, "grad_norm": 1.3858641296342686, "learning_rate": 3.268876435724728e-06, "loss": 0.29, "step": 18058 }, { "epoch": 1.2484618043553404, "grad_norm": 0.6608802916610462, "learning_rate": 3.268351307557977e-06, "loss": 0.3634, "step": 18059 }, { "epoch": 1.2485309367438644, "grad_norm": 1.1074171979379523, "learning_rate": 3.267826201094353e-06, "loss": 0.287, "step": 18060 }, { "epoch": 1.2486000691323884, "grad_norm": 0.6146613395599029, "learning_rate": 3.2673011163404343e-06, "loss": 0.2333, "step": 18061 }, { "epoch": 1.2486692015209124, "grad_norm": 0.6324826551569862, "learning_rate": 3.266776053302807e-06, "loss": 0.3115, "step": 18062 }, { "epoch": 1.2487383339094364, "grad_norm": 0.6885023594962968, "learning_rate": 3.2662510119880464e-06, "loss": 0.3035, "step": 18063 }, { "epoch": 1.2488074662979607, "grad_norm": 0.7209731828159226, "learning_rate": 3.2657259924027364e-06, "loss": 0.2597, "step": 18064 }, { "epoch": 1.2488765986864847, "grad_norm": 0.6154514318211086, "learning_rate": 3.2652009945534552e-06, "loss": 0.2657, "step": 18065 }, { "epoch": 1.2489457310750087, "grad_norm": 0.6696967195650875, "learning_rate": 3.2646760184467856e-06, "loss": 0.4203, "step": 18066 }, { "epoch": 1.2490148634635327, "grad_norm": 0.6472669343491149, "learning_rate": 3.264151064089304e-06, "loss": 0.2562, "step": 18067 }, { "epoch": 1.2490839958520568, "grad_norm": 0.6638451611651178, "learning_rate": 3.2636261314875907e-06, "loss": 0.2436, "step": 18068 }, { "epoch": 1.2491531282405808, "grad_norm": 0.798994796089363, "learning_rate": 3.263101220648225e-06, "loss": 0.3225, "step": 18069 }, { "epoch": 1.2492222606291048, "grad_norm": 0.6315933965913092, "learning_rate": 3.2625763315777876e-06, "loss": 0.3102, "step": 18070 }, { "epoch": 1.2492913930176288, "grad_norm": 0.677133286754776, "learning_rate": 3.2620514642828533e-06, "loss": 0.2105, "step": 18071 }, { "epoch": 1.2493605254061528, "grad_norm": 0.6536103621017193, "learning_rate": 3.2615266187700036e-06, "loss": 0.2723, "step": 18072 }, { "epoch": 1.2494296577946769, "grad_norm": 0.669960400218249, "learning_rate": 3.261001795045814e-06, "loss": 0.2943, "step": 18073 }, { "epoch": 1.2494987901832009, "grad_norm": 0.6272873637939615, "learning_rate": 3.2604769931168645e-06, "loss": 0.2263, "step": 18074 }, { "epoch": 1.2495679225717249, "grad_norm": 0.6062645502128913, "learning_rate": 3.259952212989731e-06, "loss": 0.2059, "step": 18075 }, { "epoch": 1.249637054960249, "grad_norm": 0.7029232519259807, "learning_rate": 3.2594274546709923e-06, "loss": 0.3236, "step": 18076 }, { "epoch": 1.249706187348773, "grad_norm": 0.705318590147904, "learning_rate": 3.2589027181672235e-06, "loss": 0.3511, "step": 18077 }, { "epoch": 1.249775319737297, "grad_norm": 0.7078190579107019, "learning_rate": 3.258378003485002e-06, "loss": 0.2806, "step": 18078 }, { "epoch": 1.249844452125821, "grad_norm": 0.6189362895643007, "learning_rate": 3.257853310630904e-06, "loss": 0.2824, "step": 18079 }, { "epoch": 1.249913584514345, "grad_norm": 0.8673848060646072, "learning_rate": 3.2573286396115073e-06, "loss": 0.4029, "step": 18080 }, { "epoch": 1.249982716902869, "grad_norm": 0.7005629277602584, "learning_rate": 3.256803990433384e-06, "loss": 0.284, "step": 18081 }, { "epoch": 1.250051849291393, "grad_norm": 0.7446583553103108, "learning_rate": 3.256279363103113e-06, "loss": 0.2574, "step": 18082 }, { "epoch": 1.250120981679917, "grad_norm": 0.6942694895989311, "learning_rate": 3.2557547576272676e-06, "loss": 0.2763, "step": 18083 }, { "epoch": 1.250190114068441, "grad_norm": 0.6516950644704754, "learning_rate": 3.2552301740124253e-06, "loss": 0.3472, "step": 18084 }, { "epoch": 1.250259246456965, "grad_norm": 0.6518971915042904, "learning_rate": 3.2547056122651573e-06, "loss": 0.2514, "step": 18085 }, { "epoch": 1.250328378845489, "grad_norm": 0.7105696002425657, "learning_rate": 3.254181072392041e-06, "loss": 0.2177, "step": 18086 }, { "epoch": 1.250397511234013, "grad_norm": 0.7151065710752046, "learning_rate": 3.253656554399648e-06, "loss": 0.4011, "step": 18087 }, { "epoch": 1.2504666436225371, "grad_norm": 0.7636800450714674, "learning_rate": 3.253132058294556e-06, "loss": 0.2898, "step": 18088 }, { "epoch": 1.2505357760110611, "grad_norm": 0.6270659081954651, "learning_rate": 3.252607584083333e-06, "loss": 0.2584, "step": 18089 }, { "epoch": 1.2506049083995852, "grad_norm": 0.6642946413580031, "learning_rate": 3.2520831317725576e-06, "loss": 0.2888, "step": 18090 }, { "epoch": 1.2506740407881092, "grad_norm": 0.6775653455270217, "learning_rate": 3.2515587013687998e-06, "loss": 0.345, "step": 18091 }, { "epoch": 1.2507431731766332, "grad_norm": 0.6633454959544453, "learning_rate": 3.2510342928786342e-06, "loss": 0.2265, "step": 18092 }, { "epoch": 1.2508123055651572, "grad_norm": 0.6142526820356751, "learning_rate": 3.2505099063086344e-06, "loss": 0.1995, "step": 18093 }, { "epoch": 1.2508814379536812, "grad_norm": 0.6639583667830411, "learning_rate": 3.249985541665368e-06, "loss": 0.347, "step": 18094 }, { "epoch": 1.2509505703422052, "grad_norm": 0.6438168772383653, "learning_rate": 3.249461198955411e-06, "loss": 0.3352, "step": 18095 }, { "epoch": 1.2510197027307293, "grad_norm": 0.6406567517780037, "learning_rate": 3.2489368781853326e-06, "loss": 0.216, "step": 18096 }, { "epoch": 1.2510888351192535, "grad_norm": 0.6229336638085508, "learning_rate": 3.2484125793617077e-06, "loss": 0.2578, "step": 18097 }, { "epoch": 1.2511579675077775, "grad_norm": 0.8584115971782954, "learning_rate": 3.2478883024911033e-06, "loss": 0.3355, "step": 18098 }, { "epoch": 1.2512270998963015, "grad_norm": 0.7534080066618974, "learning_rate": 3.2473640475800936e-06, "loss": 0.2546, "step": 18099 }, { "epoch": 1.2512962322848256, "grad_norm": 0.9253080376082166, "learning_rate": 3.2468398146352465e-06, "loss": 0.2006, "step": 18100 }, { "epoch": 1.2513653646733496, "grad_norm": 0.6676050976382057, "learning_rate": 3.2463156036631354e-06, "loss": 0.2977, "step": 18101 }, { "epoch": 1.2514344970618736, "grad_norm": 0.7060307731701958, "learning_rate": 3.2457914146703274e-06, "loss": 0.3094, "step": 18102 }, { "epoch": 1.2515036294503976, "grad_norm": 0.6647186518335659, "learning_rate": 3.245267247663394e-06, "loss": 0.2263, "step": 18103 }, { "epoch": 1.2515727618389216, "grad_norm": 0.9478300377675342, "learning_rate": 3.244743102648903e-06, "loss": 0.2531, "step": 18104 }, { "epoch": 1.2516418942274457, "grad_norm": 0.6826533740987962, "learning_rate": 3.2442189796334266e-06, "loss": 0.3692, "step": 18105 }, { "epoch": 1.2517110266159697, "grad_norm": 0.7159782263021389, "learning_rate": 3.2436948786235304e-06, "loss": 0.284, "step": 18106 }, { "epoch": 1.2517801590044937, "grad_norm": 0.6513884358114324, "learning_rate": 3.2431707996257856e-06, "loss": 0.2427, "step": 18107 }, { "epoch": 1.2518492913930177, "grad_norm": 0.7268123398790882, "learning_rate": 3.242646742646759e-06, "loss": 0.2984, "step": 18108 }, { "epoch": 1.2519184237815417, "grad_norm": 0.7469945320955238, "learning_rate": 3.2421227076930205e-06, "loss": 0.3433, "step": 18109 }, { "epoch": 1.2519875561700657, "grad_norm": 0.6703396862587743, "learning_rate": 3.2415986947711343e-06, "loss": 0.2191, "step": 18110 }, { "epoch": 1.2520566885585898, "grad_norm": 0.6377910775369039, "learning_rate": 3.241074703887674e-06, "loss": 0.2745, "step": 18111 }, { "epoch": 1.2521258209471138, "grad_norm": 0.7200320059034635, "learning_rate": 3.2405507350492004e-06, "loss": 0.3683, "step": 18112 }, { "epoch": 1.2521949533356378, "grad_norm": 0.7291848965717713, "learning_rate": 3.2400267882622847e-06, "loss": 0.2999, "step": 18113 }, { "epoch": 1.2522640857241618, "grad_norm": 0.6017255122221122, "learning_rate": 3.2395028635334914e-06, "loss": 0.189, "step": 18114 }, { "epoch": 1.2523332181126858, "grad_norm": 0.7445751709593611, "learning_rate": 3.23897896086939e-06, "loss": 0.2861, "step": 18115 }, { "epoch": 1.2524023505012098, "grad_norm": 0.7412947666360082, "learning_rate": 3.238455080276543e-06, "loss": 0.3601, "step": 18116 }, { "epoch": 1.2524714828897339, "grad_norm": 0.6438742788616667, "learning_rate": 3.237931221761519e-06, "loss": 0.2871, "step": 18117 }, { "epoch": 1.2525406152782579, "grad_norm": 0.7259529770481323, "learning_rate": 3.237407385330881e-06, "loss": 0.2221, "step": 18118 }, { "epoch": 1.252609747666782, "grad_norm": 0.6633932276788195, "learning_rate": 3.236883570991198e-06, "loss": 0.3143, "step": 18119 }, { "epoch": 1.252678880055306, "grad_norm": 0.6606829807774673, "learning_rate": 3.2363597787490313e-06, "loss": 0.2816, "step": 18120 }, { "epoch": 1.25274801244383, "grad_norm": 0.6735114319170704, "learning_rate": 3.2358360086109485e-06, "loss": 0.2403, "step": 18121 }, { "epoch": 1.252817144832354, "grad_norm": 0.685199988535868, "learning_rate": 3.235312260583512e-06, "loss": 0.2987, "step": 18122 }, { "epoch": 1.252886277220878, "grad_norm": 0.6741271312269557, "learning_rate": 3.2347885346732888e-06, "loss": 0.3829, "step": 18123 }, { "epoch": 1.252955409609402, "grad_norm": 0.6755197509874415, "learning_rate": 3.234264830886839e-06, "loss": 0.2543, "step": 18124 }, { "epoch": 1.253024541997926, "grad_norm": 0.6697590884905454, "learning_rate": 3.23374114923073e-06, "loss": 0.2612, "step": 18125 }, { "epoch": 1.25309367438645, "grad_norm": 0.7014434327926989, "learning_rate": 3.233217489711522e-06, "loss": 0.3121, "step": 18126 }, { "epoch": 1.253162806774974, "grad_norm": 0.7258431051331813, "learning_rate": 3.232693852335782e-06, "loss": 0.3567, "step": 18127 }, { "epoch": 1.253231939163498, "grad_norm": 0.6475016398767004, "learning_rate": 3.232170237110068e-06, "loss": 0.2723, "step": 18128 }, { "epoch": 1.253301071552022, "grad_norm": 0.6708556509983885, "learning_rate": 3.2316466440409484e-06, "loss": 0.3013, "step": 18129 }, { "epoch": 1.253370203940546, "grad_norm": 0.710343646878416, "learning_rate": 3.23112307313498e-06, "loss": 0.4102, "step": 18130 }, { "epoch": 1.2534393363290701, "grad_norm": 0.6741834275204983, "learning_rate": 3.230599524398728e-06, "loss": 0.2707, "step": 18131 }, { "epoch": 1.2535084687175941, "grad_norm": 0.683245658178686, "learning_rate": 3.230075997838753e-06, "loss": 0.2756, "step": 18132 }, { "epoch": 1.2535776011061182, "grad_norm": 0.7032138707297715, "learning_rate": 3.2295524934616185e-06, "loss": 0.3042, "step": 18133 }, { "epoch": 1.2536467334946422, "grad_norm": 0.7406383514918092, "learning_rate": 3.2290290112738816e-06, "loss": 0.3616, "step": 18134 }, { "epoch": 1.2537158658831662, "grad_norm": 0.5805723744072486, "learning_rate": 3.228505551282107e-06, "loss": 0.2364, "step": 18135 }, { "epoch": 1.2537849982716902, "grad_norm": 0.7862641178035557, "learning_rate": 3.2279821134928525e-06, "loss": 0.2596, "step": 18136 }, { "epoch": 1.2538541306602142, "grad_norm": 0.6467340458229601, "learning_rate": 3.2274586979126822e-06, "loss": 0.3338, "step": 18137 }, { "epoch": 1.2539232630487382, "grad_norm": 2.585325416737953, "learning_rate": 3.226935304548152e-06, "loss": 0.2903, "step": 18138 }, { "epoch": 1.2539923954372623, "grad_norm": 0.636360634907335, "learning_rate": 3.226411933405824e-06, "loss": 0.2257, "step": 18139 }, { "epoch": 1.2540615278257863, "grad_norm": 0.7553826822380613, "learning_rate": 3.225888584492257e-06, "loss": 0.299, "step": 18140 }, { "epoch": 1.2541306602143103, "grad_norm": 0.7490604201448179, "learning_rate": 3.2253652578140128e-06, "loss": 0.3226, "step": 18141 }, { "epoch": 1.2541997926028343, "grad_norm": 0.6292594412463269, "learning_rate": 3.2248419533776455e-06, "loss": 0.1961, "step": 18142 }, { "epoch": 1.2542689249913583, "grad_norm": 0.6470025445454708, "learning_rate": 3.2243186711897174e-06, "loss": 0.2312, "step": 18143 }, { "epoch": 1.2543380573798824, "grad_norm": 0.7026079500277336, "learning_rate": 3.2237954112567847e-06, "loss": 0.2904, "step": 18144 }, { "epoch": 1.2544071897684064, "grad_norm": 0.7182548362129614, "learning_rate": 3.223272173585409e-06, "loss": 0.3659, "step": 18145 }, { "epoch": 1.2544763221569304, "grad_norm": 0.6395132096354217, "learning_rate": 3.222748958182143e-06, "loss": 0.2077, "step": 18146 }, { "epoch": 1.2545454545454544, "grad_norm": 0.6560010446537443, "learning_rate": 3.2222257650535503e-06, "loss": 0.2615, "step": 18147 }, { "epoch": 1.2546145869339786, "grad_norm": 0.6778469453053777, "learning_rate": 3.221702594206183e-06, "loss": 0.331, "step": 18148 }, { "epoch": 1.2546837193225027, "grad_norm": 0.7159135598560765, "learning_rate": 3.2211794456466004e-06, "loss": 0.2596, "step": 18149 }, { "epoch": 1.2547528517110267, "grad_norm": 0.8841727046438194, "learning_rate": 3.2206563193813588e-06, "loss": 0.2468, "step": 18150 }, { "epoch": 1.2548219840995507, "grad_norm": 0.8419678801324312, "learning_rate": 3.2201332154170174e-06, "loss": 0.3036, "step": 18151 }, { "epoch": 1.2548911164880747, "grad_norm": 1.448401355435677, "learning_rate": 3.2196101337601272e-06, "loss": 0.2652, "step": 18152 }, { "epoch": 1.2549602488765987, "grad_norm": 0.7252353004946234, "learning_rate": 3.2190870744172477e-06, "loss": 0.2448, "step": 18153 }, { "epoch": 1.2550293812651228, "grad_norm": 0.7310502881289489, "learning_rate": 3.2185640373949324e-06, "loss": 0.2566, "step": 18154 }, { "epoch": 1.2550985136536468, "grad_norm": 0.6527711554451014, "learning_rate": 3.2180410226997393e-06, "loss": 0.3597, "step": 18155 }, { "epoch": 1.2551676460421708, "grad_norm": 0.6759106317902852, "learning_rate": 3.2175180303382203e-06, "loss": 0.3522, "step": 18156 }, { "epoch": 1.2552367784306948, "grad_norm": 0.603933834493163, "learning_rate": 3.216995060316933e-06, "loss": 0.2977, "step": 18157 }, { "epoch": 1.2553059108192188, "grad_norm": 0.6534736535587939, "learning_rate": 3.21647211264243e-06, "loss": 0.261, "step": 18158 }, { "epoch": 1.2553750432077428, "grad_norm": 0.7608029780200117, "learning_rate": 3.215949187321268e-06, "loss": 0.3377, "step": 18159 }, { "epoch": 1.2554441755962669, "grad_norm": 0.6471060496928572, "learning_rate": 3.215426284359998e-06, "loss": 0.2407, "step": 18160 }, { "epoch": 1.2555133079847909, "grad_norm": 0.6911329786706902, "learning_rate": 3.2149034037651736e-06, "loss": 0.271, "step": 18161 }, { "epoch": 1.255582440373315, "grad_norm": 0.7554392893475835, "learning_rate": 3.214380545543352e-06, "loss": 0.3184, "step": 18162 }, { "epoch": 1.255651572761839, "grad_norm": 0.7015977923007818, "learning_rate": 3.2138577097010803e-06, "loss": 0.3199, "step": 18163 }, { "epoch": 1.255720705150363, "grad_norm": 0.602208781379701, "learning_rate": 3.213334896244918e-06, "loss": 0.2364, "step": 18164 }, { "epoch": 1.255789837538887, "grad_norm": 0.6819222113691082, "learning_rate": 3.212812105181412e-06, "loss": 0.2755, "step": 18165 }, { "epoch": 1.255858969927411, "grad_norm": 1.260184966707024, "learning_rate": 3.2122893365171183e-06, "loss": 0.3138, "step": 18166 }, { "epoch": 1.255928102315935, "grad_norm": 0.7106582857659208, "learning_rate": 3.211766590258586e-06, "loss": 0.2596, "step": 18167 }, { "epoch": 1.255997234704459, "grad_norm": 0.5808280239566532, "learning_rate": 3.2112438664123703e-06, "loss": 0.2211, "step": 18168 }, { "epoch": 1.256066367092983, "grad_norm": 0.6616687258951609, "learning_rate": 3.2107211649850186e-06, "loss": 0.2962, "step": 18169 }, { "epoch": 1.256135499481507, "grad_norm": 0.7261100543547144, "learning_rate": 3.2101984859830848e-06, "loss": 0.3054, "step": 18170 }, { "epoch": 1.256204631870031, "grad_norm": 0.6515805399425123, "learning_rate": 3.209675829413118e-06, "loss": 0.2656, "step": 18171 }, { "epoch": 1.256273764258555, "grad_norm": 0.7120640715787271, "learning_rate": 3.209153195281672e-06, "loss": 0.2579, "step": 18172 }, { "epoch": 1.256342896647079, "grad_norm": 0.7364483511677757, "learning_rate": 3.2086305835952924e-06, "loss": 0.3633, "step": 18173 }, { "epoch": 1.2564120290356031, "grad_norm": 0.6888192711819168, "learning_rate": 3.208107994360533e-06, "loss": 0.2813, "step": 18174 }, { "epoch": 1.2564811614241271, "grad_norm": 0.6433194446996099, "learning_rate": 3.2075854275839413e-06, "loss": 0.2282, "step": 18175 }, { "epoch": 1.2565502938126512, "grad_norm": 0.7406916419416765, "learning_rate": 3.207062883272069e-06, "loss": 0.2958, "step": 18176 }, { "epoch": 1.2566194262011752, "grad_norm": 0.5959624084942768, "learning_rate": 3.2065403614314615e-06, "loss": 0.2914, "step": 18177 }, { "epoch": 1.2566885585896994, "grad_norm": 0.6473711410077316, "learning_rate": 3.206017862068672e-06, "loss": 0.2177, "step": 18178 }, { "epoch": 1.2567576909782234, "grad_norm": 0.6555910104791279, "learning_rate": 3.205495385190246e-06, "loss": 0.2453, "step": 18179 }, { "epoch": 1.2568268233667474, "grad_norm": 0.689044394067952, "learning_rate": 3.2049729308027345e-06, "loss": 0.3446, "step": 18180 }, { "epoch": 1.2568959557552715, "grad_norm": 0.8375191451392427, "learning_rate": 3.2044504989126807e-06, "loss": 0.2337, "step": 18181 }, { "epoch": 1.2569650881437955, "grad_norm": 0.6445642445374888, "learning_rate": 3.203928089526639e-06, "loss": 0.2027, "step": 18182 }, { "epoch": 1.2570342205323195, "grad_norm": 1.1172814130387083, "learning_rate": 3.2034057026511513e-06, "loss": 0.3043, "step": 18183 }, { "epoch": 1.2571033529208435, "grad_norm": 0.6700299897441434, "learning_rate": 3.202883338292768e-06, "loss": 0.2933, "step": 18184 }, { "epoch": 1.2571724853093675, "grad_norm": 0.7145567443283732, "learning_rate": 3.202360996458034e-06, "loss": 0.2348, "step": 18185 }, { "epoch": 1.2572416176978916, "grad_norm": 0.7619135986919225, "learning_rate": 3.201838677153498e-06, "loss": 0.2699, "step": 18186 }, { "epoch": 1.2573107500864156, "grad_norm": 0.7198494395498918, "learning_rate": 3.2013163803857044e-06, "loss": 0.3462, "step": 18187 }, { "epoch": 1.2573798824749396, "grad_norm": 0.6832661842105199, "learning_rate": 3.200794106161201e-06, "loss": 0.3095, "step": 18188 }, { "epoch": 1.2574490148634636, "grad_norm": 0.7262640227219869, "learning_rate": 3.200271854486531e-06, "loss": 0.2464, "step": 18189 }, { "epoch": 1.2575181472519876, "grad_norm": 0.6566492927629339, "learning_rate": 3.199749625368244e-06, "loss": 0.304, "step": 18190 }, { "epoch": 1.2575872796405116, "grad_norm": 0.7842360342473267, "learning_rate": 3.199227418812881e-06, "loss": 0.3423, "step": 18191 }, { "epoch": 1.2576564120290357, "grad_norm": 0.6256832429596272, "learning_rate": 3.198705234826989e-06, "loss": 0.2777, "step": 18192 }, { "epoch": 1.2577255444175597, "grad_norm": 0.6713497229596741, "learning_rate": 3.1981830734171117e-06, "loss": 0.2982, "step": 18193 }, { "epoch": 1.2577946768060837, "grad_norm": 0.7271610375988092, "learning_rate": 3.1976609345897957e-06, "loss": 0.3062, "step": 18194 }, { "epoch": 1.2578638091946077, "grad_norm": 0.6609442712163492, "learning_rate": 3.197138818351582e-06, "loss": 0.3512, "step": 18195 }, { "epoch": 1.2579329415831317, "grad_norm": 0.7028053507592249, "learning_rate": 3.1966167247090167e-06, "loss": 0.2253, "step": 18196 }, { "epoch": 1.2580020739716558, "grad_norm": 0.6535461989805773, "learning_rate": 3.1960946536686416e-06, "loss": 0.2193, "step": 18197 }, { "epoch": 1.2580712063601798, "grad_norm": 0.6286128895299692, "learning_rate": 3.1955726052370028e-06, "loss": 0.303, "step": 18198 }, { "epoch": 1.2581403387487038, "grad_norm": 0.6659257662146455, "learning_rate": 3.1950505794206375e-06, "loss": 0.2727, "step": 18199 }, { "epoch": 1.2582094711372278, "grad_norm": 0.6376945095530346, "learning_rate": 3.194528576226096e-06, "loss": 0.2853, "step": 18200 }, { "epoch": 1.2582786035257518, "grad_norm": 0.8809576985636015, "learning_rate": 3.1940065956599146e-06, "loss": 0.272, "step": 18201 }, { "epoch": 1.2583477359142758, "grad_norm": 0.7544339797744509, "learning_rate": 3.193484637728639e-06, "loss": 0.325, "step": 18202 }, { "epoch": 1.2584168683027999, "grad_norm": 0.7107974083029425, "learning_rate": 3.1929627024388077e-06, "loss": 0.2739, "step": 18203 }, { "epoch": 1.2584860006913239, "grad_norm": 0.8185635011958953, "learning_rate": 3.1924407897969667e-06, "loss": 0.2492, "step": 18204 }, { "epoch": 1.258555133079848, "grad_norm": 0.6544779972520431, "learning_rate": 3.1919188998096525e-06, "loss": 0.3396, "step": 18205 }, { "epoch": 1.258624265468372, "grad_norm": 0.8160088549349469, "learning_rate": 3.1913970324834088e-06, "loss": 0.2627, "step": 18206 }, { "epoch": 1.258693397856896, "grad_norm": 0.8659209790203423, "learning_rate": 3.190875187824775e-06, "loss": 0.2542, "step": 18207 }, { "epoch": 1.25876253024542, "grad_norm": 0.7242199772067582, "learning_rate": 3.190353365840294e-06, "loss": 0.2636, "step": 18208 }, { "epoch": 1.258831662633944, "grad_norm": 0.6838308564084608, "learning_rate": 3.1898315665365033e-06, "loss": 0.3673, "step": 18209 }, { "epoch": 1.258900795022468, "grad_norm": 3.4646512420484923, "learning_rate": 3.1893097899199434e-06, "loss": 0.2601, "step": 18210 }, { "epoch": 1.258969927410992, "grad_norm": 0.6394831869274749, "learning_rate": 3.1887880359971528e-06, "loss": 0.2533, "step": 18211 }, { "epoch": 1.259039059799516, "grad_norm": 0.7765319433874129, "learning_rate": 3.188266304774674e-06, "loss": 0.3725, "step": 18212 }, { "epoch": 1.25910819218804, "grad_norm": 0.6714492598406031, "learning_rate": 3.1877445962590426e-06, "loss": 0.2946, "step": 18213 }, { "epoch": 1.259177324576564, "grad_norm": 0.6279312472974861, "learning_rate": 3.1872229104567994e-06, "loss": 0.2379, "step": 18214 }, { "epoch": 1.259246456965088, "grad_norm": 0.6371837278540213, "learning_rate": 3.1867012473744808e-06, "loss": 0.3011, "step": 18215 }, { "epoch": 1.259315589353612, "grad_norm": 0.7342986054514921, "learning_rate": 3.186179607018628e-06, "loss": 0.3869, "step": 18216 }, { "epoch": 1.2593847217421361, "grad_norm": 0.6908562590879245, "learning_rate": 3.1856579893957734e-06, "loss": 0.2369, "step": 18217 }, { "epoch": 1.2594538541306601, "grad_norm": 0.8138812644326884, "learning_rate": 3.185136394512461e-06, "loss": 0.2714, "step": 18218 }, { "epoch": 1.2595229865191842, "grad_norm": 0.769852524917846, "learning_rate": 3.1846148223752243e-06, "loss": 0.3363, "step": 18219 }, { "epoch": 1.2595921189077082, "grad_norm": 0.707982175048757, "learning_rate": 3.1840932729906015e-06, "loss": 0.3889, "step": 18220 }, { "epoch": 1.2596612512962322, "grad_norm": 0.6833274265893166, "learning_rate": 3.1835717463651285e-06, "loss": 0.2738, "step": 18221 }, { "epoch": 1.2597303836847562, "grad_norm": 0.6173152841940048, "learning_rate": 3.1830502425053434e-06, "loss": 0.2008, "step": 18222 }, { "epoch": 1.2597995160732802, "grad_norm": 0.6525865351830674, "learning_rate": 3.182528761417779e-06, "loss": 0.301, "step": 18223 }, { "epoch": 1.2598686484618042, "grad_norm": 0.6697539703714311, "learning_rate": 3.182007303108974e-06, "loss": 0.2564, "step": 18224 }, { "epoch": 1.2599377808503283, "grad_norm": 0.6624542065714005, "learning_rate": 3.181485867585464e-06, "loss": 0.2904, "step": 18225 }, { "epoch": 1.2600069132388523, "grad_norm": 1.0510244178873611, "learning_rate": 3.1809644548537822e-06, "loss": 0.2895, "step": 18226 }, { "epoch": 1.2600760456273763, "grad_norm": 0.6780498828007816, "learning_rate": 3.180443064920465e-06, "loss": 0.3382, "step": 18227 }, { "epoch": 1.2601451780159003, "grad_norm": 0.6409168153098908, "learning_rate": 3.1799216977920466e-06, "loss": 0.2313, "step": 18228 }, { "epoch": 1.2602143104044246, "grad_norm": 0.8339272143857964, "learning_rate": 3.179400353475063e-06, "loss": 0.2242, "step": 18229 }, { "epoch": 1.2602834427929486, "grad_norm": 0.7103333155668431, "learning_rate": 3.1788790319760454e-06, "loss": 0.3301, "step": 18230 }, { "epoch": 1.2603525751814726, "grad_norm": 0.6983306112993087, "learning_rate": 3.178357733301529e-06, "loss": 0.2366, "step": 18231 }, { "epoch": 1.2604217075699966, "grad_norm": 0.6495619961412669, "learning_rate": 3.1778364574580478e-06, "loss": 0.2167, "step": 18232 }, { "epoch": 1.2604908399585206, "grad_norm": 0.6852730016520991, "learning_rate": 3.1773152044521366e-06, "loss": 0.2999, "step": 18233 }, { "epoch": 1.2605599723470446, "grad_norm": 0.7248406828075629, "learning_rate": 3.1767939742903233e-06, "loss": 0.4241, "step": 18234 }, { "epoch": 1.2606291047355687, "grad_norm": 0.6873817029163352, "learning_rate": 3.1762727669791473e-06, "loss": 0.3003, "step": 18235 }, { "epoch": 1.2606982371240927, "grad_norm": 0.6301282204594846, "learning_rate": 3.175751582525135e-06, "loss": 0.3, "step": 18236 }, { "epoch": 1.2607673695126167, "grad_norm": 0.6509922909932551, "learning_rate": 3.1752304209348228e-06, "loss": 0.3426, "step": 18237 }, { "epoch": 1.2608365019011407, "grad_norm": 0.6818008517123172, "learning_rate": 3.1747092822147397e-06, "loss": 0.3313, "step": 18238 }, { "epoch": 1.2609056342896647, "grad_norm": 0.5929801408071113, "learning_rate": 3.17418816637142e-06, "loss": 0.2347, "step": 18239 }, { "epoch": 1.2609747666781888, "grad_norm": 0.7002764172189585, "learning_rate": 3.1736670734113913e-06, "loss": 0.3154, "step": 18240 }, { "epoch": 1.2610438990667128, "grad_norm": 0.644033850868364, "learning_rate": 3.1731460033411877e-06, "loss": 0.3419, "step": 18241 }, { "epoch": 1.2611130314552368, "grad_norm": 0.6908342535323915, "learning_rate": 3.172624956167337e-06, "loss": 0.2832, "step": 18242 }, { "epoch": 1.2611821638437608, "grad_norm": 0.6153286330473422, "learning_rate": 3.172103931896375e-06, "loss": 0.2449, "step": 18243 }, { "epoch": 1.2612512962322848, "grad_norm": 0.7049236745081904, "learning_rate": 3.1715829305348255e-06, "loss": 0.3244, "step": 18244 }, { "epoch": 1.2613204286208088, "grad_norm": 0.7096826765993437, "learning_rate": 3.1710619520892214e-06, "loss": 0.3481, "step": 18245 }, { "epoch": 1.2613895610093329, "grad_norm": 0.6801181769635283, "learning_rate": 3.1705409965660916e-06, "loss": 0.2925, "step": 18246 }, { "epoch": 1.2614586933978569, "grad_norm": 0.6545231555410166, "learning_rate": 3.170020063971967e-06, "loss": 0.2522, "step": 18247 }, { "epoch": 1.261527825786381, "grad_norm": 0.6791173602213839, "learning_rate": 3.1694991543133734e-06, "loss": 0.3805, "step": 18248 }, { "epoch": 1.261596958174905, "grad_norm": 0.7922762563399492, "learning_rate": 3.168978267596843e-06, "loss": 0.2429, "step": 18249 }, { "epoch": 1.261666090563429, "grad_norm": 0.5793702969210109, "learning_rate": 3.1684574038289005e-06, "loss": 0.2491, "step": 18250 }, { "epoch": 1.261735222951953, "grad_norm": 0.6211130552511119, "learning_rate": 3.167936563016078e-06, "loss": 0.3192, "step": 18251 }, { "epoch": 1.261804355340477, "grad_norm": 0.6751920837535316, "learning_rate": 3.167415745164898e-06, "loss": 0.3129, "step": 18252 }, { "epoch": 1.261873487729001, "grad_norm": 0.8703887708992651, "learning_rate": 3.1668949502818945e-06, "loss": 0.2305, "step": 18253 }, { "epoch": 1.261942620117525, "grad_norm": 0.7387180680684184, "learning_rate": 3.1663741783735902e-06, "loss": 0.2857, "step": 18254 }, { "epoch": 1.262011752506049, "grad_norm": 0.7269440193192067, "learning_rate": 3.165853429446514e-06, "loss": 0.3943, "step": 18255 }, { "epoch": 1.262080884894573, "grad_norm": 0.7293605320122201, "learning_rate": 3.16533270350719e-06, "loss": 0.2999, "step": 18256 }, { "epoch": 1.262150017283097, "grad_norm": 0.6480840782025116, "learning_rate": 3.1648120005621487e-06, "loss": 0.2484, "step": 18257 }, { "epoch": 1.262219149671621, "grad_norm": 0.644606402069625, "learning_rate": 3.164291320617913e-06, "loss": 0.2549, "step": 18258 }, { "epoch": 1.262288282060145, "grad_norm": 0.67468231460205, "learning_rate": 3.1637706636810095e-06, "loss": 0.3658, "step": 18259 }, { "epoch": 1.2623574144486693, "grad_norm": 0.7211946494035075, "learning_rate": 3.1632500297579625e-06, "loss": 0.2543, "step": 18260 }, { "epoch": 1.2624265468371934, "grad_norm": 1.515921491238675, "learning_rate": 3.1627294188553017e-06, "loss": 0.2521, "step": 18261 }, { "epoch": 1.2624956792257174, "grad_norm": 0.7405405148846229, "learning_rate": 3.162208830979546e-06, "loss": 0.3786, "step": 18262 }, { "epoch": 1.2625648116142414, "grad_norm": 0.697030009683303, "learning_rate": 3.161688266137225e-06, "loss": 0.2752, "step": 18263 }, { "epoch": 1.2626339440027654, "grad_norm": 0.6866531067666992, "learning_rate": 3.1611677243348594e-06, "loss": 0.2533, "step": 18264 }, { "epoch": 1.2627030763912894, "grad_norm": 0.6856776143095819, "learning_rate": 3.160647205578976e-06, "loss": 0.2669, "step": 18265 }, { "epoch": 1.2627722087798134, "grad_norm": 0.625294768337128, "learning_rate": 3.160126709876097e-06, "loss": 0.3327, "step": 18266 }, { "epoch": 1.2628413411683375, "grad_norm": 0.6319082907829578, "learning_rate": 3.1596062372327463e-06, "loss": 0.2604, "step": 18267 }, { "epoch": 1.2629104735568615, "grad_norm": 0.7122609978842509, "learning_rate": 3.159085787655446e-06, "loss": 0.2508, "step": 18268 }, { "epoch": 1.2629796059453855, "grad_norm": 0.702689965250944, "learning_rate": 3.158565361150722e-06, "loss": 0.347, "step": 18269 }, { "epoch": 1.2630487383339095, "grad_norm": 0.6967340373967879, "learning_rate": 3.158044957725093e-06, "loss": 0.2893, "step": 18270 }, { "epoch": 1.2631178707224335, "grad_norm": 0.6227826896249803, "learning_rate": 3.157524577385085e-06, "loss": 0.2142, "step": 18271 }, { "epoch": 1.2631870031109576, "grad_norm": 0.6406655307665475, "learning_rate": 3.1570042201372177e-06, "loss": 0.2827, "step": 18272 }, { "epoch": 1.2632561354994816, "grad_norm": 1.616078861120221, "learning_rate": 3.156483885988013e-06, "loss": 0.3811, "step": 18273 }, { "epoch": 1.2633252678880056, "grad_norm": 0.6483292636045755, "learning_rate": 3.1559635749439934e-06, "loss": 0.2592, "step": 18274 }, { "epoch": 1.2633944002765296, "grad_norm": 0.630596974490455, "learning_rate": 3.155443287011681e-06, "loss": 0.2683, "step": 18275 }, { "epoch": 1.2634635326650536, "grad_norm": 0.6529657712552736, "learning_rate": 3.154923022197593e-06, "loss": 0.3292, "step": 18276 }, { "epoch": 1.2635326650535776, "grad_norm": 0.6921653295418211, "learning_rate": 3.1544027805082534e-06, "loss": 0.2898, "step": 18277 }, { "epoch": 1.2636017974421017, "grad_norm": 0.7113592055839262, "learning_rate": 3.15388256195018e-06, "loss": 0.2349, "step": 18278 }, { "epoch": 1.2636709298306257, "grad_norm": 0.6975598968702234, "learning_rate": 3.1533623665298964e-06, "loss": 0.2369, "step": 18279 }, { "epoch": 1.2637400622191497, "grad_norm": 0.6875396162989604, "learning_rate": 3.1528421942539185e-06, "loss": 0.3102, "step": 18280 }, { "epoch": 1.2638091946076737, "grad_norm": 0.7053157902473717, "learning_rate": 3.1523220451287678e-06, "loss": 0.2758, "step": 18281 }, { "epoch": 1.2638783269961977, "grad_norm": 0.7147224043848448, "learning_rate": 3.1518019191609616e-06, "loss": 0.2416, "step": 18282 }, { "epoch": 1.2639474593847217, "grad_norm": 0.765414177658857, "learning_rate": 3.151281816357023e-06, "loss": 0.3494, "step": 18283 }, { "epoch": 1.2640165917732458, "grad_norm": 0.6478025920753406, "learning_rate": 3.150761736723465e-06, "loss": 0.3602, "step": 18284 }, { "epoch": 1.2640857241617698, "grad_norm": 0.6389033374396992, "learning_rate": 3.1502416802668102e-06, "loss": 0.2707, "step": 18285 }, { "epoch": 1.2641548565502938, "grad_norm": 0.606989153257188, "learning_rate": 3.149721646993573e-06, "loss": 0.2379, "step": 18286 }, { "epoch": 1.2642239889388178, "grad_norm": 0.6853552915719454, "learning_rate": 3.1492016369102752e-06, "loss": 0.3551, "step": 18287 }, { "epoch": 1.2642931213273418, "grad_norm": 0.6632852957617265, "learning_rate": 3.1486816500234295e-06, "loss": 0.2817, "step": 18288 }, { "epoch": 1.2643622537158659, "grad_norm": 0.6424330844637512, "learning_rate": 3.148161686339558e-06, "loss": 0.2152, "step": 18289 }, { "epoch": 1.2644313861043899, "grad_norm": 0.7077977563124453, "learning_rate": 3.1476417458651733e-06, "loss": 0.3002, "step": 18290 }, { "epoch": 1.264500518492914, "grad_norm": 0.7425986749048478, "learning_rate": 3.147121828606794e-06, "loss": 0.3663, "step": 18291 }, { "epoch": 1.264569650881438, "grad_norm": 0.702647229217972, "learning_rate": 3.1466019345709376e-06, "loss": 0.2619, "step": 18292 }, { "epoch": 1.264638783269962, "grad_norm": 0.6051273176853208, "learning_rate": 3.1460820637641166e-06, "loss": 0.204, "step": 18293 }, { "epoch": 1.264707915658486, "grad_norm": 0.7170774422192981, "learning_rate": 3.145562216192849e-06, "loss": 0.2981, "step": 18294 }, { "epoch": 1.26477704804701, "grad_norm": 0.6754918188069733, "learning_rate": 3.1450423918636484e-06, "loss": 0.2754, "step": 18295 }, { "epoch": 1.264846180435534, "grad_norm": 0.6790544344522584, "learning_rate": 3.1445225907830334e-06, "loss": 0.2517, "step": 18296 }, { "epoch": 1.264915312824058, "grad_norm": 0.6825786297324034, "learning_rate": 3.144002812957515e-06, "loss": 0.289, "step": 18297 }, { "epoch": 1.264984445212582, "grad_norm": 0.7398885822040432, "learning_rate": 3.1434830583936092e-06, "loss": 0.3886, "step": 18298 }, { "epoch": 1.265053577601106, "grad_norm": 0.6530597562516044, "learning_rate": 3.14296332709783e-06, "loss": 0.2186, "step": 18299 }, { "epoch": 1.26512270998963, "grad_norm": 0.6080355459956688, "learning_rate": 3.1424436190766935e-06, "loss": 0.2419, "step": 18300 }, { "epoch": 1.265191842378154, "grad_norm": 0.6484893648068278, "learning_rate": 3.1419239343367085e-06, "loss": 0.2882, "step": 18301 }, { "epoch": 1.265260974766678, "grad_norm": 0.6798822117248947, "learning_rate": 3.141404272884393e-06, "loss": 0.3266, "step": 18302 }, { "epoch": 1.265330107155202, "grad_norm": 0.6799725270388038, "learning_rate": 3.140884634726256e-06, "loss": 0.2135, "step": 18303 }, { "epoch": 1.2653992395437261, "grad_norm": 0.6172758581394459, "learning_rate": 3.140365019868815e-06, "loss": 0.2274, "step": 18304 }, { "epoch": 1.2654683719322501, "grad_norm": 0.9371805029327259, "learning_rate": 3.139845428318577e-06, "loss": 0.3186, "step": 18305 }, { "epoch": 1.2655375043207742, "grad_norm": 0.6442141792063947, "learning_rate": 3.1393258600820597e-06, "loss": 0.279, "step": 18306 }, { "epoch": 1.2656066367092982, "grad_norm": 0.677645145354795, "learning_rate": 3.13880631516577e-06, "loss": 0.1978, "step": 18307 }, { "epoch": 1.2656757690978222, "grad_norm": 0.6237861628005816, "learning_rate": 3.138286793576223e-06, "loss": 0.2108, "step": 18308 }, { "epoch": 1.2657449014863462, "grad_norm": 0.7798022093049843, "learning_rate": 3.137767295319928e-06, "loss": 0.354, "step": 18309 }, { "epoch": 1.2658140338748705, "grad_norm": 0.6867346213566516, "learning_rate": 3.1372478204033986e-06, "loss": 0.2963, "step": 18310 }, { "epoch": 1.2658831662633945, "grad_norm": 0.8368534567946553, "learning_rate": 3.1367283688331407e-06, "loss": 0.2198, "step": 18311 }, { "epoch": 1.2659522986519185, "grad_norm": 0.7024219532601058, "learning_rate": 3.1362089406156692e-06, "loss": 0.3276, "step": 18312 }, { "epoch": 1.2660214310404425, "grad_norm": 0.9093612555560493, "learning_rate": 3.135689535757491e-06, "loss": 0.3044, "step": 18313 }, { "epoch": 1.2660905634289665, "grad_norm": 0.7236276673242411, "learning_rate": 3.1351701542651205e-06, "loss": 0.2714, "step": 18314 }, { "epoch": 1.2661596958174905, "grad_norm": 0.6213370208400945, "learning_rate": 3.1346507961450612e-06, "loss": 0.233, "step": 18315 }, { "epoch": 1.2662288282060146, "grad_norm": 0.700059556657279, "learning_rate": 3.1341314614038266e-06, "loss": 0.3106, "step": 18316 }, { "epoch": 1.2662979605945386, "grad_norm": 0.9207275417443682, "learning_rate": 3.133612150047923e-06, "loss": 0.2493, "step": 18317 }, { "epoch": 1.2663670929830626, "grad_norm": 0.6285965105676584, "learning_rate": 3.133092862083863e-06, "loss": 0.2011, "step": 18318 }, { "epoch": 1.2664362253715866, "grad_norm": 0.638797445641881, "learning_rate": 3.1325735975181505e-06, "loss": 0.2653, "step": 18319 }, { "epoch": 1.2665053577601106, "grad_norm": 0.6317754431149833, "learning_rate": 3.132054356357296e-06, "loss": 0.3068, "step": 18320 }, { "epoch": 1.2665744901486347, "grad_norm": 0.7072382426486046, "learning_rate": 3.131535138607805e-06, "loss": 0.2736, "step": 18321 }, { "epoch": 1.2666436225371587, "grad_norm": 0.6664348781801872, "learning_rate": 3.13101594427619e-06, "loss": 0.2619, "step": 18322 }, { "epoch": 1.2667127549256827, "grad_norm": 0.7054809438484303, "learning_rate": 3.1304967733689506e-06, "loss": 0.3954, "step": 18323 }, { "epoch": 1.2667818873142067, "grad_norm": 0.6987925175057496, "learning_rate": 3.1299776258926018e-06, "loss": 0.2908, "step": 18324 }, { "epoch": 1.2668510197027307, "grad_norm": 0.6444965725344357, "learning_rate": 3.129458501853643e-06, "loss": 0.2715, "step": 18325 }, { "epoch": 1.2669201520912547, "grad_norm": 0.6174850614488153, "learning_rate": 3.128939401258586e-06, "loss": 0.2658, "step": 18326 }, { "epoch": 1.2669892844797788, "grad_norm": 0.7525303905038445, "learning_rate": 3.128420324113933e-06, "loss": 0.3414, "step": 18327 }, { "epoch": 1.2670584168683028, "grad_norm": 0.636187383034517, "learning_rate": 3.1279012704261934e-06, "loss": 0.2339, "step": 18328 }, { "epoch": 1.2671275492568268, "grad_norm": 0.8096524352480494, "learning_rate": 3.127382240201868e-06, "loss": 0.2714, "step": 18329 }, { "epoch": 1.2671966816453508, "grad_norm": 0.7765251133107796, "learning_rate": 3.126863233447466e-06, "loss": 0.3907, "step": 18330 }, { "epoch": 1.2672658140338748, "grad_norm": 0.8195880058148195, "learning_rate": 3.1263442501694894e-06, "loss": 0.2906, "step": 18331 }, { "epoch": 1.2673349464223989, "grad_norm": 0.6157029061951715, "learning_rate": 3.1258252903744456e-06, "loss": 0.2145, "step": 18332 }, { "epoch": 1.2674040788109229, "grad_norm": 0.6873155129618898, "learning_rate": 3.1253063540688353e-06, "loss": 0.2942, "step": 18333 }, { "epoch": 1.267473211199447, "grad_norm": 0.6933818439204816, "learning_rate": 3.124787441259166e-06, "loss": 0.3224, "step": 18334 }, { "epoch": 1.267542343587971, "grad_norm": 0.7903633023436375, "learning_rate": 3.124268551951938e-06, "loss": 0.2592, "step": 18335 }, { "epoch": 1.267611475976495, "grad_norm": 0.6934698931766553, "learning_rate": 3.123749686153659e-06, "loss": 0.284, "step": 18336 }, { "epoch": 1.267680608365019, "grad_norm": 0.7110263306326976, "learning_rate": 3.123230843870826e-06, "loss": 0.3174, "step": 18337 }, { "epoch": 1.267749740753543, "grad_norm": 0.9130964973863568, "learning_rate": 3.1227120251099475e-06, "loss": 0.3334, "step": 18338 }, { "epoch": 1.267818873142067, "grad_norm": 0.5914492957214755, "learning_rate": 3.1221932298775213e-06, "loss": 0.2405, "step": 18339 }, { "epoch": 1.267888005530591, "grad_norm": 0.6894025211731568, "learning_rate": 3.121674458180055e-06, "loss": 0.2816, "step": 18340 }, { "epoch": 1.2679571379191152, "grad_norm": 0.870433406314886, "learning_rate": 3.121155710024044e-06, "loss": 0.3506, "step": 18341 }, { "epoch": 1.2680262703076393, "grad_norm": 0.6089730978375779, "learning_rate": 3.1206369854159957e-06, "loss": 0.224, "step": 18342 }, { "epoch": 1.2680954026961633, "grad_norm": 0.6560039566973085, "learning_rate": 3.120118284362408e-06, "loss": 0.2756, "step": 18343 }, { "epoch": 1.2681645350846873, "grad_norm": 0.6838835971999799, "learning_rate": 3.119599606869783e-06, "loss": 0.3293, "step": 18344 }, { "epoch": 1.2682336674732113, "grad_norm": 0.6728020706257887, "learning_rate": 3.11908095294462e-06, "loss": 0.323, "step": 18345 }, { "epoch": 1.2683027998617353, "grad_norm": 1.2020460136849698, "learning_rate": 3.1185623225934232e-06, "loss": 0.3007, "step": 18346 }, { "epoch": 1.2683719322502593, "grad_norm": 0.6458367738086642, "learning_rate": 3.118043715822688e-06, "loss": 0.2431, "step": 18347 }, { "epoch": 1.2684410646387834, "grad_norm": 0.7647723782167379, "learning_rate": 3.117525132638917e-06, "loss": 0.3467, "step": 18348 }, { "epoch": 1.2685101970273074, "grad_norm": 0.6955932807664784, "learning_rate": 3.1170065730486082e-06, "loss": 0.2303, "step": 18349 }, { "epoch": 1.2685793294158314, "grad_norm": 0.657976871377941, "learning_rate": 3.116488037058264e-06, "loss": 0.2149, "step": 18350 }, { "epoch": 1.2686484618043554, "grad_norm": 0.6501527042987043, "learning_rate": 3.1159695246743783e-06, "loss": 0.2469, "step": 18351 }, { "epoch": 1.2687175941928794, "grad_norm": 0.7193744020114696, "learning_rate": 3.115451035903454e-06, "loss": 0.3089, "step": 18352 }, { "epoch": 1.2687867265814035, "grad_norm": 0.6398356120474377, "learning_rate": 3.1149325707519863e-06, "loss": 0.2847, "step": 18353 }, { "epoch": 1.2688558589699275, "grad_norm": 1.2728636960345174, "learning_rate": 3.1144141292264772e-06, "loss": 0.248, "step": 18354 }, { "epoch": 1.2689249913584515, "grad_norm": 0.6634332773527319, "learning_rate": 3.11389571133342e-06, "loss": 0.336, "step": 18355 }, { "epoch": 1.2689941237469755, "grad_norm": 0.6837922663444109, "learning_rate": 3.113377317079315e-06, "loss": 0.3012, "step": 18356 }, { "epoch": 1.2690632561354995, "grad_norm": 0.6724676382110255, "learning_rate": 3.1128589464706575e-06, "loss": 0.2536, "step": 18357 }, { "epoch": 1.2691323885240235, "grad_norm": 0.6961507584672166, "learning_rate": 3.1123405995139477e-06, "loss": 0.2448, "step": 18358 }, { "epoch": 1.2692015209125476, "grad_norm": 0.7303552270872748, "learning_rate": 3.111822276215678e-06, "loss": 0.3698, "step": 18359 }, { "epoch": 1.2692706533010716, "grad_norm": 0.6959275491688005, "learning_rate": 3.111303976582346e-06, "loss": 0.2443, "step": 18360 }, { "epoch": 1.2693397856895956, "grad_norm": 0.6695281722489775, "learning_rate": 3.1107857006204497e-06, "loss": 0.2897, "step": 18361 }, { "epoch": 1.2694089180781196, "grad_norm": 0.7021346138627252, "learning_rate": 3.1102674483364815e-06, "loss": 0.3396, "step": 18362 }, { "epoch": 1.2694780504666436, "grad_norm": 0.6550669778531961, "learning_rate": 3.1097492197369407e-06, "loss": 0.3246, "step": 18363 }, { "epoch": 1.2695471828551677, "grad_norm": 0.647294785444893, "learning_rate": 3.1092310148283176e-06, "loss": 0.2076, "step": 18364 }, { "epoch": 1.2696163152436917, "grad_norm": 0.63799535137056, "learning_rate": 3.108712833617111e-06, "loss": 0.2863, "step": 18365 }, { "epoch": 1.2696854476322157, "grad_norm": 0.7679097863106663, "learning_rate": 3.1081946761098126e-06, "loss": 0.4173, "step": 18366 }, { "epoch": 1.2697545800207397, "grad_norm": 0.6463024571057867, "learning_rate": 3.10767654231292e-06, "loss": 0.2593, "step": 18367 }, { "epoch": 1.2698237124092637, "grad_norm": 0.6303139268814769, "learning_rate": 3.107158432232923e-06, "loss": 0.2208, "step": 18368 }, { "epoch": 1.2698928447977877, "grad_norm": 0.7633457101529768, "learning_rate": 3.1066403458763184e-06, "loss": 0.3305, "step": 18369 }, { "epoch": 1.2699619771863118, "grad_norm": 0.6890185399750709, "learning_rate": 3.1061222832495963e-06, "loss": 0.3096, "step": 18370 }, { "epoch": 1.2700311095748358, "grad_norm": 0.6836976415519697, "learning_rate": 3.1056042443592543e-06, "loss": 0.2358, "step": 18371 }, { "epoch": 1.2701002419633598, "grad_norm": 0.6900555128542383, "learning_rate": 3.10508622921178e-06, "loss": 0.2941, "step": 18372 }, { "epoch": 1.2701693743518838, "grad_norm": 0.6909750851388085, "learning_rate": 3.1045682378136698e-06, "loss": 0.3345, "step": 18373 }, { "epoch": 1.2702385067404078, "grad_norm": 0.7083954319048688, "learning_rate": 3.1040502701714127e-06, "loss": 0.2003, "step": 18374 }, { "epoch": 1.2703076391289319, "grad_norm": 0.6356827683953199, "learning_rate": 3.103532326291504e-06, "loss": 0.2407, "step": 18375 }, { "epoch": 1.2703767715174559, "grad_norm": 0.777117892127997, "learning_rate": 3.1030144061804317e-06, "loss": 0.2763, "step": 18376 }, { "epoch": 1.27044590390598, "grad_norm": 0.8101085753332103, "learning_rate": 3.102496509844689e-06, "loss": 0.3667, "step": 18377 }, { "epoch": 1.270515036294504, "grad_norm": 0.7579853914456187, "learning_rate": 3.1019786372907657e-06, "loss": 0.264, "step": 18378 }, { "epoch": 1.270584168683028, "grad_norm": 1.0309139062176729, "learning_rate": 3.1014607885251536e-06, "loss": 0.2176, "step": 18379 }, { "epoch": 1.270653301071552, "grad_norm": 0.658546719146285, "learning_rate": 3.100942963554342e-06, "loss": 0.374, "step": 18380 }, { "epoch": 1.270722433460076, "grad_norm": 0.6238451259431705, "learning_rate": 3.1004251623848223e-06, "loss": 0.3295, "step": 18381 }, { "epoch": 1.2707915658486, "grad_norm": 0.6233898402533018, "learning_rate": 3.0999073850230822e-06, "loss": 0.2515, "step": 18382 }, { "epoch": 1.270860698237124, "grad_norm": 0.68638068541926, "learning_rate": 3.099389631475613e-06, "loss": 0.3096, "step": 18383 }, { "epoch": 1.270929830625648, "grad_norm": 0.7002152372144016, "learning_rate": 3.098871901748901e-06, "loss": 0.3483, "step": 18384 }, { "epoch": 1.270998963014172, "grad_norm": 0.7367417577669524, "learning_rate": 3.09835419584944e-06, "loss": 0.2485, "step": 18385 }, { "epoch": 1.271068095402696, "grad_norm": 0.7237737335325164, "learning_rate": 3.0978365137837134e-06, "loss": 0.2377, "step": 18386 }, { "epoch": 1.27113722779122, "grad_norm": 0.7332723258341308, "learning_rate": 3.0973188555582124e-06, "loss": 0.4035, "step": 18387 }, { "epoch": 1.271206360179744, "grad_norm": 0.6371612858422289, "learning_rate": 3.096801221179423e-06, "loss": 0.2992, "step": 18388 }, { "epoch": 1.271275492568268, "grad_norm": 0.5860879317510764, "learning_rate": 3.096283610653836e-06, "loss": 0.2128, "step": 18389 }, { "epoch": 1.2713446249567921, "grad_norm": 0.6648963282727478, "learning_rate": 3.095766023987935e-06, "loss": 0.272, "step": 18390 }, { "epoch": 1.2714137573453161, "grad_norm": 0.6932457753295157, "learning_rate": 3.0952484611882094e-06, "loss": 0.3362, "step": 18391 }, { "epoch": 1.2714828897338404, "grad_norm": 0.6542395080762609, "learning_rate": 3.094730922261143e-06, "loss": 0.2554, "step": 18392 }, { "epoch": 1.2715520221223644, "grad_norm": 0.6399761236908037, "learning_rate": 3.094213407213228e-06, "loss": 0.2399, "step": 18393 }, { "epoch": 1.2716211545108884, "grad_norm": 0.6797409527394299, "learning_rate": 3.0936959160509432e-06, "loss": 0.3281, "step": 18394 }, { "epoch": 1.2716902868994124, "grad_norm": 0.6472809399676543, "learning_rate": 3.0931784487807815e-06, "loss": 0.2604, "step": 18395 }, { "epoch": 1.2717594192879365, "grad_norm": 0.6335259190871614, "learning_rate": 3.0926610054092227e-06, "loss": 0.2393, "step": 18396 }, { "epoch": 1.2718285516764605, "grad_norm": 0.8425757221616812, "learning_rate": 3.092143585942755e-06, "loss": 0.2828, "step": 18397 }, { "epoch": 1.2718976840649845, "grad_norm": 0.6554996653894621, "learning_rate": 3.0916261903878624e-06, "loss": 0.3407, "step": 18398 }, { "epoch": 1.2719668164535085, "grad_norm": 0.7232534400952548, "learning_rate": 3.091108818751032e-06, "loss": 0.3028, "step": 18399 }, { "epoch": 1.2720359488420325, "grad_norm": 0.6798707106821026, "learning_rate": 3.090591471038743e-06, "loss": 0.2574, "step": 18400 }, { "epoch": 1.2721050812305565, "grad_norm": 0.7307067428219556, "learning_rate": 3.0900741472574847e-06, "loss": 0.2987, "step": 18401 }, { "epoch": 1.2721742136190806, "grad_norm": 0.6855637138132008, "learning_rate": 3.0895568474137365e-06, "loss": 0.3187, "step": 18402 }, { "epoch": 1.2722433460076046, "grad_norm": 0.694215792617243, "learning_rate": 3.0890395715139864e-06, "loss": 0.2905, "step": 18403 }, { "epoch": 1.2723124783961286, "grad_norm": 0.6257485083452018, "learning_rate": 3.088522319564713e-06, "loss": 0.1967, "step": 18404 }, { "epoch": 1.2723816107846526, "grad_norm": 0.7195940721698966, "learning_rate": 3.088005091572402e-06, "loss": 0.3693, "step": 18405 }, { "epoch": 1.2724507431731766, "grad_norm": 0.6702456561031768, "learning_rate": 3.087487887543534e-06, "loss": 0.2657, "step": 18406 }, { "epoch": 1.2725198755617007, "grad_norm": 0.6362408538795388, "learning_rate": 3.0869707074845943e-06, "loss": 0.2315, "step": 18407 }, { "epoch": 1.2725890079502247, "grad_norm": 1.2192872813548339, "learning_rate": 3.086453551402061e-06, "loss": 0.2693, "step": 18408 }, { "epoch": 1.2726581403387487, "grad_norm": 0.6866042514967606, "learning_rate": 3.0859364193024188e-06, "loss": 0.3813, "step": 18409 }, { "epoch": 1.2727272727272727, "grad_norm": 0.6509794034051019, "learning_rate": 3.085419311192146e-06, "loss": 0.2502, "step": 18410 }, { "epoch": 1.2727964051157967, "grad_norm": 0.6284485527582785, "learning_rate": 3.0849022270777284e-06, "loss": 0.2174, "step": 18411 }, { "epoch": 1.2728655375043207, "grad_norm": 0.7603747559264823, "learning_rate": 3.08438516696564e-06, "loss": 0.3079, "step": 18412 }, { "epoch": 1.2729346698928448, "grad_norm": 0.7432326660759974, "learning_rate": 3.0838681308623685e-06, "loss": 0.3722, "step": 18413 }, { "epoch": 1.2730038022813688, "grad_norm": 0.6756797002759873, "learning_rate": 3.083351118774388e-06, "loss": 0.2157, "step": 18414 }, { "epoch": 1.2730729346698928, "grad_norm": 0.6996524586736681, "learning_rate": 3.082834130708182e-06, "loss": 0.2713, "step": 18415 }, { "epoch": 1.2731420670584168, "grad_norm": 0.6551811644098909, "learning_rate": 3.082317166670228e-06, "loss": 0.278, "step": 18416 }, { "epoch": 1.2732111994469408, "grad_norm": 0.882116553713779, "learning_rate": 3.0818002266670076e-06, "loss": 0.2247, "step": 18417 }, { "epoch": 1.2732803318354649, "grad_norm": 0.7300979766946166, "learning_rate": 3.081283310704997e-06, "loss": 0.2333, "step": 18418 }, { "epoch": 1.2733494642239889, "grad_norm": 0.7148577518613163, "learning_rate": 3.080766418790676e-06, "loss": 0.3305, "step": 18419 }, { "epoch": 1.2734185966125129, "grad_norm": 0.6983693042397854, "learning_rate": 3.0802495509305217e-06, "loss": 0.3183, "step": 18420 }, { "epoch": 1.273487729001037, "grad_norm": 0.6110389444584448, "learning_rate": 3.0797327071310166e-06, "loss": 0.2024, "step": 18421 }, { "epoch": 1.2735568613895611, "grad_norm": 0.6989630115124292, "learning_rate": 3.079215887398632e-06, "loss": 0.2538, "step": 18422 }, { "epoch": 1.2736259937780852, "grad_norm": 0.7153136620189509, "learning_rate": 3.0786990917398507e-06, "loss": 0.324, "step": 18423 }, { "epoch": 1.2736951261666092, "grad_norm": 0.7605951287614415, "learning_rate": 3.078182320161146e-06, "loss": 0.3238, "step": 18424 }, { "epoch": 1.2737642585551332, "grad_norm": 0.603079183218407, "learning_rate": 3.0776655726689975e-06, "loss": 0.2547, "step": 18425 }, { "epoch": 1.2738333909436572, "grad_norm": 0.7349696571947929, "learning_rate": 3.077148849269881e-06, "loss": 0.3728, "step": 18426 }, { "epoch": 1.2739025233321812, "grad_norm": 0.6737378926254978, "learning_rate": 3.07663214997027e-06, "loss": 0.3198, "step": 18427 }, { "epoch": 1.2739716557207053, "grad_norm": 0.7745990390303313, "learning_rate": 3.0761154747766446e-06, "loss": 0.2523, "step": 18428 }, { "epoch": 1.2740407881092293, "grad_norm": 0.6098822620319081, "learning_rate": 3.075598823695477e-06, "loss": 0.223, "step": 18429 }, { "epoch": 1.2741099204977533, "grad_norm": 0.7509522160482437, "learning_rate": 3.0750821967332445e-06, "loss": 0.3242, "step": 18430 }, { "epoch": 1.2741790528862773, "grad_norm": 0.6927543790921523, "learning_rate": 3.0745655938964202e-06, "loss": 0.281, "step": 18431 }, { "epoch": 1.2742481852748013, "grad_norm": 0.8211061752693711, "learning_rate": 3.074049015191482e-06, "loss": 0.295, "step": 18432 }, { "epoch": 1.2743173176633253, "grad_norm": 0.5931845090350379, "learning_rate": 3.0735324606249006e-06, "loss": 0.2159, "step": 18433 }, { "epoch": 1.2743864500518494, "grad_norm": 0.6536332076753193, "learning_rate": 3.073015930203154e-06, "loss": 0.3618, "step": 18434 }, { "epoch": 1.2744555824403734, "grad_norm": 0.6497265905332463, "learning_rate": 3.072499423932712e-06, "loss": 0.2388, "step": 18435 }, { "epoch": 1.2745247148288974, "grad_norm": 0.6526564402464463, "learning_rate": 3.0719829418200514e-06, "loss": 0.2068, "step": 18436 }, { "epoch": 1.2745938472174214, "grad_norm": 0.6936944855059259, "learning_rate": 3.0714664838716424e-06, "loss": 0.3669, "step": 18437 }, { "epoch": 1.2746629796059454, "grad_norm": 0.7645954694939631, "learning_rate": 3.0709500500939616e-06, "loss": 0.3369, "step": 18438 }, { "epoch": 1.2747321119944695, "grad_norm": 0.6577680733881612, "learning_rate": 3.070433640493477e-06, "loss": 0.2842, "step": 18439 }, { "epoch": 1.2748012443829935, "grad_norm": 0.732778889050745, "learning_rate": 3.069917255076664e-06, "loss": 0.2783, "step": 18440 }, { "epoch": 1.2748703767715175, "grad_norm": 0.7129078911473331, "learning_rate": 3.069400893849994e-06, "loss": 0.3543, "step": 18441 }, { "epoch": 1.2749395091600415, "grad_norm": 0.6771585428866896, "learning_rate": 3.06888455681994e-06, "loss": 0.2093, "step": 18442 }, { "epoch": 1.2750086415485655, "grad_norm": 0.6902493332824339, "learning_rate": 3.0683682439929698e-06, "loss": 0.2108, "step": 18443 }, { "epoch": 1.2750777739370895, "grad_norm": 0.6845918217922445, "learning_rate": 3.067851955375557e-06, "loss": 0.2941, "step": 18444 }, { "epoch": 1.2751469063256136, "grad_norm": 0.6656263562577492, "learning_rate": 3.0673356909741712e-06, "loss": 0.277, "step": 18445 }, { "epoch": 1.2752160387141376, "grad_norm": 0.6517285915220513, "learning_rate": 3.0668194507952852e-06, "loss": 0.2345, "step": 18446 }, { "epoch": 1.2752851711026616, "grad_norm": 0.6415607280592205, "learning_rate": 3.066303234845366e-06, "loss": 0.2479, "step": 18447 }, { "epoch": 1.2753543034911856, "grad_norm": 0.7461193704324867, "learning_rate": 3.0657870431308857e-06, "loss": 0.4253, "step": 18448 }, { "epoch": 1.2754234358797096, "grad_norm": 0.6900057483815736, "learning_rate": 3.0652708756583128e-06, "loss": 0.2658, "step": 18449 }, { "epoch": 1.2754925682682337, "grad_norm": 0.6342862876655722, "learning_rate": 3.0647547324341166e-06, "loss": 0.2265, "step": 18450 }, { "epoch": 1.2755617006567577, "grad_norm": 0.6876767890249561, "learning_rate": 3.064238613464766e-06, "loss": 0.2904, "step": 18451 }, { "epoch": 1.2756308330452817, "grad_norm": 0.713482857117818, "learning_rate": 3.063722518756731e-06, "loss": 0.3538, "step": 18452 }, { "epoch": 1.2756999654338057, "grad_norm": 1.3355590197321623, "learning_rate": 3.063206448316478e-06, "loss": 0.2586, "step": 18453 }, { "epoch": 1.2757690978223297, "grad_norm": 0.6778742303052138, "learning_rate": 3.0626904021504754e-06, "loss": 0.2683, "step": 18454 }, { "epoch": 1.2758382302108537, "grad_norm": 0.7003502105077194, "learning_rate": 3.0621743802651915e-06, "loss": 0.3301, "step": 18455 }, { "epoch": 1.2759073625993778, "grad_norm": 0.7154241964269868, "learning_rate": 3.0616583826670955e-06, "loss": 0.2623, "step": 18456 }, { "epoch": 1.2759764949879018, "grad_norm": 0.713418885980908, "learning_rate": 3.0611424093626497e-06, "loss": 0.2237, "step": 18457 }, { "epoch": 1.2760456273764258, "grad_norm": 0.6337023815378944, "learning_rate": 3.0606264603583257e-06, "loss": 0.2326, "step": 18458 }, { "epoch": 1.2761147597649498, "grad_norm": 0.6780401527691119, "learning_rate": 3.0601105356605865e-06, "loss": 0.3156, "step": 18459 }, { "epoch": 1.2761838921534738, "grad_norm": 0.6222489023317563, "learning_rate": 3.059594635275902e-06, "loss": 0.23, "step": 18460 }, { "epoch": 1.2762530245419978, "grad_norm": 0.6622382457872189, "learning_rate": 3.059078759210734e-06, "loss": 0.2547, "step": 18461 }, { "epoch": 1.2763221569305219, "grad_norm": 0.6563248224186743, "learning_rate": 3.0585629074715505e-06, "loss": 0.3009, "step": 18462 }, { "epoch": 1.2763912893190459, "grad_norm": 0.6577763781658365, "learning_rate": 3.0580470800648153e-06, "loss": 0.3347, "step": 18463 }, { "epoch": 1.27646042170757, "grad_norm": 0.6694638514352533, "learning_rate": 3.057531276996997e-06, "loss": 0.2232, "step": 18464 }, { "epoch": 1.276529554096094, "grad_norm": 0.6717254632593279, "learning_rate": 3.057015498274555e-06, "loss": 0.2867, "step": 18465 }, { "epoch": 1.276598686484618, "grad_norm": 0.6703088192198825, "learning_rate": 3.0564997439039585e-06, "loss": 0.3982, "step": 18466 }, { "epoch": 1.276667818873142, "grad_norm": 0.6395211144253321, "learning_rate": 3.0559840138916673e-06, "loss": 0.2378, "step": 18467 }, { "epoch": 1.276736951261666, "grad_norm": 1.0628757413807621, "learning_rate": 3.055468308244149e-06, "loss": 0.3183, "step": 18468 }, { "epoch": 1.27680608365019, "grad_norm": 0.6680365056378835, "learning_rate": 3.0549526269678638e-06, "loss": 0.2519, "step": 18469 }, { "epoch": 1.276875216038714, "grad_norm": 1.0119224886047045, "learning_rate": 3.0544369700692784e-06, "loss": 0.3086, "step": 18470 }, { "epoch": 1.276944348427238, "grad_norm": 0.6910442939248065, "learning_rate": 3.0539213375548514e-06, "loss": 0.2202, "step": 18471 }, { "epoch": 1.277013480815762, "grad_norm": 0.6886977320071498, "learning_rate": 3.053405729431049e-06, "loss": 0.3032, "step": 18472 }, { "epoch": 1.2770826132042863, "grad_norm": 0.6812154996601663, "learning_rate": 3.05289014570433e-06, "loss": 0.3199, "step": 18473 }, { "epoch": 1.2771517455928103, "grad_norm": 0.6122187982112322, "learning_rate": 3.052374586381161e-06, "loss": 0.2125, "step": 18474 }, { "epoch": 1.2772208779813343, "grad_norm": 0.6408532457345344, "learning_rate": 3.051859051467999e-06, "loss": 0.2136, "step": 18475 }, { "epoch": 1.2772900103698583, "grad_norm": 0.7549736568158999, "learning_rate": 3.051343540971308e-06, "loss": 0.3425, "step": 18476 }, { "epoch": 1.2773591427583824, "grad_norm": 0.7146455100920576, "learning_rate": 3.0508280548975466e-06, "loss": 0.3294, "step": 18477 }, { "epoch": 1.2774282751469064, "grad_norm": 0.64086860108344, "learning_rate": 3.0503125932531795e-06, "loss": 0.2264, "step": 18478 }, { "epoch": 1.2774974075354304, "grad_norm": 0.6288408732629798, "learning_rate": 3.0497971560446626e-06, "loss": 0.2086, "step": 18479 }, { "epoch": 1.2775665399239544, "grad_norm": 0.7593133189571917, "learning_rate": 3.0492817432784592e-06, "loss": 0.3411, "step": 18480 }, { "epoch": 1.2776356723124784, "grad_norm": 0.6812266011201613, "learning_rate": 3.0487663549610267e-06, "loss": 0.2836, "step": 18481 }, { "epoch": 1.2777048047010025, "grad_norm": 0.6654069489267526, "learning_rate": 3.0482509910988283e-06, "loss": 0.2892, "step": 18482 }, { "epoch": 1.2777739370895265, "grad_norm": 0.6671342703109002, "learning_rate": 3.047735651698317e-06, "loss": 0.3186, "step": 18483 }, { "epoch": 1.2778430694780505, "grad_norm": 0.6957278849495687, "learning_rate": 3.047220336765959e-06, "loss": 0.3512, "step": 18484 }, { "epoch": 1.2779122018665745, "grad_norm": 0.7294598332969365, "learning_rate": 3.046705046308207e-06, "loss": 0.2623, "step": 18485 }, { "epoch": 1.2779813342550985, "grad_norm": 0.6880789558282594, "learning_rate": 3.0461897803315223e-06, "loss": 0.2193, "step": 18486 }, { "epoch": 1.2780504666436225, "grad_norm": 0.6117183609385802, "learning_rate": 3.045674538842361e-06, "loss": 0.3177, "step": 18487 }, { "epoch": 1.2781195990321466, "grad_norm": 0.6638481758337413, "learning_rate": 3.0451593218471835e-06, "loss": 0.2946, "step": 18488 }, { "epoch": 1.2781887314206706, "grad_norm": 0.613906305586193, "learning_rate": 3.0446441293524436e-06, "loss": 0.2447, "step": 18489 }, { "epoch": 1.2782578638091946, "grad_norm": 0.6953377886119189, "learning_rate": 3.044128961364601e-06, "loss": 0.2835, "step": 18490 }, { "epoch": 1.2783269961977186, "grad_norm": 0.6705416779807091, "learning_rate": 3.0436138178901126e-06, "loss": 0.3239, "step": 18491 }, { "epoch": 1.2783961285862426, "grad_norm": 0.6648319829059074, "learning_rate": 3.0430986989354317e-06, "loss": 0.2504, "step": 18492 }, { "epoch": 1.2784652609747666, "grad_norm": 0.6339794751012212, "learning_rate": 3.0425836045070167e-06, "loss": 0.2349, "step": 18493 }, { "epoch": 1.2785343933632907, "grad_norm": 0.7780098944346835, "learning_rate": 3.042068534611323e-06, "loss": 0.3104, "step": 18494 }, { "epoch": 1.2786035257518147, "grad_norm": 0.6548762730968856, "learning_rate": 3.041553489254807e-06, "loss": 0.2925, "step": 18495 }, { "epoch": 1.2786726581403387, "grad_norm": 1.2178521371401503, "learning_rate": 3.041038468443922e-06, "loss": 0.2361, "step": 18496 }, { "epoch": 1.2787417905288627, "grad_norm": 0.7205009585288149, "learning_rate": 3.0405234721851245e-06, "loss": 0.2868, "step": 18497 }, { "epoch": 1.2788109229173867, "grad_norm": 0.7126850636306474, "learning_rate": 3.040008500484867e-06, "loss": 0.3515, "step": 18498 }, { "epoch": 1.2788800553059108, "grad_norm": 0.6392513799826433, "learning_rate": 3.0394935533496083e-06, "loss": 0.2924, "step": 18499 }, { "epoch": 1.2789491876944348, "grad_norm": 0.6469803246888621, "learning_rate": 3.038978630785796e-06, "loss": 0.2537, "step": 18500 }, { "epoch": 1.2790183200829588, "grad_norm": 0.6720278761246139, "learning_rate": 3.0384637327998887e-06, "loss": 0.2509, "step": 18501 }, { "epoch": 1.2790874524714828, "grad_norm": 0.6930179744041516, "learning_rate": 3.0379488593983368e-06, "loss": 0.3079, "step": 18502 }, { "epoch": 1.279156584860007, "grad_norm": 0.6309325645140146, "learning_rate": 3.0374340105875955e-06, "loss": 0.2519, "step": 18503 }, { "epoch": 1.279225717248531, "grad_norm": 0.6532935108803579, "learning_rate": 3.0369191863741154e-06, "loss": 0.3375, "step": 18504 }, { "epoch": 1.279294849637055, "grad_norm": 0.7105339331891112, "learning_rate": 3.0364043867643522e-06, "loss": 0.3505, "step": 18505 }, { "epoch": 1.279363982025579, "grad_norm": 0.7685101887731763, "learning_rate": 3.035889611764754e-06, "loss": 0.3377, "step": 18506 }, { "epoch": 1.2794331144141031, "grad_norm": 0.6626952918836128, "learning_rate": 3.035374861381776e-06, "loss": 0.2293, "step": 18507 }, { "epoch": 1.2795022468026271, "grad_norm": 0.665782007599222, "learning_rate": 3.0348601356218656e-06, "loss": 0.2617, "step": 18508 }, { "epoch": 1.2795713791911512, "grad_norm": 0.7558019039444447, "learning_rate": 3.034345434491479e-06, "loss": 0.3351, "step": 18509 }, { "epoch": 1.2796405115796752, "grad_norm": 0.7578377097404516, "learning_rate": 3.0338307579970625e-06, "loss": 0.2056, "step": 18510 }, { "epoch": 1.2797096439681992, "grad_norm": 0.6741285073620547, "learning_rate": 3.03331610614507e-06, "loss": 0.2512, "step": 18511 }, { "epoch": 1.2797787763567232, "grad_norm": 0.8115483508036151, "learning_rate": 3.032801478941949e-06, "loss": 0.2768, "step": 18512 }, { "epoch": 1.2798479087452472, "grad_norm": 0.6945728364376432, "learning_rate": 3.0322868763941537e-06, "loss": 0.3523, "step": 18513 }, { "epoch": 1.2799170411337712, "grad_norm": 0.618301346403594, "learning_rate": 3.031772298508128e-06, "loss": 0.188, "step": 18514 }, { "epoch": 1.2799861735222953, "grad_norm": 0.6659074700612533, "learning_rate": 3.0312577452903252e-06, "loss": 0.2746, "step": 18515 }, { "epoch": 1.2800553059108193, "grad_norm": 0.8360211250293724, "learning_rate": 3.0307432167471924e-06, "loss": 0.3599, "step": 18516 }, { "epoch": 1.2801244382993433, "grad_norm": 0.7203748303712173, "learning_rate": 3.030228712885181e-06, "loss": 0.2842, "step": 18517 }, { "epoch": 1.2801935706878673, "grad_norm": 0.6967906827982845, "learning_rate": 3.0297142337107354e-06, "loss": 0.2169, "step": 18518 }, { "epoch": 1.2802627030763913, "grad_norm": 0.65484718098985, "learning_rate": 3.0291997792303064e-06, "loss": 0.3088, "step": 18519 }, { "epoch": 1.2803318354649154, "grad_norm": 0.8689650489937417, "learning_rate": 3.028685349450341e-06, "loss": 0.3041, "step": 18520 }, { "epoch": 1.2804009678534394, "grad_norm": 0.6051366667077406, "learning_rate": 3.028170944377287e-06, "loss": 0.25, "step": 18521 }, { "epoch": 1.2804701002419634, "grad_norm": 0.7291664567720538, "learning_rate": 3.02765656401759e-06, "loss": 0.2758, "step": 18522 }, { "epoch": 1.2805392326304874, "grad_norm": 0.6404114207527579, "learning_rate": 3.0271422083777013e-06, "loss": 0.3155, "step": 18523 }, { "epoch": 1.2806083650190114, "grad_norm": 0.6343177209047374, "learning_rate": 3.0266278774640616e-06, "loss": 0.2806, "step": 18524 }, { "epoch": 1.2806774974075354, "grad_norm": 0.6198400313625256, "learning_rate": 3.0261135712831215e-06, "loss": 0.2734, "step": 18525 }, { "epoch": 1.2807466297960595, "grad_norm": 0.7315122974441512, "learning_rate": 3.0255992898413233e-06, "loss": 0.3023, "step": 18526 }, { "epoch": 1.2808157621845835, "grad_norm": 0.6891370927177023, "learning_rate": 3.025085033145117e-06, "loss": 0.3583, "step": 18527 }, { "epoch": 1.2808848945731075, "grad_norm": 0.6102275477505231, "learning_rate": 3.0245708012009434e-06, "loss": 0.25, "step": 18528 }, { "epoch": 1.2809540269616315, "grad_norm": 0.6891084150322896, "learning_rate": 3.02405659401525e-06, "loss": 0.2606, "step": 18529 }, { "epoch": 1.2810231593501555, "grad_norm": 0.7169791655516689, "learning_rate": 3.0235424115944805e-06, "loss": 0.3847, "step": 18530 }, { "epoch": 1.2810922917386796, "grad_norm": 0.6795359681544484, "learning_rate": 3.0230282539450817e-06, "loss": 0.2671, "step": 18531 }, { "epoch": 1.2811614241272036, "grad_norm": 0.6312923452946665, "learning_rate": 3.022514121073494e-06, "loss": 0.2705, "step": 18532 }, { "epoch": 1.2812305565157276, "grad_norm": 0.8203381192067534, "learning_rate": 3.0220000129861633e-06, "loss": 0.3216, "step": 18533 }, { "epoch": 1.2812996889042516, "grad_norm": 0.6296943813379945, "learning_rate": 3.0214859296895317e-06, "loss": 0.3246, "step": 18534 }, { "epoch": 1.2813688212927756, "grad_norm": 0.5979523178716004, "learning_rate": 3.020971871190045e-06, "loss": 0.2969, "step": 18535 }, { "epoch": 1.2814379536812996, "grad_norm": 0.6103163110082354, "learning_rate": 3.0204578374941416e-06, "loss": 0.2173, "step": 18536 }, { "epoch": 1.2815070860698237, "grad_norm": 0.6372707162379159, "learning_rate": 3.019943828608269e-06, "loss": 0.3113, "step": 18537 }, { "epoch": 1.2815762184583477, "grad_norm": 0.6894500737646587, "learning_rate": 3.0194298445388647e-06, "loss": 0.2514, "step": 18538 }, { "epoch": 1.2816453508468717, "grad_norm": 0.6455270544014426, "learning_rate": 3.0189158852923745e-06, "loss": 0.2148, "step": 18539 }, { "epoch": 1.2817144832353957, "grad_norm": 0.7257278249217155, "learning_rate": 3.0184019508752373e-06, "loss": 0.3044, "step": 18540 }, { "epoch": 1.2817836156239197, "grad_norm": 0.6809894683164075, "learning_rate": 3.0178880412938972e-06, "loss": 0.3254, "step": 18541 }, { "epoch": 1.2818527480124438, "grad_norm": 0.6012288897205654, "learning_rate": 3.017374156554791e-06, "loss": 0.1937, "step": 18542 }, { "epoch": 1.2819218804009678, "grad_norm": 0.6941853049749765, "learning_rate": 3.0168602966643625e-06, "loss": 0.2284, "step": 18543 }, { "epoch": 1.2819910127894918, "grad_norm": 0.7128090653620133, "learning_rate": 3.0163464616290504e-06, "loss": 0.3213, "step": 18544 }, { "epoch": 1.2820601451780158, "grad_norm": 0.6734505503362466, "learning_rate": 3.015832651455297e-06, "loss": 0.2999, "step": 18545 }, { "epoch": 1.2821292775665398, "grad_norm": 0.685024193325487, "learning_rate": 3.0153188661495392e-06, "loss": 0.2172, "step": 18546 }, { "epoch": 1.2821984099550638, "grad_norm": 0.7260265041245352, "learning_rate": 3.014805105718218e-06, "loss": 0.2761, "step": 18547 }, { "epoch": 1.2822675423435879, "grad_norm": 0.7076583636880738, "learning_rate": 3.0142913701677713e-06, "loss": 0.3909, "step": 18548 }, { "epoch": 1.2823366747321119, "grad_norm": 0.633568991314662, "learning_rate": 3.0137776595046404e-06, "loss": 0.2215, "step": 18549 }, { "epoch": 1.282405807120636, "grad_norm": 0.6614931426877939, "learning_rate": 3.01326397373526e-06, "loss": 0.2517, "step": 18550 }, { "epoch": 1.28247493950916, "grad_norm": 0.6884934301068708, "learning_rate": 3.0127503128660712e-06, "loss": 0.3473, "step": 18551 }, { "epoch": 1.282544071897684, "grad_norm": 0.67732040581583, "learning_rate": 3.01223667690351e-06, "loss": 0.3938, "step": 18552 }, { "epoch": 1.282613204286208, "grad_norm": 0.7336083697398716, "learning_rate": 3.0117230658540163e-06, "loss": 0.3055, "step": 18553 }, { "epoch": 1.2826823366747322, "grad_norm": 0.6950461326422812, "learning_rate": 3.0112094797240233e-06, "loss": 0.2565, "step": 18554 }, { "epoch": 1.2827514690632562, "grad_norm": 0.7151449797176918, "learning_rate": 3.010695918519973e-06, "loss": 0.3606, "step": 18555 }, { "epoch": 1.2828206014517802, "grad_norm": 0.6547132970151582, "learning_rate": 3.010182382248298e-06, "loss": 0.2567, "step": 18556 }, { "epoch": 1.2828897338403042, "grad_norm": 0.7428467784991929, "learning_rate": 3.009668870915435e-06, "loss": 0.2714, "step": 18557 }, { "epoch": 1.2829588662288283, "grad_norm": 0.6517858227588326, "learning_rate": 3.0091553845278238e-06, "loss": 0.2076, "step": 18558 }, { "epoch": 1.2830279986173523, "grad_norm": 0.7592194319607456, "learning_rate": 3.0086419230918938e-06, "loss": 0.2912, "step": 18559 }, { "epoch": 1.2830971310058763, "grad_norm": 0.798278901741403, "learning_rate": 3.008128486614085e-06, "loss": 0.2086, "step": 18560 }, { "epoch": 1.2831662633944003, "grad_norm": 0.6801555463690134, "learning_rate": 3.0076150751008303e-06, "loss": 0.1888, "step": 18561 }, { "epoch": 1.2832353957829243, "grad_norm": 0.6422013455350996, "learning_rate": 3.0071016885585667e-06, "loss": 0.3068, "step": 18562 }, { "epoch": 1.2833045281714484, "grad_norm": 0.6204493473012961, "learning_rate": 3.006588326993725e-06, "loss": 0.2812, "step": 18563 }, { "epoch": 1.2833736605599724, "grad_norm": 0.6936739541108426, "learning_rate": 3.0060749904127422e-06, "loss": 0.2478, "step": 18564 }, { "epoch": 1.2834427929484964, "grad_norm": 0.7156835816834627, "learning_rate": 3.00556167882205e-06, "loss": 0.2743, "step": 18565 }, { "epoch": 1.2835119253370204, "grad_norm": 0.6383705406117681, "learning_rate": 3.005048392228085e-06, "loss": 0.3764, "step": 18566 }, { "epoch": 1.2835810577255444, "grad_norm": 0.7167613935912645, "learning_rate": 3.004535130637276e-06, "loss": 0.2263, "step": 18567 }, { "epoch": 1.2836501901140684, "grad_norm": 0.5958379724974362, "learning_rate": 3.0040218940560596e-06, "loss": 0.213, "step": 18568 }, { "epoch": 1.2837193225025925, "grad_norm": 0.6740866943959054, "learning_rate": 3.0035086824908654e-06, "loss": 0.2904, "step": 18569 }, { "epoch": 1.2837884548911165, "grad_norm": 0.7046145566045484, "learning_rate": 3.0029954959481295e-06, "loss": 0.3077, "step": 18570 }, { "epoch": 1.2838575872796405, "grad_norm": 1.4136160562658706, "learning_rate": 3.0024823344342786e-06, "loss": 0.2627, "step": 18571 }, { "epoch": 1.2839267196681645, "grad_norm": 0.6346621047986126, "learning_rate": 3.001969197955748e-06, "loss": 0.3, "step": 18572 }, { "epoch": 1.2839958520566885, "grad_norm": 0.7656778187165101, "learning_rate": 3.0014560865189666e-06, "loss": 0.3696, "step": 18573 }, { "epoch": 1.2840649844452126, "grad_norm": 0.6840726553227643, "learning_rate": 3.0009430001303686e-06, "loss": 0.2615, "step": 18574 }, { "epoch": 1.2841341168337366, "grad_norm": 0.7271503818152336, "learning_rate": 3.000429938796381e-06, "loss": 0.2387, "step": 18575 }, { "epoch": 1.2842032492222606, "grad_norm": 0.6830605832572688, "learning_rate": 2.9999169025234375e-06, "loss": 0.328, "step": 18576 }, { "epoch": 1.2842723816107846, "grad_norm": 0.6652570968531547, "learning_rate": 2.9994038913179646e-06, "loss": 0.3061, "step": 18577 }, { "epoch": 1.2843415139993086, "grad_norm": 0.6433194961323706, "learning_rate": 2.9988909051863944e-06, "loss": 0.2181, "step": 18578 }, { "epoch": 1.2844106463878326, "grad_norm": 0.6001495001295669, "learning_rate": 2.9983779441351547e-06, "loss": 0.2007, "step": 18579 }, { "epoch": 1.2844797787763567, "grad_norm": 0.6784402296644623, "learning_rate": 2.997865008170678e-06, "loss": 0.3726, "step": 18580 }, { "epoch": 1.2845489111648807, "grad_norm": 0.66289088796036, "learning_rate": 2.9973520972993877e-06, "loss": 0.2591, "step": 18581 }, { "epoch": 1.2846180435534047, "grad_norm": 0.5997419410609132, "learning_rate": 2.9968392115277162e-06, "loss": 0.2258, "step": 18582 }, { "epoch": 1.2846871759419287, "grad_norm": 0.6868976318345784, "learning_rate": 2.9963263508620895e-06, "loss": 0.2861, "step": 18583 }, { "epoch": 1.2847563083304527, "grad_norm": 0.6812669206171216, "learning_rate": 2.9958135153089385e-06, "loss": 0.3907, "step": 18584 }, { "epoch": 1.284825440718977, "grad_norm": 0.6952148760838992, "learning_rate": 2.9953007048746856e-06, "loss": 0.2678, "step": 18585 }, { "epoch": 1.284894573107501, "grad_norm": 0.6225488270041528, "learning_rate": 2.9947879195657625e-06, "loss": 0.2287, "step": 18586 }, { "epoch": 1.284963705496025, "grad_norm": 0.6369410935124834, "learning_rate": 2.9942751593885934e-06, "loss": 0.3381, "step": 18587 }, { "epoch": 1.285032837884549, "grad_norm": 0.6860062272661335, "learning_rate": 2.993762424349608e-06, "loss": 0.3799, "step": 18588 }, { "epoch": 1.285101970273073, "grad_norm": 0.5817020776931077, "learning_rate": 2.9932497144552274e-06, "loss": 0.1949, "step": 18589 }, { "epoch": 1.285171102661597, "grad_norm": 0.6880675595662568, "learning_rate": 2.9927370297118818e-06, "loss": 0.2505, "step": 18590 }, { "epoch": 1.285240235050121, "grad_norm": 0.7045257999236099, "learning_rate": 2.9922243701259945e-06, "loss": 0.3633, "step": 18591 }, { "epoch": 1.285309367438645, "grad_norm": 0.642792555843289, "learning_rate": 2.991711735703993e-06, "loss": 0.2508, "step": 18592 }, { "epoch": 1.2853784998271691, "grad_norm": 0.6275508320693747, "learning_rate": 2.9911991264522998e-06, "loss": 0.1919, "step": 18593 }, { "epoch": 1.2854476322156931, "grad_norm": 0.6403083371519732, "learning_rate": 2.9906865423773423e-06, "loss": 0.2538, "step": 18594 }, { "epoch": 1.2855167646042172, "grad_norm": 0.7030552597380197, "learning_rate": 2.9901739834855412e-06, "loss": 0.3526, "step": 18595 }, { "epoch": 1.2855858969927412, "grad_norm": 0.9876764776816979, "learning_rate": 2.9896614497833244e-06, "loss": 0.2475, "step": 18596 }, { "epoch": 1.2856550293812652, "grad_norm": 0.6915016245512382, "learning_rate": 2.9891489412771126e-06, "loss": 0.259, "step": 18597 }, { "epoch": 1.2857241617697892, "grad_norm": 0.7196946899325865, "learning_rate": 2.988636457973331e-06, "loss": 0.4272, "step": 18598 }, { "epoch": 1.2857932941583132, "grad_norm": 1.0379098992579796, "learning_rate": 2.9881239998784013e-06, "loss": 0.2367, "step": 18599 }, { "epoch": 1.2858624265468372, "grad_norm": 0.6554394002340298, "learning_rate": 2.987611566998747e-06, "loss": 0.2718, "step": 18600 }, { "epoch": 1.2859315589353613, "grad_norm": 0.6643018380510322, "learning_rate": 2.98709915934079e-06, "loss": 0.3014, "step": 18601 }, { "epoch": 1.2860006913238853, "grad_norm": 0.6940313932571825, "learning_rate": 2.9865867769109545e-06, "loss": 0.3546, "step": 18602 }, { "epoch": 1.2860698237124093, "grad_norm": 0.628265962233576, "learning_rate": 2.9860744197156587e-06, "loss": 0.2506, "step": 18603 }, { "epoch": 1.2861389561009333, "grad_norm": 0.5893035274326985, "learning_rate": 2.9855620877613277e-06, "loss": 0.2325, "step": 18604 }, { "epoch": 1.2862080884894573, "grad_norm": 0.6354512896749814, "learning_rate": 2.9850497810543794e-06, "loss": 0.341, "step": 18605 }, { "epoch": 1.2862772208779814, "grad_norm": 0.6481700122093003, "learning_rate": 2.984537499601238e-06, "loss": 0.2581, "step": 18606 }, { "epoch": 1.2863463532665054, "grad_norm": 0.7425846150872256, "learning_rate": 2.984025243408321e-06, "loss": 0.2677, "step": 18607 }, { "epoch": 1.2864154856550294, "grad_norm": 0.7182317226050907, "learning_rate": 2.9835130124820503e-06, "loss": 0.3271, "step": 18608 }, { "epoch": 1.2864846180435534, "grad_norm": 0.7163619993449161, "learning_rate": 2.9830008068288453e-06, "loss": 0.3621, "step": 18609 }, { "epoch": 1.2865537504320774, "grad_norm": 0.6060537633544647, "learning_rate": 2.9824886264551255e-06, "loss": 0.2379, "step": 18610 }, { "epoch": 1.2866228828206014, "grad_norm": 0.7021528013976432, "learning_rate": 2.9819764713673098e-06, "loss": 0.2241, "step": 18611 }, { "epoch": 1.2866920152091255, "grad_norm": 0.6203537358652613, "learning_rate": 2.9814643415718204e-06, "loss": 0.2982, "step": 18612 }, { "epoch": 1.2867611475976495, "grad_norm": 1.4818290187567715, "learning_rate": 2.9809522370750698e-06, "loss": 0.282, "step": 18613 }, { "epoch": 1.2868302799861735, "grad_norm": 0.6305730300939535, "learning_rate": 2.980440157883482e-06, "loss": 0.2369, "step": 18614 }, { "epoch": 1.2868994123746975, "grad_norm": 0.7371050668054435, "learning_rate": 2.979928104003471e-06, "loss": 0.2617, "step": 18615 }, { "epoch": 1.2869685447632215, "grad_norm": 0.6673482022146898, "learning_rate": 2.9794160754414594e-06, "loss": 0.3495, "step": 18616 }, { "epoch": 1.2870376771517456, "grad_norm": 0.6669259967504322, "learning_rate": 2.9789040722038583e-06, "loss": 0.2581, "step": 18617 }, { "epoch": 1.2871068095402696, "grad_norm": 0.6707976413502426, "learning_rate": 2.978392094297089e-06, "loss": 0.2521, "step": 18618 }, { "epoch": 1.2871759419287936, "grad_norm": 0.6586158168777748, "learning_rate": 2.977880141727566e-06, "loss": 0.2647, "step": 18619 }, { "epoch": 1.2872450743173176, "grad_norm": 0.8633999482511148, "learning_rate": 2.9773682145017092e-06, "loss": 0.3154, "step": 18620 }, { "epoch": 1.2873142067058416, "grad_norm": 0.6702989939199774, "learning_rate": 2.97685631262593e-06, "loss": 0.2587, "step": 18621 }, { "epoch": 1.2873833390943656, "grad_norm": 0.6748282837362163, "learning_rate": 2.9763444361066474e-06, "loss": 0.213, "step": 18622 }, { "epoch": 1.2874524714828897, "grad_norm": 0.7300546927933372, "learning_rate": 2.9758325849502757e-06, "loss": 0.3559, "step": 18623 }, { "epoch": 1.2875216038714137, "grad_norm": 0.6961185294097837, "learning_rate": 2.975320759163231e-06, "loss": 0.2589, "step": 18624 }, { "epoch": 1.2875907362599377, "grad_norm": 0.6609622148683763, "learning_rate": 2.974808958751927e-06, "loss": 0.2373, "step": 18625 }, { "epoch": 1.2876598686484617, "grad_norm": 0.6504431451615814, "learning_rate": 2.9742971837227776e-06, "loss": 0.3037, "step": 18626 }, { "epoch": 1.2877290010369857, "grad_norm": 0.7033736231556958, "learning_rate": 2.9737854340822004e-06, "loss": 0.3193, "step": 18627 }, { "epoch": 1.2877981334255098, "grad_norm": 0.7394829497136738, "learning_rate": 2.973273709836603e-06, "loss": 0.2063, "step": 18628 }, { "epoch": 1.2878672658140338, "grad_norm": 0.6472720494292081, "learning_rate": 2.972762010992406e-06, "loss": 0.2456, "step": 18629 }, { "epoch": 1.2879363982025578, "grad_norm": 0.705476422139925, "learning_rate": 2.972250337556018e-06, "loss": 0.344, "step": 18630 }, { "epoch": 1.2880055305910818, "grad_norm": 0.7711893886554511, "learning_rate": 2.971738689533854e-06, "loss": 0.322, "step": 18631 }, { "epoch": 1.2880746629796058, "grad_norm": 0.6251864974251369, "learning_rate": 2.971227066932325e-06, "loss": 0.2529, "step": 18632 }, { "epoch": 1.2881437953681298, "grad_norm": 0.6648895299951535, "learning_rate": 2.9707154697578453e-06, "loss": 0.2506, "step": 18633 }, { "epoch": 1.2882129277566539, "grad_norm": 0.7227176829975189, "learning_rate": 2.9702038980168245e-06, "loss": 0.3455, "step": 18634 }, { "epoch": 1.288282060145178, "grad_norm": 0.6449032020051741, "learning_rate": 2.9696923517156772e-06, "loss": 0.2569, "step": 18635 }, { "epoch": 1.2883511925337021, "grad_norm": 0.6466432419103608, "learning_rate": 2.9691808308608116e-06, "loss": 0.2115, "step": 18636 }, { "epoch": 1.2884203249222261, "grad_norm": 0.8150430866703947, "learning_rate": 2.9686693354586415e-06, "loss": 0.3844, "step": 18637 }, { "epoch": 1.2884894573107502, "grad_norm": 0.6528085097677769, "learning_rate": 2.9681578655155747e-06, "loss": 0.2594, "step": 18638 }, { "epoch": 1.2885585896992742, "grad_norm": 0.708593486733986, "learning_rate": 2.967646421038024e-06, "loss": 0.2212, "step": 18639 }, { "epoch": 1.2886277220877982, "grad_norm": 0.7147357362820097, "learning_rate": 2.9671350020323974e-06, "loss": 0.1905, "step": 18640 }, { "epoch": 1.2886968544763222, "grad_norm": 0.6219584561361381, "learning_rate": 2.966623608505108e-06, "loss": 0.3532, "step": 18641 }, { "epoch": 1.2887659868648462, "grad_norm": 0.7564408407267192, "learning_rate": 2.9661122404625607e-06, "loss": 0.2583, "step": 18642 }, { "epoch": 1.2888351192533702, "grad_norm": 0.7201490918681648, "learning_rate": 2.9656008979111684e-06, "loss": 0.2646, "step": 18643 }, { "epoch": 1.2889042516418943, "grad_norm": 0.687441078659778, "learning_rate": 2.9650895808573364e-06, "loss": 0.3326, "step": 18644 }, { "epoch": 1.2889733840304183, "grad_norm": 0.7120734089332817, "learning_rate": 2.9645782893074766e-06, "loss": 0.3452, "step": 18645 }, { "epoch": 1.2890425164189423, "grad_norm": 0.6687532636376792, "learning_rate": 2.964067023267994e-06, "loss": 0.2393, "step": 18646 }, { "epoch": 1.2891116488074663, "grad_norm": 0.6498162556238851, "learning_rate": 2.963555782745301e-06, "loss": 0.2471, "step": 18647 }, { "epoch": 1.2891807811959903, "grad_norm": 0.6942008737623809, "learning_rate": 2.963044567745799e-06, "loss": 0.353, "step": 18648 }, { "epoch": 1.2892499135845144, "grad_norm": 0.6024176562819246, "learning_rate": 2.9625333782758996e-06, "loss": 0.2398, "step": 18649 }, { "epoch": 1.2893190459730384, "grad_norm": 0.6433365188584161, "learning_rate": 2.9620222143420074e-06, "loss": 0.2331, "step": 18650 }, { "epoch": 1.2893881783615624, "grad_norm": 0.6921383011819869, "learning_rate": 2.9615110759505315e-06, "loss": 0.3194, "step": 18651 }, { "epoch": 1.2894573107500864, "grad_norm": 0.723157977670133, "learning_rate": 2.9609999631078752e-06, "loss": 0.3414, "step": 18652 }, { "epoch": 1.2895264431386104, "grad_norm": 0.7595945095622775, "learning_rate": 2.9604888758204465e-06, "loss": 0.2344, "step": 18653 }, { "epoch": 1.2895955755271344, "grad_norm": 0.6923628190495106, "learning_rate": 2.9599778140946485e-06, "loss": 0.303, "step": 18654 }, { "epoch": 1.2896647079156585, "grad_norm": 1.4565799349865143, "learning_rate": 2.9594667779368903e-06, "loss": 0.3662, "step": 18655 }, { "epoch": 1.2897338403041825, "grad_norm": 0.7399464707687319, "learning_rate": 2.958955767353572e-06, "loss": 0.2931, "step": 18656 }, { "epoch": 1.2898029726927065, "grad_norm": 0.689973292444381, "learning_rate": 2.958444782351102e-06, "loss": 0.2365, "step": 18657 }, { "epoch": 1.2898721050812305, "grad_norm": 0.7171441220037625, "learning_rate": 2.9579338229358826e-06, "loss": 0.3188, "step": 18658 }, { "epoch": 1.2899412374697545, "grad_norm": 0.7139764684325255, "learning_rate": 2.957422889114321e-06, "loss": 0.325, "step": 18659 }, { "epoch": 1.2900103698582785, "grad_norm": 0.6333511447626662, "learning_rate": 2.9569119808928154e-06, "loss": 0.2658, "step": 18660 }, { "epoch": 1.2900795022468026, "grad_norm": 0.5951114835131983, "learning_rate": 2.9564010982777736e-06, "loss": 0.2254, "step": 18661 }, { "epoch": 1.2901486346353266, "grad_norm": 0.6644874953621658, "learning_rate": 2.9558902412755957e-06, "loss": 0.2878, "step": 18662 }, { "epoch": 1.2902177670238506, "grad_norm": 0.6781486601331642, "learning_rate": 2.955379409892687e-06, "loss": 0.2643, "step": 18663 }, { "epoch": 1.2902868994123746, "grad_norm": 0.6496729386610144, "learning_rate": 2.9548686041354472e-06, "loss": 0.2253, "step": 18664 }, { "epoch": 1.2903560318008986, "grad_norm": 0.7296220139736067, "learning_rate": 2.954357824010282e-06, "loss": 0.2879, "step": 18665 }, { "epoch": 1.2904251641894229, "grad_norm": 0.6432965446393055, "learning_rate": 2.9538470695235896e-06, "loss": 0.3659, "step": 18666 }, { "epoch": 1.290494296577947, "grad_norm": 0.6193541600381594, "learning_rate": 2.953336340681773e-06, "loss": 0.2161, "step": 18667 }, { "epoch": 1.290563428966471, "grad_norm": 0.6559801707163148, "learning_rate": 2.9528256374912317e-06, "loss": 0.2303, "step": 18668 }, { "epoch": 1.290632561354995, "grad_norm": 0.6717047119692909, "learning_rate": 2.95231495995837e-06, "loss": 0.2737, "step": 18669 }, { "epoch": 1.290701693743519, "grad_norm": 0.6711324598750197, "learning_rate": 2.951804308089584e-06, "loss": 0.2897, "step": 18670 }, { "epoch": 1.290770826132043, "grad_norm": 3.345837304891656, "learning_rate": 2.9512936818912767e-06, "loss": 0.2594, "step": 18671 }, { "epoch": 1.290839958520567, "grad_norm": 0.7708487425725877, "learning_rate": 2.950783081369846e-06, "loss": 0.2891, "step": 18672 }, { "epoch": 1.290909090909091, "grad_norm": 0.7173071340369305, "learning_rate": 2.9502725065316952e-06, "loss": 0.3687, "step": 18673 }, { "epoch": 1.290978223297615, "grad_norm": 0.6563945640056383, "learning_rate": 2.949761957383217e-06, "loss": 0.2607, "step": 18674 }, { "epoch": 1.291047355686139, "grad_norm": 0.709844834083642, "learning_rate": 2.9492514339308166e-06, "loss": 0.2761, "step": 18675 }, { "epoch": 1.291116488074663, "grad_norm": 0.6866036272283792, "learning_rate": 2.9487409361808874e-06, "loss": 0.307, "step": 18676 }, { "epoch": 1.291185620463187, "grad_norm": 0.6607857845249888, "learning_rate": 2.948230464139833e-06, "loss": 0.2977, "step": 18677 }, { "epoch": 1.291254752851711, "grad_norm": 0.6274701620940145, "learning_rate": 2.947720017814045e-06, "loss": 0.2066, "step": 18678 }, { "epoch": 1.2913238852402351, "grad_norm": 0.6477520073848974, "learning_rate": 2.9472095972099256e-06, "loss": 0.2593, "step": 18679 }, { "epoch": 1.2913930176287591, "grad_norm": 0.71509385963516, "learning_rate": 2.9466992023338703e-06, "loss": 0.3567, "step": 18680 }, { "epoch": 1.2914621500172832, "grad_norm": 0.7296186662625392, "learning_rate": 2.946188833192276e-06, "loss": 0.3081, "step": 18681 }, { "epoch": 1.2915312824058072, "grad_norm": 0.6307258019809202, "learning_rate": 2.945678489791539e-06, "loss": 0.2457, "step": 18682 }, { "epoch": 1.2916004147943312, "grad_norm": 0.6592371843154889, "learning_rate": 2.9451681721380587e-06, "loss": 0.2434, "step": 18683 }, { "epoch": 1.2916695471828552, "grad_norm": 0.7054815192672849, "learning_rate": 2.944657880238225e-06, "loss": 0.3936, "step": 18684 }, { "epoch": 1.2917386795713792, "grad_norm": 1.2066964546497363, "learning_rate": 2.9441476140984392e-06, "loss": 0.2387, "step": 18685 }, { "epoch": 1.2918078119599032, "grad_norm": 0.6877813859534677, "learning_rate": 2.9436373737250924e-06, "loss": 0.2949, "step": 18686 }, { "epoch": 1.2918769443484273, "grad_norm": 0.6843428890704334, "learning_rate": 2.9431271591245835e-06, "loss": 0.3241, "step": 18687 }, { "epoch": 1.2919460767369513, "grad_norm": 0.6618866361194107, "learning_rate": 2.942616970303304e-06, "loss": 0.269, "step": 18688 }, { "epoch": 1.2920152091254753, "grad_norm": 0.7177860991796434, "learning_rate": 2.9421068072676484e-06, "loss": 0.2492, "step": 18689 }, { "epoch": 1.2920843415139993, "grad_norm": 0.6600379967922104, "learning_rate": 2.9415966700240116e-06, "loss": 0.2861, "step": 18690 }, { "epoch": 1.2921534739025233, "grad_norm": 0.65946269918207, "learning_rate": 2.9410865585787897e-06, "loss": 0.3733, "step": 18691 }, { "epoch": 1.2922226062910473, "grad_norm": 0.7817254354896526, "learning_rate": 2.940576472938371e-06, "loss": 0.2195, "step": 18692 }, { "epoch": 1.2922917386795714, "grad_norm": 0.7467999405383468, "learning_rate": 2.9400664131091513e-06, "loss": 0.2112, "step": 18693 }, { "epoch": 1.2923608710680954, "grad_norm": 0.7504380057689155, "learning_rate": 2.9395563790975244e-06, "loss": 0.2761, "step": 18694 }, { "epoch": 1.2924300034566194, "grad_norm": 0.6414902778982204, "learning_rate": 2.9390463709098794e-06, "loss": 0.275, "step": 18695 }, { "epoch": 1.2924991358451434, "grad_norm": 0.7225132617795695, "learning_rate": 2.938536388552611e-06, "loss": 0.2358, "step": 18696 }, { "epoch": 1.2925682682336674, "grad_norm": 0.6475161191424097, "learning_rate": 2.9380264320321096e-06, "loss": 0.2138, "step": 18697 }, { "epoch": 1.2926374006221915, "grad_norm": 0.6125106681316258, "learning_rate": 2.9375165013547678e-06, "loss": 0.3137, "step": 18698 }, { "epoch": 1.2927065330107155, "grad_norm": 0.6902527177973883, "learning_rate": 2.9370065965269746e-06, "loss": 0.2767, "step": 18699 }, { "epoch": 1.2927756653992395, "grad_norm": 0.7462145375598026, "learning_rate": 2.9364967175551247e-06, "loss": 0.2609, "step": 18700 }, { "epoch": 1.2928447977877635, "grad_norm": 0.6727287973086835, "learning_rate": 2.9359868644456034e-06, "loss": 0.3071, "step": 18701 }, { "epoch": 1.2929139301762875, "grad_norm": 0.7231888207519681, "learning_rate": 2.9354770372048046e-06, "loss": 0.3782, "step": 18702 }, { "epoch": 1.2929830625648115, "grad_norm": 0.6789026298938693, "learning_rate": 2.9349672358391163e-06, "loss": 0.2392, "step": 18703 }, { "epoch": 1.2930521949533356, "grad_norm": 0.7359040292445301, "learning_rate": 2.9344574603549304e-06, "loss": 0.2655, "step": 18704 }, { "epoch": 1.2931213273418596, "grad_norm": 0.6629080742640158, "learning_rate": 2.933947710758631e-06, "loss": 0.3701, "step": 18705 }, { "epoch": 1.2931904597303836, "grad_norm": 0.6536196652579627, "learning_rate": 2.9334379870566123e-06, "loss": 0.2465, "step": 18706 }, { "epoch": 1.2932595921189076, "grad_norm": 0.7252771981998404, "learning_rate": 2.9329282892552592e-06, "loss": 0.2692, "step": 18707 }, { "epoch": 1.2933287245074316, "grad_norm": 0.7055322197949074, "learning_rate": 2.9324186173609624e-06, "loss": 0.3122, "step": 18708 }, { "epoch": 1.2933978568959557, "grad_norm": 0.6345240589964365, "learning_rate": 2.9319089713801076e-06, "loss": 0.3138, "step": 18709 }, { "epoch": 1.2934669892844797, "grad_norm": 0.6939807194278261, "learning_rate": 2.9313993513190835e-06, "loss": 0.2993, "step": 18710 }, { "epoch": 1.2935361216730037, "grad_norm": 0.5794572697678093, "learning_rate": 2.930889757184276e-06, "loss": 0.2322, "step": 18711 }, { "epoch": 1.2936052540615277, "grad_norm": 0.697644013643368, "learning_rate": 2.930380188982075e-06, "loss": 0.3225, "step": 18712 }, { "epoch": 1.2936743864500517, "grad_norm": 0.8254539011997978, "learning_rate": 2.929870646718863e-06, "loss": 0.2651, "step": 18713 }, { "epoch": 1.2937435188385757, "grad_norm": 0.5790472342040843, "learning_rate": 2.9293611304010287e-06, "loss": 0.1947, "step": 18714 }, { "epoch": 1.2938126512270998, "grad_norm": 0.6860832873716204, "learning_rate": 2.928851640034957e-06, "loss": 0.3093, "step": 18715 }, { "epoch": 1.2938817836156238, "grad_norm": 0.7136481869933694, "learning_rate": 2.928342175627036e-06, "loss": 0.3424, "step": 18716 }, { "epoch": 1.293950916004148, "grad_norm": 1.8616594281376382, "learning_rate": 2.9278327371836446e-06, "loss": 0.2502, "step": 18717 }, { "epoch": 1.294020048392672, "grad_norm": 0.6876003944533081, "learning_rate": 2.927323324711176e-06, "loss": 0.2347, "step": 18718 }, { "epoch": 1.294089180781196, "grad_norm": 0.6695319418539802, "learning_rate": 2.9268139382160083e-06, "loss": 0.3078, "step": 18719 }, { "epoch": 1.29415831316972, "grad_norm": 0.6370335052665821, "learning_rate": 2.9263045777045296e-06, "loss": 0.2798, "step": 18720 }, { "epoch": 1.294227445558244, "grad_norm": 0.7353382635893697, "learning_rate": 2.9257952431831203e-06, "loss": 0.2723, "step": 18721 }, { "epoch": 1.2942965779467681, "grad_norm": 0.7249701390273484, "learning_rate": 2.9252859346581685e-06, "loss": 0.2844, "step": 18722 }, { "epoch": 1.2943657103352921, "grad_norm": 0.858841600575906, "learning_rate": 2.9247766521360533e-06, "loss": 0.3844, "step": 18723 }, { "epoch": 1.2944348427238161, "grad_norm": 0.6333607357284015, "learning_rate": 2.9242673956231605e-06, "loss": 0.2576, "step": 18724 }, { "epoch": 1.2945039751123402, "grad_norm": 0.6920727063811454, "learning_rate": 2.923758165125871e-06, "loss": 0.242, "step": 18725 }, { "epoch": 1.2945731075008642, "grad_norm": 0.8236346419068484, "learning_rate": 2.9232489606505683e-06, "loss": 0.3421, "step": 18726 }, { "epoch": 1.2946422398893882, "grad_norm": 0.7185543923659657, "learning_rate": 2.9227397822036323e-06, "loss": 0.3247, "step": 18727 }, { "epoch": 1.2947113722779122, "grad_norm": 0.7515675790742907, "learning_rate": 2.922230629791447e-06, "loss": 0.275, "step": 18728 }, { "epoch": 1.2947805046664362, "grad_norm": 0.6460674353289796, "learning_rate": 2.9217215034203927e-06, "loss": 0.2437, "step": 18729 }, { "epoch": 1.2948496370549603, "grad_norm": 0.7139655465354935, "learning_rate": 2.9212124030968513e-06, "loss": 0.34, "step": 18730 }, { "epoch": 1.2949187694434843, "grad_norm": 0.6639577542908822, "learning_rate": 2.9207033288272014e-06, "loss": 0.2836, "step": 18731 }, { "epoch": 1.2949879018320083, "grad_norm": 0.5840904519848974, "learning_rate": 2.920194280617825e-06, "loss": 0.1987, "step": 18732 }, { "epoch": 1.2950570342205323, "grad_norm": 0.8292512804835039, "learning_rate": 2.919685258475101e-06, "loss": 0.2916, "step": 18733 }, { "epoch": 1.2951261666090563, "grad_norm": 0.7068364462212944, "learning_rate": 2.919176262405412e-06, "loss": 0.3854, "step": 18734 }, { "epoch": 1.2951952989975803, "grad_norm": 0.6687947476104935, "learning_rate": 2.918667292415134e-06, "loss": 0.2815, "step": 18735 }, { "epoch": 1.2952644313861044, "grad_norm": 0.6996256305112029, "learning_rate": 2.9181583485106456e-06, "loss": 0.2499, "step": 18736 }, { "epoch": 1.2953335637746284, "grad_norm": 0.7690920225493163, "learning_rate": 2.9176494306983276e-06, "loss": 0.341, "step": 18737 }, { "epoch": 1.2954026961631524, "grad_norm": 0.7104231878080908, "learning_rate": 2.9171405389845605e-06, "loss": 0.3414, "step": 18738 }, { "epoch": 1.2954718285516764, "grad_norm": 0.5941930529228007, "learning_rate": 2.9166316733757173e-06, "loss": 0.2052, "step": 18739 }, { "epoch": 1.2955409609402004, "grad_norm": 0.6941728836487178, "learning_rate": 2.91612283387818e-06, "loss": 0.283, "step": 18740 }, { "epoch": 1.2956100933287245, "grad_norm": 0.6889346023494619, "learning_rate": 2.915614020498322e-06, "loss": 0.324, "step": 18741 }, { "epoch": 1.2956792257172485, "grad_norm": 0.6341536400493141, "learning_rate": 2.9151052332425247e-06, "loss": 0.2068, "step": 18742 }, { "epoch": 1.2957483581057725, "grad_norm": 0.7053952017739783, "learning_rate": 2.9145964721171602e-06, "loss": 0.2428, "step": 18743 }, { "epoch": 1.2958174904942965, "grad_norm": 0.6441166162574256, "learning_rate": 2.914087737128608e-06, "loss": 0.305, "step": 18744 }, { "epoch": 1.2958866228828205, "grad_norm": 0.6425469312615307, "learning_rate": 2.9135790282832433e-06, "loss": 0.3345, "step": 18745 }, { "epoch": 1.2959557552713445, "grad_norm": 0.6205551712104285, "learning_rate": 2.9130703455874445e-06, "loss": 0.2348, "step": 18746 }, { "epoch": 1.2960248876598688, "grad_norm": 0.7195896846460541, "learning_rate": 2.9125616890475818e-06, "loss": 0.2557, "step": 18747 }, { "epoch": 1.2960940200483928, "grad_norm": 0.6597487010486554, "learning_rate": 2.912053058670036e-06, "loss": 0.3684, "step": 18748 }, { "epoch": 1.2961631524369168, "grad_norm": 0.7147583129866623, "learning_rate": 2.911544454461176e-06, "loss": 0.3506, "step": 18749 }, { "epoch": 1.2962322848254408, "grad_norm": 0.6603612925796961, "learning_rate": 2.911035876427382e-06, "loss": 0.2319, "step": 18750 }, { "epoch": 1.2963014172139649, "grad_norm": 0.6536921449001587, "learning_rate": 2.910527324575021e-06, "loss": 0.3261, "step": 18751 }, { "epoch": 1.2963705496024889, "grad_norm": 0.7397221350605164, "learning_rate": 2.9100187989104756e-06, "loss": 0.3473, "step": 18752 }, { "epoch": 1.296439681991013, "grad_norm": 0.6692097638031196, "learning_rate": 2.9095102994401124e-06, "loss": 0.2487, "step": 18753 }, { "epoch": 1.296508814379537, "grad_norm": 0.7579639706826676, "learning_rate": 2.909001826170309e-06, "loss": 0.2421, "step": 18754 }, { "epoch": 1.296577946768061, "grad_norm": 0.6927187000555971, "learning_rate": 2.9084933791074343e-06, "loss": 0.361, "step": 18755 }, { "epoch": 1.296647079156585, "grad_norm": 0.6349367414535761, "learning_rate": 2.9079849582578646e-06, "loss": 0.2531, "step": 18756 }, { "epoch": 1.296716211545109, "grad_norm": 0.707466086786834, "learning_rate": 2.9074765636279685e-06, "loss": 0.2605, "step": 18757 }, { "epoch": 1.296785343933633, "grad_norm": 0.6574548880792072, "learning_rate": 2.906968195224119e-06, "loss": 0.2579, "step": 18758 }, { "epoch": 1.296854476322157, "grad_norm": 0.7081262484236397, "learning_rate": 2.90645985305269e-06, "loss": 0.3506, "step": 18759 }, { "epoch": 1.296923608710681, "grad_norm": 0.6397976651456532, "learning_rate": 2.9059515371200488e-06, "loss": 0.2864, "step": 18760 }, { "epoch": 1.296992741099205, "grad_norm": 0.6976482327626712, "learning_rate": 2.9054432474325696e-06, "loss": 0.2018, "step": 18761 }, { "epoch": 1.297061873487729, "grad_norm": 1.0818753799436522, "learning_rate": 2.9049349839966197e-06, "loss": 0.3182, "step": 18762 }, { "epoch": 1.297131005876253, "grad_norm": 0.767678117705714, "learning_rate": 2.9044267468185726e-06, "loss": 0.3628, "step": 18763 }, { "epoch": 1.297200138264777, "grad_norm": 0.667293990566127, "learning_rate": 2.903918535904795e-06, "loss": 0.3031, "step": 18764 }, { "epoch": 1.297269270653301, "grad_norm": 0.7207342481251594, "learning_rate": 2.9034103512616574e-06, "loss": 0.3082, "step": 18765 }, { "epoch": 1.2973384030418251, "grad_norm": 0.6969118289790681, "learning_rate": 2.9029021928955303e-06, "loss": 0.4008, "step": 18766 }, { "epoch": 1.2974075354303491, "grad_norm": 0.6871871906985636, "learning_rate": 2.902394060812783e-06, "loss": 0.2391, "step": 18767 }, { "epoch": 1.2974766678188732, "grad_norm": 0.6641779807859268, "learning_rate": 2.9018859550197813e-06, "loss": 0.2463, "step": 18768 }, { "epoch": 1.2975458002073972, "grad_norm": 0.687178350473178, "learning_rate": 2.9013778755228973e-06, "loss": 0.3175, "step": 18769 }, { "epoch": 1.2976149325959212, "grad_norm": 0.7230591384877595, "learning_rate": 2.900869822328494e-06, "loss": 0.3118, "step": 18770 }, { "epoch": 1.2976840649844452, "grad_norm": 0.6834245309229068, "learning_rate": 2.900361795442941e-06, "loss": 0.2051, "step": 18771 }, { "epoch": 1.2977531973729692, "grad_norm": 0.6653435413859279, "learning_rate": 2.8998537948726064e-06, "loss": 0.2599, "step": 18772 }, { "epoch": 1.2978223297614933, "grad_norm": 1.0978225438079816, "learning_rate": 2.8993458206238585e-06, "loss": 0.2984, "step": 18773 }, { "epoch": 1.2978914621500173, "grad_norm": 1.2411463584509177, "learning_rate": 2.8988378727030596e-06, "loss": 0.2444, "step": 18774 }, { "epoch": 1.2979605945385413, "grad_norm": 0.6549245560992351, "learning_rate": 2.8983299511165807e-06, "loss": 0.2916, "step": 18775 }, { "epoch": 1.2980297269270653, "grad_norm": 0.6906154766085248, "learning_rate": 2.897822055870783e-06, "loss": 0.2877, "step": 18776 }, { "epoch": 1.2980988593155893, "grad_norm": 0.7026764911880907, "learning_rate": 2.8973141869720366e-06, "loss": 0.3311, "step": 18777 }, { "epoch": 1.2981679917041133, "grad_norm": 0.6562640085937266, "learning_rate": 2.8968063444267027e-06, "loss": 0.2461, "step": 18778 }, { "epoch": 1.2982371240926374, "grad_norm": 0.6870899842919251, "learning_rate": 2.896298528241147e-06, "loss": 0.2699, "step": 18779 }, { "epoch": 1.2983062564811614, "grad_norm": 0.806779975766335, "learning_rate": 2.895790738421736e-06, "loss": 0.3266, "step": 18780 }, { "epoch": 1.2983753888696854, "grad_norm": 0.630114644736259, "learning_rate": 2.8952829749748345e-06, "loss": 0.2519, "step": 18781 }, { "epoch": 1.2984445212582094, "grad_norm": 0.6297533647683843, "learning_rate": 2.8947752379068026e-06, "loss": 0.2478, "step": 18782 }, { "epoch": 1.2985136536467334, "grad_norm": 0.6807909142233353, "learning_rate": 2.894267527224009e-06, "loss": 0.2626, "step": 18783 }, { "epoch": 1.2985827860352575, "grad_norm": 0.6837359991356979, "learning_rate": 2.893759842932812e-06, "loss": 0.349, "step": 18784 }, { "epoch": 1.2986519184237815, "grad_norm": 0.6656876424321246, "learning_rate": 2.893252185039578e-06, "loss": 0.2287, "step": 18785 }, { "epoch": 1.2987210508123055, "grad_norm": 0.6211991637508698, "learning_rate": 2.892744553550666e-06, "loss": 0.2145, "step": 18786 }, { "epoch": 1.2987901832008295, "grad_norm": 0.7307743807640233, "learning_rate": 2.892236948472441e-06, "loss": 0.3272, "step": 18787 }, { "epoch": 1.2988593155893535, "grad_norm": 0.6689229156110869, "learning_rate": 2.8917293698112648e-06, "loss": 0.2968, "step": 18788 }, { "epoch": 1.2989284479778775, "grad_norm": 0.6929438160874452, "learning_rate": 2.8912218175734995e-06, "loss": 0.2701, "step": 18789 }, { "epoch": 1.2989975803664016, "grad_norm": 0.6656300024592569, "learning_rate": 2.890714291765504e-06, "loss": 0.2684, "step": 18790 }, { "epoch": 1.2990667127549256, "grad_norm": 0.8029712676340958, "learning_rate": 2.890206792393643e-06, "loss": 0.383, "step": 18791 }, { "epoch": 1.2991358451434496, "grad_norm": 0.6936427086336777, "learning_rate": 2.889699319464272e-06, "loss": 0.3052, "step": 18792 }, { "epoch": 1.2992049775319736, "grad_norm": 0.6209421994270894, "learning_rate": 2.889191872983754e-06, "loss": 0.2291, "step": 18793 }, { "epoch": 1.2992741099204976, "grad_norm": 0.7700227020672424, "learning_rate": 2.8886844529584503e-06, "loss": 0.3155, "step": 18794 }, { "epoch": 1.2993432423090217, "grad_norm": 0.760017507547695, "learning_rate": 2.88817705939472e-06, "loss": 0.4327, "step": 18795 }, { "epoch": 1.2994123746975457, "grad_norm": 0.7284831622573289, "learning_rate": 2.88766969229892e-06, "loss": 0.2349, "step": 18796 }, { "epoch": 1.2994815070860697, "grad_norm": 0.660284823856693, "learning_rate": 2.887162351677412e-06, "loss": 0.2965, "step": 18797 }, { "epoch": 1.299550639474594, "grad_norm": 0.749822687990208, "learning_rate": 2.8866550375365523e-06, "loss": 0.3663, "step": 18798 }, { "epoch": 1.299619771863118, "grad_norm": 1.0761203745516932, "learning_rate": 2.8861477498827016e-06, "loss": 0.2447, "step": 18799 }, { "epoch": 1.299688904251642, "grad_norm": 0.6466144531280604, "learning_rate": 2.885640488722215e-06, "loss": 0.2292, "step": 18800 }, { "epoch": 1.299758036640166, "grad_norm": 0.7636797726740241, "learning_rate": 2.8851332540614517e-06, "loss": 0.2632, "step": 18801 }, { "epoch": 1.29982716902869, "grad_norm": 0.6588647574983854, "learning_rate": 2.884626045906769e-06, "loss": 0.2725, "step": 18802 }, { "epoch": 1.299896301417214, "grad_norm": 0.626356635751786, "learning_rate": 2.8841188642645253e-06, "loss": 0.245, "step": 18803 }, { "epoch": 1.299965433805738, "grad_norm": 0.6971605813701565, "learning_rate": 2.8836117091410736e-06, "loss": 0.2659, "step": 18804 }, { "epoch": 1.300034566194262, "grad_norm": 0.7867250703095829, "learning_rate": 2.883104580542775e-06, "loss": 0.3481, "step": 18805 }, { "epoch": 1.300103698582786, "grad_norm": 0.7414537040796985, "learning_rate": 2.8825974784759797e-06, "loss": 0.199, "step": 18806 }, { "epoch": 1.30017283097131, "grad_norm": 0.6702041457029377, "learning_rate": 2.882090402947047e-06, "loss": 0.2226, "step": 18807 }, { "epoch": 1.300241963359834, "grad_norm": 0.6649793362817829, "learning_rate": 2.8815833539623312e-06, "loss": 0.2766, "step": 18808 }, { "epoch": 1.3003110957483581, "grad_norm": 0.6619865498487556, "learning_rate": 2.88107633152819e-06, "loss": 0.3351, "step": 18809 }, { "epoch": 1.3003802281368821, "grad_norm": 0.7559450112309396, "learning_rate": 2.8805693356509734e-06, "loss": 0.3085, "step": 18810 }, { "epoch": 1.3004493605254062, "grad_norm": 0.632725922165332, "learning_rate": 2.8800623663370397e-06, "loss": 0.2534, "step": 18811 }, { "epoch": 1.3005184929139302, "grad_norm": 0.660030472819629, "learning_rate": 2.8795554235927394e-06, "loss": 0.3368, "step": 18812 }, { "epoch": 1.3005876253024542, "grad_norm": 0.6824767135615725, "learning_rate": 2.879048507424429e-06, "loss": 0.2632, "step": 18813 }, { "epoch": 1.3006567576909782, "grad_norm": 0.6661183393290525, "learning_rate": 2.8785416178384594e-06, "loss": 0.2266, "step": 18814 }, { "epoch": 1.3007258900795022, "grad_norm": 0.9294502405445549, "learning_rate": 2.878034754841184e-06, "loss": 0.253, "step": 18815 }, { "epoch": 1.3007950224680263, "grad_norm": 0.6591622964709632, "learning_rate": 2.8775279184389573e-06, "loss": 0.3846, "step": 18816 }, { "epoch": 1.3008641548565503, "grad_norm": 0.6654421735150171, "learning_rate": 2.877021108638131e-06, "loss": 0.2885, "step": 18817 }, { "epoch": 1.3009332872450743, "grad_norm": 0.6366622807815442, "learning_rate": 2.876514325445055e-06, "loss": 0.2357, "step": 18818 }, { "epoch": 1.3010024196335983, "grad_norm": 0.6873314182307186, "learning_rate": 2.876007568866085e-06, "loss": 0.2509, "step": 18819 }, { "epoch": 1.3010715520221223, "grad_norm": 0.7070548817366907, "learning_rate": 2.8755008389075667e-06, "loss": 0.3363, "step": 18820 }, { "epoch": 1.3011406844106463, "grad_norm": 0.6447856288164577, "learning_rate": 2.874994135575857e-06, "loss": 0.2499, "step": 18821 }, { "epoch": 1.3012098167991704, "grad_norm": 0.6126365135714761, "learning_rate": 2.8744874588772985e-06, "loss": 0.2498, "step": 18822 }, { "epoch": 1.3012789491876944, "grad_norm": 0.6593302960251675, "learning_rate": 2.8739808088182514e-06, "loss": 0.3941, "step": 18823 }, { "epoch": 1.3013480815762184, "grad_norm": 0.8061415654339378, "learning_rate": 2.873474185405061e-06, "loss": 0.2522, "step": 18824 }, { "epoch": 1.3014172139647424, "grad_norm": 0.64165953199239, "learning_rate": 2.8729675886440737e-06, "loss": 0.2272, "step": 18825 }, { "epoch": 1.3014863463532664, "grad_norm": 0.7123348781608703, "learning_rate": 2.8724610185416444e-06, "loss": 0.3081, "step": 18826 }, { "epoch": 1.3015554787417905, "grad_norm": 0.7366125858833964, "learning_rate": 2.871954475104118e-06, "loss": 0.3025, "step": 18827 }, { "epoch": 1.3016246111303147, "grad_norm": 0.6148194805675825, "learning_rate": 2.871447958337844e-06, "loss": 0.2452, "step": 18828 }, { "epoch": 1.3016937435188387, "grad_norm": 0.7561478475693053, "learning_rate": 2.870941468249171e-06, "loss": 0.2459, "step": 18829 }, { "epoch": 1.3017628759073627, "grad_norm": 0.6347565643117001, "learning_rate": 2.8704350048444495e-06, "loss": 0.2768, "step": 18830 }, { "epoch": 1.3018320082958867, "grad_norm": 0.6909781431365876, "learning_rate": 2.869928568130023e-06, "loss": 0.2531, "step": 18831 }, { "epoch": 1.3019011406844108, "grad_norm": 0.6430624600021149, "learning_rate": 2.8694221581122426e-06, "loss": 0.2723, "step": 18832 }, { "epoch": 1.3019702730729348, "grad_norm": 0.657249034908714, "learning_rate": 2.8689157747974515e-06, "loss": 0.2279, "step": 18833 }, { "epoch": 1.3020394054614588, "grad_norm": 0.7056059621247189, "learning_rate": 2.8684094181919995e-06, "loss": 0.3245, "step": 18834 }, { "epoch": 1.3021085378499828, "grad_norm": 0.688638387650151, "learning_rate": 2.8679030883022297e-06, "loss": 0.2695, "step": 18835 }, { "epoch": 1.3021776702385068, "grad_norm": 0.6853984141525737, "learning_rate": 2.8673967851344897e-06, "loss": 0.2368, "step": 18836 }, { "epoch": 1.3022468026270309, "grad_norm": 1.2111901115444035, "learning_rate": 2.8668905086951255e-06, "loss": 0.2964, "step": 18837 }, { "epoch": 1.3023159350155549, "grad_norm": 0.7609892760074095, "learning_rate": 2.866384258990484e-06, "loss": 0.3355, "step": 18838 }, { "epoch": 1.3023850674040789, "grad_norm": 0.6158075128710747, "learning_rate": 2.8658780360269055e-06, "loss": 0.2147, "step": 18839 }, { "epoch": 1.302454199792603, "grad_norm": 0.6896256361714082, "learning_rate": 2.86537183981074e-06, "loss": 0.2544, "step": 18840 }, { "epoch": 1.302523332181127, "grad_norm": 0.6437902050694315, "learning_rate": 2.8648656703483267e-06, "loss": 0.3572, "step": 18841 }, { "epoch": 1.302592464569651, "grad_norm": 0.829857625177589, "learning_rate": 2.8643595276460124e-06, "loss": 0.2709, "step": 18842 }, { "epoch": 1.302661596958175, "grad_norm": 0.8272713755752664, "learning_rate": 2.86385341171014e-06, "loss": 0.2331, "step": 18843 }, { "epoch": 1.302730729346699, "grad_norm": 0.6177318580144361, "learning_rate": 2.8633473225470555e-06, "loss": 0.2844, "step": 18844 }, { "epoch": 1.302799861735223, "grad_norm": 0.6562694406695719, "learning_rate": 2.8628412601630963e-06, "loss": 0.3057, "step": 18845 }, { "epoch": 1.302868994123747, "grad_norm": 0.8097735391629474, "learning_rate": 2.862335224564611e-06, "loss": 0.2612, "step": 18846 }, { "epoch": 1.302938126512271, "grad_norm": 0.6916033099624203, "learning_rate": 2.8618292157579353e-06, "loss": 0.2758, "step": 18847 }, { "epoch": 1.303007258900795, "grad_norm": 0.6601005714497289, "learning_rate": 2.861323233749418e-06, "loss": 0.3446, "step": 18848 }, { "epoch": 1.303076391289319, "grad_norm": 0.6689681189671667, "learning_rate": 2.8608172785453948e-06, "loss": 0.2991, "step": 18849 }, { "epoch": 1.303145523677843, "grad_norm": 0.6771135222469207, "learning_rate": 2.8603113501522095e-06, "loss": 0.2429, "step": 18850 }, { "epoch": 1.303214656066367, "grad_norm": 0.9960130805439058, "learning_rate": 2.8598054485762026e-06, "loss": 0.2686, "step": 18851 }, { "epoch": 1.3032837884548911, "grad_norm": 0.644701051076437, "learning_rate": 2.859299573823718e-06, "loss": 0.3143, "step": 18852 }, { "epoch": 1.3033529208434151, "grad_norm": 0.7673174852525168, "learning_rate": 2.8587937259010902e-06, "loss": 0.2452, "step": 18853 }, { "epoch": 1.3034220532319392, "grad_norm": 0.5871059059153463, "learning_rate": 2.858287904814664e-06, "loss": 0.2104, "step": 18854 }, { "epoch": 1.3034911856204632, "grad_norm": 0.7099999803513293, "learning_rate": 2.857782110570775e-06, "loss": 0.3957, "step": 18855 }, { "epoch": 1.3035603180089872, "grad_norm": 0.7682102025848109, "learning_rate": 2.8572763431757665e-06, "loss": 0.3203, "step": 18856 }, { "epoch": 1.3036294503975112, "grad_norm": 0.7088192650593101, "learning_rate": 2.8567706026359733e-06, "loss": 0.2819, "step": 18857 }, { "epoch": 1.3036985827860352, "grad_norm": 0.6450519635187678, "learning_rate": 2.856264888957735e-06, "loss": 0.2616, "step": 18858 }, { "epoch": 1.3037677151745592, "grad_norm": 0.6668291027386841, "learning_rate": 2.855759202147391e-06, "loss": 0.3268, "step": 18859 }, { "epoch": 1.3038368475630833, "grad_norm": 0.599090001070603, "learning_rate": 2.8552535422112816e-06, "loss": 0.2083, "step": 18860 }, { "epoch": 1.3039059799516073, "grad_norm": 0.5760185103323542, "learning_rate": 2.8547479091557384e-06, "loss": 0.2032, "step": 18861 }, { "epoch": 1.3039751123401313, "grad_norm": 0.6312178393329646, "learning_rate": 2.8542423029871047e-06, "loss": 0.2863, "step": 18862 }, { "epoch": 1.3040442447286553, "grad_norm": 0.6592100601016736, "learning_rate": 2.853736723711712e-06, "loss": 0.2506, "step": 18863 }, { "epoch": 1.3041133771171793, "grad_norm": 0.5972764255695981, "learning_rate": 2.8532311713358995e-06, "loss": 0.223, "step": 18864 }, { "epoch": 1.3041825095057034, "grad_norm": 0.589104536019679, "learning_rate": 2.852725645866003e-06, "loss": 0.2177, "step": 18865 }, { "epoch": 1.3042516418942274, "grad_norm": 0.7498545712773554, "learning_rate": 2.8522201473083606e-06, "loss": 0.399, "step": 18866 }, { "epoch": 1.3043207742827514, "grad_norm": 0.6595524527606376, "learning_rate": 2.8517146756693035e-06, "loss": 0.2601, "step": 18867 }, { "epoch": 1.3043899066712754, "grad_norm": 0.609602233953413, "learning_rate": 2.851209230955172e-06, "loss": 0.2558, "step": 18868 }, { "epoch": 1.3044590390597994, "grad_norm": 0.6632024126479102, "learning_rate": 2.8507038131722952e-06, "loss": 0.3182, "step": 18869 }, { "epoch": 1.3045281714483234, "grad_norm": 0.6603270220971278, "learning_rate": 2.850198422327013e-06, "loss": 0.3161, "step": 18870 }, { "epoch": 1.3045973038368475, "grad_norm": 0.5935871099143418, "learning_rate": 2.8496930584256544e-06, "loss": 0.2483, "step": 18871 }, { "epoch": 1.3046664362253715, "grad_norm": 0.6850751380368401, "learning_rate": 2.8491877214745566e-06, "loss": 0.208, "step": 18872 }, { "epoch": 1.3047355686138955, "grad_norm": 0.6732711159844411, "learning_rate": 2.8486824114800527e-06, "loss": 0.3964, "step": 18873 }, { "epoch": 1.3048047010024195, "grad_norm": 0.9700492678891126, "learning_rate": 2.848177128448477e-06, "loss": 0.2961, "step": 18874 }, { "epoch": 1.3048738333909435, "grad_norm": 0.916256917248268, "learning_rate": 2.8476718723861584e-06, "loss": 0.2606, "step": 18875 }, { "epoch": 1.3049429657794676, "grad_norm": 1.1802318493333692, "learning_rate": 2.847166643299435e-06, "loss": 0.3286, "step": 18876 }, { "epoch": 1.3050120981679916, "grad_norm": 0.627024859766466, "learning_rate": 2.8466614411946338e-06, "loss": 0.2995, "step": 18877 }, { "epoch": 1.3050812305565156, "grad_norm": 0.7082995480871389, "learning_rate": 2.8461562660780875e-06, "loss": 0.2185, "step": 18878 }, { "epoch": 1.3051503629450398, "grad_norm": 0.6560364830412567, "learning_rate": 2.8456511179561297e-06, "loss": 0.2862, "step": 18879 }, { "epoch": 1.3052194953335639, "grad_norm": 0.6359023752429898, "learning_rate": 2.845145996835092e-06, "loss": 0.3064, "step": 18880 }, { "epoch": 1.3052886277220879, "grad_norm": 0.6893207437596813, "learning_rate": 2.8446409027213017e-06, "loss": 0.2796, "step": 18881 }, { "epoch": 1.3053577601106119, "grad_norm": 1.2352412127587196, "learning_rate": 2.8441358356210933e-06, "loss": 0.2409, "step": 18882 }, { "epoch": 1.305426892499136, "grad_norm": 0.8678688224734501, "learning_rate": 2.843630795540794e-06, "loss": 0.2686, "step": 18883 }, { "epoch": 1.30549602488766, "grad_norm": 0.685500312947281, "learning_rate": 2.8431257824867353e-06, "loss": 0.3012, "step": 18884 }, { "epoch": 1.305565157276184, "grad_norm": 0.67794559968127, "learning_rate": 2.8426207964652436e-06, "loss": 0.2307, "step": 18885 }, { "epoch": 1.305634289664708, "grad_norm": 0.6434527457105239, "learning_rate": 2.842115837482651e-06, "loss": 0.2565, "step": 18886 }, { "epoch": 1.305703422053232, "grad_norm": 0.6521701624934804, "learning_rate": 2.8416109055452843e-06, "loss": 0.3205, "step": 18887 }, { "epoch": 1.305772554441756, "grad_norm": 0.6731248908669359, "learning_rate": 2.8411060006594765e-06, "loss": 0.2666, "step": 18888 }, { "epoch": 1.30584168683028, "grad_norm": 0.64065239692049, "learning_rate": 2.8406011228315496e-06, "loss": 0.2157, "step": 18889 }, { "epoch": 1.305910819218804, "grad_norm": 0.6832602191209367, "learning_rate": 2.840096272067836e-06, "loss": 0.2737, "step": 18890 }, { "epoch": 1.305979951607328, "grad_norm": 0.6648481695432303, "learning_rate": 2.839591448374661e-06, "loss": 0.3187, "step": 18891 }, { "epoch": 1.306049083995852, "grad_norm": 0.754378176724249, "learning_rate": 2.8390866517583503e-06, "loss": 0.2268, "step": 18892 }, { "epoch": 1.306118216384376, "grad_norm": 0.6135629597971717, "learning_rate": 2.8385818822252315e-06, "loss": 0.1994, "step": 18893 }, { "epoch": 1.3061873487729, "grad_norm": 0.715368643801441, "learning_rate": 2.838077139781632e-06, "loss": 0.3163, "step": 18894 }, { "epoch": 1.3062564811614241, "grad_norm": 1.0706720908009033, "learning_rate": 2.8375724244338787e-06, "loss": 0.2543, "step": 18895 }, { "epoch": 1.3063256135499481, "grad_norm": 0.6327362664095241, "learning_rate": 2.8370677361882948e-06, "loss": 0.2542, "step": 18896 }, { "epoch": 1.3063947459384722, "grad_norm": 0.6854829049042679, "learning_rate": 2.8365630750512087e-06, "loss": 0.2793, "step": 18897 }, { "epoch": 1.3064638783269962, "grad_norm": 0.7314231181035341, "learning_rate": 2.8360584410289416e-06, "loss": 0.3146, "step": 18898 }, { "epoch": 1.3065330107155202, "grad_norm": 0.7074623107252945, "learning_rate": 2.83555383412782e-06, "loss": 0.2658, "step": 18899 }, { "epoch": 1.3066021431040442, "grad_norm": 0.5853253642586606, "learning_rate": 2.8350492543541687e-06, "loss": 0.2371, "step": 18900 }, { "epoch": 1.3066712754925682, "grad_norm": 0.7227538024487893, "learning_rate": 2.8345447017143134e-06, "loss": 0.3059, "step": 18901 }, { "epoch": 1.3067404078810922, "grad_norm": 0.6857403377704925, "learning_rate": 2.834040176214574e-06, "loss": 0.3512, "step": 18902 }, { "epoch": 1.3068095402696163, "grad_norm": 0.72189766540655, "learning_rate": 2.8335356778612777e-06, "loss": 0.2693, "step": 18903 }, { "epoch": 1.3068786726581403, "grad_norm": 0.7909933234145032, "learning_rate": 2.8330312066607434e-06, "loss": 0.2621, "step": 18904 }, { "epoch": 1.3069478050466643, "grad_norm": 0.7014043147705772, "learning_rate": 2.8325267626192974e-06, "loss": 0.3748, "step": 18905 }, { "epoch": 1.3070169374351883, "grad_norm": 0.6713246187075511, "learning_rate": 2.832022345743259e-06, "loss": 0.2941, "step": 18906 }, { "epoch": 1.3070860698237123, "grad_norm": 0.6150672359084722, "learning_rate": 2.831517956038951e-06, "loss": 0.1937, "step": 18907 }, { "epoch": 1.3071552022122364, "grad_norm": 0.6820091531295632, "learning_rate": 2.831013593512696e-06, "loss": 0.2639, "step": 18908 }, { "epoch": 1.3072243346007604, "grad_norm": 0.6763338617191506, "learning_rate": 2.8305092581708166e-06, "loss": 0.3053, "step": 18909 }, { "epoch": 1.3072934669892846, "grad_norm": 0.6523023902270944, "learning_rate": 2.8300049500196304e-06, "loss": 0.2235, "step": 18910 }, { "epoch": 1.3073625993778086, "grad_norm": 0.8633276358913342, "learning_rate": 2.829500669065461e-06, "loss": 0.2859, "step": 18911 }, { "epoch": 1.3074317317663326, "grad_norm": 0.677096921513683, "learning_rate": 2.8289964153146255e-06, "loss": 0.4026, "step": 18912 }, { "epoch": 1.3075008641548567, "grad_norm": 0.7816623691245387, "learning_rate": 2.8284921887734463e-06, "loss": 0.2621, "step": 18913 }, { "epoch": 1.3075699965433807, "grad_norm": 0.643780833653233, "learning_rate": 2.8279879894482407e-06, "loss": 0.2131, "step": 18914 }, { "epoch": 1.3076391289319047, "grad_norm": 0.7494153976923332, "learning_rate": 2.8274838173453333e-06, "loss": 0.2477, "step": 18915 }, { "epoch": 1.3077082613204287, "grad_norm": 0.7006039625284471, "learning_rate": 2.8269796724710365e-06, "loss": 0.3522, "step": 18916 }, { "epoch": 1.3077773937089527, "grad_norm": 0.6978516432468872, "learning_rate": 2.8264755548316737e-06, "loss": 0.2718, "step": 18917 }, { "epoch": 1.3078465260974768, "grad_norm": 0.7076129452372392, "learning_rate": 2.8259714644335584e-06, "loss": 0.2222, "step": 18918 }, { "epoch": 1.3079156584860008, "grad_norm": 1.3010038088088913, "learning_rate": 2.8254674012830137e-06, "loss": 0.3092, "step": 18919 }, { "epoch": 1.3079847908745248, "grad_norm": 0.6588437915424866, "learning_rate": 2.8249633653863517e-06, "loss": 0.2921, "step": 18920 }, { "epoch": 1.3080539232630488, "grad_norm": 0.6190386481933882, "learning_rate": 2.8244593567498933e-06, "loss": 0.2235, "step": 18921 }, { "epoch": 1.3081230556515728, "grad_norm": 0.6570925320292696, "learning_rate": 2.8239553753799544e-06, "loss": 0.2607, "step": 18922 }, { "epoch": 1.3081921880400968, "grad_norm": 0.6407928043528853, "learning_rate": 2.8234514212828535e-06, "loss": 0.3366, "step": 18923 }, { "epoch": 1.3082613204286209, "grad_norm": 0.6264107147276675, "learning_rate": 2.8229474944649023e-06, "loss": 0.2564, "step": 18924 }, { "epoch": 1.3083304528171449, "grad_norm": 0.5746752084137542, "learning_rate": 2.8224435949324213e-06, "loss": 0.1762, "step": 18925 }, { "epoch": 1.308399585205669, "grad_norm": 0.655417015121397, "learning_rate": 2.821939722691722e-06, "loss": 0.2727, "step": 18926 }, { "epoch": 1.308468717594193, "grad_norm": 0.6916932917011741, "learning_rate": 2.8214358777491234e-06, "loss": 0.3423, "step": 18927 }, { "epoch": 1.308537849982717, "grad_norm": 0.6563242066964942, "learning_rate": 2.8209320601109357e-06, "loss": 0.2281, "step": 18928 }, { "epoch": 1.308606982371241, "grad_norm": 0.644818527773515, "learning_rate": 2.8204282697834763e-06, "loss": 0.228, "step": 18929 }, { "epoch": 1.308676114759765, "grad_norm": 0.7048191664098002, "learning_rate": 2.819924506773059e-06, "loss": 0.3906, "step": 18930 }, { "epoch": 1.308745247148289, "grad_norm": 0.5942240555735702, "learning_rate": 2.819420771086e-06, "loss": 0.2507, "step": 18931 }, { "epoch": 1.308814379536813, "grad_norm": 0.7206757938543839, "learning_rate": 2.8189170627286072e-06, "loss": 0.2193, "step": 18932 }, { "epoch": 1.308883511925337, "grad_norm": 0.6281074714067281, "learning_rate": 2.818413381707199e-06, "loss": 0.2191, "step": 18933 }, { "epoch": 1.308952644313861, "grad_norm": 0.6585991146794008, "learning_rate": 2.817909728028084e-06, "loss": 0.3655, "step": 18934 }, { "epoch": 1.309021776702385, "grad_norm": 0.630485786807226, "learning_rate": 2.817406101697576e-06, "loss": 0.2492, "step": 18935 }, { "epoch": 1.309090909090909, "grad_norm": 0.6528072661399891, "learning_rate": 2.816902502721989e-06, "loss": 0.1952, "step": 18936 }, { "epoch": 1.309160041479433, "grad_norm": 0.8309521330694366, "learning_rate": 2.816398931107633e-06, "loss": 0.326, "step": 18937 }, { "epoch": 1.3092291738679571, "grad_norm": 0.7093769706007609, "learning_rate": 2.8158953868608197e-06, "loss": 0.28, "step": 18938 }, { "epoch": 1.3092983062564811, "grad_norm": 0.7661055668023864, "learning_rate": 2.815391869987861e-06, "loss": 0.249, "step": 18939 }, { "epoch": 1.3093674386450052, "grad_norm": 0.6926012901437966, "learning_rate": 2.814888380495064e-06, "loss": 0.282, "step": 18940 }, { "epoch": 1.3094365710335292, "grad_norm": 0.6610999374108008, "learning_rate": 2.814384918388745e-06, "loss": 0.3309, "step": 18941 }, { "epoch": 1.3095057034220532, "grad_norm": 0.678905492362128, "learning_rate": 2.813881483675208e-06, "loss": 0.2798, "step": 18942 }, { "epoch": 1.3095748358105772, "grad_norm": 0.7343182598790038, "learning_rate": 2.813378076360765e-06, "loss": 0.2909, "step": 18943 }, { "epoch": 1.3096439681991012, "grad_norm": 0.6614621160973682, "learning_rate": 2.8128746964517264e-06, "loss": 0.3463, "step": 18944 }, { "epoch": 1.3097131005876252, "grad_norm": 0.7084311849724139, "learning_rate": 2.812371343954401e-06, "loss": 0.3079, "step": 18945 }, { "epoch": 1.3097822329761493, "grad_norm": 0.6708468229359309, "learning_rate": 2.811868018875096e-06, "loss": 0.2238, "step": 18946 }, { "epoch": 1.3098513653646733, "grad_norm": 0.6935746677033389, "learning_rate": 2.8113647212201224e-06, "loss": 0.3213, "step": 18947 }, { "epoch": 1.3099204977531973, "grad_norm": 0.6826630343092569, "learning_rate": 2.810861450995783e-06, "loss": 0.3544, "step": 18948 }, { "epoch": 1.3099896301417213, "grad_norm": 0.7258876809029338, "learning_rate": 2.8103582082083903e-06, "loss": 0.2693, "step": 18949 }, { "epoch": 1.3100587625302453, "grad_norm": 0.7457941708742142, "learning_rate": 2.8098549928642495e-06, "loss": 0.224, "step": 18950 }, { "epoch": 1.3101278949187694, "grad_norm": 0.6715802744247161, "learning_rate": 2.8093518049696687e-06, "loss": 0.2827, "step": 18951 }, { "epoch": 1.3101970273072934, "grad_norm": 0.678134490125035, "learning_rate": 2.808848644530953e-06, "loss": 0.3107, "step": 18952 }, { "epoch": 1.3102661596958174, "grad_norm": 0.8568611189162867, "learning_rate": 2.80834551155441e-06, "loss": 0.2492, "step": 18953 }, { "epoch": 1.3103352920843414, "grad_norm": 0.9634138447652806, "learning_rate": 2.8078424060463436e-06, "loss": 0.2583, "step": 18954 }, { "epoch": 1.3104044244728654, "grad_norm": 0.684817636480034, "learning_rate": 2.8073393280130627e-06, "loss": 0.3696, "step": 18955 }, { "epoch": 1.3104735568613894, "grad_norm": 0.7352234982190613, "learning_rate": 2.8068362774608682e-06, "loss": 0.2486, "step": 18956 }, { "epoch": 1.3105426892499135, "grad_norm": 0.6548404917886887, "learning_rate": 2.8063332543960674e-06, "loss": 0.2466, "step": 18957 }, { "epoch": 1.3106118216384375, "grad_norm": 0.7397921235995788, "learning_rate": 2.805830258824966e-06, "loss": 0.2851, "step": 18958 }, { "epoch": 1.3106809540269615, "grad_norm": 0.6422295333988088, "learning_rate": 2.805327290753864e-06, "loss": 0.3295, "step": 18959 }, { "epoch": 1.3107500864154857, "grad_norm": 0.6446279005525027, "learning_rate": 2.8048243501890716e-06, "loss": 0.2397, "step": 18960 }, { "epoch": 1.3108192188040098, "grad_norm": 0.6284149996476165, "learning_rate": 2.8043214371368853e-06, "loss": 0.2421, "step": 18961 }, { "epoch": 1.3108883511925338, "grad_norm": 0.7386176859107548, "learning_rate": 2.8038185516036133e-06, "loss": 0.3294, "step": 18962 }, { "epoch": 1.3109574835810578, "grad_norm": 0.739839529524533, "learning_rate": 2.803315693595555e-06, "loss": 0.3227, "step": 18963 }, { "epoch": 1.3110266159695818, "grad_norm": 0.6701845834852013, "learning_rate": 2.8028128631190143e-06, "loss": 0.2091, "step": 18964 }, { "epoch": 1.3110957483581058, "grad_norm": 0.6480110917706994, "learning_rate": 2.8023100601802926e-06, "loss": 0.1981, "step": 18965 }, { "epoch": 1.3111648807466298, "grad_norm": 0.7720086398683097, "learning_rate": 2.801807284785695e-06, "loss": 0.3635, "step": 18966 }, { "epoch": 1.3112340131351539, "grad_norm": 0.6972939665959293, "learning_rate": 2.8013045369415183e-06, "loss": 0.2547, "step": 18967 }, { "epoch": 1.3113031455236779, "grad_norm": 0.6819982616001297, "learning_rate": 2.800801816654067e-06, "loss": 0.2456, "step": 18968 }, { "epoch": 1.311372277912202, "grad_norm": 0.7587060128794775, "learning_rate": 2.8002991239296386e-06, "loss": 0.3457, "step": 18969 }, { "epoch": 1.311441410300726, "grad_norm": 0.6338188586808298, "learning_rate": 2.7997964587745353e-06, "loss": 0.244, "step": 18970 }, { "epoch": 1.31151054268925, "grad_norm": 0.6522868709558918, "learning_rate": 2.7992938211950572e-06, "loss": 0.225, "step": 18971 }, { "epoch": 1.311579675077774, "grad_norm": 1.0361261915694642, "learning_rate": 2.7987912111975046e-06, "loss": 0.2541, "step": 18972 }, { "epoch": 1.311648807466298, "grad_norm": 0.6977696030601193, "learning_rate": 2.798288628788175e-06, "loss": 0.3211, "step": 18973 }, { "epoch": 1.311717939854822, "grad_norm": 0.7073715269325987, "learning_rate": 2.79778607397337e-06, "loss": 0.2692, "step": 18974 }, { "epoch": 1.311787072243346, "grad_norm": 0.6177181993588319, "learning_rate": 2.797283546759384e-06, "loss": 0.2557, "step": 18975 }, { "epoch": 1.31185620463187, "grad_norm": 0.6302977679386079, "learning_rate": 2.79678104715252e-06, "loss": 0.2646, "step": 18976 }, { "epoch": 1.311925337020394, "grad_norm": 0.6611210005661674, "learning_rate": 2.7962785751590725e-06, "loss": 0.3874, "step": 18977 }, { "epoch": 1.311994469408918, "grad_norm": 0.6323474484185123, "learning_rate": 2.7957761307853397e-06, "loss": 0.2179, "step": 18978 }, { "epoch": 1.312063601797442, "grad_norm": 0.6663544679090365, "learning_rate": 2.7952737140376196e-06, "loss": 0.2401, "step": 18979 }, { "epoch": 1.312132734185966, "grad_norm": 0.667709998725611, "learning_rate": 2.794771324922211e-06, "loss": 0.3458, "step": 18980 }, { "epoch": 1.3122018665744901, "grad_norm": 0.6596975364047463, "learning_rate": 2.7942689634454066e-06, "loss": 0.3516, "step": 18981 }, { "epoch": 1.3122709989630141, "grad_norm": 0.6405310248141861, "learning_rate": 2.793766629613506e-06, "loss": 0.2181, "step": 18982 }, { "epoch": 1.3123401313515382, "grad_norm": 0.7162209658519303, "learning_rate": 2.7932643234328024e-06, "loss": 0.2852, "step": 18983 }, { "epoch": 1.3124092637400622, "grad_norm": 0.6680773654658274, "learning_rate": 2.7927620449095915e-06, "loss": 0.3275, "step": 18984 }, { "epoch": 1.3124783961285862, "grad_norm": 0.632231551437719, "learning_rate": 2.7922597940501693e-06, "loss": 0.219, "step": 18985 }, { "epoch": 1.3125475285171102, "grad_norm": 0.6558968151975475, "learning_rate": 2.7917575708608334e-06, "loss": 0.2589, "step": 18986 }, { "epoch": 1.3126166609056342, "grad_norm": 0.7571126483788402, "learning_rate": 2.7912553753478733e-06, "loss": 0.3491, "step": 18987 }, { "epoch": 1.3126857932941582, "grad_norm": 0.7418990290553553, "learning_rate": 2.7907532075175873e-06, "loss": 0.2625, "step": 18988 }, { "epoch": 1.3127549256826823, "grad_norm": 0.586150343031501, "learning_rate": 2.790251067376265e-06, "loss": 0.1955, "step": 18989 }, { "epoch": 1.3128240580712063, "grad_norm": 0.7005549375413389, "learning_rate": 2.789748954930205e-06, "loss": 0.2805, "step": 18990 }, { "epoch": 1.3128931904597305, "grad_norm": 0.6625326625375471, "learning_rate": 2.7892468701856945e-06, "loss": 0.3481, "step": 18991 }, { "epoch": 1.3129623228482545, "grad_norm": 0.644932587038677, "learning_rate": 2.78874481314903e-06, "loss": 0.2708, "step": 18992 }, { "epoch": 1.3130314552367786, "grad_norm": 0.632213466784883, "learning_rate": 2.7882427838265027e-06, "loss": 0.2593, "step": 18993 }, { "epoch": 1.3131005876253026, "grad_norm": 0.8816122736555444, "learning_rate": 2.787740782224407e-06, "loss": 0.3441, "step": 18994 }, { "epoch": 1.3131697200138266, "grad_norm": 0.7341264013592234, "learning_rate": 2.7872388083490304e-06, "loss": 0.3555, "step": 18995 }, { "epoch": 1.3132388524023506, "grad_norm": 0.6144974026427369, "learning_rate": 2.786736862206669e-06, "loss": 0.2432, "step": 18996 }, { "epoch": 1.3133079847908746, "grad_norm": 0.6794192410394492, "learning_rate": 2.7862349438036084e-06, "loss": 0.2819, "step": 18997 }, { "epoch": 1.3133771171793986, "grad_norm": 0.6535117954930991, "learning_rate": 2.785733053146145e-06, "loss": 0.3158, "step": 18998 }, { "epoch": 1.3134462495679227, "grad_norm": 0.6554554082725831, "learning_rate": 2.7852311902405644e-06, "loss": 0.3133, "step": 18999 }, { "epoch": 1.3135153819564467, "grad_norm": 0.6790979060878083, "learning_rate": 2.7847293550931576e-06, "loss": 0.2272, "step": 19000 }, { "epoch": 1.3135845143449707, "grad_norm": 0.7073319723949535, "learning_rate": 2.7842275477102156e-06, "loss": 0.3505, "step": 19001 }, { "epoch": 1.3136536467334947, "grad_norm": 0.6910346987137703, "learning_rate": 2.783725768098029e-06, "loss": 0.3573, "step": 19002 }, { "epoch": 1.3137227791220187, "grad_norm": 0.6255484641305459, "learning_rate": 2.7832240162628828e-06, "loss": 0.2197, "step": 19003 }, { "epoch": 1.3137919115105428, "grad_norm": 0.7066746181226663, "learning_rate": 2.7827222922110696e-06, "loss": 0.2722, "step": 19004 }, { "epoch": 1.3138610438990668, "grad_norm": 0.8384212145999241, "learning_rate": 2.7822205959488735e-06, "loss": 0.381, "step": 19005 }, { "epoch": 1.3139301762875908, "grad_norm": 0.7435182907711781, "learning_rate": 2.781718927482584e-06, "loss": 0.2932, "step": 19006 }, { "epoch": 1.3139993086761148, "grad_norm": 0.6896124277196298, "learning_rate": 2.7812172868184896e-06, "loss": 0.2242, "step": 19007 }, { "epoch": 1.3140684410646388, "grad_norm": 0.7083776424067502, "learning_rate": 2.780715673962878e-06, "loss": 0.2805, "step": 19008 }, { "epoch": 1.3141375734531628, "grad_norm": 0.682592828154565, "learning_rate": 2.7802140889220343e-06, "loss": 0.3327, "step": 19009 }, { "epoch": 1.3142067058416869, "grad_norm": 0.6725478912986449, "learning_rate": 2.7797125317022467e-06, "loss": 0.3239, "step": 19010 }, { "epoch": 1.3142758382302109, "grad_norm": 0.6732893085873012, "learning_rate": 2.7792110023097984e-06, "loss": 0.2398, "step": 19011 }, { "epoch": 1.314344970618735, "grad_norm": 0.7384712073145947, "learning_rate": 2.778709500750979e-06, "loss": 0.3066, "step": 19012 }, { "epoch": 1.314414103007259, "grad_norm": 0.6471379738653193, "learning_rate": 2.77820802703207e-06, "loss": 0.2686, "step": 19013 }, { "epoch": 1.314483235395783, "grad_norm": 0.604838831594377, "learning_rate": 2.7777065811593584e-06, "loss": 0.2213, "step": 19014 }, { "epoch": 1.314552367784307, "grad_norm": 0.7151927140991575, "learning_rate": 2.777205163139129e-06, "loss": 0.2549, "step": 19015 }, { "epoch": 1.314621500172831, "grad_norm": 0.6219878399841663, "learning_rate": 2.776703772977668e-06, "loss": 0.3686, "step": 19016 }, { "epoch": 1.314690632561355, "grad_norm": 0.8557280458349023, "learning_rate": 2.7762024106812557e-06, "loss": 0.3232, "step": 19017 }, { "epoch": 1.314759764949879, "grad_norm": 0.79594061099379, "learning_rate": 2.7757010762561797e-06, "loss": 0.2157, "step": 19018 }, { "epoch": 1.314828897338403, "grad_norm": 0.6918706864737729, "learning_rate": 2.7751997697087184e-06, "loss": 0.3026, "step": 19019 }, { "epoch": 1.314898029726927, "grad_norm": 0.7344670179937125, "learning_rate": 2.774698491045159e-06, "loss": 0.3026, "step": 19020 }, { "epoch": 1.314967162115451, "grad_norm": 0.6067774568470453, "learning_rate": 2.774197240271782e-06, "loss": 0.2084, "step": 19021 }, { "epoch": 1.315036294503975, "grad_norm": 0.6425697113095026, "learning_rate": 2.7736960173948724e-06, "loss": 0.2376, "step": 19022 }, { "epoch": 1.315105426892499, "grad_norm": 0.6586977950780456, "learning_rate": 2.7731948224207107e-06, "loss": 0.3037, "step": 19023 }, { "epoch": 1.3151745592810231, "grad_norm": 0.7341631961995997, "learning_rate": 2.772693655355576e-06, "loss": 0.2784, "step": 19024 }, { "epoch": 1.3152436916695471, "grad_norm": 0.821356093216859, "learning_rate": 2.7721925162057526e-06, "loss": 0.2318, "step": 19025 }, { "epoch": 1.3153128240580712, "grad_norm": 0.8006653194269565, "learning_rate": 2.7716914049775195e-06, "loss": 0.2644, "step": 19026 }, { "epoch": 1.3153819564465952, "grad_norm": 0.7023877520807292, "learning_rate": 2.7711903216771573e-06, "loss": 0.3186, "step": 19027 }, { "epoch": 1.3154510888351192, "grad_norm": 0.6181570554087304, "learning_rate": 2.7706892663109473e-06, "loss": 0.243, "step": 19028 }, { "epoch": 1.3155202212236432, "grad_norm": 0.6222128448613143, "learning_rate": 2.7701882388851707e-06, "loss": 0.2454, "step": 19029 }, { "epoch": 1.3155893536121672, "grad_norm": 0.9188873076582557, "learning_rate": 2.7696872394061035e-06, "loss": 0.3511, "step": 19030 }, { "epoch": 1.3156584860006912, "grad_norm": 0.6662064106428972, "learning_rate": 2.7691862678800286e-06, "loss": 0.2234, "step": 19031 }, { "epoch": 1.3157276183892153, "grad_norm": 0.6766047308135292, "learning_rate": 2.7686853243132205e-06, "loss": 0.2396, "step": 19032 }, { "epoch": 1.3157967507777393, "grad_norm": 0.631128546281243, "learning_rate": 2.768184408711962e-06, "loss": 0.2605, "step": 19033 }, { "epoch": 1.3158658831662633, "grad_norm": 0.692137410453589, "learning_rate": 2.767683521082527e-06, "loss": 0.3241, "step": 19034 }, { "epoch": 1.3159350155547873, "grad_norm": 0.6927054258193959, "learning_rate": 2.7671826614311953e-06, "loss": 0.2451, "step": 19035 }, { "epoch": 1.3160041479433113, "grad_norm": 0.7159369755753704, "learning_rate": 2.7666818297642445e-06, "loss": 0.237, "step": 19036 }, { "epoch": 1.3160732803318353, "grad_norm": 0.7127058659568483, "learning_rate": 2.766181026087953e-06, "loss": 0.3898, "step": 19037 }, { "epoch": 1.3161424127203594, "grad_norm": 0.6838016902475558, "learning_rate": 2.7656802504085945e-06, "loss": 0.326, "step": 19038 }, { "epoch": 1.3162115451088834, "grad_norm": 0.6744128991970463, "learning_rate": 2.7651795027324486e-06, "loss": 0.2474, "step": 19039 }, { "epoch": 1.3162806774974074, "grad_norm": 0.7009065657378005, "learning_rate": 2.764678783065787e-06, "loss": 0.2412, "step": 19040 }, { "epoch": 1.3163498098859314, "grad_norm": 0.7308494709655099, "learning_rate": 2.7641780914148886e-06, "loss": 0.3272, "step": 19041 }, { "epoch": 1.3164189422744557, "grad_norm": 0.7086951707981163, "learning_rate": 2.763677427786027e-06, "loss": 0.2074, "step": 19042 }, { "epoch": 1.3164880746629797, "grad_norm": 0.6757220278616892, "learning_rate": 2.76317679218548e-06, "loss": 0.2045, "step": 19043 }, { "epoch": 1.3165572070515037, "grad_norm": 0.6985289754515284, "learning_rate": 2.7626761846195185e-06, "loss": 0.2443, "step": 19044 }, { "epoch": 1.3166263394400277, "grad_norm": 0.6722723524504173, "learning_rate": 2.7621756050944204e-06, "loss": 0.2993, "step": 19045 }, { "epoch": 1.3166954718285517, "grad_norm": 0.6399131187108211, "learning_rate": 2.761675053616455e-06, "loss": 0.2527, "step": 19046 }, { "epoch": 1.3167646042170758, "grad_norm": 0.6495570768799545, "learning_rate": 2.7611745301919e-06, "loss": 0.2952, "step": 19047 }, { "epoch": 1.3168337366055998, "grad_norm": 0.7288683440880573, "learning_rate": 2.760674034827026e-06, "loss": 0.3438, "step": 19048 }, { "epoch": 1.3169028689941238, "grad_norm": 0.690821353700562, "learning_rate": 2.760173567528106e-06, "loss": 0.2762, "step": 19049 }, { "epoch": 1.3169720013826478, "grad_norm": 0.6857807396209293, "learning_rate": 2.7596731283014132e-06, "loss": 0.2153, "step": 19050 }, { "epoch": 1.3170411337711718, "grad_norm": 0.6719809426679468, "learning_rate": 2.759172717153222e-06, "loss": 0.3149, "step": 19051 }, { "epoch": 1.3171102661596958, "grad_norm": 0.714663184088348, "learning_rate": 2.7586723340898005e-06, "loss": 0.3225, "step": 19052 }, { "epoch": 1.3171793985482199, "grad_norm": 0.6736166205224787, "learning_rate": 2.7581719791174223e-06, "loss": 0.2446, "step": 19053 }, { "epoch": 1.3172485309367439, "grad_norm": 0.6460056292479059, "learning_rate": 2.7576716522423564e-06, "loss": 0.1955, "step": 19054 }, { "epoch": 1.317317663325268, "grad_norm": 0.7006513287184329, "learning_rate": 2.7571713534708743e-06, "loss": 0.3152, "step": 19055 }, { "epoch": 1.317386795713792, "grad_norm": 0.6418593081981194, "learning_rate": 2.756671082809247e-06, "loss": 0.2377, "step": 19056 }, { "epoch": 1.317455928102316, "grad_norm": 0.7481256123722073, "learning_rate": 2.7561708402637454e-06, "loss": 0.3224, "step": 19057 }, { "epoch": 1.31752506049084, "grad_norm": 0.6595990539523023, "learning_rate": 2.755670625840637e-06, "loss": 0.2981, "step": 19058 }, { "epoch": 1.317594192879364, "grad_norm": 0.7066837338347383, "learning_rate": 2.7551704395461943e-06, "loss": 0.4452, "step": 19059 }, { "epoch": 1.317663325267888, "grad_norm": 0.8331793455296557, "learning_rate": 2.754670281386681e-06, "loss": 0.2856, "step": 19060 }, { "epoch": 1.317732457656412, "grad_norm": 0.9761879259008955, "learning_rate": 2.7541701513683717e-06, "loss": 0.2386, "step": 19061 }, { "epoch": 1.317801590044936, "grad_norm": 0.7501945008735278, "learning_rate": 2.7536700494975295e-06, "loss": 0.3312, "step": 19062 }, { "epoch": 1.31787072243346, "grad_norm": 0.6207423548556594, "learning_rate": 2.753169975780424e-06, "loss": 0.2802, "step": 19063 }, { "epoch": 1.317939854821984, "grad_norm": 0.7093172171687335, "learning_rate": 2.752669930223324e-06, "loss": 0.2328, "step": 19064 }, { "epoch": 1.318008987210508, "grad_norm": 0.7280916268338467, "learning_rate": 2.7521699128324977e-06, "loss": 0.2505, "step": 19065 }, { "epoch": 1.318078119599032, "grad_norm": 0.707371123731682, "learning_rate": 2.7516699236142074e-06, "loss": 0.3869, "step": 19066 }, { "epoch": 1.3181472519875561, "grad_norm": 0.6141644535886827, "learning_rate": 2.7511699625747247e-06, "loss": 0.2129, "step": 19067 }, { "epoch": 1.3182163843760801, "grad_norm": 0.9018853389364793, "learning_rate": 2.7506700297203115e-06, "loss": 0.2154, "step": 19068 }, { "epoch": 1.3182855167646041, "grad_norm": 0.6758083635308264, "learning_rate": 2.7501701250572364e-06, "loss": 0.3092, "step": 19069 }, { "epoch": 1.3183546491531282, "grad_norm": 0.6659075797563216, "learning_rate": 2.7496702485917632e-06, "loss": 0.3306, "step": 19070 }, { "epoch": 1.3184237815416522, "grad_norm": 0.636821206816401, "learning_rate": 2.7491704003301566e-06, "loss": 0.233, "step": 19071 }, { "epoch": 1.3184929139301764, "grad_norm": 0.691609181744419, "learning_rate": 2.7486705802786828e-06, "loss": 0.2318, "step": 19072 }, { "epoch": 1.3185620463187004, "grad_norm": 0.6302196729770485, "learning_rate": 2.748170788443607e-06, "loss": 0.3474, "step": 19073 }, { "epoch": 1.3186311787072245, "grad_norm": 0.7038312884721802, "learning_rate": 2.74767102483119e-06, "loss": 0.2163, "step": 19074 }, { "epoch": 1.3187003110957485, "grad_norm": 0.6345923690609051, "learning_rate": 2.7471712894476998e-06, "loss": 0.202, "step": 19075 }, { "epoch": 1.3187694434842725, "grad_norm": 0.6078414536718649, "learning_rate": 2.746671582299395e-06, "loss": 0.2894, "step": 19076 }, { "epoch": 1.3188385758727965, "grad_norm": 0.7106538064880014, "learning_rate": 2.7461719033925404e-06, "loss": 0.3499, "step": 19077 }, { "epoch": 1.3189077082613205, "grad_norm": 0.6499298180442881, "learning_rate": 2.7456722527333984e-06, "loss": 0.2438, "step": 19078 }, { "epoch": 1.3189768406498446, "grad_norm": 0.7133152326225267, "learning_rate": 2.7451726303282346e-06, "loss": 0.3016, "step": 19079 }, { "epoch": 1.3190459730383686, "grad_norm": 0.6589199098978974, "learning_rate": 2.7446730361833053e-06, "loss": 0.3245, "step": 19080 }, { "epoch": 1.3191151054268926, "grad_norm": 0.6554613801765566, "learning_rate": 2.744173470304877e-06, "loss": 0.2878, "step": 19081 }, { "epoch": 1.3191842378154166, "grad_norm": 0.6557851207953596, "learning_rate": 2.743673932699206e-06, "loss": 0.2381, "step": 19082 }, { "epoch": 1.3192533702039406, "grad_norm": 0.6722404209828926, "learning_rate": 2.743174423372558e-06, "loss": 0.2945, "step": 19083 }, { "epoch": 1.3193225025924646, "grad_norm": 0.722640337454711, "learning_rate": 2.7426749423311894e-06, "loss": 0.3169, "step": 19084 }, { "epoch": 1.3193916349809887, "grad_norm": 0.7361597504981904, "learning_rate": 2.742175489581362e-06, "loss": 0.254, "step": 19085 }, { "epoch": 1.3194607673695127, "grad_norm": 0.7028205826030661, "learning_rate": 2.741676065129335e-06, "loss": 0.2837, "step": 19086 }, { "epoch": 1.3195298997580367, "grad_norm": 0.6813968745476231, "learning_rate": 2.7411766689813713e-06, "loss": 0.3363, "step": 19087 }, { "epoch": 1.3195990321465607, "grad_norm": 0.7120231148404907, "learning_rate": 2.7406773011437237e-06, "loss": 0.2992, "step": 19088 }, { "epoch": 1.3196681645350847, "grad_norm": 0.6617739575884012, "learning_rate": 2.7401779616226578e-06, "loss": 0.2255, "step": 19089 }, { "epoch": 1.3197372969236087, "grad_norm": 0.69192184023388, "learning_rate": 2.7396786504244267e-06, "loss": 0.2961, "step": 19090 }, { "epoch": 1.3198064293121328, "grad_norm": 0.6884424703785155, "learning_rate": 2.739179367555287e-06, "loss": 0.3523, "step": 19091 }, { "epoch": 1.3198755617006568, "grad_norm": 0.6838973266387492, "learning_rate": 2.7386801130215035e-06, "loss": 0.2826, "step": 19092 }, { "epoch": 1.3199446940891808, "grad_norm": 0.6045569686069584, "learning_rate": 2.7381808868293264e-06, "loss": 0.2205, "step": 19093 }, { "epoch": 1.3200138264777048, "grad_norm": 0.6395212072514821, "learning_rate": 2.7376816889850177e-06, "loss": 0.3687, "step": 19094 }, { "epoch": 1.3200829588662288, "grad_norm": 0.7215974169968609, "learning_rate": 2.7371825194948303e-06, "loss": 0.2904, "step": 19095 }, { "epoch": 1.3201520912547529, "grad_norm": 0.6551763551646245, "learning_rate": 2.736683378365024e-06, "loss": 0.2566, "step": 19096 }, { "epoch": 1.3202212236432769, "grad_norm": 0.6074243766865406, "learning_rate": 2.7361842656018497e-06, "loss": 0.2598, "step": 19097 }, { "epoch": 1.320290356031801, "grad_norm": 0.6869213217706577, "learning_rate": 2.735685181211566e-06, "loss": 0.3621, "step": 19098 }, { "epoch": 1.320359488420325, "grad_norm": 0.7002523861986857, "learning_rate": 2.7351861252004275e-06, "loss": 0.2073, "step": 19099 }, { "epoch": 1.320428620808849, "grad_norm": 0.6794292563066264, "learning_rate": 2.734687097574691e-06, "loss": 0.2352, "step": 19100 }, { "epoch": 1.320497753197373, "grad_norm": 0.6644971685392494, "learning_rate": 2.734188098340608e-06, "loss": 0.2602, "step": 19101 }, { "epoch": 1.320566885585897, "grad_norm": 0.7512717096300076, "learning_rate": 2.7336891275044346e-06, "loss": 0.2866, "step": 19102 }, { "epoch": 1.320636017974421, "grad_norm": 0.6568444134684818, "learning_rate": 2.733190185072422e-06, "loss": 0.2306, "step": 19103 }, { "epoch": 1.320705150362945, "grad_norm": 0.6059704058939248, "learning_rate": 2.7326912710508274e-06, "loss": 0.2176, "step": 19104 }, { "epoch": 1.320774282751469, "grad_norm": 0.6783474134231039, "learning_rate": 2.7321923854458994e-06, "loss": 0.3377, "step": 19105 }, { "epoch": 1.320843415139993, "grad_norm": 0.606525821585611, "learning_rate": 2.7316935282638925e-06, "loss": 0.2572, "step": 19106 }, { "epoch": 1.320912547528517, "grad_norm": 0.643746487732473, "learning_rate": 2.7311946995110596e-06, "loss": 0.2741, "step": 19107 }, { "epoch": 1.320981679917041, "grad_norm": 0.7206733537582463, "learning_rate": 2.7306958991936538e-06, "loss": 0.2871, "step": 19108 }, { "epoch": 1.321050812305565, "grad_norm": 0.6555928614759217, "learning_rate": 2.730197127317924e-06, "loss": 0.3615, "step": 19109 }, { "epoch": 1.321119944694089, "grad_norm": 0.6588283246874457, "learning_rate": 2.7296983838901237e-06, "loss": 0.2536, "step": 19110 }, { "epoch": 1.3211890770826131, "grad_norm": 0.6028735775529942, "learning_rate": 2.729199668916501e-06, "loss": 0.2082, "step": 19111 }, { "epoch": 1.3212582094711371, "grad_norm": 0.7422123760830507, "learning_rate": 2.7287009824033077e-06, "loss": 0.3559, "step": 19112 }, { "epoch": 1.3213273418596612, "grad_norm": 1.0363971851382563, "learning_rate": 2.7282023243567944e-06, "loss": 0.2874, "step": 19113 }, { "epoch": 1.3213964742481852, "grad_norm": 0.5664669480110215, "learning_rate": 2.7277036947832135e-06, "loss": 0.1948, "step": 19114 }, { "epoch": 1.3214656066367092, "grad_norm": 1.0424546522670948, "learning_rate": 2.727205093688809e-06, "loss": 0.261, "step": 19115 }, { "epoch": 1.3215347390252332, "grad_norm": 0.6608428847326908, "learning_rate": 2.726706521079835e-06, "loss": 0.3171, "step": 19116 }, { "epoch": 1.3216038714137572, "grad_norm": 0.6929067773409692, "learning_rate": 2.7262079769625348e-06, "loss": 0.2611, "step": 19117 }, { "epoch": 1.3216730038022813, "grad_norm": 0.69575068135839, "learning_rate": 2.7257094613431635e-06, "loss": 0.265, "step": 19118 }, { "epoch": 1.3217421361908053, "grad_norm": 0.7464546337014142, "learning_rate": 2.725210974227962e-06, "loss": 0.3513, "step": 19119 }, { "epoch": 1.3218112685793293, "grad_norm": 0.7028970641233182, "learning_rate": 2.7247125156231826e-06, "loss": 0.3099, "step": 19120 }, { "epoch": 1.3218804009678533, "grad_norm": 0.65113506365959, "learning_rate": 2.7242140855350706e-06, "loss": 0.219, "step": 19121 }, { "epoch": 1.3219495333563773, "grad_norm": 1.0583634226062253, "learning_rate": 2.7237156839698757e-06, "loss": 0.2478, "step": 19122 }, { "epoch": 1.3220186657449016, "grad_norm": 0.6843425380441952, "learning_rate": 2.723217310933841e-06, "loss": 0.3202, "step": 19123 }, { "epoch": 1.3220877981334256, "grad_norm": 0.6820143654694019, "learning_rate": 2.7227189664332156e-06, "loss": 0.2708, "step": 19124 }, { "epoch": 1.3221569305219496, "grad_norm": 0.5880132507794383, "learning_rate": 2.722220650474242e-06, "loss": 0.2069, "step": 19125 }, { "epoch": 1.3222260629104736, "grad_norm": 0.6409236039965944, "learning_rate": 2.7217223630631684e-06, "loss": 0.2708, "step": 19126 }, { "epoch": 1.3222951952989976, "grad_norm": 0.6726490366086577, "learning_rate": 2.7212241042062386e-06, "loss": 0.2705, "step": 19127 }, { "epoch": 1.3223643276875217, "grad_norm": 0.5841747290230704, "learning_rate": 2.7207258739097005e-06, "loss": 0.2233, "step": 19128 }, { "epoch": 1.3224334600760457, "grad_norm": 0.6497118053086786, "learning_rate": 2.7202276721797936e-06, "loss": 0.2328, "step": 19129 }, { "epoch": 1.3225025924645697, "grad_norm": 0.7375228751928661, "learning_rate": 2.719729499022766e-06, "loss": 0.322, "step": 19130 }, { "epoch": 1.3225717248530937, "grad_norm": 0.7111573205390814, "learning_rate": 2.719231354444859e-06, "loss": 0.2487, "step": 19131 }, { "epoch": 1.3226408572416177, "grad_norm": 0.6536288141144286, "learning_rate": 2.718733238452318e-06, "loss": 0.2517, "step": 19132 }, { "epoch": 1.3227099896301417, "grad_norm": 0.6761807956111272, "learning_rate": 2.7182351510513825e-06, "loss": 0.2634, "step": 19133 }, { "epoch": 1.3227791220186658, "grad_norm": 0.6611402509100991, "learning_rate": 2.717737092248298e-06, "loss": 0.3724, "step": 19134 }, { "epoch": 1.3228482544071898, "grad_norm": 0.7113832756532528, "learning_rate": 2.7172390620493054e-06, "loss": 0.2349, "step": 19135 }, { "epoch": 1.3229173867957138, "grad_norm": 0.7290605755628928, "learning_rate": 2.7167410604606504e-06, "loss": 0.3154, "step": 19136 }, { "epoch": 1.3229865191842378, "grad_norm": 0.6893001244514697, "learning_rate": 2.7162430874885687e-06, "loss": 0.3192, "step": 19137 }, { "epoch": 1.3230556515727618, "grad_norm": 0.7628965208611317, "learning_rate": 2.715745143139307e-06, "loss": 0.2585, "step": 19138 }, { "epoch": 1.3231247839612859, "grad_norm": 0.5978627569886433, "learning_rate": 2.715247227419101e-06, "loss": 0.1888, "step": 19139 }, { "epoch": 1.3231939163498099, "grad_norm": 0.6354820888169859, "learning_rate": 2.714749340334195e-06, "loss": 0.2834, "step": 19140 }, { "epoch": 1.323263048738334, "grad_norm": 0.675592433464899, "learning_rate": 2.714251481890827e-06, "loss": 0.3461, "step": 19141 }, { "epoch": 1.323332181126858, "grad_norm": 0.7059485717756461, "learning_rate": 2.713753652095237e-06, "loss": 0.2961, "step": 19142 }, { "epoch": 1.323401313515382, "grad_norm": 0.7180722210524576, "learning_rate": 2.7132558509536656e-06, "loss": 0.2823, "step": 19143 }, { "epoch": 1.323470445903906, "grad_norm": 1.6977850846646814, "learning_rate": 2.712758078472353e-06, "loss": 0.2698, "step": 19144 }, { "epoch": 1.32353957829243, "grad_norm": 0.7201085771589301, "learning_rate": 2.7122603346575336e-06, "loss": 0.3331, "step": 19145 }, { "epoch": 1.323608710680954, "grad_norm": 0.6605995621102673, "learning_rate": 2.7117626195154505e-06, "loss": 0.2328, "step": 19146 }, { "epoch": 1.323677843069478, "grad_norm": 0.6040810113660483, "learning_rate": 2.7112649330523377e-06, "loss": 0.2411, "step": 19147 }, { "epoch": 1.323746975458002, "grad_norm": 0.8661671002007131, "learning_rate": 2.710767275274434e-06, "loss": 0.3414, "step": 19148 }, { "epoch": 1.323816107846526, "grad_norm": 0.7150724842428914, "learning_rate": 2.7102696461879773e-06, "loss": 0.2761, "step": 19149 }, { "epoch": 1.32388524023505, "grad_norm": 0.6786422228945288, "learning_rate": 2.7097720457992074e-06, "loss": 0.2345, "step": 19150 }, { "epoch": 1.323954372623574, "grad_norm": 0.6947453490293511, "learning_rate": 2.7092744741143545e-06, "loss": 0.2906, "step": 19151 }, { "epoch": 1.324023505012098, "grad_norm": 0.6979062700011159, "learning_rate": 2.7087769311396606e-06, "loss": 0.3518, "step": 19152 }, { "epoch": 1.3240926374006223, "grad_norm": 0.6824602615332983, "learning_rate": 2.708279416881357e-06, "loss": 0.2592, "step": 19153 }, { "epoch": 1.3241617697891463, "grad_norm": 0.7213085125919787, "learning_rate": 2.7077819313456834e-06, "loss": 0.2268, "step": 19154 }, { "epoch": 1.3242309021776704, "grad_norm": 0.6952606064781984, "learning_rate": 2.7072844745388706e-06, "loss": 0.2827, "step": 19155 }, { "epoch": 1.3243000345661944, "grad_norm": 0.73661347326606, "learning_rate": 2.7067870464671546e-06, "loss": 0.2879, "step": 19156 }, { "epoch": 1.3243691669547184, "grad_norm": 0.634752144494064, "learning_rate": 2.7062896471367734e-06, "loss": 0.1827, "step": 19157 }, { "epoch": 1.3244382993432424, "grad_norm": 0.6740862805027993, "learning_rate": 2.7057922765539557e-06, "loss": 0.2769, "step": 19158 }, { "epoch": 1.3245074317317664, "grad_norm": 0.709264214898951, "learning_rate": 2.7052949347249397e-06, "loss": 0.3361, "step": 19159 }, { "epoch": 1.3245765641202905, "grad_norm": 2.3923721926036863, "learning_rate": 2.7047976216559547e-06, "loss": 0.2863, "step": 19160 }, { "epoch": 1.3246456965088145, "grad_norm": 0.6769520268035946, "learning_rate": 2.7043003373532372e-06, "loss": 0.216, "step": 19161 }, { "epoch": 1.3247148288973385, "grad_norm": 0.7732469278209748, "learning_rate": 2.703803081823015e-06, "loss": 0.3068, "step": 19162 }, { "epoch": 1.3247839612858625, "grad_norm": 0.7679926091547156, "learning_rate": 2.703305855071526e-06, "loss": 0.289, "step": 19163 }, { "epoch": 1.3248530936743865, "grad_norm": 0.6210555967307324, "learning_rate": 2.702808657104999e-06, "loss": 0.2209, "step": 19164 }, { "epoch": 1.3249222260629105, "grad_norm": 0.6276804199769508, "learning_rate": 2.7023114879296676e-06, "loss": 0.2172, "step": 19165 }, { "epoch": 1.3249913584514346, "grad_norm": 0.788824688071742, "learning_rate": 2.7018143475517587e-06, "loss": 0.3753, "step": 19166 }, { "epoch": 1.3250604908399586, "grad_norm": 0.7063547702529285, "learning_rate": 2.701317235977508e-06, "loss": 0.2697, "step": 19167 }, { "epoch": 1.3251296232284826, "grad_norm": 0.680176081083699, "learning_rate": 2.7008201532131416e-06, "loss": 0.2078, "step": 19168 }, { "epoch": 1.3251987556170066, "grad_norm": 0.8922178586137504, "learning_rate": 2.700323099264891e-06, "loss": 0.3186, "step": 19169 }, { "epoch": 1.3252678880055306, "grad_norm": 0.6717363353156901, "learning_rate": 2.699826074138987e-06, "loss": 0.2953, "step": 19170 }, { "epoch": 1.3253370203940547, "grad_norm": 0.7001237243426988, "learning_rate": 2.6993290778416603e-06, "loss": 0.232, "step": 19171 }, { "epoch": 1.3254061527825787, "grad_norm": 0.6776748116738209, "learning_rate": 2.698832110379136e-06, "loss": 0.2318, "step": 19172 }, { "epoch": 1.3254752851711027, "grad_norm": 0.6625154880293967, "learning_rate": 2.6983351717576454e-06, "loss": 0.2967, "step": 19173 }, { "epoch": 1.3255444175596267, "grad_norm": 0.7322248064822054, "learning_rate": 2.6978382619834154e-06, "loss": 0.2708, "step": 19174 }, { "epoch": 1.3256135499481507, "grad_norm": 0.6417746169023293, "learning_rate": 2.697341381062676e-06, "loss": 0.2412, "step": 19175 }, { "epoch": 1.3256826823366747, "grad_norm": 0.6650125866037828, "learning_rate": 2.696844529001651e-06, "loss": 0.2762, "step": 19176 }, { "epoch": 1.3257518147251988, "grad_norm": 0.6956699222825307, "learning_rate": 2.696347705806569e-06, "loss": 0.3808, "step": 19177 }, { "epoch": 1.3258209471137228, "grad_norm": 0.6083043580515949, "learning_rate": 2.6958509114836586e-06, "loss": 0.2548, "step": 19178 }, { "epoch": 1.3258900795022468, "grad_norm": 0.73569852048405, "learning_rate": 2.6953541460391464e-06, "loss": 0.2667, "step": 19179 }, { "epoch": 1.3259592118907708, "grad_norm": 0.6685805875702798, "learning_rate": 2.6948574094792544e-06, "loss": 0.3571, "step": 19180 }, { "epoch": 1.3260283442792948, "grad_norm": 0.6724487988317348, "learning_rate": 2.6943607018102146e-06, "loss": 0.2945, "step": 19181 }, { "epoch": 1.3260974766678189, "grad_norm": 0.6105734431103516, "learning_rate": 2.6938640230382463e-06, "loss": 0.2526, "step": 19182 }, { "epoch": 1.3261666090563429, "grad_norm": 0.7161117702278395, "learning_rate": 2.6933673731695765e-06, "loss": 0.2511, "step": 19183 }, { "epoch": 1.326235741444867, "grad_norm": 0.7872409573654362, "learning_rate": 2.6928707522104304e-06, "loss": 0.3586, "step": 19184 }, { "epoch": 1.326304873833391, "grad_norm": 0.6581507433075839, "learning_rate": 2.6923741601670346e-06, "loss": 0.2038, "step": 19185 }, { "epoch": 1.326374006221915, "grad_norm": 0.6849037601367517, "learning_rate": 2.691877597045608e-06, "loss": 0.2579, "step": 19186 }, { "epoch": 1.326443138610439, "grad_norm": 0.6481091920580614, "learning_rate": 2.6913810628523797e-06, "loss": 0.303, "step": 19187 }, { "epoch": 1.326512270998963, "grad_norm": 0.5981690626074436, "learning_rate": 2.690884557593567e-06, "loss": 0.2711, "step": 19188 }, { "epoch": 1.326581403387487, "grad_norm": 0.5891035264386775, "learning_rate": 2.6903880812753975e-06, "loss": 0.2069, "step": 19189 }, { "epoch": 1.326650535776011, "grad_norm": 0.6605733304130419, "learning_rate": 2.6898916339040903e-06, "loss": 0.2056, "step": 19190 }, { "epoch": 1.326719668164535, "grad_norm": 0.7103641091812657, "learning_rate": 2.6893952154858684e-06, "loss": 0.322, "step": 19191 }, { "epoch": 1.326788800553059, "grad_norm": 0.7611629052736062, "learning_rate": 2.6888988260269543e-06, "loss": 0.2511, "step": 19192 }, { "epoch": 1.326857932941583, "grad_norm": 0.6343045513969681, "learning_rate": 2.6884024655335705e-06, "loss": 0.2243, "step": 19193 }, { "epoch": 1.326927065330107, "grad_norm": 0.7535616918126764, "learning_rate": 2.687906134011935e-06, "loss": 0.306, "step": 19194 }, { "epoch": 1.326996197718631, "grad_norm": 0.7192881739426401, "learning_rate": 2.687409831468272e-06, "loss": 0.3628, "step": 19195 }, { "epoch": 1.327065330107155, "grad_norm": 0.6403128202788415, "learning_rate": 2.6869135579087977e-06, "loss": 0.2121, "step": 19196 }, { "epoch": 1.3271344624956791, "grad_norm": 0.7276479006132568, "learning_rate": 2.6864173133397363e-06, "loss": 0.2745, "step": 19197 }, { "epoch": 1.3272035948842031, "grad_norm": 0.6939648202854476, "learning_rate": 2.685921097767301e-06, "loss": 0.3598, "step": 19198 }, { "epoch": 1.3272727272727272, "grad_norm": 0.6211150474686143, "learning_rate": 2.6854249111977195e-06, "loss": 0.2286, "step": 19199 }, { "epoch": 1.3273418596612512, "grad_norm": 0.7173701059526281, "learning_rate": 2.6849287536372037e-06, "loss": 0.1742, "step": 19200 }, { "epoch": 1.3274109920497752, "grad_norm": 0.6588063342310025, "learning_rate": 2.684432625091976e-06, "loss": 0.2819, "step": 19201 }, { "epoch": 1.3274801244382992, "grad_norm": 0.653750680032508, "learning_rate": 2.6839365255682516e-06, "loss": 0.309, "step": 19202 }, { "epoch": 1.3275492568268232, "grad_norm": 0.6758512974784989, "learning_rate": 2.6834404550722514e-06, "loss": 0.2443, "step": 19203 }, { "epoch": 1.3276183892153475, "grad_norm": 0.6952525521506084, "learning_rate": 2.6829444136101897e-06, "loss": 0.2581, "step": 19204 }, { "epoch": 1.3276875216038715, "grad_norm": 0.636969962416319, "learning_rate": 2.6824484011882845e-06, "loss": 0.3768, "step": 19205 }, { "epoch": 1.3277566539923955, "grad_norm": 0.7402539072094304, "learning_rate": 2.6819524178127522e-06, "loss": 0.3233, "step": 19206 }, { "epoch": 1.3278257863809195, "grad_norm": 0.6173652304140548, "learning_rate": 2.6814564634898123e-06, "loss": 0.2619, "step": 19207 }, { "epoch": 1.3278949187694435, "grad_norm": 0.670487472450326, "learning_rate": 2.6809605382256755e-06, "loss": 0.2264, "step": 19208 }, { "epoch": 1.3279640511579676, "grad_norm": 0.7767463867685521, "learning_rate": 2.6804646420265624e-06, "loss": 0.3323, "step": 19209 }, { "epoch": 1.3280331835464916, "grad_norm": 0.6756694228628606, "learning_rate": 2.679968774898683e-06, "loss": 0.242, "step": 19210 }, { "epoch": 1.3281023159350156, "grad_norm": 0.6361679923569297, "learning_rate": 2.679472936848257e-06, "loss": 0.2105, "step": 19211 }, { "epoch": 1.3281714483235396, "grad_norm": 0.6622708378247519, "learning_rate": 2.6789771278814947e-06, "loss": 0.3175, "step": 19212 }, { "epoch": 1.3282405807120636, "grad_norm": 0.6932113186088324, "learning_rate": 2.678481348004611e-06, "loss": 0.2957, "step": 19213 }, { "epoch": 1.3283097131005877, "grad_norm": 0.637768535345906, "learning_rate": 2.6779855972238223e-06, "loss": 0.2688, "step": 19214 }, { "epoch": 1.3283788454891117, "grad_norm": 0.6773774628844984, "learning_rate": 2.6774898755453414e-06, "loss": 0.2786, "step": 19215 }, { "epoch": 1.3284479778776357, "grad_norm": 0.683698590059351, "learning_rate": 2.6769941829753786e-06, "loss": 0.3521, "step": 19216 }, { "epoch": 1.3285171102661597, "grad_norm": 0.7496207913747129, "learning_rate": 2.6764985195201497e-06, "loss": 0.2892, "step": 19217 }, { "epoch": 1.3285862426546837, "grad_norm": 0.671412931427573, "learning_rate": 2.6760028851858632e-06, "loss": 0.2606, "step": 19218 }, { "epoch": 1.3286553750432077, "grad_norm": 0.7363305394608003, "learning_rate": 2.675507279978734e-06, "loss": 0.3398, "step": 19219 }, { "epoch": 1.3287245074317318, "grad_norm": 0.7463943742664694, "learning_rate": 2.675011703904973e-06, "loss": 0.3107, "step": 19220 }, { "epoch": 1.3287936398202558, "grad_norm": 0.9028133589250005, "learning_rate": 2.6745161569707922e-06, "loss": 0.2475, "step": 19221 }, { "epoch": 1.3288627722087798, "grad_norm": 0.8521237323726747, "learning_rate": 2.6740206391824e-06, "loss": 0.2366, "step": 19222 }, { "epoch": 1.3289319045973038, "grad_norm": 0.6813009932861195, "learning_rate": 2.673525150546009e-06, "loss": 0.3645, "step": 19223 }, { "epoch": 1.3290010369858278, "grad_norm": 0.6648627887771048, "learning_rate": 2.67302969106783e-06, "loss": 0.2773, "step": 19224 }, { "epoch": 1.3290701693743519, "grad_norm": 0.6878510811612938, "learning_rate": 2.672534260754068e-06, "loss": 0.2455, "step": 19225 }, { "epoch": 1.3291393017628759, "grad_norm": 1.1323550208659245, "learning_rate": 2.672038859610936e-06, "loss": 0.3165, "step": 19226 }, { "epoch": 1.3292084341513999, "grad_norm": 0.8157867989974442, "learning_rate": 2.671543487644642e-06, "loss": 0.3264, "step": 19227 }, { "epoch": 1.329277566539924, "grad_norm": 0.6886490440773012, "learning_rate": 2.671048144861398e-06, "loss": 0.237, "step": 19228 }, { "epoch": 1.329346698928448, "grad_norm": 0.8708448366953786, "learning_rate": 2.6705528312674066e-06, "loss": 0.2194, "step": 19229 }, { "epoch": 1.329415831316972, "grad_norm": 0.7067054107441713, "learning_rate": 2.6700575468688805e-06, "loss": 0.3982, "step": 19230 }, { "epoch": 1.329484963705496, "grad_norm": 0.6194097777305505, "learning_rate": 2.669562291672023e-06, "loss": 0.3036, "step": 19231 }, { "epoch": 1.32955409609402, "grad_norm": 0.6391462290069283, "learning_rate": 2.6690670656830464e-06, "loss": 0.2429, "step": 19232 }, { "epoch": 1.329623228482544, "grad_norm": 0.6778979614838417, "learning_rate": 2.668571868908149e-06, "loss": 0.2989, "step": 19233 }, { "epoch": 1.329692360871068, "grad_norm": 0.8717078531019505, "learning_rate": 2.668076701353547e-06, "loss": 0.3849, "step": 19234 }, { "epoch": 1.3297614932595923, "grad_norm": 0.6557632768708046, "learning_rate": 2.6675815630254404e-06, "loss": 0.2663, "step": 19235 }, { "epoch": 1.3298306256481163, "grad_norm": 0.6584625467473235, "learning_rate": 2.6670864539300377e-06, "loss": 0.1958, "step": 19236 }, { "epoch": 1.3298997580366403, "grad_norm": 0.7479835102652002, "learning_rate": 2.6665913740735422e-06, "loss": 0.2951, "step": 19237 }, { "epoch": 1.3299688904251643, "grad_norm": 0.7195303817293545, "learning_rate": 2.6660963234621607e-06, "loss": 0.2987, "step": 19238 }, { "epoch": 1.3300380228136883, "grad_norm": 0.6518411542982665, "learning_rate": 2.6656013021020956e-06, "loss": 0.2027, "step": 19239 }, { "epoch": 1.3301071552022123, "grad_norm": 0.6006095798231232, "learning_rate": 2.665106309999551e-06, "loss": 0.2566, "step": 19240 }, { "epoch": 1.3301762875907364, "grad_norm": 0.6501713648772104, "learning_rate": 2.664611347160733e-06, "loss": 0.3527, "step": 19241 }, { "epoch": 1.3302454199792604, "grad_norm": 0.693746371024791, "learning_rate": 2.6641164135918462e-06, "loss": 0.2349, "step": 19242 }, { "epoch": 1.3303145523677844, "grad_norm": 0.6482860936049502, "learning_rate": 2.6636215092990892e-06, "loss": 0.2483, "step": 19243 }, { "epoch": 1.3303836847563084, "grad_norm": 0.7531771912421658, "learning_rate": 2.663126634288669e-06, "loss": 0.3736, "step": 19244 }, { "epoch": 1.3304528171448324, "grad_norm": 0.6590530614982306, "learning_rate": 2.662631788566784e-06, "loss": 0.3176, "step": 19245 }, { "epoch": 1.3305219495333565, "grad_norm": 0.6409070525626329, "learning_rate": 2.6621369721396408e-06, "loss": 0.2068, "step": 19246 }, { "epoch": 1.3305910819218805, "grad_norm": 0.7167209860928948, "learning_rate": 2.6616421850134366e-06, "loss": 0.3278, "step": 19247 }, { "epoch": 1.3306602143104045, "grad_norm": 0.6785838466751685, "learning_rate": 2.6611474271943737e-06, "loss": 0.4087, "step": 19248 }, { "epoch": 1.3307293466989285, "grad_norm": 0.7487887911268589, "learning_rate": 2.660652698688655e-06, "loss": 0.2566, "step": 19249 }, { "epoch": 1.3307984790874525, "grad_norm": 0.5809713740183259, "learning_rate": 2.6601579995024817e-06, "loss": 0.243, "step": 19250 }, { "epoch": 1.3308676114759765, "grad_norm": 0.8782209849036333, "learning_rate": 2.6596633296420497e-06, "loss": 0.2918, "step": 19251 }, { "epoch": 1.3309367438645006, "grad_norm": 0.7310880317594587, "learning_rate": 2.659168689113564e-06, "loss": 0.3112, "step": 19252 }, { "epoch": 1.3310058762530246, "grad_norm": 0.6728110130798466, "learning_rate": 2.658674077923219e-06, "loss": 0.2674, "step": 19253 }, { "epoch": 1.3310750086415486, "grad_norm": 0.6948442060721159, "learning_rate": 2.6581794960772163e-06, "loss": 0.2624, "step": 19254 }, { "epoch": 1.3311441410300726, "grad_norm": 0.6802300921972171, "learning_rate": 2.657684943581754e-06, "loss": 0.2998, "step": 19255 }, { "epoch": 1.3312132734185966, "grad_norm": 0.6678599601464063, "learning_rate": 2.657190420443033e-06, "loss": 0.2533, "step": 19256 }, { "epoch": 1.3312824058071207, "grad_norm": 0.7137903163072563, "learning_rate": 2.656695926667248e-06, "loss": 0.2656, "step": 19257 }, { "epoch": 1.3313515381956447, "grad_norm": 0.631768107803646, "learning_rate": 2.656201462260599e-06, "loss": 0.2478, "step": 19258 }, { "epoch": 1.3314206705841687, "grad_norm": 0.7461132255495275, "learning_rate": 2.655707027229281e-06, "loss": 0.4364, "step": 19259 }, { "epoch": 1.3314898029726927, "grad_norm": 0.6319297231485222, "learning_rate": 2.655212621579493e-06, "loss": 0.2313, "step": 19260 }, { "epoch": 1.3315589353612167, "grad_norm": 0.5780482340163469, "learning_rate": 2.6547182453174275e-06, "loss": 0.2157, "step": 19261 }, { "epoch": 1.3316280677497407, "grad_norm": 0.7005197394942275, "learning_rate": 2.654223898449284e-06, "loss": 0.3349, "step": 19262 }, { "epoch": 1.3316972001382648, "grad_norm": 0.7124359263898662, "learning_rate": 2.653729580981258e-06, "loss": 0.3037, "step": 19263 }, { "epoch": 1.3317663325267888, "grad_norm": 0.6598961347990214, "learning_rate": 2.653235292919547e-06, "loss": 0.2245, "step": 19264 }, { "epoch": 1.3318354649153128, "grad_norm": 0.6354608020794239, "learning_rate": 2.6527410342703407e-06, "loss": 0.2069, "step": 19265 }, { "epoch": 1.3319045973038368, "grad_norm": 0.6815129991754998, "learning_rate": 2.65224680503984e-06, "loss": 0.3874, "step": 19266 }, { "epoch": 1.3319737296923608, "grad_norm": 0.6233067569131522, "learning_rate": 2.6517526052342322e-06, "loss": 0.2449, "step": 19267 }, { "epoch": 1.3320428620808848, "grad_norm": 0.6349228960986121, "learning_rate": 2.6512584348597182e-06, "loss": 0.1941, "step": 19268 }, { "epoch": 1.3321119944694089, "grad_norm": 0.6844448357188694, "learning_rate": 2.6507642939224833e-06, "loss": 0.3066, "step": 19269 }, { "epoch": 1.3321811268579329, "grad_norm": 0.6623179649718762, "learning_rate": 2.6502701824287303e-06, "loss": 0.3435, "step": 19270 }, { "epoch": 1.332250259246457, "grad_norm": 0.680585008333963, "learning_rate": 2.6497761003846446e-06, "loss": 0.2702, "step": 19271 }, { "epoch": 1.332319391634981, "grad_norm": 0.778009089997674, "learning_rate": 2.6492820477964236e-06, "loss": 0.2876, "step": 19272 }, { "epoch": 1.332388524023505, "grad_norm": 0.6773908614055015, "learning_rate": 2.648788024670255e-06, "loss": 0.3325, "step": 19273 }, { "epoch": 1.332457656412029, "grad_norm": 0.6793782972353208, "learning_rate": 2.648294031012335e-06, "loss": 0.2824, "step": 19274 }, { "epoch": 1.332526788800553, "grad_norm": 0.6113899768917601, "learning_rate": 2.64780006682885e-06, "loss": 0.2079, "step": 19275 }, { "epoch": 1.332595921189077, "grad_norm": 0.6792235939900688, "learning_rate": 2.647306132125994e-06, "loss": 0.2815, "step": 19276 }, { "epoch": 1.332665053577601, "grad_norm": 0.6464898778952921, "learning_rate": 2.646812226909956e-06, "loss": 0.333, "step": 19277 }, { "epoch": 1.332734185966125, "grad_norm": 0.6724908104033619, "learning_rate": 2.6463183511869306e-06, "loss": 0.237, "step": 19278 }, { "epoch": 1.332803318354649, "grad_norm": 0.6563087803365962, "learning_rate": 2.6458245049631015e-06, "loss": 0.2315, "step": 19279 }, { "epoch": 1.332872450743173, "grad_norm": 0.7287353468325434, "learning_rate": 2.645330688244663e-06, "loss": 0.3676, "step": 19280 }, { "epoch": 1.332941583131697, "grad_norm": 0.6530610270561577, "learning_rate": 2.6448369010378005e-06, "loss": 0.2533, "step": 19281 }, { "epoch": 1.333010715520221, "grad_norm": 0.6035760654479468, "learning_rate": 2.6443431433487067e-06, "loss": 0.2419, "step": 19282 }, { "epoch": 1.3330798479087451, "grad_norm": 0.6777262139152019, "learning_rate": 2.643849415183566e-06, "loss": 0.3101, "step": 19283 }, { "epoch": 1.3331489802972691, "grad_norm": 0.7683024894031496, "learning_rate": 2.643355716548567e-06, "loss": 0.3745, "step": 19284 }, { "epoch": 1.3332181126857934, "grad_norm": 0.6358890869900439, "learning_rate": 2.6428620474498996e-06, "loss": 0.2218, "step": 19285 }, { "epoch": 1.3332872450743174, "grad_norm": 0.6553321415745857, "learning_rate": 2.6423684078937516e-06, "loss": 0.2923, "step": 19286 }, { "epoch": 1.3333563774628414, "grad_norm": 0.7357488044554721, "learning_rate": 2.6418747978863056e-06, "loss": 0.3577, "step": 19287 }, { "epoch": 1.3334255098513654, "grad_norm": 0.6888319670977654, "learning_rate": 2.6413812174337535e-06, "loss": 0.325, "step": 19288 }, { "epoch": 1.3334946422398894, "grad_norm": 0.6335148146285025, "learning_rate": 2.6408876665422768e-06, "loss": 0.2008, "step": 19289 }, { "epoch": 1.3335637746284135, "grad_norm": 0.7835614349083568, "learning_rate": 2.6403941452180627e-06, "loss": 0.2495, "step": 19290 }, { "epoch": 1.3336329070169375, "grad_norm": 0.7143888936427712, "learning_rate": 2.639900653467299e-06, "loss": 0.3698, "step": 19291 }, { "epoch": 1.3337020394054615, "grad_norm": 0.6594527438318148, "learning_rate": 2.639407191296167e-06, "loss": 0.2311, "step": 19292 }, { "epoch": 1.3337711717939855, "grad_norm": 0.6621895359618606, "learning_rate": 2.638913758710855e-06, "loss": 0.2775, "step": 19293 }, { "epoch": 1.3338403041825095, "grad_norm": 0.73851474354318, "learning_rate": 2.6384203557175435e-06, "loss": 0.3918, "step": 19294 }, { "epoch": 1.3339094365710336, "grad_norm": 0.6588249881465725, "learning_rate": 2.6379269823224206e-06, "loss": 0.3186, "step": 19295 }, { "epoch": 1.3339785689595576, "grad_norm": 1.6387671946231561, "learning_rate": 2.637433638531666e-06, "loss": 0.2236, "step": 19296 }, { "epoch": 1.3340477013480816, "grad_norm": 0.6581341438937353, "learning_rate": 2.636940324351464e-06, "loss": 0.271, "step": 19297 }, { "epoch": 1.3341168337366056, "grad_norm": 0.7233164750583309, "learning_rate": 2.636447039787998e-06, "loss": 0.3529, "step": 19298 }, { "epoch": 1.3341859661251296, "grad_norm": 2.297929783088524, "learning_rate": 2.6359537848474526e-06, "loss": 0.234, "step": 19299 }, { "epoch": 1.3342550985136536, "grad_norm": 0.6341833331270232, "learning_rate": 2.635460559536006e-06, "loss": 0.2297, "step": 19300 }, { "epoch": 1.3343242309021777, "grad_norm": 0.701192454456256, "learning_rate": 2.6349673638598427e-06, "loss": 0.3279, "step": 19301 }, { "epoch": 1.3343933632907017, "grad_norm": 0.6667909581548995, "learning_rate": 2.634474197825142e-06, "loss": 0.3498, "step": 19302 }, { "epoch": 1.3344624956792257, "grad_norm": 0.5899100452694974, "learning_rate": 2.6339810614380866e-06, "loss": 0.2367, "step": 19303 }, { "epoch": 1.3345316280677497, "grad_norm": 0.6127963625872692, "learning_rate": 2.633487954704853e-06, "loss": 0.2356, "step": 19304 }, { "epoch": 1.3346007604562737, "grad_norm": 0.6224252576152346, "learning_rate": 2.632994877631629e-06, "loss": 0.3436, "step": 19305 }, { "epoch": 1.3346698928447978, "grad_norm": 0.6844714815144645, "learning_rate": 2.6325018302245877e-06, "loss": 0.2944, "step": 19306 }, { "epoch": 1.3347390252333218, "grad_norm": 0.6145497173717264, "learning_rate": 2.632008812489912e-06, "loss": 0.2042, "step": 19307 }, { "epoch": 1.3348081576218458, "grad_norm": 0.6507739913664617, "learning_rate": 2.631515824433779e-06, "loss": 0.2323, "step": 19308 }, { "epoch": 1.3348772900103698, "grad_norm": 0.7389538762575468, "learning_rate": 2.6310228660623705e-06, "loss": 0.3993, "step": 19309 }, { "epoch": 1.3349464223988938, "grad_norm": 0.6938473493589099, "learning_rate": 2.6305299373818604e-06, "loss": 0.2584, "step": 19310 }, { "epoch": 1.3350155547874178, "grad_norm": 0.7633233615107339, "learning_rate": 2.6300370383984296e-06, "loss": 0.2958, "step": 19311 }, { "epoch": 1.3350846871759419, "grad_norm": 0.7600813176746737, "learning_rate": 2.6295441691182543e-06, "loss": 0.287, "step": 19312 }, { "epoch": 1.3351538195644659, "grad_norm": 0.6868280733009208, "learning_rate": 2.6290513295475152e-06, "loss": 0.252, "step": 19313 }, { "epoch": 1.33522295195299, "grad_norm": 0.6378344466083407, "learning_rate": 2.628558519692384e-06, "loss": 0.2168, "step": 19314 }, { "epoch": 1.335292084341514, "grad_norm": 0.6760357834422354, "learning_rate": 2.628065739559042e-06, "loss": 0.2483, "step": 19315 }, { "epoch": 1.3353612167300382, "grad_norm": 0.7037200687931842, "learning_rate": 2.6275729891536613e-06, "loss": 0.393, "step": 19316 }, { "epoch": 1.3354303491185622, "grad_norm": 0.6319506350579386, "learning_rate": 2.6270802684824202e-06, "loss": 0.223, "step": 19317 }, { "epoch": 1.3354994815070862, "grad_norm": 0.6182036893303761, "learning_rate": 2.6265875775514927e-06, "loss": 0.2468, "step": 19318 }, { "epoch": 1.3355686138956102, "grad_norm": 0.7374816852280702, "learning_rate": 2.626094916367054e-06, "loss": 0.3529, "step": 19319 }, { "epoch": 1.3356377462841342, "grad_norm": 0.7411987462964604, "learning_rate": 2.625602284935279e-06, "loss": 0.2908, "step": 19320 }, { "epoch": 1.3357068786726582, "grad_norm": 0.6291436183602689, "learning_rate": 2.6251096832623436e-06, "loss": 0.2624, "step": 19321 }, { "epoch": 1.3357760110611823, "grad_norm": 0.68918443366717, "learning_rate": 2.624617111354419e-06, "loss": 0.2302, "step": 19322 }, { "epoch": 1.3358451434497063, "grad_norm": 0.6651281483510336, "learning_rate": 2.6241245692176808e-06, "loss": 0.368, "step": 19323 }, { "epoch": 1.3359142758382303, "grad_norm": 0.661091015066464, "learning_rate": 2.6236320568583e-06, "loss": 0.2406, "step": 19324 }, { "epoch": 1.3359834082267543, "grad_norm": 0.6431285087362296, "learning_rate": 2.6231395742824505e-06, "loss": 0.2021, "step": 19325 }, { "epoch": 1.3360525406152783, "grad_norm": 0.6816773568280792, "learning_rate": 2.6226471214963044e-06, "loss": 0.2681, "step": 19326 }, { "epoch": 1.3361216730038024, "grad_norm": 0.6860063797462803, "learning_rate": 2.6221546985060364e-06, "loss": 0.2869, "step": 19327 }, { "epoch": 1.3361908053923264, "grad_norm": 0.8646263455113206, "learning_rate": 2.6216623053178137e-06, "loss": 0.2647, "step": 19328 }, { "epoch": 1.3362599377808504, "grad_norm": 0.6699709573406497, "learning_rate": 2.6211699419378113e-06, "loss": 0.2427, "step": 19329 }, { "epoch": 1.3363290701693744, "grad_norm": 0.6846021101152727, "learning_rate": 2.6206776083721967e-06, "loss": 0.2592, "step": 19330 }, { "epoch": 1.3363982025578984, "grad_norm": 0.7778121108525983, "learning_rate": 2.620185304627144e-06, "loss": 0.3181, "step": 19331 }, { "epoch": 1.3364673349464224, "grad_norm": 0.750057435945189, "learning_rate": 2.6196930307088205e-06, "loss": 0.2696, "step": 19332 }, { "epoch": 1.3365364673349465, "grad_norm": 0.9254945738369948, "learning_rate": 2.619200786623396e-06, "loss": 0.3075, "step": 19333 }, { "epoch": 1.3366055997234705, "grad_norm": 0.7337131357948377, "learning_rate": 2.6187085723770412e-06, "loss": 0.4363, "step": 19334 }, { "epoch": 1.3366747321119945, "grad_norm": 0.6038720573643043, "learning_rate": 2.618216387975927e-06, "loss": 0.2424, "step": 19335 }, { "epoch": 1.3367438645005185, "grad_norm": 0.6757312103902907, "learning_rate": 2.6177242334262177e-06, "loss": 0.2473, "step": 19336 }, { "epoch": 1.3368129968890425, "grad_norm": 0.6646274117680204, "learning_rate": 2.617232108734086e-06, "loss": 0.3513, "step": 19337 }, { "epoch": 1.3368821292775666, "grad_norm": 2.128861014911243, "learning_rate": 2.616740013905695e-06, "loss": 0.2387, "step": 19338 }, { "epoch": 1.3369512616660906, "grad_norm": 0.6206903538400057, "learning_rate": 2.616247948947217e-06, "loss": 0.2197, "step": 19339 }, { "epoch": 1.3370203940546146, "grad_norm": 0.6895185095675394, "learning_rate": 2.6157559138648135e-06, "loss": 0.2852, "step": 19340 }, { "epoch": 1.3370895264431386, "grad_norm": 0.6538471997587189, "learning_rate": 2.615263908664658e-06, "loss": 0.3647, "step": 19341 }, { "epoch": 1.3371586588316626, "grad_norm": 0.6764247758015444, "learning_rate": 2.614771933352912e-06, "loss": 0.2364, "step": 19342 }, { "epoch": 1.3372277912201866, "grad_norm": 0.6576632547254199, "learning_rate": 2.6142799879357457e-06, "loss": 0.2751, "step": 19343 }, { "epoch": 1.3372969236087107, "grad_norm": 0.6387990511079695, "learning_rate": 2.61378807241932e-06, "loss": 0.2836, "step": 19344 }, { "epoch": 1.3373660559972347, "grad_norm": 0.6789238522728072, "learning_rate": 2.6132961868098037e-06, "loss": 0.2428, "step": 19345 }, { "epoch": 1.3374351883857587, "grad_norm": 0.7966266310068003, "learning_rate": 2.6128043311133588e-06, "loss": 0.2591, "step": 19346 }, { "epoch": 1.3375043207742827, "grad_norm": 0.7202523135693057, "learning_rate": 2.612312505336152e-06, "loss": 0.3308, "step": 19347 }, { "epoch": 1.3375734531628067, "grad_norm": 0.6886868739123976, "learning_rate": 2.6118207094843463e-06, "loss": 0.3336, "step": 19348 }, { "epoch": 1.3376425855513308, "grad_norm": 0.6983330788517047, "learning_rate": 2.6113289435641086e-06, "loss": 0.2705, "step": 19349 }, { "epoch": 1.3377117179398548, "grad_norm": 0.6399752527124373, "learning_rate": 2.6108372075815974e-06, "loss": 0.2404, "step": 19350 }, { "epoch": 1.3377808503283788, "grad_norm": 0.6624095074019258, "learning_rate": 2.6103455015429803e-06, "loss": 0.2816, "step": 19351 }, { "epoch": 1.3378499827169028, "grad_norm": 1.1698192137873011, "learning_rate": 2.6098538254544157e-06, "loss": 0.3473, "step": 19352 }, { "epoch": 1.3379191151054268, "grad_norm": 0.7054448221588162, "learning_rate": 2.6093621793220704e-06, "loss": 0.3065, "step": 19353 }, { "epoch": 1.3379882474939508, "grad_norm": 0.7246439424048594, "learning_rate": 2.6088705631521015e-06, "loss": 0.2465, "step": 19354 }, { "epoch": 1.3380573798824749, "grad_norm": 0.775393328432504, "learning_rate": 2.6083789769506728e-06, "loss": 0.3105, "step": 19355 }, { "epoch": 1.3381265122709989, "grad_norm": 0.7048863995996092, "learning_rate": 2.6078874207239478e-06, "loss": 0.2512, "step": 19356 }, { "epoch": 1.338195644659523, "grad_norm": 0.6484850791318975, "learning_rate": 2.607395894478083e-06, "loss": 0.2062, "step": 19357 }, { "epoch": 1.338264777048047, "grad_norm": 0.6646321769805849, "learning_rate": 2.606904398219243e-06, "loss": 0.2624, "step": 19358 }, { "epoch": 1.338333909436571, "grad_norm": 0.7097636138931637, "learning_rate": 2.6064129319535835e-06, "loss": 0.3295, "step": 19359 }, { "epoch": 1.338403041825095, "grad_norm": 0.6028019320721463, "learning_rate": 2.605921495687267e-06, "loss": 0.2452, "step": 19360 }, { "epoch": 1.338472174213619, "grad_norm": 0.6417871082816226, "learning_rate": 2.6054300894264517e-06, "loss": 0.2408, "step": 19361 }, { "epoch": 1.338541306602143, "grad_norm": 0.7108150424297117, "learning_rate": 2.6049387131772984e-06, "loss": 0.3377, "step": 19362 }, { "epoch": 1.338610438990667, "grad_norm": 0.7254439519139565, "learning_rate": 2.6044473669459622e-06, "loss": 0.3273, "step": 19363 }, { "epoch": 1.338679571379191, "grad_norm": 0.5971903275572613, "learning_rate": 2.6039560507386064e-06, "loss": 0.1799, "step": 19364 }, { "epoch": 1.338748703767715, "grad_norm": 0.6891979836566441, "learning_rate": 2.6034647645613823e-06, "loss": 0.2816, "step": 19365 }, { "epoch": 1.338817836156239, "grad_norm": 0.6285997486942065, "learning_rate": 2.602973508420453e-06, "loss": 0.3717, "step": 19366 }, { "epoch": 1.3388869685447633, "grad_norm": 0.6962846523368847, "learning_rate": 2.6024822823219716e-06, "loss": 0.266, "step": 19367 }, { "epoch": 1.3389561009332873, "grad_norm": 0.852932714478528, "learning_rate": 2.6019910862720952e-06, "loss": 0.2099, "step": 19368 }, { "epoch": 1.3390252333218113, "grad_norm": 0.7146062717589986, "learning_rate": 2.601499920276982e-06, "loss": 0.3044, "step": 19369 }, { "epoch": 1.3390943657103354, "grad_norm": 0.7729658628920851, "learning_rate": 2.601008784342789e-06, "loss": 0.23, "step": 19370 }, { "epoch": 1.3391634980988594, "grad_norm": 0.6789726138501557, "learning_rate": 2.600517678475667e-06, "loss": 0.2501, "step": 19371 }, { "epoch": 1.3392326304873834, "grad_norm": 0.6301819978537238, "learning_rate": 2.6000266026817762e-06, "loss": 0.2334, "step": 19372 }, { "epoch": 1.3393017628759074, "grad_norm": 0.6722265488485298, "learning_rate": 2.5995355569672664e-06, "loss": 0.3751, "step": 19373 }, { "epoch": 1.3393708952644314, "grad_norm": 0.6992888352592848, "learning_rate": 2.5990445413382975e-06, "loss": 0.2657, "step": 19374 }, { "epoch": 1.3394400276529554, "grad_norm": 0.5828430187174082, "learning_rate": 2.5985535558010168e-06, "loss": 0.2588, "step": 19375 }, { "epoch": 1.3395091600414795, "grad_norm": 0.742396599446336, "learning_rate": 2.5980626003615844e-06, "loss": 0.3108, "step": 19376 }, { "epoch": 1.3395782924300035, "grad_norm": 0.6646762494294338, "learning_rate": 2.5975716750261498e-06, "loss": 0.283, "step": 19377 }, { "epoch": 1.3396474248185275, "grad_norm": 0.602215086003522, "learning_rate": 2.5970807798008683e-06, "loss": 0.2544, "step": 19378 }, { "epoch": 1.3397165572070515, "grad_norm": 0.7370892926329474, "learning_rate": 2.5965899146918894e-06, "loss": 0.2506, "step": 19379 }, { "epoch": 1.3397856895955755, "grad_norm": 0.6931983581894346, "learning_rate": 2.5960990797053694e-06, "loss": 0.3528, "step": 19380 }, { "epoch": 1.3398548219840996, "grad_norm": 0.8241322278472816, "learning_rate": 2.5956082748474547e-06, "loss": 0.32, "step": 19381 }, { "epoch": 1.3399239543726236, "grad_norm": 0.6510961043699748, "learning_rate": 2.5951175001243003e-06, "loss": 0.2451, "step": 19382 }, { "epoch": 1.3399930867611476, "grad_norm": 0.6699123230873765, "learning_rate": 2.594626755542055e-06, "loss": 0.2752, "step": 19383 }, { "epoch": 1.3400622191496716, "grad_norm": 0.7216378739815714, "learning_rate": 2.594136041106874e-06, "loss": 0.3738, "step": 19384 }, { "epoch": 1.3401313515381956, "grad_norm": 0.6428541606265337, "learning_rate": 2.593645356824901e-06, "loss": 0.2288, "step": 19385 }, { "epoch": 1.3402004839267196, "grad_norm": 0.6653304456701535, "learning_rate": 2.5931547027022918e-06, "loss": 0.2723, "step": 19386 }, { "epoch": 1.3402696163152437, "grad_norm": 0.6810822987243161, "learning_rate": 2.5926640787451917e-06, "loss": 0.3458, "step": 19387 }, { "epoch": 1.3403387487037677, "grad_norm": 0.6763526672247095, "learning_rate": 2.5921734849597526e-06, "loss": 0.3092, "step": 19388 }, { "epoch": 1.3404078810922917, "grad_norm": 0.6749524912020941, "learning_rate": 2.59168292135212e-06, "loss": 0.2256, "step": 19389 }, { "epoch": 1.3404770134808157, "grad_norm": 0.6583296217800921, "learning_rate": 2.5911923879284445e-06, "loss": 0.2733, "step": 19390 }, { "epoch": 1.3405461458693397, "grad_norm": 0.6911184917649102, "learning_rate": 2.5907018846948733e-06, "loss": 0.3477, "step": 19391 }, { "epoch": 1.3406152782578638, "grad_norm": 0.6329314621891514, "learning_rate": 2.5902114116575565e-06, "loss": 0.2849, "step": 19392 }, { "epoch": 1.3406844106463878, "grad_norm": 0.8031732692663578, "learning_rate": 2.5897209688226378e-06, "loss": 0.2258, "step": 19393 }, { "epoch": 1.3407535430349118, "grad_norm": 0.7345740224405726, "learning_rate": 2.589230556196266e-06, "loss": 0.3182, "step": 19394 }, { "epoch": 1.3408226754234358, "grad_norm": 0.672089049629186, "learning_rate": 2.588740173784586e-06, "loss": 0.2624, "step": 19395 }, { "epoch": 1.3408918078119598, "grad_norm": 0.6831716733292951, "learning_rate": 2.5882498215937445e-06, "loss": 0.214, "step": 19396 }, { "epoch": 1.340960940200484, "grad_norm": 0.6406616299224311, "learning_rate": 2.587759499629888e-06, "loss": 0.2533, "step": 19397 }, { "epoch": 1.341030072589008, "grad_norm": 0.7568922625400756, "learning_rate": 2.5872692078991636e-06, "loss": 0.3535, "step": 19398 }, { "epoch": 1.341099204977532, "grad_norm": 0.6365539064720983, "learning_rate": 2.586778946407711e-06, "loss": 0.2399, "step": 19399 }, { "epoch": 1.3411683373660561, "grad_norm": 0.6002455065910788, "learning_rate": 2.58628871516168e-06, "loss": 0.2648, "step": 19400 }, { "epoch": 1.3412374697545801, "grad_norm": 0.6802531528754214, "learning_rate": 2.5857985141672104e-06, "loss": 0.2579, "step": 19401 }, { "epoch": 1.3413066021431042, "grad_norm": 0.6604918693398344, "learning_rate": 2.5853083434304506e-06, "loss": 0.3542, "step": 19402 }, { "epoch": 1.3413757345316282, "grad_norm": 0.6551811971809164, "learning_rate": 2.584818202957539e-06, "loss": 0.2087, "step": 19403 }, { "epoch": 1.3414448669201522, "grad_norm": 0.7194462376002716, "learning_rate": 2.5843280927546215e-06, "loss": 0.2632, "step": 19404 }, { "epoch": 1.3415139993086762, "grad_norm": 0.7067088903662347, "learning_rate": 2.583838012827841e-06, "loss": 0.3583, "step": 19405 }, { "epoch": 1.3415831316972002, "grad_norm": 0.680332298556535, "learning_rate": 2.5833479631833398e-06, "loss": 0.2708, "step": 19406 }, { "epoch": 1.3416522640857242, "grad_norm": 0.6046289500670526, "learning_rate": 2.5828579438272576e-06, "loss": 0.212, "step": 19407 }, { "epoch": 1.3417213964742483, "grad_norm": 0.725292667836324, "learning_rate": 2.58236795476574e-06, "loss": 0.2619, "step": 19408 }, { "epoch": 1.3417905288627723, "grad_norm": 0.7028067098020172, "learning_rate": 2.5818779960049234e-06, "loss": 0.3541, "step": 19409 }, { "epoch": 1.3418596612512963, "grad_norm": 0.7855262262279724, "learning_rate": 2.581388067550953e-06, "loss": 0.3668, "step": 19410 }, { "epoch": 1.3419287936398203, "grad_norm": 0.5602426941995967, "learning_rate": 2.580898169409962e-06, "loss": 0.1943, "step": 19411 }, { "epoch": 1.3419979260283443, "grad_norm": 0.6526123763199131, "learning_rate": 2.5804083015881e-06, "loss": 0.3381, "step": 19412 }, { "epoch": 1.3420670584168684, "grad_norm": 0.7423841853339492, "learning_rate": 2.5799184640915004e-06, "loss": 0.3245, "step": 19413 }, { "epoch": 1.3421361908053924, "grad_norm": 0.6155573630906499, "learning_rate": 2.5794286569263055e-06, "loss": 0.2603, "step": 19414 }, { "epoch": 1.3422053231939164, "grad_norm": 0.7256471549958668, "learning_rate": 2.578938880098651e-06, "loss": 0.2987, "step": 19415 }, { "epoch": 1.3422744555824404, "grad_norm": 0.7319502047657425, "learning_rate": 2.5784491336146787e-06, "loss": 0.3635, "step": 19416 }, { "epoch": 1.3423435879709644, "grad_norm": 0.6597028633142378, "learning_rate": 2.5779594174805233e-06, "loss": 0.2197, "step": 19417 }, { "epoch": 1.3424127203594884, "grad_norm": 0.7638542875549217, "learning_rate": 2.5774697317023245e-06, "loss": 0.2621, "step": 19418 }, { "epoch": 1.3424818527480125, "grad_norm": 0.7702052843337134, "learning_rate": 2.5769800762862186e-06, "loss": 0.3638, "step": 19419 }, { "epoch": 1.3425509851365365, "grad_norm": 0.6860790342370404, "learning_rate": 2.5764904512383458e-06, "loss": 0.3401, "step": 19420 }, { "epoch": 1.3426201175250605, "grad_norm": 0.7696522101587435, "learning_rate": 2.5760008565648385e-06, "loss": 0.2191, "step": 19421 }, { "epoch": 1.3426892499135845, "grad_norm": 0.7005270680185495, "learning_rate": 2.5755112922718366e-06, "loss": 0.2796, "step": 19422 }, { "epoch": 1.3427583823021085, "grad_norm": 0.6779767015792864, "learning_rate": 2.5750217583654734e-06, "loss": 0.3406, "step": 19423 }, { "epoch": 1.3428275146906326, "grad_norm": 0.7133059677033854, "learning_rate": 2.5745322548518835e-06, "loss": 0.2662, "step": 19424 }, { "epoch": 1.3428966470791566, "grad_norm": 0.6327954916331843, "learning_rate": 2.5740427817372036e-06, "loss": 0.2397, "step": 19425 }, { "epoch": 1.3429657794676806, "grad_norm": 0.7187439542276252, "learning_rate": 2.573553339027568e-06, "loss": 0.2945, "step": 19426 }, { "epoch": 1.3430349118562046, "grad_norm": 0.8625433325192498, "learning_rate": 2.5730639267291126e-06, "loss": 0.3276, "step": 19427 }, { "epoch": 1.3431040442447286, "grad_norm": 0.7219795508264117, "learning_rate": 2.5725745448479678e-06, "loss": 0.2382, "step": 19428 }, { "epoch": 1.3431731766332526, "grad_norm": 0.7225591296002009, "learning_rate": 2.572085193390271e-06, "loss": 0.2462, "step": 19429 }, { "epoch": 1.3432423090217767, "grad_norm": 0.6830226164141201, "learning_rate": 2.5715958723621527e-06, "loss": 0.3396, "step": 19430 }, { "epoch": 1.3433114414103007, "grad_norm": 0.7490482051724319, "learning_rate": 2.5711065817697457e-06, "loss": 0.3505, "step": 19431 }, { "epoch": 1.3433805737988247, "grad_norm": 0.5663543061380931, "learning_rate": 2.570617321619183e-06, "loss": 0.2689, "step": 19432 }, { "epoch": 1.3434497061873487, "grad_norm": 0.6723718564520695, "learning_rate": 2.570128091916598e-06, "loss": 0.2459, "step": 19433 }, { "epoch": 1.3435188385758727, "grad_norm": 0.6363518419274004, "learning_rate": 2.5696388926681205e-06, "loss": 0.308, "step": 19434 }, { "epoch": 1.3435879709643967, "grad_norm": 0.6512633284712498, "learning_rate": 2.5691497238798836e-06, "loss": 0.2177, "step": 19435 }, { "epoch": 1.3436571033529208, "grad_norm": 0.5730478243872794, "learning_rate": 2.568660585558015e-06, "loss": 0.2217, "step": 19436 }, { "epoch": 1.3437262357414448, "grad_norm": 0.7348421726036029, "learning_rate": 2.5681714777086485e-06, "loss": 0.3594, "step": 19437 }, { "epoch": 1.3437953681299688, "grad_norm": 0.6899253469305119, "learning_rate": 2.567682400337912e-06, "loss": 0.3344, "step": 19438 }, { "epoch": 1.3438645005184928, "grad_norm": 0.6862977590946123, "learning_rate": 2.5671933534519345e-06, "loss": 0.2605, "step": 19439 }, { "epoch": 1.3439336329070168, "grad_norm": 0.6829704836755045, "learning_rate": 2.5667043370568476e-06, "loss": 0.2172, "step": 19440 }, { "epoch": 1.3440027652955409, "grad_norm": 0.6952647452904864, "learning_rate": 2.566215351158781e-06, "loss": 0.3618, "step": 19441 }, { "epoch": 1.3440718976840649, "grad_norm": 0.7752825552537158, "learning_rate": 2.5657263957638602e-06, "loss": 0.3061, "step": 19442 }, { "epoch": 1.344141030072589, "grad_norm": 0.7340299152587331, "learning_rate": 2.5652374708782167e-06, "loss": 0.3287, "step": 19443 }, { "epoch": 1.344210162461113, "grad_norm": 0.6441000664976143, "learning_rate": 2.5647485765079743e-06, "loss": 0.2941, "step": 19444 }, { "epoch": 1.344279294849637, "grad_norm": 0.6202947999683159, "learning_rate": 2.564259712659265e-06, "loss": 0.2853, "step": 19445 }, { "epoch": 1.344348427238161, "grad_norm": 0.6307530436118878, "learning_rate": 2.5637708793382087e-06, "loss": 0.2371, "step": 19446 }, { "epoch": 1.344417559626685, "grad_norm": 0.6879554532414683, "learning_rate": 2.5632820765509414e-06, "loss": 0.2747, "step": 19447 }, { "epoch": 1.3444866920152092, "grad_norm": 0.7170261208576806, "learning_rate": 2.5627933043035822e-06, "loss": 0.3527, "step": 19448 }, { "epoch": 1.3445558244037332, "grad_norm": 0.8984164340939147, "learning_rate": 2.5623045626022623e-06, "loss": 0.3315, "step": 19449 }, { "epoch": 1.3446249567922572, "grad_norm": 0.6191851668659548, "learning_rate": 2.5618158514531023e-06, "loss": 0.222, "step": 19450 }, { "epoch": 1.3446940891807813, "grad_norm": 0.6566627758194138, "learning_rate": 2.5613271708622313e-06, "loss": 0.3133, "step": 19451 }, { "epoch": 1.3447632215693053, "grad_norm": 0.8630106408185549, "learning_rate": 2.56083852083577e-06, "loss": 0.2517, "step": 19452 }, { "epoch": 1.3448323539578293, "grad_norm": 0.6450734676491279, "learning_rate": 2.560349901379845e-06, "loss": 0.1785, "step": 19453 }, { "epoch": 1.3449014863463533, "grad_norm": 0.6559967402532081, "learning_rate": 2.55986131250058e-06, "loss": 0.242, "step": 19454 }, { "epoch": 1.3449706187348773, "grad_norm": 0.7141583405488614, "learning_rate": 2.5593727542041015e-06, "loss": 0.3404, "step": 19455 }, { "epoch": 1.3450397511234014, "grad_norm": 0.6784029603642222, "learning_rate": 2.5588842264965273e-06, "loss": 0.2943, "step": 19456 }, { "epoch": 1.3451088835119254, "grad_norm": 0.8271966901328018, "learning_rate": 2.5583957293839855e-06, "loss": 0.2644, "step": 19457 }, { "epoch": 1.3451780159004494, "grad_norm": 0.7554306247673223, "learning_rate": 2.557907262872594e-06, "loss": 0.2671, "step": 19458 }, { "epoch": 1.3452471482889734, "grad_norm": 0.6360888342109409, "learning_rate": 2.5574188269684786e-06, "loss": 0.3184, "step": 19459 }, { "epoch": 1.3453162806774974, "grad_norm": 0.7143690979056637, "learning_rate": 2.556930421677757e-06, "loss": 0.2463, "step": 19460 }, { "epoch": 1.3453854130660214, "grad_norm": 0.6781496084625411, "learning_rate": 2.556442047006553e-06, "loss": 0.2006, "step": 19461 }, { "epoch": 1.3454545454545455, "grad_norm": 0.7016244206658816, "learning_rate": 2.555953702960987e-06, "loss": 0.3919, "step": 19462 }, { "epoch": 1.3455236778430695, "grad_norm": 0.6918634124831731, "learning_rate": 2.5554653895471825e-06, "loss": 0.2557, "step": 19463 }, { "epoch": 1.3455928102315935, "grad_norm": 0.6794676849499796, "learning_rate": 2.5549771067712543e-06, "loss": 0.3129, "step": 19464 }, { "epoch": 1.3456619426201175, "grad_norm": 0.7040952093863909, "learning_rate": 2.5544888546393264e-06, "loss": 0.2314, "step": 19465 }, { "epoch": 1.3457310750086415, "grad_norm": 0.8058365207375855, "learning_rate": 2.554000633157515e-06, "loss": 0.3317, "step": 19466 }, { "epoch": 1.3458002073971655, "grad_norm": 0.6743268238527519, "learning_rate": 2.55351244233194e-06, "loss": 0.2394, "step": 19467 }, { "epoch": 1.3458693397856896, "grad_norm": 0.6523903433509881, "learning_rate": 2.553024282168721e-06, "loss": 0.2374, "step": 19468 }, { "epoch": 1.3459384721742136, "grad_norm": 0.5865158833040403, "learning_rate": 2.552536152673978e-06, "loss": 0.2441, "step": 19469 }, { "epoch": 1.3460076045627376, "grad_norm": 0.7209986711765072, "learning_rate": 2.552048053853825e-06, "loss": 0.3201, "step": 19470 }, { "epoch": 1.3460767369512616, "grad_norm": 0.6162106165092023, "learning_rate": 2.5515599857143824e-06, "loss": 0.2529, "step": 19471 }, { "epoch": 1.3461458693397856, "grad_norm": 0.6834809519349067, "learning_rate": 2.551071948261764e-06, "loss": 0.2765, "step": 19472 }, { "epoch": 1.3462150017283097, "grad_norm": 0.6443276412914591, "learning_rate": 2.5505839415020905e-06, "loss": 0.3754, "step": 19473 }, { "epoch": 1.3462841341168337, "grad_norm": 0.7219581813915835, "learning_rate": 2.5500959654414736e-06, "loss": 0.2449, "step": 19474 }, { "epoch": 1.3463532665053577, "grad_norm": 0.6815942516440758, "learning_rate": 2.549608020086033e-06, "loss": 0.242, "step": 19475 }, { "epoch": 1.3464223988938817, "grad_norm": 0.6488205865507812, "learning_rate": 2.5491201054418817e-06, "loss": 0.2482, "step": 19476 }, { "epoch": 1.3464915312824057, "grad_norm": 0.6643265049281358, "learning_rate": 2.5486322215151383e-06, "loss": 0.29, "step": 19477 }, { "epoch": 1.34656066367093, "grad_norm": 0.6687784891747399, "learning_rate": 2.5481443683119138e-06, "loss": 0.1979, "step": 19478 }, { "epoch": 1.346629796059454, "grad_norm": 1.0372698676467607, "learning_rate": 2.5476565458383263e-06, "loss": 0.2195, "step": 19479 }, { "epoch": 1.346698928447978, "grad_norm": 0.6505036015889211, "learning_rate": 2.547168754100485e-06, "loss": 0.3419, "step": 19480 }, { "epoch": 1.346768060836502, "grad_norm": 0.762632661249884, "learning_rate": 2.546680993104508e-06, "loss": 0.3019, "step": 19481 }, { "epoch": 1.346837193225026, "grad_norm": 0.6886788116117378, "learning_rate": 2.5461932628565034e-06, "loss": 0.1918, "step": 19482 }, { "epoch": 1.34690632561355, "grad_norm": 0.6713270265977406, "learning_rate": 2.5457055633625907e-06, "loss": 0.2674, "step": 19483 }, { "epoch": 1.346975458002074, "grad_norm": 0.7260516389587768, "learning_rate": 2.5452178946288763e-06, "loss": 0.407, "step": 19484 }, { "epoch": 1.347044590390598, "grad_norm": 0.6846630149126341, "learning_rate": 2.5447302566614777e-06, "loss": 0.2327, "step": 19485 }, { "epoch": 1.3471137227791221, "grad_norm": 0.6574635741700405, "learning_rate": 2.544242649466502e-06, "loss": 0.25, "step": 19486 }, { "epoch": 1.3471828551676461, "grad_norm": 0.6863747062023354, "learning_rate": 2.543755073050063e-06, "loss": 0.3608, "step": 19487 }, { "epoch": 1.3472519875561701, "grad_norm": 0.6784752245120796, "learning_rate": 2.5432675274182694e-06, "loss": 0.2936, "step": 19488 }, { "epoch": 1.3473211199446942, "grad_norm": 0.6226081112386477, "learning_rate": 2.5427800125772334e-06, "loss": 0.2289, "step": 19489 }, { "epoch": 1.3473902523332182, "grad_norm": 0.6959744102987935, "learning_rate": 2.542292528533067e-06, "loss": 0.2763, "step": 19490 }, { "epoch": 1.3474593847217422, "grad_norm": 0.6810793025771178, "learning_rate": 2.541805075291875e-06, "loss": 0.3422, "step": 19491 }, { "epoch": 1.3475285171102662, "grad_norm": 0.7441903585707366, "learning_rate": 2.5413176528597717e-06, "loss": 0.2382, "step": 19492 }, { "epoch": 1.3475976494987902, "grad_norm": 0.5705415020691453, "learning_rate": 2.5408302612428624e-06, "loss": 0.2206, "step": 19493 }, { "epoch": 1.3476667818873143, "grad_norm": 0.6398515528353086, "learning_rate": 2.5403429004472586e-06, "loss": 0.2799, "step": 19494 }, { "epoch": 1.3477359142758383, "grad_norm": 0.6569793669281723, "learning_rate": 2.539855570479065e-06, "loss": 0.2943, "step": 19495 }, { "epoch": 1.3478050466643623, "grad_norm": 0.7125733107188363, "learning_rate": 2.539368271344392e-06, "loss": 0.2713, "step": 19496 }, { "epoch": 1.3478741790528863, "grad_norm": 0.7137227024211321, "learning_rate": 2.538881003049347e-06, "loss": 0.3396, "step": 19497 }, { "epoch": 1.3479433114414103, "grad_norm": 0.6556355195083517, "learning_rate": 2.538393765600038e-06, "loss": 0.3389, "step": 19498 }, { "epoch": 1.3480124438299343, "grad_norm": 0.6265285092582149, "learning_rate": 2.5379065590025674e-06, "loss": 0.2462, "step": 19499 }, { "epoch": 1.3480815762184584, "grad_norm": 0.655608335349181, "learning_rate": 2.537419383263048e-06, "loss": 0.2384, "step": 19500 }, { "epoch": 1.3481507086069824, "grad_norm": 0.6419562690961657, "learning_rate": 2.536932238387578e-06, "loss": 0.3219, "step": 19501 }, { "epoch": 1.3482198409955064, "grad_norm": 0.7069290959092215, "learning_rate": 2.5364451243822687e-06, "loss": 0.3472, "step": 19502 }, { "epoch": 1.3482889733840304, "grad_norm": 0.7281557976152438, "learning_rate": 2.5359580412532227e-06, "loss": 0.2346, "step": 19503 }, { "epoch": 1.3483581057725544, "grad_norm": 1.3721139872412886, "learning_rate": 2.5354709890065465e-06, "loss": 0.272, "step": 19504 }, { "epoch": 1.3484272381610785, "grad_norm": 0.9087421223925152, "learning_rate": 2.534983967648342e-06, "loss": 0.331, "step": 19505 }, { "epoch": 1.3484963705496025, "grad_norm": 0.7287570409365842, "learning_rate": 2.5344969771847167e-06, "loss": 0.3253, "step": 19506 }, { "epoch": 1.3485655029381265, "grad_norm": 0.6115206075692453, "learning_rate": 2.5340100176217696e-06, "loss": 0.2258, "step": 19507 }, { "epoch": 1.3486346353266505, "grad_norm": 1.0460834150965843, "learning_rate": 2.533523088965608e-06, "loss": 0.2401, "step": 19508 }, { "epoch": 1.3487037677151745, "grad_norm": 0.6763148089919573, "learning_rate": 2.533036191222331e-06, "loss": 0.3683, "step": 19509 }, { "epoch": 1.3487729001036985, "grad_norm": 0.6416356319529831, "learning_rate": 2.5325493243980437e-06, "loss": 0.2523, "step": 19510 }, { "epoch": 1.3488420324922226, "grad_norm": 0.630065428872407, "learning_rate": 2.5320624884988467e-06, "loss": 0.2513, "step": 19511 }, { "epoch": 1.3489111648807466, "grad_norm": 0.7332100150047942, "learning_rate": 2.5315756835308436e-06, "loss": 0.3124, "step": 19512 }, { "epoch": 1.3489802972692706, "grad_norm": 0.9781949263199893, "learning_rate": 2.5310889095001324e-06, "loss": 0.2916, "step": 19513 }, { "epoch": 1.3490494296577946, "grad_norm": 0.626451561383102, "learning_rate": 2.5306021664128176e-06, "loss": 0.2668, "step": 19514 }, { "epoch": 1.3491185620463186, "grad_norm": 0.6560361238879736, "learning_rate": 2.530115454274996e-06, "loss": 0.2248, "step": 19515 }, { "epoch": 1.3491876944348427, "grad_norm": 0.6142898057336116, "learning_rate": 2.5296287730927717e-06, "loss": 0.3017, "step": 19516 }, { "epoch": 1.3492568268233667, "grad_norm": 0.696435461985679, "learning_rate": 2.52914212287224e-06, "loss": 0.2739, "step": 19517 }, { "epoch": 1.3493259592118907, "grad_norm": 1.2031054304456326, "learning_rate": 2.5286555036195026e-06, "loss": 0.2424, "step": 19518 }, { "epoch": 1.3493950916004147, "grad_norm": 0.6471502933881099, "learning_rate": 2.528168915340658e-06, "loss": 0.3563, "step": 19519 }, { "epoch": 1.3494642239889387, "grad_norm": 0.666522021017496, "learning_rate": 2.5276823580418063e-06, "loss": 0.2859, "step": 19520 }, { "epoch": 1.3495333563774627, "grad_norm": 0.6550667681362516, "learning_rate": 2.5271958317290433e-06, "loss": 0.2101, "step": 19521 }, { "epoch": 1.3496024887659868, "grad_norm": 0.835828236409424, "learning_rate": 2.526709336408469e-06, "loss": 0.2809, "step": 19522 }, { "epoch": 1.3496716211545108, "grad_norm": 0.6738774754005503, "learning_rate": 2.526222872086177e-06, "loss": 0.382, "step": 19523 }, { "epoch": 1.3497407535430348, "grad_norm": 0.722605649186704, "learning_rate": 2.5257364387682674e-06, "loss": 0.2974, "step": 19524 }, { "epoch": 1.3498098859315588, "grad_norm": 0.6554255846257396, "learning_rate": 2.5252500364608357e-06, "loss": 0.2241, "step": 19525 }, { "epoch": 1.3498790183200828, "grad_norm": 0.659364572490736, "learning_rate": 2.52476366516998e-06, "loss": 0.3177, "step": 19526 }, { "epoch": 1.3499481507086069, "grad_norm": 0.6763945866790403, "learning_rate": 2.5242773249017932e-06, "loss": 0.2877, "step": 19527 }, { "epoch": 1.3500172830971309, "grad_norm": 0.6732358970368467, "learning_rate": 2.5237910156623734e-06, "loss": 0.231, "step": 19528 }, { "epoch": 1.3500864154856551, "grad_norm": 0.6663106961066719, "learning_rate": 2.523304737457813e-06, "loss": 0.2383, "step": 19529 }, { "epoch": 1.3501555478741791, "grad_norm": 0.7439791511935631, "learning_rate": 2.5228184902942086e-06, "loss": 0.3929, "step": 19530 }, { "epoch": 1.3502246802627031, "grad_norm": 0.7767680660774254, "learning_rate": 2.5223322741776537e-06, "loss": 0.2588, "step": 19531 }, { "epoch": 1.3502938126512272, "grad_norm": 0.6080543667757228, "learning_rate": 2.5218460891142405e-06, "loss": 0.2271, "step": 19532 }, { "epoch": 1.3503629450397512, "grad_norm": 0.6605854060010086, "learning_rate": 2.5213599351100645e-06, "loss": 0.2905, "step": 19533 }, { "epoch": 1.3504320774282752, "grad_norm": 0.8049066837900166, "learning_rate": 2.5208738121712203e-06, "loss": 0.325, "step": 19534 }, { "epoch": 1.3505012098167992, "grad_norm": 0.7873045345690013, "learning_rate": 2.5203877203037964e-06, "loss": 0.2763, "step": 19535 }, { "epoch": 1.3505703422053232, "grad_norm": 0.7038705915089422, "learning_rate": 2.5199016595138902e-06, "loss": 0.2106, "step": 19536 }, { "epoch": 1.3506394745938473, "grad_norm": 0.7135020295469537, "learning_rate": 2.5194156298075877e-06, "loss": 0.3236, "step": 19537 }, { "epoch": 1.3507086069823713, "grad_norm": 0.6770737572354947, "learning_rate": 2.5189296311909845e-06, "loss": 0.272, "step": 19538 }, { "epoch": 1.3507777393708953, "grad_norm": 0.6555528224163314, "learning_rate": 2.5184436636701703e-06, "loss": 0.2654, "step": 19539 }, { "epoch": 1.3508468717594193, "grad_norm": 0.7033269338356708, "learning_rate": 2.517957727251238e-06, "loss": 0.3283, "step": 19540 }, { "epoch": 1.3509160041479433, "grad_norm": 0.663650623099165, "learning_rate": 2.517471821940275e-06, "loss": 0.3017, "step": 19541 }, { "epoch": 1.3509851365364673, "grad_norm": 1.1379574729408666, "learning_rate": 2.5169859477433732e-06, "loss": 0.2255, "step": 19542 }, { "epoch": 1.3510542689249914, "grad_norm": 0.6624714269738303, "learning_rate": 2.5165001046666204e-06, "loss": 0.2871, "step": 19543 }, { "epoch": 1.3511234013135154, "grad_norm": 0.7606090725010661, "learning_rate": 2.516014292716108e-06, "loss": 0.2916, "step": 19544 }, { "epoch": 1.3511925337020394, "grad_norm": 0.795443425565084, "learning_rate": 2.5155285118979223e-06, "loss": 0.2508, "step": 19545 }, { "epoch": 1.3512616660905634, "grad_norm": 0.6975233430524769, "learning_rate": 2.515042762218153e-06, "loss": 0.265, "step": 19546 }, { "epoch": 1.3513307984790874, "grad_norm": 0.7117001085329027, "learning_rate": 2.514557043682887e-06, "loss": 0.2341, "step": 19547 }, { "epoch": 1.3513999308676115, "grad_norm": 0.6776947487303691, "learning_rate": 2.5140713562982162e-06, "loss": 0.3324, "step": 19548 }, { "epoch": 1.3514690632561355, "grad_norm": 0.6904613130168751, "learning_rate": 2.5135857000702217e-06, "loss": 0.2534, "step": 19549 }, { "epoch": 1.3515381956446595, "grad_norm": 0.6569689817652969, "learning_rate": 2.513100075004996e-06, "loss": 0.2238, "step": 19550 }, { "epoch": 1.3516073280331835, "grad_norm": 0.6358258262722735, "learning_rate": 2.5126144811086208e-06, "loss": 0.2909, "step": 19551 }, { "epoch": 1.3516764604217075, "grad_norm": 0.6122315311179127, "learning_rate": 2.5121289183871857e-06, "loss": 0.3054, "step": 19552 }, { "epoch": 1.3517455928102315, "grad_norm": 0.7345902861251382, "learning_rate": 2.5116433868467706e-06, "loss": 0.2778, "step": 19553 }, { "epoch": 1.3518147251987556, "grad_norm": 0.6779445486421434, "learning_rate": 2.5111578864934692e-06, "loss": 0.2102, "step": 19554 }, { "epoch": 1.3518838575872796, "grad_norm": 0.6579050283483543, "learning_rate": 2.510672417333362e-06, "loss": 0.32, "step": 19555 }, { "epoch": 1.3519529899758036, "grad_norm": 0.6590858167696175, "learning_rate": 2.510186979372532e-06, "loss": 0.2625, "step": 19556 }, { "epoch": 1.3520221223643276, "grad_norm": 0.6476015815424512, "learning_rate": 2.509701572617067e-06, "loss": 0.2468, "step": 19557 }, { "epoch": 1.3520912547528516, "grad_norm": 0.7456065911539766, "learning_rate": 2.5092161970730467e-06, "loss": 0.3166, "step": 19558 }, { "epoch": 1.3521603871413757, "grad_norm": 0.6935021366715542, "learning_rate": 2.508730852746556e-06, "loss": 0.3162, "step": 19559 }, { "epoch": 1.3522295195299, "grad_norm": 0.659118475667934, "learning_rate": 2.5082455396436782e-06, "loss": 0.2417, "step": 19560 }, { "epoch": 1.352298651918424, "grad_norm": 0.7005690984426128, "learning_rate": 2.507760257770498e-06, "loss": 0.2365, "step": 19561 }, { "epoch": 1.352367784306948, "grad_norm": 0.6429037851403958, "learning_rate": 2.507275007133093e-06, "loss": 0.3151, "step": 19562 }, { "epoch": 1.352436916695472, "grad_norm": 0.6978895683867117, "learning_rate": 2.5067897877375503e-06, "loss": 0.3124, "step": 19563 }, { "epoch": 1.352506049083996, "grad_norm": 0.6435303798560112, "learning_rate": 2.5063045995899448e-06, "loss": 0.2433, "step": 19564 }, { "epoch": 1.35257518147252, "grad_norm": 0.7371335410712746, "learning_rate": 2.5058194426963646e-06, "loss": 0.2862, "step": 19565 }, { "epoch": 1.352644313861044, "grad_norm": 0.7681558427407841, "learning_rate": 2.505334317062883e-06, "loss": 0.3654, "step": 19566 }, { "epoch": 1.352713446249568, "grad_norm": 1.3196215619990004, "learning_rate": 2.504849222695585e-06, "loss": 0.2714, "step": 19567 }, { "epoch": 1.352782578638092, "grad_norm": 0.6196553976930425, "learning_rate": 2.5043641596005487e-06, "loss": 0.2631, "step": 19568 }, { "epoch": 1.352851711026616, "grad_norm": 0.9111691313896259, "learning_rate": 2.5038791277838558e-06, "loss": 0.3241, "step": 19569 }, { "epoch": 1.35292084341514, "grad_norm": 0.6739836375310219, "learning_rate": 2.5033941272515824e-06, "loss": 0.3464, "step": 19570 }, { "epoch": 1.352989975803664, "grad_norm": 0.6492437102362246, "learning_rate": 2.5029091580098093e-06, "loss": 0.2469, "step": 19571 }, { "epoch": 1.353059108192188, "grad_norm": 0.7330729606147928, "learning_rate": 2.5024242200646122e-06, "loss": 0.2526, "step": 19572 }, { "epoch": 1.3531282405807121, "grad_norm": 0.6173046184030087, "learning_rate": 2.5019393134220697e-06, "loss": 0.309, "step": 19573 }, { "epoch": 1.3531973729692361, "grad_norm": 0.681656676589561, "learning_rate": 2.5014544380882606e-06, "loss": 0.2777, "step": 19574 }, { "epoch": 1.3532665053577602, "grad_norm": 0.7113979538693005, "learning_rate": 2.500969594069263e-06, "loss": 0.2391, "step": 19575 }, { "epoch": 1.3533356377462842, "grad_norm": 0.878133499224185, "learning_rate": 2.5004847813711497e-06, "loss": 0.2715, "step": 19576 }, { "epoch": 1.3534047701348082, "grad_norm": 0.7752000817582089, "learning_rate": 2.5000000000000015e-06, "loss": 0.3275, "step": 19577 }, { "epoch": 1.3534739025233322, "grad_norm": 0.7261580014699083, "learning_rate": 2.499515249961889e-06, "loss": 0.2642, "step": 19578 }, { "epoch": 1.3535430349118562, "grad_norm": 0.6495200084994034, "learning_rate": 2.4990305312628932e-06, "loss": 0.2284, "step": 19579 }, { "epoch": 1.3536121673003803, "grad_norm": 0.6886965257800817, "learning_rate": 2.4985458439090843e-06, "loss": 0.3893, "step": 19580 }, { "epoch": 1.3536812996889043, "grad_norm": 0.7345765296951348, "learning_rate": 2.4980611879065397e-06, "loss": 0.3676, "step": 19581 }, { "epoch": 1.3537504320774283, "grad_norm": 0.8911613557233709, "learning_rate": 2.4975765632613323e-06, "loss": 0.2321, "step": 19582 }, { "epoch": 1.3538195644659523, "grad_norm": 0.7370697138920252, "learning_rate": 2.49709196997954e-06, "loss": 0.295, "step": 19583 }, { "epoch": 1.3538886968544763, "grad_norm": 0.7322700896769218, "learning_rate": 2.4966074080672304e-06, "loss": 0.3811, "step": 19584 }, { "epoch": 1.3539578292430003, "grad_norm": 0.643423943838944, "learning_rate": 2.4961228775304815e-06, "loss": 0.288, "step": 19585 }, { "epoch": 1.3540269616315244, "grad_norm": 0.6850949297912714, "learning_rate": 2.495638378375362e-06, "loss": 0.2069, "step": 19586 }, { "epoch": 1.3540960940200484, "grad_norm": 0.7176052756774398, "learning_rate": 2.4951539106079487e-06, "loss": 0.3561, "step": 19587 }, { "epoch": 1.3541652264085724, "grad_norm": 0.6880168850861533, "learning_rate": 2.4946694742343084e-06, "loss": 0.2972, "step": 19588 }, { "epoch": 1.3542343587970964, "grad_norm": 0.6391699011843037, "learning_rate": 2.4941850692605157e-06, "loss": 0.2412, "step": 19589 }, { "epoch": 1.3543034911856204, "grad_norm": 0.7267664812184857, "learning_rate": 2.4937006956926413e-06, "loss": 0.2309, "step": 19590 }, { "epoch": 1.3543726235741445, "grad_norm": 0.7178288920022653, "learning_rate": 2.4932163535367576e-06, "loss": 0.3818, "step": 19591 }, { "epoch": 1.3544417559626685, "grad_norm": 0.7330600179737683, "learning_rate": 2.4927320427989315e-06, "loss": 0.2103, "step": 19592 }, { "epoch": 1.3545108883511925, "grad_norm": 0.6763463132026244, "learning_rate": 2.492247763485237e-06, "loss": 0.266, "step": 19593 }, { "epoch": 1.3545800207397165, "grad_norm": 0.7379869517118692, "learning_rate": 2.4917635156017393e-06, "loss": 0.3411, "step": 19594 }, { "epoch": 1.3546491531282405, "grad_norm": 0.6808231183288199, "learning_rate": 2.49127929915451e-06, "loss": 0.3131, "step": 19595 }, { "epoch": 1.3547182855167645, "grad_norm": 0.7474306209726458, "learning_rate": 2.490795114149617e-06, "loss": 0.2784, "step": 19596 }, { "epoch": 1.3547874179052886, "grad_norm": 0.6315160085440741, "learning_rate": 2.4903109605931323e-06, "loss": 0.2397, "step": 19597 }, { "epoch": 1.3548565502938126, "grad_norm": 0.7414761848045127, "learning_rate": 2.489826838491119e-06, "loss": 0.3538, "step": 19598 }, { "epoch": 1.3549256826823366, "grad_norm": 0.6869369724231147, "learning_rate": 2.489342747849648e-06, "loss": 0.2493, "step": 19599 }, { "epoch": 1.3549948150708606, "grad_norm": 0.7904007498902343, "learning_rate": 2.4888586886747834e-06, "loss": 0.1991, "step": 19600 }, { "epoch": 1.3550639474593846, "grad_norm": 0.6668799714211342, "learning_rate": 2.4883746609725956e-06, "loss": 0.3286, "step": 19601 }, { "epoch": 1.3551330798479087, "grad_norm": 0.6799924179367728, "learning_rate": 2.487890664749148e-06, "loss": 0.2916, "step": 19602 }, { "epoch": 1.3552022122364327, "grad_norm": 0.608810739970884, "learning_rate": 2.4874067000105073e-06, "loss": 0.277, "step": 19603 }, { "epoch": 1.3552713446249567, "grad_norm": 0.7996253093423662, "learning_rate": 2.4869227667627393e-06, "loss": 0.2659, "step": 19604 }, { "epoch": 1.3553404770134807, "grad_norm": 0.7338607485854723, "learning_rate": 2.486438865011912e-06, "loss": 0.3172, "step": 19605 }, { "epoch": 1.3554096094020047, "grad_norm": 0.7171379210000306, "learning_rate": 2.485954994764086e-06, "loss": 0.2533, "step": 19606 }, { "epoch": 1.3554787417905287, "grad_norm": 0.6510485484786944, "learning_rate": 2.48547115602533e-06, "loss": 0.2006, "step": 19607 }, { "epoch": 1.3555478741790528, "grad_norm": 0.7119463729458585, "learning_rate": 2.484987348801703e-06, "loss": 0.2439, "step": 19608 }, { "epoch": 1.3556170065675768, "grad_norm": 0.8581801000545453, "learning_rate": 2.4845035730992717e-06, "loss": 0.3062, "step": 19609 }, { "epoch": 1.3556861389561008, "grad_norm": 0.6508677701952309, "learning_rate": 2.484019828924099e-06, "loss": 0.2575, "step": 19610 }, { "epoch": 1.355755271344625, "grad_norm": 0.5887236418889525, "learning_rate": 2.4835361162822498e-06, "loss": 0.1844, "step": 19611 }, { "epoch": 1.355824403733149, "grad_norm": 0.7341121719941323, "learning_rate": 2.483052435179782e-06, "loss": 0.3701, "step": 19612 }, { "epoch": 1.355893536121673, "grad_norm": 0.6777160422236, "learning_rate": 2.4825687856227627e-06, "loss": 0.2996, "step": 19613 }, { "epoch": 1.355962668510197, "grad_norm": 0.6414659622576261, "learning_rate": 2.482085167617249e-06, "loss": 0.2347, "step": 19614 }, { "epoch": 1.356031800898721, "grad_norm": 0.7458225253244036, "learning_rate": 2.481601581169306e-06, "loss": 0.2557, "step": 19615 }, { "epoch": 1.3561009332872451, "grad_norm": 0.6691782371961283, "learning_rate": 2.4811180262849917e-06, "loss": 0.3933, "step": 19616 }, { "epoch": 1.3561700656757691, "grad_norm": 0.9833666435935974, "learning_rate": 2.4806345029703677e-06, "loss": 0.2312, "step": 19617 }, { "epoch": 1.3562391980642932, "grad_norm": 0.6782620160626751, "learning_rate": 2.4801510112314935e-06, "loss": 0.2493, "step": 19618 }, { "epoch": 1.3563083304528172, "grad_norm": 0.746939623270187, "learning_rate": 2.4796675510744316e-06, "loss": 0.3406, "step": 19619 }, { "epoch": 1.3563774628413412, "grad_norm": 0.8080178115187134, "learning_rate": 2.4791841225052375e-06, "loss": 0.3592, "step": 19620 }, { "epoch": 1.3564465952298652, "grad_norm": 0.7348413506540464, "learning_rate": 2.4787007255299738e-06, "loss": 0.2124, "step": 19621 }, { "epoch": 1.3565157276183892, "grad_norm": 0.701384689204881, "learning_rate": 2.478217360154697e-06, "loss": 0.2876, "step": 19622 }, { "epoch": 1.3565848600069133, "grad_norm": 0.6673046977311022, "learning_rate": 2.477734026385464e-06, "loss": 0.363, "step": 19623 }, { "epoch": 1.3566539923954373, "grad_norm": 0.6291269190172931, "learning_rate": 2.477250724228333e-06, "loss": 0.2708, "step": 19624 }, { "epoch": 1.3567231247839613, "grad_norm": 0.5989812201169626, "learning_rate": 2.476767453689362e-06, "loss": 0.2061, "step": 19625 }, { "epoch": 1.3567922571724853, "grad_norm": 0.6602031580169133, "learning_rate": 2.4762842147746104e-06, "loss": 0.2621, "step": 19626 }, { "epoch": 1.3568613895610093, "grad_norm": 0.6732652339764763, "learning_rate": 2.47580100749013e-06, "loss": 0.3421, "step": 19627 }, { "epoch": 1.3569305219495333, "grad_norm": 0.6744264518704663, "learning_rate": 2.475317831841982e-06, "loss": 0.31, "step": 19628 }, { "epoch": 1.3569996543380574, "grad_norm": 0.7225969994937183, "learning_rate": 2.474834687836217e-06, "loss": 0.2513, "step": 19629 }, { "epoch": 1.3570687867265814, "grad_norm": 0.6720930215986257, "learning_rate": 2.4743515754788927e-06, "loss": 0.3393, "step": 19630 }, { "epoch": 1.3571379191151054, "grad_norm": 0.6935876338020485, "learning_rate": 2.4738684947760643e-06, "loss": 0.2681, "step": 19631 }, { "epoch": 1.3572070515036294, "grad_norm": 0.5859291836043433, "learning_rate": 2.4733854457337884e-06, "loss": 0.244, "step": 19632 }, { "epoch": 1.3572761838921534, "grad_norm": 0.690925159605895, "learning_rate": 2.472902428358115e-06, "loss": 0.2707, "step": 19633 }, { "epoch": 1.3573453162806774, "grad_norm": 0.7181212201611373, "learning_rate": 2.472419442655102e-06, "loss": 0.3747, "step": 19634 }, { "epoch": 1.3574144486692015, "grad_norm": 0.6789936643584814, "learning_rate": 2.4719364886307984e-06, "loss": 0.2571, "step": 19635 }, { "epoch": 1.3574835810577255, "grad_norm": 0.6718199788333004, "learning_rate": 2.471453566291261e-06, "loss": 0.217, "step": 19636 }, { "epoch": 1.3575527134462495, "grad_norm": 0.7752225931252075, "learning_rate": 2.4709706756425395e-06, "loss": 0.3608, "step": 19637 }, { "epoch": 1.3576218458347735, "grad_norm": 0.6838858135574329, "learning_rate": 2.4704878166906865e-06, "loss": 0.3169, "step": 19638 }, { "epoch": 1.3576909782232975, "grad_norm": 0.8956056427534366, "learning_rate": 2.4700049894417555e-06, "loss": 0.2787, "step": 19639 }, { "epoch": 1.3577601106118216, "grad_norm": 0.7260599477541074, "learning_rate": 2.4695221939017985e-06, "loss": 0.3575, "step": 19640 }, { "epoch": 1.3578292430003458, "grad_norm": 0.7048324539920566, "learning_rate": 2.469039430076863e-06, "loss": 0.3545, "step": 19641 }, { "epoch": 1.3578983753888698, "grad_norm": 0.7709138705741668, "learning_rate": 2.468556697973003e-06, "loss": 0.2368, "step": 19642 }, { "epoch": 1.3579675077773938, "grad_norm": 1.0178142406673494, "learning_rate": 2.468073997596266e-06, "loss": 0.2569, "step": 19643 }, { "epoch": 1.3580366401659179, "grad_norm": 0.6689388773715296, "learning_rate": 2.467591328952703e-06, "loss": 0.3317, "step": 19644 }, { "epoch": 1.3581057725544419, "grad_norm": 0.6228395090271466, "learning_rate": 2.4671086920483633e-06, "loss": 0.2611, "step": 19645 }, { "epoch": 1.3581749049429659, "grad_norm": 0.6910686187770798, "learning_rate": 2.4666260868892976e-06, "loss": 0.2533, "step": 19646 }, { "epoch": 1.35824403733149, "grad_norm": 0.663847726540644, "learning_rate": 2.4661435134815515e-06, "loss": 0.3238, "step": 19647 }, { "epoch": 1.358313169720014, "grad_norm": 0.692131509024596, "learning_rate": 2.465660971831176e-06, "loss": 0.3817, "step": 19648 }, { "epoch": 1.358382302108538, "grad_norm": 0.6957609925209789, "learning_rate": 2.465178461944216e-06, "loss": 0.2359, "step": 19649 }, { "epoch": 1.358451434497062, "grad_norm": 0.6758236082235977, "learning_rate": 2.464695983826722e-06, "loss": 0.2819, "step": 19650 }, { "epoch": 1.358520566885586, "grad_norm": 0.6623494245822308, "learning_rate": 2.4642135374847374e-06, "loss": 0.2979, "step": 19651 }, { "epoch": 1.35858969927411, "grad_norm": 0.7680759758625786, "learning_rate": 2.463731122924311e-06, "loss": 0.3109, "step": 19652 }, { "epoch": 1.358658831662634, "grad_norm": 0.713258777724015, "learning_rate": 2.4632487401514886e-06, "loss": 0.2527, "step": 19653 }, { "epoch": 1.358727964051158, "grad_norm": 0.6335221263984027, "learning_rate": 2.4627663891723186e-06, "loss": 0.2566, "step": 19654 }, { "epoch": 1.358797096439682, "grad_norm": 0.7037326975711653, "learning_rate": 2.4622840699928415e-06, "loss": 0.3302, "step": 19655 }, { "epoch": 1.358866228828206, "grad_norm": 0.6783448232262184, "learning_rate": 2.461801782619107e-06, "loss": 0.3316, "step": 19656 }, { "epoch": 1.35893536121673, "grad_norm": 0.6664139836513738, "learning_rate": 2.4613195270571555e-06, "loss": 0.2383, "step": 19657 }, { "epoch": 1.359004493605254, "grad_norm": 0.6798311418429485, "learning_rate": 2.4608373033130355e-06, "loss": 0.2638, "step": 19658 }, { "epoch": 1.3590736259937781, "grad_norm": 0.7078878725802124, "learning_rate": 2.4603551113927866e-06, "loss": 0.3743, "step": 19659 }, { "epoch": 1.3591427583823021, "grad_norm": 0.6970330257982721, "learning_rate": 2.459872951302454e-06, "loss": 0.2415, "step": 19660 }, { "epoch": 1.3592118907708262, "grad_norm": 0.7358535285928024, "learning_rate": 2.4593908230480805e-06, "loss": 0.2521, "step": 19661 }, { "epoch": 1.3592810231593502, "grad_norm": 0.6422931425523107, "learning_rate": 2.4589087266357113e-06, "loss": 0.3351, "step": 19662 }, { "epoch": 1.3593501555478742, "grad_norm": 0.7294359614417893, "learning_rate": 2.458426662071385e-06, "loss": 0.2708, "step": 19663 }, { "epoch": 1.3594192879363982, "grad_norm": 0.7453783825196878, "learning_rate": 2.4579446293611463e-06, "loss": 0.1834, "step": 19664 }, { "epoch": 1.3594884203249222, "grad_norm": 0.6887789589716616, "learning_rate": 2.4574626285110338e-06, "loss": 0.2856, "step": 19665 }, { "epoch": 1.3595575527134462, "grad_norm": 0.6668187082845761, "learning_rate": 2.456980659527089e-06, "loss": 0.3567, "step": 19666 }, { "epoch": 1.3596266851019703, "grad_norm": 0.6781039088357923, "learning_rate": 2.456498722415354e-06, "loss": 0.2731, "step": 19667 }, { "epoch": 1.3596958174904943, "grad_norm": 0.8209366739870289, "learning_rate": 2.4560168171818705e-06, "loss": 0.2772, "step": 19668 }, { "epoch": 1.3597649498790183, "grad_norm": 0.730102450931341, "learning_rate": 2.4555349438326735e-06, "loss": 0.3187, "step": 19669 }, { "epoch": 1.3598340822675423, "grad_norm": 0.7447276476921416, "learning_rate": 2.455053102373808e-06, "loss": 0.2524, "step": 19670 }, { "epoch": 1.3599032146560663, "grad_norm": 0.6702494898383483, "learning_rate": 2.454571292811308e-06, "loss": 0.2324, "step": 19671 }, { "epoch": 1.3599723470445904, "grad_norm": 0.6290688697096604, "learning_rate": 2.4540895151512163e-06, "loss": 0.2265, "step": 19672 }, { "epoch": 1.3600414794331144, "grad_norm": 0.6787827477706772, "learning_rate": 2.453607769399567e-06, "loss": 0.347, "step": 19673 }, { "epoch": 1.3601106118216384, "grad_norm": 0.7395180051424988, "learning_rate": 2.4531260555624003e-06, "loss": 0.3062, "step": 19674 }, { "epoch": 1.3601797442101624, "grad_norm": 0.6612521505654548, "learning_rate": 2.4526443736457533e-06, "loss": 0.2457, "step": 19675 }, { "epoch": 1.3602488765986864, "grad_norm": 0.7177879931222219, "learning_rate": 2.452162723655665e-06, "loss": 0.2932, "step": 19676 }, { "epoch": 1.3603180089872104, "grad_norm": 0.6930284626947393, "learning_rate": 2.451681105598168e-06, "loss": 0.404, "step": 19677 }, { "epoch": 1.3603871413757345, "grad_norm": 0.6621921806652229, "learning_rate": 2.4511995194793033e-06, "loss": 0.2405, "step": 19678 }, { "epoch": 1.3604562737642585, "grad_norm": 0.6672432922975287, "learning_rate": 2.450717965305101e-06, "loss": 0.3055, "step": 19679 }, { "epoch": 1.3605254061527825, "grad_norm": 0.6990847310914275, "learning_rate": 2.4502364430816006e-06, "loss": 0.3185, "step": 19680 }, { "epoch": 1.3605945385413065, "grad_norm": 0.6511943890956244, "learning_rate": 2.449754952814836e-06, "loss": 0.2831, "step": 19681 }, { "epoch": 1.3606636709298305, "grad_norm": 0.8000652993692616, "learning_rate": 2.4492734945108436e-06, "loss": 0.1944, "step": 19682 }, { "epoch": 1.3607328033183546, "grad_norm": 0.6743798361455728, "learning_rate": 2.4487920681756537e-06, "loss": 0.2785, "step": 19683 }, { "epoch": 1.3608019357068786, "grad_norm": 0.6606291965952905, "learning_rate": 2.448310673815305e-06, "loss": 0.3371, "step": 19684 }, { "epoch": 1.3608710680954026, "grad_norm": 1.1147166720334873, "learning_rate": 2.4478293114358258e-06, "loss": 0.2555, "step": 19685 }, { "epoch": 1.3609402004839266, "grad_norm": 0.6431279480574331, "learning_rate": 2.4473479810432543e-06, "loss": 0.2229, "step": 19686 }, { "epoch": 1.3610093328724506, "grad_norm": 0.7398198353912151, "learning_rate": 2.4468666826436183e-06, "loss": 0.3264, "step": 19687 }, { "epoch": 1.3610784652609746, "grad_norm": 0.6507460238348611, "learning_rate": 2.446385416242952e-06, "loss": 0.2404, "step": 19688 }, { "epoch": 1.3611475976494987, "grad_norm": 0.5788350944466492, "learning_rate": 2.445904181847289e-06, "loss": 0.2082, "step": 19689 }, { "epoch": 1.3612167300380227, "grad_norm": 0.6636822052230924, "learning_rate": 2.445422979462657e-06, "loss": 0.2498, "step": 19690 }, { "epoch": 1.3612858624265467, "grad_norm": 0.6666905500934441, "learning_rate": 2.444941809095092e-06, "loss": 0.3524, "step": 19691 }, { "epoch": 1.361354994815071, "grad_norm": 0.6593757045724784, "learning_rate": 2.444460670750619e-06, "loss": 0.2377, "step": 19692 }, { "epoch": 1.361424127203595, "grad_norm": 0.6265468908516358, "learning_rate": 2.443979564435273e-06, "loss": 0.2427, "step": 19693 }, { "epoch": 1.361493259592119, "grad_norm": 0.7985184863415722, "learning_rate": 2.44349849015508e-06, "loss": 0.2986, "step": 19694 }, { "epoch": 1.361562391980643, "grad_norm": 0.6877001821880305, "learning_rate": 2.4430174479160712e-06, "loss": 0.2867, "step": 19695 }, { "epoch": 1.361631524369167, "grad_norm": 0.6950769464147877, "learning_rate": 2.4425364377242757e-06, "loss": 0.2568, "step": 19696 }, { "epoch": 1.361700656757691, "grad_norm": 0.6956357857464863, "learning_rate": 2.442055459585723e-06, "loss": 0.2221, "step": 19697 }, { "epoch": 1.361769789146215, "grad_norm": 0.7198962863843947, "learning_rate": 2.4415745135064395e-06, "loss": 0.3207, "step": 19698 }, { "epoch": 1.361838921534739, "grad_norm": 0.6618686115475996, "learning_rate": 2.441093599492455e-06, "loss": 0.2703, "step": 19699 }, { "epoch": 1.361908053923263, "grad_norm": 0.6167812027569578, "learning_rate": 2.440612717549794e-06, "loss": 0.2474, "step": 19700 }, { "epoch": 1.361977186311787, "grad_norm": 0.7448542344435074, "learning_rate": 2.440131867684485e-06, "loss": 0.3284, "step": 19701 }, { "epoch": 1.3620463187003111, "grad_norm": 0.7174395105622128, "learning_rate": 2.4396510499025546e-06, "loss": 0.2963, "step": 19702 }, { "epoch": 1.3621154510888351, "grad_norm": 0.7852546286336805, "learning_rate": 2.439170264210031e-06, "loss": 0.3123, "step": 19703 }, { "epoch": 1.3621845834773592, "grad_norm": 0.7180560716013317, "learning_rate": 2.4386895106129367e-06, "loss": 0.2987, "step": 19704 }, { "epoch": 1.3622537158658832, "grad_norm": 0.6510044816807097, "learning_rate": 2.4382087891173e-06, "loss": 0.3117, "step": 19705 }, { "epoch": 1.3623228482544072, "grad_norm": 0.6343227126104348, "learning_rate": 2.4377280997291424e-06, "loss": 0.305, "step": 19706 }, { "epoch": 1.3623919806429312, "grad_norm": 0.6473791314811682, "learning_rate": 2.437247442454493e-06, "loss": 0.2776, "step": 19707 }, { "epoch": 1.3624611130314552, "grad_norm": 0.6996538362920011, "learning_rate": 2.436766817299371e-06, "loss": 0.2673, "step": 19708 }, { "epoch": 1.3625302454199792, "grad_norm": 1.4514582697949607, "learning_rate": 2.436286224269803e-06, "loss": 0.3689, "step": 19709 }, { "epoch": 1.3625993778085033, "grad_norm": 0.6161017102101763, "learning_rate": 2.435805663371812e-06, "loss": 0.2438, "step": 19710 }, { "epoch": 1.3626685101970273, "grad_norm": 0.6531921942180322, "learning_rate": 2.435325134611423e-06, "loss": 0.1821, "step": 19711 }, { "epoch": 1.3627376425855513, "grad_norm": 0.711716881720561, "learning_rate": 2.434844637994654e-06, "loss": 0.3567, "step": 19712 }, { "epoch": 1.3628067749740753, "grad_norm": 0.7293206787029171, "learning_rate": 2.434364173527532e-06, "loss": 0.2695, "step": 19713 }, { "epoch": 1.3628759073625993, "grad_norm": 0.6360285239401059, "learning_rate": 2.433883741216075e-06, "loss": 0.2079, "step": 19714 }, { "epoch": 1.3629450397511234, "grad_norm": 0.7080794002315056, "learning_rate": 2.4334033410663053e-06, "loss": 0.2604, "step": 19715 }, { "epoch": 1.3630141721396474, "grad_norm": 0.6638333680685264, "learning_rate": 2.4329229730842445e-06, "loss": 0.409, "step": 19716 }, { "epoch": 1.3630833045281714, "grad_norm": 0.601347378048484, "learning_rate": 2.4324426372759157e-06, "loss": 0.2161, "step": 19717 }, { "epoch": 1.3631524369166954, "grad_norm": 0.6854037118425427, "learning_rate": 2.4319623336473337e-06, "loss": 0.2288, "step": 19718 }, { "epoch": 1.3632215693052194, "grad_norm": 0.669736194923801, "learning_rate": 2.431482062204523e-06, "loss": 0.2903, "step": 19719 }, { "epoch": 1.3632907016937434, "grad_norm": 0.7323699500666518, "learning_rate": 2.431001822953499e-06, "loss": 0.3401, "step": 19720 }, { "epoch": 1.3633598340822675, "grad_norm": 0.671891085182419, "learning_rate": 2.4305216159002855e-06, "loss": 0.2213, "step": 19721 }, { "epoch": 1.3634289664707917, "grad_norm": 0.6339084177503908, "learning_rate": 2.4300414410508947e-06, "loss": 0.2659, "step": 19722 }, { "epoch": 1.3634980988593157, "grad_norm": 0.7254057488523442, "learning_rate": 2.4295612984113493e-06, "loss": 0.3555, "step": 19723 }, { "epoch": 1.3635672312478397, "grad_norm": 0.8609257017558515, "learning_rate": 2.429081187987665e-06, "loss": 0.2827, "step": 19724 }, { "epoch": 1.3636363636363638, "grad_norm": 0.6438276838972096, "learning_rate": 2.4286011097858624e-06, "loss": 0.2269, "step": 19725 }, { "epoch": 1.3637054960248878, "grad_norm": 0.6929112473147527, "learning_rate": 2.428121063811954e-06, "loss": 0.2829, "step": 19726 }, { "epoch": 1.3637746284134118, "grad_norm": 0.6715658628680067, "learning_rate": 2.4276410500719604e-06, "loss": 0.3428, "step": 19727 }, { "epoch": 1.3638437608019358, "grad_norm": 0.6116812941293664, "learning_rate": 2.4271610685718934e-06, "loss": 0.2232, "step": 19728 }, { "epoch": 1.3639128931904598, "grad_norm": 0.6829040737624058, "learning_rate": 2.426681119317773e-06, "loss": 0.2771, "step": 19729 }, { "epoch": 1.3639820255789838, "grad_norm": 0.6820514395937237, "learning_rate": 2.4262012023156105e-06, "loss": 0.3679, "step": 19730 }, { "epoch": 1.3640511579675079, "grad_norm": 0.8072956969078235, "learning_rate": 2.4257213175714233e-06, "loss": 0.2834, "step": 19731 }, { "epoch": 1.3641202903560319, "grad_norm": 0.6880575267063415, "learning_rate": 2.4252414650912247e-06, "loss": 0.2307, "step": 19732 }, { "epoch": 1.364189422744556, "grad_norm": 0.7163086445552381, "learning_rate": 2.424761644881032e-06, "loss": 0.3455, "step": 19733 }, { "epoch": 1.36425855513308, "grad_norm": 0.6971686870551819, "learning_rate": 2.424281856946854e-06, "loss": 0.3208, "step": 19734 }, { "epoch": 1.364327687521604, "grad_norm": 0.649600652678354, "learning_rate": 2.4238021012947085e-06, "loss": 0.2518, "step": 19735 }, { "epoch": 1.364396819910128, "grad_norm": 0.7124885858373893, "learning_rate": 2.423322377930604e-06, "loss": 0.2478, "step": 19736 }, { "epoch": 1.364465952298652, "grad_norm": 0.6569359919804837, "learning_rate": 2.4228426868605554e-06, "loss": 0.3071, "step": 19737 }, { "epoch": 1.364535084687176, "grad_norm": 1.0333890208565937, "learning_rate": 2.4223630280905745e-06, "loss": 0.2962, "step": 19738 }, { "epoch": 1.3646042170757, "grad_norm": 1.3724267222544337, "learning_rate": 2.421883401626675e-06, "loss": 0.2493, "step": 19739 }, { "epoch": 1.364673349464224, "grad_norm": 0.6115471299907739, "learning_rate": 2.4214038074748646e-06, "loss": 0.2506, "step": 19740 }, { "epoch": 1.364742481852748, "grad_norm": 0.7622107295085637, "learning_rate": 2.4209242456411575e-06, "loss": 0.4102, "step": 19741 }, { "epoch": 1.364811614241272, "grad_norm": 0.6800371009790749, "learning_rate": 2.4204447161315605e-06, "loss": 0.2569, "step": 19742 }, { "epoch": 1.364880746629796, "grad_norm": 0.6332815165346152, "learning_rate": 2.4199652189520874e-06, "loss": 0.2179, "step": 19743 }, { "epoch": 1.36494987901832, "grad_norm": 0.7153089443077755, "learning_rate": 2.4194857541087443e-06, "loss": 0.3515, "step": 19744 }, { "epoch": 1.3650190114068441, "grad_norm": 0.6887743685606464, "learning_rate": 2.4190063216075416e-06, "loss": 0.2904, "step": 19745 }, { "epoch": 1.3650881437953681, "grad_norm": 0.635613277648579, "learning_rate": 2.4185269214544886e-06, "loss": 0.2044, "step": 19746 }, { "epoch": 1.3651572761838922, "grad_norm": 0.6496891259244513, "learning_rate": 2.4180475536555963e-06, "loss": 0.2486, "step": 19747 }, { "epoch": 1.3652264085724162, "grad_norm": 0.6399487256774156, "learning_rate": 2.417568218216868e-06, "loss": 0.3358, "step": 19748 }, { "epoch": 1.3652955409609402, "grad_norm": 0.7230132196359803, "learning_rate": 2.4170889151443156e-06, "loss": 0.2859, "step": 19749 }, { "epoch": 1.3653646733494642, "grad_norm": 0.6219032319928921, "learning_rate": 2.4166096444439424e-06, "loss": 0.2413, "step": 19750 }, { "epoch": 1.3654338057379882, "grad_norm": 0.744709638281111, "learning_rate": 2.4161304061217568e-06, "loss": 0.2886, "step": 19751 }, { "epoch": 1.3655029381265122, "grad_norm": 0.661972506154893, "learning_rate": 2.4156512001837655e-06, "loss": 0.3234, "step": 19752 }, { "epoch": 1.3655720705150363, "grad_norm": 0.674938237738495, "learning_rate": 2.415172026635976e-06, "loss": 0.2594, "step": 19753 }, { "epoch": 1.3656412029035603, "grad_norm": 0.768352534184806, "learning_rate": 2.414692885484391e-06, "loss": 0.2709, "step": 19754 }, { "epoch": 1.3657103352920843, "grad_norm": 0.6989460176975705, "learning_rate": 2.414213776735019e-06, "loss": 0.3321, "step": 19755 }, { "epoch": 1.3657794676806083, "grad_norm": 0.6491520752023962, "learning_rate": 2.4137347003938622e-06, "loss": 0.2833, "step": 19756 }, { "epoch": 1.3658486000691323, "grad_norm": 0.6451955101389415, "learning_rate": 2.413255656466924e-06, "loss": 0.1995, "step": 19757 }, { "epoch": 1.3659177324576564, "grad_norm": 0.6659205313526431, "learning_rate": 2.4127766449602097e-06, "loss": 0.2906, "step": 19758 }, { "epoch": 1.3659868648461804, "grad_norm": 0.6570796138453139, "learning_rate": 2.412297665879723e-06, "loss": 0.382, "step": 19759 }, { "epoch": 1.3660559972347044, "grad_norm": 0.693833248173277, "learning_rate": 2.411818719231469e-06, "loss": 0.2156, "step": 19760 }, { "epoch": 1.3661251296232284, "grad_norm": 0.7441610894646354, "learning_rate": 2.411339805021447e-06, "loss": 0.2118, "step": 19761 }, { "epoch": 1.3661942620117524, "grad_norm": 0.6840969694809901, "learning_rate": 2.4108609232556624e-06, "loss": 0.3382, "step": 19762 }, { "epoch": 1.3662633944002764, "grad_norm": 0.7030222458154516, "learning_rate": 2.4103820739401133e-06, "loss": 0.3309, "step": 19763 }, { "epoch": 1.3663325267888005, "grad_norm": 0.6519536091624809, "learning_rate": 2.4099032570808064e-06, "loss": 0.2777, "step": 19764 }, { "epoch": 1.3664016591773245, "grad_norm": 0.6930035081885925, "learning_rate": 2.4094244726837373e-06, "loss": 0.31, "step": 19765 }, { "epoch": 1.3664707915658485, "grad_norm": 0.6739467234029315, "learning_rate": 2.40894572075491e-06, "loss": 0.3706, "step": 19766 }, { "epoch": 1.3665399239543725, "grad_norm": 0.6603087632555312, "learning_rate": 2.4084670013003235e-06, "loss": 0.2639, "step": 19767 }, { "epoch": 1.3666090563428965, "grad_norm": 0.627011192031872, "learning_rate": 2.407988314325981e-06, "loss": 0.2366, "step": 19768 }, { "epoch": 1.3666781887314206, "grad_norm": 0.6443588923048849, "learning_rate": 2.4075096598378768e-06, "loss": 0.2663, "step": 19769 }, { "epoch": 1.3667473211199446, "grad_norm": 0.6586736886661632, "learning_rate": 2.407031037842014e-06, "loss": 0.2958, "step": 19770 }, { "epoch": 1.3668164535084686, "grad_norm": 0.6597894367616509, "learning_rate": 2.4065524483443886e-06, "loss": 0.2535, "step": 19771 }, { "epoch": 1.3668855858969926, "grad_norm": 0.7216113594034368, "learning_rate": 2.4060738913510003e-06, "loss": 0.2552, "step": 19772 }, { "epoch": 1.3669547182855168, "grad_norm": 0.6797725549153085, "learning_rate": 2.405595366867846e-06, "loss": 0.3399, "step": 19773 }, { "epoch": 1.3670238506740409, "grad_norm": 0.6839192233301268, "learning_rate": 2.405116874900926e-06, "loss": 0.2811, "step": 19774 }, { "epoch": 1.3670929830625649, "grad_norm": 0.6344918455412473, "learning_rate": 2.404638415456233e-06, "loss": 0.2413, "step": 19775 }, { "epoch": 1.367162115451089, "grad_norm": 0.632488579736002, "learning_rate": 2.4041599885397683e-06, "loss": 0.2757, "step": 19776 }, { "epoch": 1.367231247839613, "grad_norm": 0.6625114797996516, "learning_rate": 2.4036815941575235e-06, "loss": 0.2631, "step": 19777 }, { "epoch": 1.367300380228137, "grad_norm": 0.642187376577321, "learning_rate": 2.4032032323154986e-06, "loss": 0.2166, "step": 19778 }, { "epoch": 1.367369512616661, "grad_norm": 0.7069624016075511, "learning_rate": 2.4027249030196847e-06, "loss": 0.258, "step": 19779 }, { "epoch": 1.367438645005185, "grad_norm": 0.6816455672570315, "learning_rate": 2.4022466062760797e-06, "loss": 0.3255, "step": 19780 }, { "epoch": 1.367507777393709, "grad_norm": 0.7062746347730131, "learning_rate": 2.401768342090677e-06, "loss": 0.298, "step": 19781 }, { "epoch": 1.367576909782233, "grad_norm": 0.6050359101781426, "learning_rate": 2.4012901104694736e-06, "loss": 0.2445, "step": 19782 }, { "epoch": 1.367646042170757, "grad_norm": 0.6643537394104388, "learning_rate": 2.4008119114184593e-06, "loss": 0.2447, "step": 19783 }, { "epoch": 1.367715174559281, "grad_norm": 0.6860494889564855, "learning_rate": 2.400333744943631e-06, "loss": 0.316, "step": 19784 }, { "epoch": 1.367784306947805, "grad_norm": 0.6661241612211295, "learning_rate": 2.3998556110509775e-06, "loss": 0.267, "step": 19785 }, { "epoch": 1.367853439336329, "grad_norm": 0.6130898917961347, "learning_rate": 2.3993775097464943e-06, "loss": 0.2163, "step": 19786 }, { "epoch": 1.367922571724853, "grad_norm": 0.6693084760070167, "learning_rate": 2.3988994410361725e-06, "loss": 0.3949, "step": 19787 }, { "epoch": 1.3679917041133771, "grad_norm": 1.1478076726484812, "learning_rate": 2.398421404926007e-06, "loss": 0.2862, "step": 19788 }, { "epoch": 1.3680608365019011, "grad_norm": 0.6108623329687722, "learning_rate": 2.3979434014219843e-06, "loss": 0.2623, "step": 19789 }, { "epoch": 1.3681299688904252, "grad_norm": 0.6647791942699731, "learning_rate": 2.397465430530099e-06, "loss": 0.2955, "step": 19790 }, { "epoch": 1.3681991012789492, "grad_norm": 0.7109570840879493, "learning_rate": 2.3969874922563385e-06, "loss": 0.4006, "step": 19791 }, { "epoch": 1.3682682336674732, "grad_norm": 0.7485606275761966, "learning_rate": 2.3965095866066963e-06, "loss": 0.2191, "step": 19792 }, { "epoch": 1.3683373660559972, "grad_norm": 0.6486363968367517, "learning_rate": 2.3960317135871588e-06, "loss": 0.2361, "step": 19793 }, { "epoch": 1.3684064984445212, "grad_norm": 0.7452850728847993, "learning_rate": 2.3955538732037163e-06, "loss": 0.3879, "step": 19794 }, { "epoch": 1.3684756308330452, "grad_norm": 0.6597447231102174, "learning_rate": 2.3950760654623585e-06, "loss": 0.291, "step": 19795 }, { "epoch": 1.3685447632215693, "grad_norm": 0.8562368842296674, "learning_rate": 2.394598290369075e-06, "loss": 0.2116, "step": 19796 }, { "epoch": 1.3686138956100933, "grad_norm": 0.6570004625996924, "learning_rate": 2.3941205479298505e-06, "loss": 0.2851, "step": 19797 }, { "epoch": 1.3686830279986173, "grad_norm": 0.7147842297516663, "learning_rate": 2.3936428381506766e-06, "loss": 0.3254, "step": 19798 }, { "epoch": 1.3687521603871413, "grad_norm": 0.6360352140514214, "learning_rate": 2.3931651610375366e-06, "loss": 0.3315, "step": 19799 }, { "epoch": 1.3688212927756653, "grad_norm": 0.624250501079502, "learning_rate": 2.3926875165964213e-06, "loss": 0.3, "step": 19800 }, { "epoch": 1.3688904251641894, "grad_norm": 0.6661456641387741, "learning_rate": 2.3922099048333124e-06, "loss": 0.2682, "step": 19801 }, { "epoch": 1.3689595575527134, "grad_norm": 0.680358606505168, "learning_rate": 2.3917323257541996e-06, "loss": 0.3075, "step": 19802 }, { "epoch": 1.3690286899412374, "grad_norm": 0.6745278409531308, "learning_rate": 2.3912547793650664e-06, "loss": 0.2574, "step": 19803 }, { "epoch": 1.3690978223297616, "grad_norm": 0.7809943383780851, "learning_rate": 2.3907772656719017e-06, "loss": 0.2854, "step": 19804 }, { "epoch": 1.3691669547182856, "grad_norm": 0.7595897013594819, "learning_rate": 2.3902997846806853e-06, "loss": 0.3498, "step": 19805 }, { "epoch": 1.3692360871068097, "grad_norm": 0.8185946065293678, "learning_rate": 2.389822336397406e-06, "loss": 0.2319, "step": 19806 }, { "epoch": 1.3693052194953337, "grad_norm": 0.6229705667610564, "learning_rate": 2.3893449208280434e-06, "loss": 0.2565, "step": 19807 }, { "epoch": 1.3693743518838577, "grad_norm": 0.6559790332086455, "learning_rate": 2.3888675379785835e-06, "loss": 0.2968, "step": 19808 }, { "epoch": 1.3694434842723817, "grad_norm": 0.746968264476762, "learning_rate": 2.38839018785501e-06, "loss": 0.3798, "step": 19809 }, { "epoch": 1.3695126166609057, "grad_norm": 0.9462175478659137, "learning_rate": 2.3879128704633054e-06, "loss": 0.2597, "step": 19810 }, { "epoch": 1.3695817490494298, "grad_norm": 0.722435054931461, "learning_rate": 2.3874355858094505e-06, "loss": 0.2261, "step": 19811 }, { "epoch": 1.3696508814379538, "grad_norm": 0.7608807601285918, "learning_rate": 2.3869583338994297e-06, "loss": 0.308, "step": 19812 }, { "epoch": 1.3697200138264778, "grad_norm": 0.6559282794897028, "learning_rate": 2.3864811147392216e-06, "loss": 0.2713, "step": 19813 }, { "epoch": 1.3697891462150018, "grad_norm": 0.654179081761343, "learning_rate": 2.3860039283348096e-06, "loss": 0.2582, "step": 19814 }, { "epoch": 1.3698582786035258, "grad_norm": 0.6985770151132823, "learning_rate": 2.3855267746921722e-06, "loss": 0.2697, "step": 19815 }, { "epoch": 1.3699274109920498, "grad_norm": 0.6811141238270295, "learning_rate": 2.385049653817291e-06, "loss": 0.3109, "step": 19816 }, { "epoch": 1.3699965433805739, "grad_norm": 0.6386344936557384, "learning_rate": 2.3845725657161455e-06, "loss": 0.2354, "step": 19817 }, { "epoch": 1.3700656757690979, "grad_norm": 0.6637084945699971, "learning_rate": 2.3840955103947176e-06, "loss": 0.2268, "step": 19818 }, { "epoch": 1.370134808157622, "grad_norm": 0.7665330588274745, "learning_rate": 2.3836184878589822e-06, "loss": 0.2944, "step": 19819 }, { "epoch": 1.370203940546146, "grad_norm": 0.6580825706956712, "learning_rate": 2.383141498114922e-06, "loss": 0.3054, "step": 19820 }, { "epoch": 1.37027307293467, "grad_norm": 0.6130086828313084, "learning_rate": 2.3826645411685135e-06, "loss": 0.2136, "step": 19821 }, { "epoch": 1.370342205323194, "grad_norm": 0.6456761258851336, "learning_rate": 2.3821876170257296e-06, "loss": 0.2383, "step": 19822 }, { "epoch": 1.370411337711718, "grad_norm": 0.7369280303388187, "learning_rate": 2.3817107256925564e-06, "loss": 0.3862, "step": 19823 }, { "epoch": 1.370480470100242, "grad_norm": 0.7141721487047302, "learning_rate": 2.3812338671749647e-06, "loss": 0.2998, "step": 19824 }, { "epoch": 1.370549602488766, "grad_norm": 0.6355074530673093, "learning_rate": 2.380757041478935e-06, "loss": 0.2294, "step": 19825 }, { "epoch": 1.37061873487729, "grad_norm": 0.7657688278384284, "learning_rate": 2.380280248610439e-06, "loss": 0.3027, "step": 19826 }, { "epoch": 1.370687867265814, "grad_norm": 0.6815279879484984, "learning_rate": 2.3798034885754577e-06, "loss": 0.2657, "step": 19827 }, { "epoch": 1.370756999654338, "grad_norm": 0.6307220227400246, "learning_rate": 2.379326761379962e-06, "loss": 0.2607, "step": 19828 }, { "epoch": 1.370826132042862, "grad_norm": 0.6282032401360623, "learning_rate": 2.3788500670299282e-06, "loss": 0.2095, "step": 19829 }, { "epoch": 1.370895264431386, "grad_norm": 0.7258925798780207, "learning_rate": 2.3783734055313306e-06, "loss": 0.3747, "step": 19830 }, { "epoch": 1.3709643968199101, "grad_norm": 0.7307221872055373, "learning_rate": 2.3778967768901464e-06, "loss": 0.2842, "step": 19831 }, { "epoch": 1.3710335292084341, "grad_norm": 0.5922476473087405, "learning_rate": 2.3774201811123447e-06, "loss": 0.1911, "step": 19832 }, { "epoch": 1.3711026615969581, "grad_norm": 0.6872297341701319, "learning_rate": 2.376943618203903e-06, "loss": 0.2567, "step": 19833 }, { "epoch": 1.3711717939854822, "grad_norm": 0.706975335424844, "learning_rate": 2.3764670881707894e-06, "loss": 0.3609, "step": 19834 }, { "epoch": 1.3712409263740062, "grad_norm": 0.6752735410634729, "learning_rate": 2.3759905910189812e-06, "loss": 0.3024, "step": 19835 }, { "epoch": 1.3713100587625302, "grad_norm": 0.5970445383340038, "learning_rate": 2.375514126754446e-06, "loss": 0.183, "step": 19836 }, { "epoch": 1.3713791911510542, "grad_norm": 0.6717124478805226, "learning_rate": 2.3750376953831585e-06, "loss": 0.2948, "step": 19837 }, { "epoch": 1.3714483235395782, "grad_norm": 0.6833164651545965, "learning_rate": 2.374561296911089e-06, "loss": 0.2757, "step": 19838 }, { "epoch": 1.3715174559281023, "grad_norm": 0.6516310241435146, "learning_rate": 2.3740849313442096e-06, "loss": 0.2677, "step": 19839 }, { "epoch": 1.3715865883166263, "grad_norm": 0.6668268968731403, "learning_rate": 2.373608598688487e-06, "loss": 0.3137, "step": 19840 }, { "epoch": 1.3716557207051503, "grad_norm": 1.0938081608465033, "learning_rate": 2.373132298949896e-06, "loss": 0.3277, "step": 19841 }, { "epoch": 1.3717248530936743, "grad_norm": 0.6888698543781399, "learning_rate": 2.372656032134402e-06, "loss": 0.2253, "step": 19842 }, { "epoch": 1.3717939854821983, "grad_norm": 0.6293753997318421, "learning_rate": 2.3721797982479766e-06, "loss": 0.2261, "step": 19843 }, { "epoch": 1.3718631178707223, "grad_norm": 0.6793755712324828, "learning_rate": 2.371703597296587e-06, "loss": 0.3534, "step": 19844 }, { "epoch": 1.3719322502592464, "grad_norm": 0.7525151126456351, "learning_rate": 2.371227429286205e-06, "loss": 0.2622, "step": 19845 }, { "epoch": 1.3720013826477704, "grad_norm": 0.7185746180715854, "learning_rate": 2.3707512942227933e-06, "loss": 0.2602, "step": 19846 }, { "epoch": 1.3720705150362944, "grad_norm": 0.8734625894957411, "learning_rate": 2.370275192112324e-06, "loss": 0.2488, "step": 19847 }, { "epoch": 1.3721396474248184, "grad_norm": 0.658026001806067, "learning_rate": 2.369799122960761e-06, "loss": 0.3678, "step": 19848 }, { "epoch": 1.3722087798133424, "grad_norm": 0.6260684844483768, "learning_rate": 2.369323086774073e-06, "loss": 0.3048, "step": 19849 }, { "epoch": 1.3722779122018665, "grad_norm": 0.6889390452156954, "learning_rate": 2.3688470835582247e-06, "loss": 0.2523, "step": 19850 }, { "epoch": 1.3723470445903905, "grad_norm": 0.8559292686507824, "learning_rate": 2.3683711133191817e-06, "loss": 0.2881, "step": 19851 }, { "epoch": 1.3724161769789145, "grad_norm": 0.658451828104012, "learning_rate": 2.367895176062911e-06, "loss": 0.3225, "step": 19852 }, { "epoch": 1.3724853093674385, "grad_norm": 0.7260640511067824, "learning_rate": 2.367419271795379e-06, "loss": 0.2018, "step": 19853 }, { "epoch": 1.3725544417559628, "grad_norm": 0.6672966811535119, "learning_rate": 2.366943400522546e-06, "loss": 0.2476, "step": 19854 }, { "epoch": 1.3726235741444868, "grad_norm": 0.699823521699089, "learning_rate": 2.366467562250381e-06, "loss": 0.417, "step": 19855 }, { "epoch": 1.3726927065330108, "grad_norm": 0.7537306407021652, "learning_rate": 2.3659917569848436e-06, "loss": 0.2644, "step": 19856 }, { "epoch": 1.3727618389215348, "grad_norm": 0.6613414215239037, "learning_rate": 2.3655159847318983e-06, "loss": 0.2673, "step": 19857 }, { "epoch": 1.3728309713100588, "grad_norm": 0.6561091854929306, "learning_rate": 2.365040245497509e-06, "loss": 0.2132, "step": 19858 }, { "epoch": 1.3729001036985828, "grad_norm": 0.6418057043783539, "learning_rate": 2.3645645392876397e-06, "loss": 0.2711, "step": 19859 }, { "epoch": 1.3729692360871069, "grad_norm": 0.6797624457059744, "learning_rate": 2.364088866108249e-06, "loss": 0.2285, "step": 19860 }, { "epoch": 1.3730383684756309, "grad_norm": 0.7169111927266739, "learning_rate": 2.363613225965302e-06, "loss": 0.2288, "step": 19861 }, { "epoch": 1.373107500864155, "grad_norm": 0.7184024472624141, "learning_rate": 2.363137618864757e-06, "loss": 0.2807, "step": 19862 }, { "epoch": 1.373176633252679, "grad_norm": 0.6775973855882323, "learning_rate": 2.3626620448125776e-06, "loss": 0.2758, "step": 19863 }, { "epoch": 1.373245765641203, "grad_norm": 0.8128930064396812, "learning_rate": 2.362186503814721e-06, "loss": 0.2235, "step": 19864 }, { "epoch": 1.373314898029727, "grad_norm": 0.740101592205797, "learning_rate": 2.36171099587715e-06, "loss": 0.302, "step": 19865 }, { "epoch": 1.373384030418251, "grad_norm": 0.7723344401896604, "learning_rate": 2.361235521005823e-06, "loss": 0.3406, "step": 19866 }, { "epoch": 1.373453162806775, "grad_norm": 0.685586835043066, "learning_rate": 2.3607600792067015e-06, "loss": 0.2453, "step": 19867 }, { "epoch": 1.373522295195299, "grad_norm": 0.7152432932790096, "learning_rate": 2.36028467048574e-06, "loss": 0.2714, "step": 19868 }, { "epoch": 1.373591427583823, "grad_norm": 0.6746964216875109, "learning_rate": 2.3598092948489022e-06, "loss": 0.3321, "step": 19869 }, { "epoch": 1.373660559972347, "grad_norm": 0.6870012505683557, "learning_rate": 2.359333952302141e-06, "loss": 0.2906, "step": 19870 }, { "epoch": 1.373729692360871, "grad_norm": 0.6594744412376075, "learning_rate": 2.3588586428514177e-06, "loss": 0.2571, "step": 19871 }, { "epoch": 1.373798824749395, "grad_norm": 0.9244913970644619, "learning_rate": 2.3583833665026867e-06, "loss": 0.2611, "step": 19872 }, { "epoch": 1.373867957137919, "grad_norm": 0.8485738699943018, "learning_rate": 2.3579081232619054e-06, "loss": 0.3478, "step": 19873 }, { "epoch": 1.3739370895264431, "grad_norm": 0.6547036246054595, "learning_rate": 2.3574329131350314e-06, "loss": 0.2187, "step": 19874 }, { "epoch": 1.3740062219149671, "grad_norm": 0.7204372425408616, "learning_rate": 2.3569577361280215e-06, "loss": 0.2702, "step": 19875 }, { "epoch": 1.3740753543034911, "grad_norm": 0.660736147695708, "learning_rate": 2.356482592246828e-06, "loss": 0.2724, "step": 19876 }, { "epoch": 1.3741444866920152, "grad_norm": 0.7097961386953557, "learning_rate": 2.3560074814974094e-06, "loss": 0.343, "step": 19877 }, { "epoch": 1.3742136190805392, "grad_norm": 0.6090701805027067, "learning_rate": 2.355532403885717e-06, "loss": 0.2186, "step": 19878 }, { "epoch": 1.3742827514690632, "grad_norm": 0.7062260864099142, "learning_rate": 2.3550573594177064e-06, "loss": 0.2958, "step": 19879 }, { "epoch": 1.3743518838575872, "grad_norm": 0.7143992820940208, "learning_rate": 2.3545823480993313e-06, "loss": 0.3588, "step": 19880 }, { "epoch": 1.3744210162461112, "grad_norm": 0.6760470146671151, "learning_rate": 2.354107369936548e-06, "loss": 0.2894, "step": 19881 }, { "epoch": 1.3744901486346353, "grad_norm": 0.6327861207219202, "learning_rate": 2.353632424935305e-06, "loss": 0.217, "step": 19882 }, { "epoch": 1.3745592810231593, "grad_norm": 0.7076812713933115, "learning_rate": 2.353157513101558e-06, "loss": 0.2749, "step": 19883 }, { "epoch": 1.3746284134116833, "grad_norm": 0.7189115117412669, "learning_rate": 2.3526826344412574e-06, "loss": 0.3674, "step": 19884 }, { "epoch": 1.3746975458002075, "grad_norm": 0.8126853737181617, "learning_rate": 2.3522077889603566e-06, "loss": 0.2763, "step": 19885 }, { "epoch": 1.3747666781887316, "grad_norm": 0.6491708249557528, "learning_rate": 2.3517329766648044e-06, "loss": 0.2454, "step": 19886 }, { "epoch": 1.3748358105772556, "grad_norm": 0.7043024630862187, "learning_rate": 2.351258197560554e-06, "loss": 0.3618, "step": 19887 }, { "epoch": 1.3749049429657796, "grad_norm": 0.6857636666712592, "learning_rate": 2.350783451653556e-06, "loss": 0.317, "step": 19888 }, { "epoch": 1.3749740753543036, "grad_norm": 0.6574692790665159, "learning_rate": 2.350308738949758e-06, "loss": 0.3026, "step": 19889 }, { "epoch": 1.3750432077428276, "grad_norm": 0.6599755745579866, "learning_rate": 2.3498340594551133e-06, "loss": 0.2273, "step": 19890 }, { "epoch": 1.3751123401313516, "grad_norm": 0.695829050654815, "learning_rate": 2.3493594131755674e-06, "loss": 0.3496, "step": 19891 }, { "epoch": 1.3751814725198757, "grad_norm": 0.6721517119809616, "learning_rate": 2.3488848001170735e-06, "loss": 0.2777, "step": 19892 }, { "epoch": 1.3752506049083997, "grad_norm": 0.5865997037228415, "learning_rate": 2.348410220285573e-06, "loss": 0.2256, "step": 19893 }, { "epoch": 1.3753197372969237, "grad_norm": 0.659023746515287, "learning_rate": 2.347935673687022e-06, "loss": 0.2854, "step": 19894 }, { "epoch": 1.3753888696854477, "grad_norm": 0.7092222089582015, "learning_rate": 2.3474611603273635e-06, "loss": 0.3232, "step": 19895 }, { "epoch": 1.3754580020739717, "grad_norm": 0.5814284382542763, "learning_rate": 2.346986680212547e-06, "loss": 0.2305, "step": 19896 }, { "epoch": 1.3755271344624957, "grad_norm": 0.7096245856385992, "learning_rate": 2.3465122333485165e-06, "loss": 0.2592, "step": 19897 }, { "epoch": 1.3755962668510198, "grad_norm": 0.6967347378091625, "learning_rate": 2.346037819741221e-06, "loss": 0.3897, "step": 19898 }, { "epoch": 1.3756653992395438, "grad_norm": 0.545037408375988, "learning_rate": 2.3455634393966038e-06, "loss": 0.2193, "step": 19899 }, { "epoch": 1.3757345316280678, "grad_norm": 0.6656199193339262, "learning_rate": 2.3450890923206117e-06, "loss": 0.2179, "step": 19900 }, { "epoch": 1.3758036640165918, "grad_norm": 0.6956880153244241, "learning_rate": 2.3446147785191897e-06, "loss": 0.2873, "step": 19901 }, { "epoch": 1.3758727964051158, "grad_norm": 0.7584671736031187, "learning_rate": 2.3441404979982846e-06, "loss": 0.3567, "step": 19902 }, { "epoch": 1.3759419287936399, "grad_norm": 0.6436339606615208, "learning_rate": 2.3436662507638377e-06, "loss": 0.2553, "step": 19903 }, { "epoch": 1.3760110611821639, "grad_norm": 0.6848139983742001, "learning_rate": 2.343192036821795e-06, "loss": 0.2392, "step": 19904 }, { "epoch": 1.376080193570688, "grad_norm": 0.7090263872477108, "learning_rate": 2.3427178561780967e-06, "loss": 0.3113, "step": 19905 }, { "epoch": 1.376149325959212, "grad_norm": 0.742139230119738, "learning_rate": 2.342243708838691e-06, "loss": 0.2398, "step": 19906 }, { "epoch": 1.376218458347736, "grad_norm": 1.6142624091719495, "learning_rate": 2.3417695948095143e-06, "loss": 0.256, "step": 19907 }, { "epoch": 1.37628759073626, "grad_norm": 0.8521219234907057, "learning_rate": 2.341295514096512e-06, "loss": 0.3463, "step": 19908 }, { "epoch": 1.376356723124784, "grad_norm": 0.6674024920970808, "learning_rate": 2.3408214667056266e-06, "loss": 0.3475, "step": 19909 }, { "epoch": 1.376425855513308, "grad_norm": 0.712319333050714, "learning_rate": 2.3403474526428e-06, "loss": 0.26, "step": 19910 }, { "epoch": 1.376494987901832, "grad_norm": 0.7138119468381879, "learning_rate": 2.33987347191397e-06, "loss": 0.2305, "step": 19911 }, { "epoch": 1.376564120290356, "grad_norm": 0.6514949104104273, "learning_rate": 2.3393995245250808e-06, "loss": 0.345, "step": 19912 }, { "epoch": 1.37663325267888, "grad_norm": 0.7414689022785965, "learning_rate": 2.3389256104820677e-06, "loss": 0.3375, "step": 19913 }, { "epoch": 1.376702385067404, "grad_norm": 0.6254224648025452, "learning_rate": 2.3384517297908745e-06, "loss": 0.2358, "step": 19914 }, { "epoch": 1.376771517455928, "grad_norm": 0.7393007794909591, "learning_rate": 2.3379778824574384e-06, "loss": 0.3166, "step": 19915 }, { "epoch": 1.376840649844452, "grad_norm": 0.6706307993429043, "learning_rate": 2.337504068487701e-06, "loss": 0.3457, "step": 19916 }, { "epoch": 1.376909782232976, "grad_norm": 0.6535719939008172, "learning_rate": 2.3370302878875966e-06, "loss": 0.2717, "step": 19917 }, { "epoch": 1.3769789146215001, "grad_norm": 0.7355353996678419, "learning_rate": 2.336556540663067e-06, "loss": 0.2381, "step": 19918 }, { "epoch": 1.3770480470100241, "grad_norm": 0.7040441609374459, "learning_rate": 2.336082826820046e-06, "loss": 0.366, "step": 19919 }, { "epoch": 1.3771171793985482, "grad_norm": 0.68175017425982, "learning_rate": 2.3356091463644754e-06, "loss": 0.2665, "step": 19920 }, { "epoch": 1.3771863117870722, "grad_norm": 0.6476060065710315, "learning_rate": 2.3351354993022874e-06, "loss": 0.2363, "step": 19921 }, { "epoch": 1.3772554441755962, "grad_norm": 0.7396521047900942, "learning_rate": 2.33466188563942e-06, "loss": 0.2714, "step": 19922 }, { "epoch": 1.3773245765641202, "grad_norm": 0.6262903817379558, "learning_rate": 2.3341883053818093e-06, "loss": 0.3875, "step": 19923 }, { "epoch": 1.3773937089526442, "grad_norm": 0.663396850582297, "learning_rate": 2.3337147585353923e-06, "loss": 0.2215, "step": 19924 }, { "epoch": 1.3774628413411683, "grad_norm": 0.6243561444552116, "learning_rate": 2.3332412451061014e-06, "loss": 0.2304, "step": 19925 }, { "epoch": 1.3775319737296923, "grad_norm": 0.712070333461631, "learning_rate": 2.3327677650998743e-06, "loss": 0.3149, "step": 19926 }, { "epoch": 1.3776011061182163, "grad_norm": 0.6767388347922193, "learning_rate": 2.3322943185226414e-06, "loss": 0.3102, "step": 19927 }, { "epoch": 1.3776702385067403, "grad_norm": 0.6957584022037727, "learning_rate": 2.3318209053803404e-06, "loss": 0.2329, "step": 19928 }, { "epoch": 1.3777393708952643, "grad_norm": 0.646425075874306, "learning_rate": 2.3313475256788985e-06, "loss": 0.2264, "step": 19929 }, { "epoch": 1.3778085032837883, "grad_norm": 0.7085606962959573, "learning_rate": 2.3308741794242572e-06, "loss": 0.3959, "step": 19930 }, { "epoch": 1.3778776356723124, "grad_norm": 0.7544878472491283, "learning_rate": 2.3304008666223428e-06, "loss": 0.3034, "step": 19931 }, { "epoch": 1.3779467680608364, "grad_norm": 0.6568722798827149, "learning_rate": 2.329927587279091e-06, "loss": 0.24, "step": 19932 }, { "epoch": 1.3780159004493604, "grad_norm": 0.7092684181282365, "learning_rate": 2.3294543414004308e-06, "loss": 0.2972, "step": 19933 }, { "epoch": 1.3780850328378844, "grad_norm": 0.7289577305435169, "learning_rate": 2.328981128992296e-06, "loss": 0.4158, "step": 19934 }, { "epoch": 1.3781541652264084, "grad_norm": 0.715649029582731, "learning_rate": 2.3285079500606137e-06, "loss": 0.2426, "step": 19935 }, { "epoch": 1.3782232976149327, "grad_norm": 0.632659287185453, "learning_rate": 2.3280348046113177e-06, "loss": 0.2312, "step": 19936 }, { "epoch": 1.3782924300034567, "grad_norm": 0.753454157635723, "learning_rate": 2.327561692650337e-06, "loss": 0.2893, "step": 19937 }, { "epoch": 1.3783615623919807, "grad_norm": 0.6334676245349068, "learning_rate": 2.3270886141836026e-06, "loss": 0.2486, "step": 19938 }, { "epoch": 1.3784306947805047, "grad_norm": 0.6591748772628543, "learning_rate": 2.326615569217041e-06, "loss": 0.2218, "step": 19939 }, { "epoch": 1.3784998271690287, "grad_norm": 0.6781284679318997, "learning_rate": 2.326142557756584e-06, "loss": 0.2554, "step": 19940 }, { "epoch": 1.3785689595575528, "grad_norm": 0.7815956197608851, "learning_rate": 2.325669579808156e-06, "loss": 0.3593, "step": 19941 }, { "epoch": 1.3786380919460768, "grad_norm": 0.6531848646940516, "learning_rate": 2.32519663537769e-06, "loss": 0.2623, "step": 19942 }, { "epoch": 1.3787072243346008, "grad_norm": 0.5990370709242429, "learning_rate": 2.3247237244711083e-06, "loss": 0.224, "step": 19943 }, { "epoch": 1.3787763567231248, "grad_norm": 0.7367171050068432, "learning_rate": 2.324250847094341e-06, "loss": 0.356, "step": 19944 }, { "epoch": 1.3788454891116488, "grad_norm": 0.6561106918638183, "learning_rate": 2.323778003253314e-06, "loss": 0.2845, "step": 19945 }, { "epoch": 1.3789146215001729, "grad_norm": 0.6477567062670185, "learning_rate": 2.3233051929539557e-06, "loss": 0.2396, "step": 19946 }, { "epoch": 1.3789837538886969, "grad_norm": 0.6840811453269741, "learning_rate": 2.3228324162021883e-06, "loss": 0.2766, "step": 19947 }, { "epoch": 1.379052886277221, "grad_norm": 0.6896944679126419, "learning_rate": 2.3223596730039405e-06, "loss": 0.3568, "step": 19948 }, { "epoch": 1.379122018665745, "grad_norm": 0.6556975352831304, "learning_rate": 2.3218869633651334e-06, "loss": 0.2379, "step": 19949 }, { "epoch": 1.379191151054269, "grad_norm": 0.6672298511271965, "learning_rate": 2.321414287291695e-06, "loss": 0.2444, "step": 19950 }, { "epoch": 1.379260283442793, "grad_norm": 1.0779488940481363, "learning_rate": 2.3209416447895477e-06, "loss": 0.2857, "step": 19951 }, { "epoch": 1.379329415831317, "grad_norm": 0.6647976020628418, "learning_rate": 2.3204690358646183e-06, "loss": 0.3644, "step": 19952 }, { "epoch": 1.379398548219841, "grad_norm": 0.6955092209759774, "learning_rate": 2.319996460522825e-06, "loss": 0.2562, "step": 19953 }, { "epoch": 1.379467680608365, "grad_norm": 0.6637948001902991, "learning_rate": 2.319523918770096e-06, "loss": 0.2124, "step": 19954 }, { "epoch": 1.379536812996889, "grad_norm": 0.7527048571648208, "learning_rate": 2.3190514106123507e-06, "loss": 0.3883, "step": 19955 }, { "epoch": 1.379605945385413, "grad_norm": 0.8337170702820185, "learning_rate": 2.31857893605551e-06, "loss": 0.3263, "step": 19956 }, { "epoch": 1.379675077773937, "grad_norm": 0.6879043506279313, "learning_rate": 2.3181064951054965e-06, "loss": 0.2563, "step": 19957 }, { "epoch": 1.379744210162461, "grad_norm": 0.8183004041096581, "learning_rate": 2.3176340877682328e-06, "loss": 0.2952, "step": 19958 }, { "epoch": 1.379813342550985, "grad_norm": 0.6404982191802161, "learning_rate": 2.3171617140496406e-06, "loss": 0.2986, "step": 19959 }, { "epoch": 1.379882474939509, "grad_norm": 0.6572678392153202, "learning_rate": 2.316689373955636e-06, "loss": 0.2416, "step": 19960 }, { "epoch": 1.3799516073280331, "grad_norm": 0.6653455040254738, "learning_rate": 2.316217067492144e-06, "loss": 0.2602, "step": 19961 }, { "epoch": 1.3800207397165571, "grad_norm": 0.6820235031270517, "learning_rate": 2.3157447946650803e-06, "loss": 0.3511, "step": 19962 }, { "epoch": 1.3800898721050812, "grad_norm": 0.7111984532950553, "learning_rate": 2.315272555480366e-06, "loss": 0.2658, "step": 19963 }, { "epoch": 1.3801590044936052, "grad_norm": 0.6639425103708753, "learning_rate": 2.3148003499439163e-06, "loss": 0.1812, "step": 19964 }, { "epoch": 1.3802281368821292, "grad_norm": 0.7467416339580206, "learning_rate": 2.314328178061656e-06, "loss": 0.2383, "step": 19965 }, { "epoch": 1.3802972692706534, "grad_norm": 0.6580226340327371, "learning_rate": 2.313856039839497e-06, "loss": 0.3393, "step": 19966 }, { "epoch": 1.3803664016591775, "grad_norm": 0.714104809484156, "learning_rate": 2.313383935283361e-06, "loss": 0.2102, "step": 19967 }, { "epoch": 1.3804355340477015, "grad_norm": 0.7254989489804501, "learning_rate": 2.3129118643991606e-06, "loss": 0.3088, "step": 19968 }, { "epoch": 1.3805046664362255, "grad_norm": 0.6902172496291883, "learning_rate": 2.3124398271928167e-06, "loss": 0.3203, "step": 19969 }, { "epoch": 1.3805737988247495, "grad_norm": 0.6608091779274669, "learning_rate": 2.3119678236702414e-06, "loss": 0.3481, "step": 19970 }, { "epoch": 1.3806429312132735, "grad_norm": 0.6918961642223015, "learning_rate": 2.311495853837353e-06, "loss": 0.2411, "step": 19971 }, { "epoch": 1.3807120636017975, "grad_norm": 0.6932507593745738, "learning_rate": 2.3110239177000653e-06, "loss": 0.2335, "step": 19972 }, { "epoch": 1.3807811959903216, "grad_norm": 0.7706555234245123, "learning_rate": 2.3105520152642968e-06, "loss": 0.3583, "step": 19973 }, { "epoch": 1.3808503283788456, "grad_norm": 0.6704258216142963, "learning_rate": 2.3100801465359574e-06, "loss": 0.2138, "step": 19974 }, { "epoch": 1.3809194607673696, "grad_norm": 0.628873367455302, "learning_rate": 2.3096083115209643e-06, "loss": 0.2573, "step": 19975 }, { "epoch": 1.3809885931558936, "grad_norm": 0.7349356862476552, "learning_rate": 2.3091365102252283e-06, "loss": 0.2972, "step": 19976 }, { "epoch": 1.3810577255444176, "grad_norm": 0.6687277934840262, "learning_rate": 2.3086647426546656e-06, "loss": 0.3364, "step": 19977 }, { "epoch": 1.3811268579329417, "grad_norm": 0.8286813113511512, "learning_rate": 2.308193008815186e-06, "loss": 0.2664, "step": 19978 }, { "epoch": 1.3811959903214657, "grad_norm": 0.6590314188175382, "learning_rate": 2.3077213087127026e-06, "loss": 0.2239, "step": 19979 }, { "epoch": 1.3812651227099897, "grad_norm": 0.7391045865242788, "learning_rate": 2.307249642353129e-06, "loss": 0.3763, "step": 19980 }, { "epoch": 1.3813342550985137, "grad_norm": 0.7790363275715236, "learning_rate": 2.3067780097423768e-06, "loss": 0.2748, "step": 19981 }, { "epoch": 1.3814033874870377, "grad_norm": 0.649113291559498, "learning_rate": 2.3063064108863543e-06, "loss": 0.2261, "step": 19982 }, { "epoch": 1.3814725198755617, "grad_norm": 0.6883819866966867, "learning_rate": 2.3058348457909756e-06, "loss": 0.2542, "step": 19983 }, { "epoch": 1.3815416522640858, "grad_norm": 0.6851935589220678, "learning_rate": 2.305363314462148e-06, "loss": 0.3262, "step": 19984 }, { "epoch": 1.3816107846526098, "grad_norm": 0.5972267002686461, "learning_rate": 2.3048918169057816e-06, "loss": 0.2068, "step": 19985 }, { "epoch": 1.3816799170411338, "grad_norm": 0.6865266579311681, "learning_rate": 2.304420353127787e-06, "loss": 0.2586, "step": 19986 }, { "epoch": 1.3817490494296578, "grad_norm": 0.6519150078931955, "learning_rate": 2.3039489231340745e-06, "loss": 0.3021, "step": 19987 }, { "epoch": 1.3818181818181818, "grad_norm": 0.7720981365876871, "learning_rate": 2.303477526930549e-06, "loss": 0.2577, "step": 19988 }, { "epoch": 1.3818873142067059, "grad_norm": 0.5919367081774155, "learning_rate": 2.3030061645231226e-06, "loss": 0.2129, "step": 19989 }, { "epoch": 1.3819564465952299, "grad_norm": 1.0864192877080348, "learning_rate": 2.302534835917699e-06, "loss": 0.3136, "step": 19990 }, { "epoch": 1.3820255789837539, "grad_norm": 0.6898095604284019, "learning_rate": 2.30206354112019e-06, "loss": 0.3246, "step": 19991 }, { "epoch": 1.382094711372278, "grad_norm": 0.7365473030021829, "learning_rate": 2.3015922801364965e-06, "loss": 0.1922, "step": 19992 }, { "epoch": 1.382163843760802, "grad_norm": 0.6896966144092885, "learning_rate": 2.3011210529725294e-06, "loss": 0.2178, "step": 19993 }, { "epoch": 1.382232976149326, "grad_norm": 0.7195238143586029, "learning_rate": 2.3006498596341932e-06, "loss": 0.3532, "step": 19994 }, { "epoch": 1.38230210853785, "grad_norm": 0.663909966749042, "learning_rate": 2.3001787001273957e-06, "loss": 0.2952, "step": 19995 }, { "epoch": 1.382371240926374, "grad_norm": 0.8128253492551425, "learning_rate": 2.2997075744580377e-06, "loss": 0.2602, "step": 19996 }, { "epoch": 1.382440373314898, "grad_norm": 0.666050437494548, "learning_rate": 2.2992364826320292e-06, "loss": 0.2247, "step": 19997 }, { "epoch": 1.382509505703422, "grad_norm": 0.6701753349414292, "learning_rate": 2.2987654246552688e-06, "loss": 0.2927, "step": 19998 }, { "epoch": 1.382578638091946, "grad_norm": 0.6351840570843326, "learning_rate": 2.298294400533666e-06, "loss": 0.2377, "step": 19999 }, { "epoch": 1.38264777048047, "grad_norm": 0.8113208598394176, "learning_rate": 2.297823410273117e-06, "loss": 0.2371, "step": 20000 }, { "epoch": 1.382716902868994, "grad_norm": 0.6746510561623729, "learning_rate": 2.297352453879533e-06, "loss": 0.2896, "step": 20001 }, { "epoch": 1.382786035257518, "grad_norm": 0.6787571127486752, "learning_rate": 2.296881531358812e-06, "loss": 0.3593, "step": 20002 }, { "epoch": 1.382855167646042, "grad_norm": 0.6291390741515617, "learning_rate": 2.296410642716858e-06, "loss": 0.2271, "step": 20003 }, { "epoch": 1.3829243000345661, "grad_norm": 0.7488346880517734, "learning_rate": 2.295939787959571e-06, "loss": 0.2596, "step": 20004 }, { "epoch": 1.3829934324230901, "grad_norm": 0.7125044331449576, "learning_rate": 2.2954689670928535e-06, "loss": 0.3226, "step": 20005 }, { "epoch": 1.3830625648116142, "grad_norm": 0.6723995608475211, "learning_rate": 2.2949981801226057e-06, "loss": 0.2486, "step": 20006 }, { "epoch": 1.3831316972001382, "grad_norm": 0.715387189346909, "learning_rate": 2.2945274270547275e-06, "loss": 0.2471, "step": 20007 }, { "epoch": 1.3832008295886622, "grad_norm": 0.6651387453413874, "learning_rate": 2.29405670789512e-06, "loss": 0.2545, "step": 20008 }, { "epoch": 1.3832699619771862, "grad_norm": 0.6404398444547115, "learning_rate": 2.2935860226496847e-06, "loss": 0.324, "step": 20009 }, { "epoch": 1.3833390943657102, "grad_norm": 0.7001866444339683, "learning_rate": 2.293115371324317e-06, "loss": 0.1964, "step": 20010 }, { "epoch": 1.3834082267542342, "grad_norm": 0.6252340421955951, "learning_rate": 2.2926447539249192e-06, "loss": 0.2462, "step": 20011 }, { "epoch": 1.3834773591427583, "grad_norm": 0.7086111389666985, "learning_rate": 2.292174170457386e-06, "loss": 0.2696, "step": 20012 }, { "epoch": 1.3835464915312823, "grad_norm": 0.7118519909263852, "learning_rate": 2.2917036209276193e-06, "loss": 0.2793, "step": 20013 }, { "epoch": 1.3836156239198063, "grad_norm": 0.8449140312128159, "learning_rate": 2.2912331053415126e-06, "loss": 0.2246, "step": 20014 }, { "epoch": 1.3836847563083303, "grad_norm": 0.9368102575850249, "learning_rate": 2.2907626237049646e-06, "loss": 0.2722, "step": 20015 }, { "epoch": 1.3837538886968543, "grad_norm": 0.6526153525366265, "learning_rate": 2.2902921760238727e-06, "loss": 0.3907, "step": 20016 }, { "epoch": 1.3838230210853786, "grad_norm": 0.6765212996331148, "learning_rate": 2.289821762304134e-06, "loss": 0.2279, "step": 20017 }, { "epoch": 1.3838921534739026, "grad_norm": 0.6574648490799888, "learning_rate": 2.2893513825516417e-06, "loss": 0.1968, "step": 20018 }, { "epoch": 1.3839612858624266, "grad_norm": 0.7001406475245224, "learning_rate": 2.288881036772294e-06, "loss": 0.322, "step": 20019 }, { "epoch": 1.3840304182509506, "grad_norm": 0.6790383923925689, "learning_rate": 2.288410724971983e-06, "loss": 0.2736, "step": 20020 }, { "epoch": 1.3840995506394747, "grad_norm": 0.623027327427293, "learning_rate": 2.2879404471566037e-06, "loss": 0.2492, "step": 20021 }, { "epoch": 1.3841686830279987, "grad_norm": 0.6114266854719212, "learning_rate": 2.2874702033320536e-06, "loss": 0.2388, "step": 20022 }, { "epoch": 1.3842378154165227, "grad_norm": 0.6785476969421365, "learning_rate": 2.2869999935042215e-06, "loss": 0.3692, "step": 20023 }, { "epoch": 1.3843069478050467, "grad_norm": 0.6628257661679627, "learning_rate": 2.2865298176790046e-06, "loss": 0.2747, "step": 20024 }, { "epoch": 1.3843760801935707, "grad_norm": 0.7244478941995345, "learning_rate": 2.286059675862292e-06, "loss": 0.2968, "step": 20025 }, { "epoch": 1.3844452125820947, "grad_norm": 0.7504195835179903, "learning_rate": 2.2855895680599806e-06, "loss": 0.2827, "step": 20026 }, { "epoch": 1.3845143449706188, "grad_norm": 0.7183793490416862, "learning_rate": 2.2851194942779574e-06, "loss": 0.3356, "step": 20027 }, { "epoch": 1.3845834773591428, "grad_norm": 0.5889449765759026, "learning_rate": 2.284649454522117e-06, "loss": 0.2495, "step": 20028 }, { "epoch": 1.3846526097476668, "grad_norm": 0.6294078656704669, "learning_rate": 2.28417944879835e-06, "loss": 0.2098, "step": 20029 }, { "epoch": 1.3847217421361908, "grad_norm": 0.7294578959003449, "learning_rate": 2.2837094771125485e-06, "loss": 0.3056, "step": 20030 }, { "epoch": 1.3847908745247148, "grad_norm": 0.68680493863286, "learning_rate": 2.283239539470599e-06, "loss": 0.238, "step": 20031 }, { "epoch": 1.3848600069132389, "grad_norm": 0.6093332709690821, "learning_rate": 2.282769635878396e-06, "loss": 0.2314, "step": 20032 }, { "epoch": 1.3849291393017629, "grad_norm": 0.6528479871292892, "learning_rate": 2.282299766341825e-06, "loss": 0.2671, "step": 20033 }, { "epoch": 1.3849982716902869, "grad_norm": 0.6759091707588449, "learning_rate": 2.281829930866778e-06, "loss": 0.3548, "step": 20034 }, { "epoch": 1.385067404078811, "grad_norm": 0.6546708526757897, "learning_rate": 2.281360129459138e-06, "loss": 0.2522, "step": 20035 }, { "epoch": 1.385136536467335, "grad_norm": 0.612099246354873, "learning_rate": 2.2808903621248014e-06, "loss": 0.2183, "step": 20036 }, { "epoch": 1.385205668855859, "grad_norm": 0.9708163291433035, "learning_rate": 2.2804206288696497e-06, "loss": 0.3032, "step": 20037 }, { "epoch": 1.385274801244383, "grad_norm": 0.7049132031328581, "learning_rate": 2.279950929699573e-06, "loss": 0.2992, "step": 20038 }, { "epoch": 1.385343933632907, "grad_norm": 0.64427887398951, "learning_rate": 2.279481264620457e-06, "loss": 0.2079, "step": 20039 }, { "epoch": 1.385413066021431, "grad_norm": 0.6571176290257987, "learning_rate": 2.279011633638189e-06, "loss": 0.257, "step": 20040 }, { "epoch": 1.385482198409955, "grad_norm": 0.6679891544834626, "learning_rate": 2.2785420367586526e-06, "loss": 0.4008, "step": 20041 }, { "epoch": 1.385551330798479, "grad_norm": 0.6485636412329182, "learning_rate": 2.278072473987735e-06, "loss": 0.2593, "step": 20042 }, { "epoch": 1.385620463187003, "grad_norm": 0.6416340680429758, "learning_rate": 2.277602945331322e-06, "loss": 0.2218, "step": 20043 }, { "epoch": 1.385689595575527, "grad_norm": 0.655334873994452, "learning_rate": 2.2771334507952998e-06, "loss": 0.2801, "step": 20044 }, { "epoch": 1.385758727964051, "grad_norm": 0.6636674136780281, "learning_rate": 2.276663990385548e-06, "loss": 0.2839, "step": 20045 }, { "epoch": 1.385827860352575, "grad_norm": 0.6939011534778182, "learning_rate": 2.2761945641079553e-06, "loss": 0.2292, "step": 20046 }, { "epoch": 1.3858969927410993, "grad_norm": 0.6597164748860166, "learning_rate": 2.2757251719684013e-06, "loss": 0.2688, "step": 20047 }, { "epoch": 1.3859661251296234, "grad_norm": 0.6808530615249258, "learning_rate": 2.275255813972772e-06, "loss": 0.2944, "step": 20048 }, { "epoch": 1.3860352575181474, "grad_norm": 0.6723215873721867, "learning_rate": 2.2747864901269467e-06, "loss": 0.2533, "step": 20049 }, { "epoch": 1.3861043899066714, "grad_norm": 0.6298350964252897, "learning_rate": 2.2743172004368095e-06, "loss": 0.2388, "step": 20050 }, { "epoch": 1.3861735222951954, "grad_norm": 0.8257776978035962, "learning_rate": 2.2738479449082425e-06, "loss": 0.3398, "step": 20051 }, { "epoch": 1.3862426546837194, "grad_norm": 0.6762006737181043, "learning_rate": 2.273378723547128e-06, "loss": 0.3426, "step": 20052 }, { "epoch": 1.3863117870722435, "grad_norm": 0.7012891301008393, "learning_rate": 2.272909536359343e-06, "loss": 0.2081, "step": 20053 }, { "epoch": 1.3863809194607675, "grad_norm": 0.7492335966798417, "learning_rate": 2.2724403833507725e-06, "loss": 0.2565, "step": 20054 }, { "epoch": 1.3864500518492915, "grad_norm": 0.7120261895761297, "learning_rate": 2.2719712645272924e-06, "loss": 0.344, "step": 20055 }, { "epoch": 1.3865191842378155, "grad_norm": 0.7854743320398612, "learning_rate": 2.2715021798947837e-06, "loss": 0.2826, "step": 20056 }, { "epoch": 1.3865883166263395, "grad_norm": 0.6632219648440036, "learning_rate": 2.2710331294591263e-06, "loss": 0.2171, "step": 20057 }, { "epoch": 1.3866574490148635, "grad_norm": 0.7644455764471304, "learning_rate": 2.2705641132262006e-06, "loss": 0.3129, "step": 20058 }, { "epoch": 1.3867265814033876, "grad_norm": 0.6960470540282392, "learning_rate": 2.270095131201881e-06, "loss": 0.3364, "step": 20059 }, { "epoch": 1.3867957137919116, "grad_norm": 0.667969624507974, "learning_rate": 2.2696261833920484e-06, "loss": 0.2397, "step": 20060 }, { "epoch": 1.3868648461804356, "grad_norm": 0.6564979470163355, "learning_rate": 2.269157269802578e-06, "loss": 0.218, "step": 20061 }, { "epoch": 1.3869339785689596, "grad_norm": 0.6737513452797903, "learning_rate": 2.2686883904393492e-06, "loss": 0.3395, "step": 20062 }, { "epoch": 1.3870031109574836, "grad_norm": 0.6926181751235226, "learning_rate": 2.2682195453082355e-06, "loss": 0.286, "step": 20063 }, { "epoch": 1.3870722433460076, "grad_norm": 0.6203275490656782, "learning_rate": 2.2677507344151147e-06, "loss": 0.2458, "step": 20064 }, { "epoch": 1.3871413757345317, "grad_norm": 0.6849472729126498, "learning_rate": 2.267281957765862e-06, "loss": 0.2961, "step": 20065 }, { "epoch": 1.3872105081230557, "grad_norm": 0.6780593766727536, "learning_rate": 2.266813215366356e-06, "loss": 0.2876, "step": 20066 }, { "epoch": 1.3872796405115797, "grad_norm": 0.6689781806250249, "learning_rate": 2.266344507222466e-06, "loss": 0.2714, "step": 20067 }, { "epoch": 1.3873487729001037, "grad_norm": 0.7334886959659153, "learning_rate": 2.2658758333400715e-06, "loss": 0.2374, "step": 20068 }, { "epoch": 1.3874179052886277, "grad_norm": 0.6329153765509486, "learning_rate": 2.2654071937250425e-06, "loss": 0.2723, "step": 20069 }, { "epoch": 1.3874870376771518, "grad_norm": 0.8234498882233859, "learning_rate": 2.264938588383256e-06, "loss": 0.3395, "step": 20070 }, { "epoch": 1.3875561700656758, "grad_norm": 0.7483177473834096, "learning_rate": 2.264470017320579e-06, "loss": 0.3181, "step": 20071 }, { "epoch": 1.3876253024541998, "grad_norm": 0.6150010242105279, "learning_rate": 2.2640014805428928e-06, "loss": 0.2225, "step": 20072 }, { "epoch": 1.3876944348427238, "grad_norm": 0.7222611348333347, "learning_rate": 2.263532978056064e-06, "loss": 0.4496, "step": 20073 }, { "epoch": 1.3877635672312478, "grad_norm": 0.6847844733653201, "learning_rate": 2.263064509865967e-06, "loss": 0.2573, "step": 20074 }, { "epoch": 1.3878326996197718, "grad_norm": 0.6072680931703386, "learning_rate": 2.2625960759784704e-06, "loss": 0.1989, "step": 20075 }, { "epoch": 1.3879018320082959, "grad_norm": 0.7685875077086155, "learning_rate": 2.262127676399449e-06, "loss": 0.3076, "step": 20076 }, { "epoch": 1.3879709643968199, "grad_norm": 0.7419705607529508, "learning_rate": 2.261659311134769e-06, "loss": 0.3171, "step": 20077 }, { "epoch": 1.388040096785344, "grad_norm": 0.6941303281327064, "learning_rate": 2.2611909801903033e-06, "loss": 0.3005, "step": 20078 }, { "epoch": 1.388109229173868, "grad_norm": 0.6584924408178092, "learning_rate": 2.2607226835719205e-06, "loss": 0.2754, "step": 20079 }, { "epoch": 1.388178361562392, "grad_norm": 0.7268922828382144, "learning_rate": 2.260254421285493e-06, "loss": 0.357, "step": 20080 }, { "epoch": 1.388247493950916, "grad_norm": 0.7395883980941059, "learning_rate": 2.2597861933368842e-06, "loss": 0.2762, "step": 20081 }, { "epoch": 1.38831662633944, "grad_norm": 0.6303987794988346, "learning_rate": 2.2593179997319676e-06, "loss": 0.2273, "step": 20082 }, { "epoch": 1.388385758727964, "grad_norm": 0.681295380331153, "learning_rate": 2.2588498404766066e-06, "loss": 0.24, "step": 20083 }, { "epoch": 1.388454891116488, "grad_norm": 0.7279222031651571, "learning_rate": 2.2583817155766737e-06, "loss": 0.3347, "step": 20084 }, { "epoch": 1.388524023505012, "grad_norm": 0.7382008711283152, "learning_rate": 2.257913625038031e-06, "loss": 0.2913, "step": 20085 }, { "epoch": 1.388593155893536, "grad_norm": 0.6325171103456357, "learning_rate": 2.2574455688665476e-06, "loss": 0.2245, "step": 20086 }, { "epoch": 1.38866228828206, "grad_norm": 0.7109067469334166, "learning_rate": 2.2569775470680914e-06, "loss": 0.3305, "step": 20087 }, { "epoch": 1.388731420670584, "grad_norm": 0.7369709310504784, "learning_rate": 2.2565095596485252e-06, "loss": 0.2991, "step": 20088 }, { "epoch": 1.388800553059108, "grad_norm": 0.676772319262382, "learning_rate": 2.2560416066137173e-06, "loss": 0.2141, "step": 20089 }, { "epoch": 1.3888696854476321, "grad_norm": 0.6974631211484051, "learning_rate": 2.2555736879695296e-06, "loss": 0.3005, "step": 20090 }, { "epoch": 1.3889388178361561, "grad_norm": 0.7071233937137449, "learning_rate": 2.2551058037218273e-06, "loss": 0.3243, "step": 20091 }, { "epoch": 1.3890079502246802, "grad_norm": 0.6764374063916829, "learning_rate": 2.2546379538764757e-06, "loss": 0.2667, "step": 20092 }, { "epoch": 1.3890770826132042, "grad_norm": 0.9447254586476687, "learning_rate": 2.2541701384393393e-06, "loss": 0.1983, "step": 20093 }, { "epoch": 1.3891462150017282, "grad_norm": 0.6456810760935838, "learning_rate": 2.2537023574162787e-06, "loss": 0.3324, "step": 20094 }, { "epoch": 1.3892153473902522, "grad_norm": 0.8700206437072266, "learning_rate": 2.2532346108131604e-06, "loss": 0.2938, "step": 20095 }, { "epoch": 1.3892844797787762, "grad_norm": 0.6413632910217278, "learning_rate": 2.252766898635842e-06, "loss": 0.2331, "step": 20096 }, { "epoch": 1.3893536121673002, "grad_norm": 0.680493182276813, "learning_rate": 2.2522992208901894e-06, "loss": 0.2745, "step": 20097 }, { "epoch": 1.3894227445558245, "grad_norm": 0.6225006998451478, "learning_rate": 2.251831577582061e-06, "loss": 0.3584, "step": 20098 }, { "epoch": 1.3894918769443485, "grad_norm": 0.6945247585761264, "learning_rate": 2.2513639687173195e-06, "loss": 0.2493, "step": 20099 }, { "epoch": 1.3895610093328725, "grad_norm": 0.6546417775888934, "learning_rate": 2.250896394301826e-06, "loss": 0.2522, "step": 20100 }, { "epoch": 1.3896301417213965, "grad_norm": 0.7343350506150106, "learning_rate": 2.2504288543414415e-06, "loss": 0.3766, "step": 20101 }, { "epoch": 1.3896992741099206, "grad_norm": 0.67680902838469, "learning_rate": 2.2499613488420224e-06, "loss": 0.3012, "step": 20102 }, { "epoch": 1.3897684064984446, "grad_norm": 0.7047451115976969, "learning_rate": 2.2494938778094322e-06, "loss": 0.2443, "step": 20103 }, { "epoch": 1.3898375388869686, "grad_norm": 0.7641483438892264, "learning_rate": 2.2490264412495262e-06, "loss": 0.222, "step": 20104 }, { "epoch": 1.3899066712754926, "grad_norm": 0.7196311453644354, "learning_rate": 2.2485590391681662e-06, "loss": 0.3564, "step": 20105 }, { "epoch": 1.3899758036640166, "grad_norm": 0.6751562340190087, "learning_rate": 2.248091671571205e-06, "loss": 0.2564, "step": 20106 }, { "epoch": 1.3900449360525406, "grad_norm": 0.8059971686243982, "learning_rate": 2.2476243384645064e-06, "loss": 0.2924, "step": 20107 }, { "epoch": 1.3901140684410647, "grad_norm": 0.758380863450055, "learning_rate": 2.247157039853924e-06, "loss": 0.2679, "step": 20108 }, { "epoch": 1.3901832008295887, "grad_norm": 0.6470923118326494, "learning_rate": 2.2466897757453165e-06, "loss": 0.3244, "step": 20109 }, { "epoch": 1.3902523332181127, "grad_norm": 0.7388819154440759, "learning_rate": 2.246222546144538e-06, "loss": 0.2718, "step": 20110 }, { "epoch": 1.3903214656066367, "grad_norm": 0.6362210189580305, "learning_rate": 2.245755351057447e-06, "loss": 0.1884, "step": 20111 }, { "epoch": 1.3903905979951607, "grad_norm": 0.8275365825690482, "learning_rate": 2.245288190489896e-06, "loss": 0.3474, "step": 20112 }, { "epoch": 1.3904597303836848, "grad_norm": 0.7407094173543373, "learning_rate": 2.2448210644477415e-06, "loss": 0.3557, "step": 20113 }, { "epoch": 1.3905288627722088, "grad_norm": 0.7006980257427623, "learning_rate": 2.244353972936838e-06, "loss": 0.2307, "step": 20114 }, { "epoch": 1.3905979951607328, "grad_norm": 0.6716393820327634, "learning_rate": 2.2438869159630416e-06, "loss": 0.2304, "step": 20115 }, { "epoch": 1.3906671275492568, "grad_norm": 0.6551897026002156, "learning_rate": 2.2434198935322023e-06, "loss": 0.3864, "step": 20116 }, { "epoch": 1.3907362599377808, "grad_norm": 0.6305447415690985, "learning_rate": 2.242952905650177e-06, "loss": 0.2394, "step": 20117 }, { "epoch": 1.3908053923263048, "grad_norm": 0.6711276948738304, "learning_rate": 2.2424859523228153e-06, "loss": 0.2177, "step": 20118 }, { "epoch": 1.3908745247148289, "grad_norm": 0.6889282521332785, "learning_rate": 2.2420190335559727e-06, "loss": 0.2876, "step": 20119 }, { "epoch": 1.3909436571033529, "grad_norm": 0.6877736165715539, "learning_rate": 2.2415521493554983e-06, "loss": 0.2654, "step": 20120 }, { "epoch": 1.391012789491877, "grad_norm": 0.5949343920851721, "learning_rate": 2.241085299727245e-06, "loss": 0.1609, "step": 20121 }, { "epoch": 1.391081921880401, "grad_norm": 0.6463827494630798, "learning_rate": 2.2406184846770634e-06, "loss": 0.1942, "step": 20122 }, { "epoch": 1.391151054268925, "grad_norm": 0.652067077875428, "learning_rate": 2.2401517042108073e-06, "loss": 0.3115, "step": 20123 }, { "epoch": 1.391220186657449, "grad_norm": 0.6691772442510057, "learning_rate": 2.2396849583343224e-06, "loss": 0.2064, "step": 20124 }, { "epoch": 1.391289319045973, "grad_norm": 1.2164430649715519, "learning_rate": 2.2392182470534617e-06, "loss": 0.3006, "step": 20125 }, { "epoch": 1.391358451434497, "grad_norm": 0.654747277350977, "learning_rate": 2.238751570374072e-06, "loss": 0.2783, "step": 20126 }, { "epoch": 1.391427583823021, "grad_norm": 0.6776459859062365, "learning_rate": 2.2382849283020043e-06, "loss": 0.3675, "step": 20127 }, { "epoch": 1.391496716211545, "grad_norm": 0.7215852125329462, "learning_rate": 2.237818320843106e-06, "loss": 0.277, "step": 20128 }, { "epoch": 1.3915658486000693, "grad_norm": 0.7154419508226872, "learning_rate": 2.2373517480032275e-06, "loss": 0.2595, "step": 20129 }, { "epoch": 1.3916349809885933, "grad_norm": 0.6052506116349674, "learning_rate": 2.2368852097882126e-06, "loss": 0.308, "step": 20130 }, { "epoch": 1.3917041133771173, "grad_norm": 0.7538632269506442, "learning_rate": 2.2364187062039128e-06, "loss": 0.3138, "step": 20131 }, { "epoch": 1.3917732457656413, "grad_norm": 0.6808866419387843, "learning_rate": 2.235952237256171e-06, "loss": 0.2337, "step": 20132 }, { "epoch": 1.3918423781541653, "grad_norm": 0.6679353927153473, "learning_rate": 2.235485802950837e-06, "loss": 0.2486, "step": 20133 }, { "epoch": 1.3919115105426894, "grad_norm": 0.6930768878294369, "learning_rate": 2.2350194032937527e-06, "loss": 0.3516, "step": 20134 }, { "epoch": 1.3919806429312134, "grad_norm": 0.7557442134020221, "learning_rate": 2.234553038290766e-06, "loss": 0.2258, "step": 20135 }, { "epoch": 1.3920497753197374, "grad_norm": 0.7029594651080658, "learning_rate": 2.2340867079477224e-06, "loss": 0.2685, "step": 20136 }, { "epoch": 1.3921189077082614, "grad_norm": 0.660847895868594, "learning_rate": 2.233620412270467e-06, "loss": 0.3136, "step": 20137 }, { "epoch": 1.3921880400967854, "grad_norm": 0.7005470937791477, "learning_rate": 2.2331541512648412e-06, "loss": 0.3209, "step": 20138 }, { "epoch": 1.3922571724853094, "grad_norm": 0.5811514766099479, "learning_rate": 2.2326879249366927e-06, "loss": 0.2052, "step": 20139 }, { "epoch": 1.3923263048738335, "grad_norm": 0.702740147064869, "learning_rate": 2.2322217332918604e-06, "loss": 0.2999, "step": 20140 }, { "epoch": 1.3923954372623575, "grad_norm": 0.7243111753539121, "learning_rate": 2.2317555763361914e-06, "loss": 0.3532, "step": 20141 }, { "epoch": 1.3924645696508815, "grad_norm": 0.7447526592569118, "learning_rate": 2.2312894540755225e-06, "loss": 0.2361, "step": 20142 }, { "epoch": 1.3925337020394055, "grad_norm": 0.6491402863721456, "learning_rate": 2.2308233665157026e-06, "loss": 0.2269, "step": 20143 }, { "epoch": 1.3926028344279295, "grad_norm": 0.7490165878139381, "learning_rate": 2.2303573136625686e-06, "loss": 0.3291, "step": 20144 }, { "epoch": 1.3926719668164536, "grad_norm": 0.7289857439941902, "learning_rate": 2.2298912955219645e-06, "loss": 0.305, "step": 20145 }, { "epoch": 1.3927410992049776, "grad_norm": 0.720749448403378, "learning_rate": 2.2294253120997277e-06, "loss": 0.2566, "step": 20146 }, { "epoch": 1.3928102315935016, "grad_norm": 0.9467283739872658, "learning_rate": 2.2289593634017022e-06, "loss": 0.2824, "step": 20147 }, { "epoch": 1.3928793639820256, "grad_norm": 0.6444887188261109, "learning_rate": 2.228493449433724e-06, "loss": 0.3153, "step": 20148 }, { "epoch": 1.3929484963705496, "grad_norm": 0.7892844117828611, "learning_rate": 2.2280275702016344e-06, "loss": 0.2367, "step": 20149 }, { "epoch": 1.3930176287590736, "grad_norm": 0.6451965909433051, "learning_rate": 2.227561725711273e-06, "loss": 0.1988, "step": 20150 }, { "epoch": 1.3930867611475977, "grad_norm": 0.6946050176633117, "learning_rate": 2.227095915968479e-06, "loss": 0.2687, "step": 20151 }, { "epoch": 1.3931558935361217, "grad_norm": 0.874708935150623, "learning_rate": 2.2266301409790876e-06, "loss": 0.3134, "step": 20152 }, { "epoch": 1.3932250259246457, "grad_norm": 0.6932640474951616, "learning_rate": 2.2261644007489398e-06, "loss": 0.2364, "step": 20153 }, { "epoch": 1.3932941583131697, "grad_norm": 0.6206399903121514, "learning_rate": 2.225698695283871e-06, "loss": 0.2135, "step": 20154 }, { "epoch": 1.3933632907016937, "grad_norm": 0.6351524509256249, "learning_rate": 2.225233024589717e-06, "loss": 0.3508, "step": 20155 }, { "epoch": 1.3934324230902178, "grad_norm": 0.7406408184312953, "learning_rate": 2.2247673886723143e-06, "loss": 0.2571, "step": 20156 }, { "epoch": 1.3935015554787418, "grad_norm": 0.7517481100864994, "learning_rate": 2.2243017875375e-06, "loss": 0.1992, "step": 20157 }, { "epoch": 1.3935706878672658, "grad_norm": 0.7128013860888396, "learning_rate": 2.223836221191111e-06, "loss": 0.2442, "step": 20158 }, { "epoch": 1.3936398202557898, "grad_norm": 0.6316841443315833, "learning_rate": 2.2233706896389797e-06, "loss": 0.2881, "step": 20159 }, { "epoch": 1.3937089526443138, "grad_norm": 0.6627222548453242, "learning_rate": 2.2229051928869427e-06, "loss": 0.2915, "step": 20160 }, { "epoch": 1.3937780850328378, "grad_norm": 0.732687421850809, "learning_rate": 2.2224397309408317e-06, "loss": 0.2232, "step": 20161 }, { "epoch": 1.3938472174213619, "grad_norm": 0.6948598787780619, "learning_rate": 2.221974303806482e-06, "loss": 0.3513, "step": 20162 }, { "epoch": 1.3939163498098859, "grad_norm": 0.7085314951363318, "learning_rate": 2.2215089114897266e-06, "loss": 0.3108, "step": 20163 }, { "epoch": 1.39398548219841, "grad_norm": 0.6253413367721662, "learning_rate": 2.2210435539964e-06, "loss": 0.2276, "step": 20164 }, { "epoch": 1.394054614586934, "grad_norm": 0.6506191080133977, "learning_rate": 2.2205782313323316e-06, "loss": 0.2666, "step": 20165 }, { "epoch": 1.394123746975458, "grad_norm": 0.6783843677867581, "learning_rate": 2.220112943503357e-06, "loss": 0.3255, "step": 20166 }, { "epoch": 1.394192879363982, "grad_norm": 0.7869747130021872, "learning_rate": 2.219647690515304e-06, "loss": 0.2571, "step": 20167 }, { "epoch": 1.394262011752506, "grad_norm": 0.6222744180487587, "learning_rate": 2.2191824723740068e-06, "loss": 0.2297, "step": 20168 }, { "epoch": 1.39433114414103, "grad_norm": 0.7225312699360552, "learning_rate": 2.2187172890852932e-06, "loss": 0.3656, "step": 20169 }, { "epoch": 1.394400276529554, "grad_norm": 0.7478134640958893, "learning_rate": 2.218252140654995e-06, "loss": 0.3087, "step": 20170 }, { "epoch": 1.394469408918078, "grad_norm": 0.7050858328637836, "learning_rate": 2.2177870270889417e-06, "loss": 0.2728, "step": 20171 }, { "epoch": 1.394538541306602, "grad_norm": 0.6568900904002143, "learning_rate": 2.2173219483929648e-06, "loss": 0.2257, "step": 20172 }, { "epoch": 1.394607673695126, "grad_norm": 1.02833186377192, "learning_rate": 2.2168569045728897e-06, "loss": 0.3534, "step": 20173 }, { "epoch": 1.39467680608365, "grad_norm": 0.7235001965340757, "learning_rate": 2.216391895634549e-06, "loss": 0.2649, "step": 20174 }, { "epoch": 1.394745938472174, "grad_norm": 0.6822551773571431, "learning_rate": 2.2159269215837653e-06, "loss": 0.216, "step": 20175 }, { "epoch": 1.3948150708606981, "grad_norm": 0.7016541648664192, "learning_rate": 2.215461982426372e-06, "loss": 0.3066, "step": 20176 }, { "epoch": 1.3948842032492221, "grad_norm": 0.7100492646512774, "learning_rate": 2.214997078168189e-06, "loss": 0.2884, "step": 20177 }, { "epoch": 1.3949533356377462, "grad_norm": 0.6847855488229203, "learning_rate": 2.2145322088150516e-06, "loss": 0.2311, "step": 20178 }, { "epoch": 1.3950224680262704, "grad_norm": 0.6746305950631607, "learning_rate": 2.21406737437278e-06, "loss": 0.2402, "step": 20179 }, { "epoch": 1.3950916004147944, "grad_norm": 0.6853274466730241, "learning_rate": 2.213602574847205e-06, "loss": 0.315, "step": 20180 }, { "epoch": 1.3951607328033184, "grad_norm": 0.7580791559966275, "learning_rate": 2.2131378102441464e-06, "loss": 0.3044, "step": 20181 }, { "epoch": 1.3952298651918424, "grad_norm": 0.6010084153983699, "learning_rate": 2.212673080569434e-06, "loss": 0.1966, "step": 20182 }, { "epoch": 1.3952989975803665, "grad_norm": 0.6897498604303219, "learning_rate": 2.212208385828889e-06, "loss": 0.323, "step": 20183 }, { "epoch": 1.3953681299688905, "grad_norm": 0.7059460566707859, "learning_rate": 2.2117437260283357e-06, "loss": 0.3666, "step": 20184 }, { "epoch": 1.3954372623574145, "grad_norm": 0.6454026250562976, "learning_rate": 2.2112791011736e-06, "loss": 0.2234, "step": 20185 }, { "epoch": 1.3955063947459385, "grad_norm": 0.8601118619124097, "learning_rate": 2.2108145112705054e-06, "loss": 0.1993, "step": 20186 }, { "epoch": 1.3955755271344625, "grad_norm": 0.712882333668123, "learning_rate": 2.210349956324872e-06, "loss": 0.3356, "step": 20187 }, { "epoch": 1.3956446595229866, "grad_norm": 0.7744079364460781, "learning_rate": 2.2098854363425248e-06, "loss": 0.2542, "step": 20188 }, { "epoch": 1.3957137919115106, "grad_norm": 0.6518642356003873, "learning_rate": 2.209420951329283e-06, "loss": 0.2405, "step": 20189 }, { "epoch": 1.3957829243000346, "grad_norm": 0.6854916811684387, "learning_rate": 2.2089565012909715e-06, "loss": 0.2785, "step": 20190 }, { "epoch": 1.3958520566885586, "grad_norm": 0.6566563855112438, "learning_rate": 2.208492086233407e-06, "loss": 0.3294, "step": 20191 }, { "epoch": 1.3959211890770826, "grad_norm": 0.6626269462967337, "learning_rate": 2.208027706162413e-06, "loss": 0.2749, "step": 20192 }, { "epoch": 1.3959903214656066, "grad_norm": 0.6093481858151942, "learning_rate": 2.2075633610838095e-06, "loss": 0.2379, "step": 20193 }, { "epoch": 1.3960594538541307, "grad_norm": 0.6835744238303407, "learning_rate": 2.207099051003417e-06, "loss": 0.329, "step": 20194 }, { "epoch": 1.3961285862426547, "grad_norm": 0.6847014635960189, "learning_rate": 2.2066347759270524e-06, "loss": 0.2949, "step": 20195 }, { "epoch": 1.3961977186311787, "grad_norm": 0.6611111028607445, "learning_rate": 2.2061705358605373e-06, "loss": 0.209, "step": 20196 }, { "epoch": 1.3962668510197027, "grad_norm": 0.7011467259655152, "learning_rate": 2.205706330809687e-06, "loss": 0.2345, "step": 20197 }, { "epoch": 1.3963359834082267, "grad_norm": 0.8006405876762779, "learning_rate": 2.2052421607803216e-06, "loss": 0.3757, "step": 20198 }, { "epoch": 1.3964051157967508, "grad_norm": 0.7086976414546928, "learning_rate": 2.204778025778258e-06, "loss": 0.2638, "step": 20199 }, { "epoch": 1.3964742481852748, "grad_norm": 0.6625345121540931, "learning_rate": 2.2043139258093156e-06, "loss": 0.2101, "step": 20200 }, { "epoch": 1.3965433805737988, "grad_norm": 0.6702471333457639, "learning_rate": 2.203849860879307e-06, "loss": 0.2748, "step": 20201 }, { "epoch": 1.3966125129623228, "grad_norm": 0.743796383217785, "learning_rate": 2.203385830994052e-06, "loss": 0.2848, "step": 20202 }, { "epoch": 1.3966816453508468, "grad_norm": 0.6230490024264088, "learning_rate": 2.2029218361593635e-06, "loss": 0.2356, "step": 20203 }, { "epoch": 1.3967507777393708, "grad_norm": 0.6642723619397188, "learning_rate": 2.20245787638106e-06, "loss": 0.236, "step": 20204 }, { "epoch": 1.3968199101278949, "grad_norm": 0.7253377321258776, "learning_rate": 2.2019939516649527e-06, "loss": 0.3113, "step": 20205 }, { "epoch": 1.3968890425164189, "grad_norm": 0.6773401191364206, "learning_rate": 2.2015300620168572e-06, "loss": 0.2602, "step": 20206 }, { "epoch": 1.396958174904943, "grad_norm": 0.6325843511147663, "learning_rate": 2.2010662074425894e-06, "loss": 0.2147, "step": 20207 }, { "epoch": 1.397027307293467, "grad_norm": 0.6966506571281179, "learning_rate": 2.2006023879479625e-06, "loss": 0.294, "step": 20208 }, { "epoch": 1.397096439681991, "grad_norm": 0.6905145364993994, "learning_rate": 2.2001386035387878e-06, "loss": 0.3609, "step": 20209 }, { "epoch": 1.3971655720705152, "grad_norm": 0.689512577665307, "learning_rate": 2.1996748542208807e-06, "loss": 0.197, "step": 20210 }, { "epoch": 1.3972347044590392, "grad_norm": 0.693031260817963, "learning_rate": 2.1992111400000506e-06, "loss": 0.2724, "step": 20211 }, { "epoch": 1.3973038368475632, "grad_norm": 0.6564897694499277, "learning_rate": 2.198747460882112e-06, "loss": 0.2898, "step": 20212 }, { "epoch": 1.3973729692360872, "grad_norm": 0.7119570764250972, "learning_rate": 2.1982838168728708e-06, "loss": 0.2364, "step": 20213 }, { "epoch": 1.3974421016246112, "grad_norm": 0.630585697369159, "learning_rate": 2.1978202079781465e-06, "loss": 0.2475, "step": 20214 }, { "epoch": 1.3975112340131353, "grad_norm": 0.6777883279295801, "learning_rate": 2.1973566342037427e-06, "loss": 0.2785, "step": 20215 }, { "epoch": 1.3975803664016593, "grad_norm": 0.8457715706001832, "learning_rate": 2.1968930955554746e-06, "loss": 0.3643, "step": 20216 }, { "epoch": 1.3976494987901833, "grad_norm": 0.7132039811275765, "learning_rate": 2.1964295920391465e-06, "loss": 0.2543, "step": 20217 }, { "epoch": 1.3977186311787073, "grad_norm": 0.620912070176874, "learning_rate": 2.195966123660573e-06, "loss": 0.2452, "step": 20218 }, { "epoch": 1.3977877635672313, "grad_norm": 0.6453209400195435, "learning_rate": 2.1955026904255576e-06, "loss": 0.2895, "step": 20219 }, { "epoch": 1.3978568959557554, "grad_norm": 0.7065783006436407, "learning_rate": 2.195039292339911e-06, "loss": 0.31, "step": 20220 }, { "epoch": 1.3979260283442794, "grad_norm": 0.7103369378218551, "learning_rate": 2.194575929409444e-06, "loss": 0.2239, "step": 20221 }, { "epoch": 1.3979951607328034, "grad_norm": 0.7080808462403593, "learning_rate": 2.194112601639958e-06, "loss": 0.2704, "step": 20222 }, { "epoch": 1.3980642931213274, "grad_norm": 0.6845915327094197, "learning_rate": 2.1936493090372657e-06, "loss": 0.3422, "step": 20223 }, { "epoch": 1.3981334255098514, "grad_norm": 0.8854008542677966, "learning_rate": 2.1931860516071697e-06, "loss": 0.2495, "step": 20224 }, { "epoch": 1.3982025578983754, "grad_norm": 0.6609755038376578, "learning_rate": 2.1927228293554793e-06, "loss": 0.2292, "step": 20225 }, { "epoch": 1.3982716902868995, "grad_norm": 0.6370447265739909, "learning_rate": 2.192259642287996e-06, "loss": 0.2503, "step": 20226 }, { "epoch": 1.3983408226754235, "grad_norm": 0.7179349453621351, "learning_rate": 2.1917964904105278e-06, "loss": 0.3348, "step": 20227 }, { "epoch": 1.3984099550639475, "grad_norm": 0.6328229352947785, "learning_rate": 2.1913333737288794e-06, "loss": 0.2356, "step": 20228 }, { "epoch": 1.3984790874524715, "grad_norm": 0.6371691002652488, "learning_rate": 2.190870292248856e-06, "loss": 0.2525, "step": 20229 }, { "epoch": 1.3985482198409955, "grad_norm": 0.6893232421468377, "learning_rate": 2.1904072459762594e-06, "loss": 0.337, "step": 20230 }, { "epoch": 1.3986173522295196, "grad_norm": 0.7545938197172438, "learning_rate": 2.1899442349168955e-06, "loss": 0.2883, "step": 20231 }, { "epoch": 1.3986864846180436, "grad_norm": 0.6779992240746149, "learning_rate": 2.1894812590765636e-06, "loss": 0.2646, "step": 20232 }, { "epoch": 1.3987556170065676, "grad_norm": 0.7376072531625276, "learning_rate": 2.189018318461069e-06, "loss": 0.3229, "step": 20233 }, { "epoch": 1.3988247493950916, "grad_norm": 0.7982083731604753, "learning_rate": 2.188555413076213e-06, "loss": 0.3405, "step": 20234 }, { "epoch": 1.3988938817836156, "grad_norm": 0.6789688168278118, "learning_rate": 2.1880925429277998e-06, "loss": 0.2242, "step": 20235 }, { "epoch": 1.3989630141721396, "grad_norm": 0.6798745101778321, "learning_rate": 2.1876297080216262e-06, "loss": 0.2271, "step": 20236 }, { "epoch": 1.3990321465606637, "grad_norm": 0.727287401047664, "learning_rate": 2.1871669083634976e-06, "loss": 0.3322, "step": 20237 }, { "epoch": 1.3991012789491877, "grad_norm": 0.7506987286628695, "learning_rate": 2.1867041439592107e-06, "loss": 0.2639, "step": 20238 }, { "epoch": 1.3991704113377117, "grad_norm": 0.7353452648309485, "learning_rate": 2.186241414814568e-06, "loss": 0.2552, "step": 20239 }, { "epoch": 1.3992395437262357, "grad_norm": 0.7526350955817983, "learning_rate": 2.185778720935366e-06, "loss": 0.293, "step": 20240 }, { "epoch": 1.3993086761147597, "grad_norm": 0.7270287855241935, "learning_rate": 2.1853160623274054e-06, "loss": 0.3649, "step": 20241 }, { "epoch": 1.3993778085032837, "grad_norm": 0.7068837525849337, "learning_rate": 2.1848534389964855e-06, "loss": 0.2764, "step": 20242 }, { "epoch": 1.3994469408918078, "grad_norm": 0.7859297614411155, "learning_rate": 2.1843908509484053e-06, "loss": 0.1895, "step": 20243 }, { "epoch": 1.3995160732803318, "grad_norm": 0.700110160848808, "learning_rate": 2.1839282981889594e-06, "loss": 0.3355, "step": 20244 }, { "epoch": 1.3995852056688558, "grad_norm": 0.6881157173544867, "learning_rate": 2.183465780723948e-06, "loss": 0.2455, "step": 20245 }, { "epoch": 1.3996543380573798, "grad_norm": 0.6316917530695267, "learning_rate": 2.1830032985591654e-06, "loss": 0.2342, "step": 20246 }, { "epoch": 1.3997234704459038, "grad_norm": 0.7203525049457549, "learning_rate": 2.182540851700411e-06, "loss": 0.2186, "step": 20247 }, { "epoch": 1.3997926028344279, "grad_norm": 0.7489916946672134, "learning_rate": 2.1820784401534773e-06, "loss": 0.3664, "step": 20248 }, { "epoch": 1.3998617352229519, "grad_norm": 0.8246317647796688, "learning_rate": 2.181616063924162e-06, "loss": 0.2602, "step": 20249 }, { "epoch": 1.399930867611476, "grad_norm": 0.6895476425784813, "learning_rate": 2.181153723018259e-06, "loss": 0.2982, "step": 20250 }, { "epoch": 1.4, "grad_norm": 0.7306773056759022, "learning_rate": 2.1806914174415657e-06, "loss": 0.2808, "step": 20251 }, { "epoch": 1.400069132388524, "grad_norm": 0.6835432313187714, "learning_rate": 2.1802291471998725e-06, "loss": 0.3202, "step": 20252 }, { "epoch": 1.400138264777048, "grad_norm": 0.6862034894228841, "learning_rate": 2.179766912298977e-06, "loss": 0.1953, "step": 20253 }, { "epoch": 1.400207397165572, "grad_norm": 0.7164002192005715, "learning_rate": 2.1793047127446675e-06, "loss": 0.2653, "step": 20254 }, { "epoch": 1.400276529554096, "grad_norm": 0.7051821365319462, "learning_rate": 2.17884254854274e-06, "loss": 0.3795, "step": 20255 }, { "epoch": 1.40034566194262, "grad_norm": 0.763399269862698, "learning_rate": 2.1783804196989872e-06, "loss": 0.2908, "step": 20256 }, { "epoch": 1.400414794331144, "grad_norm": 0.6131315801558753, "learning_rate": 2.1779183262192017e-06, "loss": 0.2194, "step": 20257 }, { "epoch": 1.400483926719668, "grad_norm": 0.6939333569426018, "learning_rate": 2.177456268109172e-06, "loss": 0.3494, "step": 20258 }, { "epoch": 1.400553059108192, "grad_norm": 0.7442288171841184, "learning_rate": 2.176994245374693e-06, "loss": 0.3413, "step": 20259 }, { "epoch": 1.400622191496716, "grad_norm": 0.6675117158824451, "learning_rate": 2.1765322580215512e-06, "loss": 0.2338, "step": 20260 }, { "epoch": 1.4006913238852403, "grad_norm": 0.6231432136404931, "learning_rate": 2.1760703060555407e-06, "loss": 0.2782, "step": 20261 }, { "epoch": 1.4007604562737643, "grad_norm": 0.7021999281054297, "learning_rate": 2.1756083894824477e-06, "loss": 0.3497, "step": 20262 }, { "epoch": 1.4008295886622883, "grad_norm": 0.6803981743004726, "learning_rate": 2.1751465083080632e-06, "loss": 0.3093, "step": 20263 }, { "epoch": 1.4008987210508124, "grad_norm": 0.6003352965074471, "learning_rate": 2.1746846625381757e-06, "loss": 0.2142, "step": 20264 }, { "epoch": 1.4009678534393364, "grad_norm": 0.6672262329035834, "learning_rate": 2.174222852178576e-06, "loss": 0.2335, "step": 20265 }, { "epoch": 1.4010369858278604, "grad_norm": 1.9172396090823303, "learning_rate": 2.1737610772350482e-06, "loss": 0.3857, "step": 20266 }, { "epoch": 1.4011061182163844, "grad_norm": 0.850264721829832, "learning_rate": 2.173299337713383e-06, "loss": 0.249, "step": 20267 }, { "epoch": 1.4011752506049084, "grad_norm": 0.8784821810004013, "learning_rate": 2.172837633619365e-06, "loss": 0.2508, "step": 20268 }, { "epoch": 1.4012443829934325, "grad_norm": 0.6821754164026328, "learning_rate": 2.172375964958782e-06, "loss": 0.3453, "step": 20269 }, { "epoch": 1.4013135153819565, "grad_norm": 0.7464125429581384, "learning_rate": 2.1719143317374196e-06, "loss": 0.2882, "step": 20270 }, { "epoch": 1.4013826477704805, "grad_norm": 0.7647590505314477, "learning_rate": 2.1714527339610663e-06, "loss": 0.3086, "step": 20271 }, { "epoch": 1.4014517801590045, "grad_norm": 0.6598176504603822, "learning_rate": 2.1709911716355033e-06, "loss": 0.2896, "step": 20272 }, { "epoch": 1.4015209125475285, "grad_norm": 0.7027520898110449, "learning_rate": 2.1705296447665193e-06, "loss": 0.3405, "step": 20273 }, { "epoch": 1.4015900449360525, "grad_norm": 0.6761903652620685, "learning_rate": 2.170068153359895e-06, "loss": 0.2574, "step": 20274 }, { "epoch": 1.4016591773245766, "grad_norm": 0.5942910672503446, "learning_rate": 2.1696066974214175e-06, "loss": 0.2114, "step": 20275 }, { "epoch": 1.4017283097131006, "grad_norm": 0.701032120115472, "learning_rate": 2.1691452769568677e-06, "loss": 0.2954, "step": 20276 }, { "epoch": 1.4017974421016246, "grad_norm": 0.6814998322034416, "learning_rate": 2.16868389197203e-06, "loss": 0.2863, "step": 20277 }, { "epoch": 1.4018665744901486, "grad_norm": 0.6871213978700796, "learning_rate": 2.1682225424726864e-06, "loss": 0.2509, "step": 20278 }, { "epoch": 1.4019357068786726, "grad_norm": 0.6255909689834591, "learning_rate": 2.167761228464622e-06, "loss": 0.275, "step": 20279 }, { "epoch": 1.4020048392671967, "grad_norm": 0.6993892527399582, "learning_rate": 2.167299949953614e-06, "loss": 0.383, "step": 20280 }, { "epoch": 1.4020739716557207, "grad_norm": 0.6675687153208715, "learning_rate": 2.166838706945448e-06, "loss": 0.2738, "step": 20281 }, { "epoch": 1.4021431040442447, "grad_norm": 0.646912458928974, "learning_rate": 2.1663774994459014e-06, "loss": 0.2638, "step": 20282 }, { "epoch": 1.4022122364327687, "grad_norm": 0.6591126511178685, "learning_rate": 2.1659163274607575e-06, "loss": 0.2454, "step": 20283 }, { "epoch": 1.4022813688212927, "grad_norm": 0.7051674335709243, "learning_rate": 2.165455190995791e-06, "loss": 0.3796, "step": 20284 }, { "epoch": 1.4023505012098167, "grad_norm": 0.6490162922657698, "learning_rate": 2.164994090056789e-06, "loss": 0.2491, "step": 20285 }, { "epoch": 1.4024196335983408, "grad_norm": 0.6555366309881008, "learning_rate": 2.164533024649525e-06, "loss": 0.2174, "step": 20286 }, { "epoch": 1.4024887659868648, "grad_norm": 0.6739305661851559, "learning_rate": 2.164071994779781e-06, "loss": 0.3718, "step": 20287 }, { "epoch": 1.4025578983753888, "grad_norm": 0.6602179263296579, "learning_rate": 2.1636110004533337e-06, "loss": 0.3108, "step": 20288 }, { "epoch": 1.4026270307639128, "grad_norm": 0.6716476314720544, "learning_rate": 2.1631500416759592e-06, "loss": 0.214, "step": 20289 }, { "epoch": 1.4026961631524368, "grad_norm": 0.6186275520887538, "learning_rate": 2.1626891184534364e-06, "loss": 0.2373, "step": 20290 }, { "epoch": 1.402765295540961, "grad_norm": 0.7239014006238893, "learning_rate": 2.162228230791542e-06, "loss": 0.423, "step": 20291 }, { "epoch": 1.402834427929485, "grad_norm": 0.6510612090357015, "learning_rate": 2.1617673786960542e-06, "loss": 0.2514, "step": 20292 }, { "epoch": 1.4029035603180091, "grad_norm": 0.669836032329796, "learning_rate": 2.1613065621727454e-06, "loss": 0.2948, "step": 20293 }, { "epoch": 1.4029726927065331, "grad_norm": 0.6597307877922389, "learning_rate": 2.1608457812273945e-06, "loss": 0.3079, "step": 20294 }, { "epoch": 1.4030418250950571, "grad_norm": 0.728392462671372, "learning_rate": 2.1603850358657737e-06, "loss": 0.3394, "step": 20295 }, { "epoch": 1.4031109574835812, "grad_norm": 0.6444348537451802, "learning_rate": 2.1599243260936605e-06, "loss": 0.2819, "step": 20296 }, { "epoch": 1.4031800898721052, "grad_norm": 0.7239398728471542, "learning_rate": 2.159463651916826e-06, "loss": 0.252, "step": 20297 }, { "epoch": 1.4032492222606292, "grad_norm": 0.6802820612724056, "learning_rate": 2.159003013341045e-06, "loss": 0.3413, "step": 20298 }, { "epoch": 1.4033183546491532, "grad_norm": 0.6866626528324363, "learning_rate": 2.1585424103720915e-06, "loss": 0.2576, "step": 20299 }, { "epoch": 1.4033874870376772, "grad_norm": 0.7401855743431164, "learning_rate": 2.15808184301574e-06, "loss": 0.2564, "step": 20300 }, { "epoch": 1.4034566194262013, "grad_norm": 0.6803366723969323, "learning_rate": 2.157621311277759e-06, "loss": 0.3091, "step": 20301 }, { "epoch": 1.4035257518147253, "grad_norm": 0.6799143800858379, "learning_rate": 2.1571608151639243e-06, "loss": 0.3092, "step": 20302 }, { "epoch": 1.4035948842032493, "grad_norm": 0.6839567391724513, "learning_rate": 2.156700354680004e-06, "loss": 0.2362, "step": 20303 }, { "epoch": 1.4036640165917733, "grad_norm": 0.6886534979563669, "learning_rate": 2.15623992983177e-06, "loss": 0.2566, "step": 20304 }, { "epoch": 1.4037331489802973, "grad_norm": 0.6865305139888174, "learning_rate": 2.155779540624994e-06, "loss": 0.4006, "step": 20305 }, { "epoch": 1.4038022813688213, "grad_norm": 0.7300924956602469, "learning_rate": 2.155319187065447e-06, "loss": 0.2486, "step": 20306 }, { "epoch": 1.4038714137573454, "grad_norm": 0.6180199606970727, "learning_rate": 2.1548588691588963e-06, "loss": 0.1889, "step": 20307 }, { "epoch": 1.4039405461458694, "grad_norm": 0.6848148405292435, "learning_rate": 2.1543985869111138e-06, "loss": 0.3246, "step": 20308 }, { "epoch": 1.4040096785343934, "grad_norm": 0.7453879399198553, "learning_rate": 2.1539383403278653e-06, "loss": 0.3833, "step": 20309 }, { "epoch": 1.4040788109229174, "grad_norm": 0.6462408881913152, "learning_rate": 2.153478129414922e-06, "loss": 0.2403, "step": 20310 }, { "epoch": 1.4041479433114414, "grad_norm": 0.8729466728337031, "learning_rate": 2.153017954178049e-06, "loss": 0.2059, "step": 20311 }, { "epoch": 1.4042170756999655, "grad_norm": 0.670666252954292, "learning_rate": 2.152557814623015e-06, "loss": 0.3321, "step": 20312 }, { "epoch": 1.4042862080884895, "grad_norm": 0.7533448738169963, "learning_rate": 2.1520977107555873e-06, "loss": 0.3138, "step": 20313 }, { "epoch": 1.4043553404770135, "grad_norm": 0.6746298809849052, "learning_rate": 2.151637642581535e-06, "loss": 0.234, "step": 20314 }, { "epoch": 1.4044244728655375, "grad_norm": 0.6546857295764229, "learning_rate": 2.151177610106619e-06, "loss": 0.2642, "step": 20315 }, { "epoch": 1.4044936052540615, "grad_norm": 0.6703702455300656, "learning_rate": 2.15071761333661e-06, "loss": 0.3682, "step": 20316 }, { "epoch": 1.4045627376425855, "grad_norm": 0.8320392511339025, "learning_rate": 2.1502576522772687e-06, "loss": 0.2937, "step": 20317 }, { "epoch": 1.4046318700311096, "grad_norm": 0.7322705396934279, "learning_rate": 2.1497977269343643e-06, "loss": 0.2464, "step": 20318 }, { "epoch": 1.4047010024196336, "grad_norm": 0.6694479028048351, "learning_rate": 2.1493378373136574e-06, "loss": 0.2883, "step": 20319 }, { "epoch": 1.4047701348081576, "grad_norm": 0.7246484027933937, "learning_rate": 2.1488779834209133e-06, "loss": 0.3237, "step": 20320 }, { "epoch": 1.4048392671966816, "grad_norm": 0.7610902737253172, "learning_rate": 2.1484181652618956e-06, "loss": 0.2224, "step": 20321 }, { "epoch": 1.4049083995852056, "grad_norm": 0.6800530605922693, "learning_rate": 2.147958382842369e-06, "loss": 0.2837, "step": 20322 }, { "epoch": 1.4049775319737297, "grad_norm": 0.6598272182073165, "learning_rate": 2.1474986361680927e-06, "loss": 0.3054, "step": 20323 }, { "epoch": 1.4050466643622537, "grad_norm": 0.7335538804889036, "learning_rate": 2.147038925244832e-06, "loss": 0.2787, "step": 20324 }, { "epoch": 1.4051157967507777, "grad_norm": 0.6284041719589222, "learning_rate": 2.1465792500783457e-06, "loss": 0.2453, "step": 20325 }, { "epoch": 1.4051849291393017, "grad_norm": 0.79793199913177, "learning_rate": 2.146119610674396e-06, "loss": 0.2594, "step": 20326 }, { "epoch": 1.4052540615278257, "grad_norm": 0.748388357995299, "learning_rate": 2.145660007038744e-06, "loss": 0.3736, "step": 20327 }, { "epoch": 1.4053231939163497, "grad_norm": 0.6784089697514639, "learning_rate": 2.145200439177152e-06, "loss": 0.2793, "step": 20328 }, { "epoch": 1.4053923263048738, "grad_norm": 0.6657674804584544, "learning_rate": 2.144740907095376e-06, "loss": 0.2407, "step": 20329 }, { "epoch": 1.4054614586933978, "grad_norm": 0.6717622187337, "learning_rate": 2.1442814107991795e-06, "loss": 0.3759, "step": 20330 }, { "epoch": 1.4055305910819218, "grad_norm": 0.760685501282159, "learning_rate": 2.1438219502943168e-06, "loss": 0.2896, "step": 20331 }, { "epoch": 1.4055997234704458, "grad_norm": 0.6502897191069154, "learning_rate": 2.143362525586551e-06, "loss": 0.2332, "step": 20332 }, { "epoch": 1.4056688558589698, "grad_norm": 0.6954605452212587, "learning_rate": 2.142903136681636e-06, "loss": 0.2955, "step": 20333 }, { "epoch": 1.4057379882474939, "grad_norm": 1.3264359126980718, "learning_rate": 2.1424437835853317e-06, "loss": 0.2946, "step": 20334 }, { "epoch": 1.4058071206360179, "grad_norm": 0.7778720216712036, "learning_rate": 2.141984466303395e-06, "loss": 0.2885, "step": 20335 }, { "epoch": 1.4058762530245419, "grad_norm": 0.7273977698384507, "learning_rate": 2.141525184841585e-06, "loss": 0.2175, "step": 20336 }, { "epoch": 1.405945385413066, "grad_norm": 0.6965616190492413, "learning_rate": 2.1410659392056528e-06, "loss": 0.3278, "step": 20337 }, { "epoch": 1.40601451780159, "grad_norm": 0.8108641273358739, "learning_rate": 2.1406067294013594e-06, "loss": 0.2858, "step": 20338 }, { "epoch": 1.406083650190114, "grad_norm": 0.6441171946518338, "learning_rate": 2.1401475554344558e-06, "loss": 0.2318, "step": 20339 }, { "epoch": 1.406152782578638, "grad_norm": 0.7616888336737843, "learning_rate": 2.139688417310699e-06, "loss": 0.2584, "step": 20340 }, { "epoch": 1.406221914967162, "grad_norm": 0.7048784550967753, "learning_rate": 2.1392293150358434e-06, "loss": 0.3178, "step": 20341 }, { "epoch": 1.4062910473556862, "grad_norm": 0.6939128534706024, "learning_rate": 2.1387702486156454e-06, "loss": 0.2816, "step": 20342 }, { "epoch": 1.4063601797442102, "grad_norm": 0.6867347254931642, "learning_rate": 2.1383112180558535e-06, "loss": 0.2108, "step": 20343 }, { "epoch": 1.4064293121327343, "grad_norm": 0.671170834998543, "learning_rate": 2.1378522233622263e-06, "loss": 0.3465, "step": 20344 }, { "epoch": 1.4064984445212583, "grad_norm": 0.6822269672448384, "learning_rate": 2.1373932645405114e-06, "loss": 0.3027, "step": 20345 }, { "epoch": 1.4065675769097823, "grad_norm": 0.6021252689687991, "learning_rate": 2.1369343415964655e-06, "loss": 0.2182, "step": 20346 }, { "epoch": 1.4066367092983063, "grad_norm": 0.7269963095526562, "learning_rate": 2.1364754545358363e-06, "loss": 0.256, "step": 20347 }, { "epoch": 1.4067058416868303, "grad_norm": 0.7739651783931719, "learning_rate": 2.136016603364377e-06, "loss": 0.3901, "step": 20348 }, { "epoch": 1.4067749740753543, "grad_norm": 0.716326434651488, "learning_rate": 2.1355577880878387e-06, "loss": 0.2887, "step": 20349 }, { "epoch": 1.4068441064638784, "grad_norm": 0.7507753881434729, "learning_rate": 2.1350990087119734e-06, "loss": 0.2831, "step": 20350 }, { "epoch": 1.4069132388524024, "grad_norm": 0.6897875078284939, "learning_rate": 2.1346402652425273e-06, "loss": 0.2688, "step": 20351 }, { "epoch": 1.4069823712409264, "grad_norm": 0.6868095912670631, "learning_rate": 2.134181557685254e-06, "loss": 0.3219, "step": 20352 }, { "epoch": 1.4070515036294504, "grad_norm": 0.661786470812358, "learning_rate": 2.133722886045901e-06, "loss": 0.2496, "step": 20353 }, { "epoch": 1.4071206360179744, "grad_norm": 0.7022538907060293, "learning_rate": 2.1332642503302146e-06, "loss": 0.2476, "step": 20354 }, { "epoch": 1.4071897684064985, "grad_norm": 0.8890080569506572, "learning_rate": 2.1328056505439448e-06, "loss": 0.3387, "step": 20355 }, { "epoch": 1.4072589007950225, "grad_norm": 0.7190014636123994, "learning_rate": 2.1323470866928393e-06, "loss": 0.2697, "step": 20356 }, { "epoch": 1.4073280331835465, "grad_norm": 0.6471598136534962, "learning_rate": 2.131888558782647e-06, "loss": 0.2476, "step": 20357 }, { "epoch": 1.4073971655720705, "grad_norm": 0.7070767801275626, "learning_rate": 2.1314300668191122e-06, "loss": 0.2809, "step": 20358 }, { "epoch": 1.4074662979605945, "grad_norm": 0.6420396969198255, "learning_rate": 2.1309716108079836e-06, "loss": 0.3241, "step": 20359 }, { "epoch": 1.4075354303491185, "grad_norm": 0.6483577067343526, "learning_rate": 2.130513190755004e-06, "loss": 0.1939, "step": 20360 }, { "epoch": 1.4076045627376426, "grad_norm": 0.6858775815838604, "learning_rate": 2.130054806665921e-06, "loss": 0.2816, "step": 20361 }, { "epoch": 1.4076736951261666, "grad_norm": 0.6344197921739789, "learning_rate": 2.1295964585464794e-06, "loss": 0.2924, "step": 20362 }, { "epoch": 1.4077428275146906, "grad_norm": 0.7146052023428449, "learning_rate": 2.129138146402426e-06, "loss": 0.287, "step": 20363 }, { "epoch": 1.4078119599032146, "grad_norm": 0.8385679264866921, "learning_rate": 2.1286798702394995e-06, "loss": 0.2362, "step": 20364 }, { "epoch": 1.4078810922917386, "grad_norm": 0.6792472484056253, "learning_rate": 2.1282216300634497e-06, "loss": 0.2187, "step": 20365 }, { "epoch": 1.4079502246802627, "grad_norm": 0.6481561442240577, "learning_rate": 2.1277634258800147e-06, "loss": 0.3699, "step": 20366 }, { "epoch": 1.4080193570687867, "grad_norm": 0.6840917020057562, "learning_rate": 2.127305257694941e-06, "loss": 0.2772, "step": 20367 }, { "epoch": 1.4080884894573107, "grad_norm": 0.6240939033578399, "learning_rate": 2.1268471255139676e-06, "loss": 0.2052, "step": 20368 }, { "epoch": 1.4081576218458347, "grad_norm": 0.7470092368687122, "learning_rate": 2.126389029342838e-06, "loss": 0.3499, "step": 20369 }, { "epoch": 1.4082267542343587, "grad_norm": 0.6686190241700519, "learning_rate": 2.125930969187294e-06, "loss": 0.2966, "step": 20370 }, { "epoch": 1.4082958866228827, "grad_norm": 0.6221111550919961, "learning_rate": 2.1254729450530783e-06, "loss": 0.2633, "step": 20371 }, { "epoch": 1.408365019011407, "grad_norm": 0.6589049064283412, "learning_rate": 2.125014956945927e-06, "loss": 0.2389, "step": 20372 }, { "epoch": 1.408434151399931, "grad_norm": 0.6544089696747923, "learning_rate": 2.1245570048715845e-06, "loss": 0.3252, "step": 20373 }, { "epoch": 1.408503283788455, "grad_norm": 0.7135876100177037, "learning_rate": 2.124099088835787e-06, "loss": 0.2505, "step": 20374 }, { "epoch": 1.408572416176979, "grad_norm": 0.6126810550836144, "learning_rate": 2.1236412088442753e-06, "loss": 0.1829, "step": 20375 }, { "epoch": 1.408641548565503, "grad_norm": 0.6956195843026394, "learning_rate": 2.1231833649027877e-06, "loss": 0.317, "step": 20376 }, { "epoch": 1.408710680954027, "grad_norm": 0.6750673863397416, "learning_rate": 2.1227255570170646e-06, "loss": 0.2611, "step": 20377 }, { "epoch": 1.408779813342551, "grad_norm": 0.6241380285798312, "learning_rate": 2.1222677851928397e-06, "loss": 0.2455, "step": 20378 }, { "epoch": 1.408848945731075, "grad_norm": 0.6223327963430688, "learning_rate": 2.1218100494358553e-06, "loss": 0.207, "step": 20379 }, { "epoch": 1.4089180781195991, "grad_norm": 0.7072086011786936, "learning_rate": 2.121352349751843e-06, "loss": 0.3848, "step": 20380 }, { "epoch": 1.4089872105081231, "grad_norm": 0.7007003465399884, "learning_rate": 2.1208946861465434e-06, "loss": 0.2102, "step": 20381 }, { "epoch": 1.4090563428966472, "grad_norm": 0.6708855681666515, "learning_rate": 2.1204370586256893e-06, "loss": 0.2255, "step": 20382 }, { "epoch": 1.4091254752851712, "grad_norm": 0.7200132199031885, "learning_rate": 2.1199794671950185e-06, "loss": 0.3072, "step": 20383 }, { "epoch": 1.4091946076736952, "grad_norm": 1.1290994915861645, "learning_rate": 2.119521911860265e-06, "loss": 0.3899, "step": 20384 }, { "epoch": 1.4092637400622192, "grad_norm": 0.7193482767555374, "learning_rate": 2.1190643926271654e-06, "loss": 0.2853, "step": 20385 }, { "epoch": 1.4093328724507432, "grad_norm": 0.6685586335819336, "learning_rate": 2.118606909501451e-06, "loss": 0.2635, "step": 20386 }, { "epoch": 1.4094020048392673, "grad_norm": 0.6889545417344628, "learning_rate": 2.118149462488859e-06, "loss": 0.3517, "step": 20387 }, { "epoch": 1.4094711372277913, "grad_norm": 0.6961550574783687, "learning_rate": 2.1176920515951183e-06, "loss": 0.2778, "step": 20388 }, { "epoch": 1.4095402696163153, "grad_norm": 0.9840248074739959, "learning_rate": 2.1172346768259654e-06, "loss": 0.2327, "step": 20389 }, { "epoch": 1.4096094020048393, "grad_norm": 0.7239709211524527, "learning_rate": 2.1167773381871303e-06, "loss": 0.3231, "step": 20390 }, { "epoch": 1.4096785343933633, "grad_norm": 0.8987150869931179, "learning_rate": 2.1163200356843453e-06, "loss": 0.3426, "step": 20391 }, { "epoch": 1.4097476667818873, "grad_norm": 0.7573704263618266, "learning_rate": 2.115862769323343e-06, "loss": 0.2415, "step": 20392 }, { "epoch": 1.4098167991704114, "grad_norm": 0.7469905959025629, "learning_rate": 2.1154055391098555e-06, "loss": 0.2569, "step": 20393 }, { "epoch": 1.4098859315589354, "grad_norm": 0.7201138113869909, "learning_rate": 2.11494834504961e-06, "loss": 0.3034, "step": 20394 }, { "epoch": 1.4099550639474594, "grad_norm": 0.6916535090305367, "learning_rate": 2.11449118714834e-06, "loss": 0.3406, "step": 20395 }, { "epoch": 1.4100241963359834, "grad_norm": 0.6918411046354311, "learning_rate": 2.1140340654117726e-06, "loss": 0.2874, "step": 20396 }, { "epoch": 1.4100933287245074, "grad_norm": 0.6932611568959707, "learning_rate": 2.1135769798456366e-06, "loss": 0.2887, "step": 20397 }, { "epoch": 1.4101624611130315, "grad_norm": 0.7511239479877072, "learning_rate": 2.1131199304556633e-06, "loss": 0.2854, "step": 20398 }, { "epoch": 1.4102315935015555, "grad_norm": 0.7517533524031635, "learning_rate": 2.112662917247581e-06, "loss": 0.2196, "step": 20399 }, { "epoch": 1.4103007258900795, "grad_norm": 0.6653437049685929, "learning_rate": 2.1122059402271143e-06, "loss": 0.2461, "step": 20400 }, { "epoch": 1.4103698582786035, "grad_norm": 0.7426115136250357, "learning_rate": 2.1117489993999952e-06, "loss": 0.3259, "step": 20401 }, { "epoch": 1.4104389906671275, "grad_norm": 0.700015495065431, "learning_rate": 2.1112920947719455e-06, "loss": 0.2999, "step": 20402 }, { "epoch": 1.4105081230556515, "grad_norm": 0.6645414218224421, "learning_rate": 2.110835226348697e-06, "loss": 0.2561, "step": 20403 }, { "epoch": 1.4105772554441756, "grad_norm": 0.6081052076294777, "learning_rate": 2.1103783941359703e-06, "loss": 0.2618, "step": 20404 }, { "epoch": 1.4106463878326996, "grad_norm": 0.9384098139687577, "learning_rate": 2.1099215981394937e-06, "loss": 0.3492, "step": 20405 }, { "epoch": 1.4107155202212236, "grad_norm": 0.680240434604507, "learning_rate": 2.109464838364993e-06, "loss": 0.2465, "step": 20406 }, { "epoch": 1.4107846526097476, "grad_norm": 0.6951566003583747, "learning_rate": 2.1090081148181936e-06, "loss": 0.2272, "step": 20407 }, { "epoch": 1.4108537849982716, "grad_norm": 0.6563913760084623, "learning_rate": 2.1085514275048165e-06, "loss": 0.2214, "step": 20408 }, { "epoch": 1.4109229173867956, "grad_norm": 0.6832670294376573, "learning_rate": 2.1080947764305887e-06, "loss": 0.3432, "step": 20409 }, { "epoch": 1.4109920497753197, "grad_norm": 0.5785159313521858, "learning_rate": 2.1076381616012305e-06, "loss": 0.206, "step": 20410 }, { "epoch": 1.4110611821638437, "grad_norm": 0.6345839572332989, "learning_rate": 2.1071815830224664e-06, "loss": 0.2157, "step": 20411 }, { "epoch": 1.4111303145523677, "grad_norm": 0.6894459316158306, "learning_rate": 2.1067250407000184e-06, "loss": 0.3296, "step": 20412 }, { "epoch": 1.4111994469408917, "grad_norm": 0.6365616585083133, "learning_rate": 2.1062685346396107e-06, "loss": 0.2726, "step": 20413 }, { "epoch": 1.4112685793294157, "grad_norm": 0.6712687464115971, "learning_rate": 2.105812064846961e-06, "loss": 0.2211, "step": 20414 }, { "epoch": 1.4113377117179398, "grad_norm": 0.7150300386355375, "learning_rate": 2.105355631327794e-06, "loss": 0.2507, "step": 20415 }, { "epoch": 1.4114068441064638, "grad_norm": 0.7102098604317529, "learning_rate": 2.1048992340878267e-06, "loss": 0.3649, "step": 20416 }, { "epoch": 1.4114759764949878, "grad_norm": 0.7646215036722481, "learning_rate": 2.1044428731327827e-06, "loss": 0.3073, "step": 20417 }, { "epoch": 1.4115451088835118, "grad_norm": 0.6211003940824646, "learning_rate": 2.103986548468378e-06, "loss": 0.2417, "step": 20418 }, { "epoch": 1.4116142412720358, "grad_norm": 0.6794547838360582, "learning_rate": 2.103530260100334e-06, "loss": 0.344, "step": 20419 }, { "epoch": 1.4116833736605598, "grad_norm": 0.7438765291123735, "learning_rate": 2.103074008034371e-06, "loss": 0.2465, "step": 20420 }, { "epoch": 1.4117525060490839, "grad_norm": 0.6265532747669701, "learning_rate": 2.1026177922762036e-06, "loss": 0.2112, "step": 20421 }, { "epoch": 1.4118216384376079, "grad_norm": 0.6611083841781364, "learning_rate": 2.102161612831553e-06, "loss": 0.252, "step": 20422 }, { "epoch": 1.4118907708261321, "grad_norm": 0.7135122883844457, "learning_rate": 2.101705469706134e-06, "loss": 0.3181, "step": 20423 }, { "epoch": 1.4119599032146561, "grad_norm": 0.7339014338752936, "learning_rate": 2.1012493629056663e-06, "loss": 0.2653, "step": 20424 }, { "epoch": 1.4120290356031802, "grad_norm": 0.7322361522594791, "learning_rate": 2.1007932924358625e-06, "loss": 0.2654, "step": 20425 }, { "epoch": 1.4120981679917042, "grad_norm": 0.6715706954649966, "learning_rate": 2.100337258302442e-06, "loss": 0.2838, "step": 20426 }, { "epoch": 1.4121673003802282, "grad_norm": 0.6502310471483419, "learning_rate": 2.099881260511118e-06, "loss": 0.3033, "step": 20427 }, { "epoch": 1.4122364327687522, "grad_norm": 0.7774817072465043, "learning_rate": 2.0994252990676095e-06, "loss": 0.2451, "step": 20428 }, { "epoch": 1.4123055651572762, "grad_norm": 0.6289632895891792, "learning_rate": 2.0989693739776263e-06, "loss": 0.2548, "step": 20429 }, { "epoch": 1.4123746975458003, "grad_norm": 0.6631428565801745, "learning_rate": 2.098513485246887e-06, "loss": 0.3225, "step": 20430 }, { "epoch": 1.4124438299343243, "grad_norm": 0.7048017845822648, "learning_rate": 2.0980576328811014e-06, "loss": 0.2376, "step": 20431 }, { "epoch": 1.4125129623228483, "grad_norm": 0.7090371853409714, "learning_rate": 2.0976018168859845e-06, "loss": 0.2352, "step": 20432 }, { "epoch": 1.4125820947113723, "grad_norm": 0.6711570663581292, "learning_rate": 2.0971460372672493e-06, "loss": 0.3093, "step": 20433 }, { "epoch": 1.4126512270998963, "grad_norm": 0.7173519225157261, "learning_rate": 2.09669029403061e-06, "loss": 0.3424, "step": 20434 }, { "epoch": 1.4127203594884203, "grad_norm": 0.6974577785395116, "learning_rate": 2.096234587181775e-06, "loss": 0.2971, "step": 20435 }, { "epoch": 1.4127894918769444, "grad_norm": 0.7128810082687643, "learning_rate": 2.095778916726459e-06, "loss": 0.2546, "step": 20436 }, { "epoch": 1.4128586242654684, "grad_norm": 0.6971176074426055, "learning_rate": 2.0953232826703703e-06, "loss": 0.3475, "step": 20437 }, { "epoch": 1.4129277566539924, "grad_norm": 0.7322800921426096, "learning_rate": 2.0948676850192213e-06, "loss": 0.3079, "step": 20438 }, { "epoch": 1.4129968890425164, "grad_norm": 0.662221004805022, "learning_rate": 2.09441212377872e-06, "loss": 0.2101, "step": 20439 }, { "epoch": 1.4130660214310404, "grad_norm": 0.6840665830973361, "learning_rate": 2.093956598954578e-06, "loss": 0.2855, "step": 20440 }, { "epoch": 1.4131351538195644, "grad_norm": 0.698015285143426, "learning_rate": 2.093501110552504e-06, "loss": 0.3639, "step": 20441 }, { "epoch": 1.4132042862080885, "grad_norm": 0.696164004971701, "learning_rate": 2.093045658578208e-06, "loss": 0.2282, "step": 20442 }, { "epoch": 1.4132734185966125, "grad_norm": 0.7195711361061902, "learning_rate": 2.092590243037395e-06, "loss": 0.2691, "step": 20443 }, { "epoch": 1.4133425509851365, "grad_norm": 0.6704345570465465, "learning_rate": 2.0921348639357776e-06, "loss": 0.331, "step": 20444 }, { "epoch": 1.4134116833736605, "grad_norm": 0.7312868575448613, "learning_rate": 2.0916795212790575e-06, "loss": 0.3192, "step": 20445 }, { "epoch": 1.4134808157621845, "grad_norm": 0.7626899384803195, "learning_rate": 2.091224215072945e-06, "loss": 0.186, "step": 20446 }, { "epoch": 1.4135499481507086, "grad_norm": 0.68758125403106, "learning_rate": 2.0907689453231455e-06, "loss": 0.2494, "step": 20447 }, { "epoch": 1.4136190805392326, "grad_norm": 0.6863224414888933, "learning_rate": 2.0903137120353678e-06, "loss": 0.3698, "step": 20448 }, { "epoch": 1.4136882129277566, "grad_norm": 1.955220164913584, "learning_rate": 2.0898585152153133e-06, "loss": 0.2974, "step": 20449 }, { "epoch": 1.4137573453162806, "grad_norm": 0.6148984404892556, "learning_rate": 2.089403354868691e-06, "loss": 0.2432, "step": 20450 }, { "epoch": 1.4138264777048046, "grad_norm": 0.6901270872178961, "learning_rate": 2.088948231001201e-06, "loss": 0.338, "step": 20451 }, { "epoch": 1.4138956100933286, "grad_norm": 0.6784686622629843, "learning_rate": 2.0884931436185524e-06, "loss": 0.333, "step": 20452 }, { "epoch": 1.4139647424818527, "grad_norm": 0.6740646077091684, "learning_rate": 2.088038092726444e-06, "loss": 0.2204, "step": 20453 }, { "epoch": 1.414033874870377, "grad_norm": 0.6197596088555435, "learning_rate": 2.0875830783305817e-06, "loss": 0.199, "step": 20454 }, { "epoch": 1.414103007258901, "grad_norm": 0.6784644993766108, "learning_rate": 2.0871281004366685e-06, "loss": 0.3464, "step": 20455 }, { "epoch": 1.414172139647425, "grad_norm": 0.6567156775891823, "learning_rate": 2.0866731590504074e-06, "loss": 0.2876, "step": 20456 }, { "epoch": 1.414241272035949, "grad_norm": 0.6912757228560247, "learning_rate": 2.086218254177498e-06, "loss": 0.2264, "step": 20457 }, { "epoch": 1.414310404424473, "grad_norm": 0.6880329605775458, "learning_rate": 2.0857633858236443e-06, "loss": 0.3147, "step": 20458 }, { "epoch": 1.414379536812997, "grad_norm": 0.6710779038766002, "learning_rate": 2.0853085539945443e-06, "loss": 0.3552, "step": 20459 }, { "epoch": 1.414448669201521, "grad_norm": 0.6286429262440972, "learning_rate": 2.084853758695902e-06, "loss": 0.2241, "step": 20460 }, { "epoch": 1.414517801590045, "grad_norm": 0.6323143088322369, "learning_rate": 2.0843989999334136e-06, "loss": 0.2048, "step": 20461 }, { "epoch": 1.414586933978569, "grad_norm": 0.7249065879488171, "learning_rate": 2.0839442777127804e-06, "loss": 0.3637, "step": 20462 }, { "epoch": 1.414656066367093, "grad_norm": 0.6889493809681907, "learning_rate": 2.083489592039702e-06, "loss": 0.3431, "step": 20463 }, { "epoch": 1.414725198755617, "grad_norm": 0.6384505812700371, "learning_rate": 2.0830349429198783e-06, "loss": 0.1804, "step": 20464 }, { "epoch": 1.414794331144141, "grad_norm": 0.7396906917959064, "learning_rate": 2.082580330359004e-06, "loss": 0.2426, "step": 20465 }, { "epoch": 1.4148634635326651, "grad_norm": 0.6518030757245216, "learning_rate": 2.0821257543627816e-06, "loss": 0.3309, "step": 20466 }, { "epoch": 1.4149325959211891, "grad_norm": 0.7012402230493843, "learning_rate": 2.0816712149369034e-06, "loss": 0.2597, "step": 20467 }, { "epoch": 1.4150017283097132, "grad_norm": 0.6687033277878025, "learning_rate": 2.0812167120870684e-06, "loss": 0.2306, "step": 20468 }, { "epoch": 1.4150708606982372, "grad_norm": 0.6571690268883745, "learning_rate": 2.080762245818973e-06, "loss": 0.3074, "step": 20469 }, { "epoch": 1.4151399930867612, "grad_norm": 0.7797243861008137, "learning_rate": 2.0803078161383156e-06, "loss": 0.3576, "step": 20470 }, { "epoch": 1.4152091254752852, "grad_norm": 0.7411546486182993, "learning_rate": 2.079853423050787e-06, "loss": 0.2987, "step": 20471 }, { "epoch": 1.4152782578638092, "grad_norm": 0.9578864974597144, "learning_rate": 2.0793990665620866e-06, "loss": 0.2359, "step": 20472 }, { "epoch": 1.4153473902523332, "grad_norm": 0.658341871241085, "learning_rate": 2.078944746677905e-06, "loss": 0.3203, "step": 20473 }, { "epoch": 1.4154165226408573, "grad_norm": 0.6387521617166702, "learning_rate": 2.0784904634039398e-06, "loss": 0.2573, "step": 20474 }, { "epoch": 1.4154856550293813, "grad_norm": 0.644673548354487, "learning_rate": 2.078036216745881e-06, "loss": 0.2386, "step": 20475 }, { "epoch": 1.4155547874179053, "grad_norm": 0.6143090565130431, "learning_rate": 2.0775820067094244e-06, "loss": 0.2812, "step": 20476 }, { "epoch": 1.4156239198064293, "grad_norm": 0.7110394221846862, "learning_rate": 2.0771278333002614e-06, "loss": 0.349, "step": 20477 }, { "epoch": 1.4156930521949533, "grad_norm": 0.7541446535955302, "learning_rate": 2.076673696524087e-06, "loss": 0.2904, "step": 20478 }, { "epoch": 1.4157621845834774, "grad_norm": 0.6749339955318774, "learning_rate": 2.076219596386589e-06, "loss": 0.3009, "step": 20479 }, { "epoch": 1.4158313169720014, "grad_norm": 0.6717267522611385, "learning_rate": 2.0757655328934622e-06, "loss": 0.384, "step": 20480 }, { "epoch": 1.4159004493605254, "grad_norm": 0.6367178813329144, "learning_rate": 2.075311506050395e-06, "loss": 0.2683, "step": 20481 }, { "epoch": 1.4159695817490494, "grad_norm": 0.6915976843114069, "learning_rate": 2.074857515863078e-06, "loss": 0.2632, "step": 20482 }, { "epoch": 1.4160387141375734, "grad_norm": 0.652457762765466, "learning_rate": 2.0744035623372017e-06, "loss": 0.2602, "step": 20483 }, { "epoch": 1.4161078465260974, "grad_norm": 0.7023017998565395, "learning_rate": 2.073949645478458e-06, "loss": 0.3437, "step": 20484 }, { "epoch": 1.4161769789146215, "grad_norm": 0.7244568062311247, "learning_rate": 2.073495765292532e-06, "loss": 0.239, "step": 20485 }, { "epoch": 1.4162461113031455, "grad_norm": 0.6341867145867612, "learning_rate": 2.0730419217851155e-06, "loss": 0.2497, "step": 20486 }, { "epoch": 1.4163152436916695, "grad_norm": 0.7421544305913521, "learning_rate": 2.0725881149618954e-06, "loss": 0.3068, "step": 20487 }, { "epoch": 1.4163843760801935, "grad_norm": 0.6990601771126437, "learning_rate": 2.0721343448285567e-06, "loss": 0.3124, "step": 20488 }, { "epoch": 1.4164535084687175, "grad_norm": 0.6496607159251028, "learning_rate": 2.07168061139079e-06, "loss": 0.2797, "step": 20489 }, { "epoch": 1.4165226408572416, "grad_norm": 0.6964417754889912, "learning_rate": 2.07122691465428e-06, "loss": 0.3016, "step": 20490 }, { "epoch": 1.4165917732457656, "grad_norm": 0.719509085211321, "learning_rate": 2.0707732546247168e-06, "loss": 0.3728, "step": 20491 }, { "epoch": 1.4166609056342896, "grad_norm": 0.6872010141867182, "learning_rate": 2.0703196313077805e-06, "loss": 0.2719, "step": 20492 }, { "epoch": 1.4167300380228136, "grad_norm": 0.7487356552117224, "learning_rate": 2.0698660447091618e-06, "loss": 0.3163, "step": 20493 }, { "epoch": 1.4167991704113376, "grad_norm": 0.7316673738171583, "learning_rate": 2.069412494834541e-06, "loss": 0.2832, "step": 20494 }, { "epoch": 1.4168683027998616, "grad_norm": 0.7277012705562615, "learning_rate": 2.068958981689607e-06, "loss": 0.3318, "step": 20495 }, { "epoch": 1.4169374351883857, "grad_norm": 0.7354205674266363, "learning_rate": 2.0685055052800394e-06, "loss": 0.2317, "step": 20496 }, { "epoch": 1.4170065675769097, "grad_norm": 0.749882759194849, "learning_rate": 2.0680520656115246e-06, "loss": 0.1938, "step": 20497 }, { "epoch": 1.4170756999654337, "grad_norm": 0.6956193293298751, "learning_rate": 2.067598662689744e-06, "loss": 0.3071, "step": 20498 }, { "epoch": 1.4171448323539577, "grad_norm": 0.6297172344856883, "learning_rate": 2.0671452965203824e-06, "loss": 0.2482, "step": 20499 }, { "epoch": 1.4172139647424817, "grad_norm": 0.6320835620736317, "learning_rate": 2.0666919671091197e-06, "loss": 0.2557, "step": 20500 }, { "epoch": 1.4172830971310058, "grad_norm": 0.6681658519381112, "learning_rate": 2.06623867446164e-06, "loss": 0.2597, "step": 20501 }, { "epoch": 1.4173522295195298, "grad_norm": 0.6532671765529161, "learning_rate": 2.0657854185836214e-06, "loss": 0.3054, "step": 20502 }, { "epoch": 1.4174213619080538, "grad_norm": 0.7014745284095174, "learning_rate": 2.065332199480746e-06, "loss": 0.2446, "step": 20503 }, { "epoch": 1.417490494296578, "grad_norm": 0.6524830198362107, "learning_rate": 2.064879017158694e-06, "loss": 0.1944, "step": 20504 }, { "epoch": 1.417559626685102, "grad_norm": 0.6561324036075333, "learning_rate": 2.0644258716231473e-06, "loss": 0.3146, "step": 20505 }, { "epoch": 1.417628759073626, "grad_norm": 0.7858577977542812, "learning_rate": 2.0639727628797824e-06, "loss": 0.2329, "step": 20506 }, { "epoch": 1.41769789146215, "grad_norm": 0.6456041060710905, "learning_rate": 2.063519690934281e-06, "loss": 0.2566, "step": 20507 }, { "epoch": 1.417767023850674, "grad_norm": 0.6662018895958691, "learning_rate": 2.063066655792318e-06, "loss": 0.2849, "step": 20508 }, { "epoch": 1.4178361562391981, "grad_norm": 0.7361574423537197, "learning_rate": 2.062613657459575e-06, "loss": 0.3942, "step": 20509 }, { "epoch": 1.4179052886277221, "grad_norm": 0.6337170672777247, "learning_rate": 2.0621606959417255e-06, "loss": 0.2447, "step": 20510 }, { "epoch": 1.4179744210162462, "grad_norm": 0.6298747092892235, "learning_rate": 2.06170777124445e-06, "loss": 0.2388, "step": 20511 }, { "epoch": 1.4180435534047702, "grad_norm": 0.6666320705315555, "learning_rate": 2.061254883373423e-06, "loss": 0.3284, "step": 20512 }, { "epoch": 1.4181126857932942, "grad_norm": 0.7535491099087445, "learning_rate": 2.0608020323343235e-06, "loss": 0.294, "step": 20513 }, { "epoch": 1.4181818181818182, "grad_norm": 0.6916396342333925, "learning_rate": 2.0603492181328245e-06, "loss": 0.2821, "step": 20514 }, { "epoch": 1.4182509505703422, "grad_norm": 0.766223825696327, "learning_rate": 2.0598964407746025e-06, "loss": 0.2717, "step": 20515 }, { "epoch": 1.4183200829588662, "grad_norm": 0.6754923038888987, "learning_rate": 2.0594437002653304e-06, "loss": 0.3778, "step": 20516 }, { "epoch": 1.4183892153473903, "grad_norm": 0.6119422719990959, "learning_rate": 2.058990996610684e-06, "loss": 0.2704, "step": 20517 }, { "epoch": 1.4184583477359143, "grad_norm": 0.692629835728206, "learning_rate": 2.0585383298163376e-06, "loss": 0.2407, "step": 20518 }, { "epoch": 1.4185274801244383, "grad_norm": 0.6394886748425804, "learning_rate": 2.058085699887965e-06, "loss": 0.3334, "step": 20519 }, { "epoch": 1.4185966125129623, "grad_norm": 0.7002779369425066, "learning_rate": 2.0576331068312362e-06, "loss": 0.324, "step": 20520 }, { "epoch": 1.4186657449014863, "grad_norm": 0.7121951857607242, "learning_rate": 2.0571805506518278e-06, "loss": 0.2178, "step": 20521 }, { "epoch": 1.4187348772900104, "grad_norm": 0.9753666966051067, "learning_rate": 2.0567280313554073e-06, "loss": 0.2867, "step": 20522 }, { "epoch": 1.4188040096785344, "grad_norm": 0.6769154090931874, "learning_rate": 2.0562755489476507e-06, "loss": 0.4162, "step": 20523 }, { "epoch": 1.4188731420670584, "grad_norm": 0.678621598374531, "learning_rate": 2.055823103434224e-06, "loss": 0.2061, "step": 20524 }, { "epoch": 1.4189422744555824, "grad_norm": 0.6475192027777067, "learning_rate": 2.0553706948208012e-06, "loss": 0.2267, "step": 20525 }, { "epoch": 1.4190114068441064, "grad_norm": 0.7768920373790202, "learning_rate": 2.054918323113052e-06, "loss": 0.303, "step": 20526 }, { "epoch": 1.4190805392326304, "grad_norm": 0.7328087028998932, "learning_rate": 2.0544659883166474e-06, "loss": 0.3358, "step": 20527 }, { "epoch": 1.4191496716211545, "grad_norm": 0.6952190345681115, "learning_rate": 2.0540136904372537e-06, "loss": 0.2172, "step": 20528 }, { "epoch": 1.4192188040096785, "grad_norm": 0.6150324552566268, "learning_rate": 2.0535614294805424e-06, "loss": 0.2788, "step": 20529 }, { "epoch": 1.4192879363982025, "grad_norm": 0.8368198787024411, "learning_rate": 2.053109205452179e-06, "loss": 0.3881, "step": 20530 }, { "epoch": 1.4193570687867265, "grad_norm": 0.7055242442372597, "learning_rate": 2.052657018357834e-06, "loss": 0.3453, "step": 20531 }, { "epoch": 1.4194262011752505, "grad_norm": 0.6958601866374668, "learning_rate": 2.052204868203172e-06, "loss": 0.2316, "step": 20532 }, { "epoch": 1.4194953335637746, "grad_norm": 0.689712519310876, "learning_rate": 2.0517527549938614e-06, "loss": 0.2737, "step": 20533 }, { "epoch": 1.4195644659522986, "grad_norm": 0.7141246197391843, "learning_rate": 2.0513006787355688e-06, "loss": 0.3317, "step": 20534 }, { "epoch": 1.4196335983408228, "grad_norm": 0.6256725907605206, "learning_rate": 2.0508486394339615e-06, "loss": 0.2391, "step": 20535 }, { "epoch": 1.4197027307293468, "grad_norm": 0.6558833239579702, "learning_rate": 2.050396637094702e-06, "loss": 0.2284, "step": 20536 }, { "epoch": 1.4197718631178708, "grad_norm": 0.6518961492673349, "learning_rate": 2.0499446717234585e-06, "loss": 0.3052, "step": 20537 }, { "epoch": 1.4198409955063949, "grad_norm": 0.5969721264230864, "learning_rate": 2.049492743325892e-06, "loss": 0.2725, "step": 20538 }, { "epoch": 1.4199101278949189, "grad_norm": 0.6230828088647284, "learning_rate": 2.049040851907669e-06, "loss": 0.207, "step": 20539 }, { "epoch": 1.419979260283443, "grad_norm": 0.698073123717563, "learning_rate": 2.0485889974744523e-06, "loss": 0.2863, "step": 20540 }, { "epoch": 1.420048392671967, "grad_norm": 0.6794076125448935, "learning_rate": 2.048137180031907e-06, "loss": 0.3589, "step": 20541 }, { "epoch": 1.420117525060491, "grad_norm": 0.6598217387632692, "learning_rate": 2.047685399585693e-06, "loss": 0.2405, "step": 20542 }, { "epoch": 1.420186657449015, "grad_norm": 0.6516058056550282, "learning_rate": 2.047233656141476e-06, "loss": 0.2379, "step": 20543 }, { "epoch": 1.420255789837539, "grad_norm": 0.7819414646897257, "learning_rate": 2.0467819497049128e-06, "loss": 0.3211, "step": 20544 }, { "epoch": 1.420324922226063, "grad_norm": 0.6650580364202386, "learning_rate": 2.04633028028167e-06, "loss": 0.269, "step": 20545 }, { "epoch": 1.420394054614587, "grad_norm": 0.5853167155847279, "learning_rate": 2.0458786478774045e-06, "loss": 0.2179, "step": 20546 }, { "epoch": 1.420463187003111, "grad_norm": 0.7021687972568245, "learning_rate": 2.045427052497778e-06, "loss": 0.2994, "step": 20547 }, { "epoch": 1.420532319391635, "grad_norm": 0.8102086283278479, "learning_rate": 2.044975494148451e-06, "loss": 0.3524, "step": 20548 }, { "epoch": 1.420601451780159, "grad_norm": 0.8440484573237846, "learning_rate": 2.0445239728350845e-06, "loss": 0.283, "step": 20549 }, { "epoch": 1.420670584168683, "grad_norm": 0.633098583570341, "learning_rate": 2.044072488563334e-06, "loss": 0.1947, "step": 20550 }, { "epoch": 1.420739716557207, "grad_norm": 0.6338484764416912, "learning_rate": 2.043621041338861e-06, "loss": 0.2753, "step": 20551 }, { "epoch": 1.4208088489457311, "grad_norm": 0.6910008251237021, "learning_rate": 2.043169631167321e-06, "loss": 0.3493, "step": 20552 }, { "epoch": 1.4208779813342551, "grad_norm": 0.6723779894395665, "learning_rate": 2.0427182580543726e-06, "loss": 0.2199, "step": 20553 }, { "epoch": 1.4209471137227792, "grad_norm": 0.6895034468462566, "learning_rate": 2.0422669220056755e-06, "loss": 0.2765, "step": 20554 }, { "epoch": 1.4210162461113032, "grad_norm": 0.7222144486683438, "learning_rate": 2.041815623026882e-06, "loss": 0.3371, "step": 20555 }, { "epoch": 1.4210853784998272, "grad_norm": 0.8184144263574373, "learning_rate": 2.0413643611236527e-06, "loss": 0.2132, "step": 20556 }, { "epoch": 1.4211545108883512, "grad_norm": 0.6876505841106174, "learning_rate": 2.0409131363016395e-06, "loss": 0.2104, "step": 20557 }, { "epoch": 1.4212236432768752, "grad_norm": 0.707414824878513, "learning_rate": 2.040461948566501e-06, "loss": 0.2971, "step": 20558 }, { "epoch": 1.4212927756653992, "grad_norm": 0.6690338151399056, "learning_rate": 2.0400107979238885e-06, "loss": 0.3233, "step": 20559 }, { "epoch": 1.4213619080539233, "grad_norm": 0.5890126156246611, "learning_rate": 2.0395596843794587e-06, "loss": 0.2533, "step": 20560 }, { "epoch": 1.4214310404424473, "grad_norm": 0.6545777436972732, "learning_rate": 2.039108607938865e-06, "loss": 0.2344, "step": 20561 }, { "epoch": 1.4215001728309713, "grad_norm": 0.6689588027018855, "learning_rate": 2.0386575686077625e-06, "loss": 0.3277, "step": 20562 }, { "epoch": 1.4215693052194953, "grad_norm": 0.7351215987600046, "learning_rate": 2.038206566391801e-06, "loss": 0.337, "step": 20563 }, { "epoch": 1.4216384376080193, "grad_norm": 0.6424866463566419, "learning_rate": 2.0377556012966364e-06, "loss": 0.2736, "step": 20564 }, { "epoch": 1.4217075699965434, "grad_norm": 0.7014682209464302, "learning_rate": 2.0373046733279172e-06, "loss": 0.2545, "step": 20565 }, { "epoch": 1.4217767023850674, "grad_norm": 0.6190330248497178, "learning_rate": 2.036853782491299e-06, "loss": 0.3428, "step": 20566 }, { "epoch": 1.4218458347735914, "grad_norm": 0.7114819165059757, "learning_rate": 2.0364029287924283e-06, "loss": 0.2218, "step": 20567 }, { "epoch": 1.4219149671621154, "grad_norm": 0.5947117463350352, "learning_rate": 2.035952112236958e-06, "loss": 0.2486, "step": 20568 }, { "epoch": 1.4219840995506394, "grad_norm": 0.7081569897774647, "learning_rate": 2.035501332830539e-06, "loss": 0.3398, "step": 20569 }, { "epoch": 1.4220532319391634, "grad_norm": 0.6790845796129897, "learning_rate": 2.0350505905788216e-06, "loss": 0.3191, "step": 20570 }, { "epoch": 1.4221223643276875, "grad_norm": 0.663779760245316, "learning_rate": 2.034599885487452e-06, "loss": 0.193, "step": 20571 }, { "epoch": 1.4221914967162115, "grad_norm": 0.6645485461110048, "learning_rate": 2.034149217562083e-06, "loss": 0.2512, "step": 20572 }, { "epoch": 1.4222606291047355, "grad_norm": 0.6582271904082356, "learning_rate": 2.0336985868083594e-06, "loss": 0.3098, "step": 20573 }, { "epoch": 1.4223297614932595, "grad_norm": 0.6702004945992688, "learning_rate": 2.0332479932319293e-06, "loss": 0.2577, "step": 20574 }, { "epoch": 1.4223988938817835, "grad_norm": 0.6382238292043234, "learning_rate": 2.0327974368384418e-06, "loss": 0.2866, "step": 20575 }, { "epoch": 1.4224680262703076, "grad_norm": 0.7510785040477669, "learning_rate": 2.0323469176335453e-06, "loss": 0.2783, "step": 20576 }, { "epoch": 1.4225371586588316, "grad_norm": 0.666886112607005, "learning_rate": 2.031896435622882e-06, "loss": 0.3057, "step": 20577 }, { "epoch": 1.4226062910473556, "grad_norm": 1.0047142708017283, "learning_rate": 2.031445990812102e-06, "loss": 0.2164, "step": 20578 }, { "epoch": 1.4226754234358796, "grad_norm": 0.6052592564686861, "learning_rate": 2.0309955832068466e-06, "loss": 0.219, "step": 20579 }, { "epoch": 1.4227445558244036, "grad_norm": 0.7354500838273084, "learning_rate": 2.0305452128127658e-06, "loss": 0.4132, "step": 20580 }, { "epoch": 1.4228136882129276, "grad_norm": 0.5972090210869827, "learning_rate": 2.030094879635499e-06, "loss": 0.2449, "step": 20581 }, { "epoch": 1.4228828206014517, "grad_norm": 0.6476905410412789, "learning_rate": 2.029644583680693e-06, "loss": 0.2134, "step": 20582 }, { "epoch": 1.4229519529899757, "grad_norm": 0.6865973082150967, "learning_rate": 2.0291943249539918e-06, "loss": 0.3112, "step": 20583 }, { "epoch": 1.4230210853784997, "grad_norm": 0.7400540208256943, "learning_rate": 2.0287441034610393e-06, "loss": 0.3404, "step": 20584 }, { "epoch": 1.4230902177670237, "grad_norm": 0.6747732944703363, "learning_rate": 2.028293919207476e-06, "loss": 0.2376, "step": 20585 }, { "epoch": 1.423159350155548, "grad_norm": 0.6479222997278363, "learning_rate": 2.0278437721989463e-06, "loss": 0.2607, "step": 20586 }, { "epoch": 1.423228482544072, "grad_norm": 0.8632952433952216, "learning_rate": 2.027393662441089e-06, "loss": 0.3244, "step": 20587 }, { "epoch": 1.423297614932596, "grad_norm": 0.776400407627454, "learning_rate": 2.026943589939548e-06, "loss": 0.2768, "step": 20588 }, { "epoch": 1.42336674732112, "grad_norm": 0.7144381251841685, "learning_rate": 2.0264935546999627e-06, "loss": 0.2262, "step": 20589 }, { "epoch": 1.423435879709644, "grad_norm": 0.6486775818767375, "learning_rate": 2.0260435567279763e-06, "loss": 0.2075, "step": 20590 }, { "epoch": 1.423505012098168, "grad_norm": 0.658200204910351, "learning_rate": 2.0255935960292245e-06, "loss": 0.3589, "step": 20591 }, { "epoch": 1.423574144486692, "grad_norm": 0.6456336484407246, "learning_rate": 2.0251436726093508e-06, "loss": 0.2066, "step": 20592 }, { "epoch": 1.423643276875216, "grad_norm": 0.7914504726860784, "learning_rate": 2.024693786473991e-06, "loss": 0.227, "step": 20593 }, { "epoch": 1.42371240926374, "grad_norm": 1.4478391326821192, "learning_rate": 2.0242439376287865e-06, "loss": 0.3393, "step": 20594 }, { "epoch": 1.4237815416522641, "grad_norm": 0.70365681027205, "learning_rate": 2.023794126079372e-06, "loss": 0.3336, "step": 20595 }, { "epoch": 1.4238506740407881, "grad_norm": 0.6507645578223874, "learning_rate": 2.0233443518313868e-06, "loss": 0.2136, "step": 20596 }, { "epoch": 1.4239198064293122, "grad_norm": 0.6341525525058854, "learning_rate": 2.022894614890468e-06, "loss": 0.2494, "step": 20597 }, { "epoch": 1.4239889388178362, "grad_norm": 0.721338665717891, "learning_rate": 2.022444915262254e-06, "loss": 0.3709, "step": 20598 }, { "epoch": 1.4240580712063602, "grad_norm": 0.8014287649263822, "learning_rate": 2.021995252952378e-06, "loss": 0.2678, "step": 20599 }, { "epoch": 1.4241272035948842, "grad_norm": 0.6861220974575188, "learning_rate": 2.021545627966478e-06, "loss": 0.229, "step": 20600 }, { "epoch": 1.4241963359834082, "grad_norm": 0.7250129344734414, "learning_rate": 2.021096040310187e-06, "loss": 0.328, "step": 20601 }, { "epoch": 1.4242654683719322, "grad_norm": 0.6731081708772861, "learning_rate": 2.020646489989143e-06, "loss": 0.3037, "step": 20602 }, { "epoch": 1.4243346007604563, "grad_norm": 0.6793504035079312, "learning_rate": 2.0201969770089765e-06, "loss": 0.2654, "step": 20603 }, { "epoch": 1.4244037331489803, "grad_norm": 0.650052120653613, "learning_rate": 2.0197475013753226e-06, "loss": 0.2202, "step": 20604 }, { "epoch": 1.4244728655375043, "grad_norm": 0.7243655448440159, "learning_rate": 2.019298063093816e-06, "loss": 0.3468, "step": 20605 }, { "epoch": 1.4245419979260283, "grad_norm": 0.6621124443668255, "learning_rate": 2.0188486621700903e-06, "loss": 0.2392, "step": 20606 }, { "epoch": 1.4246111303145523, "grad_norm": 0.6330532377935357, "learning_rate": 2.0183992986097746e-06, "loss": 0.2402, "step": 20607 }, { "epoch": 1.4246802627030763, "grad_norm": 0.6942679315519412, "learning_rate": 2.0179499724185048e-06, "loss": 0.2667, "step": 20608 }, { "epoch": 1.4247493950916004, "grad_norm": 0.7640663938173119, "learning_rate": 2.0175006836019086e-06, "loss": 0.3736, "step": 20609 }, { "epoch": 1.4248185274801244, "grad_norm": 0.7676969388386782, "learning_rate": 2.017051432165619e-06, "loss": 0.3223, "step": 20610 }, { "epoch": 1.4248876598686484, "grad_norm": 0.584905795597375, "learning_rate": 2.016602218115266e-06, "loss": 0.2166, "step": 20611 }, { "epoch": 1.4249567922571724, "grad_norm": 0.6773035134570263, "learning_rate": 2.016153041456482e-06, "loss": 0.4064, "step": 20612 }, { "epoch": 1.4250259246456964, "grad_norm": 0.737332658093654, "learning_rate": 2.0157039021948925e-06, "loss": 0.2244, "step": 20613 }, { "epoch": 1.4250950570342205, "grad_norm": 0.6641591337992886, "learning_rate": 2.015254800336132e-06, "loss": 0.2367, "step": 20614 }, { "epoch": 1.4251641894227445, "grad_norm": 0.7660472126779647, "learning_rate": 2.014805735885823e-06, "loss": 0.2834, "step": 20615 }, { "epoch": 1.4252333218112687, "grad_norm": 0.739806044606754, "learning_rate": 2.0143567088495984e-06, "loss": 0.3701, "step": 20616 }, { "epoch": 1.4253024541997927, "grad_norm": 0.7719335019995482, "learning_rate": 2.0139077192330836e-06, "loss": 0.2518, "step": 20617 }, { "epoch": 1.4253715865883168, "grad_norm": 0.6649117392906849, "learning_rate": 2.013458767041906e-06, "loss": 0.2816, "step": 20618 }, { "epoch": 1.4254407189768408, "grad_norm": 0.6871251934064457, "learning_rate": 2.013009852281695e-06, "loss": 0.3157, "step": 20619 }, { "epoch": 1.4255098513653648, "grad_norm": 0.7251778344733065, "learning_rate": 2.012560974958073e-06, "loss": 0.3076, "step": 20620 }, { "epoch": 1.4255789837538888, "grad_norm": 0.6494717528211206, "learning_rate": 2.01211213507667e-06, "loss": 0.1986, "step": 20621 }, { "epoch": 1.4256481161424128, "grad_norm": 0.7745300748133963, "learning_rate": 2.0116633326431073e-06, "loss": 0.2891, "step": 20622 }, { "epoch": 1.4257172485309368, "grad_norm": 0.716655983255954, "learning_rate": 2.0112145676630137e-06, "loss": 0.3379, "step": 20623 }, { "epoch": 1.4257863809194609, "grad_norm": 0.7330570370995705, "learning_rate": 2.0107658401420077e-06, "loss": 0.2631, "step": 20624 }, { "epoch": 1.4258555133079849, "grad_norm": 0.6452576950712926, "learning_rate": 2.0103171500857216e-06, "loss": 0.2372, "step": 20625 }, { "epoch": 1.425924645696509, "grad_norm": 0.6329404879797313, "learning_rate": 2.0098684974997723e-06, "loss": 0.2557, "step": 20626 }, { "epoch": 1.425993778085033, "grad_norm": 0.7149960997240989, "learning_rate": 2.0094198823897877e-06, "loss": 0.3051, "step": 20627 }, { "epoch": 1.426062910473557, "grad_norm": 0.8864038323160269, "learning_rate": 2.008971304761386e-06, "loss": 0.3008, "step": 20628 }, { "epoch": 1.426132042862081, "grad_norm": 0.6910951073525887, "learning_rate": 2.0085227646201923e-06, "loss": 0.2327, "step": 20629 }, { "epoch": 1.426201175250605, "grad_norm": 0.739958364549845, "learning_rate": 2.008074261971826e-06, "loss": 0.3686, "step": 20630 }, { "epoch": 1.426270307639129, "grad_norm": 0.7008090581659682, "learning_rate": 2.007625796821909e-06, "loss": 0.2717, "step": 20631 }, { "epoch": 1.426339440027653, "grad_norm": 0.6634597393613186, "learning_rate": 2.0071773691760634e-06, "loss": 0.2311, "step": 20632 }, { "epoch": 1.426408572416177, "grad_norm": 0.6953221632587918, "learning_rate": 2.0067289790399096e-06, "loss": 0.3053, "step": 20633 }, { "epoch": 1.426477704804701, "grad_norm": 0.6358534806119513, "learning_rate": 2.006280626419065e-06, "loss": 0.3329, "step": 20634 }, { "epoch": 1.426546837193225, "grad_norm": 0.7228113906624581, "learning_rate": 2.0058323113191518e-06, "loss": 0.2782, "step": 20635 }, { "epoch": 1.426615969581749, "grad_norm": 0.6229185627901879, "learning_rate": 2.0053840337457857e-06, "loss": 0.243, "step": 20636 }, { "epoch": 1.426685101970273, "grad_norm": 0.6846819285156772, "learning_rate": 2.004935793704588e-06, "loss": 0.342, "step": 20637 }, { "epoch": 1.4267542343587971, "grad_norm": 0.7088106058060106, "learning_rate": 2.0044875912011737e-06, "loss": 0.2972, "step": 20638 }, { "epoch": 1.4268233667473211, "grad_norm": 0.7237851221842703, "learning_rate": 2.0040394262411616e-06, "loss": 0.2326, "step": 20639 }, { "epoch": 1.4268924991358451, "grad_norm": 0.7261419988484271, "learning_rate": 2.0035912988301687e-06, "loss": 0.2535, "step": 20640 }, { "epoch": 1.4269616315243692, "grad_norm": 0.7022050049080315, "learning_rate": 2.0031432089738134e-06, "loss": 0.3674, "step": 20641 }, { "epoch": 1.4270307639128932, "grad_norm": 0.6776553629469223, "learning_rate": 2.0026951566777082e-06, "loss": 0.2766, "step": 20642 }, { "epoch": 1.4270998963014172, "grad_norm": 0.64849383680874, "learning_rate": 2.0022471419474714e-06, "loss": 0.217, "step": 20643 }, { "epoch": 1.4271690286899412, "grad_norm": 0.7085066662807538, "learning_rate": 2.001799164788716e-06, "loss": 0.3569, "step": 20644 }, { "epoch": 1.4272381610784652, "grad_norm": 0.7084532029276912, "learning_rate": 2.001351225207057e-06, "loss": 0.2458, "step": 20645 }, { "epoch": 1.4273072934669893, "grad_norm": 0.623538332100142, "learning_rate": 2.0009033232081093e-06, "loss": 0.2281, "step": 20646 }, { "epoch": 1.4273764258555133, "grad_norm": 0.9492131650381177, "learning_rate": 2.0004554587974885e-06, "loss": 0.2559, "step": 20647 }, { "epoch": 1.4274455582440373, "grad_norm": 0.6774631738262271, "learning_rate": 2.0000076319808033e-06, "loss": 0.3092, "step": 20648 }, { "epoch": 1.4275146906325613, "grad_norm": 0.6378057791725628, "learning_rate": 1.99955984276367e-06, "loss": 0.2448, "step": 20649 }, { "epoch": 1.4275838230210853, "grad_norm": 0.6340603494977642, "learning_rate": 1.9991120911516985e-06, "loss": 0.212, "step": 20650 }, { "epoch": 1.4276529554096093, "grad_norm": 0.6840620935251791, "learning_rate": 1.9986643771505032e-06, "loss": 0.2746, "step": 20651 }, { "epoch": 1.4277220877981334, "grad_norm": 0.7261977508289371, "learning_rate": 1.998216700765692e-06, "loss": 0.3044, "step": 20652 }, { "epoch": 1.4277912201866574, "grad_norm": 0.6205803961465902, "learning_rate": 1.9977690620028778e-06, "loss": 0.2541, "step": 20653 }, { "epoch": 1.4278603525751814, "grad_norm": 0.6508472625483759, "learning_rate": 1.9973214608676707e-06, "loss": 0.2255, "step": 20654 }, { "epoch": 1.4279294849637054, "grad_norm": 1.4081559454772217, "learning_rate": 1.996873897365682e-06, "loss": 0.3399, "step": 20655 }, { "epoch": 1.4279986173522294, "grad_norm": 0.6606406788132881, "learning_rate": 1.9964263715025177e-06, "loss": 0.203, "step": 20656 }, { "epoch": 1.4280677497407535, "grad_norm": 0.6195008145070792, "learning_rate": 1.9959788832837914e-06, "loss": 0.196, "step": 20657 }, { "epoch": 1.4281368821292775, "grad_norm": 0.7508986385781948, "learning_rate": 1.995531432715106e-06, "loss": 0.3379, "step": 20658 }, { "epoch": 1.4282060145178015, "grad_norm": 0.637412262570378, "learning_rate": 1.995084019802075e-06, "loss": 0.3082, "step": 20659 }, { "epoch": 1.4282751469063255, "grad_norm": 0.684004426176891, "learning_rate": 1.994636644550299e-06, "loss": 0.2358, "step": 20660 }, { "epoch": 1.4283442792948495, "grad_norm": 0.7035078783130522, "learning_rate": 1.9941893069653934e-06, "loss": 0.2268, "step": 20661 }, { "epoch": 1.4284134116833735, "grad_norm": 0.7045564554761538, "learning_rate": 1.993742007052959e-06, "loss": 0.3892, "step": 20662 }, { "epoch": 1.4284825440718976, "grad_norm": 0.669730798746497, "learning_rate": 1.9932947448186053e-06, "loss": 0.299, "step": 20663 }, { "epoch": 1.4285516764604216, "grad_norm": 0.7082394714437596, "learning_rate": 1.992847520267934e-06, "loss": 0.2227, "step": 20664 }, { "epoch": 1.4286208088489456, "grad_norm": 0.6887864486666317, "learning_rate": 1.992400333406555e-06, "loss": 0.252, "step": 20665 }, { "epoch": 1.4286899412374696, "grad_norm": 0.7291752631828055, "learning_rate": 1.9919531842400684e-06, "loss": 0.3976, "step": 20666 }, { "epoch": 1.4287590736259939, "grad_norm": 0.6872109058056073, "learning_rate": 1.991506072774081e-06, "loss": 0.2816, "step": 20667 }, { "epoch": 1.4288282060145179, "grad_norm": 0.6898946214926784, "learning_rate": 1.9910589990141953e-06, "loss": 0.2408, "step": 20668 }, { "epoch": 1.428897338403042, "grad_norm": 0.7132078630583747, "learning_rate": 1.9906119629660182e-06, "loss": 0.3673, "step": 20669 }, { "epoch": 1.428966470791566, "grad_norm": 0.7408386475198022, "learning_rate": 1.990164964635147e-06, "loss": 0.2826, "step": 20670 }, { "epoch": 1.42903560318009, "grad_norm": 0.6688155551092114, "learning_rate": 1.9897180040271883e-06, "loss": 0.2458, "step": 20671 }, { "epoch": 1.429104735568614, "grad_norm": 0.6974673869912852, "learning_rate": 1.989271081147741e-06, "loss": 0.2849, "step": 20672 }, { "epoch": 1.429173867957138, "grad_norm": 0.8200163588598974, "learning_rate": 1.9888241960024094e-06, "loss": 0.2855, "step": 20673 }, { "epoch": 1.429243000345662, "grad_norm": 0.6382850589621304, "learning_rate": 1.9883773485967918e-06, "loss": 0.2241, "step": 20674 }, { "epoch": 1.429312132734186, "grad_norm": 0.664804007371357, "learning_rate": 1.987930538936489e-06, "loss": 0.2543, "step": 20675 }, { "epoch": 1.42938126512271, "grad_norm": 0.6560831470319712, "learning_rate": 1.9874837670271014e-06, "loss": 0.2304, "step": 20676 }, { "epoch": 1.429450397511234, "grad_norm": 0.8508869299479604, "learning_rate": 1.9870370328742306e-06, "loss": 0.3417, "step": 20677 }, { "epoch": 1.429519529899758, "grad_norm": 0.6951433956307891, "learning_rate": 1.9865903364834723e-06, "loss": 0.2093, "step": 20678 }, { "epoch": 1.429588662288282, "grad_norm": 0.69163212732766, "learning_rate": 1.9861436778604278e-06, "loss": 0.2507, "step": 20679 }, { "epoch": 1.429657794676806, "grad_norm": 0.7906771011160445, "learning_rate": 1.9856970570106922e-06, "loss": 0.3816, "step": 20680 }, { "epoch": 1.42972692706533, "grad_norm": 0.7506943903771199, "learning_rate": 1.985250473939865e-06, "loss": 0.291, "step": 20681 }, { "epoch": 1.4297960594538541, "grad_norm": 0.6538120295736997, "learning_rate": 1.9848039286535426e-06, "loss": 0.2938, "step": 20682 }, { "epoch": 1.4298651918423781, "grad_norm": 0.7476059351731406, "learning_rate": 1.984357421157324e-06, "loss": 0.2689, "step": 20683 }, { "epoch": 1.4299343242309022, "grad_norm": 0.7390589092215081, "learning_rate": 1.9839109514568017e-06, "loss": 0.375, "step": 20684 }, { "epoch": 1.4300034566194262, "grad_norm": 0.6445833312380681, "learning_rate": 1.9834645195575754e-06, "loss": 0.2488, "step": 20685 }, { "epoch": 1.4300725890079502, "grad_norm": 0.6363631791122663, "learning_rate": 1.9830181254652374e-06, "loss": 0.2161, "step": 20686 }, { "epoch": 1.4301417213964742, "grad_norm": 0.7101529211385894, "learning_rate": 1.982571769185382e-06, "loss": 0.3129, "step": 20687 }, { "epoch": 1.4302108537849982, "grad_norm": 0.6959968118783557, "learning_rate": 1.9821254507236044e-06, "loss": 0.3218, "step": 20688 }, { "epoch": 1.4302799861735223, "grad_norm": 0.6373013154424164, "learning_rate": 1.9816791700854983e-06, "loss": 0.2292, "step": 20689 }, { "epoch": 1.4303491185620463, "grad_norm": 0.6934831695755469, "learning_rate": 1.9812329272766595e-06, "loss": 0.2632, "step": 20690 }, { "epoch": 1.4304182509505703, "grad_norm": 0.744164007875652, "learning_rate": 1.9807867223026767e-06, "loss": 0.3766, "step": 20691 }, { "epoch": 1.4304873833390943, "grad_norm": 0.6286465979560281, "learning_rate": 1.9803405551691463e-06, "loss": 0.2397, "step": 20692 }, { "epoch": 1.4305565157276183, "grad_norm": 0.6358325806110029, "learning_rate": 1.9798944258816566e-06, "loss": 0.2125, "step": 20693 }, { "epoch": 1.4306256481161423, "grad_norm": 0.7478059223431689, "learning_rate": 1.979448334445802e-06, "loss": 0.2874, "step": 20694 }, { "epoch": 1.4306947805046664, "grad_norm": 0.7257053810479107, "learning_rate": 1.979002280867169e-06, "loss": 0.3226, "step": 20695 }, { "epoch": 1.4307639128931904, "grad_norm": 0.9703679338264248, "learning_rate": 1.9785562651513547e-06, "loss": 0.2433, "step": 20696 }, { "epoch": 1.4308330452817146, "grad_norm": 0.7114729333011879, "learning_rate": 1.9781102873039433e-06, "loss": 0.2682, "step": 20697 }, { "epoch": 1.4309021776702386, "grad_norm": 0.6704492631051617, "learning_rate": 1.977664347330529e-06, "loss": 0.3655, "step": 20698 }, { "epoch": 1.4309713100587627, "grad_norm": 0.8264546752926257, "learning_rate": 1.977218445236696e-06, "loss": 0.2511, "step": 20699 }, { "epoch": 1.4310404424472867, "grad_norm": 0.6287332164796651, "learning_rate": 1.976772581028038e-06, "loss": 0.2485, "step": 20700 }, { "epoch": 1.4311095748358107, "grad_norm": 0.7059022708688166, "learning_rate": 1.9763267547101383e-06, "loss": 0.3208, "step": 20701 }, { "epoch": 1.4311787072243347, "grad_norm": 0.7229442787146686, "learning_rate": 1.975880966288587e-06, "loss": 0.3156, "step": 20702 }, { "epoch": 1.4312478396128587, "grad_norm": 0.7684350924829507, "learning_rate": 1.975435215768971e-06, "loss": 0.2383, "step": 20703 }, { "epoch": 1.4313169720013827, "grad_norm": 0.6674030617265422, "learning_rate": 1.9749895031568784e-06, "loss": 0.232, "step": 20704 }, { "epoch": 1.4313861043899068, "grad_norm": 0.7288225382016367, "learning_rate": 1.974543828457892e-06, "loss": 0.4002, "step": 20705 }, { "epoch": 1.4314552367784308, "grad_norm": 0.7301742220201427, "learning_rate": 1.974098191677602e-06, "loss": 0.3179, "step": 20706 }, { "epoch": 1.4315243691669548, "grad_norm": 0.5941895768772465, "learning_rate": 1.9736525928215896e-06, "loss": 0.2671, "step": 20707 }, { "epoch": 1.4315935015554788, "grad_norm": 0.6975773823560563, "learning_rate": 1.9732070318954424e-06, "loss": 0.3179, "step": 20708 }, { "epoch": 1.4316626339440028, "grad_norm": 0.6965076532532158, "learning_rate": 1.972761508904742e-06, "loss": 0.3408, "step": 20709 }, { "epoch": 1.4317317663325269, "grad_norm": 0.7579941937933964, "learning_rate": 1.9723160238550733e-06, "loss": 0.2342, "step": 20710 }, { "epoch": 1.4318008987210509, "grad_norm": 0.6547380559738016, "learning_rate": 1.971870576752021e-06, "loss": 0.2299, "step": 20711 }, { "epoch": 1.431870031109575, "grad_norm": 0.7106045985894194, "learning_rate": 1.9714251676011675e-06, "loss": 0.3369, "step": 20712 }, { "epoch": 1.431939163498099, "grad_norm": 0.7180421556676397, "learning_rate": 1.970979796408094e-06, "loss": 0.292, "step": 20713 }, { "epoch": 1.432008295886623, "grad_norm": 0.632713442338633, "learning_rate": 1.9705344631783845e-06, "loss": 0.219, "step": 20714 }, { "epoch": 1.432077428275147, "grad_norm": 0.6679420281005873, "learning_rate": 1.970089167917618e-06, "loss": 0.2411, "step": 20715 }, { "epoch": 1.432146560663671, "grad_norm": 0.6335748802016684, "learning_rate": 1.9696439106313762e-06, "loss": 0.345, "step": 20716 }, { "epoch": 1.432215693052195, "grad_norm": 0.6631508630361035, "learning_rate": 1.9691986913252405e-06, "loss": 0.2455, "step": 20717 }, { "epoch": 1.432284825440719, "grad_norm": 0.7202572504944199, "learning_rate": 1.9687535100047915e-06, "loss": 0.2797, "step": 20718 }, { "epoch": 1.432353957829243, "grad_norm": 0.7275476875660863, "learning_rate": 1.968308366675607e-06, "loss": 0.3, "step": 20719 }, { "epoch": 1.432423090217767, "grad_norm": 0.6732952832162927, "learning_rate": 1.967863261343268e-06, "loss": 0.3376, "step": 20720 }, { "epoch": 1.432492222606291, "grad_norm": 0.6602641494877123, "learning_rate": 1.9674181940133502e-06, "loss": 0.248, "step": 20721 }, { "epoch": 1.432561354994815, "grad_norm": 0.6789541276777541, "learning_rate": 1.966973164691435e-06, "loss": 0.2558, "step": 20722 }, { "epoch": 1.432630487383339, "grad_norm": 0.9067652967488735, "learning_rate": 1.966528173383097e-06, "loss": 0.3689, "step": 20723 }, { "epoch": 1.432699619771863, "grad_norm": 0.6787984884203255, "learning_rate": 1.966083220093915e-06, "loss": 0.2472, "step": 20724 }, { "epoch": 1.4327687521603871, "grad_norm": 0.7640114333740422, "learning_rate": 1.9656383048294654e-06, "loss": 0.27, "step": 20725 }, { "epoch": 1.4328378845489111, "grad_norm": 0.677885885866811, "learning_rate": 1.9651934275953268e-06, "loss": 0.2788, "step": 20726 }, { "epoch": 1.4329070169374352, "grad_norm": 0.7765547794146181, "learning_rate": 1.96474858839707e-06, "loss": 0.335, "step": 20727 }, { "epoch": 1.4329761493259592, "grad_norm": 0.648001249476014, "learning_rate": 1.9643037872402756e-06, "loss": 0.2618, "step": 20728 }, { "epoch": 1.4330452817144832, "grad_norm": 0.6904362210966533, "learning_rate": 1.9638590241305133e-06, "loss": 0.247, "step": 20729 }, { "epoch": 1.4331144141030072, "grad_norm": 0.7680055713016144, "learning_rate": 1.9634142990733624e-06, "loss": 0.3963, "step": 20730 }, { "epoch": 1.4331835464915312, "grad_norm": 0.6602378073623068, "learning_rate": 1.9629696120743902e-06, "loss": 0.2585, "step": 20731 }, { "epoch": 1.4332526788800553, "grad_norm": 0.7050942751296162, "learning_rate": 1.962524963139178e-06, "loss": 0.2359, "step": 20732 }, { "epoch": 1.4333218112685793, "grad_norm": 0.6603285533676947, "learning_rate": 1.9620803522732928e-06, "loss": 0.2503, "step": 20733 }, { "epoch": 1.4333909436571033, "grad_norm": 0.6846383385288545, "learning_rate": 1.9616357794823106e-06, "loss": 0.3737, "step": 20734 }, { "epoch": 1.4334600760456273, "grad_norm": 0.6848973777813875, "learning_rate": 1.9611912447718003e-06, "loss": 0.2216, "step": 20735 }, { "epoch": 1.4335292084341513, "grad_norm": 0.6429610971870032, "learning_rate": 1.9607467481473358e-06, "loss": 0.2178, "step": 20736 }, { "epoch": 1.4335983408226753, "grad_norm": 0.7093890941679059, "learning_rate": 1.960302289614486e-06, "loss": 0.3299, "step": 20737 }, { "epoch": 1.4336674732111994, "grad_norm": 0.6113310104470256, "learning_rate": 1.9598578691788222e-06, "loss": 0.2379, "step": 20738 }, { "epoch": 1.4337366055997234, "grad_norm": 0.5683549093208945, "learning_rate": 1.9594134868459147e-06, "loss": 0.2149, "step": 20739 }, { "epoch": 1.4338057379882474, "grad_norm": 0.6734553101207291, "learning_rate": 1.958969142621334e-06, "loss": 0.2733, "step": 20740 }, { "epoch": 1.4338748703767714, "grad_norm": 0.9233871035272616, "learning_rate": 1.9585248365106472e-06, "loss": 0.3205, "step": 20741 }, { "epoch": 1.4339440027652954, "grad_norm": 0.6356703987144396, "learning_rate": 1.9580805685194253e-06, "loss": 0.2949, "step": 20742 }, { "epoch": 1.4340131351538195, "grad_norm": 0.6345323380132358, "learning_rate": 1.957636338653233e-06, "loss": 0.1977, "step": 20743 }, { "epoch": 1.4340822675423435, "grad_norm": 0.6787648202219573, "learning_rate": 1.9571921469176426e-06, "loss": 0.2546, "step": 20744 }, { "epoch": 1.4341513999308675, "grad_norm": 0.6725279262672551, "learning_rate": 1.956747993318215e-06, "loss": 0.2892, "step": 20745 }, { "epoch": 1.4342205323193915, "grad_norm": 0.6383223214220738, "learning_rate": 1.956303877860522e-06, "loss": 0.1926, "step": 20746 }, { "epoch": 1.4342896647079155, "grad_norm": 0.7496326715240571, "learning_rate": 1.9558598005501276e-06, "loss": 0.2579, "step": 20747 }, { "epoch": 1.4343587970964398, "grad_norm": 0.6922804161405179, "learning_rate": 1.9554157613926e-06, "loss": 0.334, "step": 20748 }, { "epoch": 1.4344279294849638, "grad_norm": 0.6287217388629229, "learning_rate": 1.9549717603935e-06, "loss": 0.2174, "step": 20749 }, { "epoch": 1.4344970618734878, "grad_norm": 0.590094512973951, "learning_rate": 1.9545277975583976e-06, "loss": 0.2067, "step": 20750 }, { "epoch": 1.4345661942620118, "grad_norm": 0.7448561741592095, "learning_rate": 1.9540838728928523e-06, "loss": 0.2937, "step": 20751 }, { "epoch": 1.4346353266505358, "grad_norm": 0.7267159773845854, "learning_rate": 1.95363998640243e-06, "loss": 0.3608, "step": 20752 }, { "epoch": 1.4347044590390599, "grad_norm": 0.6565153465782727, "learning_rate": 1.9531961380926954e-06, "loss": 0.2229, "step": 20753 }, { "epoch": 1.4347735914275839, "grad_norm": 1.957594505449758, "learning_rate": 1.9527523279692083e-06, "loss": 0.2095, "step": 20754 }, { "epoch": 1.434842723816108, "grad_norm": 0.6824886595645787, "learning_rate": 1.952308556037535e-06, "loss": 0.3251, "step": 20755 }, { "epoch": 1.434911856204632, "grad_norm": 0.6788227034185329, "learning_rate": 1.9518648223032327e-06, "loss": 0.2262, "step": 20756 }, { "epoch": 1.434980988593156, "grad_norm": 0.5838629513909963, "learning_rate": 1.951421126771867e-06, "loss": 0.2374, "step": 20757 }, { "epoch": 1.43505012098168, "grad_norm": 0.6568426198384145, "learning_rate": 1.9509774694489956e-06, "loss": 0.2444, "step": 20758 }, { "epoch": 1.435119253370204, "grad_norm": 0.8091089591666736, "learning_rate": 1.9505338503401807e-06, "loss": 0.3263, "step": 20759 }, { "epoch": 1.435188385758728, "grad_norm": 0.6377333979003709, "learning_rate": 1.950090269450981e-06, "loss": 0.2683, "step": 20760 }, { "epoch": 1.435257518147252, "grad_norm": 0.6261057433886986, "learning_rate": 1.9496467267869595e-06, "loss": 0.2441, "step": 20761 }, { "epoch": 1.435326650535776, "grad_norm": 0.6548712076597117, "learning_rate": 1.9492032223536705e-06, "loss": 0.3199, "step": 20762 }, { "epoch": 1.4353957829243, "grad_norm": 0.7195111719753645, "learning_rate": 1.948759756156677e-06, "loss": 0.3145, "step": 20763 }, { "epoch": 1.435464915312824, "grad_norm": 0.6304124144787554, "learning_rate": 1.948316328201532e-06, "loss": 0.2108, "step": 20764 }, { "epoch": 1.435534047701348, "grad_norm": 0.6853091630559593, "learning_rate": 1.947872938493799e-06, "loss": 0.2767, "step": 20765 }, { "epoch": 1.435603180089872, "grad_norm": 0.6736487094467427, "learning_rate": 1.9474295870390275e-06, "loss": 0.3807, "step": 20766 }, { "epoch": 1.435672312478396, "grad_norm": 0.643529728527013, "learning_rate": 1.9469862738427823e-06, "loss": 0.2475, "step": 20767 }, { "epoch": 1.4357414448669201, "grad_norm": 0.6993815938173328, "learning_rate": 1.946542998910614e-06, "loss": 0.1793, "step": 20768 }, { "epoch": 1.4358105772554441, "grad_norm": 0.7024514485351895, "learning_rate": 1.946099762248082e-06, "loss": 0.3389, "step": 20769 }, { "epoch": 1.4358797096439682, "grad_norm": 0.6167430162746032, "learning_rate": 1.945656563860738e-06, "loss": 0.2614, "step": 20770 }, { "epoch": 1.4359488420324922, "grad_norm": 0.6561940576920549, "learning_rate": 1.9452134037541393e-06, "loss": 0.2534, "step": 20771 }, { "epoch": 1.4360179744210162, "grad_norm": 0.6713510150970879, "learning_rate": 1.944770281933838e-06, "loss": 0.2622, "step": 20772 }, { "epoch": 1.4360871068095402, "grad_norm": 0.6386954418998607, "learning_rate": 1.9443271984053886e-06, "loss": 0.3313, "step": 20773 }, { "epoch": 1.4361562391980642, "grad_norm": 0.6933686844111303, "learning_rate": 1.9438841531743445e-06, "loss": 0.3264, "step": 20774 }, { "epoch": 1.4362253715865883, "grad_norm": 0.672252263202336, "learning_rate": 1.9434411462462603e-06, "loss": 0.2078, "step": 20775 }, { "epoch": 1.4362945039751123, "grad_norm": 0.9131431091740076, "learning_rate": 1.9429981776266853e-06, "loss": 0.2861, "step": 20776 }, { "epoch": 1.4363636363636363, "grad_norm": 0.729162120749992, "learning_rate": 1.9425552473211744e-06, "loss": 0.3364, "step": 20777 }, { "epoch": 1.4364327687521603, "grad_norm": 0.6695449088029368, "learning_rate": 1.9421123553352756e-06, "loss": 0.2556, "step": 20778 }, { "epoch": 1.4365019011406845, "grad_norm": 0.6778421627950652, "learning_rate": 1.941669501674543e-06, "loss": 0.2519, "step": 20779 }, { "epoch": 1.4365710335292086, "grad_norm": 0.6680270586165494, "learning_rate": 1.941226686344523e-06, "loss": 0.3344, "step": 20780 }, { "epoch": 1.4366401659177326, "grad_norm": 0.7477300455840075, "learning_rate": 1.940783909350769e-06, "loss": 0.3399, "step": 20781 }, { "epoch": 1.4367092983062566, "grad_norm": 0.7128397366741294, "learning_rate": 1.9403411706988284e-06, "loss": 0.2574, "step": 20782 }, { "epoch": 1.4367784306947806, "grad_norm": 0.7495245926282416, "learning_rate": 1.939898470394253e-06, "loss": 0.3403, "step": 20783 }, { "epoch": 1.4368475630833046, "grad_norm": 0.6861421210272576, "learning_rate": 1.939455808442588e-06, "loss": 0.357, "step": 20784 }, { "epoch": 1.4369166954718287, "grad_norm": 0.8061290354278813, "learning_rate": 1.939013184849384e-06, "loss": 0.22, "step": 20785 }, { "epoch": 1.4369858278603527, "grad_norm": 0.7061566000403451, "learning_rate": 1.9385705996201854e-06, "loss": 0.2743, "step": 20786 }, { "epoch": 1.4370549602488767, "grad_norm": 0.6694074961178678, "learning_rate": 1.9381280527605405e-06, "loss": 0.2971, "step": 20787 }, { "epoch": 1.4371240926374007, "grad_norm": 0.7052300980031138, "learning_rate": 1.9376855442759965e-06, "loss": 0.2733, "step": 20788 }, { "epoch": 1.4371932250259247, "grad_norm": 0.610145705299309, "learning_rate": 1.9372430741721015e-06, "loss": 0.2053, "step": 20789 }, { "epoch": 1.4372623574144487, "grad_norm": 0.6550754406921452, "learning_rate": 1.9368006424543974e-06, "loss": 0.2355, "step": 20790 }, { "epoch": 1.4373314898029728, "grad_norm": 0.7711854625640285, "learning_rate": 1.9363582491284316e-06, "loss": 0.322, "step": 20791 }, { "epoch": 1.4374006221914968, "grad_norm": 0.7067777054080825, "learning_rate": 1.9359158941997463e-06, "loss": 0.2362, "step": 20792 }, { "epoch": 1.4374697545800208, "grad_norm": 0.8451281968381719, "learning_rate": 1.9354735776738894e-06, "loss": 0.2101, "step": 20793 }, { "epoch": 1.4375388869685448, "grad_norm": 0.6876966348979547, "learning_rate": 1.9350312995564e-06, "loss": 0.3186, "step": 20794 }, { "epoch": 1.4376080193570688, "grad_norm": 0.7694873100445305, "learning_rate": 1.934589059852824e-06, "loss": 0.311, "step": 20795 }, { "epoch": 1.4376771517455929, "grad_norm": 0.692686074686146, "learning_rate": 1.9341468585687036e-06, "loss": 0.2777, "step": 20796 }, { "epoch": 1.4377462841341169, "grad_norm": 0.6549237812750976, "learning_rate": 1.9337046957095824e-06, "loss": 0.2375, "step": 20797 }, { "epoch": 1.4378154165226409, "grad_norm": 0.7169579935693529, "learning_rate": 1.9332625712809994e-06, "loss": 0.3675, "step": 20798 }, { "epoch": 1.437884548911165, "grad_norm": 0.6450620200522521, "learning_rate": 1.9328204852884986e-06, "loss": 0.2184, "step": 20799 }, { "epoch": 1.437953681299689, "grad_norm": 0.742673910336545, "learning_rate": 1.9323784377376183e-06, "loss": 0.2289, "step": 20800 }, { "epoch": 1.438022813688213, "grad_norm": 0.7886502341793262, "learning_rate": 1.931936428633901e-06, "loss": 0.2923, "step": 20801 }, { "epoch": 1.438091946076737, "grad_norm": 0.7828332074036983, "learning_rate": 1.931494457982882e-06, "loss": 0.3506, "step": 20802 }, { "epoch": 1.438161078465261, "grad_norm": 0.6450973455502799, "learning_rate": 1.9310525257901073e-06, "loss": 0.193, "step": 20803 }, { "epoch": 1.438230210853785, "grad_norm": 0.6367501033275873, "learning_rate": 1.930610632061111e-06, "loss": 0.2317, "step": 20804 }, { "epoch": 1.438299343242309, "grad_norm": 0.6874362188622203, "learning_rate": 1.930168776801434e-06, "loss": 0.3313, "step": 20805 }, { "epoch": 1.438368475630833, "grad_norm": 0.6731382360340531, "learning_rate": 1.9297269600166123e-06, "loss": 0.2368, "step": 20806 }, { "epoch": 1.438437608019357, "grad_norm": 0.6656375690661239, "learning_rate": 1.9292851817121855e-06, "loss": 0.2241, "step": 20807 }, { "epoch": 1.438506740407881, "grad_norm": 0.7337771900329177, "learning_rate": 1.928843441893687e-06, "loss": 0.3044, "step": 20808 }, { "epoch": 1.438575872796405, "grad_norm": 0.6961360002268024, "learning_rate": 1.928401740566655e-06, "loss": 0.3311, "step": 20809 }, { "epoch": 1.438645005184929, "grad_norm": 0.7869471050958089, "learning_rate": 1.927960077736626e-06, "loss": 0.2159, "step": 20810 }, { "epoch": 1.4387141375734531, "grad_norm": 0.6352658063818976, "learning_rate": 1.927518453409137e-06, "loss": 0.2624, "step": 20811 }, { "epoch": 1.4387832699619771, "grad_norm": 0.713947652913434, "learning_rate": 1.927076867589719e-06, "loss": 0.3156, "step": 20812 }, { "epoch": 1.4388524023505012, "grad_norm": 0.7535533269277369, "learning_rate": 1.9266353202839107e-06, "loss": 0.2597, "step": 20813 }, { "epoch": 1.4389215347390252, "grad_norm": 0.7140830205714933, "learning_rate": 1.9261938114972418e-06, "loss": 0.243, "step": 20814 }, { "epoch": 1.4389906671275492, "grad_norm": 0.7472245335426407, "learning_rate": 1.9257523412352495e-06, "loss": 0.3383, "step": 20815 }, { "epoch": 1.4390597995160732, "grad_norm": 0.6911099323535379, "learning_rate": 1.9253109095034646e-06, "loss": 0.3809, "step": 20816 }, { "epoch": 1.4391289319045972, "grad_norm": 0.8095421900813291, "learning_rate": 1.92486951630742e-06, "loss": 0.2568, "step": 20817 }, { "epoch": 1.4391980642931212, "grad_norm": 0.6080980763082111, "learning_rate": 1.924428161652648e-06, "loss": 0.2278, "step": 20818 }, { "epoch": 1.4392671966816453, "grad_norm": 0.7189646738745049, "learning_rate": 1.923986845544682e-06, "loss": 0.3273, "step": 20819 }, { "epoch": 1.4393363290701693, "grad_norm": 0.6628706169953364, "learning_rate": 1.923545567989052e-06, "loss": 0.3296, "step": 20820 }, { "epoch": 1.4394054614586933, "grad_norm": 0.6685179685034395, "learning_rate": 1.923104328991286e-06, "loss": 0.2068, "step": 20821 }, { "epoch": 1.4394745938472173, "grad_norm": 0.6886430481485388, "learning_rate": 1.9226631285569164e-06, "loss": 0.2869, "step": 20822 }, { "epoch": 1.4395437262357413, "grad_norm": 0.6979642256128367, "learning_rate": 1.9222219666914726e-06, "loss": 0.3407, "step": 20823 }, { "epoch": 1.4396128586242654, "grad_norm": 0.7403902399931622, "learning_rate": 1.9217808434004857e-06, "loss": 0.2579, "step": 20824 }, { "epoch": 1.4396819910127894, "grad_norm": 0.667463505548362, "learning_rate": 1.921339758689481e-06, "loss": 0.2575, "step": 20825 }, { "epoch": 1.4397511234013134, "grad_norm": 0.6340999970148986, "learning_rate": 1.9208987125639898e-06, "loss": 0.2865, "step": 20826 }, { "epoch": 1.4398202557898374, "grad_norm": 0.6184284251952292, "learning_rate": 1.920457705029537e-06, "loss": 0.2537, "step": 20827 }, { "epoch": 1.4398893881783614, "grad_norm": 0.766866159574017, "learning_rate": 1.9200167360916528e-06, "loss": 0.201, "step": 20828 }, { "epoch": 1.4399585205668857, "grad_norm": 0.6241434429807505, "learning_rate": 1.9195758057558605e-06, "loss": 0.2658, "step": 20829 }, { "epoch": 1.4400276529554097, "grad_norm": 0.6695422644640351, "learning_rate": 1.9191349140276884e-06, "loss": 0.3151, "step": 20830 }, { "epoch": 1.4400967853439337, "grad_norm": 0.737158621524891, "learning_rate": 1.9186940609126615e-06, "loss": 0.2506, "step": 20831 }, { "epoch": 1.4401659177324577, "grad_norm": 0.6902783825193974, "learning_rate": 1.9182532464163077e-06, "loss": 0.2715, "step": 20832 }, { "epoch": 1.4402350501209817, "grad_norm": 0.6583186695529408, "learning_rate": 1.9178124705441486e-06, "loss": 0.2951, "step": 20833 }, { "epoch": 1.4403041825095058, "grad_norm": 0.6660428502801833, "learning_rate": 1.917371733301711e-06, "loss": 0.3028, "step": 20834 }, { "epoch": 1.4403733148980298, "grad_norm": 0.6620438568348062, "learning_rate": 1.9169310346945157e-06, "loss": 0.247, "step": 20835 }, { "epoch": 1.4404424472865538, "grad_norm": 0.6111427672427622, "learning_rate": 1.91649037472809e-06, "loss": 0.2041, "step": 20836 }, { "epoch": 1.4405115796750778, "grad_norm": 0.7257197548092446, "learning_rate": 1.9160497534079513e-06, "loss": 0.3809, "step": 20837 }, { "epoch": 1.4405807120636018, "grad_norm": 0.6674896495073581, "learning_rate": 1.9156091707396287e-06, "loss": 0.262, "step": 20838 }, { "epoch": 1.4406498444521258, "grad_norm": 0.7437371686267722, "learning_rate": 1.915168626728639e-06, "loss": 0.281, "step": 20839 }, { "epoch": 1.4407189768406499, "grad_norm": 0.6991234703501878, "learning_rate": 1.914728121380507e-06, "loss": 0.2735, "step": 20840 }, { "epoch": 1.4407881092291739, "grad_norm": 0.7308053327111346, "learning_rate": 1.9142876547007507e-06, "loss": 0.3638, "step": 20841 }, { "epoch": 1.440857241617698, "grad_norm": 0.7478748435819088, "learning_rate": 1.9138472266948933e-06, "loss": 0.2325, "step": 20842 }, { "epoch": 1.440926374006222, "grad_norm": 0.6481643426078049, "learning_rate": 1.9134068373684516e-06, "loss": 0.2387, "step": 20843 }, { "epoch": 1.440995506394746, "grad_norm": 0.7552996176078969, "learning_rate": 1.9129664867269477e-06, "loss": 0.396, "step": 20844 }, { "epoch": 1.44106463878327, "grad_norm": 0.6900171164984185, "learning_rate": 1.9125261747758993e-06, "loss": 0.297, "step": 20845 }, { "epoch": 1.441133771171794, "grad_norm": 0.6864716090055384, "learning_rate": 1.912085901520827e-06, "loss": 0.2368, "step": 20846 }, { "epoch": 1.441202903560318, "grad_norm": 0.7009528724237506, "learning_rate": 1.9116456669672463e-06, "loss": 0.2852, "step": 20847 }, { "epoch": 1.441272035948842, "grad_norm": 0.6849910526257431, "learning_rate": 1.911205471120677e-06, "loss": 0.3312, "step": 20848 }, { "epoch": 1.441341168337366, "grad_norm": 0.6309007823371119, "learning_rate": 1.9107653139866327e-06, "loss": 0.2655, "step": 20849 }, { "epoch": 1.44141030072589, "grad_norm": 0.6286077383992923, "learning_rate": 1.9103251955706344e-06, "loss": 0.2348, "step": 20850 }, { "epoch": 1.441479433114414, "grad_norm": 0.7205308126457194, "learning_rate": 1.9098851158781945e-06, "loss": 0.2722, "step": 20851 }, { "epoch": 1.441548565502938, "grad_norm": 0.6427040928495322, "learning_rate": 1.9094450749148296e-06, "loss": 0.2941, "step": 20852 }, { "epoch": 1.441617697891462, "grad_norm": 0.6675790012206969, "learning_rate": 1.909005072686056e-06, "loss": 0.2841, "step": 20853 }, { "epoch": 1.4416868302799861, "grad_norm": 0.6314560789096239, "learning_rate": 1.908565109197389e-06, "loss": 0.2082, "step": 20854 }, { "epoch": 1.4417559626685101, "grad_norm": 0.6604816340739256, "learning_rate": 1.90812518445434e-06, "loss": 0.3264, "step": 20855 }, { "epoch": 1.4418250950570342, "grad_norm": 0.821487596492993, "learning_rate": 1.907685298462425e-06, "loss": 0.2802, "step": 20856 }, { "epoch": 1.4418942274455582, "grad_norm": 0.6692108710389395, "learning_rate": 1.9072454512271553e-06, "loss": 0.2372, "step": 20857 }, { "epoch": 1.4419633598340822, "grad_norm": 0.704489334929, "learning_rate": 1.9068056427540444e-06, "loss": 0.2476, "step": 20858 }, { "epoch": 1.4420324922226062, "grad_norm": 0.7561292687865327, "learning_rate": 1.9063658730486051e-06, "loss": 0.3433, "step": 20859 }, { "epoch": 1.4421016246111305, "grad_norm": 0.6774739891118736, "learning_rate": 1.90592614211635e-06, "loss": 0.251, "step": 20860 }, { "epoch": 1.4421707569996545, "grad_norm": 0.5786760212381465, "learning_rate": 1.9054864499627873e-06, "loss": 0.2226, "step": 20861 }, { "epoch": 1.4422398893881785, "grad_norm": 0.709277097847072, "learning_rate": 1.9050467965934316e-06, "loss": 0.2799, "step": 20862 }, { "epoch": 1.4423090217767025, "grad_norm": 0.6787993860062453, "learning_rate": 1.9046071820137896e-06, "loss": 0.3191, "step": 20863 }, { "epoch": 1.4423781541652265, "grad_norm": 0.661240553167182, "learning_rate": 1.9041676062293735e-06, "loss": 0.2435, "step": 20864 }, { "epoch": 1.4424472865537505, "grad_norm": 0.7068249363202287, "learning_rate": 1.9037280692456905e-06, "loss": 0.2898, "step": 20865 }, { "epoch": 1.4425164189422746, "grad_norm": 0.6769951683777405, "learning_rate": 1.9032885710682507e-06, "loss": 0.3757, "step": 20866 }, { "epoch": 1.4425855513307986, "grad_norm": 0.6563555669821123, "learning_rate": 1.9028491117025621e-06, "loss": 0.2671, "step": 20867 }, { "epoch": 1.4426546837193226, "grad_norm": 0.5818965769854789, "learning_rate": 1.9024096911541345e-06, "loss": 0.1777, "step": 20868 }, { "epoch": 1.4427238161078466, "grad_norm": 0.6821302462218027, "learning_rate": 1.901970309428472e-06, "loss": 0.3313, "step": 20869 }, { "epoch": 1.4427929484963706, "grad_norm": 0.7715330402956811, "learning_rate": 1.901530966531085e-06, "loss": 0.2797, "step": 20870 }, { "epoch": 1.4428620808848946, "grad_norm": 0.6874120313100678, "learning_rate": 1.9010916624674757e-06, "loss": 0.2115, "step": 20871 }, { "epoch": 1.4429312132734187, "grad_norm": 0.7173278028098594, "learning_rate": 1.900652397243154e-06, "loss": 0.3275, "step": 20872 }, { "epoch": 1.4430003456619427, "grad_norm": 0.6709379926090772, "learning_rate": 1.9002131708636195e-06, "loss": 0.3699, "step": 20873 }, { "epoch": 1.4430694780504667, "grad_norm": 0.6683027687970163, "learning_rate": 1.8997739833343853e-06, "loss": 0.2903, "step": 20874 }, { "epoch": 1.4431386104389907, "grad_norm": 0.8207412499639296, "learning_rate": 1.8993348346609497e-06, "loss": 0.2225, "step": 20875 }, { "epoch": 1.4432077428275147, "grad_norm": 0.6929792256662517, "learning_rate": 1.8988957248488199e-06, "loss": 0.3125, "step": 20876 }, { "epoch": 1.4432768752160388, "grad_norm": 0.7178922420313864, "learning_rate": 1.8984566539034965e-06, "loss": 0.31, "step": 20877 }, { "epoch": 1.4433460076045628, "grad_norm": 0.697001066838978, "learning_rate": 1.8980176218304857e-06, "loss": 0.2242, "step": 20878 }, { "epoch": 1.4434151399930868, "grad_norm": 0.7389224494307808, "learning_rate": 1.897578628635287e-06, "loss": 0.2738, "step": 20879 }, { "epoch": 1.4434842723816108, "grad_norm": 0.7583796024879077, "learning_rate": 1.8971396743234033e-06, "loss": 0.3537, "step": 20880 }, { "epoch": 1.4435534047701348, "grad_norm": 0.8772205571523037, "learning_rate": 1.896700758900336e-06, "loss": 0.2898, "step": 20881 }, { "epoch": 1.4436225371586588, "grad_norm": 0.7459591539080863, "learning_rate": 1.8962618823715889e-06, "loss": 0.2165, "step": 20882 }, { "epoch": 1.4436916695471829, "grad_norm": 0.739677331948655, "learning_rate": 1.8958230447426584e-06, "loss": 0.2827, "step": 20883 }, { "epoch": 1.4437608019357069, "grad_norm": 0.6803365840359187, "learning_rate": 1.8953842460190486e-06, "loss": 0.3527, "step": 20884 }, { "epoch": 1.443829934324231, "grad_norm": 0.8379358362347508, "learning_rate": 1.894945486206256e-06, "loss": 0.2781, "step": 20885 }, { "epoch": 1.443899066712755, "grad_norm": 0.7010073232833826, "learning_rate": 1.8945067653097793e-06, "loss": 0.2206, "step": 20886 }, { "epoch": 1.443968199101279, "grad_norm": 0.6169709177772131, "learning_rate": 1.8940680833351184e-06, "loss": 0.2936, "step": 20887 }, { "epoch": 1.444037331489803, "grad_norm": 0.6740886610131027, "learning_rate": 1.8936294402877708e-06, "loss": 0.2686, "step": 20888 }, { "epoch": 1.444106463878327, "grad_norm": 0.6933187728102405, "learning_rate": 1.8931908361732366e-06, "loss": 0.2349, "step": 20889 }, { "epoch": 1.444175596266851, "grad_norm": 1.4150991982354255, "learning_rate": 1.8927522709970092e-06, "loss": 0.2282, "step": 20890 }, { "epoch": 1.444244728655375, "grad_norm": 0.6947721574756557, "learning_rate": 1.8923137447645884e-06, "loss": 0.3465, "step": 20891 }, { "epoch": 1.444313861043899, "grad_norm": 0.723588343696309, "learning_rate": 1.8918752574814675e-06, "loss": 0.2649, "step": 20892 }, { "epoch": 1.444382993432423, "grad_norm": 0.7197899723927502, "learning_rate": 1.8914368091531432e-06, "loss": 0.2355, "step": 20893 }, { "epoch": 1.444452125820947, "grad_norm": 2.0770005714099162, "learning_rate": 1.8909983997851117e-06, "loss": 0.3219, "step": 20894 }, { "epoch": 1.444521258209471, "grad_norm": 0.6928060360671714, "learning_rate": 1.890560029382868e-06, "loss": 0.3032, "step": 20895 }, { "epoch": 1.444590390597995, "grad_norm": 0.6576854828752796, "learning_rate": 1.8901216979519038e-06, "loss": 0.2453, "step": 20896 }, { "epoch": 1.4446595229865191, "grad_norm": 1.1219373392935645, "learning_rate": 1.8896834054977165e-06, "loss": 0.2452, "step": 20897 }, { "epoch": 1.4447286553750431, "grad_norm": 0.6875397954012245, "learning_rate": 1.8892451520257948e-06, "loss": 0.3545, "step": 20898 }, { "epoch": 1.4447977877635672, "grad_norm": 0.6517844567904094, "learning_rate": 1.888806937541635e-06, "loss": 0.2126, "step": 20899 }, { "epoch": 1.4448669201520912, "grad_norm": 0.6490591317877011, "learning_rate": 1.8883687620507273e-06, "loss": 0.2929, "step": 20900 }, { "epoch": 1.4449360525406152, "grad_norm": 0.7011658828674726, "learning_rate": 1.8879306255585638e-06, "loss": 0.3123, "step": 20901 }, { "epoch": 1.4450051849291392, "grad_norm": 0.7299875009633651, "learning_rate": 1.8874925280706363e-06, "loss": 0.29, "step": 20902 }, { "epoch": 1.4450743173176632, "grad_norm": 0.6440198395398997, "learning_rate": 1.887054469592437e-06, "loss": 0.2289, "step": 20903 }, { "epoch": 1.4451434497061872, "grad_norm": 0.6239520726269864, "learning_rate": 1.8866164501294531e-06, "loss": 0.2569, "step": 20904 }, { "epoch": 1.4452125820947113, "grad_norm": 0.7250216970943996, "learning_rate": 1.8861784696871777e-06, "loss": 0.3162, "step": 20905 }, { "epoch": 1.4452817144832353, "grad_norm": 0.8490658799425743, "learning_rate": 1.8857405282710961e-06, "loss": 0.2742, "step": 20906 }, { "epoch": 1.4453508468717593, "grad_norm": 0.6727132644580164, "learning_rate": 1.8853026258867013e-06, "loss": 0.1748, "step": 20907 }, { "epoch": 1.4454199792602833, "grad_norm": 0.6984540504844617, "learning_rate": 1.8848647625394768e-06, "loss": 0.2922, "step": 20908 }, { "epoch": 1.4454891116488073, "grad_norm": 0.6732032972070368, "learning_rate": 1.8844269382349157e-06, "loss": 0.3545, "step": 20909 }, { "epoch": 1.4455582440373314, "grad_norm": 0.7094322231437017, "learning_rate": 1.8839891529785021e-06, "loss": 0.2265, "step": 20910 }, { "epoch": 1.4456273764258556, "grad_norm": 0.6222111223388372, "learning_rate": 1.8835514067757248e-06, "loss": 0.2127, "step": 20911 }, { "epoch": 1.4456965088143796, "grad_norm": 0.7450376326441026, "learning_rate": 1.8831136996320675e-06, "loss": 0.3364, "step": 20912 }, { "epoch": 1.4457656412029036, "grad_norm": 0.7328171249229413, "learning_rate": 1.8826760315530195e-06, "loss": 0.3164, "step": 20913 }, { "epoch": 1.4458347735914276, "grad_norm": 0.8696442648439757, "learning_rate": 1.8822384025440627e-06, "loss": 0.2501, "step": 20914 }, { "epoch": 1.4459039059799517, "grad_norm": 0.6512736222326545, "learning_rate": 1.8818008126106836e-06, "loss": 0.2305, "step": 20915 }, { "epoch": 1.4459730383684757, "grad_norm": 0.7150861955768384, "learning_rate": 1.8813632617583665e-06, "loss": 0.3898, "step": 20916 }, { "epoch": 1.4460421707569997, "grad_norm": 0.9314459046282723, "learning_rate": 1.880925749992597e-06, "loss": 0.2446, "step": 20917 }, { "epoch": 1.4461113031455237, "grad_norm": 0.6742279108093946, "learning_rate": 1.8804882773188554e-06, "loss": 0.2238, "step": 20918 }, { "epoch": 1.4461804355340477, "grad_norm": 0.6838416724355715, "learning_rate": 1.8800508437426284e-06, "loss": 0.3157, "step": 20919 }, { "epoch": 1.4462495679225718, "grad_norm": 0.7829711947595244, "learning_rate": 1.8796134492693941e-06, "loss": 0.3614, "step": 20920 }, { "epoch": 1.4463187003110958, "grad_norm": 0.6321156183561863, "learning_rate": 1.8791760939046383e-06, "loss": 0.1992, "step": 20921 }, { "epoch": 1.4463878326996198, "grad_norm": 1.1646474681388528, "learning_rate": 1.878738777653839e-06, "loss": 0.2719, "step": 20922 }, { "epoch": 1.4464569650881438, "grad_norm": 0.7585330364381409, "learning_rate": 1.8783015005224792e-06, "loss": 0.31, "step": 20923 }, { "epoch": 1.4465260974766678, "grad_norm": 0.6636321992216068, "learning_rate": 1.8778642625160392e-06, "loss": 0.279, "step": 20924 }, { "epoch": 1.4465952298651918, "grad_norm": 0.6875727414354766, "learning_rate": 1.877427063640001e-06, "loss": 0.2176, "step": 20925 }, { "epoch": 1.4466643622537159, "grad_norm": 0.7483311147846543, "learning_rate": 1.8769899038998396e-06, "loss": 0.285, "step": 20926 }, { "epoch": 1.4467334946422399, "grad_norm": 0.8226751615867757, "learning_rate": 1.8765527833010388e-06, "loss": 0.4282, "step": 20927 }, { "epoch": 1.446802627030764, "grad_norm": 0.7060156632764533, "learning_rate": 1.8761157018490728e-06, "loss": 0.2483, "step": 20928 }, { "epoch": 1.446871759419288, "grad_norm": 0.6508151069374474, "learning_rate": 1.8756786595494219e-06, "loss": 0.2166, "step": 20929 }, { "epoch": 1.446940891807812, "grad_norm": 0.623486202854948, "learning_rate": 1.8752416564075632e-06, "loss": 0.2789, "step": 20930 }, { "epoch": 1.447010024196336, "grad_norm": 0.7761222866234558, "learning_rate": 1.874804692428976e-06, "loss": 0.2681, "step": 20931 }, { "epoch": 1.44707915658486, "grad_norm": 0.6605121649288933, "learning_rate": 1.8743677676191329e-06, "loss": 0.2569, "step": 20932 }, { "epoch": 1.447148288973384, "grad_norm": 0.676881614504483, "learning_rate": 1.8739308819835133e-06, "loss": 0.2561, "step": 20933 }, { "epoch": 1.447217421361908, "grad_norm": 0.6815865997287582, "learning_rate": 1.87349403552759e-06, "loss": 0.3785, "step": 20934 }, { "epoch": 1.447286553750432, "grad_norm": 0.677460262802902, "learning_rate": 1.873057228256841e-06, "loss": 0.2737, "step": 20935 }, { "epoch": 1.447355686138956, "grad_norm": 0.6144203502600455, "learning_rate": 1.8726204601767378e-06, "loss": 0.2503, "step": 20936 }, { "epoch": 1.44742481852748, "grad_norm": 0.6907624254759559, "learning_rate": 1.872183731292756e-06, "loss": 0.2962, "step": 20937 }, { "epoch": 1.447493950916004, "grad_norm": 1.7931461435625782, "learning_rate": 1.8717470416103695e-06, "loss": 0.3542, "step": 20938 }, { "epoch": 1.447563083304528, "grad_norm": 0.7609209049447357, "learning_rate": 1.8713103911350527e-06, "loss": 0.2087, "step": 20939 }, { "epoch": 1.4476322156930521, "grad_norm": 0.7473510596834825, "learning_rate": 1.8708737798722748e-06, "loss": 0.269, "step": 20940 }, { "epoch": 1.4477013480815764, "grad_norm": 1.1259873267458973, "learning_rate": 1.870437207827512e-06, "loss": 0.3573, "step": 20941 }, { "epoch": 1.4477704804701004, "grad_norm": 0.6885027344568647, "learning_rate": 1.8700006750062322e-06, "loss": 0.2218, "step": 20942 }, { "epoch": 1.4478396128586244, "grad_norm": 0.6775359936591606, "learning_rate": 1.86956418141391e-06, "loss": 0.2347, "step": 20943 }, { "epoch": 1.4479087452471484, "grad_norm": 0.6546003156809537, "learning_rate": 1.8691277270560104e-06, "loss": 0.3003, "step": 20944 }, { "epoch": 1.4479778776356724, "grad_norm": 0.8147285522000035, "learning_rate": 1.8686913119380108e-06, "loss": 0.257, "step": 20945 }, { "epoch": 1.4480470100241964, "grad_norm": 0.7492362539039757, "learning_rate": 1.8682549360653763e-06, "loss": 0.3023, "step": 20946 }, { "epoch": 1.4481161424127205, "grad_norm": 0.8265092108886639, "learning_rate": 1.8678185994435788e-06, "loss": 0.2222, "step": 20947 }, { "epoch": 1.4481852748012445, "grad_norm": 0.6579164372246606, "learning_rate": 1.8673823020780834e-06, "loss": 0.3404, "step": 20948 }, { "epoch": 1.4482544071897685, "grad_norm": 0.7327866821448291, "learning_rate": 1.866946043974363e-06, "loss": 0.2276, "step": 20949 }, { "epoch": 1.4483235395782925, "grad_norm": 0.6588951069373289, "learning_rate": 1.8665098251378804e-06, "loss": 0.2287, "step": 20950 }, { "epoch": 1.4483926719668165, "grad_norm": 0.6190334457709457, "learning_rate": 1.8660736455741052e-06, "loss": 0.2457, "step": 20951 }, { "epoch": 1.4484618043553406, "grad_norm": 0.6888908986210355, "learning_rate": 1.8656375052885062e-06, "loss": 0.322, "step": 20952 }, { "epoch": 1.4485309367438646, "grad_norm": 0.6300920111300635, "learning_rate": 1.8652014042865457e-06, "loss": 0.192, "step": 20953 }, { "epoch": 1.4486000691323886, "grad_norm": 0.7368152620385985, "learning_rate": 1.8647653425736929e-06, "loss": 0.2797, "step": 20954 }, { "epoch": 1.4486692015209126, "grad_norm": 0.7327689375645431, "learning_rate": 1.8643293201554097e-06, "loss": 0.3712, "step": 20955 }, { "epoch": 1.4487383339094366, "grad_norm": 0.6766326804224856, "learning_rate": 1.8638933370371648e-06, "loss": 0.2109, "step": 20956 }, { "epoch": 1.4488074662979606, "grad_norm": 0.7253961368244064, "learning_rate": 1.8634573932244182e-06, "loss": 0.2204, "step": 20957 }, { "epoch": 1.4488765986864847, "grad_norm": 0.6591308142528046, "learning_rate": 1.8630214887226356e-06, "loss": 0.2208, "step": 20958 }, { "epoch": 1.4489457310750087, "grad_norm": 0.692598005167786, "learning_rate": 1.8625856235372808e-06, "loss": 0.3314, "step": 20959 }, { "epoch": 1.4490148634635327, "grad_norm": 0.701142420075469, "learning_rate": 1.8621497976738173e-06, "loss": 0.2442, "step": 20960 }, { "epoch": 1.4490839958520567, "grad_norm": 0.5995457869980827, "learning_rate": 1.8617140111377053e-06, "loss": 0.2131, "step": 20961 }, { "epoch": 1.4491531282405807, "grad_norm": 0.671896547762976, "learning_rate": 1.861278263934409e-06, "loss": 0.3238, "step": 20962 }, { "epoch": 1.4492222606291048, "grad_norm": 0.7718480678833572, "learning_rate": 1.860842556069387e-06, "loss": 0.2834, "step": 20963 }, { "epoch": 1.4492913930176288, "grad_norm": 0.6789551661416413, "learning_rate": 1.8604068875481014e-06, "loss": 0.2681, "step": 20964 }, { "epoch": 1.4493605254061528, "grad_norm": 0.7054780068317543, "learning_rate": 1.859971258376012e-06, "loss": 0.3006, "step": 20965 }, { "epoch": 1.4494296577946768, "grad_norm": 0.7106790685053467, "learning_rate": 1.859535668558582e-06, "loss": 0.3795, "step": 20966 }, { "epoch": 1.4494987901832008, "grad_norm": 0.6101613308869243, "learning_rate": 1.8591001181012659e-06, "loss": 0.2514, "step": 20967 }, { "epoch": 1.4495679225717248, "grad_norm": 0.6700505649639074, "learning_rate": 1.858664607009526e-06, "loss": 0.2471, "step": 20968 }, { "epoch": 1.4496370549602489, "grad_norm": 0.6865310900992665, "learning_rate": 1.8582291352888177e-06, "loss": 0.2911, "step": 20969 }, { "epoch": 1.4497061873487729, "grad_norm": 0.6958003487110742, "learning_rate": 1.8577937029446019e-06, "loss": 0.2703, "step": 20970 }, { "epoch": 1.449775319737297, "grad_norm": 0.6564635297627788, "learning_rate": 1.8573583099823334e-06, "loss": 0.2688, "step": 20971 }, { "epoch": 1.449844452125821, "grad_norm": 0.6339368249343109, "learning_rate": 1.8569229564074699e-06, "loss": 0.251, "step": 20972 }, { "epoch": 1.449913584514345, "grad_norm": 0.7375574838419683, "learning_rate": 1.8564876422254679e-06, "loss": 0.3241, "step": 20973 }, { "epoch": 1.449982716902869, "grad_norm": 0.6505647164422119, "learning_rate": 1.8560523674417853e-06, "loss": 0.3149, "step": 20974 }, { "epoch": 1.450051849291393, "grad_norm": 0.7972081067007338, "learning_rate": 1.8556171320618738e-06, "loss": 0.2276, "step": 20975 }, { "epoch": 1.450120981679917, "grad_norm": 0.7479292771261011, "learning_rate": 1.8551819360911922e-06, "loss": 0.2813, "step": 20976 }, { "epoch": 1.450190114068441, "grad_norm": 0.8007224465208079, "learning_rate": 1.854746779535191e-06, "loss": 0.3091, "step": 20977 }, { "epoch": 1.450259246456965, "grad_norm": 0.6520393327452765, "learning_rate": 1.8543116623993279e-06, "loss": 0.1907, "step": 20978 }, { "epoch": 1.450328378845489, "grad_norm": 0.7587540199353999, "learning_rate": 1.8538765846890504e-06, "loss": 0.2979, "step": 20979 }, { "epoch": 1.450397511234013, "grad_norm": 0.6713442128586484, "learning_rate": 1.8534415464098194e-06, "loss": 0.3359, "step": 20980 }, { "epoch": 1.450466643622537, "grad_norm": 0.7165231932801138, "learning_rate": 1.8530065475670812e-06, "loss": 0.2943, "step": 20981 }, { "epoch": 1.450535776011061, "grad_norm": 0.7184094137265544, "learning_rate": 1.8525715881662915e-06, "loss": 0.2433, "step": 20982 }, { "epoch": 1.4506049083995851, "grad_norm": 0.6850512265857917, "learning_rate": 1.8521366682128988e-06, "loss": 0.2842, "step": 20983 }, { "epoch": 1.4506740407881091, "grad_norm": 0.7246525239954884, "learning_rate": 1.8517017877123566e-06, "loss": 0.3832, "step": 20984 }, { "epoch": 1.4507431731766331, "grad_norm": 0.6548336986330832, "learning_rate": 1.8512669466701127e-06, "loss": 0.209, "step": 20985 }, { "epoch": 1.4508123055651572, "grad_norm": 0.6337217801720412, "learning_rate": 1.8508321450916184e-06, "loss": 0.2572, "step": 20986 }, { "epoch": 1.4508814379536812, "grad_norm": 0.6700686064027762, "learning_rate": 1.8503973829823235e-06, "loss": 0.3446, "step": 20987 }, { "epoch": 1.4509505703422052, "grad_norm": 0.6766621476993698, "learning_rate": 1.8499626603476777e-06, "loss": 0.2953, "step": 20988 }, { "epoch": 1.4510197027307292, "grad_norm": 0.6993401022400718, "learning_rate": 1.8495279771931273e-06, "loss": 0.2792, "step": 20989 }, { "epoch": 1.4510888351192532, "grad_norm": 0.6601243739950378, "learning_rate": 1.8490933335241235e-06, "loss": 0.2804, "step": 20990 }, { "epoch": 1.4511579675077773, "grad_norm": 0.7444669174588723, "learning_rate": 1.8486587293461094e-06, "loss": 0.3501, "step": 20991 }, { "epoch": 1.4512270998963015, "grad_norm": 0.6340538544884766, "learning_rate": 1.848224164664536e-06, "loss": 0.2257, "step": 20992 }, { "epoch": 1.4512962322848255, "grad_norm": 0.6173543386498236, "learning_rate": 1.8477896394848467e-06, "loss": 0.182, "step": 20993 }, { "epoch": 1.4513653646733495, "grad_norm": 0.650243469996965, "learning_rate": 1.8473551538124883e-06, "loss": 0.2569, "step": 20994 }, { "epoch": 1.4514344970618736, "grad_norm": 0.7982216087464755, "learning_rate": 1.8469207076529077e-06, "loss": 0.2854, "step": 20995 }, { "epoch": 1.4515036294503976, "grad_norm": 0.6311710669625589, "learning_rate": 1.8464863010115503e-06, "loss": 0.2443, "step": 20996 }, { "epoch": 1.4515727618389216, "grad_norm": 0.6694450624579926, "learning_rate": 1.8460519338938576e-06, "loss": 0.2806, "step": 20997 }, { "epoch": 1.4516418942274456, "grad_norm": 0.6922522359507164, "learning_rate": 1.8456176063052777e-06, "loss": 0.3279, "step": 20998 }, { "epoch": 1.4517110266159696, "grad_norm": 0.8326372270029427, "learning_rate": 1.8451833182512496e-06, "loss": 0.2044, "step": 20999 }, { "epoch": 1.4517801590044936, "grad_norm": 0.8201319434509846, "learning_rate": 1.8447490697372188e-06, "loss": 0.2489, "step": 21000 }, { "epoch": 1.4518492913930177, "grad_norm": 0.6686393649667024, "learning_rate": 1.844314860768628e-06, "loss": 0.2479, "step": 21001 }, { "epoch": 1.4519184237815417, "grad_norm": 0.7133087403278268, "learning_rate": 1.84388069135092e-06, "loss": 0.3087, "step": 21002 }, { "epoch": 1.4519875561700657, "grad_norm": 0.634519363591294, "learning_rate": 1.8434465614895336e-06, "loss": 0.2602, "step": 21003 }, { "epoch": 1.4520566885585897, "grad_norm": 0.6567894680836686, "learning_rate": 1.8430124711899133e-06, "loss": 0.2124, "step": 21004 }, { "epoch": 1.4521258209471137, "grad_norm": 0.6661064265470648, "learning_rate": 1.8425784204574958e-06, "loss": 0.3463, "step": 21005 }, { "epoch": 1.4521949533356378, "grad_norm": 0.7459714407337876, "learning_rate": 1.8421444092977253e-06, "loss": 0.3412, "step": 21006 }, { "epoch": 1.4522640857241618, "grad_norm": 0.6983340590802948, "learning_rate": 1.8417104377160371e-06, "loss": 0.2346, "step": 21007 }, { "epoch": 1.4523332181126858, "grad_norm": 0.6432105139696115, "learning_rate": 1.8412765057178728e-06, "loss": 0.2473, "step": 21008 }, { "epoch": 1.4524023505012098, "grad_norm": 0.6852569376991479, "learning_rate": 1.84084261330867e-06, "loss": 0.3675, "step": 21009 }, { "epoch": 1.4524714828897338, "grad_norm": 0.6378030727872048, "learning_rate": 1.8404087604938691e-06, "loss": 0.2386, "step": 21010 }, { "epoch": 1.4525406152782578, "grad_norm": 0.6298004902277469, "learning_rate": 1.839974947278904e-06, "loss": 0.2019, "step": 21011 }, { "epoch": 1.4526097476667819, "grad_norm": 0.7053257255880805, "learning_rate": 1.8395411736692154e-06, "loss": 0.3417, "step": 21012 }, { "epoch": 1.4526788800553059, "grad_norm": 0.649254542960834, "learning_rate": 1.839107439670237e-06, "loss": 0.2743, "step": 21013 }, { "epoch": 1.45274801244383, "grad_norm": 0.7061904505895639, "learning_rate": 1.838673745287407e-06, "loss": 0.2621, "step": 21014 }, { "epoch": 1.452817144832354, "grad_norm": 0.7631316939125982, "learning_rate": 1.8382400905261565e-06, "loss": 0.2564, "step": 21015 }, { "epoch": 1.452886277220878, "grad_norm": 0.67629836743662, "learning_rate": 1.8378064753919277e-06, "loss": 0.376, "step": 21016 }, { "epoch": 1.452955409609402, "grad_norm": 0.6472288738059624, "learning_rate": 1.8373728998901497e-06, "loss": 0.1935, "step": 21017 }, { "epoch": 1.453024541997926, "grad_norm": 0.6379256342805696, "learning_rate": 1.8369393640262606e-06, "loss": 0.2384, "step": 21018 }, { "epoch": 1.45309367438645, "grad_norm": 0.6938838567847522, "learning_rate": 1.8365058678056913e-06, "loss": 0.2766, "step": 21019 }, { "epoch": 1.453162806774974, "grad_norm": 0.7857804033375044, "learning_rate": 1.8360724112338741e-06, "loss": 0.3124, "step": 21020 }, { "epoch": 1.453231939163498, "grad_norm": 0.5782637916220624, "learning_rate": 1.8356389943162423e-06, "loss": 0.2403, "step": 21021 }, { "epoch": 1.4533010715520223, "grad_norm": 0.6738899221496346, "learning_rate": 1.8352056170582295e-06, "loss": 0.2295, "step": 21022 }, { "epoch": 1.4533702039405463, "grad_norm": 0.7181025681107346, "learning_rate": 1.834772279465268e-06, "loss": 0.3706, "step": 21023 }, { "epoch": 1.4534393363290703, "grad_norm": 0.6990600453794007, "learning_rate": 1.8343389815427849e-06, "loss": 0.26, "step": 21024 }, { "epoch": 1.4535084687175943, "grad_norm": 0.6790262682272032, "learning_rate": 1.833905723296216e-06, "loss": 0.2072, "step": 21025 }, { "epoch": 1.4535776011061183, "grad_norm": 0.6856812619234431, "learning_rate": 1.8334725047309866e-06, "loss": 0.2952, "step": 21026 }, { "epoch": 1.4536467334946424, "grad_norm": 0.7023245847858459, "learning_rate": 1.8330393258525297e-06, "loss": 0.3632, "step": 21027 }, { "epoch": 1.4537158658831664, "grad_norm": 0.6464787686872325, "learning_rate": 1.832606186666272e-06, "loss": 0.2652, "step": 21028 }, { "epoch": 1.4537849982716904, "grad_norm": 0.6674404098575728, "learning_rate": 1.8321730871776428e-06, "loss": 0.2938, "step": 21029 }, { "epoch": 1.4538541306602144, "grad_norm": 0.7002558997977498, "learning_rate": 1.831740027392071e-06, "loss": 0.3241, "step": 21030 }, { "epoch": 1.4539232630487384, "grad_norm": 0.727326197509352, "learning_rate": 1.8313070073149858e-06, "loss": 0.3206, "step": 21031 }, { "epoch": 1.4539923954372624, "grad_norm": 0.6676375268052863, "learning_rate": 1.8308740269518104e-06, "loss": 0.2314, "step": 21032 }, { "epoch": 1.4540615278257865, "grad_norm": 0.677298943699338, "learning_rate": 1.8304410863079757e-06, "loss": 0.3025, "step": 21033 }, { "epoch": 1.4541306602143105, "grad_norm": 0.6805120838605057, "learning_rate": 1.830008185388904e-06, "loss": 0.3071, "step": 21034 }, { "epoch": 1.4541997926028345, "grad_norm": 0.7150804103480194, "learning_rate": 1.8295753242000218e-06, "loss": 0.2509, "step": 21035 }, { "epoch": 1.4542689249913585, "grad_norm": 0.6959827890543144, "learning_rate": 1.8291425027467557e-06, "loss": 0.2495, "step": 21036 }, { "epoch": 1.4543380573798825, "grad_norm": 0.7065177745103179, "learning_rate": 1.8287097210345312e-06, "loss": 0.3506, "step": 21037 }, { "epoch": 1.4544071897684065, "grad_norm": 0.7042134375027765, "learning_rate": 1.8282769790687693e-06, "loss": 0.3127, "step": 21038 }, { "epoch": 1.4544763221569306, "grad_norm": 0.6461031529558042, "learning_rate": 1.827844276854897e-06, "loss": 0.2367, "step": 21039 }, { "epoch": 1.4545454545454546, "grad_norm": 0.6958967501665516, "learning_rate": 1.8274116143983338e-06, "loss": 0.2398, "step": 21040 }, { "epoch": 1.4546145869339786, "grad_norm": 0.639771547040487, "learning_rate": 1.826978991704506e-06, "loss": 0.3658, "step": 21041 }, { "epoch": 1.4546837193225026, "grad_norm": 0.6668375950743168, "learning_rate": 1.8265464087788326e-06, "loss": 0.2268, "step": 21042 }, { "epoch": 1.4547528517110266, "grad_norm": 0.6656173239492704, "learning_rate": 1.8261138656267368e-06, "loss": 0.256, "step": 21043 }, { "epoch": 1.4548219840995507, "grad_norm": 0.6931298287704993, "learning_rate": 1.825681362253639e-06, "loss": 0.3121, "step": 21044 }, { "epoch": 1.4548911164880747, "grad_norm": 0.6783496342569008, "learning_rate": 1.825248898664963e-06, "loss": 0.2824, "step": 21045 }, { "epoch": 1.4549602488765987, "grad_norm": 0.6358942183974715, "learning_rate": 1.824816474866124e-06, "loss": 0.248, "step": 21046 }, { "epoch": 1.4550293812651227, "grad_norm": 0.731780547917214, "learning_rate": 1.8243840908625465e-06, "loss": 0.2369, "step": 21047 }, { "epoch": 1.4550985136536467, "grad_norm": 0.6980855631924533, "learning_rate": 1.8239517466596452e-06, "loss": 0.351, "step": 21048 }, { "epoch": 1.4551676460421707, "grad_norm": 0.6887888039447034, "learning_rate": 1.8235194422628422e-06, "loss": 0.2567, "step": 21049 }, { "epoch": 1.4552367784306948, "grad_norm": 0.7482809834163099, "learning_rate": 1.8230871776775532e-06, "loss": 0.231, "step": 21050 }, { "epoch": 1.4553059108192188, "grad_norm": 0.7525586844213793, "learning_rate": 1.822654952909197e-06, "loss": 0.3343, "step": 21051 }, { "epoch": 1.4553750432077428, "grad_norm": 0.6833470360740379, "learning_rate": 1.8222227679631904e-06, "loss": 0.3105, "step": 21052 }, { "epoch": 1.4554441755962668, "grad_norm": 0.7633812214468967, "learning_rate": 1.8217906228449521e-06, "loss": 0.2256, "step": 21053 }, { "epoch": 1.4555133079847908, "grad_norm": 1.0654884564608775, "learning_rate": 1.8213585175598947e-06, "loss": 0.2684, "step": 21054 }, { "epoch": 1.4555824403733149, "grad_norm": 0.6943707901097121, "learning_rate": 1.8209264521134379e-06, "loss": 0.3224, "step": 21055 }, { "epoch": 1.4556515727618389, "grad_norm": 0.6832879548841235, "learning_rate": 1.820494426510993e-06, "loss": 0.2939, "step": 21056 }, { "epoch": 1.455720705150363, "grad_norm": 0.650657400899635, "learning_rate": 1.820062440757976e-06, "loss": 0.2332, "step": 21057 }, { "epoch": 1.455789837538887, "grad_norm": 0.7566159215791864, "learning_rate": 1.8196304948598015e-06, "loss": 0.2477, "step": 21058 }, { "epoch": 1.455858969927411, "grad_norm": 0.7551439388803066, "learning_rate": 1.819198588821885e-06, "loss": 0.3547, "step": 21059 }, { "epoch": 1.455928102315935, "grad_norm": 0.6958076278493255, "learning_rate": 1.8187667226496364e-06, "loss": 0.2141, "step": 21060 }, { "epoch": 1.455997234704459, "grad_norm": 0.7367063258494816, "learning_rate": 1.818334896348471e-06, "loss": 0.2421, "step": 21061 }, { "epoch": 1.456066367092983, "grad_norm": 0.7025782570258445, "learning_rate": 1.8179031099237982e-06, "loss": 0.3523, "step": 21062 }, { "epoch": 1.456135499481507, "grad_norm": 0.6641249357277138, "learning_rate": 1.8174713633810326e-06, "loss": 0.304, "step": 21063 }, { "epoch": 1.456204631870031, "grad_norm": 0.6639381381917808, "learning_rate": 1.8170396567255826e-06, "loss": 0.2527, "step": 21064 }, { "epoch": 1.456273764258555, "grad_norm": 0.9445025361112673, "learning_rate": 1.8166079899628602e-06, "loss": 0.2599, "step": 21065 }, { "epoch": 1.456342896647079, "grad_norm": 0.6944069489378624, "learning_rate": 1.816176363098276e-06, "loss": 0.3617, "step": 21066 }, { "epoch": 1.456412029035603, "grad_norm": 0.7179247111199771, "learning_rate": 1.8157447761372403e-06, "loss": 0.2452, "step": 21067 }, { "epoch": 1.456481161424127, "grad_norm": 0.6435377950479122, "learning_rate": 1.8153132290851599e-06, "loss": 0.2291, "step": 21068 }, { "epoch": 1.456550293812651, "grad_norm": 0.6823091935942868, "learning_rate": 1.8148817219474469e-06, "loss": 0.3401, "step": 21069 }, { "epoch": 1.4566194262011751, "grad_norm": 0.7175202372853406, "learning_rate": 1.8144502547295051e-06, "loss": 0.3107, "step": 21070 }, { "epoch": 1.4566885585896991, "grad_norm": 0.6569874484451839, "learning_rate": 1.814018827436745e-06, "loss": 0.2491, "step": 21071 }, { "epoch": 1.4567576909782232, "grad_norm": 0.7127070201976337, "learning_rate": 1.8135874400745728e-06, "loss": 0.2353, "step": 21072 }, { "epoch": 1.4568268233667474, "grad_norm": 0.69929427802316, "learning_rate": 1.8131560926483976e-06, "loss": 0.3736, "step": 21073 }, { "epoch": 1.4568959557552714, "grad_norm": 0.6522710255228519, "learning_rate": 1.812724785163622e-06, "loss": 0.2178, "step": 21074 }, { "epoch": 1.4569650881437954, "grad_norm": 0.6730977457671433, "learning_rate": 1.8122935176256546e-06, "loss": 0.1789, "step": 21075 }, { "epoch": 1.4570342205323195, "grad_norm": 0.7000852931928228, "learning_rate": 1.811862290039898e-06, "loss": 0.3083, "step": 21076 }, { "epoch": 1.4571033529208435, "grad_norm": 0.7150085569597446, "learning_rate": 1.8114311024117593e-06, "loss": 0.3429, "step": 21077 }, { "epoch": 1.4571724853093675, "grad_norm": 0.6372584188640215, "learning_rate": 1.81099995474664e-06, "loss": 0.2171, "step": 21078 }, { "epoch": 1.4572416176978915, "grad_norm": 0.6804231240871011, "learning_rate": 1.8105688470499456e-06, "loss": 0.1945, "step": 21079 }, { "epoch": 1.4573107500864155, "grad_norm": 0.7740509232072422, "learning_rate": 1.8101377793270786e-06, "loss": 0.3285, "step": 21080 }, { "epoch": 1.4573798824749395, "grad_norm": 0.7102192820524774, "learning_rate": 1.8097067515834437e-06, "loss": 0.2952, "step": 21081 }, { "epoch": 1.4574490148634636, "grad_norm": 0.7797547116101835, "learning_rate": 1.8092757638244397e-06, "loss": 0.2843, "step": 21082 }, { "epoch": 1.4575181472519876, "grad_norm": 0.7097102895837021, "learning_rate": 1.8088448160554712e-06, "loss": 0.2875, "step": 21083 }, { "epoch": 1.4575872796405116, "grad_norm": 0.6868294155356368, "learning_rate": 1.8084139082819368e-06, "loss": 0.3238, "step": 21084 }, { "epoch": 1.4576564120290356, "grad_norm": 0.6693042466036477, "learning_rate": 1.8079830405092396e-06, "loss": 0.2436, "step": 21085 }, { "epoch": 1.4577255444175596, "grad_norm": 0.7373489324506665, "learning_rate": 1.8075522127427775e-06, "loss": 0.2161, "step": 21086 }, { "epoch": 1.4577946768060837, "grad_norm": 0.7286013750744998, "learning_rate": 1.8071214249879509e-06, "loss": 0.4032, "step": 21087 }, { "epoch": 1.4578638091946077, "grad_norm": 0.7148832871345369, "learning_rate": 1.8066906772501614e-06, "loss": 0.2753, "step": 21088 }, { "epoch": 1.4579329415831317, "grad_norm": 0.7990371812156203, "learning_rate": 1.8062599695348032e-06, "loss": 0.2695, "step": 21089 }, { "epoch": 1.4580020739716557, "grad_norm": 0.6423910542073126, "learning_rate": 1.8058293018472788e-06, "loss": 0.2476, "step": 21090 }, { "epoch": 1.4580712063601797, "grad_norm": 0.6780985214992827, "learning_rate": 1.8053986741929824e-06, "loss": 0.3862, "step": 21091 }, { "epoch": 1.4581403387487037, "grad_norm": 1.0100528044321082, "learning_rate": 1.8049680865773122e-06, "loss": 0.2209, "step": 21092 }, { "epoch": 1.4582094711372278, "grad_norm": 0.6182874227671116, "learning_rate": 1.8045375390056658e-06, "loss": 0.2215, "step": 21093 }, { "epoch": 1.4582786035257518, "grad_norm": 0.6684422345767067, "learning_rate": 1.8041070314834396e-06, "loss": 0.3317, "step": 21094 }, { "epoch": 1.4583477359142758, "grad_norm": 1.0875100278583396, "learning_rate": 1.8036765640160274e-06, "loss": 0.2557, "step": 21095 }, { "epoch": 1.4584168683027998, "grad_norm": 0.6807612526361234, "learning_rate": 1.8032461366088272e-06, "loss": 0.2663, "step": 21096 }, { "epoch": 1.4584860006913238, "grad_norm": 0.6690122033383317, "learning_rate": 1.8028157492672293e-06, "loss": 0.2737, "step": 21097 }, { "epoch": 1.4585551330798479, "grad_norm": 0.6394958518609377, "learning_rate": 1.8023854019966331e-06, "loss": 0.3269, "step": 21098 }, { "epoch": 1.4586242654683719, "grad_norm": 0.6935920035107722, "learning_rate": 1.8019550948024271e-06, "loss": 0.2582, "step": 21099 }, { "epoch": 1.458693397856896, "grad_norm": 0.8594460570228716, "learning_rate": 1.8015248276900066e-06, "loss": 0.2716, "step": 21100 }, { "epoch": 1.45876253024542, "grad_norm": 0.7103459350268885, "learning_rate": 1.8010946006647651e-06, "loss": 0.2987, "step": 21101 }, { "epoch": 1.458831662633944, "grad_norm": 0.6932552422485717, "learning_rate": 1.8006644137320956e-06, "loss": 0.2913, "step": 21102 }, { "epoch": 1.458900795022468, "grad_norm": 0.8562549517880951, "learning_rate": 1.8002342668973867e-06, "loss": 0.2106, "step": 21103 }, { "epoch": 1.4589699274109922, "grad_norm": 0.6824243969601163, "learning_rate": 1.7998041601660322e-06, "loss": 0.2532, "step": 21104 }, { "epoch": 1.4590390597995162, "grad_norm": 0.7639782057589551, "learning_rate": 1.7993740935434202e-06, "loss": 0.335, "step": 21105 }, { "epoch": 1.4591081921880402, "grad_norm": 0.6975322520628253, "learning_rate": 1.7989440670349417e-06, "loss": 0.2644, "step": 21106 }, { "epoch": 1.4591773245765642, "grad_norm": 0.6536811707712302, "learning_rate": 1.7985140806459872e-06, "loss": 0.238, "step": 21107 }, { "epoch": 1.4592464569650883, "grad_norm": 0.722162488656726, "learning_rate": 1.798084134381947e-06, "loss": 0.2283, "step": 21108 }, { "epoch": 1.4593155893536123, "grad_norm": 0.6804462100690244, "learning_rate": 1.7976542282482062e-06, "loss": 0.307, "step": 21109 }, { "epoch": 1.4593847217421363, "grad_norm": 0.6739670008594507, "learning_rate": 1.7972243622501568e-06, "loss": 0.2114, "step": 21110 }, { "epoch": 1.4594538541306603, "grad_norm": 0.6412356455546644, "learning_rate": 1.7967945363931827e-06, "loss": 0.2415, "step": 21111 }, { "epoch": 1.4595229865191843, "grad_norm": 0.6810476533846862, "learning_rate": 1.7963647506826741e-06, "loss": 0.3337, "step": 21112 }, { "epoch": 1.4595921189077083, "grad_norm": 0.7228822999727375, "learning_rate": 1.7959350051240148e-06, "loss": 0.2763, "step": 21113 }, { "epoch": 1.4596612512962324, "grad_norm": 0.6485244819631972, "learning_rate": 1.7955052997225924e-06, "loss": 0.2475, "step": 21114 }, { "epoch": 1.4597303836847564, "grad_norm": 0.6587837481868556, "learning_rate": 1.7950756344837923e-06, "loss": 0.2568, "step": 21115 }, { "epoch": 1.4597995160732804, "grad_norm": 0.6536050638310291, "learning_rate": 1.7946460094130015e-06, "loss": 0.3413, "step": 21116 }, { "epoch": 1.4598686484618044, "grad_norm": 0.6593208682614244, "learning_rate": 1.7942164245156008e-06, "loss": 0.2101, "step": 21117 }, { "epoch": 1.4599377808503284, "grad_norm": 0.6469642219992835, "learning_rate": 1.7937868797969776e-06, "loss": 0.2527, "step": 21118 }, { "epoch": 1.4600069132388525, "grad_norm": 0.8213672452455079, "learning_rate": 1.7933573752625132e-06, "loss": 0.2758, "step": 21119 }, { "epoch": 1.4600760456273765, "grad_norm": 0.6247011604054127, "learning_rate": 1.792927910917593e-06, "loss": 0.2396, "step": 21120 }, { "epoch": 1.4601451780159005, "grad_norm": 0.7000887880411067, "learning_rate": 1.7924984867675959e-06, "loss": 0.2371, "step": 21121 }, { "epoch": 1.4602143104044245, "grad_norm": 0.8223443566448142, "learning_rate": 1.7920691028179072e-06, "loss": 0.3424, "step": 21122 }, { "epoch": 1.4602834427929485, "grad_norm": 0.7346204740280078, "learning_rate": 1.791639759073907e-06, "loss": 0.392, "step": 21123 }, { "epoch": 1.4603525751814725, "grad_norm": 0.7140675057140567, "learning_rate": 1.7912104555409782e-06, "loss": 0.2999, "step": 21124 }, { "epoch": 1.4604217075699966, "grad_norm": 0.6462164115906872, "learning_rate": 1.790781192224499e-06, "loss": 0.1939, "step": 21125 }, { "epoch": 1.4604908399585206, "grad_norm": 0.6388333684722267, "learning_rate": 1.7903519691298516e-06, "loss": 0.316, "step": 21126 }, { "epoch": 1.4605599723470446, "grad_norm": 0.8382755205577543, "learning_rate": 1.789922786262413e-06, "loss": 0.3358, "step": 21127 }, { "epoch": 1.4606291047355686, "grad_norm": 0.6543006766911635, "learning_rate": 1.7894936436275634e-06, "loss": 0.2361, "step": 21128 }, { "epoch": 1.4606982371240926, "grad_norm": 0.7232482281852193, "learning_rate": 1.7890645412306818e-06, "loss": 0.2823, "step": 21129 }, { "epoch": 1.4607673695126167, "grad_norm": 0.7107312768685343, "learning_rate": 1.7886354790771477e-06, "loss": 0.3335, "step": 21130 }, { "epoch": 1.4608365019011407, "grad_norm": 0.7204197379070952, "learning_rate": 1.7882064571723351e-06, "loss": 0.2997, "step": 21131 }, { "epoch": 1.4609056342896647, "grad_norm": 1.1432007388966643, "learning_rate": 1.7877774755216243e-06, "loss": 0.2572, "step": 21132 }, { "epoch": 1.4609747666781887, "grad_norm": 0.8614006074734218, "learning_rate": 1.7873485341303892e-06, "loss": 0.2865, "step": 21133 }, { "epoch": 1.4610438990667127, "grad_norm": 0.712610385880455, "learning_rate": 1.7869196330040083e-06, "loss": 0.3442, "step": 21134 }, { "epoch": 1.4611130314552367, "grad_norm": 0.672850171344419, "learning_rate": 1.7864907721478542e-06, "loss": 0.2425, "step": 21135 }, { "epoch": 1.4611821638437608, "grad_norm": 0.6252225337342031, "learning_rate": 1.7860619515673034e-06, "loss": 0.2032, "step": 21136 }, { "epoch": 1.4612512962322848, "grad_norm": 0.695774345120333, "learning_rate": 1.7856331712677304e-06, "loss": 0.3012, "step": 21137 }, { "epoch": 1.4613204286208088, "grad_norm": 0.6387685693734608, "learning_rate": 1.7852044312545113e-06, "loss": 0.2762, "step": 21138 }, { "epoch": 1.4613895610093328, "grad_norm": 0.7251975751709664, "learning_rate": 1.784775731533016e-06, "loss": 0.2967, "step": 21139 }, { "epoch": 1.4614586933978568, "grad_norm": 0.7746391074280997, "learning_rate": 1.7843470721086208e-06, "loss": 0.2652, "step": 21140 }, { "epoch": 1.4615278257863809, "grad_norm": 0.6802216432712739, "learning_rate": 1.7839184529866943e-06, "loss": 0.3322, "step": 21141 }, { "epoch": 1.4615969581749049, "grad_norm": 0.687435724995116, "learning_rate": 1.7834898741726114e-06, "loss": 0.2277, "step": 21142 }, { "epoch": 1.4616660905634289, "grad_norm": 0.6423556982455532, "learning_rate": 1.783061335671742e-06, "loss": 0.249, "step": 21143 }, { "epoch": 1.461735222951953, "grad_norm": 0.6677069331872127, "learning_rate": 1.7826328374894602e-06, "loss": 0.3134, "step": 21144 }, { "epoch": 1.461804355340477, "grad_norm": 0.8196354859982006, "learning_rate": 1.7822043796311316e-06, "loss": 0.2851, "step": 21145 }, { "epoch": 1.461873487729001, "grad_norm": 0.7089882232449937, "learning_rate": 1.7817759621021313e-06, "loss": 0.2384, "step": 21146 }, { "epoch": 1.461942620117525, "grad_norm": 0.6061529193746974, "learning_rate": 1.7813475849078237e-06, "loss": 0.262, "step": 21147 }, { "epoch": 1.462011752506049, "grad_norm": 0.6936901320466969, "learning_rate": 1.7809192480535826e-06, "loss": 0.3033, "step": 21148 }, { "epoch": 1.462080884894573, "grad_norm": 0.6715853630270335, "learning_rate": 1.7804909515447722e-06, "loss": 0.2596, "step": 21149 }, { "epoch": 1.462150017283097, "grad_norm": 0.708200575717223, "learning_rate": 1.7800626953867616e-06, "loss": 0.2523, "step": 21150 }, { "epoch": 1.462219149671621, "grad_norm": 1.2863223257007197, "learning_rate": 1.7796344795849213e-06, "loss": 0.2599, "step": 21151 }, { "epoch": 1.462288282060145, "grad_norm": 0.6482993940688044, "learning_rate": 1.7792063041446133e-06, "loss": 0.3132, "step": 21152 }, { "epoch": 1.462357414448669, "grad_norm": 0.7311331497706288, "learning_rate": 1.7787781690712091e-06, "loss": 0.2511, "step": 21153 }, { "epoch": 1.462426546837193, "grad_norm": 0.6178975710455075, "learning_rate": 1.77835007437007e-06, "loss": 0.2277, "step": 21154 }, { "epoch": 1.4624956792257173, "grad_norm": 0.6912108240531932, "learning_rate": 1.7779220200465646e-06, "loss": 0.351, "step": 21155 }, { "epoch": 1.4625648116142413, "grad_norm": 0.6605940138838166, "learning_rate": 1.777494006106056e-06, "loss": 0.2567, "step": 21156 }, { "epoch": 1.4626339440027654, "grad_norm": 0.6152200792566918, "learning_rate": 1.777066032553908e-06, "loss": 0.2173, "step": 21157 }, { "epoch": 1.4627030763912894, "grad_norm": 0.7185848155120765, "learning_rate": 1.7766380993954863e-06, "loss": 0.3128, "step": 21158 }, { "epoch": 1.4627722087798134, "grad_norm": 0.6877986078131115, "learning_rate": 1.7762102066361558e-06, "loss": 0.3265, "step": 21159 }, { "epoch": 1.4628413411683374, "grad_norm": 0.693051635114518, "learning_rate": 1.7757823542812746e-06, "loss": 0.2555, "step": 21160 }, { "epoch": 1.4629104735568614, "grad_norm": 0.6878204737193169, "learning_rate": 1.77535454233621e-06, "loss": 0.1844, "step": 21161 }, { "epoch": 1.4629796059453855, "grad_norm": 0.6237063135942795, "learning_rate": 1.77492677080632e-06, "loss": 0.2901, "step": 21162 }, { "epoch": 1.4630487383339095, "grad_norm": 0.7822326200779144, "learning_rate": 1.7744990396969674e-06, "loss": 0.3352, "step": 21163 }, { "epoch": 1.4631178707224335, "grad_norm": 0.7723136733606767, "learning_rate": 1.7740713490135132e-06, "loss": 0.2755, "step": 21164 }, { "epoch": 1.4631870031109575, "grad_norm": 1.210960815790591, "learning_rate": 1.7736436987613193e-06, "loss": 0.2959, "step": 21165 }, { "epoch": 1.4632561354994815, "grad_norm": 0.9356967492676879, "learning_rate": 1.7732160889457423e-06, "loss": 0.3498, "step": 21166 }, { "epoch": 1.4633252678880055, "grad_norm": 0.6263630186509657, "learning_rate": 1.7727885195721445e-06, "loss": 0.2239, "step": 21167 }, { "epoch": 1.4633944002765296, "grad_norm": 0.6814567041418563, "learning_rate": 1.7723609906458822e-06, "loss": 0.2234, "step": 21168 }, { "epoch": 1.4634635326650536, "grad_norm": 0.7599741576325151, "learning_rate": 1.771933502172316e-06, "loss": 0.298, "step": 21169 }, { "epoch": 1.4635326650535776, "grad_norm": 0.7071105874005621, "learning_rate": 1.7715060541568013e-06, "loss": 0.3452, "step": 21170 }, { "epoch": 1.4636017974421016, "grad_norm": 0.6794975061639783, "learning_rate": 1.771078646604697e-06, "loss": 0.2449, "step": 21171 }, { "epoch": 1.4636709298306256, "grad_norm": 1.151613230920043, "learning_rate": 1.7706512795213594e-06, "loss": 0.2852, "step": 21172 }, { "epoch": 1.4637400622191497, "grad_norm": 0.6707792942326463, "learning_rate": 1.7702239529121468e-06, "loss": 0.3882, "step": 21173 }, { "epoch": 1.4638091946076737, "grad_norm": 0.6940447384067159, "learning_rate": 1.7697966667824113e-06, "loss": 0.2535, "step": 21174 }, { "epoch": 1.4638783269961977, "grad_norm": 0.6298217295381737, "learning_rate": 1.7693694211375112e-06, "loss": 0.2586, "step": 21175 }, { "epoch": 1.4639474593847217, "grad_norm": 0.6882884682083958, "learning_rate": 1.768942215982799e-06, "loss": 0.343, "step": 21176 }, { "epoch": 1.4640165917732457, "grad_norm": 0.7677104773213939, "learning_rate": 1.7685150513236299e-06, "loss": 0.3183, "step": 21177 }, { "epoch": 1.4640857241617697, "grad_norm": 0.6840684298115729, "learning_rate": 1.7680879271653578e-06, "loss": 0.2305, "step": 21178 }, { "epoch": 1.4641548565502938, "grad_norm": 1.197548309835825, "learning_rate": 1.7676608435133375e-06, "loss": 0.2695, "step": 21179 }, { "epoch": 1.4642239889388178, "grad_norm": 0.7546577229292373, "learning_rate": 1.767233800372919e-06, "loss": 0.3299, "step": 21180 }, { "epoch": 1.4642931213273418, "grad_norm": 0.6969272996669115, "learning_rate": 1.7668067977494568e-06, "loss": 0.2551, "step": 21181 }, { "epoch": 1.4643622537158658, "grad_norm": 0.6468084274752873, "learning_rate": 1.7663798356482997e-06, "loss": 0.2448, "step": 21182 }, { "epoch": 1.4644313861043898, "grad_norm": 0.7433883036548324, "learning_rate": 1.7659529140748027e-06, "loss": 0.2979, "step": 21183 }, { "epoch": 1.4645005184929138, "grad_norm": 0.7261968724059749, "learning_rate": 1.7655260330343126e-06, "loss": 0.3657, "step": 21184 }, { "epoch": 1.464569650881438, "grad_norm": 1.0244089670239926, "learning_rate": 1.765099192532182e-06, "loss": 0.2878, "step": 21185 }, { "epoch": 1.464638783269962, "grad_norm": 0.6931239016213933, "learning_rate": 1.7646723925737596e-06, "loss": 0.2875, "step": 21186 }, { "epoch": 1.4647079156584861, "grad_norm": 0.7236821683045811, "learning_rate": 1.764245633164397e-06, "loss": 0.3318, "step": 21187 }, { "epoch": 1.4647770480470101, "grad_norm": 0.7487574237239016, "learning_rate": 1.763818914309439e-06, "loss": 0.2792, "step": 21188 }, { "epoch": 1.4648461804355342, "grad_norm": 0.6593613379951927, "learning_rate": 1.7633922360142375e-06, "loss": 0.2399, "step": 21189 }, { "epoch": 1.4649153128240582, "grad_norm": 0.6649209087099559, "learning_rate": 1.762965598284137e-06, "loss": 0.2631, "step": 21190 }, { "epoch": 1.4649844452125822, "grad_norm": 0.6204043031545092, "learning_rate": 1.762539001124487e-06, "loss": 0.3289, "step": 21191 }, { "epoch": 1.4650535776011062, "grad_norm": 0.7104291556589075, "learning_rate": 1.7621124445406318e-06, "loss": 0.22, "step": 21192 }, { "epoch": 1.4651227099896302, "grad_norm": 0.6463851038564644, "learning_rate": 1.761685928537919e-06, "loss": 0.2393, "step": 21193 }, { "epoch": 1.4651918423781543, "grad_norm": 0.7268653309000738, "learning_rate": 1.7612594531216936e-06, "loss": 0.2682, "step": 21194 }, { "epoch": 1.4652609747666783, "grad_norm": 0.7128149483745951, "learning_rate": 1.7608330182973033e-06, "loss": 0.2979, "step": 21195 }, { "epoch": 1.4653301071552023, "grad_norm": 0.6786014994075114, "learning_rate": 1.7604066240700884e-06, "loss": 0.2369, "step": 21196 }, { "epoch": 1.4653992395437263, "grad_norm": 0.6918785759822782, "learning_rate": 1.7599802704453971e-06, "loss": 0.2631, "step": 21197 }, { "epoch": 1.4654683719322503, "grad_norm": 0.6442307590749005, "learning_rate": 1.7595539574285697e-06, "loss": 0.3095, "step": 21198 }, { "epoch": 1.4655375043207743, "grad_norm": 0.729301683594298, "learning_rate": 1.75912768502495e-06, "loss": 0.2954, "step": 21199 }, { "epoch": 1.4656066367092984, "grad_norm": 0.6201360078665679, "learning_rate": 1.7587014532398817e-06, "loss": 0.2613, "step": 21200 }, { "epoch": 1.4656757690978224, "grad_norm": 0.6675152185306535, "learning_rate": 1.7582752620787074e-06, "loss": 0.3094, "step": 21201 }, { "epoch": 1.4657449014863464, "grad_norm": 0.7362004973032633, "learning_rate": 1.7578491115467666e-06, "loss": 0.3117, "step": 21202 }, { "epoch": 1.4658140338748704, "grad_norm": 0.6797126689503129, "learning_rate": 1.7574230016494026e-06, "loss": 0.2224, "step": 21203 }, { "epoch": 1.4658831662633944, "grad_norm": 0.6808159208550425, "learning_rate": 1.7569969323919528e-06, "loss": 0.2085, "step": 21204 }, { "epoch": 1.4659522986519185, "grad_norm": 0.7141239673900278, "learning_rate": 1.7565709037797606e-06, "loss": 0.3465, "step": 21205 }, { "epoch": 1.4660214310404425, "grad_norm": 0.8351836015465081, "learning_rate": 1.756144915818163e-06, "loss": 0.3681, "step": 21206 }, { "epoch": 1.4660905634289665, "grad_norm": 0.7349488913107484, "learning_rate": 1.7557189685124992e-06, "loss": 0.2787, "step": 21207 }, { "epoch": 1.4661596958174905, "grad_norm": 0.6795836307076324, "learning_rate": 1.7552930618681091e-06, "loss": 0.2837, "step": 21208 }, { "epoch": 1.4662288282060145, "grad_norm": 0.6758751796487855, "learning_rate": 1.7548671958903312e-06, "loss": 0.3302, "step": 21209 }, { "epoch": 1.4662979605945385, "grad_norm": 0.6804744023411151, "learning_rate": 1.7544413705845004e-06, "loss": 0.241, "step": 21210 }, { "epoch": 1.4663670929830626, "grad_norm": 0.6406014107749118, "learning_rate": 1.7540155859559565e-06, "loss": 0.2236, "step": 21211 }, { "epoch": 1.4664362253715866, "grad_norm": 0.6321347619726121, "learning_rate": 1.7535898420100333e-06, "loss": 0.303, "step": 21212 }, { "epoch": 1.4665053577601106, "grad_norm": 0.7054580884807666, "learning_rate": 1.7531641387520676e-06, "loss": 0.2783, "step": 21213 }, { "epoch": 1.4665744901486346, "grad_norm": 0.6974374750145003, "learning_rate": 1.7527384761873957e-06, "loss": 0.2773, "step": 21214 }, { "epoch": 1.4666436225371586, "grad_norm": 0.6833101266504455, "learning_rate": 1.752312854321353e-06, "loss": 0.2712, "step": 21215 }, { "epoch": 1.4667127549256826, "grad_norm": 0.7280834426579027, "learning_rate": 1.7518872731592712e-06, "loss": 0.3185, "step": 21216 }, { "epoch": 1.4667818873142067, "grad_norm": 0.8534133768126033, "learning_rate": 1.7514617327064881e-06, "loss": 0.2145, "step": 21217 }, { "epoch": 1.4668510197027307, "grad_norm": 0.676512715919475, "learning_rate": 1.7510362329683344e-06, "loss": 0.2399, "step": 21218 }, { "epoch": 1.4669201520912547, "grad_norm": 0.8454843816963438, "learning_rate": 1.750610773950142e-06, "loss": 0.2815, "step": 21219 }, { "epoch": 1.4669892844797787, "grad_norm": 0.7304834060320176, "learning_rate": 1.7501853556572452e-06, "loss": 0.2856, "step": 21220 }, { "epoch": 1.4670584168683027, "grad_norm": 0.6572125930097872, "learning_rate": 1.7497599780949747e-06, "loss": 0.2105, "step": 21221 }, { "epoch": 1.4671275492568268, "grad_norm": 0.6725352383298145, "learning_rate": 1.7493346412686646e-06, "loss": 0.2575, "step": 21222 }, { "epoch": 1.4671966816453508, "grad_norm": 0.6639498583206638, "learning_rate": 1.748909345183642e-06, "loss": 0.3442, "step": 21223 }, { "epoch": 1.4672658140338748, "grad_norm": 0.7074334963061318, "learning_rate": 1.7484840898452405e-06, "loss": 0.2171, "step": 21224 }, { "epoch": 1.4673349464223988, "grad_norm": 0.6310130666488495, "learning_rate": 1.7480588752587868e-06, "loss": 0.2264, "step": 21225 }, { "epoch": 1.4674040788109228, "grad_norm": 0.7046360759628064, "learning_rate": 1.7476337014296136e-06, "loss": 0.3321, "step": 21226 }, { "epoch": 1.4674732111994468, "grad_norm": 0.6937146028554393, "learning_rate": 1.7472085683630457e-06, "loss": 0.2969, "step": 21227 }, { "epoch": 1.4675423435879709, "grad_norm": 0.6572028056478956, "learning_rate": 1.7467834760644136e-06, "loss": 0.2488, "step": 21228 }, { "epoch": 1.4676114759764949, "grad_norm": 0.7128876114394729, "learning_rate": 1.7463584245390453e-06, "loss": 0.2769, "step": 21229 }, { "epoch": 1.467680608365019, "grad_norm": 0.6898048461443225, "learning_rate": 1.7459334137922696e-06, "loss": 0.3524, "step": 21230 }, { "epoch": 1.467749740753543, "grad_norm": 0.769964217069777, "learning_rate": 1.7455084438294095e-06, "loss": 0.3135, "step": 21231 }, { "epoch": 1.467818873142067, "grad_norm": 0.620969102557468, "learning_rate": 1.7450835146557949e-06, "loss": 0.2431, "step": 21232 }, { "epoch": 1.467888005530591, "grad_norm": 0.6949926212914439, "learning_rate": 1.744658626276748e-06, "loss": 0.2701, "step": 21233 }, { "epoch": 1.467957137919115, "grad_norm": 0.7684338149045781, "learning_rate": 1.7442337786975955e-06, "loss": 0.3906, "step": 21234 }, { "epoch": 1.468026270307639, "grad_norm": 0.6470535645725128, "learning_rate": 1.743808971923663e-06, "loss": 0.246, "step": 21235 }, { "epoch": 1.4680954026961632, "grad_norm": 0.708617781284184, "learning_rate": 1.7433842059602752e-06, "loss": 0.2549, "step": 21236 }, { "epoch": 1.4681645350846872, "grad_norm": 0.6799082167243002, "learning_rate": 1.7429594808127532e-06, "loss": 0.3747, "step": 21237 }, { "epoch": 1.4682336674732113, "grad_norm": 0.6711162642878586, "learning_rate": 1.7425347964864236e-06, "loss": 0.2705, "step": 21238 }, { "epoch": 1.4683027998617353, "grad_norm": 0.627002633577259, "learning_rate": 1.7421101529866053e-06, "loss": 0.1902, "step": 21239 }, { "epoch": 1.4683719322502593, "grad_norm": 0.6829761562689636, "learning_rate": 1.7416855503186236e-06, "loss": 0.2309, "step": 21240 }, { "epoch": 1.4684410646387833, "grad_norm": 0.7309420836431222, "learning_rate": 1.741260988487798e-06, "loss": 0.3497, "step": 21241 }, { "epoch": 1.4685101970273073, "grad_norm": 0.7504082732656945, "learning_rate": 1.7408364674994493e-06, "loss": 0.2735, "step": 21242 }, { "epoch": 1.4685793294158314, "grad_norm": 0.6377354837299518, "learning_rate": 1.7404119873588998e-06, "loss": 0.2174, "step": 21243 }, { "epoch": 1.4686484618043554, "grad_norm": 0.721410189213514, "learning_rate": 1.7399875480714712e-06, "loss": 0.3377, "step": 21244 }, { "epoch": 1.4687175941928794, "grad_norm": 0.698841104943872, "learning_rate": 1.7395631496424786e-06, "loss": 0.2928, "step": 21245 }, { "epoch": 1.4687867265814034, "grad_norm": 0.7143635192873948, "learning_rate": 1.739138792077245e-06, "loss": 0.2598, "step": 21246 }, { "epoch": 1.4688558589699274, "grad_norm": 0.6748473602806913, "learning_rate": 1.7387144753810863e-06, "loss": 0.2371, "step": 21247 }, { "epoch": 1.4689249913584514, "grad_norm": 0.7189872278970324, "learning_rate": 1.738290199559321e-06, "loss": 0.3648, "step": 21248 }, { "epoch": 1.4689941237469755, "grad_norm": 0.6967040581981839, "learning_rate": 1.737865964617268e-06, "loss": 0.238, "step": 21249 }, { "epoch": 1.4690632561354995, "grad_norm": 0.6650794070331575, "learning_rate": 1.7374417705602447e-06, "loss": 0.2759, "step": 21250 }, { "epoch": 1.4691323885240235, "grad_norm": 0.710585862359399, "learning_rate": 1.7370176173935649e-06, "loss": 0.2324, "step": 21251 }, { "epoch": 1.4692015209125475, "grad_norm": 0.7240160590193192, "learning_rate": 1.736593505122548e-06, "loss": 0.2775, "step": 21252 }, { "epoch": 1.4692706533010715, "grad_norm": 0.6605938666329721, "learning_rate": 1.736169433752506e-06, "loss": 0.2466, "step": 21253 }, { "epoch": 1.4693397856895956, "grad_norm": 1.070919101391314, "learning_rate": 1.7357454032887571e-06, "loss": 0.2383, "step": 21254 }, { "epoch": 1.4694089180781196, "grad_norm": 0.6746399793341681, "learning_rate": 1.7353214137366125e-06, "loss": 0.3043, "step": 21255 }, { "epoch": 1.4694780504666436, "grad_norm": 0.6889759791063137, "learning_rate": 1.7348974651013883e-06, "loss": 0.291, "step": 21256 }, { "epoch": 1.4695471828551676, "grad_norm": 0.6230930404976096, "learning_rate": 1.7344735573883974e-06, "loss": 0.1989, "step": 21257 }, { "epoch": 1.4696163152436916, "grad_norm": 0.7365613077860913, "learning_rate": 1.7340496906029541e-06, "loss": 0.3188, "step": 21258 }, { "epoch": 1.4696854476322156, "grad_norm": 0.8549665301628493, "learning_rate": 1.7336258647503685e-06, "loss": 0.3616, "step": 21259 }, { "epoch": 1.4697545800207397, "grad_norm": 0.6619620578773533, "learning_rate": 1.733202079835955e-06, "loss": 0.2398, "step": 21260 }, { "epoch": 1.4698237124092637, "grad_norm": 0.6879923709133555, "learning_rate": 1.7327783358650218e-06, "loss": 0.2511, "step": 21261 }, { "epoch": 1.4698928447977877, "grad_norm": 0.6745934070330658, "learning_rate": 1.7323546328428836e-06, "loss": 0.3366, "step": 21262 }, { "epoch": 1.4699619771863117, "grad_norm": 0.7337695800004541, "learning_rate": 1.7319309707748467e-06, "loss": 0.3167, "step": 21263 }, { "epoch": 1.4700311095748357, "grad_norm": 0.6662650278684021, "learning_rate": 1.7315073496662233e-06, "loss": 0.2547, "step": 21264 }, { "epoch": 1.4701002419633598, "grad_norm": 0.7319641893058992, "learning_rate": 1.731083769522322e-06, "loss": 0.2744, "step": 21265 }, { "epoch": 1.470169374351884, "grad_norm": 0.7006716315507335, "learning_rate": 1.7306602303484538e-06, "loss": 0.3073, "step": 21266 }, { "epoch": 1.470238506740408, "grad_norm": 0.6545863574998115, "learning_rate": 1.7302367321499236e-06, "loss": 0.2623, "step": 21267 }, { "epoch": 1.470307639128932, "grad_norm": 0.6494394347589413, "learning_rate": 1.7298132749320423e-06, "loss": 0.2075, "step": 21268 }, { "epoch": 1.470376771517456, "grad_norm": 0.6767621979583347, "learning_rate": 1.729389858700114e-06, "loss": 0.3019, "step": 21269 }, { "epoch": 1.47044590390598, "grad_norm": 0.7737887850514966, "learning_rate": 1.7289664834594471e-06, "loss": 0.3105, "step": 21270 }, { "epoch": 1.470515036294504, "grad_norm": 0.6787453507973537, "learning_rate": 1.7285431492153482e-06, "loss": 0.3004, "step": 21271 }, { "epoch": 1.470584168683028, "grad_norm": 0.6782191053299398, "learning_rate": 1.7281198559731243e-06, "loss": 0.3057, "step": 21272 }, { "epoch": 1.4706533010715521, "grad_norm": 0.7241608129654762, "learning_rate": 1.7276966037380777e-06, "loss": 0.3483, "step": 21273 }, { "epoch": 1.4707224334600761, "grad_norm": 0.6843359188100155, "learning_rate": 1.7272733925155155e-06, "loss": 0.2738, "step": 21274 }, { "epoch": 1.4707915658486002, "grad_norm": 0.728578207079096, "learning_rate": 1.7268502223107397e-06, "loss": 0.222, "step": 21275 }, { "epoch": 1.4708606982371242, "grad_norm": 0.8164897758232831, "learning_rate": 1.7264270931290567e-06, "loss": 0.2693, "step": 21276 }, { "epoch": 1.4709298306256482, "grad_norm": 0.6426436302507862, "learning_rate": 1.7260040049757665e-06, "loss": 0.3029, "step": 21277 }, { "epoch": 1.4709989630141722, "grad_norm": 0.6706310381299218, "learning_rate": 1.7255809578561734e-06, "loss": 0.2262, "step": 21278 }, { "epoch": 1.4710680954026962, "grad_norm": 2.4250526850064, "learning_rate": 1.7251579517755796e-06, "loss": 0.2584, "step": 21279 }, { "epoch": 1.4711372277912202, "grad_norm": 1.0355193954607085, "learning_rate": 1.7247349867392883e-06, "loss": 0.3208, "step": 21280 }, { "epoch": 1.4712063601797443, "grad_norm": 0.7339226676089472, "learning_rate": 1.7243120627525971e-06, "loss": 0.2474, "step": 21281 }, { "epoch": 1.4712754925682683, "grad_norm": 0.6997768143216447, "learning_rate": 1.7238891798208102e-06, "loss": 0.2475, "step": 21282 }, { "epoch": 1.4713446249567923, "grad_norm": 0.7202344387605636, "learning_rate": 1.7234663379492244e-06, "loss": 0.3113, "step": 21283 }, { "epoch": 1.4714137573453163, "grad_norm": 0.686671255651116, "learning_rate": 1.7230435371431409e-06, "loss": 0.3065, "step": 21284 }, { "epoch": 1.4714828897338403, "grad_norm": 0.6694358147823232, "learning_rate": 1.7226207774078602e-06, "loss": 0.2069, "step": 21285 }, { "epoch": 1.4715520221223644, "grad_norm": 0.6561170227099873, "learning_rate": 1.7221980587486775e-06, "loss": 0.2101, "step": 21286 }, { "epoch": 1.4716211545108884, "grad_norm": 0.7963769585061171, "learning_rate": 1.7217753811708943e-06, "loss": 0.3505, "step": 21287 }, { "epoch": 1.4716902868994124, "grad_norm": 0.6844260839369389, "learning_rate": 1.721352744679805e-06, "loss": 0.2823, "step": 21288 }, { "epoch": 1.4717594192879364, "grad_norm": 0.6172021218947946, "learning_rate": 1.7209301492807096e-06, "loss": 0.2079, "step": 21289 }, { "epoch": 1.4718285516764604, "grad_norm": 0.8188735344697243, "learning_rate": 1.7205075949789013e-06, "loss": 0.3119, "step": 21290 }, { "epoch": 1.4718976840649844, "grad_norm": 0.6881869492964593, "learning_rate": 1.7200850817796778e-06, "loss": 0.3346, "step": 21291 }, { "epoch": 1.4719668164535085, "grad_norm": 0.6557963229647636, "learning_rate": 1.7196626096883346e-06, "loss": 0.2399, "step": 21292 }, { "epoch": 1.4720359488420325, "grad_norm": 0.6238677298920818, "learning_rate": 1.7192401787101681e-06, "loss": 0.2379, "step": 21293 }, { "epoch": 1.4721050812305565, "grad_norm": 0.777740187065516, "learning_rate": 1.7188177888504699e-06, "loss": 0.3016, "step": 21294 }, { "epoch": 1.4721742136190805, "grad_norm": 0.7246762026088658, "learning_rate": 1.7183954401145363e-06, "loss": 0.3219, "step": 21295 }, { "epoch": 1.4722433460076045, "grad_norm": 0.7308654225690684, "learning_rate": 1.7179731325076582e-06, "loss": 0.2225, "step": 21296 }, { "epoch": 1.4723124783961286, "grad_norm": 0.8400921182637565, "learning_rate": 1.7175508660351315e-06, "loss": 0.2147, "step": 21297 }, { "epoch": 1.4723816107846526, "grad_norm": 0.6338759008937548, "learning_rate": 1.7171286407022453e-06, "loss": 0.3432, "step": 21298 }, { "epoch": 1.4724507431731766, "grad_norm": 0.7341656486941321, "learning_rate": 1.7167064565142927e-06, "loss": 0.2619, "step": 21299 }, { "epoch": 1.4725198755617006, "grad_norm": 1.485781613247825, "learning_rate": 1.7162843134765654e-06, "loss": 0.1746, "step": 21300 }, { "epoch": 1.4725890079502246, "grad_norm": 0.6741446148548835, "learning_rate": 1.7158622115943562e-06, "loss": 0.2783, "step": 21301 }, { "epoch": 1.4726581403387486, "grad_norm": 0.6838184897514792, "learning_rate": 1.7154401508729518e-06, "loss": 0.3147, "step": 21302 }, { "epoch": 1.4727272727272727, "grad_norm": 0.6168340505792168, "learning_rate": 1.715018131317645e-06, "loss": 0.2383, "step": 21303 }, { "epoch": 1.4727964051157967, "grad_norm": 0.8068579624417204, "learning_rate": 1.7145961529337218e-06, "loss": 0.2396, "step": 21304 }, { "epoch": 1.4728655375043207, "grad_norm": 0.7217654657599338, "learning_rate": 1.7141742157264728e-06, "loss": 0.4283, "step": 21305 }, { "epoch": 1.4729346698928447, "grad_norm": 0.7462061047112829, "learning_rate": 1.7137523197011863e-06, "loss": 0.2345, "step": 21306 }, { "epoch": 1.4730038022813687, "grad_norm": 0.6710709931248726, "learning_rate": 1.7133304648631515e-06, "loss": 0.2473, "step": 21307 }, { "epoch": 1.4730729346698928, "grad_norm": 0.6357098519760438, "learning_rate": 1.7129086512176524e-06, "loss": 0.2777, "step": 21308 }, { "epoch": 1.4731420670584168, "grad_norm": 0.6985798076347686, "learning_rate": 1.712486878769979e-06, "loss": 0.3335, "step": 21309 }, { "epoch": 1.4732111994469408, "grad_norm": 0.6137244117126549, "learning_rate": 1.712065147525414e-06, "loss": 0.2659, "step": 21310 }, { "epoch": 1.4732803318354648, "grad_norm": 0.64340997290843, "learning_rate": 1.7116434574892465e-06, "loss": 0.2545, "step": 21311 }, { "epoch": 1.4733494642239888, "grad_norm": 0.6257596023555035, "learning_rate": 1.7112218086667582e-06, "loss": 0.2911, "step": 21312 }, { "epoch": 1.4734185966125128, "grad_norm": 0.7028751869303015, "learning_rate": 1.7108002010632356e-06, "loss": 0.2407, "step": 21313 }, { "epoch": 1.4734877290010369, "grad_norm": 0.6754566975441445, "learning_rate": 1.7103786346839623e-06, "loss": 0.2109, "step": 21314 }, { "epoch": 1.4735568613895609, "grad_norm": 0.7741527844335444, "learning_rate": 1.7099571095342244e-06, "loss": 0.3184, "step": 21315 }, { "epoch": 1.473625993778085, "grad_norm": 0.675884920590337, "learning_rate": 1.7095356256193008e-06, "loss": 0.3852, "step": 21316 }, { "epoch": 1.4736951261666091, "grad_norm": 0.7305689376108152, "learning_rate": 1.7091141829444774e-06, "loss": 0.2552, "step": 21317 }, { "epoch": 1.4737642585551332, "grad_norm": 0.6705086857231046, "learning_rate": 1.7086927815150333e-06, "loss": 0.2384, "step": 21318 }, { "epoch": 1.4738333909436572, "grad_norm": 0.6811411706731925, "learning_rate": 1.7082714213362522e-06, "loss": 0.2969, "step": 21319 }, { "epoch": 1.4739025233321812, "grad_norm": 0.6537975886990934, "learning_rate": 1.7078501024134142e-06, "loss": 0.3198, "step": 21320 }, { "epoch": 1.4739716557207052, "grad_norm": 0.6897841674262645, "learning_rate": 1.7074288247518011e-06, "loss": 0.2195, "step": 21321 }, { "epoch": 1.4740407881092292, "grad_norm": 0.7733523602705823, "learning_rate": 1.7070075883566912e-06, "loss": 0.2265, "step": 21322 }, { "epoch": 1.4741099204977532, "grad_norm": 0.6780850323021086, "learning_rate": 1.7065863932333654e-06, "loss": 0.3101, "step": 21323 }, { "epoch": 1.4741790528862773, "grad_norm": 0.6757141815864761, "learning_rate": 1.7061652393871003e-06, "loss": 0.2361, "step": 21324 }, { "epoch": 1.4742481852748013, "grad_norm": 0.6466733510182268, "learning_rate": 1.7057441268231778e-06, "loss": 0.2641, "step": 21325 }, { "epoch": 1.4743173176633253, "grad_norm": 0.70340167721968, "learning_rate": 1.7053230555468719e-06, "loss": 0.2521, "step": 21326 }, { "epoch": 1.4743864500518493, "grad_norm": 0.7198464944384577, "learning_rate": 1.7049020255634618e-06, "loss": 0.3874, "step": 21327 }, { "epoch": 1.4744555824403733, "grad_norm": 0.6632380194810505, "learning_rate": 1.7044810368782245e-06, "loss": 0.2501, "step": 21328 }, { "epoch": 1.4745247148288974, "grad_norm": 1.3035653012128743, "learning_rate": 1.7040600894964383e-06, "loss": 0.2274, "step": 21329 }, { "epoch": 1.4745938472174214, "grad_norm": 0.6693796315042354, "learning_rate": 1.703639183423375e-06, "loss": 0.3724, "step": 21330 }, { "epoch": 1.4746629796059454, "grad_norm": 0.7459697352665847, "learning_rate": 1.7032183186643136e-06, "loss": 0.2873, "step": 21331 }, { "epoch": 1.4747321119944694, "grad_norm": 0.6451524301339425, "learning_rate": 1.7027974952245253e-06, "loss": 0.2579, "step": 21332 }, { "epoch": 1.4748012443829934, "grad_norm": 0.6592044096213294, "learning_rate": 1.7023767131092883e-06, "loss": 0.2343, "step": 21333 }, { "epoch": 1.4748703767715174, "grad_norm": 0.7238734090399718, "learning_rate": 1.7019559723238727e-06, "loss": 0.3575, "step": 21334 }, { "epoch": 1.4749395091600415, "grad_norm": 0.7800691372816263, "learning_rate": 1.7015352728735535e-06, "loss": 0.236, "step": 21335 }, { "epoch": 1.4750086415485655, "grad_norm": 0.6428850090197427, "learning_rate": 1.7011146147636027e-06, "loss": 0.1883, "step": 21336 }, { "epoch": 1.4750777739370895, "grad_norm": 0.6754886899265319, "learning_rate": 1.7006939979992954e-06, "loss": 0.3729, "step": 21337 }, { "epoch": 1.4751469063256135, "grad_norm": 0.668544055642521, "learning_rate": 1.7002734225858986e-06, "loss": 0.2729, "step": 21338 }, { "epoch": 1.4752160387141375, "grad_norm": 0.696661951313769, "learning_rate": 1.6998528885286881e-06, "loss": 0.2446, "step": 21339 }, { "epoch": 1.4752851711026616, "grad_norm": 0.6657978116649892, "learning_rate": 1.6994323958329301e-06, "loss": 0.2516, "step": 21340 }, { "epoch": 1.4753543034911856, "grad_norm": 1.1708507662717231, "learning_rate": 1.6990119445038972e-06, "loss": 0.3221, "step": 21341 }, { "epoch": 1.4754234358797096, "grad_norm": 0.684876620750843, "learning_rate": 1.6985915345468584e-06, "loss": 0.2308, "step": 21342 }, { "epoch": 1.4754925682682336, "grad_norm": 0.7753235229171596, "learning_rate": 1.6981711659670853e-06, "loss": 0.2134, "step": 21343 }, { "epoch": 1.4755617006567576, "grad_norm": 0.703399868592247, "learning_rate": 1.6977508387698422e-06, "loss": 0.3081, "step": 21344 }, { "epoch": 1.4756308330452816, "grad_norm": 0.7671777742131459, "learning_rate": 1.6973305529604012e-06, "loss": 0.347, "step": 21345 }, { "epoch": 1.4756999654338057, "grad_norm": 0.7704036908385279, "learning_rate": 1.6969103085440259e-06, "loss": 0.2901, "step": 21346 }, { "epoch": 1.4757690978223297, "grad_norm": 0.7192994060162989, "learning_rate": 1.696490105525987e-06, "loss": 0.2312, "step": 21347 }, { "epoch": 1.475838230210854, "grad_norm": 0.7363173155713056, "learning_rate": 1.6960699439115475e-06, "loss": 0.3929, "step": 21348 }, { "epoch": 1.475907362599378, "grad_norm": 0.70581364119186, "learning_rate": 1.6956498237059748e-06, "loss": 0.2603, "step": 21349 }, { "epoch": 1.475976494987902, "grad_norm": 0.6254795495521954, "learning_rate": 1.695229744914535e-06, "loss": 0.2166, "step": 21350 }, { "epoch": 1.476045627376426, "grad_norm": 0.730517430403254, "learning_rate": 1.6948097075424945e-06, "loss": 0.356, "step": 21351 }, { "epoch": 1.47611475976495, "grad_norm": 0.7132120523870299, "learning_rate": 1.6943897115951157e-06, "loss": 0.3629, "step": 21352 }, { "epoch": 1.476183892153474, "grad_norm": 1.1457690256912811, "learning_rate": 1.6939697570776614e-06, "loss": 0.2216, "step": 21353 }, { "epoch": 1.476253024541998, "grad_norm": 1.7077232011589916, "learning_rate": 1.693549843995398e-06, "loss": 0.2882, "step": 21354 }, { "epoch": 1.476322156930522, "grad_norm": 0.7859980355983525, "learning_rate": 1.6931299723535831e-06, "loss": 0.3227, "step": 21355 }, { "epoch": 1.476391289319046, "grad_norm": 0.6595913491907857, "learning_rate": 1.692710142157486e-06, "loss": 0.2988, "step": 21356 }, { "epoch": 1.47646042170757, "grad_norm": 0.7005030683313739, "learning_rate": 1.6922903534123635e-06, "loss": 0.2783, "step": 21357 }, { "epoch": 1.476529554096094, "grad_norm": 0.7501037279262173, "learning_rate": 1.6918706061234802e-06, "loss": 0.2535, "step": 21358 }, { "epoch": 1.4765986864846181, "grad_norm": 0.6812808339463194, "learning_rate": 1.691450900296094e-06, "loss": 0.3941, "step": 21359 }, { "epoch": 1.4766678188731421, "grad_norm": 0.6377922819258692, "learning_rate": 1.6910312359354675e-06, "loss": 0.1952, "step": 21360 }, { "epoch": 1.4767369512616662, "grad_norm": 0.7272163823771699, "learning_rate": 1.6906116130468575e-06, "loss": 0.2931, "step": 21361 }, { "epoch": 1.4768060836501902, "grad_norm": 0.6875317689022821, "learning_rate": 1.6901920316355254e-06, "loss": 0.3555, "step": 21362 }, { "epoch": 1.4768752160387142, "grad_norm": 0.7611964819211944, "learning_rate": 1.6897724917067293e-06, "loss": 0.3078, "step": 21363 }, { "epoch": 1.4769443484272382, "grad_norm": 0.611306805072476, "learning_rate": 1.6893529932657298e-06, "loss": 0.1816, "step": 21364 }, { "epoch": 1.4770134808157622, "grad_norm": 0.6933421713338498, "learning_rate": 1.6889335363177806e-06, "loss": 0.2577, "step": 21365 }, { "epoch": 1.4770826132042862, "grad_norm": 0.746597756818253, "learning_rate": 1.688514120868142e-06, "loss": 0.3818, "step": 21366 }, { "epoch": 1.4771517455928103, "grad_norm": 0.659417466669737, "learning_rate": 1.6880947469220682e-06, "loss": 0.273, "step": 21367 }, { "epoch": 1.4772208779813343, "grad_norm": 0.6432411924433207, "learning_rate": 1.6876754144848172e-06, "loss": 0.311, "step": 21368 }, { "epoch": 1.4772900103698583, "grad_norm": 0.6467751525403179, "learning_rate": 1.687256123561643e-06, "loss": 0.2937, "step": 21369 }, { "epoch": 1.4773591427583823, "grad_norm": 0.6786791419396431, "learning_rate": 1.6868368741578006e-06, "loss": 0.289, "step": 21370 }, { "epoch": 1.4774282751469063, "grad_norm": 0.7157541866304451, "learning_rate": 1.686417666278546e-06, "loss": 0.2343, "step": 21371 }, { "epoch": 1.4774974075354304, "grad_norm": 0.6899395269781344, "learning_rate": 1.685998499929134e-06, "loss": 0.2307, "step": 21372 }, { "epoch": 1.4775665399239544, "grad_norm": 1.6923208245437378, "learning_rate": 1.6855793751148154e-06, "loss": 0.2895, "step": 21373 }, { "epoch": 1.4776356723124784, "grad_norm": 0.733553240594739, "learning_rate": 1.6851602918408454e-06, "loss": 0.2151, "step": 21374 }, { "epoch": 1.4777048047010024, "grad_norm": 1.3256972280556236, "learning_rate": 1.6847412501124748e-06, "loss": 0.2355, "step": 21375 }, { "epoch": 1.4777739370895264, "grad_norm": 0.6984514257345235, "learning_rate": 1.684322249934956e-06, "loss": 0.2637, "step": 21376 }, { "epoch": 1.4778430694780504, "grad_norm": 0.6864992585155003, "learning_rate": 1.6839032913135401e-06, "loss": 0.3567, "step": 21377 }, { "epoch": 1.4779122018665745, "grad_norm": 0.7647986094886782, "learning_rate": 1.6834843742534807e-06, "loss": 0.2591, "step": 21378 }, { "epoch": 1.4779813342550985, "grad_norm": 0.5818968341640575, "learning_rate": 1.6830654987600247e-06, "loss": 0.2253, "step": 21379 }, { "epoch": 1.4780504666436225, "grad_norm": 0.7503116958070845, "learning_rate": 1.6826466648384248e-06, "loss": 0.2949, "step": 21380 }, { "epoch": 1.4781195990321465, "grad_norm": 0.7411753045433821, "learning_rate": 1.6822278724939268e-06, "loss": 0.2782, "step": 21381 }, { "epoch": 1.4781887314206705, "grad_norm": 0.8574869626802102, "learning_rate": 1.6818091217317838e-06, "loss": 0.2596, "step": 21382 }, { "epoch": 1.4782578638091945, "grad_norm": 0.6896049662030531, "learning_rate": 1.68139041255724e-06, "loss": 0.2827, "step": 21383 }, { "epoch": 1.4783269961977186, "grad_norm": 0.8216615072389389, "learning_rate": 1.6809717449755447e-06, "loss": 0.3929, "step": 21384 }, { "epoch": 1.4783961285862426, "grad_norm": 0.6472058534341474, "learning_rate": 1.6805531189919461e-06, "loss": 0.2262, "step": 21385 }, { "epoch": 1.4784652609747666, "grad_norm": 0.7144951380360026, "learning_rate": 1.6801345346116915e-06, "loss": 0.2529, "step": 21386 }, { "epoch": 1.4785343933632906, "grad_norm": 0.7249205348042317, "learning_rate": 1.6797159918400247e-06, "loss": 0.3355, "step": 21387 }, { "epoch": 1.4786035257518146, "grad_norm": 0.6707658356277514, "learning_rate": 1.6792974906821942e-06, "loss": 0.3093, "step": 21388 }, { "epoch": 1.4786726581403387, "grad_norm": 0.6702918542278491, "learning_rate": 1.6788790311434416e-06, "loss": 0.2031, "step": 21389 }, { "epoch": 1.4787417905288627, "grad_norm": 0.6977512272054904, "learning_rate": 1.678460613229016e-06, "loss": 0.3001, "step": 21390 }, { "epoch": 1.4788109229173867, "grad_norm": 0.6490609646754303, "learning_rate": 1.6780422369441552e-06, "loss": 0.3175, "step": 21391 }, { "epoch": 1.4788800553059107, "grad_norm": 0.6896186165124513, "learning_rate": 1.6776239022941099e-06, "loss": 0.2496, "step": 21392 }, { "epoch": 1.4789491876944347, "grad_norm": 0.6531734017128447, "learning_rate": 1.6772056092841183e-06, "loss": 0.2296, "step": 21393 }, { "epoch": 1.4790183200829587, "grad_norm": 0.6354843574243326, "learning_rate": 1.676787357919426e-06, "loss": 0.2893, "step": 21394 }, { "epoch": 1.4790874524714828, "grad_norm": 0.7118336170988927, "learning_rate": 1.6763691482052723e-06, "loss": 0.2607, "step": 21395 }, { "epoch": 1.4791565848600068, "grad_norm": 0.6414691636878571, "learning_rate": 1.675950980146902e-06, "loss": 0.2726, "step": 21396 }, { "epoch": 1.4792257172485308, "grad_norm": 0.6845298271517212, "learning_rate": 1.6755328537495519e-06, "loss": 0.2278, "step": 21397 }, { "epoch": 1.479294849637055, "grad_norm": 0.6927548362736023, "learning_rate": 1.6751147690184649e-06, "loss": 0.3653, "step": 21398 }, { "epoch": 1.479363982025579, "grad_norm": 0.6687071030881195, "learning_rate": 1.6746967259588808e-06, "loss": 0.2081, "step": 21399 }, { "epoch": 1.479433114414103, "grad_norm": 0.6692940344596705, "learning_rate": 1.6742787245760405e-06, "loss": 0.2479, "step": 21400 }, { "epoch": 1.479502246802627, "grad_norm": 0.6780461494260934, "learning_rate": 1.6738607648751804e-06, "loss": 0.2975, "step": 21401 }, { "epoch": 1.4795713791911511, "grad_norm": 0.626939186184497, "learning_rate": 1.6734428468615404e-06, "loss": 0.2809, "step": 21402 }, { "epoch": 1.4796405115796751, "grad_norm": 0.6468256426780006, "learning_rate": 1.6730249705403567e-06, "loss": 0.2288, "step": 21403 }, { "epoch": 1.4797096439681992, "grad_norm": 0.7455080005111754, "learning_rate": 1.6726071359168694e-06, "loss": 0.2231, "step": 21404 }, { "epoch": 1.4797787763567232, "grad_norm": 0.7652967354284598, "learning_rate": 1.672189342996312e-06, "loss": 0.4135, "step": 21405 }, { "epoch": 1.4798479087452472, "grad_norm": 0.7832451277601926, "learning_rate": 1.6717715917839228e-06, "loss": 0.3352, "step": 21406 }, { "epoch": 1.4799170411337712, "grad_norm": 0.6442779484061418, "learning_rate": 1.6713538822849368e-06, "loss": 0.2396, "step": 21407 }, { "epoch": 1.4799861735222952, "grad_norm": 0.7088476401528785, "learning_rate": 1.670936214504591e-06, "loss": 0.2507, "step": 21408 }, { "epoch": 1.4800553059108192, "grad_norm": 0.7619305666773322, "learning_rate": 1.670518588448118e-06, "loss": 0.3163, "step": 21409 }, { "epoch": 1.4801244382993433, "grad_norm": 0.6779875693324577, "learning_rate": 1.6701010041207538e-06, "loss": 0.2398, "step": 21410 }, { "epoch": 1.4801935706878673, "grad_norm": 0.6301560532944427, "learning_rate": 1.6696834615277292e-06, "loss": 0.2105, "step": 21411 }, { "epoch": 1.4802627030763913, "grad_norm": 0.7359333952909928, "learning_rate": 1.6692659606742795e-06, "loss": 0.3713, "step": 21412 }, { "epoch": 1.4803318354649153, "grad_norm": 1.4592938295066178, "learning_rate": 1.668848501565637e-06, "loss": 0.2717, "step": 21413 }, { "epoch": 1.4804009678534393, "grad_norm": 0.6063623445571298, "learning_rate": 1.6684310842070357e-06, "loss": 0.2435, "step": 21414 }, { "epoch": 1.4804701002419633, "grad_norm": 0.6766060584459208, "learning_rate": 1.668013708603703e-06, "loss": 0.306, "step": 21415 }, { "epoch": 1.4805392326304874, "grad_norm": 0.7453253127850542, "learning_rate": 1.6675963747608748e-06, "loss": 0.3521, "step": 21416 }, { "epoch": 1.4806083650190114, "grad_norm": 0.8040908194132884, "learning_rate": 1.6671790826837786e-06, "loss": 0.2521, "step": 21417 }, { "epoch": 1.4806774974075354, "grad_norm": 0.6346047162200165, "learning_rate": 1.666761832377643e-06, "loss": 0.2729, "step": 21418 }, { "epoch": 1.4807466297960594, "grad_norm": 0.7005101259225938, "learning_rate": 1.6663446238477e-06, "loss": 0.3552, "step": 21419 }, { "epoch": 1.4808157621845834, "grad_norm": 0.7720714795247572, "learning_rate": 1.6659274570991774e-06, "loss": 0.3395, "step": 21420 }, { "epoch": 1.4808848945731075, "grad_norm": 0.7096496422628485, "learning_rate": 1.6655103321373061e-06, "loss": 0.3042, "step": 21421 }, { "epoch": 1.4809540269616315, "grad_norm": 0.7998168680568234, "learning_rate": 1.6650932489673106e-06, "loss": 0.2696, "step": 21422 }, { "epoch": 1.4810231593501555, "grad_norm": 0.6980472065173959, "learning_rate": 1.6646762075944212e-06, "loss": 0.3594, "step": 21423 }, { "epoch": 1.4810922917386795, "grad_norm": 0.6223656993205781, "learning_rate": 1.6642592080238617e-06, "loss": 0.2027, "step": 21424 }, { "epoch": 1.4811614241272035, "grad_norm": 0.6733709224062457, "learning_rate": 1.663842250260862e-06, "loss": 0.2355, "step": 21425 }, { "epoch": 1.4812305565157275, "grad_norm": 0.7403524080246391, "learning_rate": 1.6634253343106426e-06, "loss": 0.2742, "step": 21426 }, { "epoch": 1.4812996889042516, "grad_norm": 0.7309821902040007, "learning_rate": 1.6630084601784357e-06, "loss": 0.3646, "step": 21427 }, { "epoch": 1.4813688212927756, "grad_norm": 0.6409668675864071, "learning_rate": 1.6625916278694615e-06, "loss": 0.2106, "step": 21428 }, { "epoch": 1.4814379536812998, "grad_norm": 0.7360622374084714, "learning_rate": 1.6621748373889463e-06, "loss": 0.2943, "step": 21429 }, { "epoch": 1.4815070860698238, "grad_norm": 0.6502744230136416, "learning_rate": 1.6617580887421113e-06, "loss": 0.3449, "step": 21430 }, { "epoch": 1.4815762184583479, "grad_norm": 0.7262031562860032, "learning_rate": 1.6613413819341833e-06, "loss": 0.3512, "step": 21431 }, { "epoch": 1.4816453508468719, "grad_norm": 0.7140406442484263, "learning_rate": 1.6609247169703808e-06, "loss": 0.2281, "step": 21432 }, { "epoch": 1.481714483235396, "grad_norm": 0.6289937675653638, "learning_rate": 1.6605080938559288e-06, "loss": 0.1995, "step": 21433 }, { "epoch": 1.48178361562392, "grad_norm": 0.7369144351820148, "learning_rate": 1.6600915125960477e-06, "loss": 0.4086, "step": 21434 }, { "epoch": 1.481852748012444, "grad_norm": 0.6926531614770599, "learning_rate": 1.659674973195961e-06, "loss": 0.2357, "step": 21435 }, { "epoch": 1.481921880400968, "grad_norm": 0.6919116019532858, "learning_rate": 1.6592584756608854e-06, "loss": 0.1881, "step": 21436 }, { "epoch": 1.481991012789492, "grad_norm": 0.6220817724031377, "learning_rate": 1.6588420199960453e-06, "loss": 0.3425, "step": 21437 }, { "epoch": 1.482060145178016, "grad_norm": 0.8418032482026263, "learning_rate": 1.6584256062066555e-06, "loss": 0.3308, "step": 21438 }, { "epoch": 1.48212927756654, "grad_norm": 0.6279304811682409, "learning_rate": 1.65800923429794e-06, "loss": 0.2106, "step": 21439 }, { "epoch": 1.482198409955064, "grad_norm": 0.739627327349985, "learning_rate": 1.6575929042751126e-06, "loss": 0.3312, "step": 21440 }, { "epoch": 1.482267542343588, "grad_norm": 1.1097059455557465, "learning_rate": 1.6571766161433932e-06, "loss": 0.348, "step": 21441 }, { "epoch": 1.482336674732112, "grad_norm": 0.6762357026800897, "learning_rate": 1.6567603699079993e-06, "loss": 0.2251, "step": 21442 }, { "epoch": 1.482405807120636, "grad_norm": 0.603830552308974, "learning_rate": 1.6563441655741496e-06, "loss": 0.2631, "step": 21443 }, { "epoch": 1.48247493950916, "grad_norm": 0.7127076445808833, "learning_rate": 1.6559280031470576e-06, "loss": 0.3244, "step": 21444 }, { "epoch": 1.4825440718976841, "grad_norm": 0.624049043263474, "learning_rate": 1.6555118826319416e-06, "loss": 0.2548, "step": 21445 }, { "epoch": 1.4826132042862081, "grad_norm": 0.794413791816555, "learning_rate": 1.6550958040340137e-06, "loss": 0.2155, "step": 21446 }, { "epoch": 1.4826823366747321, "grad_norm": 0.6907814582556568, "learning_rate": 1.6546797673584914e-06, "loss": 0.3018, "step": 21447 }, { "epoch": 1.4827514690632562, "grad_norm": 0.6863765269776307, "learning_rate": 1.6542637726105876e-06, "loss": 0.3584, "step": 21448 }, { "epoch": 1.4828206014517802, "grad_norm": 0.7606149456897529, "learning_rate": 1.6538478197955182e-06, "loss": 0.2665, "step": 21449 }, { "epoch": 1.4828897338403042, "grad_norm": 0.6520209099402996, "learning_rate": 1.6534319089184936e-06, "loss": 0.2158, "step": 21450 }, { "epoch": 1.4829588662288282, "grad_norm": 0.7246134788268527, "learning_rate": 1.6530160399847295e-06, "loss": 0.2795, "step": 21451 }, { "epoch": 1.4830279986173522, "grad_norm": 0.6845431572166243, "learning_rate": 1.6526002129994351e-06, "loss": 0.2928, "step": 21452 }, { "epoch": 1.4830971310058763, "grad_norm": 0.7490063854445141, "learning_rate": 1.652184427967824e-06, "loss": 0.2396, "step": 21453 }, { "epoch": 1.4831662633944003, "grad_norm": 0.6998728602398605, "learning_rate": 1.6517686848951058e-06, "loss": 0.2183, "step": 21454 }, { "epoch": 1.4832353957829243, "grad_norm": 0.6432374902968484, "learning_rate": 1.6513529837864917e-06, "loss": 0.3351, "step": 21455 }, { "epoch": 1.4833045281714483, "grad_norm": 0.6617218843270715, "learning_rate": 1.650937324647192e-06, "loss": 0.2606, "step": 21456 }, { "epoch": 1.4833736605599723, "grad_norm": 0.6385453781695821, "learning_rate": 1.6505217074824176e-06, "loss": 0.2347, "step": 21457 }, { "epoch": 1.4834427929484963, "grad_norm": 0.6934377521860848, "learning_rate": 1.650106132297375e-06, "loss": 0.2861, "step": 21458 }, { "epoch": 1.4835119253370204, "grad_norm": 0.9054437345174655, "learning_rate": 1.6496905990972755e-06, "loss": 0.3643, "step": 21459 }, { "epoch": 1.4835810577255444, "grad_norm": 0.671164479924674, "learning_rate": 1.6492751078873236e-06, "loss": 0.2314, "step": 21460 }, { "epoch": 1.4836501901140684, "grad_norm": 1.0555909767619258, "learning_rate": 1.64885965867273e-06, "loss": 0.2639, "step": 21461 }, { "epoch": 1.4837193225025924, "grad_norm": 0.7017340766969273, "learning_rate": 1.6484442514586974e-06, "loss": 0.3476, "step": 21462 }, { "epoch": 1.4837884548911164, "grad_norm": 0.6771730851969493, "learning_rate": 1.6480288862504379e-06, "loss": 0.3449, "step": 21463 }, { "epoch": 1.4838575872796405, "grad_norm": 0.7023742557430268, "learning_rate": 1.647613563053153e-06, "loss": 0.1732, "step": 21464 }, { "epoch": 1.4839267196681645, "grad_norm": 0.8478267262448049, "learning_rate": 1.647198281872051e-06, "loss": 0.2755, "step": 21465 }, { "epoch": 1.4839958520566885, "grad_norm": 0.6410126440176547, "learning_rate": 1.646783042712334e-06, "loss": 0.3464, "step": 21466 }, { "epoch": 1.4840649844452125, "grad_norm": 0.6167021777809493, "learning_rate": 1.6463678455792087e-06, "loss": 0.2096, "step": 21467 }, { "epoch": 1.4841341168337365, "grad_norm": 0.6200217447445039, "learning_rate": 1.645952690477876e-06, "loss": 0.2347, "step": 21468 }, { "epoch": 1.4842032492222605, "grad_norm": 0.6597155583437474, "learning_rate": 1.6455375774135408e-06, "loss": 0.2742, "step": 21469 }, { "epoch": 1.4842723816107846, "grad_norm": 0.671077733947077, "learning_rate": 1.6451225063914061e-06, "loss": 0.314, "step": 21470 }, { "epoch": 1.4843415139993086, "grad_norm": 0.6903093151122165, "learning_rate": 1.644707477416675e-06, "loss": 0.2364, "step": 21471 }, { "epoch": 1.4844106463878326, "grad_norm": 0.6954585119313225, "learning_rate": 1.6442924904945463e-06, "loss": 0.2217, "step": 21472 }, { "epoch": 1.4844797787763566, "grad_norm": 0.7058860191469594, "learning_rate": 1.643877545630224e-06, "loss": 0.386, "step": 21473 }, { "epoch": 1.4845489111648806, "grad_norm": 0.7296521631996077, "learning_rate": 1.6434626428289063e-06, "loss": 0.2716, "step": 21474 }, { "epoch": 1.4846180435534047, "grad_norm": 0.6725744978111254, "learning_rate": 1.6430477820957957e-06, "loss": 0.2382, "step": 21475 }, { "epoch": 1.4846871759419287, "grad_norm": 0.7704453208177573, "learning_rate": 1.6426329634360888e-06, "loss": 0.3162, "step": 21476 }, { "epoch": 1.4847563083304527, "grad_norm": 0.7032054784691616, "learning_rate": 1.6422181868549864e-06, "loss": 0.3361, "step": 21477 }, { "epoch": 1.4848254407189767, "grad_norm": 0.6964952111144359, "learning_rate": 1.6418034523576865e-06, "loss": 0.302, "step": 21478 }, { "epoch": 1.4848945731075007, "grad_norm": 0.6729543261493816, "learning_rate": 1.6413887599493893e-06, "loss": 0.2288, "step": 21479 }, { "epoch": 1.484963705496025, "grad_norm": 0.7365425912272694, "learning_rate": 1.6409741096352887e-06, "loss": 0.4271, "step": 21480 }, { "epoch": 1.485032837884549, "grad_norm": 0.6474326711852755, "learning_rate": 1.6405595014205844e-06, "loss": 0.3166, "step": 21481 }, { "epoch": 1.485101970273073, "grad_norm": 0.7030971537630445, "learning_rate": 1.6401449353104704e-06, "loss": 0.3012, "step": 21482 }, { "epoch": 1.485171102661597, "grad_norm": 0.6373797315458357, "learning_rate": 1.639730411310143e-06, "loss": 0.2728, "step": 21483 }, { "epoch": 1.485240235050121, "grad_norm": 0.8816834259982457, "learning_rate": 1.6393159294248006e-06, "loss": 0.3533, "step": 21484 }, { "epoch": 1.485309367438645, "grad_norm": 0.7179204030074917, "learning_rate": 1.638901489659634e-06, "loss": 0.2238, "step": 21485 }, { "epoch": 1.485378499827169, "grad_norm": 0.6512763498721877, "learning_rate": 1.6384870920198403e-06, "loss": 0.2342, "step": 21486 }, { "epoch": 1.485447632215693, "grad_norm": 0.7265451390766949, "learning_rate": 1.6380727365106103e-06, "loss": 0.2472, "step": 21487 }, { "epoch": 1.485516764604217, "grad_norm": 0.7011925001982107, "learning_rate": 1.6376584231371406e-06, "loss": 0.3102, "step": 21488 }, { "epoch": 1.4855858969927411, "grad_norm": 0.6378342919638266, "learning_rate": 1.6372441519046205e-06, "loss": 0.2208, "step": 21489 }, { "epoch": 1.4856550293812651, "grad_norm": 0.7256835996661393, "learning_rate": 1.6368299228182445e-06, "loss": 0.2776, "step": 21490 }, { "epoch": 1.4857241617697892, "grad_norm": 0.6564969842170382, "learning_rate": 1.6364157358832029e-06, "loss": 0.3189, "step": 21491 }, { "epoch": 1.4857932941583132, "grad_norm": 0.7217570861679109, "learning_rate": 1.6360015911046895e-06, "loss": 0.2601, "step": 21492 }, { "epoch": 1.4858624265468372, "grad_norm": 0.6436070080728791, "learning_rate": 1.6355874884878908e-06, "loss": 0.2233, "step": 21493 }, { "epoch": 1.4859315589353612, "grad_norm": 0.7625236451053381, "learning_rate": 1.635173428038001e-06, "loss": 0.3598, "step": 21494 }, { "epoch": 1.4860006913238852, "grad_norm": 0.6994992961423612, "learning_rate": 1.6347594097602053e-06, "loss": 0.3225, "step": 21495 }, { "epoch": 1.4860698237124093, "grad_norm": 0.6498980600763382, "learning_rate": 1.634345433659697e-06, "loss": 0.2796, "step": 21496 }, { "epoch": 1.4861389561009333, "grad_norm": 0.763571371602789, "learning_rate": 1.6339314997416588e-06, "loss": 0.202, "step": 21497 }, { "epoch": 1.4862080884894573, "grad_norm": 0.7258601727847146, "learning_rate": 1.6335176080112857e-06, "loss": 0.3851, "step": 21498 }, { "epoch": 1.4862772208779813, "grad_norm": 0.652015182817096, "learning_rate": 1.6331037584737597e-06, "loss": 0.261, "step": 21499 }, { "epoch": 1.4863463532665053, "grad_norm": 0.7248315819719994, "learning_rate": 1.6326899511342714e-06, "loss": 0.2013, "step": 21500 }, { "epoch": 1.4864154856550293, "grad_norm": 0.7363377688843298, "learning_rate": 1.632276185998004e-06, "loss": 0.2887, "step": 21501 }, { "epoch": 1.4864846180435534, "grad_norm": 0.6861171342296691, "learning_rate": 1.6318624630701456e-06, "loss": 0.3004, "step": 21502 }, { "epoch": 1.4865537504320774, "grad_norm": 0.7443801657464203, "learning_rate": 1.6314487823558795e-06, "loss": 0.2187, "step": 21503 }, { "epoch": 1.4866228828206014, "grad_norm": 0.6952746978661687, "learning_rate": 1.6310351438603911e-06, "loss": 0.3116, "step": 21504 }, { "epoch": 1.4866920152091254, "grad_norm": 0.6697505611431134, "learning_rate": 1.6306215475888654e-06, "loss": 0.328, "step": 21505 }, { "epoch": 1.4867611475976494, "grad_norm": 0.6720120323728396, "learning_rate": 1.630207993546487e-06, "loss": 0.2462, "step": 21506 }, { "epoch": 1.4868302799861735, "grad_norm": 0.6339231537893114, "learning_rate": 1.629794481738436e-06, "loss": 0.2498, "step": 21507 }, { "epoch": 1.4868994123746975, "grad_norm": 0.775153404810505, "learning_rate": 1.6293810121698983e-06, "loss": 0.2991, "step": 21508 }, { "epoch": 1.4869685447632215, "grad_norm": 0.6691504201877384, "learning_rate": 1.6289675848460535e-06, "loss": 0.338, "step": 21509 }, { "epoch": 1.4870376771517457, "grad_norm": 0.8637949312511032, "learning_rate": 1.6285541997720849e-06, "loss": 0.2463, "step": 21510 }, { "epoch": 1.4871068095402697, "grad_norm": 0.7180123029753017, "learning_rate": 1.6281408569531716e-06, "loss": 0.2764, "step": 21511 }, { "epoch": 1.4871759419287938, "grad_norm": 0.7086067426053323, "learning_rate": 1.6277275563944955e-06, "loss": 0.32, "step": 21512 }, { "epoch": 1.4872450743173178, "grad_norm": 0.7028158414912216, "learning_rate": 1.6273142981012364e-06, "loss": 0.2874, "step": 21513 }, { "epoch": 1.4873142067058418, "grad_norm": 0.7228293166460574, "learning_rate": 1.6269010820785747e-06, "loss": 0.2308, "step": 21514 }, { "epoch": 1.4873833390943658, "grad_norm": 0.7087451960715914, "learning_rate": 1.6264879083316875e-06, "loss": 0.2665, "step": 21515 }, { "epoch": 1.4874524714828898, "grad_norm": 0.7149349153056002, "learning_rate": 1.6260747768657553e-06, "loss": 0.3607, "step": 21516 }, { "epoch": 1.4875216038714139, "grad_norm": 0.8449236505547242, "learning_rate": 1.6256616876859533e-06, "loss": 0.2057, "step": 21517 }, { "epoch": 1.4875907362599379, "grad_norm": 1.0696294959531731, "learning_rate": 1.6252486407974599e-06, "loss": 0.2516, "step": 21518 }, { "epoch": 1.487659868648462, "grad_norm": 0.7287126190412468, "learning_rate": 1.624835636205453e-06, "loss": 0.3188, "step": 21519 }, { "epoch": 1.487729001036986, "grad_norm": 0.7146468441427324, "learning_rate": 1.6244226739151092e-06, "loss": 0.3089, "step": 21520 }, { "epoch": 1.48779813342551, "grad_norm": 0.6729743189886147, "learning_rate": 1.6240097539316013e-06, "loss": 0.2743, "step": 21521 }, { "epoch": 1.487867265814034, "grad_norm": 1.5771691387606395, "learning_rate": 1.6235968762601085e-06, "loss": 0.2893, "step": 21522 }, { "epoch": 1.487936398202558, "grad_norm": 0.7011265594651462, "learning_rate": 1.6231840409058019e-06, "loss": 0.3277, "step": 21523 }, { "epoch": 1.488005530591082, "grad_norm": 0.6517022529708054, "learning_rate": 1.6227712478738584e-06, "loss": 0.2564, "step": 21524 }, { "epoch": 1.488074662979606, "grad_norm": 0.6212559616404776, "learning_rate": 1.622358497169449e-06, "loss": 0.2719, "step": 21525 }, { "epoch": 1.48814379536813, "grad_norm": 0.6578963921726809, "learning_rate": 1.621945788797748e-06, "loss": 0.2426, "step": 21526 }, { "epoch": 1.488212927756654, "grad_norm": 0.736787060045783, "learning_rate": 1.6215331227639286e-06, "loss": 0.3629, "step": 21527 }, { "epoch": 1.488282060145178, "grad_norm": 0.6295024143853131, "learning_rate": 1.6211204990731633e-06, "loss": 0.2399, "step": 21528 }, { "epoch": 1.488351192533702, "grad_norm": 0.813295515942535, "learning_rate": 1.6207079177306212e-06, "loss": 0.2756, "step": 21529 }, { "epoch": 1.488420324922226, "grad_norm": 0.6773081882433623, "learning_rate": 1.6202953787414765e-06, "loss": 0.3524, "step": 21530 }, { "epoch": 1.48848945731075, "grad_norm": 0.7180579639238993, "learning_rate": 1.6198828821108964e-06, "loss": 0.2855, "step": 21531 }, { "epoch": 1.4885585896992741, "grad_norm": 0.7294500536209229, "learning_rate": 1.6194704278440538e-06, "loss": 0.2459, "step": 21532 }, { "epoch": 1.4886277220877981, "grad_norm": 0.8190259464676103, "learning_rate": 1.6190580159461133e-06, "loss": 0.2786, "step": 21533 }, { "epoch": 1.4886968544763222, "grad_norm": 0.7143613125003843, "learning_rate": 1.61864564642225e-06, "loss": 0.4033, "step": 21534 }, { "epoch": 1.4887659868648462, "grad_norm": 0.6562455256191901, "learning_rate": 1.6182333192776278e-06, "loss": 0.2587, "step": 21535 }, { "epoch": 1.4888351192533702, "grad_norm": 0.6070748660853371, "learning_rate": 1.6178210345174178e-06, "loss": 0.2184, "step": 21536 }, { "epoch": 1.4889042516418942, "grad_norm": 0.6894281081585236, "learning_rate": 1.6174087921467835e-06, "loss": 0.3306, "step": 21537 }, { "epoch": 1.4889733840304182, "grad_norm": 0.6556701768236747, "learning_rate": 1.6169965921708953e-06, "loss": 0.2816, "step": 21538 }, { "epoch": 1.4890425164189423, "grad_norm": 0.6834955869353118, "learning_rate": 1.6165844345949155e-06, "loss": 0.2364, "step": 21539 }, { "epoch": 1.4891116488074663, "grad_norm": 0.8052270287919784, "learning_rate": 1.616172319424012e-06, "loss": 0.2453, "step": 21540 }, { "epoch": 1.4891807811959903, "grad_norm": 0.72850543357287, "learning_rate": 1.6157602466633499e-06, "loss": 0.3399, "step": 21541 }, { "epoch": 1.4892499135845143, "grad_norm": 0.577553768561654, "learning_rate": 1.615348216318095e-06, "loss": 0.1963, "step": 21542 }, { "epoch": 1.4893190459730383, "grad_norm": 0.6645603098184326, "learning_rate": 1.614936228393409e-06, "loss": 0.2314, "step": 21543 }, { "epoch": 1.4893881783615623, "grad_norm": 0.7230538516008352, "learning_rate": 1.6145242828944574e-06, "loss": 0.3381, "step": 21544 }, { "epoch": 1.4894573107500864, "grad_norm": 0.6595274677442601, "learning_rate": 1.6141123798264014e-06, "loss": 0.2725, "step": 21545 }, { "epoch": 1.4895264431386104, "grad_norm": 0.6546815810123419, "learning_rate": 1.6137005191944055e-06, "loss": 0.2343, "step": 21546 }, { "epoch": 1.4895955755271344, "grad_norm": 0.6790421561459741, "learning_rate": 1.6132887010036285e-06, "loss": 0.2456, "step": 21547 }, { "epoch": 1.4896647079156584, "grad_norm": 0.630163604837799, "learning_rate": 1.612876925259234e-06, "loss": 0.3317, "step": 21548 }, { "epoch": 1.4897338403041824, "grad_norm": 0.7103910677088981, "learning_rate": 1.6124651919663831e-06, "loss": 0.2941, "step": 21549 }, { "epoch": 1.4898029726927065, "grad_norm": 0.6541153142403909, "learning_rate": 1.6120535011302368e-06, "loss": 0.2136, "step": 21550 }, { "epoch": 1.4898721050812305, "grad_norm": 0.7124290141572626, "learning_rate": 1.6116418527559542e-06, "loss": 0.3136, "step": 21551 }, { "epoch": 1.4899412374697545, "grad_norm": 0.6544836647069512, "learning_rate": 1.6112302468486922e-06, "loss": 0.3018, "step": 21552 }, { "epoch": 1.4900103698582785, "grad_norm": 0.6856851271593298, "learning_rate": 1.610818683413612e-06, "loss": 0.2684, "step": 21553 }, { "epoch": 1.4900795022468025, "grad_norm": 0.6579042498613393, "learning_rate": 1.610407162455871e-06, "loss": 0.242, "step": 21554 }, { "epoch": 1.4901486346353265, "grad_norm": 0.7323888477347373, "learning_rate": 1.6099956839806286e-06, "loss": 0.373, "step": 21555 }, { "epoch": 1.4902177670238506, "grad_norm": 0.7373225824540234, "learning_rate": 1.6095842479930396e-06, "loss": 0.2719, "step": 21556 }, { "epoch": 1.4902868994123746, "grad_norm": 0.7209668050441201, "learning_rate": 1.6091728544982627e-06, "loss": 0.2044, "step": 21557 }, { "epoch": 1.4903560318008986, "grad_norm": 0.735338806192552, "learning_rate": 1.6087615035014515e-06, "loss": 0.2574, "step": 21558 }, { "epoch": 1.4904251641894226, "grad_norm": 0.7138538269420659, "learning_rate": 1.6083501950077641e-06, "loss": 0.3813, "step": 21559 }, { "epoch": 1.4904942965779466, "grad_norm": 0.6912350119962617, "learning_rate": 1.607938929022353e-06, "loss": 0.2577, "step": 21560 }, { "epoch": 1.4905634289664709, "grad_norm": 0.6825981085721243, "learning_rate": 1.6075277055503741e-06, "loss": 0.2777, "step": 21561 }, { "epoch": 1.490632561354995, "grad_norm": 0.7087136952251971, "learning_rate": 1.607116524596981e-06, "loss": 0.3006, "step": 21562 }, { "epoch": 1.490701693743519, "grad_norm": 0.712369332396043, "learning_rate": 1.6067053861673293e-06, "loss": 0.3103, "step": 21563 }, { "epoch": 1.490770826132043, "grad_norm": 0.6849675754158766, "learning_rate": 1.6062942902665685e-06, "loss": 0.2312, "step": 21564 }, { "epoch": 1.490839958520567, "grad_norm": 0.6795359411854239, "learning_rate": 1.6058832368998534e-06, "loss": 0.2468, "step": 21565 }, { "epoch": 1.490909090909091, "grad_norm": 0.7084269611878211, "learning_rate": 1.6054722260723338e-06, "loss": 0.3363, "step": 21566 }, { "epoch": 1.490978223297615, "grad_norm": 0.7597603353995619, "learning_rate": 1.605061257789164e-06, "loss": 0.309, "step": 21567 }, { "epoch": 1.491047355686139, "grad_norm": 0.6550433127431323, "learning_rate": 1.6046503320554884e-06, "loss": 0.2034, "step": 21568 }, { "epoch": 1.491116488074663, "grad_norm": 0.7357250255796552, "learning_rate": 1.6042394488764656e-06, "loss": 0.316, "step": 21569 }, { "epoch": 1.491185620463187, "grad_norm": 0.6639332678946893, "learning_rate": 1.6038286082572397e-06, "loss": 0.3145, "step": 21570 }, { "epoch": 1.491254752851711, "grad_norm": 0.6797542057518047, "learning_rate": 1.6034178102029628e-06, "loss": 0.2529, "step": 21571 }, { "epoch": 1.491323885240235, "grad_norm": 0.6504370532691128, "learning_rate": 1.6030070547187808e-06, "loss": 0.2089, "step": 21572 }, { "epoch": 1.491393017628759, "grad_norm": 0.7132225095172435, "learning_rate": 1.6025963418098451e-06, "loss": 0.3171, "step": 21573 }, { "epoch": 1.491462150017283, "grad_norm": 0.7305214336800631, "learning_rate": 1.6021856714812988e-06, "loss": 0.3028, "step": 21574 }, { "epoch": 1.4915312824058071, "grad_norm": 0.7927942149238932, "learning_rate": 1.6017750437382922e-06, "loss": 0.2835, "step": 21575 }, { "epoch": 1.4916004147943311, "grad_norm": 0.7764208724962277, "learning_rate": 1.601364458585971e-06, "loss": 0.3413, "step": 21576 }, { "epoch": 1.4916695471828552, "grad_norm": 0.7366154400100478, "learning_rate": 1.6009539160294824e-06, "loss": 0.4024, "step": 21577 }, { "epoch": 1.4917386795713792, "grad_norm": 0.7055860136491924, "learning_rate": 1.6005434160739692e-06, "loss": 0.2678, "step": 21578 }, { "epoch": 1.4918078119599032, "grad_norm": 1.1360715595026334, "learning_rate": 1.600132958724579e-06, "loss": 0.2867, "step": 21579 }, { "epoch": 1.4918769443484272, "grad_norm": 0.7193363821664563, "learning_rate": 1.599722543986454e-06, "loss": 0.3312, "step": 21580 }, { "epoch": 1.4919460767369512, "grad_norm": 0.6694657668893064, "learning_rate": 1.5993121718647398e-06, "loss": 0.2877, "step": 21581 }, { "epoch": 1.4920152091254753, "grad_norm": 0.6539724543712679, "learning_rate": 1.598901842364577e-06, "loss": 0.2142, "step": 21582 }, { "epoch": 1.4920843415139993, "grad_norm": 0.7140193991072062, "learning_rate": 1.5984915554911101e-06, "loss": 0.2318, "step": 21583 }, { "epoch": 1.4921534739025233, "grad_norm": 0.6772343076565847, "learning_rate": 1.5980813112494813e-06, "loss": 0.3593, "step": 21584 }, { "epoch": 1.4922226062910473, "grad_norm": 0.9137905470594915, "learning_rate": 1.5976711096448344e-06, "loss": 0.2031, "step": 21585 }, { "epoch": 1.4922917386795713, "grad_norm": 0.7434191454023971, "learning_rate": 1.5972609506823062e-06, "loss": 0.2246, "step": 21586 }, { "epoch": 1.4923608710680953, "grad_norm": 0.6946351569325361, "learning_rate": 1.5968508343670413e-06, "loss": 0.3191, "step": 21587 }, { "epoch": 1.4924300034566194, "grad_norm": 0.7730430609518782, "learning_rate": 1.5964407607041765e-06, "loss": 0.2741, "step": 21588 }, { "epoch": 1.4924991358451434, "grad_norm": 0.7027281800175028, "learning_rate": 1.596030729698853e-06, "loss": 0.2988, "step": 21589 }, { "epoch": 1.4925682682336674, "grad_norm": 0.6798004386431721, "learning_rate": 1.5956207413562097e-06, "loss": 0.229, "step": 21590 }, { "epoch": 1.4926374006221916, "grad_norm": 0.8140673950043488, "learning_rate": 1.5952107956813862e-06, "loss": 0.3877, "step": 21591 }, { "epoch": 1.4927065330107157, "grad_norm": 0.7207670442662351, "learning_rate": 1.5948008926795178e-06, "loss": 0.1995, "step": 21592 }, { "epoch": 1.4927756653992397, "grad_norm": 0.6233949757697863, "learning_rate": 1.594391032355745e-06, "loss": 0.188, "step": 21593 }, { "epoch": 1.4928447977877637, "grad_norm": 0.6668983632176126, "learning_rate": 1.5939812147152017e-06, "loss": 0.2406, "step": 21594 }, { "epoch": 1.4929139301762877, "grad_norm": 4.05450479168399, "learning_rate": 1.5935714397630269e-06, "loss": 0.279, "step": 21595 }, { "epoch": 1.4929830625648117, "grad_norm": 0.6800628935666501, "learning_rate": 1.5931617075043532e-06, "loss": 0.2481, "step": 21596 }, { "epoch": 1.4930521949533357, "grad_norm": 0.6743156877710561, "learning_rate": 1.5927520179443178e-06, "loss": 0.2494, "step": 21597 }, { "epoch": 1.4931213273418598, "grad_norm": 0.578420136561084, "learning_rate": 1.592342371088056e-06, "loss": 0.2797, "step": 21598 }, { "epoch": 1.4931904597303838, "grad_norm": 0.6232260804527591, "learning_rate": 1.5919327669407019e-06, "loss": 0.2768, "step": 21599 }, { "epoch": 1.4932595921189078, "grad_norm": 0.7128923278893597, "learning_rate": 1.5915232055073877e-06, "loss": 0.2121, "step": 21600 }, { "epoch": 1.4933287245074318, "grad_norm": 0.6696861695670747, "learning_rate": 1.591113686793249e-06, "loss": 0.3078, "step": 21601 }, { "epoch": 1.4933978568959558, "grad_norm": 0.7446281059634574, "learning_rate": 1.5907042108034149e-06, "loss": 0.3236, "step": 21602 }, { "epoch": 1.4934669892844799, "grad_norm": 0.6777485232898929, "learning_rate": 1.590294777543021e-06, "loss": 0.2055, "step": 21603 }, { "epoch": 1.4935361216730039, "grad_norm": 0.6881479955779096, "learning_rate": 1.5898853870171944e-06, "loss": 0.3056, "step": 21604 }, { "epoch": 1.4936052540615279, "grad_norm": 0.7064876457589476, "learning_rate": 1.5894760392310714e-06, "loss": 0.3061, "step": 21605 }, { "epoch": 1.493674386450052, "grad_norm": 0.6344600193460973, "learning_rate": 1.5890667341897792e-06, "loss": 0.2484, "step": 21606 }, { "epoch": 1.493743518838576, "grad_norm": 0.662281614830341, "learning_rate": 1.5886574718984492e-06, "loss": 0.1953, "step": 21607 }, { "epoch": 1.4938126512271, "grad_norm": 0.7199360849821544, "learning_rate": 1.5882482523622089e-06, "loss": 0.2646, "step": 21608 }, { "epoch": 1.493881783615624, "grad_norm": 0.6637480682596828, "learning_rate": 1.5878390755861894e-06, "loss": 0.3571, "step": 21609 }, { "epoch": 1.493950916004148, "grad_norm": 0.6467161108610326, "learning_rate": 1.5874299415755162e-06, "loss": 0.2376, "step": 21610 }, { "epoch": 1.494020048392672, "grad_norm": 0.8294519104432576, "learning_rate": 1.587020850335319e-06, "loss": 0.2324, "step": 21611 }, { "epoch": 1.494089180781196, "grad_norm": 0.7110852538832921, "learning_rate": 1.5866118018707244e-06, "loss": 0.3593, "step": 21612 }, { "epoch": 1.49415831316972, "grad_norm": 0.7229575631147683, "learning_rate": 1.5862027961868615e-06, "loss": 0.2585, "step": 21613 }, { "epoch": 1.494227445558244, "grad_norm": 0.7062134070350178, "learning_rate": 1.5857938332888523e-06, "loss": 0.2438, "step": 21614 }, { "epoch": 1.494296577946768, "grad_norm": 0.695977221074949, "learning_rate": 1.5853849131818261e-06, "loss": 0.2244, "step": 21615 }, { "epoch": 1.494365710335292, "grad_norm": 0.6416291955314226, "learning_rate": 1.584976035870905e-06, "loss": 0.3143, "step": 21616 }, { "epoch": 1.494434842723816, "grad_norm": 0.7127143502560782, "learning_rate": 1.5845672013612163e-06, "loss": 0.2521, "step": 21617 }, { "epoch": 1.4945039751123401, "grad_norm": 0.6095561796813616, "learning_rate": 1.584158409657881e-06, "loss": 0.2338, "step": 21618 }, { "epoch": 1.4945731075008641, "grad_norm": 0.7361332504092697, "learning_rate": 1.5837496607660241e-06, "loss": 0.2872, "step": 21619 }, { "epoch": 1.4946422398893882, "grad_norm": 0.7130434330290601, "learning_rate": 1.5833409546907703e-06, "loss": 0.3483, "step": 21620 }, { "epoch": 1.4947113722779122, "grad_norm": 0.6200238252476419, "learning_rate": 1.5829322914372387e-06, "loss": 0.2159, "step": 21621 }, { "epoch": 1.4947805046664362, "grad_norm": 0.7083168324911051, "learning_rate": 1.5825236710105545e-06, "loss": 0.2787, "step": 21622 }, { "epoch": 1.4948496370549602, "grad_norm": 0.6850523999264985, "learning_rate": 1.5821150934158359e-06, "loss": 0.3547, "step": 21623 }, { "epoch": 1.4949187694434842, "grad_norm": 0.6866733246518785, "learning_rate": 1.5817065586582048e-06, "loss": 0.2936, "step": 21624 }, { "epoch": 1.4949879018320082, "grad_norm": 0.6512808292797052, "learning_rate": 1.5812980667427818e-06, "loss": 0.2505, "step": 21625 }, { "epoch": 1.4950570342205323, "grad_norm": 0.6889860546962271, "learning_rate": 1.5808896176746885e-06, "loss": 0.2382, "step": 21626 }, { "epoch": 1.4951261666090563, "grad_norm": 0.714122722798543, "learning_rate": 1.58048121145904e-06, "loss": 0.39, "step": 21627 }, { "epoch": 1.4951952989975803, "grad_norm": 0.6219116403381942, "learning_rate": 1.5800728481009592e-06, "loss": 0.238, "step": 21628 }, { "epoch": 1.4952644313861043, "grad_norm": 0.6485140016677997, "learning_rate": 1.5796645276055606e-06, "loss": 0.278, "step": 21629 }, { "epoch": 1.4953335637746283, "grad_norm": 0.7919575318503153, "learning_rate": 1.5792562499779646e-06, "loss": 0.3459, "step": 21630 }, { "epoch": 1.4954026961631524, "grad_norm": 0.6878699306320756, "learning_rate": 1.5788480152232855e-06, "loss": 0.3221, "step": 21631 }, { "epoch": 1.4954718285516764, "grad_norm": 0.646892935227554, "learning_rate": 1.5784398233466408e-06, "loss": 0.1863, "step": 21632 }, { "epoch": 1.4955409609402004, "grad_norm": 0.7703478994532119, "learning_rate": 1.578031674353147e-06, "loss": 0.3036, "step": 21633 }, { "epoch": 1.4956100933287244, "grad_norm": 0.7504105324375131, "learning_rate": 1.5776235682479213e-06, "loss": 0.2931, "step": 21634 }, { "epoch": 1.4956792257172484, "grad_norm": 0.6707421006132397, "learning_rate": 1.5772155050360749e-06, "loss": 0.2168, "step": 21635 }, { "epoch": 1.4957483581057724, "grad_norm": 0.6410355052994859, "learning_rate": 1.5768074847227255e-06, "loss": 0.1898, "step": 21636 }, { "epoch": 1.4958174904942965, "grad_norm": 0.7950188898738322, "learning_rate": 1.576399507312984e-06, "loss": 0.3515, "step": 21637 }, { "epoch": 1.4958866228828205, "grad_norm": 0.6864073847607177, "learning_rate": 1.5759915728119662e-06, "loss": 0.2744, "step": 21638 }, { "epoch": 1.4959557552713445, "grad_norm": 0.7453303859708728, "learning_rate": 1.5755836812247805e-06, "loss": 0.2236, "step": 21639 }, { "epoch": 1.4960248876598685, "grad_norm": 0.6752399610406384, "learning_rate": 1.5751758325565453e-06, "loss": 0.2296, "step": 21640 }, { "epoch": 1.4960940200483925, "grad_norm": 0.7351948962138289, "learning_rate": 1.5747680268123677e-06, "loss": 0.3501, "step": 21641 }, { "epoch": 1.4961631524369168, "grad_norm": 0.6626716436370628, "learning_rate": 1.574360263997362e-06, "loss": 0.2339, "step": 21642 }, { "epoch": 1.4962322848254408, "grad_norm": 0.9778405000041065, "learning_rate": 1.5739525441166353e-06, "loss": 0.1894, "step": 21643 }, { "epoch": 1.4963014172139648, "grad_norm": 0.6807852121525882, "learning_rate": 1.5735448671753012e-06, "loss": 0.2721, "step": 21644 }, { "epoch": 1.4963705496024888, "grad_norm": 0.7675315030721982, "learning_rate": 1.5731372331784655e-06, "loss": 0.3238, "step": 21645 }, { "epoch": 1.4964396819910128, "grad_norm": 0.6646100715644517, "learning_rate": 1.5727296421312394e-06, "loss": 0.2293, "step": 21646 }, { "epoch": 1.4965088143795369, "grad_norm": 0.7597445043212575, "learning_rate": 1.5723220940387306e-06, "loss": 0.2115, "step": 21647 }, { "epoch": 1.4965779467680609, "grad_norm": 0.702225945587316, "learning_rate": 1.5719145889060495e-06, "loss": 0.3214, "step": 21648 }, { "epoch": 1.496647079156585, "grad_norm": 0.6547614652580914, "learning_rate": 1.5715071267382997e-06, "loss": 0.2076, "step": 21649 }, { "epoch": 1.496716211545109, "grad_norm": 0.6753619881354648, "learning_rate": 1.571099707540591e-06, "loss": 0.2716, "step": 21650 }, { "epoch": 1.496785343933633, "grad_norm": 0.7217155592780267, "learning_rate": 1.5706923313180266e-06, "loss": 0.2974, "step": 21651 }, { "epoch": 1.496854476322157, "grad_norm": 0.7320635009107069, "learning_rate": 1.5702849980757157e-06, "loss": 0.3237, "step": 21652 }, { "epoch": 1.496923608710681, "grad_norm": 0.6718266175557005, "learning_rate": 1.56987770781876e-06, "loss": 0.2271, "step": 21653 }, { "epoch": 1.496992741099205, "grad_norm": 0.6273149067549548, "learning_rate": 1.5694704605522659e-06, "loss": 0.1996, "step": 21654 }, { "epoch": 1.497061873487729, "grad_norm": 0.7287700072476669, "learning_rate": 1.569063256281338e-06, "loss": 0.3846, "step": 21655 }, { "epoch": 1.497131005876253, "grad_norm": 0.7570164088872211, "learning_rate": 1.5686560950110802e-06, "loss": 0.2687, "step": 21656 }, { "epoch": 1.497200138264777, "grad_norm": 0.675376552002287, "learning_rate": 1.5682489767465935e-06, "loss": 0.2745, "step": 21657 }, { "epoch": 1.497269270653301, "grad_norm": 0.6656755122215764, "learning_rate": 1.5678419014929834e-06, "loss": 0.251, "step": 21658 }, { "epoch": 1.497338403041825, "grad_norm": 0.6870260059358556, "learning_rate": 1.5674348692553487e-06, "loss": 0.3347, "step": 21659 }, { "epoch": 1.497407535430349, "grad_norm": 0.6674289772702817, "learning_rate": 1.5670278800387917e-06, "loss": 0.2276, "step": 21660 }, { "epoch": 1.4974766678188731, "grad_norm": 0.7132679504463099, "learning_rate": 1.5666209338484144e-06, "loss": 0.2703, "step": 21661 }, { "epoch": 1.4975458002073971, "grad_norm": 0.691133807306703, "learning_rate": 1.5662140306893175e-06, "loss": 0.3373, "step": 21662 }, { "epoch": 1.4976149325959212, "grad_norm": 0.9329293492138881, "learning_rate": 1.5658071705665988e-06, "loss": 0.2726, "step": 21663 }, { "epoch": 1.4976840649844452, "grad_norm": 0.6893732356837131, "learning_rate": 1.56540035348536e-06, "loss": 0.2442, "step": 21664 }, { "epoch": 1.4977531973729692, "grad_norm": 0.652569622472628, "learning_rate": 1.564993579450697e-06, "loss": 0.2513, "step": 21665 }, { "epoch": 1.4978223297614932, "grad_norm": 0.8799573353526078, "learning_rate": 1.5645868484677108e-06, "loss": 0.3597, "step": 21666 }, { "epoch": 1.4978914621500172, "grad_norm": 0.8441066936916309, "learning_rate": 1.5641801605414963e-06, "loss": 0.2394, "step": 21667 }, { "epoch": 1.4979605945385412, "grad_norm": 0.634702821976981, "learning_rate": 1.5637735156771517e-06, "loss": 0.2235, "step": 21668 }, { "epoch": 1.4980297269270653, "grad_norm": 0.6867536229743463, "learning_rate": 1.563366913879774e-06, "loss": 0.3287, "step": 21669 }, { "epoch": 1.4980988593155893, "grad_norm": 0.6909425493886533, "learning_rate": 1.5629603551544609e-06, "loss": 0.323, "step": 21670 }, { "epoch": 1.4981679917041133, "grad_norm": 0.6138930735504924, "learning_rate": 1.5625538395063044e-06, "loss": 0.2166, "step": 21671 }, { "epoch": 1.4982371240926373, "grad_norm": 0.6412716992219867, "learning_rate": 1.5621473669404026e-06, "loss": 0.2214, "step": 21672 }, { "epoch": 1.4983062564811616, "grad_norm": 0.7475499264624503, "learning_rate": 1.561740937461847e-06, "loss": 0.3277, "step": 21673 }, { "epoch": 1.4983753888696856, "grad_norm": 0.8353266331686815, "learning_rate": 1.561334551075734e-06, "loss": 0.2259, "step": 21674 }, { "epoch": 1.4984445212582096, "grad_norm": 0.7698859600647403, "learning_rate": 1.560928207787153e-06, "loss": 0.2645, "step": 21675 }, { "epoch": 1.4985136536467336, "grad_norm": 0.667934366013684, "learning_rate": 1.560521907601203e-06, "loss": 0.2768, "step": 21676 }, { "epoch": 1.4985827860352576, "grad_norm": 0.6629005647864494, "learning_rate": 1.560115650522971e-06, "loss": 0.3047, "step": 21677 }, { "epoch": 1.4986519184237816, "grad_norm": 0.6846060909981382, "learning_rate": 1.5597094365575522e-06, "loss": 0.2129, "step": 21678 }, { "epoch": 1.4987210508123057, "grad_norm": 0.7852643443992349, "learning_rate": 1.5593032657100344e-06, "loss": 0.2496, "step": 21679 }, { "epoch": 1.4987901832008297, "grad_norm": 0.7215990125556868, "learning_rate": 1.5588971379855117e-06, "loss": 0.4083, "step": 21680 }, { "epoch": 1.4988593155893537, "grad_norm": 0.7312093135173137, "learning_rate": 1.5584910533890713e-06, "loss": 0.284, "step": 21681 }, { "epoch": 1.4989284479778777, "grad_norm": 0.6449159822825091, "learning_rate": 1.5580850119258033e-06, "loss": 0.2547, "step": 21682 }, { "epoch": 1.4989975803664017, "grad_norm": 0.6747178046717862, "learning_rate": 1.5576790136007991e-06, "loss": 0.255, "step": 21683 }, { "epoch": 1.4990667127549258, "grad_norm": 0.7077295205418118, "learning_rate": 1.5572730584191437e-06, "loss": 0.3584, "step": 21684 }, { "epoch": 1.4991358451434498, "grad_norm": 0.699951698456211, "learning_rate": 1.556867146385928e-06, "loss": 0.2738, "step": 21685 }, { "epoch": 1.4992049775319738, "grad_norm": 0.7669791074223011, "learning_rate": 1.5564612775062372e-06, "loss": 0.2535, "step": 21686 }, { "epoch": 1.4992741099204978, "grad_norm": 0.711809007243265, "learning_rate": 1.55605545178516e-06, "loss": 0.3255, "step": 21687 }, { "epoch": 1.4993432423090218, "grad_norm": 0.6961021825968983, "learning_rate": 1.5556496692277806e-06, "loss": 0.2712, "step": 21688 }, { "epoch": 1.4994123746975458, "grad_norm": 0.6782519898024177, "learning_rate": 1.5552439298391854e-06, "loss": 0.2484, "step": 21689 }, { "epoch": 1.4994815070860699, "grad_norm": 0.6656652444952458, "learning_rate": 1.5548382336244604e-06, "loss": 0.2616, "step": 21690 }, { "epoch": 1.4995506394745939, "grad_norm": 0.6455124788797638, "learning_rate": 1.5544325805886918e-06, "loss": 0.335, "step": 21691 }, { "epoch": 1.499619771863118, "grad_norm": 0.7516200481191844, "learning_rate": 1.5540269707369598e-06, "loss": 0.2664, "step": 21692 }, { "epoch": 1.499688904251642, "grad_norm": 0.6678770455577198, "learning_rate": 1.5536214040743519e-06, "loss": 0.2431, "step": 21693 }, { "epoch": 1.499758036640166, "grad_norm": 0.7409866544116382, "learning_rate": 1.5532158806059477e-06, "loss": 0.3168, "step": 21694 }, { "epoch": 1.49982716902869, "grad_norm": 0.7140253355697433, "learning_rate": 1.5528104003368316e-06, "loss": 0.2965, "step": 21695 }, { "epoch": 1.499896301417214, "grad_norm": 0.6216336712736307, "learning_rate": 1.552404963272085e-06, "loss": 0.1911, "step": 21696 }, { "epoch": 1.499965433805738, "grad_norm": 0.6652683688997819, "learning_rate": 1.5519995694167916e-06, "loss": 0.2462, "step": 21697 }, { "epoch": 1.500034566194262, "grad_norm": 0.7867391292286652, "learning_rate": 1.5515942187760286e-06, "loss": 0.314, "step": 21698 }, { "epoch": 1.500103698582786, "grad_norm": 0.7342142302432675, "learning_rate": 1.5511889113548795e-06, "loss": 0.2228, "step": 21699 }, { "epoch": 1.50017283097131, "grad_norm": 1.259795102859746, "learning_rate": 1.5507836471584215e-06, "loss": 0.2333, "step": 21700 }, { "epoch": 1.500241963359834, "grad_norm": 0.7240696146177897, "learning_rate": 1.5503784261917366e-06, "loss": 0.3072, "step": 21701 }, { "epoch": 1.500311095748358, "grad_norm": 1.0670299921375537, "learning_rate": 1.5499732484599007e-06, "loss": 0.3077, "step": 21702 }, { "epoch": 1.500380228136882, "grad_norm": 0.6310419552779777, "learning_rate": 1.5495681139679924e-06, "loss": 0.207, "step": 21703 }, { "epoch": 1.5004493605254061, "grad_norm": 0.7027463448705211, "learning_rate": 1.5491630227210908e-06, "loss": 0.2083, "step": 21704 }, { "epoch": 1.5005184929139301, "grad_norm": 0.6528952240796516, "learning_rate": 1.5487579747242743e-06, "loss": 0.2943, "step": 21705 }, { "epoch": 1.5005876253024542, "grad_norm": 0.7884065899258936, "learning_rate": 1.5483529699826154e-06, "loss": 0.2726, "step": 21706 }, { "epoch": 1.5006567576909782, "grad_norm": 0.673021446251974, "learning_rate": 1.547948008501194e-06, "loss": 0.2477, "step": 21707 }, { "epoch": 1.5007258900795022, "grad_norm": 0.6737129411240567, "learning_rate": 1.5475430902850825e-06, "loss": 0.2446, "step": 21708 }, { "epoch": 1.5007950224680262, "grad_norm": 0.7604661147531812, "learning_rate": 1.5471382153393589e-06, "loss": 0.4056, "step": 21709 }, { "epoch": 1.5008641548565502, "grad_norm": 0.6549032559089638, "learning_rate": 1.5467333836690923e-06, "loss": 0.2321, "step": 21710 }, { "epoch": 1.5009332872450742, "grad_norm": 0.5699719256502132, "learning_rate": 1.5463285952793633e-06, "loss": 0.2067, "step": 21711 }, { "epoch": 1.5010024196335983, "grad_norm": 0.6942143913188465, "learning_rate": 1.5459238501752405e-06, "loss": 0.3446, "step": 21712 }, { "epoch": 1.5010715520221223, "grad_norm": 0.7538491281344666, "learning_rate": 1.5455191483617998e-06, "loss": 0.2618, "step": 21713 }, { "epoch": 1.5011406844106463, "grad_norm": 0.6655671662039292, "learning_rate": 1.54511448984411e-06, "loss": 0.2327, "step": 21714 }, { "epoch": 1.5012098167991703, "grad_norm": 0.697205252157004, "learning_rate": 1.5447098746272466e-06, "loss": 0.2696, "step": 21715 }, { "epoch": 1.5012789491876943, "grad_norm": 0.6433937774738335, "learning_rate": 1.5443053027162764e-06, "loss": 0.3632, "step": 21716 }, { "epoch": 1.5013480815762184, "grad_norm": 0.6828933868384519, "learning_rate": 1.5439007741162726e-06, "loss": 0.2355, "step": 21717 }, { "epoch": 1.5014172139647424, "grad_norm": 0.6123579310693688, "learning_rate": 1.5434962888323052e-06, "loss": 0.2142, "step": 21718 }, { "epoch": 1.5014863463532664, "grad_norm": 0.6993043273928028, "learning_rate": 1.5430918468694445e-06, "loss": 0.344, "step": 21719 }, { "epoch": 1.5015554787417904, "grad_norm": 0.7559368522941781, "learning_rate": 1.5426874482327569e-06, "loss": 0.2881, "step": 21720 }, { "epoch": 1.5016246111303144, "grad_norm": 0.6582418170230153, "learning_rate": 1.5422830929273142e-06, "loss": 0.2517, "step": 21721 }, { "epoch": 1.5016937435188384, "grad_norm": 0.7798354615980838, "learning_rate": 1.541878780958181e-06, "loss": 0.2635, "step": 21722 }, { "epoch": 1.5017628759073625, "grad_norm": 0.7090790451756862, "learning_rate": 1.541474512330427e-06, "loss": 0.2911, "step": 21723 }, { "epoch": 1.5018320082958865, "grad_norm": 0.6840738868935191, "learning_rate": 1.5410702870491161e-06, "loss": 0.2944, "step": 21724 }, { "epoch": 1.5019011406844105, "grad_norm": 0.8734160220828776, "learning_rate": 1.5406661051193172e-06, "loss": 0.2539, "step": 21725 }, { "epoch": 1.5019702730729345, "grad_norm": 0.6826375638489413, "learning_rate": 1.540261966546095e-06, "loss": 0.279, "step": 21726 }, { "epoch": 1.5020394054614585, "grad_norm": 0.690709132176132, "learning_rate": 1.5398578713345164e-06, "loss": 0.3632, "step": 21727 }, { "epoch": 1.5021085378499828, "grad_norm": 0.687136956016977, "learning_rate": 1.5394538194896426e-06, "loss": 0.2787, "step": 21728 }, { "epoch": 1.5021776702385068, "grad_norm": 0.7346291047943071, "learning_rate": 1.5390498110165415e-06, "loss": 0.2036, "step": 21729 }, { "epoch": 1.5022468026270308, "grad_norm": 1.6090692227859484, "learning_rate": 1.538645845920273e-06, "loss": 0.2992, "step": 21730 }, { "epoch": 1.5023159350155548, "grad_norm": 0.7242419870588997, "learning_rate": 1.5382419242059016e-06, "loss": 0.2705, "step": 21731 }, { "epoch": 1.5023850674040788, "grad_norm": 0.7667132241037363, "learning_rate": 1.5378380458784898e-06, "loss": 0.2093, "step": 21732 }, { "epoch": 1.5024541997926029, "grad_norm": 0.7431897895446699, "learning_rate": 1.5374342109431012e-06, "loss": 0.1983, "step": 21733 }, { "epoch": 1.5025233321811269, "grad_norm": 0.6943728341512878, "learning_rate": 1.537030419404794e-06, "loss": 0.3461, "step": 21734 }, { "epoch": 1.502592464569651, "grad_norm": 0.7372768454943015, "learning_rate": 1.5366266712686328e-06, "loss": 0.24, "step": 21735 }, { "epoch": 1.502661596958175, "grad_norm": 0.6534177474090429, "learning_rate": 1.5362229665396734e-06, "loss": 0.2331, "step": 21736 }, { "epoch": 1.502730729346699, "grad_norm": 0.6915173428084089, "learning_rate": 1.5358193052229791e-06, "loss": 0.3662, "step": 21737 }, { "epoch": 1.502799861735223, "grad_norm": 0.964370369698348, "learning_rate": 1.5354156873236064e-06, "loss": 0.3054, "step": 21738 }, { "epoch": 1.502868994123747, "grad_norm": 0.646430159576939, "learning_rate": 1.5350121128466156e-06, "loss": 0.2592, "step": 21739 }, { "epoch": 1.502938126512271, "grad_norm": 0.776777856759007, "learning_rate": 1.5346085817970641e-06, "loss": 0.2877, "step": 21740 }, { "epoch": 1.503007258900795, "grad_norm": 0.657550233315199, "learning_rate": 1.5342050941800114e-06, "loss": 0.3534, "step": 21741 }, { "epoch": 1.503076391289319, "grad_norm": 0.7554664381979651, "learning_rate": 1.5338016500005109e-06, "loss": 0.3051, "step": 21742 }, { "epoch": 1.503145523677843, "grad_norm": 0.6586828962355146, "learning_rate": 1.5333982492636228e-06, "loss": 0.2459, "step": 21743 }, { "epoch": 1.503214656066367, "grad_norm": 0.7747493596517412, "learning_rate": 1.5329948919743998e-06, "loss": 0.3509, "step": 21744 }, { "epoch": 1.503283788454891, "grad_norm": 0.7220615514860601, "learning_rate": 1.5325915781379002e-06, "loss": 0.3134, "step": 21745 }, { "epoch": 1.503352920843415, "grad_norm": 0.6098208935916088, "learning_rate": 1.5321883077591742e-06, "loss": 0.2406, "step": 21746 }, { "epoch": 1.5034220532319391, "grad_norm": 0.7735086847836383, "learning_rate": 1.531785080843282e-06, "loss": 0.2728, "step": 21747 }, { "epoch": 1.5034911856204631, "grad_norm": 0.8034088753937578, "learning_rate": 1.5313818973952732e-06, "loss": 0.3159, "step": 21748 }, { "epoch": 1.5035603180089874, "grad_norm": 0.698263522259775, "learning_rate": 1.5309787574202039e-06, "loss": 0.2886, "step": 21749 }, { "epoch": 1.5036294503975114, "grad_norm": 0.6286321779438747, "learning_rate": 1.530575660923125e-06, "loss": 0.2013, "step": 21750 }, { "epoch": 1.5036985827860354, "grad_norm": 0.710523680143464, "learning_rate": 1.5301726079090872e-06, "loss": 0.2662, "step": 21751 }, { "epoch": 1.5037677151745594, "grad_norm": 0.7675332856816687, "learning_rate": 1.5297695983831435e-06, "loss": 0.3558, "step": 21752 }, { "epoch": 1.5038368475630834, "grad_norm": 1.0689253751557575, "learning_rate": 1.5293666323503453e-06, "loss": 0.2109, "step": 21753 }, { "epoch": 1.5039059799516075, "grad_norm": 0.6830963405058892, "learning_rate": 1.528963709815744e-06, "loss": 0.2338, "step": 21754 }, { "epoch": 1.5039751123401315, "grad_norm": 0.6569106657141907, "learning_rate": 1.5285608307843869e-06, "loss": 0.305, "step": 21755 }, { "epoch": 1.5040442447286555, "grad_norm": 0.7364179190468014, "learning_rate": 1.5281579952613258e-06, "loss": 0.2813, "step": 21756 }, { "epoch": 1.5041133771171795, "grad_norm": 0.7242926946808614, "learning_rate": 1.5277552032516074e-06, "loss": 0.2325, "step": 21757 }, { "epoch": 1.5041825095057035, "grad_norm": 0.6555188107160527, "learning_rate": 1.5273524547602825e-06, "loss": 0.2751, "step": 21758 }, { "epoch": 1.5042516418942276, "grad_norm": 0.791998482621792, "learning_rate": 1.5269497497923958e-06, "loss": 0.3453, "step": 21759 }, { "epoch": 1.5043207742827516, "grad_norm": 1.1425687022636672, "learning_rate": 1.526547088352996e-06, "loss": 0.2242, "step": 21760 }, { "epoch": 1.5043899066712756, "grad_norm": 0.6453440526723041, "learning_rate": 1.5261444704471295e-06, "loss": 0.2666, "step": 21761 }, { "epoch": 1.5044590390597996, "grad_norm": 0.661886748323758, "learning_rate": 1.5257418960798447e-06, "loss": 0.2875, "step": 21762 }, { "epoch": 1.5045281714483236, "grad_norm": 0.8066692753887232, "learning_rate": 1.5253393652561832e-06, "loss": 0.3125, "step": 21763 }, { "epoch": 1.5045973038368476, "grad_norm": 0.6645353826569371, "learning_rate": 1.5249368779811935e-06, "loss": 0.2434, "step": 21764 }, { "epoch": 1.5046664362253717, "grad_norm": 0.812025892640445, "learning_rate": 1.5245344342599173e-06, "loss": 0.3001, "step": 21765 }, { "epoch": 1.5047355686138957, "grad_norm": 0.7360977165848386, "learning_rate": 1.5241320340973998e-06, "loss": 0.4022, "step": 21766 }, { "epoch": 1.5048047010024197, "grad_norm": 0.7381607520672014, "learning_rate": 1.5237296774986838e-06, "loss": 0.2163, "step": 21767 }, { "epoch": 1.5048738333909437, "grad_norm": 0.6153776137573601, "learning_rate": 1.5233273644688146e-06, "loss": 0.2265, "step": 21768 }, { "epoch": 1.5049429657794677, "grad_norm": 0.6945846525036816, "learning_rate": 1.5229250950128306e-06, "loss": 0.3353, "step": 21769 }, { "epoch": 1.5050120981679918, "grad_norm": 0.7239582929035188, "learning_rate": 1.5225228691357774e-06, "loss": 0.2899, "step": 21770 }, { "epoch": 1.5050812305565158, "grad_norm": 0.6825355287319846, "learning_rate": 1.5221206868426924e-06, "loss": 0.2272, "step": 21771 }, { "epoch": 1.5051503629450398, "grad_norm": 0.7528532256887428, "learning_rate": 1.5217185481386194e-06, "loss": 0.2325, "step": 21772 }, { "epoch": 1.5052194953335638, "grad_norm": 0.6751409891333984, "learning_rate": 1.5213164530285956e-06, "loss": 0.3734, "step": 21773 }, { "epoch": 1.5052886277220878, "grad_norm": 0.9299781501341099, "learning_rate": 1.5209144015176624e-06, "loss": 0.2399, "step": 21774 }, { "epoch": 1.5053577601106118, "grad_norm": 0.7640649877104522, "learning_rate": 1.520512393610858e-06, "loss": 0.2495, "step": 21775 }, { "epoch": 1.5054268924991359, "grad_norm": 0.6955957377025342, "learning_rate": 1.5201104293132236e-06, "loss": 0.2646, "step": 21776 }, { "epoch": 1.5054960248876599, "grad_norm": 0.739360560994146, "learning_rate": 1.5197085086297925e-06, "loss": 0.2856, "step": 21777 }, { "epoch": 1.505565157276184, "grad_norm": 0.6990266902705782, "learning_rate": 1.5193066315656063e-06, "loss": 0.2024, "step": 21778 }, { "epoch": 1.505634289664708, "grad_norm": 5.008421080415361, "learning_rate": 1.5189047981256982e-06, "loss": 0.2477, "step": 21779 }, { "epoch": 1.505703422053232, "grad_norm": 0.7781947805511377, "learning_rate": 1.5185030083151082e-06, "loss": 0.3328, "step": 21780 }, { "epoch": 1.505772554441756, "grad_norm": 0.703365674968188, "learning_rate": 1.5181012621388675e-06, "loss": 0.224, "step": 21781 }, { "epoch": 1.50584168683028, "grad_norm": 0.6265869828077753, "learning_rate": 1.5176995596020144e-06, "loss": 0.2642, "step": 21782 }, { "epoch": 1.505910819218804, "grad_norm": 0.6999415972643441, "learning_rate": 1.5172979007095829e-06, "loss": 0.2831, "step": 21783 }, { "epoch": 1.505979951607328, "grad_norm": 0.7116395007097154, "learning_rate": 1.5168962854666086e-06, "loss": 0.3455, "step": 21784 }, { "epoch": 1.506049083995852, "grad_norm": 0.6263147491647124, "learning_rate": 1.5164947138781216e-06, "loss": 0.2634, "step": 21785 }, { "epoch": 1.506118216384376, "grad_norm": 0.6706477457393495, "learning_rate": 1.5160931859491584e-06, "loss": 0.2272, "step": 21786 }, { "epoch": 1.5061873487729, "grad_norm": 0.8315659430300018, "learning_rate": 1.5156917016847482e-06, "loss": 0.3571, "step": 21787 }, { "epoch": 1.506256481161424, "grad_norm": 0.6869652923204598, "learning_rate": 1.515290261089925e-06, "loss": 0.2555, "step": 21788 }, { "epoch": 1.506325613549948, "grad_norm": 0.7125502760472012, "learning_rate": 1.514888864169719e-06, "loss": 0.2184, "step": 21789 }, { "epoch": 1.5063947459384721, "grad_norm": 1.6051840804778963, "learning_rate": 1.514487510929164e-06, "loss": 0.273, "step": 21790 }, { "epoch": 1.5064638783269961, "grad_norm": 0.90321657882047, "learning_rate": 1.5140862013732854e-06, "loss": 0.3927, "step": 21791 }, { "epoch": 1.5065330107155201, "grad_norm": 0.7056226915642624, "learning_rate": 1.5136849355071177e-06, "loss": 0.2021, "step": 21792 }, { "epoch": 1.5066021431040442, "grad_norm": 0.8322443505375808, "learning_rate": 1.5132837133356858e-06, "loss": 0.2182, "step": 21793 }, { "epoch": 1.5066712754925682, "grad_norm": 0.6936913278610323, "learning_rate": 1.512882534864022e-06, "loss": 0.3506, "step": 21794 }, { "epoch": 1.5067404078810922, "grad_norm": 0.7034237402726823, "learning_rate": 1.5124814000971504e-06, "loss": 0.3158, "step": 21795 }, { "epoch": 1.5068095402696162, "grad_norm": 0.6626964765343208, "learning_rate": 1.512080309040101e-06, "loss": 0.2252, "step": 21796 }, { "epoch": 1.5068786726581402, "grad_norm": 0.7383896320398402, "learning_rate": 1.5116792616979008e-06, "loss": 0.2138, "step": 21797 }, { "epoch": 1.5069478050466643, "grad_norm": 0.6973449247373582, "learning_rate": 1.5112782580755769e-06, "loss": 0.3165, "step": 21798 }, { "epoch": 1.5070169374351883, "grad_norm": 0.6912390456338298, "learning_rate": 1.5108772981781533e-06, "loss": 0.2755, "step": 21799 }, { "epoch": 1.5070860698237123, "grad_norm": 0.7018333055833749, "learning_rate": 1.5104763820106572e-06, "loss": 0.1762, "step": 21800 }, { "epoch": 1.5071552022122363, "grad_norm": 0.7536913104082795, "learning_rate": 1.5100755095781105e-06, "loss": 0.297, "step": 21801 }, { "epoch": 1.5072243346007603, "grad_norm": 0.680754763424306, "learning_rate": 1.5096746808855395e-06, "loss": 0.3371, "step": 21802 }, { "epoch": 1.5072934669892843, "grad_norm": 0.6763949860272721, "learning_rate": 1.5092738959379676e-06, "loss": 0.2746, "step": 21803 }, { "epoch": 1.5073625993778084, "grad_norm": 0.6616857439741322, "learning_rate": 1.50887315474042e-06, "loss": 0.2475, "step": 21804 }, { "epoch": 1.5074317317663324, "grad_norm": 0.6524339325276269, "learning_rate": 1.5084724572979148e-06, "loss": 0.3074, "step": 21805 }, { "epoch": 1.5075008641548564, "grad_norm": 0.6847881370103532, "learning_rate": 1.5080718036154784e-06, "loss": 0.2223, "step": 21806 }, { "epoch": 1.5075699965433804, "grad_norm": 0.6554579725722932, "learning_rate": 1.5076711936981292e-06, "loss": 0.2031, "step": 21807 }, { "epoch": 1.5076391289319044, "grad_norm": 0.7113300627452693, "learning_rate": 1.5072706275508903e-06, "loss": 0.3277, "step": 21808 }, { "epoch": 1.5077082613204285, "grad_norm": 0.7154291256689903, "learning_rate": 1.5068701051787792e-06, "loss": 0.3243, "step": 21809 }, { "epoch": 1.5077773937089527, "grad_norm": 0.6934204059576669, "learning_rate": 1.5064696265868183e-06, "loss": 0.2976, "step": 21810 }, { "epoch": 1.5078465260974767, "grad_norm": 0.6679307609740429, "learning_rate": 1.5060691917800257e-06, "loss": 0.291, "step": 21811 }, { "epoch": 1.5079156584860007, "grad_norm": 0.7540284649138919, "learning_rate": 1.5056688007634223e-06, "loss": 0.3075, "step": 21812 }, { "epoch": 1.5079847908745247, "grad_norm": 0.7193884517072684, "learning_rate": 1.5052684535420226e-06, "loss": 0.2855, "step": 21813 }, { "epoch": 1.5080539232630488, "grad_norm": 0.6509871746954686, "learning_rate": 1.5048681501208484e-06, "loss": 0.2441, "step": 21814 }, { "epoch": 1.5081230556515728, "grad_norm": 0.7260991254790436, "learning_rate": 1.5044678905049127e-06, "loss": 0.2689, "step": 21815 }, { "epoch": 1.5081921880400968, "grad_norm": 0.7103453061890421, "learning_rate": 1.5040676746992356e-06, "loss": 0.356, "step": 21816 }, { "epoch": 1.5082613204286208, "grad_norm": 0.692312863245067, "learning_rate": 1.5036675027088293e-06, "loss": 0.2421, "step": 21817 }, { "epoch": 1.5083304528171448, "grad_norm": 0.7057075550470359, "learning_rate": 1.5032673745387122e-06, "loss": 0.2268, "step": 21818 }, { "epoch": 1.5083995852056689, "grad_norm": 0.7682621476907073, "learning_rate": 1.5028672901938996e-06, "loss": 0.2853, "step": 21819 }, { "epoch": 1.5084687175941929, "grad_norm": 0.724255906325693, "learning_rate": 1.502467249679403e-06, "loss": 0.3112, "step": 21820 }, { "epoch": 1.508537849982717, "grad_norm": 0.6184229875119356, "learning_rate": 1.502067253000239e-06, "loss": 0.2487, "step": 21821 }, { "epoch": 1.508606982371241, "grad_norm": 0.8607185905488981, "learning_rate": 1.5016673001614186e-06, "loss": 0.3159, "step": 21822 }, { "epoch": 1.508676114759765, "grad_norm": 0.7535161080593317, "learning_rate": 1.5012673911679553e-06, "loss": 0.3651, "step": 21823 }, { "epoch": 1.508745247148289, "grad_norm": 0.6795984944413311, "learning_rate": 1.5008675260248617e-06, "loss": 0.2598, "step": 21824 }, { "epoch": 1.508814379536813, "grad_norm": 0.6648322078191601, "learning_rate": 1.5004677047371501e-06, "loss": 0.2747, "step": 21825 }, { "epoch": 1.508883511925337, "grad_norm": 0.6777132074770529, "learning_rate": 1.50006792730983e-06, "loss": 0.2873, "step": 21826 }, { "epoch": 1.508952644313861, "grad_norm": 0.7332166591211418, "learning_rate": 1.4996681937479135e-06, "loss": 0.3434, "step": 21827 }, { "epoch": 1.509021776702385, "grad_norm": 0.6723227365978903, "learning_rate": 1.499268504056408e-06, "loss": 0.216, "step": 21828 }, { "epoch": 1.509090909090909, "grad_norm": 0.6363627605114115, "learning_rate": 1.4988688582403266e-06, "loss": 0.2449, "step": 21829 }, { "epoch": 1.5091600414794333, "grad_norm": 0.6963607019512731, "learning_rate": 1.4984692563046737e-06, "loss": 0.3567, "step": 21830 }, { "epoch": 1.5092291738679573, "grad_norm": 0.7716782803061703, "learning_rate": 1.4980696982544602e-06, "loss": 0.2817, "step": 21831 }, { "epoch": 1.5092983062564813, "grad_norm": 1.1343587095803889, "learning_rate": 1.497670184094694e-06, "loss": 0.2192, "step": 21832 }, { "epoch": 1.5093674386450053, "grad_norm": 0.6853826397298548, "learning_rate": 1.4972707138303826e-06, "loss": 0.3325, "step": 21833 }, { "epoch": 1.5094365710335294, "grad_norm": 0.6684262681874271, "learning_rate": 1.4968712874665309e-06, "loss": 0.3475, "step": 21834 }, { "epoch": 1.5095057034220534, "grad_norm": 0.7213719311149783, "learning_rate": 1.4964719050081472e-06, "loss": 0.2513, "step": 21835 }, { "epoch": 1.5095748358105774, "grad_norm": 0.6920605610919982, "learning_rate": 1.496072566460235e-06, "loss": 0.22, "step": 21836 }, { "epoch": 1.5096439681991014, "grad_norm": 0.7262285065750316, "learning_rate": 1.4956732718277994e-06, "loss": 0.3336, "step": 21837 }, { "epoch": 1.5097131005876254, "grad_norm": 0.6309133804402545, "learning_rate": 1.4952740211158457e-06, "loss": 0.2511, "step": 21838 }, { "epoch": 1.5097822329761494, "grad_norm": 0.6881748768042465, "learning_rate": 1.4948748143293795e-06, "loss": 0.303, "step": 21839 }, { "epoch": 1.5098513653646735, "grad_norm": 0.7339064360282372, "learning_rate": 1.4944756514734005e-06, "loss": 0.2986, "step": 21840 }, { "epoch": 1.5099204977531975, "grad_norm": 0.720891249726863, "learning_rate": 1.494076532552915e-06, "loss": 0.3575, "step": 21841 }, { "epoch": 1.5099896301417215, "grad_norm": 1.0915141367951455, "learning_rate": 1.493677457572922e-06, "loss": 0.2078, "step": 21842 }, { "epoch": 1.5100587625302455, "grad_norm": 0.6702131377914906, "learning_rate": 1.4932784265384265e-06, "loss": 0.2101, "step": 21843 }, { "epoch": 1.5101278949187695, "grad_norm": 0.7217520759121011, "learning_rate": 1.492879439454426e-06, "loss": 0.34, "step": 21844 }, { "epoch": 1.5101970273072935, "grad_norm": 0.6760077009231668, "learning_rate": 1.4924804963259231e-06, "loss": 0.2678, "step": 21845 }, { "epoch": 1.5102661596958176, "grad_norm": 0.7031308676818486, "learning_rate": 1.4920815971579178e-06, "loss": 0.2208, "step": 21846 }, { "epoch": 1.5103352920843416, "grad_norm": 0.7094276914948002, "learning_rate": 1.4916827419554108e-06, "loss": 0.2946, "step": 21847 }, { "epoch": 1.5104044244728656, "grad_norm": 0.7227935995239118, "learning_rate": 1.491283930723399e-06, "loss": 0.3623, "step": 21848 }, { "epoch": 1.5104735568613896, "grad_norm": 0.8652077957161437, "learning_rate": 1.4908851634668825e-06, "loss": 0.2647, "step": 21849 }, { "epoch": 1.5105426892499136, "grad_norm": 0.6520826848156083, "learning_rate": 1.4904864401908563e-06, "loss": 0.2636, "step": 21850 }, { "epoch": 1.5106118216384377, "grad_norm": 0.7581205972410244, "learning_rate": 1.4900877609003217e-06, "loss": 0.2824, "step": 21851 }, { "epoch": 1.5106809540269617, "grad_norm": 0.6656275087222906, "learning_rate": 1.4896891256002717e-06, "loss": 0.3367, "step": 21852 }, { "epoch": 1.5107500864154857, "grad_norm": 0.6388125600069267, "learning_rate": 1.4892905342957037e-06, "loss": 0.2164, "step": 21853 }, { "epoch": 1.5108192188040097, "grad_norm": 0.636542819721244, "learning_rate": 1.488891986991614e-06, "loss": 0.2874, "step": 21854 }, { "epoch": 1.5108883511925337, "grad_norm": 0.7030835631986476, "learning_rate": 1.4884934836929987e-06, "loss": 0.343, "step": 21855 }, { "epoch": 1.5109574835810577, "grad_norm": 0.6679855619309762, "learning_rate": 1.4880950244048497e-06, "loss": 0.2729, "step": 21856 }, { "epoch": 1.5110266159695818, "grad_norm": 0.7535347267555247, "learning_rate": 1.4876966091321637e-06, "loss": 0.2521, "step": 21857 }, { "epoch": 1.5110957483581058, "grad_norm": 0.6755427602202678, "learning_rate": 1.4872982378799317e-06, "loss": 0.2617, "step": 21858 }, { "epoch": 1.5111648807466298, "grad_norm": 0.721908467274933, "learning_rate": 1.4868999106531467e-06, "loss": 0.3514, "step": 21859 }, { "epoch": 1.5112340131351538, "grad_norm": 0.7831400821884297, "learning_rate": 1.4865016274568023e-06, "loss": 0.2264, "step": 21860 }, { "epoch": 1.5113031455236778, "grad_norm": 0.7097221979042104, "learning_rate": 1.486103388295892e-06, "loss": 0.2643, "step": 21861 }, { "epoch": 1.5113722779122019, "grad_norm": 0.7572113088980423, "learning_rate": 1.4857051931754024e-06, "loss": 0.3652, "step": 21862 }, { "epoch": 1.5114414103007259, "grad_norm": 0.7114707866563577, "learning_rate": 1.4853070421003284e-06, "loss": 0.2973, "step": 21863 }, { "epoch": 1.51151054268925, "grad_norm": 0.6106644822759143, "learning_rate": 1.4849089350756573e-06, "loss": 0.211, "step": 21864 }, { "epoch": 1.511579675077774, "grad_norm": 0.7029286152164045, "learning_rate": 1.4845108721063812e-06, "loss": 0.2894, "step": 21865 }, { "epoch": 1.511648807466298, "grad_norm": 0.7197676633392489, "learning_rate": 1.4841128531974858e-06, "loss": 0.376, "step": 21866 }, { "epoch": 1.511717939854822, "grad_norm": 0.6292803384824388, "learning_rate": 1.4837148783539618e-06, "loss": 0.2372, "step": 21867 }, { "epoch": 1.511787072243346, "grad_norm": 0.9978329996260776, "learning_rate": 1.4833169475807963e-06, "loss": 0.249, "step": 21868 }, { "epoch": 1.51185620463187, "grad_norm": 0.7451337904643756, "learning_rate": 1.4829190608829786e-06, "loss": 0.3173, "step": 21869 }, { "epoch": 1.511925337020394, "grad_norm": 0.7665866481852235, "learning_rate": 1.482521218265493e-06, "loss": 0.307, "step": 21870 }, { "epoch": 1.511994469408918, "grad_norm": 0.6924645524141524, "learning_rate": 1.4821234197333278e-06, "loss": 0.2163, "step": 21871 }, { "epoch": 1.512063601797442, "grad_norm": 0.7377071215964477, "learning_rate": 1.4817256652914657e-06, "loss": 0.3043, "step": 21872 }, { "epoch": 1.512132734185966, "grad_norm": 0.8314904854332882, "learning_rate": 1.4813279549448943e-06, "loss": 0.3333, "step": 21873 }, { "epoch": 1.51220186657449, "grad_norm": 0.6334407411993728, "learning_rate": 1.4809302886985977e-06, "loss": 0.2457, "step": 21874 }, { "epoch": 1.512270998963014, "grad_norm": 0.615763846189226, "learning_rate": 1.4805326665575609e-06, "loss": 0.2044, "step": 21875 }, { "epoch": 1.512340131351538, "grad_norm": 0.7070067965772437, "learning_rate": 1.4801350885267652e-06, "loss": 0.3215, "step": 21876 }, { "epoch": 1.5124092637400621, "grad_norm": 0.7783573980876622, "learning_rate": 1.4797375546111957e-06, "loss": 0.2968, "step": 21877 }, { "epoch": 1.5124783961285861, "grad_norm": 0.7044844970427953, "learning_rate": 1.479340064815833e-06, "loss": 0.2577, "step": 21878 }, { "epoch": 1.5125475285171102, "grad_norm": 0.8306715766148023, "learning_rate": 1.4789426191456607e-06, "loss": 0.2647, "step": 21879 }, { "epoch": 1.5126166609056342, "grad_norm": 0.7434739619492302, "learning_rate": 1.4785452176056581e-06, "loss": 0.336, "step": 21880 }, { "epoch": 1.5126857932941582, "grad_norm": 0.7096619480548519, "learning_rate": 1.478147860200806e-06, "loss": 0.2561, "step": 21881 }, { "epoch": 1.5127549256826822, "grad_norm": 0.7655606373633057, "learning_rate": 1.4777505469360858e-06, "loss": 0.2113, "step": 21882 }, { "epoch": 1.5128240580712062, "grad_norm": 0.6914651434448105, "learning_rate": 1.477353277816479e-06, "loss": 0.3065, "step": 21883 }, { "epoch": 1.5128931904597303, "grad_norm": 0.7468186428334826, "learning_rate": 1.4769560528469618e-06, "loss": 0.4172, "step": 21884 }, { "epoch": 1.5129623228482543, "grad_norm": 0.6699151343707253, "learning_rate": 1.4765588720325113e-06, "loss": 0.2438, "step": 21885 }, { "epoch": 1.5130314552367783, "grad_norm": 0.8068036015321081, "learning_rate": 1.47616173537811e-06, "loss": 0.3032, "step": 21886 }, { "epoch": 1.5131005876253023, "grad_norm": 0.6949305273512814, "learning_rate": 1.4757646428887307e-06, "loss": 0.3792, "step": 21887 }, { "epoch": 1.5131697200138263, "grad_norm": 0.6901650754220279, "learning_rate": 1.4753675945693523e-06, "loss": 0.2917, "step": 21888 }, { "epoch": 1.5132388524023503, "grad_norm": 0.7155805697861721, "learning_rate": 1.4749705904249512e-06, "loss": 0.2472, "step": 21889 }, { "epoch": 1.5133079847908744, "grad_norm": 0.7054400614927895, "learning_rate": 1.4745736304605046e-06, "loss": 0.3165, "step": 21890 }, { "epoch": 1.5133771171793986, "grad_norm": 0.6696052657610723, "learning_rate": 1.4741767146809842e-06, "loss": 0.3692, "step": 21891 }, { "epoch": 1.5134462495679226, "grad_norm": 0.7193854534045474, "learning_rate": 1.4737798430913686e-06, "loss": 0.2593, "step": 21892 }, { "epoch": 1.5135153819564466, "grad_norm": 0.6070035440944843, "learning_rate": 1.4733830156966283e-06, "loss": 0.1926, "step": 21893 }, { "epoch": 1.5135845143449707, "grad_norm": 0.6783550202330823, "learning_rate": 1.472986232501738e-06, "loss": 0.2984, "step": 21894 }, { "epoch": 1.5136536467334947, "grad_norm": 0.6972495265677765, "learning_rate": 1.4725894935116714e-06, "loss": 0.258, "step": 21895 }, { "epoch": 1.5137227791220187, "grad_norm": 0.6843551475041024, "learning_rate": 1.4721927987314021e-06, "loss": 0.2739, "step": 21896 }, { "epoch": 1.5137919115105427, "grad_norm": 0.7505226641656848, "learning_rate": 1.4717961481658989e-06, "loss": 0.2733, "step": 21897 }, { "epoch": 1.5138610438990667, "grad_norm": 0.779746788928747, "learning_rate": 1.471399541820136e-06, "loss": 0.3595, "step": 21898 }, { "epoch": 1.5139301762875907, "grad_norm": 0.6852164526893326, "learning_rate": 1.471002979699081e-06, "loss": 0.2085, "step": 21899 }, { "epoch": 1.5139993086761148, "grad_norm": 0.6796840963694977, "learning_rate": 1.4706064618077075e-06, "loss": 0.2358, "step": 21900 }, { "epoch": 1.5140684410646388, "grad_norm": 0.7767935767804532, "learning_rate": 1.4702099881509823e-06, "loss": 0.2908, "step": 21901 }, { "epoch": 1.5141375734531628, "grad_norm": 0.7020809018595838, "learning_rate": 1.4698135587338759e-06, "loss": 0.302, "step": 21902 }, { "epoch": 1.5142067058416868, "grad_norm": 0.7426547262848943, "learning_rate": 1.469417173561356e-06, "loss": 0.2475, "step": 21903 }, { "epoch": 1.5142758382302108, "grad_norm": 0.6404493438101507, "learning_rate": 1.469020832638393e-06, "loss": 0.2097, "step": 21904 }, { "epoch": 1.5143449706187349, "grad_norm": 0.7184733972797321, "learning_rate": 1.4686245359699514e-06, "loss": 0.3701, "step": 21905 }, { "epoch": 1.5144141030072589, "grad_norm": 0.7421906103974121, "learning_rate": 1.468228283561e-06, "loss": 0.1804, "step": 21906 }, { "epoch": 1.514483235395783, "grad_norm": 0.645707661232941, "learning_rate": 1.4678320754165032e-06, "loss": 0.2181, "step": 21907 }, { "epoch": 1.514552367784307, "grad_norm": 0.7200443838824222, "learning_rate": 1.4674359115414282e-06, "loss": 0.3054, "step": 21908 }, { "epoch": 1.514621500172831, "grad_norm": 0.7552487070707884, "learning_rate": 1.4670397919407397e-06, "loss": 0.3067, "step": 21909 }, { "epoch": 1.514690632561355, "grad_norm": 0.7250867618271419, "learning_rate": 1.4666437166194037e-06, "loss": 0.2096, "step": 21910 }, { "epoch": 1.5147597649498792, "grad_norm": 0.6531778324149164, "learning_rate": 1.466247685582382e-06, "loss": 0.2609, "step": 21911 }, { "epoch": 1.5148288973384032, "grad_norm": 0.7166324049759952, "learning_rate": 1.465851698834641e-06, "loss": 0.2999, "step": 21912 }, { "epoch": 1.5148980297269272, "grad_norm": 0.7262119941669053, "learning_rate": 1.46545575638114e-06, "loss": 0.281, "step": 21913 }, { "epoch": 1.5149671621154512, "grad_norm": 0.6696526423027093, "learning_rate": 1.465059858226845e-06, "loss": 0.2398, "step": 21914 }, { "epoch": 1.5150362945039753, "grad_norm": 0.6945484965013847, "learning_rate": 1.4646640043767152e-06, "loss": 0.2183, "step": 21915 }, { "epoch": 1.5151054268924993, "grad_norm": 0.8236299672405214, "learning_rate": 1.4642681948357123e-06, "loss": 0.3912, "step": 21916 }, { "epoch": 1.5151745592810233, "grad_norm": 0.6464155283255189, "learning_rate": 1.4638724296087987e-06, "loss": 0.2724, "step": 21917 }, { "epoch": 1.5152436916695473, "grad_norm": 0.62467753051517, "learning_rate": 1.4634767087009344e-06, "loss": 0.2041, "step": 21918 }, { "epoch": 1.5153128240580713, "grad_norm": 0.6955148362573877, "learning_rate": 1.4630810321170774e-06, "loss": 0.3004, "step": 21919 }, { "epoch": 1.5153819564465953, "grad_norm": 0.7019166221845029, "learning_rate": 1.4626853998621888e-06, "loss": 0.2478, "step": 21920 }, { "epoch": 1.5154510888351194, "grad_norm": 0.6260043188068869, "learning_rate": 1.4622898119412249e-06, "loss": 0.2071, "step": 21921 }, { "epoch": 1.5155202212236434, "grad_norm": 0.6337644218208612, "learning_rate": 1.4618942683591463e-06, "loss": 0.2917, "step": 21922 }, { "epoch": 1.5155893536121674, "grad_norm": 0.6671957066504342, "learning_rate": 1.4614987691209076e-06, "loss": 0.3332, "step": 21923 }, { "epoch": 1.5156584860006914, "grad_norm": 0.6879024606651537, "learning_rate": 1.4611033142314668e-06, "loss": 0.2817, "step": 21924 }, { "epoch": 1.5157276183892154, "grad_norm": 0.6222352810638018, "learning_rate": 1.4607079036957806e-06, "loss": 0.2123, "step": 21925 }, { "epoch": 1.5157967507777395, "grad_norm": 0.6986956392754999, "learning_rate": 1.4603125375188065e-06, "loss": 0.3144, "step": 21926 }, { "epoch": 1.5158658831662635, "grad_norm": 0.700702898810992, "learning_rate": 1.4599172157054957e-06, "loss": 0.3457, "step": 21927 }, { "epoch": 1.5159350155547875, "grad_norm": 0.616308059008706, "learning_rate": 1.4595219382608073e-06, "loss": 0.1884, "step": 21928 }, { "epoch": 1.5160041479433115, "grad_norm": 0.9061168302362043, "learning_rate": 1.4591267051896913e-06, "loss": 0.2342, "step": 21929 }, { "epoch": 1.5160732803318355, "grad_norm": 0.7627380450133382, "learning_rate": 1.4587315164971034e-06, "loss": 0.3809, "step": 21930 }, { "epoch": 1.5161424127203595, "grad_norm": 0.7148740283156048, "learning_rate": 1.4583363721879957e-06, "loss": 0.244, "step": 21931 }, { "epoch": 1.5162115451088836, "grad_norm": 0.6378357899690268, "learning_rate": 1.457941272267323e-06, "loss": 0.2468, "step": 21932 }, { "epoch": 1.5162806774974076, "grad_norm": 0.6718820643720397, "learning_rate": 1.457546216740034e-06, "loss": 0.2576, "step": 21933 }, { "epoch": 1.5163498098859316, "grad_norm": 0.7016179990198772, "learning_rate": 1.4571512056110825e-06, "loss": 0.3458, "step": 21934 }, { "epoch": 1.5164189422744556, "grad_norm": 0.7222301386767587, "learning_rate": 1.4567562388854167e-06, "loss": 0.2378, "step": 21935 }, { "epoch": 1.5164880746629796, "grad_norm": 0.6761172431776231, "learning_rate": 1.4563613165679902e-06, "loss": 0.2592, "step": 21936 }, { "epoch": 1.5165572070515037, "grad_norm": 0.6973192824021189, "learning_rate": 1.455966438663749e-06, "loss": 0.3172, "step": 21937 }, { "epoch": 1.5166263394400277, "grad_norm": 1.4749493813747716, "learning_rate": 1.4555716051776437e-06, "loss": 0.2879, "step": 21938 }, { "epoch": 1.5166954718285517, "grad_norm": 0.6992901696657847, "learning_rate": 1.4551768161146229e-06, "loss": 0.2796, "step": 21939 }, { "epoch": 1.5167646042170757, "grad_norm": 0.6286858182620744, "learning_rate": 1.4547820714796363e-06, "loss": 0.2604, "step": 21940 }, { "epoch": 1.5168337366055997, "grad_norm": 0.7581184237168587, "learning_rate": 1.4543873712776284e-06, "loss": 0.3366, "step": 21941 }, { "epoch": 1.5169028689941237, "grad_norm": 0.7556923281323493, "learning_rate": 1.4539927155135486e-06, "loss": 0.2081, "step": 21942 }, { "epoch": 1.5169720013826478, "grad_norm": 0.7578937138135675, "learning_rate": 1.453598104192341e-06, "loss": 0.2462, "step": 21943 }, { "epoch": 1.5170411337711718, "grad_norm": 0.7709467374972433, "learning_rate": 1.4532035373189519e-06, "loss": 0.2939, "step": 21944 }, { "epoch": 1.5171102661596958, "grad_norm": 0.7384771169068332, "learning_rate": 1.4528090148983271e-06, "loss": 0.3432, "step": 21945 }, { "epoch": 1.5171793985482198, "grad_norm": 0.606675189866502, "learning_rate": 1.452414536935412e-06, "loss": 0.2242, "step": 21946 }, { "epoch": 1.5172485309367438, "grad_norm": 0.6757795686207683, "learning_rate": 1.4520201034351488e-06, "loss": 0.2734, "step": 21947 }, { "epoch": 1.5173176633252679, "grad_norm": 0.6129720045960936, "learning_rate": 1.451625714402483e-06, "loss": 0.3271, "step": 21948 }, { "epoch": 1.5173867957137919, "grad_norm": 1.8107437272636497, "learning_rate": 1.4512313698423568e-06, "loss": 0.2018, "step": 21949 }, { "epoch": 1.5174559281023159, "grad_norm": 0.6499750296229884, "learning_rate": 1.4508370697597113e-06, "loss": 0.2428, "step": 21950 }, { "epoch": 1.51752506049084, "grad_norm": 0.7421427965794942, "learning_rate": 1.4504428141594889e-06, "loss": 0.4127, "step": 21951 }, { "epoch": 1.517594192879364, "grad_norm": 0.6416035374787825, "learning_rate": 1.450048603046631e-06, "loss": 0.3034, "step": 21952 }, { "epoch": 1.517663325267888, "grad_norm": 0.6375695631459686, "learning_rate": 1.4496544364260806e-06, "loss": 0.2113, "step": 21953 }, { "epoch": 1.517732457656412, "grad_norm": 0.7068572261842001, "learning_rate": 1.4492603143027745e-06, "loss": 0.2401, "step": 21954 }, { "epoch": 1.517801590044936, "grad_norm": 0.7101256179314996, "learning_rate": 1.4488662366816552e-06, "loss": 0.3266, "step": 21955 }, { "epoch": 1.51787072243346, "grad_norm": 0.6519379372168964, "learning_rate": 1.4484722035676584e-06, "loss": 0.256, "step": 21956 }, { "epoch": 1.517939854821984, "grad_norm": 0.6094881771602341, "learning_rate": 1.448078214965727e-06, "loss": 0.2445, "step": 21957 }, { "epoch": 1.518008987210508, "grad_norm": 0.6635131138731768, "learning_rate": 1.4476842708807942e-06, "loss": 0.2826, "step": 21958 }, { "epoch": 1.518078119599032, "grad_norm": 0.7522657493592845, "learning_rate": 1.4472903713178e-06, "loss": 0.3575, "step": 21959 }, { "epoch": 1.518147251987556, "grad_norm": 0.6476786285790337, "learning_rate": 1.4468965162816805e-06, "loss": 0.2294, "step": 21960 }, { "epoch": 1.51821638437608, "grad_norm": 0.6574908585035005, "learning_rate": 1.4465027057773744e-06, "loss": 0.2431, "step": 21961 }, { "epoch": 1.518285516764604, "grad_norm": 0.6941561913471, "learning_rate": 1.4461089398098138e-06, "loss": 0.3774, "step": 21962 }, { "epoch": 1.5183546491531281, "grad_norm": 0.7255298753139479, "learning_rate": 1.4457152183839369e-06, "loss": 0.3359, "step": 21963 }, { "epoch": 1.5184237815416521, "grad_norm": 1.0542576433339401, "learning_rate": 1.445321541504675e-06, "loss": 0.2344, "step": 21964 }, { "epoch": 1.5184929139301762, "grad_norm": 0.752578657021154, "learning_rate": 1.444927909176964e-06, "loss": 0.293, "step": 21965 }, { "epoch": 1.5185620463187002, "grad_norm": 0.726055785341499, "learning_rate": 1.4445343214057379e-06, "loss": 0.3117, "step": 21966 }, { "epoch": 1.5186311787072242, "grad_norm": 0.7334401340721958, "learning_rate": 1.4441407781959306e-06, "loss": 0.2262, "step": 21967 }, { "epoch": 1.5187003110957482, "grad_norm": 0.6474566595022471, "learning_rate": 1.4437472795524704e-06, "loss": 0.1984, "step": 21968 }, { "epoch": 1.5187694434842722, "grad_norm": 0.6733422827789664, "learning_rate": 1.4433538254802943e-06, "loss": 0.3036, "step": 21969 }, { "epoch": 1.5188385758727962, "grad_norm": 0.6786287968447846, "learning_rate": 1.4429604159843286e-06, "loss": 0.3065, "step": 21970 }, { "epoch": 1.5189077082613203, "grad_norm": 0.6850097829317985, "learning_rate": 1.4425670510695077e-06, "loss": 0.2299, "step": 21971 }, { "epoch": 1.5189768406498445, "grad_norm": 0.6452759102529044, "learning_rate": 1.4421737307407597e-06, "loss": 0.2074, "step": 21972 }, { "epoch": 1.5190459730383685, "grad_norm": 0.6804529837336314, "learning_rate": 1.441780455003014e-06, "loss": 0.3479, "step": 21973 }, { "epoch": 1.5191151054268925, "grad_norm": 0.678648683172191, "learning_rate": 1.4413872238612004e-06, "loss": 0.1999, "step": 21974 }, { "epoch": 1.5191842378154166, "grad_norm": 0.620648127670314, "learning_rate": 1.440994037320249e-06, "loss": 0.1963, "step": 21975 }, { "epoch": 1.5192533702039406, "grad_norm": 0.6687693537192625, "learning_rate": 1.4406008953850842e-06, "loss": 0.3051, "step": 21976 }, { "epoch": 1.5193225025924646, "grad_norm": 0.7199478927948553, "learning_rate": 1.440207798060637e-06, "loss": 0.3077, "step": 21977 }, { "epoch": 1.5193916349809886, "grad_norm": 0.6722431554728769, "learning_rate": 1.4398147453518308e-06, "loss": 0.2526, "step": 21978 }, { "epoch": 1.5194607673695126, "grad_norm": 0.8193228683074025, "learning_rate": 1.4394217372635932e-06, "loss": 0.304, "step": 21979 }, { "epoch": 1.5195298997580367, "grad_norm": 0.6824794666120396, "learning_rate": 1.4390287738008502e-06, "loss": 0.3033, "step": 21980 }, { "epoch": 1.5195990321465607, "grad_norm": 0.7098812312732092, "learning_rate": 1.438635854968528e-06, "loss": 0.2462, "step": 21981 }, { "epoch": 1.5196681645350847, "grad_norm": 0.629411883222801, "learning_rate": 1.438242980771548e-06, "loss": 0.2339, "step": 21982 }, { "epoch": 1.5197372969236087, "grad_norm": 0.6765996943293723, "learning_rate": 1.4378501512148384e-06, "loss": 0.2385, "step": 21983 }, { "epoch": 1.5198064293121327, "grad_norm": 0.7554401172388234, "learning_rate": 1.4374573663033182e-06, "loss": 0.3413, "step": 21984 }, { "epoch": 1.5198755617006567, "grad_norm": 0.6815156092618868, "learning_rate": 1.4370646260419135e-06, "loss": 0.283, "step": 21985 }, { "epoch": 1.5199446940891808, "grad_norm": 0.6474303271852219, "learning_rate": 1.436671930435544e-06, "loss": 0.1978, "step": 21986 }, { "epoch": 1.5200138264777048, "grad_norm": 0.6720938525086823, "learning_rate": 1.4362792794891328e-06, "loss": 0.3159, "step": 21987 }, { "epoch": 1.5200829588662288, "grad_norm": 0.661715970277341, "learning_rate": 1.4358866732076016e-06, "loss": 0.2205, "step": 21988 }, { "epoch": 1.5201520912547528, "grad_norm": 0.6926362239801916, "learning_rate": 1.4354941115958716e-06, "loss": 0.2446, "step": 21989 }, { "epoch": 1.5202212236432768, "grad_norm": 0.8961188654465302, "learning_rate": 1.4351015946588604e-06, "loss": 0.2296, "step": 21990 }, { "epoch": 1.5202903560318008, "grad_norm": 0.6366213015661185, "learning_rate": 1.4347091224014897e-06, "loss": 0.3382, "step": 21991 }, { "epoch": 1.5203594884203249, "grad_norm": 0.7278732569423386, "learning_rate": 1.4343166948286769e-06, "loss": 0.3327, "step": 21992 }, { "epoch": 1.520428620808849, "grad_norm": 0.6967718763142731, "learning_rate": 1.4339243119453423e-06, "loss": 0.2208, "step": 21993 }, { "epoch": 1.5204977531973731, "grad_norm": 0.7113506966049045, "learning_rate": 1.4335319737564008e-06, "loss": 0.3286, "step": 21994 }, { "epoch": 1.5205668855858971, "grad_norm": 0.7205227632670298, "learning_rate": 1.4331396802667718e-06, "loss": 0.2867, "step": 21995 }, { "epoch": 1.5206360179744212, "grad_norm": 0.7227725297690893, "learning_rate": 1.432747431481371e-06, "loss": 0.2817, "step": 21996 }, { "epoch": 1.5207051503629452, "grad_norm": 0.770259125546978, "learning_rate": 1.4323552274051172e-06, "loss": 0.263, "step": 21997 }, { "epoch": 1.5207742827514692, "grad_norm": 0.6882565228408035, "learning_rate": 1.431963068042922e-06, "loss": 0.3357, "step": 21998 }, { "epoch": 1.5208434151399932, "grad_norm": 0.817123734611449, "learning_rate": 1.431570953399704e-06, "loss": 0.2383, "step": 21999 }, { "epoch": 1.5209125475285172, "grad_norm": 0.7040174142968637, "learning_rate": 1.431178883480374e-06, "loss": 0.2331, "step": 22000 }, { "epoch": 1.5209816799170413, "grad_norm": 0.6827820883876758, "learning_rate": 1.4307868582898489e-06, "loss": 0.2561, "step": 22001 }, { "epoch": 1.5210508123055653, "grad_norm": 0.7371161564160666, "learning_rate": 1.43039487783304e-06, "loss": 0.3329, "step": 22002 }, { "epoch": 1.5211199446940893, "grad_norm": 0.6495551948297156, "learning_rate": 1.4300029421148625e-06, "loss": 0.2463, "step": 22003 }, { "epoch": 1.5211890770826133, "grad_norm": 0.8957238747374828, "learning_rate": 1.429611051140226e-06, "loss": 0.2034, "step": 22004 }, { "epoch": 1.5212582094711373, "grad_norm": 0.8834722627761639, "learning_rate": 1.4292192049140451e-06, "loss": 0.3339, "step": 22005 }, { "epoch": 1.5213273418596613, "grad_norm": 0.6250372452244548, "learning_rate": 1.4288274034412275e-06, "loss": 0.239, "step": 22006 }, { "epoch": 1.5213964742481854, "grad_norm": 0.6480002595989928, "learning_rate": 1.4284356467266868e-06, "loss": 0.2457, "step": 22007 }, { "epoch": 1.5214656066367094, "grad_norm": 0.7434562265831407, "learning_rate": 1.42804393477533e-06, "loss": 0.2748, "step": 22008 }, { "epoch": 1.5215347390252334, "grad_norm": 0.7390688021112538, "learning_rate": 1.4276522675920678e-06, "loss": 0.3475, "step": 22009 }, { "epoch": 1.5216038714137574, "grad_norm": 0.6990313636065814, "learning_rate": 1.4272606451818094e-06, "loss": 0.2762, "step": 22010 }, { "epoch": 1.5216730038022814, "grad_norm": 0.6862682758777009, "learning_rate": 1.4268690675494645e-06, "loss": 0.2844, "step": 22011 }, { "epoch": 1.5217421361908055, "grad_norm": 0.7378468134467366, "learning_rate": 1.4264775346999383e-06, "loss": 0.3312, "step": 22012 }, { "epoch": 1.5218112685793295, "grad_norm": 0.7057125719691197, "learning_rate": 1.4260860466381398e-06, "loss": 0.274, "step": 22013 }, { "epoch": 1.5218804009678535, "grad_norm": 0.6656319702419887, "learning_rate": 1.4256946033689735e-06, "loss": 0.2154, "step": 22014 }, { "epoch": 1.5219495333563775, "grad_norm": 0.7410923763870736, "learning_rate": 1.4253032048973464e-06, "loss": 0.3298, "step": 22015 }, { "epoch": 1.5220186657449015, "grad_norm": 0.7168741599440924, "learning_rate": 1.4249118512281661e-06, "loss": 0.3652, "step": 22016 }, { "epoch": 1.5220877981334255, "grad_norm": 0.6589958512097897, "learning_rate": 1.4245205423663344e-06, "loss": 0.2917, "step": 22017 }, { "epoch": 1.5221569305219496, "grad_norm": 0.7283638590547058, "learning_rate": 1.424129278316758e-06, "loss": 0.2775, "step": 22018 }, { "epoch": 1.5222260629104736, "grad_norm": 0.6962768467000942, "learning_rate": 1.4237380590843386e-06, "loss": 0.3532, "step": 22019 }, { "epoch": 1.5222951952989976, "grad_norm": 0.7380621623568405, "learning_rate": 1.4233468846739818e-06, "loss": 0.3289, "step": 22020 }, { "epoch": 1.5223643276875216, "grad_norm": 0.6942973905008324, "learning_rate": 1.4229557550905875e-06, "loss": 0.2091, "step": 22021 }, { "epoch": 1.5224334600760456, "grad_norm": 0.8107722972242097, "learning_rate": 1.4225646703390593e-06, "loss": 0.2298, "step": 22022 }, { "epoch": 1.5225025924645696, "grad_norm": 0.7179813101114678, "learning_rate": 1.4221736304242989e-06, "loss": 0.3471, "step": 22023 }, { "epoch": 1.5225717248530937, "grad_norm": 0.6763457761867439, "learning_rate": 1.4217826353512081e-06, "loss": 0.248, "step": 22024 }, { "epoch": 1.5226408572416177, "grad_norm": 0.7052500482240265, "learning_rate": 1.4213916851246857e-06, "loss": 0.2877, "step": 22025 }, { "epoch": 1.5227099896301417, "grad_norm": 0.7547721675369095, "learning_rate": 1.4210007797496333e-06, "loss": 0.2612, "step": 22026 }, { "epoch": 1.5227791220186657, "grad_norm": 0.6599493413023646, "learning_rate": 1.4206099192309474e-06, "loss": 0.2789, "step": 22027 }, { "epoch": 1.5228482544071897, "grad_norm": 0.6278618775411955, "learning_rate": 1.4202191035735302e-06, "loss": 0.2368, "step": 22028 }, { "epoch": 1.5229173867957138, "grad_norm": 0.700363880830378, "learning_rate": 1.4198283327822764e-06, "loss": 0.2466, "step": 22029 }, { "epoch": 1.5229865191842378, "grad_norm": 0.7482879119088166, "learning_rate": 1.419437606862085e-06, "loss": 0.359, "step": 22030 }, { "epoch": 1.5230556515727618, "grad_norm": 0.6885226792670456, "learning_rate": 1.419046925817854e-06, "loss": 0.2375, "step": 22031 }, { "epoch": 1.5231247839612858, "grad_norm": 0.658627851845254, "learning_rate": 1.418656289654481e-06, "loss": 0.2601, "step": 22032 }, { "epoch": 1.5231939163498098, "grad_norm": 0.731601797032692, "learning_rate": 1.4182656983768584e-06, "loss": 0.2779, "step": 22033 }, { "epoch": 1.5232630487383338, "grad_norm": 0.7020746767536546, "learning_rate": 1.4178751519898848e-06, "loss": 0.3572, "step": 22034 }, { "epoch": 1.5233321811268579, "grad_norm": 0.6860549553558046, "learning_rate": 1.4174846504984524e-06, "loss": 0.2494, "step": 22035 }, { "epoch": 1.5234013135153819, "grad_norm": 0.6548042336179588, "learning_rate": 1.4170941939074562e-06, "loss": 0.2062, "step": 22036 }, { "epoch": 1.523470445903906, "grad_norm": 0.708822838755654, "learning_rate": 1.4167037822217906e-06, "loss": 0.3514, "step": 22037 }, { "epoch": 1.52353957829243, "grad_norm": 0.6934266012141136, "learning_rate": 1.41631341544635e-06, "loss": 0.2469, "step": 22038 }, { "epoch": 1.523608710680954, "grad_norm": 0.6485512650777627, "learning_rate": 1.415923093586023e-06, "loss": 0.2035, "step": 22039 }, { "epoch": 1.523677843069478, "grad_norm": 0.7331965658217786, "learning_rate": 1.415532816645706e-06, "loss": 0.2825, "step": 22040 }, { "epoch": 1.523746975458002, "grad_norm": 0.6635003199359991, "learning_rate": 1.4151425846302875e-06, "loss": 0.3149, "step": 22041 }, { "epoch": 1.523816107846526, "grad_norm": 0.6372747836519894, "learning_rate": 1.4147523975446598e-06, "loss": 0.2742, "step": 22042 }, { "epoch": 1.52388524023505, "grad_norm": 0.6202620191780762, "learning_rate": 1.4143622553937114e-06, "loss": 0.2005, "step": 22043 }, { "epoch": 1.523954372623574, "grad_norm": 0.7128180416408979, "learning_rate": 1.4139721581823335e-06, "loss": 0.3127, "step": 22044 }, { "epoch": 1.524023505012098, "grad_norm": 0.6998196444331154, "learning_rate": 1.4135821059154149e-06, "loss": 0.3001, "step": 22045 }, { "epoch": 1.524092637400622, "grad_norm": 0.7225080818597717, "learning_rate": 1.4131920985978458e-06, "loss": 0.2402, "step": 22046 }, { "epoch": 1.524161769789146, "grad_norm": 0.7016321562662641, "learning_rate": 1.4128021362345118e-06, "loss": 0.2488, "step": 22047 }, { "epoch": 1.52423090217767, "grad_norm": 0.8117375872545327, "learning_rate": 1.4124122188303024e-06, "loss": 0.3667, "step": 22048 }, { "epoch": 1.5243000345661941, "grad_norm": 0.6618146813982451, "learning_rate": 1.4120223463901018e-06, "loss": 0.2671, "step": 22049 }, { "epoch": 1.5243691669547181, "grad_norm": 0.6881460254591634, "learning_rate": 1.4116325189187984e-06, "loss": 0.25, "step": 22050 }, { "epoch": 1.5244382993432422, "grad_norm": 0.7475481937801536, "learning_rate": 1.4112427364212777e-06, "loss": 0.3061, "step": 22051 }, { "epoch": 1.5245074317317662, "grad_norm": 0.6981498664851411, "learning_rate": 1.410852998902426e-06, "loss": 0.2982, "step": 22052 }, { "epoch": 1.5245765641202902, "grad_norm": 0.6660249118514756, "learning_rate": 1.4104633063671259e-06, "loss": 0.211, "step": 22053 }, { "epoch": 1.5246456965088144, "grad_norm": 0.737254368208543, "learning_rate": 1.4100736588202635e-06, "loss": 0.2704, "step": 22054 }, { "epoch": 1.5247148288973384, "grad_norm": 0.8084355073167895, "learning_rate": 1.4096840562667197e-06, "loss": 0.3427, "step": 22055 }, { "epoch": 1.5247839612858625, "grad_norm": 0.731137098671128, "learning_rate": 1.409294498711381e-06, "loss": 0.3093, "step": 22056 }, { "epoch": 1.5248530936743865, "grad_norm": 0.6060380049596739, "learning_rate": 1.4089049861591258e-06, "loss": 0.3042, "step": 22057 }, { "epoch": 1.5249222260629105, "grad_norm": 0.7894369170631743, "learning_rate": 1.4085155186148387e-06, "loss": 0.2896, "step": 22058 }, { "epoch": 1.5249913584514345, "grad_norm": 0.8270984320961831, "learning_rate": 1.4081260960833998e-06, "loss": 0.297, "step": 22059 }, { "epoch": 1.5250604908399585, "grad_norm": 0.6215936588030665, "learning_rate": 1.4077367185696916e-06, "loss": 0.1969, "step": 22060 }, { "epoch": 1.5251296232284826, "grad_norm": 0.6391511808388513, "learning_rate": 1.407347386078592e-06, "loss": 0.2054, "step": 22061 }, { "epoch": 1.5251987556170066, "grad_norm": 0.6855988320406721, "learning_rate": 1.4069580986149828e-06, "loss": 0.3213, "step": 22062 }, { "epoch": 1.5252678880055306, "grad_norm": 0.7125640417616277, "learning_rate": 1.4065688561837398e-06, "loss": 0.2788, "step": 22063 }, { "epoch": 1.5253370203940546, "grad_norm": 0.6696487147866688, "learning_rate": 1.4061796587897459e-06, "loss": 0.2292, "step": 22064 }, { "epoch": 1.5254061527825786, "grad_norm": 0.6534381197162886, "learning_rate": 1.4057905064378741e-06, "loss": 0.2422, "step": 22065 }, { "epoch": 1.5254752851711026, "grad_norm": 0.731662697922241, "learning_rate": 1.4054013991330045e-06, "loss": 0.3751, "step": 22066 }, { "epoch": 1.5255444175596267, "grad_norm": 0.6261353265614356, "learning_rate": 1.4050123368800134e-06, "loss": 0.2327, "step": 22067 }, { "epoch": 1.5256135499481507, "grad_norm": 0.6821969188545565, "learning_rate": 1.4046233196837788e-06, "loss": 0.2114, "step": 22068 }, { "epoch": 1.5256826823366747, "grad_norm": 0.6940992569675174, "learning_rate": 1.4042343475491732e-06, "loss": 0.3479, "step": 22069 }, { "epoch": 1.5257518147251987, "grad_norm": 0.7071241160814851, "learning_rate": 1.4038454204810747e-06, "loss": 0.3382, "step": 22070 }, { "epoch": 1.5258209471137227, "grad_norm": 0.778290289858316, "learning_rate": 1.4034565384843545e-06, "loss": 0.2823, "step": 22071 }, { "epoch": 1.5258900795022468, "grad_norm": 0.6821710244771592, "learning_rate": 1.4030677015638883e-06, "loss": 0.2648, "step": 22072 }, { "epoch": 1.5259592118907708, "grad_norm": 0.7156164157609379, "learning_rate": 1.4026789097245492e-06, "loss": 0.3315, "step": 22073 }, { "epoch": 1.526028344279295, "grad_norm": 0.6983924359389686, "learning_rate": 1.4022901629712122e-06, "loss": 0.2687, "step": 22074 }, { "epoch": 1.526097476667819, "grad_norm": 0.72114311753563, "learning_rate": 1.4019014613087456e-06, "loss": 0.2433, "step": 22075 }, { "epoch": 1.526166609056343, "grad_norm": 0.6655413710779737, "learning_rate": 1.4015128047420251e-06, "loss": 0.3019, "step": 22076 }, { "epoch": 1.526235741444867, "grad_norm": 0.9582368068904172, "learning_rate": 1.4011241932759184e-06, "loss": 0.3263, "step": 22077 }, { "epoch": 1.526304873833391, "grad_norm": 0.7376509271932892, "learning_rate": 1.4007356269152983e-06, "loss": 0.226, "step": 22078 }, { "epoch": 1.526374006221915, "grad_norm": 0.6520412603684771, "learning_rate": 1.4003471056650331e-06, "loss": 0.1657, "step": 22079 }, { "epoch": 1.5264431386104391, "grad_norm": 1.0311293083869568, "learning_rate": 1.3999586295299926e-06, "loss": 0.3345, "step": 22080 }, { "epoch": 1.5265122709989631, "grad_norm": 0.7066783489219229, "learning_rate": 1.3995701985150462e-06, "loss": 0.3094, "step": 22081 }, { "epoch": 1.5265814033874872, "grad_norm": 0.6588975636233032, "learning_rate": 1.399181812625064e-06, "loss": 0.258, "step": 22082 }, { "epoch": 1.5266505357760112, "grad_norm": 0.7476102036881759, "learning_rate": 1.3987934718649115e-06, "loss": 0.2969, "step": 22083 }, { "epoch": 1.5267196681645352, "grad_norm": 0.737189012785878, "learning_rate": 1.3984051762394547e-06, "loss": 0.3095, "step": 22084 }, { "epoch": 1.5267888005530592, "grad_norm": 0.6449781523562947, "learning_rate": 1.3980169257535632e-06, "loss": 0.2377, "step": 22085 }, { "epoch": 1.5268579329415832, "grad_norm": 0.670051214024792, "learning_rate": 1.3976287204120986e-06, "loss": 0.2093, "step": 22086 }, { "epoch": 1.5269270653301072, "grad_norm": 0.7296572849721508, "learning_rate": 1.3972405602199318e-06, "loss": 0.3185, "step": 22087 }, { "epoch": 1.5269961977186313, "grad_norm": 0.7453404818512286, "learning_rate": 1.3968524451819244e-06, "loss": 0.2788, "step": 22088 }, { "epoch": 1.5270653301071553, "grad_norm": 0.6434118354099012, "learning_rate": 1.3964643753029421e-06, "loss": 0.2179, "step": 22089 }, { "epoch": 1.5271344624956793, "grad_norm": 0.7388137784474867, "learning_rate": 1.3960763505878473e-06, "loss": 0.2511, "step": 22090 }, { "epoch": 1.5272035948842033, "grad_norm": 0.6771002615398597, "learning_rate": 1.3956883710415053e-06, "loss": 0.3159, "step": 22091 }, { "epoch": 1.5272727272727273, "grad_norm": 0.7554876824665979, "learning_rate": 1.395300436668776e-06, "loss": 0.2354, "step": 22092 }, { "epoch": 1.5273418596612514, "grad_norm": 0.6640263547434087, "learning_rate": 1.3949125474745223e-06, "loss": 0.2514, "step": 22093 }, { "epoch": 1.5274109920497754, "grad_norm": 0.724873857884968, "learning_rate": 1.3945247034636066e-06, "loss": 0.3135, "step": 22094 }, { "epoch": 1.5274801244382994, "grad_norm": 0.7636420070834407, "learning_rate": 1.3941369046408916e-06, "loss": 0.3459, "step": 22095 }, { "epoch": 1.5275492568268234, "grad_norm": 0.6401445781039123, "learning_rate": 1.3937491510112333e-06, "loss": 0.22, "step": 22096 }, { "epoch": 1.5276183892153474, "grad_norm": 0.62248758094438, "learning_rate": 1.393361442579495e-06, "loss": 0.2115, "step": 22097 }, { "epoch": 1.5276875216038714, "grad_norm": 0.7468254398898166, "learning_rate": 1.392973779350534e-06, "loss": 0.3733, "step": 22098 }, { "epoch": 1.5277566539923955, "grad_norm": 0.7186792183168694, "learning_rate": 1.3925861613292107e-06, "loss": 0.2398, "step": 22099 }, { "epoch": 1.5278257863809195, "grad_norm": 0.6583864597457993, "learning_rate": 1.3921985885203804e-06, "loss": 0.2644, "step": 22100 }, { "epoch": 1.5278949187694435, "grad_norm": 0.7477749551357374, "learning_rate": 1.3918110609289025e-06, "loss": 0.2949, "step": 22101 }, { "epoch": 1.5279640511579675, "grad_norm": 0.6900066036699594, "learning_rate": 1.3914235785596341e-06, "loss": 0.317, "step": 22102 }, { "epoch": 1.5280331835464915, "grad_norm": 0.7318532843293832, "learning_rate": 1.391036141417433e-06, "loss": 0.2441, "step": 22103 }, { "epoch": 1.5281023159350156, "grad_norm": 0.6075281572821626, "learning_rate": 1.3906487495071519e-06, "loss": 0.227, "step": 22104 }, { "epoch": 1.5281714483235396, "grad_norm": 0.642239618968646, "learning_rate": 1.3902614028336486e-06, "loss": 0.334, "step": 22105 }, { "epoch": 1.5282405807120636, "grad_norm": 0.7566271416735343, "learning_rate": 1.3898741014017752e-06, "loss": 0.2367, "step": 22106 }, { "epoch": 1.5283097131005876, "grad_norm": 0.6334000490500391, "learning_rate": 1.389486845216388e-06, "loss": 0.2381, "step": 22107 }, { "epoch": 1.5283788454891116, "grad_norm": 0.6141153544020973, "learning_rate": 1.3890996342823393e-06, "loss": 0.2481, "step": 22108 }, { "epoch": 1.5284479778776356, "grad_norm": 0.7733858366861555, "learning_rate": 1.3887124686044845e-06, "loss": 0.3488, "step": 22109 }, { "epoch": 1.5285171102661597, "grad_norm": 0.7041396954343232, "learning_rate": 1.388325348187673e-06, "loss": 0.2389, "step": 22110 }, { "epoch": 1.5285862426546837, "grad_norm": 0.6892511655902879, "learning_rate": 1.387938273036759e-06, "loss": 0.2062, "step": 22111 }, { "epoch": 1.5286553750432077, "grad_norm": 0.6930411276731722, "learning_rate": 1.3875512431565917e-06, "loss": 0.3245, "step": 22112 }, { "epoch": 1.5287245074317317, "grad_norm": 0.6927079266839792, "learning_rate": 1.387164258552024e-06, "loss": 0.2447, "step": 22113 }, { "epoch": 1.5287936398202557, "grad_norm": 0.8493528760154809, "learning_rate": 1.3867773192279037e-06, "loss": 0.2533, "step": 22114 }, { "epoch": 1.5288627722087798, "grad_norm": 0.7009386258934895, "learning_rate": 1.3863904251890813e-06, "loss": 0.2671, "step": 22115 }, { "epoch": 1.5289319045973038, "grad_norm": 0.72921402861893, "learning_rate": 1.386003576440406e-06, "loss": 0.3899, "step": 22116 }, { "epoch": 1.5290010369858278, "grad_norm": 2.5793821223368987, "learning_rate": 1.3856167729867288e-06, "loss": 0.2023, "step": 22117 }, { "epoch": 1.5290701693743518, "grad_norm": 0.7051530862972537, "learning_rate": 1.385230014832893e-06, "loss": 0.2441, "step": 22118 }, { "epoch": 1.5291393017628758, "grad_norm": 0.6761036217129941, "learning_rate": 1.38484330198375e-06, "loss": 0.282, "step": 22119 }, { "epoch": 1.5292084341513998, "grad_norm": 0.7251977810805441, "learning_rate": 1.3844566344441424e-06, "loss": 0.3478, "step": 22120 }, { "epoch": 1.5292775665399239, "grad_norm": 0.7265224668782996, "learning_rate": 1.384070012218921e-06, "loss": 0.2806, "step": 22121 }, { "epoch": 1.5293466989284479, "grad_norm": 0.7546308645781118, "learning_rate": 1.3836834353129252e-06, "loss": 0.265, "step": 22122 }, { "epoch": 1.529415831316972, "grad_norm": 0.7139819894086515, "learning_rate": 1.383296903731008e-06, "loss": 0.3369, "step": 22123 }, { "epoch": 1.529484963705496, "grad_norm": 0.6931409352710192, "learning_rate": 1.3829104174780073e-06, "loss": 0.2264, "step": 22124 }, { "epoch": 1.52955409609402, "grad_norm": 0.789051369343643, "learning_rate": 1.3825239765587716e-06, "loss": 0.2418, "step": 22125 }, { "epoch": 1.529623228482544, "grad_norm": 0.7148269359242353, "learning_rate": 1.3821375809781408e-06, "loss": 0.2865, "step": 22126 }, { "epoch": 1.529692360871068, "grad_norm": 0.6993980003640019, "learning_rate": 1.3817512307409602e-06, "loss": 0.3438, "step": 22127 }, { "epoch": 1.529761493259592, "grad_norm": 0.673458472840679, "learning_rate": 1.3813649258520696e-06, "loss": 0.2197, "step": 22128 }, { "epoch": 1.529830625648116, "grad_norm": 0.7400195562007229, "learning_rate": 1.3809786663163116e-06, "loss": 0.2498, "step": 22129 }, { "epoch": 1.52989975803664, "grad_norm": 0.6513088868311411, "learning_rate": 1.3805924521385277e-06, "loss": 0.3169, "step": 22130 }, { "epoch": 1.529968890425164, "grad_norm": 0.7230122662625986, "learning_rate": 1.3802062833235602e-06, "loss": 0.304, "step": 22131 }, { "epoch": 1.530038022813688, "grad_norm": 1.001786109178614, "learning_rate": 1.379820159876245e-06, "loss": 0.2296, "step": 22132 }, { "epoch": 1.530107155202212, "grad_norm": 0.6932390919958862, "learning_rate": 1.3794340818014257e-06, "loss": 0.3152, "step": 22133 }, { "epoch": 1.530176287590736, "grad_norm": 0.6604813404776045, "learning_rate": 1.379048049103937e-06, "loss": 0.3172, "step": 22134 }, { "epoch": 1.5302454199792603, "grad_norm": 0.6500234465153131, "learning_rate": 1.3786620617886209e-06, "loss": 0.2266, "step": 22135 }, { "epoch": 1.5303145523677844, "grad_norm": 0.7220851371928135, "learning_rate": 1.3782761198603118e-06, "loss": 0.2652, "step": 22136 }, { "epoch": 1.5303836847563084, "grad_norm": 0.8413720988693096, "learning_rate": 1.3778902233238484e-06, "loss": 0.3642, "step": 22137 }, { "epoch": 1.5304528171448324, "grad_norm": 0.8836071537615117, "learning_rate": 1.377504372184067e-06, "loss": 0.2839, "step": 22138 }, { "epoch": 1.5305219495333564, "grad_norm": 0.6512369813820329, "learning_rate": 1.3771185664458048e-06, "loss": 0.2002, "step": 22139 }, { "epoch": 1.5305910819218804, "grad_norm": 1.0239518445879818, "learning_rate": 1.376732806113895e-06, "loss": 0.2884, "step": 22140 }, { "epoch": 1.5306602143104044, "grad_norm": 0.7912763448779292, "learning_rate": 1.3763470911931749e-06, "loss": 0.3338, "step": 22141 }, { "epoch": 1.5307293466989285, "grad_norm": 0.6526552361193754, "learning_rate": 1.3759614216884753e-06, "loss": 0.2696, "step": 22142 }, { "epoch": 1.5307984790874525, "grad_norm": 0.7064429640942198, "learning_rate": 1.3755757976046324e-06, "loss": 0.1989, "step": 22143 }, { "epoch": 1.5308676114759765, "grad_norm": 0.7701518100821918, "learning_rate": 1.375190218946479e-06, "loss": 0.3399, "step": 22144 }, { "epoch": 1.5309367438645005, "grad_norm": 0.7410891896999278, "learning_rate": 1.3748046857188486e-06, "loss": 0.246, "step": 22145 }, { "epoch": 1.5310058762530245, "grad_norm": 0.6607295353274778, "learning_rate": 1.3744191979265704e-06, "loss": 0.2309, "step": 22146 }, { "epoch": 1.5310750086415486, "grad_norm": 0.6915948857268311, "learning_rate": 1.3740337555744792e-06, "loss": 0.2432, "step": 22147 }, { "epoch": 1.5311441410300726, "grad_norm": 0.7141417483802076, "learning_rate": 1.3736483586674022e-06, "loss": 0.3885, "step": 22148 }, { "epoch": 1.5312132734185966, "grad_norm": 0.6711356094350557, "learning_rate": 1.3732630072101732e-06, "loss": 0.2495, "step": 22149 }, { "epoch": 1.5312824058071206, "grad_norm": 0.714940168189364, "learning_rate": 1.372877701207619e-06, "loss": 0.2286, "step": 22150 }, { "epoch": 1.5313515381956446, "grad_norm": 0.7788291735072734, "learning_rate": 1.3724924406645696e-06, "loss": 0.322, "step": 22151 }, { "epoch": 1.5314206705841686, "grad_norm": 0.6847801775525197, "learning_rate": 1.3721072255858552e-06, "loss": 0.2839, "step": 22152 }, { "epoch": 1.5314898029726927, "grad_norm": 0.6118967863258775, "learning_rate": 1.3717220559763012e-06, "loss": 0.2169, "step": 22153 }, { "epoch": 1.5315589353612167, "grad_norm": 0.8482202706242497, "learning_rate": 1.3713369318407377e-06, "loss": 0.2413, "step": 22154 }, { "epoch": 1.531628067749741, "grad_norm": 0.6660503690997266, "learning_rate": 1.370951853183989e-06, "loss": 0.2698, "step": 22155 }, { "epoch": 1.531697200138265, "grad_norm": 0.6471845427401396, "learning_rate": 1.3705668200108835e-06, "loss": 0.2384, "step": 22156 }, { "epoch": 1.531766332526789, "grad_norm": 0.6964989101923126, "learning_rate": 1.3701818323262428e-06, "loss": 0.2012, "step": 22157 }, { "epoch": 1.531835464915313, "grad_norm": 0.9557423846213062, "learning_rate": 1.3697968901348984e-06, "loss": 0.2256, "step": 22158 }, { "epoch": 1.531904597303837, "grad_norm": 0.7044413109538491, "learning_rate": 1.3694119934416705e-06, "loss": 0.3262, "step": 22159 }, { "epoch": 1.531973729692361, "grad_norm": 0.6443635250157743, "learning_rate": 1.3690271422513851e-06, "loss": 0.2396, "step": 22160 }, { "epoch": 1.532042862080885, "grad_norm": 0.6720268434396712, "learning_rate": 1.3686423365688633e-06, "loss": 0.2708, "step": 22161 }, { "epoch": 1.532111994469409, "grad_norm": 0.7404022303373755, "learning_rate": 1.3682575763989309e-06, "loss": 0.3288, "step": 22162 }, { "epoch": 1.532181126857933, "grad_norm": 0.7043403332365001, "learning_rate": 1.3678728617464077e-06, "loss": 0.3084, "step": 22163 }, { "epoch": 1.532250259246457, "grad_norm": 0.5786724328678226, "learning_rate": 1.3674881926161154e-06, "loss": 0.1661, "step": 22164 }, { "epoch": 1.532319391634981, "grad_norm": 0.7861398244550659, "learning_rate": 1.3671035690128765e-06, "loss": 0.2511, "step": 22165 }, { "epoch": 1.5323885240235051, "grad_norm": 0.7084351462629861, "learning_rate": 1.366718990941513e-06, "loss": 0.3865, "step": 22166 }, { "epoch": 1.5324576564120291, "grad_norm": 0.609223164358514, "learning_rate": 1.3663344584068412e-06, "loss": 0.2216, "step": 22167 }, { "epoch": 1.5325267888005532, "grad_norm": 0.7055680803055772, "learning_rate": 1.3659499714136841e-06, "loss": 0.2409, "step": 22168 }, { "epoch": 1.5325959211890772, "grad_norm": 0.7875119592848284, "learning_rate": 1.3655655299668569e-06, "loss": 0.3301, "step": 22169 }, { "epoch": 1.5326650535776012, "grad_norm": 0.6829414899179111, "learning_rate": 1.3651811340711817e-06, "loss": 0.2762, "step": 22170 }, { "epoch": 1.5327341859661252, "grad_norm": 0.6561312372604315, "learning_rate": 1.364796783731473e-06, "loss": 0.2176, "step": 22171 }, { "epoch": 1.5328033183546492, "grad_norm": 0.6926741767229275, "learning_rate": 1.3644124789525493e-06, "loss": 0.2725, "step": 22172 }, { "epoch": 1.5328724507431732, "grad_norm": 0.7891306152758369, "learning_rate": 1.3640282197392268e-06, "loss": 0.3562, "step": 22173 }, { "epoch": 1.5329415831316973, "grad_norm": 0.6380189144462836, "learning_rate": 1.3636440060963234e-06, "loss": 0.2055, "step": 22174 }, { "epoch": 1.5330107155202213, "grad_norm": 0.6200355017682252, "learning_rate": 1.363259838028652e-06, "loss": 0.2013, "step": 22175 }, { "epoch": 1.5330798479087453, "grad_norm": 0.7527489692863486, "learning_rate": 1.36287571554103e-06, "loss": 0.2772, "step": 22176 }, { "epoch": 1.5331489802972693, "grad_norm": 0.7727473071356288, "learning_rate": 1.3624916386382682e-06, "loss": 0.3676, "step": 22177 }, { "epoch": 1.5332181126857933, "grad_norm": 0.9298938875672521, "learning_rate": 1.362107607325182e-06, "loss": 0.1922, "step": 22178 }, { "epoch": 1.5332872450743174, "grad_norm": 0.678329266116326, "learning_rate": 1.3617236216065855e-06, "loss": 0.2718, "step": 22179 }, { "epoch": 1.5333563774628414, "grad_norm": 0.7188344050469438, "learning_rate": 1.361339681487292e-06, "loss": 0.3418, "step": 22180 }, { "epoch": 1.5334255098513654, "grad_norm": 0.7382219884117385, "learning_rate": 1.3609557869721108e-06, "loss": 0.294, "step": 22181 }, { "epoch": 1.5334946422398894, "grad_norm": 0.6490766213424771, "learning_rate": 1.3605719380658555e-06, "loss": 0.275, "step": 22182 }, { "epoch": 1.5335637746284134, "grad_norm": 0.8814897711404481, "learning_rate": 1.3601881347733348e-06, "loss": 0.2526, "step": 22183 }, { "epoch": 1.5336329070169374, "grad_norm": 0.7782398361979891, "learning_rate": 1.3598043770993618e-06, "loss": 0.3532, "step": 22184 }, { "epoch": 1.5337020394054615, "grad_norm": 0.7096499178637865, "learning_rate": 1.3594206650487435e-06, "loss": 0.2245, "step": 22185 }, { "epoch": 1.5337711717939855, "grad_norm": 0.626591462916724, "learning_rate": 1.3590369986262902e-06, "loss": 0.2304, "step": 22186 }, { "epoch": 1.5338403041825095, "grad_norm": 0.7246488630492319, "learning_rate": 1.3586533778368105e-06, "loss": 0.3544, "step": 22187 }, { "epoch": 1.5339094365710335, "grad_norm": 0.7835933367910692, "learning_rate": 1.3582698026851138e-06, "loss": 0.3029, "step": 22188 }, { "epoch": 1.5339785689595575, "grad_norm": 0.656137341574036, "learning_rate": 1.357886273176005e-06, "loss": 0.3012, "step": 22189 }, { "epoch": 1.5340477013480815, "grad_norm": 0.7120275850918385, "learning_rate": 1.3575027893142935e-06, "loss": 0.284, "step": 22190 }, { "epoch": 1.5341168337366056, "grad_norm": 0.653333591904496, "learning_rate": 1.3571193511047819e-06, "loss": 0.3081, "step": 22191 }, { "epoch": 1.5341859661251296, "grad_norm": 0.6966701864649226, "learning_rate": 1.356735958552281e-06, "loss": 0.2958, "step": 22192 }, { "epoch": 1.5342550985136536, "grad_norm": 0.7070904568990888, "learning_rate": 1.3563526116615895e-06, "loss": 0.2145, "step": 22193 }, { "epoch": 1.5343242309021776, "grad_norm": 0.7014129914688458, "learning_rate": 1.3559693104375182e-06, "loss": 0.3149, "step": 22194 }, { "epoch": 1.5343933632907016, "grad_norm": 0.7460725567614488, "learning_rate": 1.3555860548848676e-06, "loss": 0.2997, "step": 22195 }, { "epoch": 1.5344624956792257, "grad_norm": 0.7258896272753838, "learning_rate": 1.3552028450084437e-06, "loss": 0.2471, "step": 22196 }, { "epoch": 1.5345316280677497, "grad_norm": 0.757305173381716, "learning_rate": 1.354819680813046e-06, "loss": 0.2316, "step": 22197 }, { "epoch": 1.5346007604562737, "grad_norm": 0.7351671536270802, "learning_rate": 1.35443656230348e-06, "loss": 0.3353, "step": 22198 }, { "epoch": 1.5346698928447977, "grad_norm": 0.7209084451225551, "learning_rate": 1.3540534894845442e-06, "loss": 0.2565, "step": 22199 }, { "epoch": 1.5347390252333217, "grad_norm": 0.6984359244796746, "learning_rate": 1.3536704623610413e-06, "loss": 0.2144, "step": 22200 }, { "epoch": 1.5348081576218457, "grad_norm": 0.6959392131671149, "learning_rate": 1.3532874809377728e-06, "loss": 0.2959, "step": 22201 }, { "epoch": 1.5348772900103698, "grad_norm": 0.8356512344507189, "learning_rate": 1.352904545219539e-06, "loss": 0.3183, "step": 22202 }, { "epoch": 1.5349464223988938, "grad_norm": 0.7128362437855618, "learning_rate": 1.352521655211136e-06, "loss": 0.2323, "step": 22203 }, { "epoch": 1.5350155547874178, "grad_norm": 0.6163483207579279, "learning_rate": 1.3521388109173673e-06, "loss": 0.2592, "step": 22204 }, { "epoch": 1.5350846871759418, "grad_norm": 0.7131167477037724, "learning_rate": 1.3517560123430261e-06, "loss": 0.3315, "step": 22205 }, { "epoch": 1.5351538195644658, "grad_norm": 0.7029114029787069, "learning_rate": 1.3513732594929151e-06, "loss": 0.283, "step": 22206 }, { "epoch": 1.5352229519529899, "grad_norm": 0.672844108505, "learning_rate": 1.3509905523718274e-06, "loss": 0.1995, "step": 22207 }, { "epoch": 1.5352920843415139, "grad_norm": 0.6265169339633837, "learning_rate": 1.350607890984561e-06, "loss": 0.2639, "step": 22208 }, { "epoch": 1.535361216730038, "grad_norm": 0.6822455648150859, "learning_rate": 1.3502252753359124e-06, "loss": 0.3764, "step": 22209 }, { "epoch": 1.535430349118562, "grad_norm": 0.6606874653271549, "learning_rate": 1.349842705430678e-06, "loss": 0.2385, "step": 22210 }, { "epoch": 1.535499481507086, "grad_norm": 0.679608248097612, "learning_rate": 1.3494601812736497e-06, "loss": 0.2686, "step": 22211 }, { "epoch": 1.53556861389561, "grad_norm": 0.6812330350765287, "learning_rate": 1.3490777028696245e-06, "loss": 0.3083, "step": 22212 }, { "epoch": 1.535637746284134, "grad_norm": 0.9652811119722934, "learning_rate": 1.3486952702233936e-06, "loss": 0.2405, "step": 22213 }, { "epoch": 1.535706878672658, "grad_norm": 0.6193992345720687, "learning_rate": 1.3483128833397518e-06, "loss": 0.2402, "step": 22214 }, { "epoch": 1.535776011061182, "grad_norm": 0.7122405089687303, "learning_rate": 1.3479305422234923e-06, "loss": 0.2621, "step": 22215 }, { "epoch": 1.5358451434497062, "grad_norm": 0.9951731052576792, "learning_rate": 1.347548246879405e-06, "loss": 0.322, "step": 22216 }, { "epoch": 1.5359142758382303, "grad_norm": 0.5997359646749104, "learning_rate": 1.3471659973122837e-06, "loss": 0.2365, "step": 22217 }, { "epoch": 1.5359834082267543, "grad_norm": 0.6903109753396244, "learning_rate": 1.3467837935269163e-06, "loss": 0.3124, "step": 22218 }, { "epoch": 1.5360525406152783, "grad_norm": 0.68371946664378, "learning_rate": 1.3464016355280963e-06, "loss": 0.3474, "step": 22219 }, { "epoch": 1.5361216730038023, "grad_norm": 0.7021740271515541, "learning_rate": 1.3460195233206097e-06, "loss": 0.2565, "step": 22220 }, { "epoch": 1.5361908053923263, "grad_norm": 0.6650550401957623, "learning_rate": 1.345637456909248e-06, "loss": 0.1819, "step": 22221 }, { "epoch": 1.5362599377808503, "grad_norm": 0.6338743857837844, "learning_rate": 1.3452554362987991e-06, "loss": 0.2219, "step": 22222 }, { "epoch": 1.5363290701693744, "grad_norm": 0.7056752181532344, "learning_rate": 1.3448734614940523e-06, "loss": 0.3377, "step": 22223 }, { "epoch": 1.5363982025578984, "grad_norm": 0.6866232479822442, "learning_rate": 1.3444915324997926e-06, "loss": 0.2392, "step": 22224 }, { "epoch": 1.5364673349464224, "grad_norm": 0.6995397496587551, "learning_rate": 1.3441096493208094e-06, "loss": 0.2502, "step": 22225 }, { "epoch": 1.5365364673349464, "grad_norm": 0.87005477583046, "learning_rate": 1.3437278119618863e-06, "loss": 0.2657, "step": 22226 }, { "epoch": 1.5366055997234704, "grad_norm": 0.705832959334592, "learning_rate": 1.3433460204278115e-06, "loss": 0.3816, "step": 22227 }, { "epoch": 1.5366747321119945, "grad_norm": 0.6044783690732969, "learning_rate": 1.3429642747233656e-06, "loss": 0.2212, "step": 22228 }, { "epoch": 1.5367438645005185, "grad_norm": 0.8439531566910242, "learning_rate": 1.3425825748533395e-06, "loss": 0.2945, "step": 22229 }, { "epoch": 1.5368129968890425, "grad_norm": 0.7090030992673464, "learning_rate": 1.3422009208225124e-06, "loss": 0.3486, "step": 22230 }, { "epoch": 1.5368821292775665, "grad_norm": 0.6852283053978814, "learning_rate": 1.3418193126356698e-06, "loss": 0.2786, "step": 22231 }, { "epoch": 1.5369512616660905, "grad_norm": 0.6695587641561823, "learning_rate": 1.3414377502975933e-06, "loss": 0.2739, "step": 22232 }, { "epoch": 1.5370203940546145, "grad_norm": 0.706628875395982, "learning_rate": 1.3410562338130666e-06, "loss": 0.2083, "step": 22233 }, { "epoch": 1.5370895264431386, "grad_norm": 0.7071107970388723, "learning_rate": 1.3406747631868688e-06, "loss": 0.3295, "step": 22234 }, { "epoch": 1.5371586588316626, "grad_norm": 0.6578782972955194, "learning_rate": 1.3402933384237832e-06, "loss": 0.2355, "step": 22235 }, { "epoch": 1.5372277912201868, "grad_norm": 0.6542911319633699, "learning_rate": 1.3399119595285887e-06, "loss": 0.2521, "step": 22236 }, { "epoch": 1.5372969236087108, "grad_norm": 0.7082041880924234, "learning_rate": 1.339530626506068e-06, "loss": 0.2978, "step": 22237 }, { "epoch": 1.5373660559972349, "grad_norm": 0.646824448489502, "learning_rate": 1.3391493393609967e-06, "loss": 0.2895, "step": 22238 }, { "epoch": 1.5374351883857589, "grad_norm": 0.6880383383104172, "learning_rate": 1.338768098098157e-06, "loss": 0.204, "step": 22239 }, { "epoch": 1.537504320774283, "grad_norm": 0.7032609250293934, "learning_rate": 1.338386902722324e-06, "loss": 0.2219, "step": 22240 }, { "epoch": 1.537573453162807, "grad_norm": 0.8103891753766703, "learning_rate": 1.3380057532382784e-06, "loss": 0.3838, "step": 22241 }, { "epoch": 1.537642585551331, "grad_norm": 0.6240789025252926, "learning_rate": 1.337624649650794e-06, "loss": 0.2106, "step": 22242 }, { "epoch": 1.537711717939855, "grad_norm": 0.6442337166940352, "learning_rate": 1.3372435919646487e-06, "loss": 0.2378, "step": 22243 }, { "epoch": 1.537780850328379, "grad_norm": 0.685493648483991, "learning_rate": 1.336862580184618e-06, "loss": 0.3029, "step": 22244 }, { "epoch": 1.537849982716903, "grad_norm": 0.6860191720422087, "learning_rate": 1.3364816143154796e-06, "loss": 0.2698, "step": 22245 }, { "epoch": 1.537919115105427, "grad_norm": 0.6500686965741849, "learning_rate": 1.3361006943620053e-06, "loss": 0.1955, "step": 22246 }, { "epoch": 1.537988247493951, "grad_norm": 0.691324128812562, "learning_rate": 1.3357198203289712e-06, "loss": 0.2556, "step": 22247 }, { "epoch": 1.538057379882475, "grad_norm": 0.6866509113202333, "learning_rate": 1.3353389922211486e-06, "loss": 0.3891, "step": 22248 }, { "epoch": 1.538126512270999, "grad_norm": 0.7108736783112144, "learning_rate": 1.3349582100433117e-06, "loss": 0.2192, "step": 22249 }, { "epoch": 1.538195644659523, "grad_norm": 0.6835661850095889, "learning_rate": 1.3345774738002331e-06, "loss": 0.2403, "step": 22250 }, { "epoch": 1.538264777048047, "grad_norm": 0.6650444227473249, "learning_rate": 1.3341967834966867e-06, "loss": 0.2673, "step": 22251 }, { "epoch": 1.5383339094365711, "grad_norm": 0.7083575581388776, "learning_rate": 1.3338161391374394e-06, "loss": 0.2991, "step": 22252 }, { "epoch": 1.5384030418250951, "grad_norm": 0.6439699323042891, "learning_rate": 1.3334355407272665e-06, "loss": 0.2463, "step": 22253 }, { "epoch": 1.5384721742136191, "grad_norm": 0.6428719821382091, "learning_rate": 1.3330549882709336e-06, "loss": 0.2037, "step": 22254 }, { "epoch": 1.5385413066021432, "grad_norm": 0.6614865218848637, "learning_rate": 1.332674481773214e-06, "loss": 0.3167, "step": 22255 }, { "epoch": 1.5386104389906672, "grad_norm": 0.6573772723320341, "learning_rate": 1.332294021238874e-06, "loss": 0.2717, "step": 22256 }, { "epoch": 1.5386795713791912, "grad_norm": 0.7109885027169129, "learning_rate": 1.3319136066726828e-06, "loss": 0.2634, "step": 22257 }, { "epoch": 1.5387487037677152, "grad_norm": 0.9457601530611834, "learning_rate": 1.3315332380794089e-06, "loss": 0.2725, "step": 22258 }, { "epoch": 1.5388178361562392, "grad_norm": 0.7252393668097961, "learning_rate": 1.3311529154638209e-06, "loss": 0.3317, "step": 22259 }, { "epoch": 1.5388869685447633, "grad_norm": 0.6608735195351115, "learning_rate": 1.330772638830682e-06, "loss": 0.2481, "step": 22260 }, { "epoch": 1.5389561009332873, "grad_norm": 0.6201484731431918, "learning_rate": 1.3303924081847614e-06, "loss": 0.2163, "step": 22261 }, { "epoch": 1.5390252333218113, "grad_norm": 0.7865541840298089, "learning_rate": 1.3300122235308215e-06, "loss": 0.3545, "step": 22262 }, { "epoch": 1.5390943657103353, "grad_norm": 1.0146541622234997, "learning_rate": 1.3296320848736311e-06, "loss": 0.2674, "step": 22263 }, { "epoch": 1.5391634980988593, "grad_norm": 0.8284674008871287, "learning_rate": 1.3292519922179486e-06, "loss": 0.1997, "step": 22264 }, { "epoch": 1.5392326304873833, "grad_norm": 0.7261139762889167, "learning_rate": 1.3288719455685455e-06, "loss": 0.2153, "step": 22265 }, { "epoch": 1.5393017628759074, "grad_norm": 0.7620707179470083, "learning_rate": 1.3284919449301793e-06, "loss": 0.3172, "step": 22266 }, { "epoch": 1.5393708952644314, "grad_norm": 0.7189496716325035, "learning_rate": 1.3281119903076156e-06, "loss": 0.2356, "step": 22267 }, { "epoch": 1.5394400276529554, "grad_norm": 0.6651327571331703, "learning_rate": 1.3277320817056145e-06, "loss": 0.2098, "step": 22268 }, { "epoch": 1.5395091600414794, "grad_norm": 0.70484399204088, "learning_rate": 1.3273522191289396e-06, "loss": 0.3312, "step": 22269 }, { "epoch": 1.5395782924300034, "grad_norm": 0.667809888588621, "learning_rate": 1.326972402582349e-06, "loss": 0.2959, "step": 22270 }, { "epoch": 1.5396474248185275, "grad_norm": 0.6631891950130618, "learning_rate": 1.326592632070604e-06, "loss": 0.2245, "step": 22271 }, { "epoch": 1.5397165572070515, "grad_norm": 0.7091333719514672, "learning_rate": 1.3262129075984652e-06, "loss": 0.2374, "step": 22272 }, { "epoch": 1.5397856895955755, "grad_norm": 0.691444480432675, "learning_rate": 1.3258332291706933e-06, "loss": 0.3104, "step": 22273 }, { "epoch": 1.5398548219840995, "grad_norm": 0.7321202174269675, "learning_rate": 1.3254535967920428e-06, "loss": 0.2998, "step": 22274 }, { "epoch": 1.5399239543726235, "grad_norm": 0.6136004516416912, "learning_rate": 1.325074010467276e-06, "loss": 0.2043, "step": 22275 }, { "epoch": 1.5399930867611475, "grad_norm": 0.6702278233355513, "learning_rate": 1.3246944702011467e-06, "loss": 0.2414, "step": 22276 }, { "epoch": 1.5400622191496716, "grad_norm": 0.6890557118781077, "learning_rate": 1.3243149759984146e-06, "loss": 0.3073, "step": 22277 }, { "epoch": 1.5401313515381956, "grad_norm": 0.7711021415418994, "learning_rate": 1.3239355278638338e-06, "loss": 0.2254, "step": 22278 }, { "epoch": 1.5402004839267196, "grad_norm": 1.0085131449547253, "learning_rate": 1.3235561258021606e-06, "loss": 0.2649, "step": 22279 }, { "epoch": 1.5402696163152436, "grad_norm": 0.7059859432430156, "learning_rate": 1.3231767698181502e-06, "loss": 0.3346, "step": 22280 }, { "epoch": 1.5403387487037676, "grad_norm": 0.697051165845958, "learning_rate": 1.3227974599165594e-06, "loss": 0.2713, "step": 22281 }, { "epoch": 1.5404078810922917, "grad_norm": 0.6385081610550865, "learning_rate": 1.3224181961021399e-06, "loss": 0.2415, "step": 22282 }, { "epoch": 1.5404770134808157, "grad_norm": 0.6571737183040302, "learning_rate": 1.322038978379644e-06, "loss": 0.2565, "step": 22283 }, { "epoch": 1.5405461458693397, "grad_norm": 0.6713082529909705, "learning_rate": 1.3216598067538255e-06, "loss": 0.3527, "step": 22284 }, { "epoch": 1.5406152782578637, "grad_norm": 0.6799149402584556, "learning_rate": 1.3212806812294372e-06, "loss": 0.2392, "step": 22285 }, { "epoch": 1.5406844106463877, "grad_norm": 0.6612163053577709, "learning_rate": 1.320901601811232e-06, "loss": 0.2544, "step": 22286 }, { "epoch": 1.5407535430349117, "grad_norm": 0.7167087447178375, "learning_rate": 1.3205225685039586e-06, "loss": 0.3108, "step": 22287 }, { "epoch": 1.5408226754234358, "grad_norm": 0.7132953505010562, "learning_rate": 1.3201435813123691e-06, "loss": 0.2838, "step": 22288 }, { "epoch": 1.5408918078119598, "grad_norm": 0.7468872430712321, "learning_rate": 1.3197646402412122e-06, "loss": 0.2498, "step": 22289 }, { "epoch": 1.5409609402004838, "grad_norm": 0.7443125623055061, "learning_rate": 1.3193857452952391e-06, "loss": 0.2631, "step": 22290 }, { "epoch": 1.5410300725890078, "grad_norm": 0.6797411102846398, "learning_rate": 1.3190068964791958e-06, "loss": 0.3654, "step": 22291 }, { "epoch": 1.5410992049775318, "grad_norm": 0.78109160819336, "learning_rate": 1.3186280937978319e-06, "loss": 0.2454, "step": 22292 }, { "epoch": 1.5411683373660559, "grad_norm": 0.6550109329339033, "learning_rate": 1.3182493372558953e-06, "loss": 0.2251, "step": 22293 }, { "epoch": 1.5412374697545799, "grad_norm": 0.7180099550525268, "learning_rate": 1.3178706268581338e-06, "loss": 0.3048, "step": 22294 }, { "epoch": 1.5413066021431039, "grad_norm": 0.7560427417849911, "learning_rate": 1.317491962609292e-06, "loss": 0.3275, "step": 22295 }, { "epoch": 1.541375734531628, "grad_norm": 0.779173334413438, "learning_rate": 1.3171133445141183e-06, "loss": 0.2328, "step": 22296 }, { "epoch": 1.5414448669201521, "grad_norm": 0.6691352622991628, "learning_rate": 1.316734772577355e-06, "loss": 0.2279, "step": 22297 }, { "epoch": 1.5415139993086762, "grad_norm": 0.7051764155504306, "learning_rate": 1.3163562468037495e-06, "loss": 0.3305, "step": 22298 }, { "epoch": 1.5415831316972002, "grad_norm": 0.79601583314703, "learning_rate": 1.315977767198041e-06, "loss": 0.2616, "step": 22299 }, { "epoch": 1.5416522640857242, "grad_norm": 0.6925748180837087, "learning_rate": 1.3155993337649808e-06, "loss": 0.2164, "step": 22300 }, { "epoch": 1.5417213964742482, "grad_norm": 0.7008792696466735, "learning_rate": 1.3152209465093058e-06, "loss": 0.2815, "step": 22301 }, { "epoch": 1.5417905288627722, "grad_norm": 0.6548240948361396, "learning_rate": 1.3148426054357622e-06, "loss": 0.3379, "step": 22302 }, { "epoch": 1.5418596612512963, "grad_norm": 0.6549439386441053, "learning_rate": 1.3144643105490885e-06, "loss": 0.2875, "step": 22303 }, { "epoch": 1.5419287936398203, "grad_norm": 0.6831142325762359, "learning_rate": 1.314086061854029e-06, "loss": 0.2803, "step": 22304 }, { "epoch": 1.5419979260283443, "grad_norm": 0.6611667728251341, "learning_rate": 1.3137078593553215e-06, "loss": 0.349, "step": 22305 }, { "epoch": 1.5420670584168683, "grad_norm": 0.714382970625566, "learning_rate": 1.3133297030577076e-06, "loss": 0.282, "step": 22306 }, { "epoch": 1.5421361908053923, "grad_norm": 0.6915900134096026, "learning_rate": 1.3129515929659264e-06, "loss": 0.2334, "step": 22307 }, { "epoch": 1.5422053231939163, "grad_norm": 0.7024852295891658, "learning_rate": 1.3125735290847191e-06, "loss": 0.2909, "step": 22308 }, { "epoch": 1.5422744555824404, "grad_norm": 1.1956433503341772, "learning_rate": 1.3121955114188201e-06, "loss": 0.4106, "step": 22309 }, { "epoch": 1.5423435879709644, "grad_norm": 0.7105321041689179, "learning_rate": 1.311817539972971e-06, "loss": 0.2303, "step": 22310 }, { "epoch": 1.5424127203594884, "grad_norm": 0.6736068185226999, "learning_rate": 1.3114396147519054e-06, "loss": 0.1981, "step": 22311 }, { "epoch": 1.5424818527480124, "grad_norm": 0.6626511023037036, "learning_rate": 1.3110617357603633e-06, "loss": 0.2783, "step": 22312 }, { "epoch": 1.5425509851365364, "grad_norm": 0.7283781434578164, "learning_rate": 1.3106839030030776e-06, "loss": 0.3565, "step": 22313 }, { "epoch": 1.5426201175250605, "grad_norm": 0.7434836827935546, "learning_rate": 1.3103061164847857e-06, "loss": 0.2389, "step": 22314 }, { "epoch": 1.5426892499135845, "grad_norm": 0.7041694333582813, "learning_rate": 1.3099283762102217e-06, "loss": 0.2519, "step": 22315 }, { "epoch": 1.5427583823021085, "grad_norm": 0.6493436363468852, "learning_rate": 1.309550682184122e-06, "loss": 0.3032, "step": 22316 }, { "epoch": 1.5428275146906325, "grad_norm": 0.8916602286573362, "learning_rate": 1.3091730344112168e-06, "loss": 0.2487, "step": 22317 }, { "epoch": 1.5428966470791567, "grad_norm": 1.24579144033064, "learning_rate": 1.3087954328962426e-06, "loss": 0.2469, "step": 22318 }, { "epoch": 1.5429657794676808, "grad_norm": 0.7172313901579784, "learning_rate": 1.3084178776439293e-06, "loss": 0.2624, "step": 22319 }, { "epoch": 1.5430349118562048, "grad_norm": 0.7060615150495338, "learning_rate": 1.3080403686590093e-06, "loss": 0.2844, "step": 22320 }, { "epoch": 1.5431040442447288, "grad_norm": 0.7178140766783804, "learning_rate": 1.3076629059462154e-06, "loss": 0.2942, "step": 22321 }, { "epoch": 1.5431731766332528, "grad_norm": 0.7035331422820916, "learning_rate": 1.3072854895102788e-06, "loss": 0.21, "step": 22322 }, { "epoch": 1.5432423090217768, "grad_norm": 0.6798263757947686, "learning_rate": 1.306908119355927e-06, "loss": 0.2976, "step": 22323 }, { "epoch": 1.5433114414103009, "grad_norm": 0.7276333266099693, "learning_rate": 1.306530795487893e-06, "loss": 0.2735, "step": 22324 }, { "epoch": 1.5433805737988249, "grad_norm": 0.6310927233975415, "learning_rate": 1.3061535179109025e-06, "loss": 0.2659, "step": 22325 }, { "epoch": 1.543449706187349, "grad_norm": 0.6686487862320194, "learning_rate": 1.3057762866296875e-06, "loss": 0.2844, "step": 22326 }, { "epoch": 1.543518838575873, "grad_norm": 0.646314492300467, "learning_rate": 1.3053991016489726e-06, "loss": 0.2858, "step": 22327 }, { "epoch": 1.543587970964397, "grad_norm": 0.6830837166669342, "learning_rate": 1.3050219629734868e-06, "loss": 0.243, "step": 22328 }, { "epoch": 1.543657103352921, "grad_norm": 0.6914757885573559, "learning_rate": 1.304644870607956e-06, "loss": 0.2354, "step": 22329 }, { "epoch": 1.543726235741445, "grad_norm": 0.7296850913248991, "learning_rate": 1.3042678245571094e-06, "loss": 0.3886, "step": 22330 }, { "epoch": 1.543795368129969, "grad_norm": 0.7145507324887114, "learning_rate": 1.3038908248256681e-06, "loss": 0.2756, "step": 22331 }, { "epoch": 1.543864500518493, "grad_norm": 0.6787165076711764, "learning_rate": 1.3035138714183616e-06, "loss": 0.2246, "step": 22332 }, { "epoch": 1.543933632907017, "grad_norm": 0.6773547158126585, "learning_rate": 1.30313696433991e-06, "loss": 0.2339, "step": 22333 }, { "epoch": 1.544002765295541, "grad_norm": 0.7186403277781296, "learning_rate": 1.3027601035950405e-06, "loss": 0.3177, "step": 22334 }, { "epoch": 1.544071897684065, "grad_norm": 0.6846283730913204, "learning_rate": 1.3023832891884725e-06, "loss": 0.2971, "step": 22335 }, { "epoch": 1.544141030072589, "grad_norm": 0.6982545490826434, "learning_rate": 1.3020065211249338e-06, "loss": 0.2377, "step": 22336 }, { "epoch": 1.544210162461113, "grad_norm": 0.712970874505871, "learning_rate": 1.3016297994091431e-06, "loss": 0.3337, "step": 22337 }, { "epoch": 1.544279294849637, "grad_norm": 0.657500086204924, "learning_rate": 1.3012531240458242e-06, "loss": 0.3307, "step": 22338 }, { "epoch": 1.5443484272381611, "grad_norm": 0.7131916494151099, "learning_rate": 1.3008764950396946e-06, "loss": 0.2382, "step": 22339 }, { "epoch": 1.5444175596266851, "grad_norm": 0.6708627276279732, "learning_rate": 1.3004999123954786e-06, "loss": 0.2386, "step": 22340 }, { "epoch": 1.5444866920152092, "grad_norm": 0.6816876044886735, "learning_rate": 1.3001233761178934e-06, "loss": 0.3905, "step": 22341 }, { "epoch": 1.5445558244037332, "grad_norm": 0.7065674061079056, "learning_rate": 1.299746886211658e-06, "loss": 0.2337, "step": 22342 }, { "epoch": 1.5446249567922572, "grad_norm": 0.8421639321020646, "learning_rate": 1.2993704426814924e-06, "loss": 0.226, "step": 22343 }, { "epoch": 1.5446940891807812, "grad_norm": 0.6820071659439911, "learning_rate": 1.2989940455321155e-06, "loss": 0.28, "step": 22344 }, { "epoch": 1.5447632215693052, "grad_norm": 0.7108603081666248, "learning_rate": 1.2986176947682421e-06, "loss": 0.2638, "step": 22345 }, { "epoch": 1.5448323539578293, "grad_norm": 0.790571342612539, "learning_rate": 1.2982413903945917e-06, "loss": 0.2849, "step": 22346 }, { "epoch": 1.5449014863463533, "grad_norm": 0.7293826969114482, "learning_rate": 1.2978651324158782e-06, "loss": 0.3461, "step": 22347 }, { "epoch": 1.5449706187348773, "grad_norm": 1.20494281715571, "learning_rate": 1.2974889208368202e-06, "loss": 0.3899, "step": 22348 }, { "epoch": 1.5450397511234013, "grad_norm": 0.7658845015023351, "learning_rate": 1.2971127556621293e-06, "loss": 0.2885, "step": 22349 }, { "epoch": 1.5451088835119253, "grad_norm": 0.7870756600563489, "learning_rate": 1.2967366368965223e-06, "loss": 0.2821, "step": 22350 }, { "epoch": 1.5451780159004493, "grad_norm": 0.682273081271976, "learning_rate": 1.2963605645447136e-06, "loss": 0.279, "step": 22351 }, { "epoch": 1.5452471482889734, "grad_norm": 0.7503716476383958, "learning_rate": 1.295984538611415e-06, "loss": 0.3757, "step": 22352 }, { "epoch": 1.5453162806774974, "grad_norm": 0.9481922850478329, "learning_rate": 1.2956085591013413e-06, "loss": 0.2433, "step": 22353 }, { "epoch": 1.5453854130660214, "grad_norm": 0.650237357419899, "learning_rate": 1.2952326260192017e-06, "loss": 0.2327, "step": 22354 }, { "epoch": 1.5454545454545454, "grad_norm": 0.6809394490955532, "learning_rate": 1.2948567393697104e-06, "loss": 0.3525, "step": 22355 }, { "epoch": 1.5455236778430694, "grad_norm": 0.7460839388277686, "learning_rate": 1.2944808991575768e-06, "loss": 0.2864, "step": 22356 }, { "epoch": 1.5455928102315935, "grad_norm": 0.6634318838151949, "learning_rate": 1.2941051053875148e-06, "loss": 0.2008, "step": 22357 }, { "epoch": 1.5456619426201175, "grad_norm": 0.6818196913650159, "learning_rate": 1.2937293580642296e-06, "loss": 0.2599, "step": 22358 }, { "epoch": 1.5457310750086415, "grad_norm": 0.7681976996639617, "learning_rate": 1.293353657192435e-06, "loss": 0.3585, "step": 22359 }, { "epoch": 1.5458002073971655, "grad_norm": 1.0891371372195375, "learning_rate": 1.292978002776835e-06, "loss": 0.1978, "step": 22360 }, { "epoch": 1.5458693397856895, "grad_norm": 0.6849668910263135, "learning_rate": 1.2926023948221428e-06, "loss": 0.2051, "step": 22361 }, { "epoch": 1.5459384721742135, "grad_norm": 0.7165025599084628, "learning_rate": 1.2922268333330611e-06, "loss": 0.3057, "step": 22362 }, { "epoch": 1.5460076045627376, "grad_norm": 0.6815115825695626, "learning_rate": 1.2918513183143e-06, "loss": 0.2986, "step": 22363 }, { "epoch": 1.5460767369512616, "grad_norm": 0.7533413135361392, "learning_rate": 1.291475849770565e-06, "loss": 0.2643, "step": 22364 }, { "epoch": 1.5461458693397856, "grad_norm": 0.8400167403281447, "learning_rate": 1.291100427706563e-06, "loss": 0.2808, "step": 22365 }, { "epoch": 1.5462150017283096, "grad_norm": 0.7065944483850938, "learning_rate": 1.2907250521269971e-06, "loss": 0.3448, "step": 22366 }, { "epoch": 1.5462841341168336, "grad_norm": 0.7400444294799953, "learning_rate": 1.2903497230365747e-06, "loss": 0.242, "step": 22367 }, { "epoch": 1.5463532665053576, "grad_norm": 0.5618173367700019, "learning_rate": 1.2899744404399967e-06, "loss": 0.1779, "step": 22368 }, { "epoch": 1.5464223988938817, "grad_norm": 0.7876095779046547, "learning_rate": 1.2895992043419696e-06, "loss": 0.342, "step": 22369 }, { "epoch": 1.5464915312824057, "grad_norm": 0.6606169076328383, "learning_rate": 1.2892240147471925e-06, "loss": 0.2724, "step": 22370 }, { "epoch": 1.5465606636709297, "grad_norm": 0.7191171788529367, "learning_rate": 1.2888488716603731e-06, "loss": 0.2514, "step": 22371 }, { "epoch": 1.5466297960594537, "grad_norm": 0.7114231963807187, "learning_rate": 1.2884737750862086e-06, "loss": 0.2948, "step": 22372 }, { "epoch": 1.5466989284479777, "grad_norm": 0.6899188366055768, "learning_rate": 1.288098725029403e-06, "loss": 0.3465, "step": 22373 }, { "epoch": 1.5467680608365018, "grad_norm": 0.9190469437055604, "learning_rate": 1.2877237214946547e-06, "loss": 0.2621, "step": 22374 }, { "epoch": 1.5468371932250258, "grad_norm": 0.6688891723260618, "learning_rate": 1.2873487644866662e-06, "loss": 0.2233, "step": 22375 }, { "epoch": 1.5469063256135498, "grad_norm": 0.9449032984689444, "learning_rate": 1.2869738540101334e-06, "loss": 0.2847, "step": 22376 }, { "epoch": 1.5469754580020738, "grad_norm": 0.6689589597968548, "learning_rate": 1.2865989900697579e-06, "loss": 0.3061, "step": 22377 }, { "epoch": 1.5470445903905978, "grad_norm": 0.6580122271592801, "learning_rate": 1.2862241726702363e-06, "loss": 0.2111, "step": 22378 }, { "epoch": 1.547113722779122, "grad_norm": 0.6453381266132497, "learning_rate": 1.2858494018162693e-06, "loss": 0.2676, "step": 22379 }, { "epoch": 1.547182855167646, "grad_norm": 0.6664755771612602, "learning_rate": 1.2854746775125504e-06, "loss": 0.3133, "step": 22380 }, { "epoch": 1.54725198755617, "grad_norm": 0.7017095639382064, "learning_rate": 1.2850999997637786e-06, "loss": 0.2515, "step": 22381 }, { "epoch": 1.5473211199446941, "grad_norm": 0.6605596714334527, "learning_rate": 1.2847253685746476e-06, "loss": 0.2196, "step": 22382 }, { "epoch": 1.5473902523332181, "grad_norm": 0.6878499011923113, "learning_rate": 1.2843507839498554e-06, "loss": 0.2924, "step": 22383 }, { "epoch": 1.5474593847217422, "grad_norm": 0.7555153144661299, "learning_rate": 1.2839762458940935e-06, "loss": 0.3316, "step": 22384 }, { "epoch": 1.5475285171102662, "grad_norm": 1.7775826427060986, "learning_rate": 1.2836017544120577e-06, "loss": 0.1947, "step": 22385 }, { "epoch": 1.5475976494987902, "grad_norm": 0.7199285667903268, "learning_rate": 1.2832273095084418e-06, "loss": 0.2031, "step": 22386 }, { "epoch": 1.5476667818873142, "grad_norm": 0.71314750242213, "learning_rate": 1.2828529111879407e-06, "loss": 0.3214, "step": 22387 }, { "epoch": 1.5477359142758382, "grad_norm": 0.7054569199786134, "learning_rate": 1.282478559455243e-06, "loss": 0.2807, "step": 22388 }, { "epoch": 1.5478050466643622, "grad_norm": 0.7756642536867964, "learning_rate": 1.2821042543150437e-06, "loss": 0.242, "step": 22389 }, { "epoch": 1.5478741790528863, "grad_norm": 0.6814390373011648, "learning_rate": 1.2817299957720314e-06, "loss": 0.26, "step": 22390 }, { "epoch": 1.5479433114414103, "grad_norm": 0.7614618812566817, "learning_rate": 1.2813557838308977e-06, "loss": 0.3987, "step": 22391 }, { "epoch": 1.5480124438299343, "grad_norm": 0.7399255311582243, "learning_rate": 1.2809816184963337e-06, "loss": 0.3052, "step": 22392 }, { "epoch": 1.5480815762184583, "grad_norm": 0.7164730710939559, "learning_rate": 1.2806074997730288e-06, "loss": 0.2068, "step": 22393 }, { "epoch": 1.5481507086069823, "grad_norm": 0.633997102525459, "learning_rate": 1.2802334276656708e-06, "loss": 0.3364, "step": 22394 }, { "epoch": 1.5482198409955064, "grad_norm": 0.6898623154714357, "learning_rate": 1.2798594021789496e-06, "loss": 0.2674, "step": 22395 }, { "epoch": 1.5482889733840304, "grad_norm": 0.8380864462889315, "learning_rate": 1.2794854233175503e-06, "loss": 0.2844, "step": 22396 }, { "epoch": 1.5483581057725544, "grad_norm": 0.7055278724780119, "learning_rate": 1.279111491086163e-06, "loss": 0.2621, "step": 22397 }, { "epoch": 1.5484272381610784, "grad_norm": 0.7944789410012582, "learning_rate": 1.2787376054894718e-06, "loss": 0.3561, "step": 22398 }, { "epoch": 1.5484963705496027, "grad_norm": 0.646373050444656, "learning_rate": 1.2783637665321635e-06, "loss": 0.2582, "step": 22399 }, { "epoch": 1.5485655029381267, "grad_norm": 0.6523787650391288, "learning_rate": 1.2779899742189245e-06, "loss": 0.2504, "step": 22400 }, { "epoch": 1.5486346353266507, "grad_norm": 0.6875865957014501, "learning_rate": 1.2776162285544397e-06, "loss": 0.3274, "step": 22401 }, { "epoch": 1.5487037677151747, "grad_norm": 0.8636662784448393, "learning_rate": 1.2772425295433916e-06, "loss": 0.2688, "step": 22402 }, { "epoch": 1.5487729001036987, "grad_norm": 0.6998936592523556, "learning_rate": 1.2768688771904658e-06, "loss": 0.2581, "step": 22403 }, { "epoch": 1.5488420324922227, "grad_norm": 0.6697462041924789, "learning_rate": 1.2764952715003437e-06, "loss": 0.2601, "step": 22404 }, { "epoch": 1.5489111648807468, "grad_norm": 0.697609453928188, "learning_rate": 1.2761217124777091e-06, "loss": 0.288, "step": 22405 }, { "epoch": 1.5489802972692708, "grad_norm": 0.7380001940566779, "learning_rate": 1.275748200127241e-06, "loss": 0.2654, "step": 22406 }, { "epoch": 1.5490494296577948, "grad_norm": 0.6823128879938796, "learning_rate": 1.2753747344536255e-06, "loss": 0.2386, "step": 22407 }, { "epoch": 1.5491185620463188, "grad_norm": 0.6956989520275971, "learning_rate": 1.2750013154615392e-06, "loss": 0.3125, "step": 22408 }, { "epoch": 1.5491876944348428, "grad_norm": 0.7210730757256748, "learning_rate": 1.2746279431556657e-06, "loss": 0.3387, "step": 22409 }, { "epoch": 1.5492568268233669, "grad_norm": 0.6838333802957703, "learning_rate": 1.274254617540681e-06, "loss": 0.291, "step": 22410 }, { "epoch": 1.5493259592118909, "grad_norm": 0.8202412727667806, "learning_rate": 1.2738813386212678e-06, "loss": 0.2576, "step": 22411 }, { "epoch": 1.5493950916004149, "grad_norm": 0.7033212095175164, "learning_rate": 1.2735081064021004e-06, "loss": 0.3593, "step": 22412 }, { "epoch": 1.549464223988939, "grad_norm": 0.8526175468387117, "learning_rate": 1.273134920887859e-06, "loss": 0.3062, "step": 22413 }, { "epoch": 1.549533356377463, "grad_norm": 0.7958588576723342, "learning_rate": 1.27276178208322e-06, "loss": 0.1829, "step": 22414 }, { "epoch": 1.549602488765987, "grad_norm": 0.7701109299122525, "learning_rate": 1.2723886899928622e-06, "loss": 0.2431, "step": 22415 }, { "epoch": 1.549671621154511, "grad_norm": 0.8795342608163492, "learning_rate": 1.2720156446214605e-06, "loss": 0.3698, "step": 22416 }, { "epoch": 1.549740753543035, "grad_norm": 0.6825691970707247, "learning_rate": 1.2716426459736875e-06, "loss": 0.2095, "step": 22417 }, { "epoch": 1.549809885931559, "grad_norm": 0.6795297680557465, "learning_rate": 1.2712696940542224e-06, "loss": 0.265, "step": 22418 }, { "epoch": 1.549879018320083, "grad_norm": 0.6875593215505243, "learning_rate": 1.2708967888677359e-06, "loss": 0.2973, "step": 22419 }, { "epoch": 1.549948150708607, "grad_norm": 0.7109233153574849, "learning_rate": 1.2705239304189033e-06, "loss": 0.3032, "step": 22420 }, { "epoch": 1.550017283097131, "grad_norm": 0.7587055811547809, "learning_rate": 1.2701511187123978e-06, "loss": 0.23, "step": 22421 }, { "epoch": 1.550086415485655, "grad_norm": 0.6639095100143403, "learning_rate": 1.269778353752893e-06, "loss": 0.2405, "step": 22422 }, { "epoch": 1.550155547874179, "grad_norm": 0.6836084109494798, "learning_rate": 1.269405635545058e-06, "loss": 0.3372, "step": 22423 }, { "epoch": 1.550224680262703, "grad_norm": 0.7332015202616666, "learning_rate": 1.2690329640935673e-06, "loss": 0.2039, "step": 22424 }, { "epoch": 1.5502938126512271, "grad_norm": 0.6799437959250293, "learning_rate": 1.2686603394030895e-06, "loss": 0.1922, "step": 22425 }, { "epoch": 1.5503629450397511, "grad_norm": 0.6827455708348249, "learning_rate": 1.2682877614782945e-06, "loss": 0.3481, "step": 22426 }, { "epoch": 1.5504320774282752, "grad_norm": 0.6902307698826147, "learning_rate": 1.2679152303238535e-06, "loss": 0.3259, "step": 22427 }, { "epoch": 1.5505012098167992, "grad_norm": 0.6705612170129172, "learning_rate": 1.2675427459444361e-06, "loss": 0.2672, "step": 22428 }, { "epoch": 1.5505703422053232, "grad_norm": 0.6798805005946658, "learning_rate": 1.2671703083447079e-06, "loss": 0.2178, "step": 22429 }, { "epoch": 1.5506394745938472, "grad_norm": 0.6748258708388596, "learning_rate": 1.26679791752934e-06, "loss": 0.3199, "step": 22430 }, { "epoch": 1.5507086069823712, "grad_norm": 0.6604909554212759, "learning_rate": 1.2664255735029967e-06, "loss": 0.238, "step": 22431 }, { "epoch": 1.5507777393708952, "grad_norm": 0.6542764416942985, "learning_rate": 1.266053276270347e-06, "loss": 0.1901, "step": 22432 }, { "epoch": 1.5508468717594193, "grad_norm": 0.8587814841900623, "learning_rate": 1.2656810258360553e-06, "loss": 0.3005, "step": 22433 }, { "epoch": 1.5509160041479433, "grad_norm": 0.740197496752207, "learning_rate": 1.2653088222047872e-06, "loss": 0.3281, "step": 22434 }, { "epoch": 1.5509851365364673, "grad_norm": 0.7667660122203941, "learning_rate": 1.2649366653812079e-06, "loss": 0.245, "step": 22435 }, { "epoch": 1.5510542689249913, "grad_norm": 0.6601478074923053, "learning_rate": 1.2645645553699838e-06, "loss": 0.2382, "step": 22436 }, { "epoch": 1.5511234013135153, "grad_norm": 0.6687903816765179, "learning_rate": 1.2641924921757754e-06, "loss": 0.3257, "step": 22437 }, { "epoch": 1.5511925337020394, "grad_norm": 0.6488781585740563, "learning_rate": 1.2638204758032485e-06, "loss": 0.2677, "step": 22438 }, { "epoch": 1.5512616660905634, "grad_norm": 0.6586949644671508, "learning_rate": 1.2634485062570629e-06, "loss": 0.2119, "step": 22439 }, { "epoch": 1.5513307984790874, "grad_norm": 0.7620668705673929, "learning_rate": 1.263076583541884e-06, "loss": 0.3345, "step": 22440 }, { "epoch": 1.5513999308676114, "grad_norm": 0.765399595690439, "learning_rate": 1.2627047076623677e-06, "loss": 0.342, "step": 22441 }, { "epoch": 1.5514690632561354, "grad_norm": 0.6588056110549795, "learning_rate": 1.262332878623182e-06, "loss": 0.2361, "step": 22442 }, { "epoch": 1.5515381956446594, "grad_norm": 0.6467414213938073, "learning_rate": 1.2619610964289814e-06, "loss": 0.192, "step": 22443 }, { "epoch": 1.5516073280331835, "grad_norm": 0.6937361914973698, "learning_rate": 1.2615893610844292e-06, "loss": 0.3261, "step": 22444 }, { "epoch": 1.5516764604217075, "grad_norm": 0.7326396550340047, "learning_rate": 1.2612176725941822e-06, "loss": 0.3016, "step": 22445 }, { "epoch": 1.5517455928102315, "grad_norm": 0.7052857024521066, "learning_rate": 1.2608460309629e-06, "loss": 0.2593, "step": 22446 }, { "epoch": 1.5518147251987555, "grad_norm": 0.6943590188964524, "learning_rate": 1.2604744361952392e-06, "loss": 0.2807, "step": 22447 }, { "epoch": 1.5518838575872795, "grad_norm": 1.105066149042558, "learning_rate": 1.2601028882958577e-06, "loss": 0.3649, "step": 22448 }, { "epoch": 1.5519529899758036, "grad_norm": 0.7272069342420051, "learning_rate": 1.2597313872694122e-06, "loss": 0.2577, "step": 22449 }, { "epoch": 1.5520221223643276, "grad_norm": 0.704672556905106, "learning_rate": 1.2593599331205603e-06, "loss": 0.2488, "step": 22450 }, { "epoch": 1.5520912547528516, "grad_norm": 0.6156710009756825, "learning_rate": 1.2589885258539546e-06, "loss": 0.2253, "step": 22451 }, { "epoch": 1.5521603871413756, "grad_norm": 0.7507767978316177, "learning_rate": 1.2586171654742536e-06, "loss": 0.3071, "step": 22452 }, { "epoch": 1.5522295195298996, "grad_norm": 0.7327647770453343, "learning_rate": 1.258245851986108e-06, "loss": 0.2315, "step": 22453 }, { "epoch": 1.5522986519184236, "grad_norm": 0.6923446966767393, "learning_rate": 1.2578745853941743e-06, "loss": 0.2559, "step": 22454 }, { "epoch": 1.5523677843069477, "grad_norm": 0.7960685494348786, "learning_rate": 1.2575033657031033e-06, "loss": 0.3013, "step": 22455 }, { "epoch": 1.5524369166954717, "grad_norm": 0.6527167671227119, "learning_rate": 1.257132192917549e-06, "loss": 0.2553, "step": 22456 }, { "epoch": 1.5525060490839957, "grad_norm": 0.6739583919991848, "learning_rate": 1.256761067042163e-06, "loss": 0.2045, "step": 22457 }, { "epoch": 1.5525751814725197, "grad_norm": 0.7016107510146478, "learning_rate": 1.2563899880815983e-06, "loss": 0.25, "step": 22458 }, { "epoch": 1.5526443138610437, "grad_norm": 0.7694320321199981, "learning_rate": 1.2560189560405029e-06, "loss": 0.3693, "step": 22459 }, { "epoch": 1.552713446249568, "grad_norm": 0.7021711275401803, "learning_rate": 1.2556479709235302e-06, "loss": 0.2266, "step": 22460 }, { "epoch": 1.552782578638092, "grad_norm": 0.6402901128956753, "learning_rate": 1.2552770327353264e-06, "loss": 0.2098, "step": 22461 }, { "epoch": 1.552851711026616, "grad_norm": 0.7335044943613911, "learning_rate": 1.2549061414805425e-06, "loss": 0.3354, "step": 22462 }, { "epoch": 1.55292084341514, "grad_norm": 0.9938159786733309, "learning_rate": 1.2545352971638264e-06, "loss": 0.296, "step": 22463 }, { "epoch": 1.552989975803664, "grad_norm": 0.7408814684683904, "learning_rate": 1.2541644997898278e-06, "loss": 0.244, "step": 22464 }, { "epoch": 1.553059108192188, "grad_norm": 0.776370268594921, "learning_rate": 1.2537937493631908e-06, "loss": 0.3253, "step": 22465 }, { "epoch": 1.553128240580712, "grad_norm": 0.7918740621711468, "learning_rate": 1.2534230458885655e-06, "loss": 0.3182, "step": 22466 }, { "epoch": 1.553197372969236, "grad_norm": 0.6840770270772322, "learning_rate": 1.2530523893705942e-06, "loss": 0.2427, "step": 22467 }, { "epoch": 1.5532665053577601, "grad_norm": 0.7008902676435749, "learning_rate": 1.2526817798139268e-06, "loss": 0.2143, "step": 22468 }, { "epoch": 1.5533356377462841, "grad_norm": 0.7516473618625963, "learning_rate": 1.2523112172232038e-06, "loss": 0.3877, "step": 22469 }, { "epoch": 1.5534047701348082, "grad_norm": 0.6304213228093424, "learning_rate": 1.2519407016030722e-06, "loss": 0.2313, "step": 22470 }, { "epoch": 1.5534739025233322, "grad_norm": 0.622513567064508, "learning_rate": 1.2515702329581747e-06, "loss": 0.2514, "step": 22471 }, { "epoch": 1.5535430349118562, "grad_norm": 0.6738842197585468, "learning_rate": 1.2511998112931567e-06, "loss": 0.2525, "step": 22472 }, { "epoch": 1.5536121673003802, "grad_norm": 0.7112307635399311, "learning_rate": 1.2508294366126572e-06, "loss": 0.3374, "step": 22473 }, { "epoch": 1.5536812996889042, "grad_norm": 0.7675037320354623, "learning_rate": 1.250459108921322e-06, "loss": 0.3113, "step": 22474 }, { "epoch": 1.5537504320774282, "grad_norm": 0.636492637613715, "learning_rate": 1.2500888282237889e-06, "loss": 0.2549, "step": 22475 }, { "epoch": 1.5538195644659523, "grad_norm": 0.6733717813011298, "learning_rate": 1.2497185945247015e-06, "loss": 0.2599, "step": 22476 }, { "epoch": 1.5538886968544763, "grad_norm": 0.7301442999570752, "learning_rate": 1.249348407828696e-06, "loss": 0.3331, "step": 22477 }, { "epoch": 1.5539578292430003, "grad_norm": 0.62124341576294, "learning_rate": 1.2489782681404183e-06, "loss": 0.2763, "step": 22478 }, { "epoch": 1.5540269616315243, "grad_norm": 0.6065235365469268, "learning_rate": 1.2486081754645018e-06, "loss": 0.1975, "step": 22479 }, { "epoch": 1.5540960940200486, "grad_norm": 0.7938832281791702, "learning_rate": 1.248238129805589e-06, "loss": 0.3186, "step": 22480 }, { "epoch": 1.5541652264085726, "grad_norm": 0.7025007534723177, "learning_rate": 1.247868131168316e-06, "loss": 0.2219, "step": 22481 }, { "epoch": 1.5542343587970966, "grad_norm": 0.7343975678751461, "learning_rate": 1.2474981795573182e-06, "loss": 0.2515, "step": 22482 }, { "epoch": 1.5543034911856206, "grad_norm": 0.6908309996442593, "learning_rate": 1.2471282749772346e-06, "loss": 0.3141, "step": 22483 }, { "epoch": 1.5543726235741446, "grad_norm": 0.7696593200371319, "learning_rate": 1.2467584174327001e-06, "loss": 0.3235, "step": 22484 }, { "epoch": 1.5544417559626686, "grad_norm": 0.702742179360501, "learning_rate": 1.2463886069283532e-06, "loss": 0.3111, "step": 22485 }, { "epoch": 1.5545108883511927, "grad_norm": 0.6794312924925128, "learning_rate": 1.2460188434688241e-06, "loss": 0.2887, "step": 22486 }, { "epoch": 1.5545800207397167, "grad_norm": 0.726246158782775, "learning_rate": 1.2456491270587517e-06, "loss": 0.3605, "step": 22487 }, { "epoch": 1.5546491531282407, "grad_norm": 0.7265114856179947, "learning_rate": 1.245279457702766e-06, "loss": 0.2821, "step": 22488 }, { "epoch": 1.5547182855167647, "grad_norm": 0.6292871016924046, "learning_rate": 1.2449098354055028e-06, "loss": 0.2279, "step": 22489 }, { "epoch": 1.5547874179052887, "grad_norm": 0.706129931492104, "learning_rate": 1.2445402601715922e-06, "loss": 0.2533, "step": 22490 }, { "epoch": 1.5548565502938128, "grad_norm": 0.7328866743399498, "learning_rate": 1.244170732005668e-06, "loss": 0.3683, "step": 22491 }, { "epoch": 1.5549256826823368, "grad_norm": 0.6856885283768481, "learning_rate": 1.2438012509123603e-06, "loss": 0.2524, "step": 22492 }, { "epoch": 1.5549948150708608, "grad_norm": 0.6722349957148538, "learning_rate": 1.2434318168963028e-06, "loss": 0.2118, "step": 22493 }, { "epoch": 1.5550639474593848, "grad_norm": 0.7131428793596865, "learning_rate": 1.2430624299621214e-06, "loss": 0.32, "step": 22494 }, { "epoch": 1.5551330798479088, "grad_norm": 0.714745324982619, "learning_rate": 1.2426930901144501e-06, "loss": 0.3138, "step": 22495 }, { "epoch": 1.5552022122364328, "grad_norm": 0.7074765502943102, "learning_rate": 1.242323797357914e-06, "loss": 0.2532, "step": 22496 }, { "epoch": 1.5552713446249569, "grad_norm": 0.6557322549548897, "learning_rate": 1.2419545516971432e-06, "loss": 0.199, "step": 22497 }, { "epoch": 1.5553404770134809, "grad_norm": 0.7004400381389043, "learning_rate": 1.2415853531367656e-06, "loss": 0.3456, "step": 22498 }, { "epoch": 1.555409609402005, "grad_norm": 0.684622618972294, "learning_rate": 1.2412162016814095e-06, "loss": 0.2741, "step": 22499 }, { "epoch": 1.555478741790529, "grad_norm": 0.6551461968548512, "learning_rate": 1.2408470973356996e-06, "loss": 0.242, "step": 22500 }, { "epoch": 1.555547874179053, "grad_norm": 0.6616224527857234, "learning_rate": 1.2404780401042638e-06, "loss": 0.3029, "step": 22501 }, { "epoch": 1.555617006567577, "grad_norm": 0.6940027072150711, "learning_rate": 1.240109029991725e-06, "loss": 0.2741, "step": 22502 }, { "epoch": 1.555686138956101, "grad_norm": 0.7070329384524188, "learning_rate": 1.2397400670027115e-06, "loss": 0.2656, "step": 22503 }, { "epoch": 1.555755271344625, "grad_norm": 0.6697356495228401, "learning_rate": 1.2393711511418444e-06, "loss": 0.2195, "step": 22504 }, { "epoch": 1.555824403733149, "grad_norm": 0.6390847333850713, "learning_rate": 1.2390022824137488e-06, "loss": 0.3021, "step": 22505 }, { "epoch": 1.555893536121673, "grad_norm": 0.6866165195715783, "learning_rate": 1.2386334608230472e-06, "loss": 0.3087, "step": 22506 }, { "epoch": 1.555962668510197, "grad_norm": 0.6358170571631995, "learning_rate": 1.2382646863743647e-06, "loss": 0.2325, "step": 22507 }, { "epoch": 1.556031800898721, "grad_norm": 0.8044331105887806, "learning_rate": 1.23789595907232e-06, "loss": 0.3423, "step": 22508 }, { "epoch": 1.556100933287245, "grad_norm": 0.760379795286423, "learning_rate": 1.2375272789215366e-06, "loss": 0.3313, "step": 22509 }, { "epoch": 1.556170065675769, "grad_norm": 0.7640996832643903, "learning_rate": 1.2371586459266333e-06, "loss": 0.2902, "step": 22510 }, { "epoch": 1.5562391980642931, "grad_norm": 0.6930715798858169, "learning_rate": 1.2367900600922323e-06, "loss": 0.2209, "step": 22511 }, { "epoch": 1.5563083304528171, "grad_norm": 0.7162206555691981, "learning_rate": 1.2364215214229508e-06, "loss": 0.326, "step": 22512 }, { "epoch": 1.5563774628413412, "grad_norm": 0.6872809568882564, "learning_rate": 1.2360530299234092e-06, "loss": 0.2572, "step": 22513 }, { "epoch": 1.5564465952298652, "grad_norm": 0.7338718273370755, "learning_rate": 1.2356845855982259e-06, "loss": 0.2498, "step": 22514 }, { "epoch": 1.5565157276183892, "grad_norm": 0.7607355625206381, "learning_rate": 1.2353161884520199e-06, "loss": 0.311, "step": 22515 }, { "epoch": 1.5565848600069132, "grad_norm": 1.59582805829148, "learning_rate": 1.2349478384894058e-06, "loss": 0.3567, "step": 22516 }, { "epoch": 1.5566539923954372, "grad_norm": 0.6475458539688966, "learning_rate": 1.2345795357150032e-06, "loss": 0.2303, "step": 22517 }, { "epoch": 1.5567231247839612, "grad_norm": 1.1338430325759246, "learning_rate": 1.2342112801334244e-06, "loss": 0.1951, "step": 22518 }, { "epoch": 1.5567922571724853, "grad_norm": 0.6872112814241819, "learning_rate": 1.2338430717492873e-06, "loss": 0.2755, "step": 22519 }, { "epoch": 1.5568613895610093, "grad_norm": 0.7454335139305839, "learning_rate": 1.233474910567206e-06, "loss": 0.3666, "step": 22520 }, { "epoch": 1.5569305219495333, "grad_norm": 0.6522545337801351, "learning_rate": 1.2331067965917965e-06, "loss": 0.2335, "step": 22521 }, { "epoch": 1.5569996543380573, "grad_norm": 1.1802117806899854, "learning_rate": 1.2327387298276694e-06, "loss": 0.2464, "step": 22522 }, { "epoch": 1.5570687867265813, "grad_norm": 0.7400735505928422, "learning_rate": 1.2323707102794414e-06, "loss": 0.3681, "step": 22523 }, { "epoch": 1.5571379191151054, "grad_norm": 0.637575935009042, "learning_rate": 1.2320027379517208e-06, "loss": 0.1849, "step": 22524 }, { "epoch": 1.5572070515036294, "grad_norm": 0.6406899457060279, "learning_rate": 1.2316348128491235e-06, "loss": 0.2315, "step": 22525 }, { "epoch": 1.5572761838921534, "grad_norm": 0.7342819095341321, "learning_rate": 1.2312669349762574e-06, "loss": 0.2656, "step": 22526 }, { "epoch": 1.5573453162806774, "grad_norm": 0.6704962923145088, "learning_rate": 1.2308991043377343e-06, "loss": 0.3074, "step": 22527 }, { "epoch": 1.5574144486692014, "grad_norm": 0.6469297820709343, "learning_rate": 1.2305313209381648e-06, "loss": 0.2184, "step": 22528 }, { "epoch": 1.5574835810577254, "grad_norm": 0.6552188154878228, "learning_rate": 1.2301635847821597e-06, "loss": 0.2012, "step": 22529 }, { "epoch": 1.5575527134462495, "grad_norm": 0.7406527401242632, "learning_rate": 1.229795895874325e-06, "loss": 0.3652, "step": 22530 }, { "epoch": 1.5576218458347735, "grad_norm": 0.6955391541087788, "learning_rate": 1.229428254219272e-06, "loss": 0.229, "step": 22531 }, { "epoch": 1.5576909782232975, "grad_norm": 0.6092800839817795, "learning_rate": 1.229060659821606e-06, "loss": 0.214, "step": 22532 }, { "epoch": 1.5577601106118215, "grad_norm": 0.6891864445803512, "learning_rate": 1.2286931126859347e-06, "loss": 0.2598, "step": 22533 }, { "epoch": 1.5578292430003455, "grad_norm": 0.7591802181200337, "learning_rate": 1.2283256128168647e-06, "loss": 0.3572, "step": 22534 }, { "epoch": 1.5578983753888695, "grad_norm": 0.788491459101931, "learning_rate": 1.2279581602190037e-06, "loss": 0.2312, "step": 22535 }, { "epoch": 1.5579675077773936, "grad_norm": 0.6529233174279897, "learning_rate": 1.2275907548969552e-06, "loss": 0.2358, "step": 22536 }, { "epoch": 1.5580366401659176, "grad_norm": 0.6345909820238791, "learning_rate": 1.2272233968553248e-06, "loss": 0.3127, "step": 22537 }, { "epoch": 1.5581057725544416, "grad_norm": 0.69971813382587, "learning_rate": 1.2268560860987156e-06, "loss": 0.2759, "step": 22538 }, { "epoch": 1.5581749049429656, "grad_norm": 0.6390873375789913, "learning_rate": 1.2264888226317328e-06, "loss": 0.2242, "step": 22539 }, { "epoch": 1.5582440373314896, "grad_norm": 0.7598957510104106, "learning_rate": 1.2261216064589771e-06, "loss": 0.3015, "step": 22540 }, { "epoch": 1.5583131697200139, "grad_norm": 1.0772419325233726, "learning_rate": 1.2257544375850522e-06, "loss": 0.33, "step": 22541 }, { "epoch": 1.558382302108538, "grad_norm": 0.7657136553933483, "learning_rate": 1.2253873160145602e-06, "loss": 0.2527, "step": 22542 }, { "epoch": 1.558451434497062, "grad_norm": 0.6427725515882976, "learning_rate": 1.225020241752104e-06, "loss": 0.2172, "step": 22543 }, { "epoch": 1.558520566885586, "grad_norm": 0.6626121070301229, "learning_rate": 1.22465321480228e-06, "loss": 0.2704, "step": 22544 }, { "epoch": 1.55858969927411, "grad_norm": 0.6649159117749798, "learning_rate": 1.2242862351696927e-06, "loss": 0.2742, "step": 22545 }, { "epoch": 1.558658831662634, "grad_norm": 0.6805790738569761, "learning_rate": 1.2239193028589375e-06, "loss": 0.2064, "step": 22546 }, { "epoch": 1.558727964051158, "grad_norm": 0.6547391946137764, "learning_rate": 1.2235524178746167e-06, "loss": 0.2131, "step": 22547 }, { "epoch": 1.558797096439682, "grad_norm": 0.6570854298644392, "learning_rate": 1.2231855802213255e-06, "loss": 0.3576, "step": 22548 }, { "epoch": 1.558866228828206, "grad_norm": 0.6830185675259669, "learning_rate": 1.2228187899036624e-06, "loss": 0.2407, "step": 22549 }, { "epoch": 1.55893536121673, "grad_norm": 0.729100123211208, "learning_rate": 1.2224520469262274e-06, "loss": 0.2, "step": 22550 }, { "epoch": 1.559004493605254, "grad_norm": 0.7208065965010322, "learning_rate": 1.2220853512936132e-06, "loss": 0.2702, "step": 22551 }, { "epoch": 1.559073625993778, "grad_norm": 0.7330164983073933, "learning_rate": 1.2217187030104184e-06, "loss": 0.413, "step": 22552 }, { "epoch": 1.559142758382302, "grad_norm": 0.6844980971361759, "learning_rate": 1.2213521020812353e-06, "loss": 0.3019, "step": 22553 }, { "epoch": 1.5592118907708261, "grad_norm": 0.7006518693105, "learning_rate": 1.2209855485106604e-06, "loss": 0.2119, "step": 22554 }, { "epoch": 1.5592810231593501, "grad_norm": 0.6358283794137553, "learning_rate": 1.2206190423032876e-06, "loss": 0.3095, "step": 22555 }, { "epoch": 1.5593501555478742, "grad_norm": 0.7214717974811213, "learning_rate": 1.2202525834637125e-06, "loss": 0.285, "step": 22556 }, { "epoch": 1.5594192879363982, "grad_norm": 0.8133113560511006, "learning_rate": 1.2198861719965244e-06, "loss": 0.2598, "step": 22557 }, { "epoch": 1.5594884203249222, "grad_norm": 0.7064254632981981, "learning_rate": 1.2195198079063182e-06, "loss": 0.3224, "step": 22558 }, { "epoch": 1.5595575527134462, "grad_norm": 0.721080991209548, "learning_rate": 1.2191534911976832e-06, "loss": 0.2876, "step": 22559 }, { "epoch": 1.5596266851019702, "grad_norm": 0.7945599854501364, "learning_rate": 1.218787221875214e-06, "loss": 0.2179, "step": 22560 }, { "epoch": 1.5596958174904945, "grad_norm": 0.6888671431956209, "learning_rate": 1.218420999943497e-06, "loss": 0.2524, "step": 22561 }, { "epoch": 1.5597649498790185, "grad_norm": 0.7864047019748645, "learning_rate": 1.2180548254071244e-06, "loss": 0.3349, "step": 22562 }, { "epoch": 1.5598340822675425, "grad_norm": 0.7701451903717906, "learning_rate": 1.217688698270686e-06, "loss": 0.301, "step": 22563 }, { "epoch": 1.5599032146560665, "grad_norm": 0.6631257648172055, "learning_rate": 1.217322618538771e-06, "loss": 0.277, "step": 22564 }, { "epoch": 1.5599723470445905, "grad_norm": 0.6615307960177108, "learning_rate": 1.2169565862159648e-06, "loss": 0.2357, "step": 22565 }, { "epoch": 1.5600414794331146, "grad_norm": 0.7072729501554501, "learning_rate": 1.2165906013068578e-06, "loss": 0.3633, "step": 22566 }, { "epoch": 1.5601106118216386, "grad_norm": 0.6582635068629974, "learning_rate": 1.2162246638160352e-06, "loss": 0.2255, "step": 22567 }, { "epoch": 1.5601797442101626, "grad_norm": 0.725503595583287, "learning_rate": 1.2158587737480838e-06, "loss": 0.1991, "step": 22568 }, { "epoch": 1.5602488765986866, "grad_norm": 0.7035032907962436, "learning_rate": 1.2154929311075892e-06, "loss": 0.3016, "step": 22569 }, { "epoch": 1.5603180089872106, "grad_norm": 0.7728490943263026, "learning_rate": 1.2151271358991385e-06, "loss": 0.3365, "step": 22570 }, { "epoch": 1.5603871413757346, "grad_norm": 0.7552922421804171, "learning_rate": 1.2147613881273134e-06, "loss": 0.2835, "step": 22571 }, { "epoch": 1.5604562737642587, "grad_norm": 0.6118357453595752, "learning_rate": 1.2143956877967007e-06, "loss": 0.2157, "step": 22572 }, { "epoch": 1.5605254061527827, "grad_norm": 0.7455828941508322, "learning_rate": 1.2140300349118806e-06, "loss": 0.3384, "step": 22573 }, { "epoch": 1.5605945385413067, "grad_norm": 0.5936868925005402, "learning_rate": 1.2136644294774392e-06, "loss": 0.1917, "step": 22574 }, { "epoch": 1.5606636709298307, "grad_norm": 0.7147234904449302, "learning_rate": 1.213298871497956e-06, "loss": 0.222, "step": 22575 }, { "epoch": 1.5607328033183547, "grad_norm": 0.6579850064991443, "learning_rate": 1.2129333609780136e-06, "loss": 0.2825, "step": 22576 }, { "epoch": 1.5608019357068788, "grad_norm": 0.7306172534242691, "learning_rate": 1.2125678979221933e-06, "loss": 0.2969, "step": 22577 }, { "epoch": 1.5608710680954028, "grad_norm": 0.6731509870517934, "learning_rate": 1.2122024823350764e-06, "loss": 0.2427, "step": 22578 }, { "epoch": 1.5609402004839268, "grad_norm": 0.6994408511916018, "learning_rate": 1.211837114221241e-06, "loss": 0.2377, "step": 22579 }, { "epoch": 1.5610093328724508, "grad_norm": 0.6747859694738999, "learning_rate": 1.2114717935852682e-06, "loss": 0.3298, "step": 22580 }, { "epoch": 1.5610784652609748, "grad_norm": 0.9280966547186448, "learning_rate": 1.211106520431734e-06, "loss": 0.2552, "step": 22581 }, { "epoch": 1.5611475976494988, "grad_norm": 0.670082420138394, "learning_rate": 1.2107412947652197e-06, "loss": 0.2188, "step": 22582 }, { "epoch": 1.5612167300380229, "grad_norm": 0.6702009493379318, "learning_rate": 1.2103761165902994e-06, "loss": 0.2936, "step": 22583 }, { "epoch": 1.5612858624265469, "grad_norm": 0.7050724627392534, "learning_rate": 1.2100109859115517e-06, "loss": 0.3585, "step": 22584 }, { "epoch": 1.561354994815071, "grad_norm": 0.6999963161548165, "learning_rate": 1.2096459027335522e-06, "loss": 0.1882, "step": 22585 }, { "epoch": 1.561424127203595, "grad_norm": 0.6887831468950166, "learning_rate": 1.2092808670608792e-06, "loss": 0.241, "step": 22586 }, { "epoch": 1.561493259592119, "grad_norm": 0.6732322848502804, "learning_rate": 1.2089158788981041e-06, "loss": 0.3475, "step": 22587 }, { "epoch": 1.561562391980643, "grad_norm": 0.764033682323307, "learning_rate": 1.2085509382498045e-06, "loss": 0.3242, "step": 22588 }, { "epoch": 1.561631524369167, "grad_norm": 0.7490035306223977, "learning_rate": 1.2081860451205512e-06, "loss": 0.259, "step": 22589 }, { "epoch": 1.561700656757691, "grad_norm": 0.7636935830254344, "learning_rate": 1.2078211995149192e-06, "loss": 0.2442, "step": 22590 }, { "epoch": 1.561769789146215, "grad_norm": 0.7234366889674596, "learning_rate": 1.2074564014374807e-06, "loss": 0.3894, "step": 22591 }, { "epoch": 1.561838921534739, "grad_norm": 0.7538900774700623, "learning_rate": 1.2070916508928094e-06, "loss": 0.2701, "step": 22592 }, { "epoch": 1.561908053923263, "grad_norm": 0.6591011774046331, "learning_rate": 1.2067269478854743e-06, "loss": 0.2405, "step": 22593 }, { "epoch": 1.561977186311787, "grad_norm": 0.8228317898936119, "learning_rate": 1.2063622924200492e-06, "loss": 0.3231, "step": 22594 }, { "epoch": 1.562046318700311, "grad_norm": 0.7181590868189743, "learning_rate": 1.2059976845011011e-06, "loss": 0.2671, "step": 22595 }, { "epoch": 1.562115451088835, "grad_norm": 0.6334970047964119, "learning_rate": 1.2056331241332025e-06, "loss": 0.2329, "step": 22596 }, { "epoch": 1.5621845834773591, "grad_norm": 0.669859632836726, "learning_rate": 1.2052686113209205e-06, "loss": 0.2457, "step": 22597 }, { "epoch": 1.5622537158658831, "grad_norm": 0.6909633408991264, "learning_rate": 1.2049041460688239e-06, "loss": 0.3619, "step": 22598 }, { "epoch": 1.5623228482544071, "grad_norm": 0.70657736840721, "learning_rate": 1.2045397283814814e-06, "loss": 0.2369, "step": 22599 }, { "epoch": 1.5623919806429312, "grad_norm": 0.6290881695756316, "learning_rate": 1.2041753582634614e-06, "loss": 0.2442, "step": 22600 }, { "epoch": 1.5624611130314552, "grad_norm": 0.6732068144253955, "learning_rate": 1.2038110357193283e-06, "loss": 0.3387, "step": 22601 }, { "epoch": 1.5625302454199792, "grad_norm": 0.7111546325819165, "learning_rate": 1.2034467607536503e-06, "loss": 0.3643, "step": 22602 }, { "epoch": 1.5625993778085032, "grad_norm": 0.699842064941652, "learning_rate": 1.2030825333709905e-06, "loss": 0.2469, "step": 22603 }, { "epoch": 1.5626685101970272, "grad_norm": 0.6606104799236837, "learning_rate": 1.2027183535759152e-06, "loss": 0.2173, "step": 22604 }, { "epoch": 1.5627376425855513, "grad_norm": 0.7376606305518182, "learning_rate": 1.2023542213729888e-06, "loss": 0.3524, "step": 22605 }, { "epoch": 1.5628067749740753, "grad_norm": 0.7402066275490873, "learning_rate": 1.2019901367667769e-06, "loss": 0.2776, "step": 22606 }, { "epoch": 1.5628759073625993, "grad_norm": 0.630001216547165, "learning_rate": 1.201626099761839e-06, "loss": 0.1616, "step": 22607 }, { "epoch": 1.5629450397511233, "grad_norm": 0.6667875254575805, "learning_rate": 1.2012621103627408e-06, "loss": 0.2288, "step": 22608 }, { "epoch": 1.5630141721396473, "grad_norm": 0.7189889362220255, "learning_rate": 1.2008981685740411e-06, "loss": 0.3461, "step": 22609 }, { "epoch": 1.5630833045281713, "grad_norm": 0.6742229868144736, "learning_rate": 1.200534274400305e-06, "loss": 0.177, "step": 22610 }, { "epoch": 1.5631524369166954, "grad_norm": 0.6314415288627432, "learning_rate": 1.2001704278460901e-06, "loss": 0.2162, "step": 22611 }, { "epoch": 1.5632215693052194, "grad_norm": 0.7141566609226335, "learning_rate": 1.1998066289159578e-06, "loss": 0.3075, "step": 22612 }, { "epoch": 1.5632907016937434, "grad_norm": 2.734170358892497, "learning_rate": 1.1994428776144668e-06, "loss": 0.2582, "step": 22613 }, { "epoch": 1.5633598340822674, "grad_norm": 0.6522202798764967, "learning_rate": 1.1990791739461792e-06, "loss": 0.2186, "step": 22614 }, { "epoch": 1.5634289664707914, "grad_norm": 0.680965158121057, "learning_rate": 1.1987155179156506e-06, "loss": 0.2327, "step": 22615 }, { "epoch": 1.5634980988593155, "grad_norm": 0.7802850188747448, "learning_rate": 1.1983519095274383e-06, "loss": 0.3737, "step": 22616 }, { "epoch": 1.5635672312478395, "grad_norm": 0.6868290816966381, "learning_rate": 1.1979883487861015e-06, "loss": 0.2412, "step": 22617 }, { "epoch": 1.5636363636363635, "grad_norm": 0.6646769172316964, "learning_rate": 1.1976248356961946e-06, "loss": 0.2813, "step": 22618 }, { "epoch": 1.5637054960248875, "grad_norm": 0.7156642229632996, "learning_rate": 1.1972613702622749e-06, "loss": 0.2996, "step": 22619 }, { "epoch": 1.5637746284134115, "grad_norm": 0.739052918662552, "learning_rate": 1.1968979524888967e-06, "loss": 0.3406, "step": 22620 }, { "epoch": 1.5638437608019355, "grad_norm": 0.6508426789264005, "learning_rate": 1.1965345823806185e-06, "loss": 0.228, "step": 22621 }, { "epoch": 1.5639128931904598, "grad_norm": 0.9281016790037809, "learning_rate": 1.1961712599419895e-06, "loss": 0.2693, "step": 22622 }, { "epoch": 1.5639820255789838, "grad_norm": 0.7679291329668724, "learning_rate": 1.1958079851775666e-06, "loss": 0.3857, "step": 22623 }, { "epoch": 1.5640511579675078, "grad_norm": 0.6680330399588031, "learning_rate": 1.195444758091901e-06, "loss": 0.2202, "step": 22624 }, { "epoch": 1.5641202903560318, "grad_norm": 0.6842781867709171, "learning_rate": 1.195081578689546e-06, "loss": 0.2535, "step": 22625 }, { "epoch": 1.5641894227445559, "grad_norm": 0.6535804548044207, "learning_rate": 1.1947184469750522e-06, "loss": 0.2708, "step": 22626 }, { "epoch": 1.5642585551330799, "grad_norm": 0.7276321268596344, "learning_rate": 1.1943553629529741e-06, "loss": 0.2843, "step": 22627 }, { "epoch": 1.564327687521604, "grad_norm": 0.6512314082135837, "learning_rate": 1.1939923266278585e-06, "loss": 0.2185, "step": 22628 }, { "epoch": 1.564396819910128, "grad_norm": 0.6936133320132445, "learning_rate": 1.193629338004258e-06, "loss": 0.2985, "step": 22629 }, { "epoch": 1.564465952298652, "grad_norm": 0.708105436854526, "learning_rate": 1.1932663970867197e-06, "loss": 0.3919, "step": 22630 }, { "epoch": 1.564535084687176, "grad_norm": 0.6909842065817506, "learning_rate": 1.1929035038797953e-06, "loss": 0.2462, "step": 22631 }, { "epoch": 1.5646042170757, "grad_norm": 0.6927745044708762, "learning_rate": 1.19254065838803e-06, "loss": 0.2739, "step": 22632 }, { "epoch": 1.564673349464224, "grad_norm": 0.7226393067236384, "learning_rate": 1.1921778606159729e-06, "loss": 0.2746, "step": 22633 }, { "epoch": 1.564742481852748, "grad_norm": 0.7035583986704628, "learning_rate": 1.1918151105681708e-06, "loss": 0.3225, "step": 22634 }, { "epoch": 1.564811614241272, "grad_norm": 0.6525330492957416, "learning_rate": 1.1914524082491718e-06, "loss": 0.2261, "step": 22635 }, { "epoch": 1.564880746629796, "grad_norm": 0.7334481553464222, "learning_rate": 1.1910897536635185e-06, "loss": 0.2124, "step": 22636 }, { "epoch": 1.56494987901832, "grad_norm": 0.6461800612420421, "learning_rate": 1.1907271468157593e-06, "loss": 0.3209, "step": 22637 }, { "epoch": 1.565019011406844, "grad_norm": 0.6663383110336574, "learning_rate": 1.1903645877104364e-06, "loss": 0.2934, "step": 22638 }, { "epoch": 1.565088143795368, "grad_norm": 0.6511615354939733, "learning_rate": 1.1900020763520948e-06, "loss": 0.1777, "step": 22639 }, { "epoch": 1.565157276183892, "grad_norm": 0.7968633131570485, "learning_rate": 1.1896396127452775e-06, "loss": 0.2956, "step": 22640 }, { "epoch": 1.5652264085724161, "grad_norm": 0.6774515624975819, "learning_rate": 1.1892771968945293e-06, "loss": 0.3938, "step": 22641 }, { "epoch": 1.5652955409609401, "grad_norm": 0.6646492083580696, "learning_rate": 1.1889148288043895e-06, "loss": 0.1869, "step": 22642 }, { "epoch": 1.5653646733494644, "grad_norm": 0.7142727861809278, "learning_rate": 1.1885525084794025e-06, "loss": 0.2501, "step": 22643 }, { "epoch": 1.5654338057379884, "grad_norm": 0.6724719148609102, "learning_rate": 1.1881902359241065e-06, "loss": 0.3146, "step": 22644 }, { "epoch": 1.5655029381265124, "grad_norm": 0.7622616874432401, "learning_rate": 1.1878280111430445e-06, "loss": 0.2907, "step": 22645 }, { "epoch": 1.5655720705150364, "grad_norm": 0.6368903866821773, "learning_rate": 1.1874658341407546e-06, "loss": 0.2647, "step": 22646 }, { "epoch": 1.5656412029035605, "grad_norm": 0.7346427729976011, "learning_rate": 1.1871037049217765e-06, "loss": 0.2162, "step": 22647 }, { "epoch": 1.5657103352920845, "grad_norm": 0.7489042071001818, "learning_rate": 1.1867416234906486e-06, "loss": 0.376, "step": 22648 }, { "epoch": 1.5657794676806085, "grad_norm": 0.6532553728349074, "learning_rate": 1.186379589851911e-06, "loss": 0.2288, "step": 22649 }, { "epoch": 1.5658486000691325, "grad_norm": 0.6745791509378962, "learning_rate": 1.1860176040100984e-06, "loss": 0.2592, "step": 22650 }, { "epoch": 1.5659177324576565, "grad_norm": 0.7172808822786934, "learning_rate": 1.18565566596975e-06, "loss": 0.2901, "step": 22651 }, { "epoch": 1.5659868648461805, "grad_norm": 0.7024633358672279, "learning_rate": 1.1852937757353994e-06, "loss": 0.3039, "step": 22652 }, { "epoch": 1.5660559972347046, "grad_norm": 1.4716768712432726, "learning_rate": 1.1849319333115849e-06, "loss": 0.208, "step": 22653 }, { "epoch": 1.5661251296232286, "grad_norm": 0.7078092997763593, "learning_rate": 1.184570138702839e-06, "loss": 0.2325, "step": 22654 }, { "epoch": 1.5661942620117526, "grad_norm": 0.7039700834229671, "learning_rate": 1.1842083919136977e-06, "loss": 0.3202, "step": 22655 }, { "epoch": 1.5662633944002766, "grad_norm": 0.6411316301529603, "learning_rate": 1.1838466929486947e-06, "loss": 0.2287, "step": 22656 }, { "epoch": 1.5663325267888006, "grad_norm": 0.6484442545031462, "learning_rate": 1.1834850418123646e-06, "loss": 0.2316, "step": 22657 }, { "epoch": 1.5664016591773247, "grad_norm": 0.675925052873248, "learning_rate": 1.1831234385092367e-06, "loss": 0.2519, "step": 22658 }, { "epoch": 1.5664707915658487, "grad_norm": 0.7694160755085092, "learning_rate": 1.1827618830438465e-06, "loss": 0.3086, "step": 22659 }, { "epoch": 1.5665399239543727, "grad_norm": 0.6239630748258738, "learning_rate": 1.1824003754207224e-06, "loss": 0.2047, "step": 22660 }, { "epoch": 1.5666090563428967, "grad_norm": 0.6989963792224447, "learning_rate": 1.1820389156443974e-06, "loss": 0.2487, "step": 22661 }, { "epoch": 1.5666781887314207, "grad_norm": 1.08452742334642, "learning_rate": 1.1816775037194006e-06, "loss": 0.3548, "step": 22662 }, { "epoch": 1.5667473211199447, "grad_norm": 0.7070206259945728, "learning_rate": 1.1813161396502642e-06, "loss": 0.3042, "step": 22663 }, { "epoch": 1.5668164535084688, "grad_norm": 0.6278589221489174, "learning_rate": 1.180954823441513e-06, "loss": 0.2133, "step": 22664 }, { "epoch": 1.5668855858969928, "grad_norm": 0.6903305596318595, "learning_rate": 1.1805935550976798e-06, "loss": 0.2274, "step": 22665 }, { "epoch": 1.5669547182855168, "grad_norm": 0.6438460056204043, "learning_rate": 1.1802323346232885e-06, "loss": 0.3394, "step": 22666 }, { "epoch": 1.5670238506740408, "grad_norm": 0.7170362365755593, "learning_rate": 1.1798711620228698e-06, "loss": 0.2246, "step": 22667 }, { "epoch": 1.5670929830625648, "grad_norm": 0.742052138941341, "learning_rate": 1.1795100373009476e-06, "loss": 0.2282, "step": 22668 }, { "epoch": 1.5671621154510889, "grad_norm": 0.7313587585240252, "learning_rate": 1.1791489604620486e-06, "loss": 0.3697, "step": 22669 }, { "epoch": 1.5672312478396129, "grad_norm": 0.7325326520901297, "learning_rate": 1.1787879315106992e-06, "loss": 0.2581, "step": 22670 }, { "epoch": 1.567300380228137, "grad_norm": 0.6789071940515345, "learning_rate": 1.178426950451425e-06, "loss": 0.2355, "step": 22671 }, { "epoch": 1.567369512616661, "grad_norm": 0.7555481855691264, "learning_rate": 1.1780660172887477e-06, "loss": 0.263, "step": 22672 }, { "epoch": 1.567438645005185, "grad_norm": 0.7462661152442818, "learning_rate": 1.1777051320271937e-06, "loss": 0.3466, "step": 22673 }, { "epoch": 1.567507777393709, "grad_norm": 0.6581882442765947, "learning_rate": 1.1773442946712832e-06, "loss": 0.2301, "step": 22674 }, { "epoch": 1.567576909782233, "grad_norm": 0.6248188546081301, "learning_rate": 1.1769835052255402e-06, "loss": 0.2157, "step": 22675 }, { "epoch": 1.567646042170757, "grad_norm": 0.7359188169341785, "learning_rate": 1.1766227636944865e-06, "loss": 0.3147, "step": 22676 }, { "epoch": 1.567715174559281, "grad_norm": 0.7120982142893111, "learning_rate": 1.176262070082645e-06, "loss": 0.3337, "step": 22677 }, { "epoch": 1.567784306947805, "grad_norm": 0.8378590138573813, "learning_rate": 1.1759014243945332e-06, "loss": 0.2474, "step": 22678 }, { "epoch": 1.567853439336329, "grad_norm": 0.6150882422792064, "learning_rate": 1.1755408266346734e-06, "loss": 0.2201, "step": 22679 }, { "epoch": 1.567922571724853, "grad_norm": 0.6694672541669451, "learning_rate": 1.1751802768075837e-06, "loss": 0.3226, "step": 22680 }, { "epoch": 1.567991704113377, "grad_norm": 0.694063048905644, "learning_rate": 1.1748197749177847e-06, "loss": 0.3249, "step": 22681 }, { "epoch": 1.568060836501901, "grad_norm": 0.6921881688077747, "learning_rate": 1.174459320969792e-06, "loss": 0.2227, "step": 22682 }, { "epoch": 1.568129968890425, "grad_norm": 0.7215954603302772, "learning_rate": 1.1740989149681248e-06, "loss": 0.2769, "step": 22683 }, { "epoch": 1.5681991012789491, "grad_norm": 0.7872719738589868, "learning_rate": 1.173738556917301e-06, "loss": 0.3666, "step": 22684 }, { "epoch": 1.5682682336674731, "grad_norm": 0.7803559271567387, "learning_rate": 1.1733782468218351e-06, "loss": 0.2641, "step": 22685 }, { "epoch": 1.5683373660559972, "grad_norm": 0.7193888570883844, "learning_rate": 1.1730179846862456e-06, "loss": 0.2605, "step": 22686 }, { "epoch": 1.5684064984445212, "grad_norm": 0.7326641339018283, "learning_rate": 1.1726577705150444e-06, "loss": 0.3303, "step": 22687 }, { "epoch": 1.5684756308330452, "grad_norm": 0.6792256566371219, "learning_rate": 1.1722976043127493e-06, "loss": 0.2945, "step": 22688 }, { "epoch": 1.5685447632215692, "grad_norm": 0.6514399813956115, "learning_rate": 1.1719374860838717e-06, "loss": 0.2508, "step": 22689 }, { "epoch": 1.5686138956100932, "grad_norm": 0.7413606847378267, "learning_rate": 1.1715774158329257e-06, "loss": 0.2381, "step": 22690 }, { "epoch": 1.5686830279986173, "grad_norm": 0.7209631248770073, "learning_rate": 1.171217393564425e-06, "loss": 0.3206, "step": 22691 }, { "epoch": 1.5687521603871413, "grad_norm": 0.7240613554529506, "learning_rate": 1.170857419282883e-06, "loss": 0.2783, "step": 22692 }, { "epoch": 1.5688212927756653, "grad_norm": 0.6497504498127733, "learning_rate": 1.1704974929928087e-06, "loss": 0.2398, "step": 22693 }, { "epoch": 1.5688904251641893, "grad_norm": 0.6330822093747626, "learning_rate": 1.1701376146987148e-06, "loss": 0.2898, "step": 22694 }, { "epoch": 1.5689595575527133, "grad_norm": 0.721255834980596, "learning_rate": 1.1697777844051105e-06, "loss": 0.2873, "step": 22695 }, { "epoch": 1.5690286899412373, "grad_norm": 0.682088278320918, "learning_rate": 1.1694180021165063e-06, "loss": 0.2051, "step": 22696 }, { "epoch": 1.5690978223297614, "grad_norm": 0.7837509629340362, "learning_rate": 1.1690582678374114e-06, "loss": 0.2841, "step": 22697 }, { "epoch": 1.5691669547182854, "grad_norm": 0.6764340077748353, "learning_rate": 1.1686985815723357e-06, "loss": 0.3187, "step": 22698 }, { "epoch": 1.5692360871068094, "grad_norm": 0.6189262080955191, "learning_rate": 1.1683389433257848e-06, "loss": 0.2376, "step": 22699 }, { "epoch": 1.5693052194953334, "grad_norm": 0.6674205118954242, "learning_rate": 1.1679793531022687e-06, "loss": 0.2785, "step": 22700 }, { "epoch": 1.5693743518838574, "grad_norm": 0.7027225949766132, "learning_rate": 1.1676198109062915e-06, "loss": 0.258, "step": 22701 }, { "epoch": 1.5694434842723815, "grad_norm": 0.7121951055102069, "learning_rate": 1.1672603167423618e-06, "loss": 0.3358, "step": 22702 }, { "epoch": 1.5695126166609055, "grad_norm": 0.7704229747062292, "learning_rate": 1.1669008706149832e-06, "loss": 0.24, "step": 22703 }, { "epoch": 1.5695817490494297, "grad_norm": 0.6696387604394987, "learning_rate": 1.1665414725286617e-06, "loss": 0.2063, "step": 22704 }, { "epoch": 1.5696508814379537, "grad_norm": 0.659579055523488, "learning_rate": 1.1661821224879017e-06, "loss": 0.3631, "step": 22705 }, { "epoch": 1.5697200138264777, "grad_norm": 0.7380303188488667, "learning_rate": 1.1658228204972082e-06, "loss": 0.2837, "step": 22706 }, { "epoch": 1.5697891462150018, "grad_norm": 0.6711752915858868, "learning_rate": 1.165463566561082e-06, "loss": 0.2199, "step": 22707 }, { "epoch": 1.5698582786035258, "grad_norm": 0.6769449504388285, "learning_rate": 1.1651043606840285e-06, "loss": 0.2857, "step": 22708 }, { "epoch": 1.5699274109920498, "grad_norm": 0.6807432085109417, "learning_rate": 1.1647452028705463e-06, "loss": 0.3002, "step": 22709 }, { "epoch": 1.5699965433805738, "grad_norm": 0.615968402337954, "learning_rate": 1.1643860931251395e-06, "loss": 0.2057, "step": 22710 }, { "epoch": 1.5700656757690978, "grad_norm": 0.6664066871596233, "learning_rate": 1.1640270314523072e-06, "loss": 0.2296, "step": 22711 }, { "epoch": 1.5701348081576219, "grad_norm": 0.8070702822530574, "learning_rate": 1.1636680178565524e-06, "loss": 0.3764, "step": 22712 }, { "epoch": 1.5702039405461459, "grad_norm": 0.7043515544042992, "learning_rate": 1.1633090523423713e-06, "loss": 0.2786, "step": 22713 }, { "epoch": 1.57027307293467, "grad_norm": 0.7051933205366729, "learning_rate": 1.1629501349142652e-06, "loss": 0.2215, "step": 22714 }, { "epoch": 1.570342205323194, "grad_norm": 0.6443254121372841, "learning_rate": 1.1625912655767308e-06, "loss": 0.2365, "step": 22715 }, { "epoch": 1.570411337711718, "grad_norm": 0.6766096136646385, "learning_rate": 1.162232444334268e-06, "loss": 0.331, "step": 22716 }, { "epoch": 1.570480470100242, "grad_norm": 0.6906143891478417, "learning_rate": 1.1618736711913714e-06, "loss": 0.2454, "step": 22717 }, { "epoch": 1.570549602488766, "grad_norm": 0.6051178698815527, "learning_rate": 1.1615149461525388e-06, "loss": 0.2153, "step": 22718 }, { "epoch": 1.57061873487729, "grad_norm": 0.6713485899647174, "learning_rate": 1.1611562692222662e-06, "loss": 0.3156, "step": 22719 }, { "epoch": 1.570687867265814, "grad_norm": 0.6809990274635754, "learning_rate": 1.1607976404050502e-06, "loss": 0.2621, "step": 22720 }, { "epoch": 1.570756999654338, "grad_norm": 0.8433167933258957, "learning_rate": 1.1604390597053833e-06, "loss": 0.2057, "step": 22721 }, { "epoch": 1.570826132042862, "grad_norm": 0.6407706815804671, "learning_rate": 1.1600805271277627e-06, "loss": 0.2377, "step": 22722 }, { "epoch": 1.570895264431386, "grad_norm": 0.6593543554574375, "learning_rate": 1.1597220426766775e-06, "loss": 0.3367, "step": 22723 }, { "epoch": 1.5709643968199103, "grad_norm": 0.7485636299056204, "learning_rate": 1.1593636063566254e-06, "loss": 0.2189, "step": 22724 }, { "epoch": 1.5710335292084343, "grad_norm": 0.6965395440790392, "learning_rate": 1.1590052181720952e-06, "loss": 0.263, "step": 22725 }, { "epoch": 1.5711026615969583, "grad_norm": 0.6843050100392993, "learning_rate": 1.1586468781275801e-06, "loss": 0.2674, "step": 22726 }, { "epoch": 1.5711717939854823, "grad_norm": 0.7879460058689549, "learning_rate": 1.1582885862275705e-06, "loss": 0.3183, "step": 22727 }, { "epoch": 1.5712409263740064, "grad_norm": 0.669980282047869, "learning_rate": 1.1579303424765597e-06, "loss": 0.2212, "step": 22728 }, { "epoch": 1.5713100587625304, "grad_norm": 0.6647442945547354, "learning_rate": 1.1575721468790335e-06, "loss": 0.2594, "step": 22729 }, { "epoch": 1.5713791911510544, "grad_norm": 0.7171895659869345, "learning_rate": 1.1572139994394853e-06, "loss": 0.323, "step": 22730 }, { "epoch": 1.5714483235395784, "grad_norm": 1.0628510945331993, "learning_rate": 1.1568559001624001e-06, "loss": 0.2759, "step": 22731 }, { "epoch": 1.5715174559281024, "grad_norm": 0.7013679013530856, "learning_rate": 1.1564978490522683e-06, "loss": 0.2511, "step": 22732 }, { "epoch": 1.5715865883166265, "grad_norm": 0.6770005246558329, "learning_rate": 1.1561398461135765e-06, "loss": 0.2345, "step": 22733 }, { "epoch": 1.5716557207051505, "grad_norm": 0.733161773145416, "learning_rate": 1.155781891350814e-06, "loss": 0.3418, "step": 22734 }, { "epoch": 1.5717248530936745, "grad_norm": 0.6960427355462131, "learning_rate": 1.1554239847684628e-06, "loss": 0.2475, "step": 22735 }, { "epoch": 1.5717939854821985, "grad_norm": 0.6123279698985052, "learning_rate": 1.1550661263710133e-06, "loss": 0.1764, "step": 22736 }, { "epoch": 1.5718631178707225, "grad_norm": 0.7006637491756557, "learning_rate": 1.1547083161629463e-06, "loss": 0.3064, "step": 22737 }, { "epoch": 1.5719322502592465, "grad_norm": 0.710836351472096, "learning_rate": 1.1543505541487503e-06, "loss": 0.34, "step": 22738 }, { "epoch": 1.5720013826477706, "grad_norm": 0.767499328715953, "learning_rate": 1.1539928403329053e-06, "loss": 0.236, "step": 22739 }, { "epoch": 1.5720705150362946, "grad_norm": 0.6497780978001594, "learning_rate": 1.153635174719897e-06, "loss": 0.1758, "step": 22740 }, { "epoch": 1.5721396474248186, "grad_norm": 0.7109889564723582, "learning_rate": 1.1532775573142073e-06, "loss": 0.393, "step": 22741 }, { "epoch": 1.5722087798133426, "grad_norm": 0.6395380629325202, "learning_rate": 1.15291998812032e-06, "loss": 0.2552, "step": 22742 }, { "epoch": 1.5722779122018666, "grad_norm": 0.6681095342322014, "learning_rate": 1.1525624671427137e-06, "loss": 0.206, "step": 22743 }, { "epoch": 1.5723470445903907, "grad_norm": 0.6932035009662837, "learning_rate": 1.1522049943858727e-06, "loss": 0.3195, "step": 22744 }, { "epoch": 1.5724161769789147, "grad_norm": 0.7553340905240925, "learning_rate": 1.1518475698542736e-06, "loss": 0.3263, "step": 22745 }, { "epoch": 1.5724853093674387, "grad_norm": 0.8218592579331537, "learning_rate": 1.151490193552398e-06, "loss": 0.259, "step": 22746 }, { "epoch": 1.5725544417559627, "grad_norm": 0.8164371702180636, "learning_rate": 1.1511328654847264e-06, "loss": 0.2656, "step": 22747 }, { "epoch": 1.5726235741444867, "grad_norm": 0.8500004425707494, "learning_rate": 1.1507755856557345e-06, "loss": 0.3548, "step": 22748 }, { "epoch": 1.5726927065330107, "grad_norm": 0.7185100501881099, "learning_rate": 1.1504183540699021e-06, "loss": 0.2577, "step": 22749 }, { "epoch": 1.5727618389215348, "grad_norm": 0.628593424761038, "learning_rate": 1.150061170731705e-06, "loss": 0.2643, "step": 22750 }, { "epoch": 1.5728309713100588, "grad_norm": 0.6556681596065538, "learning_rate": 1.1497040356456218e-06, "loss": 0.2912, "step": 22751 }, { "epoch": 1.5729001036985828, "grad_norm": 0.741171497982859, "learning_rate": 1.1493469488161258e-06, "loss": 0.3189, "step": 22752 }, { "epoch": 1.5729692360871068, "grad_norm": 0.7898842362185752, "learning_rate": 1.1489899102476942e-06, "loss": 0.226, "step": 22753 }, { "epoch": 1.5730383684756308, "grad_norm": 0.7086934474469767, "learning_rate": 1.1486329199448015e-06, "loss": 0.2087, "step": 22754 }, { "epoch": 1.5731075008641549, "grad_norm": 0.7095458216381629, "learning_rate": 1.1482759779119234e-06, "loss": 0.3214, "step": 22755 }, { "epoch": 1.5731766332526789, "grad_norm": 0.7559972166444953, "learning_rate": 1.147919084153531e-06, "loss": 0.2803, "step": 22756 }, { "epoch": 1.5732457656412029, "grad_norm": 0.8547353678462585, "learning_rate": 1.1475622386740998e-06, "loss": 0.1959, "step": 22757 }, { "epoch": 1.573314898029727, "grad_norm": 0.6718458128830099, "learning_rate": 1.1472054414780999e-06, "loss": 0.3051, "step": 22758 }, { "epoch": 1.573384030418251, "grad_norm": 0.7008935004759534, "learning_rate": 1.146848692570005e-06, "loss": 0.3245, "step": 22759 }, { "epoch": 1.573453162806775, "grad_norm": 0.6507693737086212, "learning_rate": 1.1464919919542845e-06, "loss": 0.2564, "step": 22760 }, { "epoch": 1.573522295195299, "grad_norm": 0.69917074445029, "learning_rate": 1.1461353396354097e-06, "loss": 0.2252, "step": 22761 }, { "epoch": 1.573591427583823, "grad_norm": 0.7011994207191434, "learning_rate": 1.1457787356178508e-06, "loss": 0.3487, "step": 22762 }, { "epoch": 1.573660559972347, "grad_norm": 0.7276849114226808, "learning_rate": 1.145422179906079e-06, "loss": 0.285, "step": 22763 }, { "epoch": 1.573729692360871, "grad_norm": 0.6554727556805523, "learning_rate": 1.14506567250456e-06, "loss": 0.2565, "step": 22764 }, { "epoch": 1.573798824749395, "grad_norm": 0.7105512300208443, "learning_rate": 1.1447092134177646e-06, "loss": 0.2187, "step": 22765 }, { "epoch": 1.573867957137919, "grad_norm": 0.6656452884731487, "learning_rate": 1.1443528026501576e-06, "loss": 0.3089, "step": 22766 }, { "epoch": 1.573937089526443, "grad_norm": 0.7473276885642195, "learning_rate": 1.1439964402062076e-06, "loss": 0.2679, "step": 22767 }, { "epoch": 1.574006221914967, "grad_norm": 0.6656598250467066, "learning_rate": 1.1436401260903812e-06, "loss": 0.2468, "step": 22768 }, { "epoch": 1.574075354303491, "grad_norm": 0.6984990013217188, "learning_rate": 1.143283860307145e-06, "loss": 0.3231, "step": 22769 }, { "epoch": 1.5741444866920151, "grad_norm": 0.7424905423735686, "learning_rate": 1.1429276428609616e-06, "loss": 0.3069, "step": 22770 }, { "epoch": 1.5742136190805391, "grad_norm": 0.6526907204668354, "learning_rate": 1.1425714737562982e-06, "loss": 0.2181, "step": 22771 }, { "epoch": 1.5742827514690632, "grad_norm": 0.6132199857442826, "learning_rate": 1.142215352997616e-06, "loss": 0.2543, "step": 22772 }, { "epoch": 1.5743518838575872, "grad_norm": 0.6688325724372731, "learning_rate": 1.141859280589382e-06, "loss": 0.3304, "step": 22773 }, { "epoch": 1.5744210162461112, "grad_norm": 0.6485514585663287, "learning_rate": 1.1415032565360545e-06, "loss": 0.2385, "step": 22774 }, { "epoch": 1.5744901486346352, "grad_norm": 0.6450192850507388, "learning_rate": 1.141147280842098e-06, "loss": 0.262, "step": 22775 }, { "epoch": 1.5745592810231592, "grad_norm": 0.7342358496822489, "learning_rate": 1.140791353511974e-06, "loss": 0.2946, "step": 22776 }, { "epoch": 1.5746284134116832, "grad_norm": 0.8508144300911595, "learning_rate": 1.140435474550145e-06, "loss": 0.3369, "step": 22777 }, { "epoch": 1.5746975458002073, "grad_norm": 0.7177296155506558, "learning_rate": 1.1400796439610678e-06, "loss": 0.2333, "step": 22778 }, { "epoch": 1.5747666781887313, "grad_norm": 0.6854273976220393, "learning_rate": 1.1397238617492056e-06, "loss": 0.2763, "step": 22779 }, { "epoch": 1.5748358105772553, "grad_norm": 1.1386884321621846, "learning_rate": 1.1393681279190144e-06, "loss": 0.4245, "step": 22780 }, { "epoch": 1.5749049429657793, "grad_norm": 0.6795056836843896, "learning_rate": 1.1390124424749543e-06, "loss": 0.3048, "step": 22781 }, { "epoch": 1.5749740753543033, "grad_norm": 0.700009750947217, "learning_rate": 1.1386568054214824e-06, "loss": 0.2089, "step": 22782 }, { "epoch": 1.5750432077428274, "grad_norm": 0.6882795663270475, "learning_rate": 1.1383012167630587e-06, "loss": 0.2622, "step": 22783 }, { "epoch": 1.5751123401313514, "grad_norm": 0.790662970646638, "learning_rate": 1.1379456765041358e-06, "loss": 0.3516, "step": 22784 }, { "epoch": 1.5751814725198756, "grad_norm": 1.0450855800093513, "learning_rate": 1.1375901846491733e-06, "loss": 0.2976, "step": 22785 }, { "epoch": 1.5752506049083996, "grad_norm": 0.7386278600620269, "learning_rate": 1.1372347412026236e-06, "loss": 0.2517, "step": 22786 }, { "epoch": 1.5753197372969237, "grad_norm": 0.7492166773969651, "learning_rate": 1.1368793461689442e-06, "loss": 0.3492, "step": 22787 }, { "epoch": 1.5753888696854477, "grad_norm": 0.7012300187819043, "learning_rate": 1.1365239995525873e-06, "loss": 0.2805, "step": 22788 }, { "epoch": 1.5754580020739717, "grad_norm": 0.6863746691044734, "learning_rate": 1.136168701358007e-06, "loss": 0.2279, "step": 22789 }, { "epoch": 1.5755271344624957, "grad_norm": 0.6723699564069415, "learning_rate": 1.1358134515896569e-06, "loss": 0.2827, "step": 22790 }, { "epoch": 1.5755962668510197, "grad_norm": 0.7011985906021573, "learning_rate": 1.1354582502519901e-06, "loss": 0.3254, "step": 22791 }, { "epoch": 1.5756653992395437, "grad_norm": 0.6976394612656852, "learning_rate": 1.1351030973494564e-06, "loss": 0.2742, "step": 22792 }, { "epoch": 1.5757345316280678, "grad_norm": 0.6849553813188656, "learning_rate": 1.1347479928865101e-06, "loss": 0.2709, "step": 22793 }, { "epoch": 1.5758036640165918, "grad_norm": 0.7121873938531825, "learning_rate": 1.1343929368675982e-06, "loss": 0.3256, "step": 22794 }, { "epoch": 1.5758727964051158, "grad_norm": 0.7220543841178805, "learning_rate": 1.1340379292971732e-06, "loss": 0.2911, "step": 22795 }, { "epoch": 1.5759419287936398, "grad_norm": 0.6745803550710298, "learning_rate": 1.1336829701796831e-06, "loss": 0.2478, "step": 22796 }, { "epoch": 1.5760110611821638, "grad_norm": 0.6189679401274685, "learning_rate": 1.1333280595195766e-06, "loss": 0.2439, "step": 22797 }, { "epoch": 1.5760801935706878, "grad_norm": 0.6612690660250226, "learning_rate": 1.1329731973213025e-06, "loss": 0.3225, "step": 22798 }, { "epoch": 1.5761493259592119, "grad_norm": 0.6979709041242803, "learning_rate": 1.13261838358931e-06, "loss": 0.2389, "step": 22799 }, { "epoch": 1.5762184583477359, "grad_norm": 0.6467066314960596, "learning_rate": 1.1322636183280428e-06, "loss": 0.2176, "step": 22800 }, { "epoch": 1.57628759073626, "grad_norm": 0.706008250474938, "learning_rate": 1.1319089015419504e-06, "loss": 0.2657, "step": 22801 }, { "epoch": 1.576356723124784, "grad_norm": 0.6951934983012543, "learning_rate": 1.1315542332354756e-06, "loss": 0.3348, "step": 22802 }, { "epoch": 1.576425855513308, "grad_norm": 0.651965431119496, "learning_rate": 1.1311996134130648e-06, "loss": 0.2138, "step": 22803 }, { "epoch": 1.576494987901832, "grad_norm": 0.6722471306238076, "learning_rate": 1.1308450420791628e-06, "loss": 0.2599, "step": 22804 }, { "epoch": 1.5765641202903562, "grad_norm": 0.8430433369562231, "learning_rate": 1.130490519238215e-06, "loss": 0.3324, "step": 22805 }, { "epoch": 1.5766332526788802, "grad_norm": 0.6724417189800377, "learning_rate": 1.1301360448946613e-06, "loss": 0.2602, "step": 22806 }, { "epoch": 1.5767023850674042, "grad_norm": 0.6463687240421007, "learning_rate": 1.1297816190529475e-06, "loss": 0.2274, "step": 22807 }, { "epoch": 1.5767715174559283, "grad_norm": 0.6802189379112851, "learning_rate": 1.1294272417175133e-06, "loss": 0.2623, "step": 22808 }, { "epoch": 1.5768406498444523, "grad_norm": 0.7215587321572587, "learning_rate": 1.129072912892803e-06, "loss": 0.3354, "step": 22809 }, { "epoch": 1.5769097822329763, "grad_norm": 0.673025162884545, "learning_rate": 1.1287186325832539e-06, "loss": 0.2126, "step": 22810 }, { "epoch": 1.5769789146215003, "grad_norm": 0.752112432999815, "learning_rate": 1.128364400793308e-06, "loss": 0.2261, "step": 22811 }, { "epoch": 1.5770480470100243, "grad_norm": 0.8042840528697024, "learning_rate": 1.1280102175274054e-06, "loss": 0.3561, "step": 22812 }, { "epoch": 1.5771171793985483, "grad_norm": 0.8878412456741308, "learning_rate": 1.1276560827899863e-06, "loss": 0.3104, "step": 22813 }, { "epoch": 1.5771863117870724, "grad_norm": 0.6710517544738057, "learning_rate": 1.127301996585487e-06, "loss": 0.1843, "step": 22814 }, { "epoch": 1.5772554441755964, "grad_norm": 0.7020714887753944, "learning_rate": 1.1269479589183451e-06, "loss": 0.2387, "step": 22815 }, { "epoch": 1.5773245765641204, "grad_norm": 0.659183596041901, "learning_rate": 1.1265939697930007e-06, "loss": 0.3419, "step": 22816 }, { "epoch": 1.5773937089526444, "grad_norm": 0.6928515624278005, "learning_rate": 1.1262400292138853e-06, "loss": 0.2725, "step": 22817 }, { "epoch": 1.5774628413411684, "grad_norm": 0.62043343308523, "learning_rate": 1.1258861371854408e-06, "loss": 0.2094, "step": 22818 }, { "epoch": 1.5775319737296924, "grad_norm": 0.6208650009825547, "learning_rate": 1.1255322937120983e-06, "loss": 0.2765, "step": 22819 }, { "epoch": 1.5776011061182165, "grad_norm": 0.7772028181270563, "learning_rate": 1.1251784987982955e-06, "loss": 0.3228, "step": 22820 }, { "epoch": 1.5776702385067405, "grad_norm": 0.6502812234495006, "learning_rate": 1.1248247524484646e-06, "loss": 0.2352, "step": 22821 }, { "epoch": 1.5777393708952645, "grad_norm": 0.6238136347156765, "learning_rate": 1.1244710546670406e-06, "loss": 0.245, "step": 22822 }, { "epoch": 1.5778085032837885, "grad_norm": 0.7886848848919193, "learning_rate": 1.1241174054584548e-06, "loss": 0.3692, "step": 22823 }, { "epoch": 1.5778776356723125, "grad_norm": 0.7129124381230213, "learning_rate": 1.1237638048271405e-06, "loss": 0.2539, "step": 22824 }, { "epoch": 1.5779467680608366, "grad_norm": 0.6781871500229353, "learning_rate": 1.1234102527775293e-06, "loss": 0.232, "step": 22825 }, { "epoch": 1.5780159004493606, "grad_norm": 0.7294393524136451, "learning_rate": 1.1230567493140538e-06, "loss": 0.3126, "step": 22826 }, { "epoch": 1.5780850328378846, "grad_norm": 0.7614741665951601, "learning_rate": 1.1227032944411415e-06, "loss": 0.3006, "step": 22827 }, { "epoch": 1.5781541652264086, "grad_norm": 0.6808005763081851, "learning_rate": 1.1223498881632262e-06, "loss": 0.2079, "step": 22828 }, { "epoch": 1.5782232976149326, "grad_norm": 0.8973426402783502, "learning_rate": 1.121996530484733e-06, "loss": 0.2252, "step": 22829 }, { "epoch": 1.5782924300034566, "grad_norm": 0.7957604161653343, "learning_rate": 1.1216432214100948e-06, "loss": 0.3255, "step": 22830 }, { "epoch": 1.5783615623919807, "grad_norm": 0.6686192210713172, "learning_rate": 1.1212899609437355e-06, "loss": 0.2699, "step": 22831 }, { "epoch": 1.5784306947805047, "grad_norm": 0.6742066785791809, "learning_rate": 1.1209367490900846e-06, "loss": 0.2461, "step": 22832 }, { "epoch": 1.5784998271690287, "grad_norm": 0.6881813397147588, "learning_rate": 1.1205835858535697e-06, "loss": 0.3148, "step": 22833 }, { "epoch": 1.5785689595575527, "grad_norm": 0.8001303809335842, "learning_rate": 1.1202304712386174e-06, "loss": 0.2842, "step": 22834 }, { "epoch": 1.5786380919460767, "grad_norm": 0.6770815499709358, "learning_rate": 1.1198774052496515e-06, "loss": 0.2251, "step": 22835 }, { "epoch": 1.5787072243346008, "grad_norm": 0.6654746819267596, "learning_rate": 1.1195243878910989e-06, "loss": 0.2398, "step": 22836 }, { "epoch": 1.5787763567231248, "grad_norm": 0.7307418410377369, "learning_rate": 1.1191714191673813e-06, "loss": 0.349, "step": 22837 }, { "epoch": 1.5788454891116488, "grad_norm": 0.7614362625690965, "learning_rate": 1.1188184990829248e-06, "loss": 0.2634, "step": 22838 }, { "epoch": 1.5789146215001728, "grad_norm": 0.719890847267573, "learning_rate": 1.1184656276421523e-06, "loss": 0.2472, "step": 22839 }, { "epoch": 1.5789837538886968, "grad_norm": 0.646082617882713, "learning_rate": 1.1181128048494872e-06, "loss": 0.2084, "step": 22840 }, { "epoch": 1.5790528862772208, "grad_norm": 0.7087153583970449, "learning_rate": 1.1177600307093494e-06, "loss": 0.3823, "step": 22841 }, { "epoch": 1.5791220186657449, "grad_norm": 1.8459721791100028, "learning_rate": 1.1174073052261631e-06, "loss": 0.2819, "step": 22842 }, { "epoch": 1.5791911510542689, "grad_norm": 0.9414697252915979, "learning_rate": 1.1170546284043454e-06, "loss": 0.2566, "step": 22843 }, { "epoch": 1.579260283442793, "grad_norm": 0.6987749766476699, "learning_rate": 1.1167020002483203e-06, "loss": 0.3184, "step": 22844 }, { "epoch": 1.579329415831317, "grad_norm": 0.6919680946438691, "learning_rate": 1.1163494207625043e-06, "loss": 0.3383, "step": 22845 }, { "epoch": 1.579398548219841, "grad_norm": 0.6339898720020329, "learning_rate": 1.1159968899513173e-06, "loss": 0.2574, "step": 22846 }, { "epoch": 1.579467680608365, "grad_norm": 0.6679877558550328, "learning_rate": 1.115644407819178e-06, "loss": 0.227, "step": 22847 }, { "epoch": 1.579536812996889, "grad_norm": 0.746073634588287, "learning_rate": 1.115291974370506e-06, "loss": 0.3554, "step": 22848 }, { "epoch": 1.579605945385413, "grad_norm": 0.903463790748959, "learning_rate": 1.1149395896097143e-06, "loss": 0.2378, "step": 22849 }, { "epoch": 1.579675077773937, "grad_norm": 0.6046345494575499, "learning_rate": 1.114587253541224e-06, "loss": 0.2106, "step": 22850 }, { "epoch": 1.579744210162461, "grad_norm": 0.6903859279040024, "learning_rate": 1.1142349661694462e-06, "loss": 0.2671, "step": 22851 }, { "epoch": 1.579813342550985, "grad_norm": 0.6890056067219481, "learning_rate": 1.1138827274988007e-06, "loss": 0.3565, "step": 22852 }, { "epoch": 1.579882474939509, "grad_norm": 0.662852820723629, "learning_rate": 1.1135305375336974e-06, "loss": 0.2374, "step": 22853 }, { "epoch": 1.579951607328033, "grad_norm": 0.659292560972561, "learning_rate": 1.1131783962785552e-06, "loss": 0.2324, "step": 22854 }, { "epoch": 1.580020739716557, "grad_norm": 0.6592765811840451, "learning_rate": 1.1128263037377852e-06, "loss": 0.2929, "step": 22855 }, { "epoch": 1.5800898721050811, "grad_norm": 0.7648399468785084, "learning_rate": 1.112474259915801e-06, "loss": 0.2457, "step": 22856 }, { "epoch": 1.5801590044936051, "grad_norm": 0.6394269446025979, "learning_rate": 1.1121222648170132e-06, "loss": 0.1823, "step": 22857 }, { "epoch": 1.5802281368821292, "grad_norm": 0.712837207811583, "learning_rate": 1.1117703184458362e-06, "loss": 0.2002, "step": 22858 }, { "epoch": 1.5802972692706532, "grad_norm": 0.7680175510203934, "learning_rate": 1.1114184208066775e-06, "loss": 0.3558, "step": 22859 }, { "epoch": 1.5803664016591772, "grad_norm": 0.7356253686866223, "learning_rate": 1.1110665719039498e-06, "loss": 0.2291, "step": 22860 }, { "epoch": 1.5804355340477012, "grad_norm": 0.768707990670477, "learning_rate": 1.1107147717420624e-06, "loss": 0.2413, "step": 22861 }, { "epoch": 1.5805046664362252, "grad_norm": 0.6920901147962353, "learning_rate": 1.1103630203254261e-06, "loss": 0.2927, "step": 22862 }, { "epoch": 1.5805737988247492, "grad_norm": 0.780618739171816, "learning_rate": 1.1100113176584464e-06, "loss": 0.3022, "step": 22863 }, { "epoch": 1.5806429312132733, "grad_norm": 0.7467747779638692, "learning_rate": 1.1096596637455343e-06, "loss": 0.2976, "step": 22864 }, { "epoch": 1.5807120636017973, "grad_norm": 0.7753902166649195, "learning_rate": 1.1093080585910942e-06, "loss": 0.2743, "step": 22865 }, { "epoch": 1.5807811959903215, "grad_norm": 0.7130864349953603, "learning_rate": 1.1089565021995358e-06, "loss": 0.3469, "step": 22866 }, { "epoch": 1.5808503283788455, "grad_norm": 0.7017685260611064, "learning_rate": 1.1086049945752624e-06, "loss": 0.2942, "step": 22867 }, { "epoch": 1.5809194607673696, "grad_norm": 0.6085675150558, "learning_rate": 1.1082535357226808e-06, "loss": 0.2412, "step": 22868 }, { "epoch": 1.5809885931558936, "grad_norm": 0.6662741761168142, "learning_rate": 1.1079021256461963e-06, "loss": 0.3412, "step": 22869 }, { "epoch": 1.5810577255444176, "grad_norm": 0.7289316536283005, "learning_rate": 1.1075507643502141e-06, "loss": 0.3499, "step": 22870 }, { "epoch": 1.5811268579329416, "grad_norm": 0.6826368190270142, "learning_rate": 1.1071994518391354e-06, "loss": 0.2936, "step": 22871 }, { "epoch": 1.5811959903214656, "grad_norm": 0.5984688479980979, "learning_rate": 1.1068481881173653e-06, "loss": 0.1902, "step": 22872 }, { "epoch": 1.5812651227099896, "grad_norm": 0.6934730782643916, "learning_rate": 1.106496973189305e-06, "loss": 0.3997, "step": 22873 }, { "epoch": 1.5813342550985137, "grad_norm": 0.758624448501218, "learning_rate": 1.1061458070593566e-06, "loss": 0.2674, "step": 22874 }, { "epoch": 1.5814033874870377, "grad_norm": 0.6957681445174777, "learning_rate": 1.1057946897319216e-06, "loss": 0.227, "step": 22875 }, { "epoch": 1.5814725198755617, "grad_norm": 0.6705034388942613, "learning_rate": 1.1054436212114016e-06, "loss": 0.243, "step": 22876 }, { "epoch": 1.5815416522640857, "grad_norm": 0.713040855413274, "learning_rate": 1.1050926015021946e-06, "loss": 0.3272, "step": 22877 }, { "epoch": 1.5816107846526097, "grad_norm": 0.7020961976548754, "learning_rate": 1.1047416306087027e-06, "loss": 0.237, "step": 22878 }, { "epoch": 1.5816799170411338, "grad_norm": 0.6157793292684395, "learning_rate": 1.1043907085353211e-06, "loss": 0.2221, "step": 22879 }, { "epoch": 1.5817490494296578, "grad_norm": 0.7227943627577783, "learning_rate": 1.1040398352864518e-06, "loss": 0.3616, "step": 22880 }, { "epoch": 1.5818181818181818, "grad_norm": 0.6916879867126382, "learning_rate": 1.1036890108664894e-06, "loss": 0.297, "step": 22881 }, { "epoch": 1.5818873142067058, "grad_norm": 0.6580899463044363, "learning_rate": 1.1033382352798317e-06, "loss": 0.1935, "step": 22882 }, { "epoch": 1.5819564465952298, "grad_norm": 2.855562031460511, "learning_rate": 1.102987508530876e-06, "loss": 0.2593, "step": 22883 }, { "epoch": 1.5820255789837538, "grad_norm": 0.8288540553208849, "learning_rate": 1.102636830624017e-06, "loss": 0.2891, "step": 22884 }, { "epoch": 1.5820947113722779, "grad_norm": 0.6578546153052488, "learning_rate": 1.102286201563651e-06, "loss": 0.2393, "step": 22885 }, { "epoch": 1.582163843760802, "grad_norm": 0.6665418942549363, "learning_rate": 1.1019356213541704e-06, "loss": 0.2088, "step": 22886 }, { "epoch": 1.5822329761493261, "grad_norm": 0.7353878300856607, "learning_rate": 1.101585089999972e-06, "loss": 0.3509, "step": 22887 }, { "epoch": 1.5823021085378501, "grad_norm": 0.6887804137490104, "learning_rate": 1.1012346075054443e-06, "loss": 0.2666, "step": 22888 }, { "epoch": 1.5823712409263742, "grad_norm": 0.6306050549803883, "learning_rate": 1.1008841738749859e-06, "loss": 0.205, "step": 22889 }, { "epoch": 1.5824403733148982, "grad_norm": 0.6878727229127293, "learning_rate": 1.1005337891129846e-06, "loss": 0.2621, "step": 22890 }, { "epoch": 1.5825095057034222, "grad_norm": 0.7293693117597254, "learning_rate": 1.1001834532238353e-06, "loss": 0.345, "step": 22891 }, { "epoch": 1.5825786380919462, "grad_norm": 0.9665517988836558, "learning_rate": 1.099833166211925e-06, "loss": 0.27, "step": 22892 }, { "epoch": 1.5826477704804702, "grad_norm": 0.6050211421185572, "learning_rate": 1.0994829280816478e-06, "loss": 0.2098, "step": 22893 }, { "epoch": 1.5827169028689942, "grad_norm": 0.7679507306771667, "learning_rate": 1.099132738837389e-06, "loss": 0.3103, "step": 22894 }, { "epoch": 1.5827860352575183, "grad_norm": 0.6936569039892768, "learning_rate": 1.0987825984835403e-06, "loss": 0.3382, "step": 22895 }, { "epoch": 1.5828551676460423, "grad_norm": 0.6684919351255124, "learning_rate": 1.09843250702449e-06, "loss": 0.2509, "step": 22896 }, { "epoch": 1.5829243000345663, "grad_norm": 0.726596174907506, "learning_rate": 1.098082464464627e-06, "loss": 0.2935, "step": 22897 }, { "epoch": 1.5829934324230903, "grad_norm": 0.7917334779351886, "learning_rate": 1.0977324708083353e-06, "loss": 0.3562, "step": 22898 }, { "epoch": 1.5830625648116143, "grad_norm": 0.700355796706111, "learning_rate": 1.0973825260600047e-06, "loss": 0.2462, "step": 22899 }, { "epoch": 1.5831316972001384, "grad_norm": 0.8111812422013166, "learning_rate": 1.097032630224018e-06, "loss": 0.2279, "step": 22900 }, { "epoch": 1.5832008295886624, "grad_norm": 0.7431087055813218, "learning_rate": 1.0966827833047639e-06, "loss": 0.3486, "step": 22901 }, { "epoch": 1.5832699619771864, "grad_norm": 0.6792612136511575, "learning_rate": 1.0963329853066235e-06, "loss": 0.3035, "step": 22902 }, { "epoch": 1.5833390943657104, "grad_norm": 0.6978843848518059, "learning_rate": 1.0959832362339828e-06, "loss": 0.2791, "step": 22903 }, { "epoch": 1.5834082267542344, "grad_norm": 0.644815434389932, "learning_rate": 1.0956335360912257e-06, "loss": 0.1921, "step": 22904 }, { "epoch": 1.5834773591427584, "grad_norm": 0.7153421953633503, "learning_rate": 1.0952838848827357e-06, "loss": 0.2963, "step": 22905 }, { "epoch": 1.5835464915312825, "grad_norm": 0.695057180303142, "learning_rate": 1.0949342826128922e-06, "loss": 0.268, "step": 22906 }, { "epoch": 1.5836156239198065, "grad_norm": 0.6789016313198907, "learning_rate": 1.0945847292860801e-06, "loss": 0.2223, "step": 22907 }, { "epoch": 1.5836847563083305, "grad_norm": 1.034444892416539, "learning_rate": 1.094235224906678e-06, "loss": 0.2089, "step": 22908 }, { "epoch": 1.5837538886968545, "grad_norm": 0.6999781889831199, "learning_rate": 1.0938857694790668e-06, "loss": 0.3812, "step": 22909 }, { "epoch": 1.5838230210853785, "grad_norm": 0.628194354672739, "learning_rate": 1.0935363630076261e-06, "loss": 0.2017, "step": 22910 }, { "epoch": 1.5838921534739026, "grad_norm": 0.6989221435295819, "learning_rate": 1.093187005496738e-06, "loss": 0.2384, "step": 22911 }, { "epoch": 1.5839612858624266, "grad_norm": 0.7036657145135603, "learning_rate": 1.092837696950777e-06, "loss": 0.3002, "step": 22912 }, { "epoch": 1.5840304182509506, "grad_norm": 0.6569191453654246, "learning_rate": 1.0924884373741245e-06, "loss": 0.2594, "step": 22913 }, { "epoch": 1.5840995506394746, "grad_norm": 0.7122607804759892, "learning_rate": 1.0921392267711546e-06, "loss": 0.2522, "step": 22914 }, { "epoch": 1.5841686830279986, "grad_norm": 0.6644045831576786, "learning_rate": 1.0917900651462465e-06, "loss": 0.2493, "step": 22915 }, { "epoch": 1.5842378154165226, "grad_norm": 0.7194054127255011, "learning_rate": 1.0914409525037745e-06, "loss": 0.3203, "step": 22916 }, { "epoch": 1.5843069478050467, "grad_norm": 0.6512583034243047, "learning_rate": 1.0910918888481153e-06, "loss": 0.2224, "step": 22917 }, { "epoch": 1.5843760801935707, "grad_norm": 0.7108489495481433, "learning_rate": 1.090742874183643e-06, "loss": 0.2807, "step": 22918 }, { "epoch": 1.5844452125820947, "grad_norm": 0.7041582606904495, "learning_rate": 1.0903939085147342e-06, "loss": 0.3529, "step": 22919 }, { "epoch": 1.5845143449706187, "grad_norm": 0.7136148444314435, "learning_rate": 1.090044991845759e-06, "loss": 0.2926, "step": 22920 }, { "epoch": 1.5845834773591427, "grad_norm": 0.6513843858556482, "learning_rate": 1.0896961241810938e-06, "loss": 0.2026, "step": 22921 }, { "epoch": 1.5846526097476668, "grad_norm": 0.733873374612485, "learning_rate": 1.089347305525108e-06, "loss": 0.2327, "step": 22922 }, { "epoch": 1.5847217421361908, "grad_norm": 0.8899388893051732, "learning_rate": 1.0889985358821765e-06, "loss": 0.3591, "step": 22923 }, { "epoch": 1.5847908745247148, "grad_norm": 0.6439779825488892, "learning_rate": 1.0886498152566654e-06, "loss": 0.2026, "step": 22924 }, { "epoch": 1.5848600069132388, "grad_norm": 0.6120363488641922, "learning_rate": 1.0883011436529523e-06, "loss": 0.2373, "step": 22925 }, { "epoch": 1.5849291393017628, "grad_norm": 0.7062123626799561, "learning_rate": 1.0879525210754016e-06, "loss": 0.3032, "step": 22926 }, { "epoch": 1.5849982716902868, "grad_norm": 0.6620802742408948, "learning_rate": 1.0876039475283861e-06, "loss": 0.2954, "step": 22927 }, { "epoch": 1.5850674040788109, "grad_norm": 0.627149666166628, "learning_rate": 1.0872554230162725e-06, "loss": 0.2433, "step": 22928 }, { "epoch": 1.5851365364673349, "grad_norm": 0.644485174795775, "learning_rate": 1.0869069475434301e-06, "loss": 0.2462, "step": 22929 }, { "epoch": 1.585205668855859, "grad_norm": 0.6880816570222431, "learning_rate": 1.0865585211142248e-06, "loss": 0.3593, "step": 22930 }, { "epoch": 1.585274801244383, "grad_norm": 1.0376119321183284, "learning_rate": 1.0862101437330248e-06, "loss": 0.2554, "step": 22931 }, { "epoch": 1.585343933632907, "grad_norm": 0.6372900171258962, "learning_rate": 1.0858618154041962e-06, "loss": 0.2459, "step": 22932 }, { "epoch": 1.585413066021431, "grad_norm": 0.6724126086309139, "learning_rate": 1.0855135361321056e-06, "loss": 0.3098, "step": 22933 }, { "epoch": 1.585482198409955, "grad_norm": 0.7117390521558573, "learning_rate": 1.085165305921116e-06, "loss": 0.3382, "step": 22934 }, { "epoch": 1.585551330798479, "grad_norm": 0.6527235541941852, "learning_rate": 1.0848171247755944e-06, "loss": 0.2788, "step": 22935 }, { "epoch": 1.585620463187003, "grad_norm": 1.326811615776703, "learning_rate": 1.0844689926999013e-06, "loss": 0.2187, "step": 22936 }, { "epoch": 1.585689595575527, "grad_norm": 0.63813878283662, "learning_rate": 1.084120909698404e-06, "loss": 0.3281, "step": 22937 }, { "epoch": 1.585758727964051, "grad_norm": 0.8195556299362826, "learning_rate": 1.0837728757754613e-06, "loss": 0.327, "step": 22938 }, { "epoch": 1.585827860352575, "grad_norm": 0.7127295317485401, "learning_rate": 1.083424890935436e-06, "loss": 0.2673, "step": 22939 }, { "epoch": 1.585896992741099, "grad_norm": 0.692357093816056, "learning_rate": 1.0830769551826908e-06, "loss": 0.2703, "step": 22940 }, { "epoch": 1.585966125129623, "grad_norm": 0.6629867221732877, "learning_rate": 1.0827290685215874e-06, "loss": 0.3964, "step": 22941 }, { "epoch": 1.5860352575181471, "grad_norm": 0.7160087683929955, "learning_rate": 1.082381230956483e-06, "loss": 0.2259, "step": 22942 }, { "epoch": 1.5861043899066711, "grad_norm": 0.6356244795651285, "learning_rate": 1.08203344249174e-06, "loss": 0.214, "step": 22943 }, { "epoch": 1.5861735222951951, "grad_norm": 0.6913415032333103, "learning_rate": 1.0816857031317145e-06, "loss": 0.2736, "step": 22944 }, { "epoch": 1.5862426546837192, "grad_norm": 0.7178987906710983, "learning_rate": 1.081338012880766e-06, "loss": 0.2869, "step": 22945 }, { "epoch": 1.5863117870722432, "grad_norm": 0.9536888936418642, "learning_rate": 1.0809903717432524e-06, "loss": 0.2257, "step": 22946 }, { "epoch": 1.5863809194607674, "grad_norm": 0.6726521295358726, "learning_rate": 1.0806427797235319e-06, "loss": 0.2922, "step": 22947 }, { "epoch": 1.5864500518492914, "grad_norm": 0.7955273735315006, "learning_rate": 1.08029523682596e-06, "loss": 0.3353, "step": 22948 }, { "epoch": 1.5865191842378155, "grad_norm": 0.7789181226582778, "learning_rate": 1.079947743054891e-06, "loss": 0.297, "step": 22949 }, { "epoch": 1.5865883166263395, "grad_norm": 0.758447751415097, "learning_rate": 1.0796002984146826e-06, "loss": 0.2901, "step": 22950 }, { "epoch": 1.5866574490148635, "grad_norm": 0.6864554091901357, "learning_rate": 1.0792529029096872e-06, "loss": 0.2224, "step": 22951 }, { "epoch": 1.5867265814033875, "grad_norm": 0.891480394161186, "learning_rate": 1.078905556544259e-06, "loss": 0.2826, "step": 22952 }, { "epoch": 1.5867957137919115, "grad_norm": 0.7297993683134391, "learning_rate": 1.0785582593227528e-06, "loss": 0.2402, "step": 22953 }, { "epoch": 1.5868648461804356, "grad_norm": 1.4724884131452376, "learning_rate": 1.0782110112495214e-06, "loss": 0.2774, "step": 22954 }, { "epoch": 1.5869339785689596, "grad_norm": 0.6892214551028502, "learning_rate": 1.0778638123289153e-06, "loss": 0.3304, "step": 22955 }, { "epoch": 1.5870031109574836, "grad_norm": 0.7592619684853051, "learning_rate": 1.0775166625652883e-06, "loss": 0.2477, "step": 22956 }, { "epoch": 1.5870722433460076, "grad_norm": 0.7213135103435567, "learning_rate": 1.077169561962989e-06, "loss": 0.2349, "step": 22957 }, { "epoch": 1.5871413757345316, "grad_norm": 0.8625680407089202, "learning_rate": 1.0768225105263697e-06, "loss": 0.2973, "step": 22958 }, { "epoch": 1.5872105081230556, "grad_norm": 0.7051974131898571, "learning_rate": 1.0764755082597762e-06, "loss": 0.3321, "step": 22959 }, { "epoch": 1.5872796405115797, "grad_norm": 0.6434076957330691, "learning_rate": 1.0761285551675627e-06, "loss": 0.2149, "step": 22960 }, { "epoch": 1.5873487729001037, "grad_norm": 0.6231612178577314, "learning_rate": 1.0757816512540743e-06, "loss": 0.2336, "step": 22961 }, { "epoch": 1.5874179052886277, "grad_norm": 0.7665818160098941, "learning_rate": 1.075434796523661e-06, "loss": 0.3202, "step": 22962 }, { "epoch": 1.5874870376771517, "grad_norm": 0.762302882572409, "learning_rate": 1.0750879909806678e-06, "loss": 0.3544, "step": 22963 }, { "epoch": 1.5875561700656757, "grad_norm": 0.7610083865577082, "learning_rate": 1.0747412346294434e-06, "loss": 0.219, "step": 22964 }, { "epoch": 1.5876253024541997, "grad_norm": 0.6434810373462322, "learning_rate": 1.0743945274743317e-06, "loss": 0.208, "step": 22965 }, { "epoch": 1.5876944348427238, "grad_norm": 0.6850342255168099, "learning_rate": 1.0740478695196781e-06, "loss": 0.3501, "step": 22966 }, { "epoch": 1.5877635672312478, "grad_norm": 0.6696117391880854, "learning_rate": 1.0737012607698294e-06, "loss": 0.2238, "step": 22967 }, { "epoch": 1.587832699619772, "grad_norm": 0.6723089491789348, "learning_rate": 1.0733547012291295e-06, "loss": 0.2164, "step": 22968 }, { "epoch": 1.587901832008296, "grad_norm": 0.7329404010011104, "learning_rate": 1.0730081909019196e-06, "loss": 0.3399, "step": 22969 }, { "epoch": 1.58797096439682, "grad_norm": 0.6872832360790961, "learning_rate": 1.0726617297925451e-06, "loss": 0.2918, "step": 22970 }, { "epoch": 1.588040096785344, "grad_norm": 0.685271821366559, "learning_rate": 1.0723153179053464e-06, "loss": 0.2549, "step": 22971 }, { "epoch": 1.588109229173868, "grad_norm": 0.7076788304494467, "learning_rate": 1.0719689552446672e-06, "loss": 0.2628, "step": 22972 }, { "epoch": 1.5881783615623921, "grad_norm": 1.3785991676074933, "learning_rate": 1.0716226418148456e-06, "loss": 0.3629, "step": 22973 }, { "epoch": 1.5882474939509161, "grad_norm": 0.7506026265116791, "learning_rate": 1.0712763776202233e-06, "loss": 0.2337, "step": 22974 }, { "epoch": 1.5883166263394402, "grad_norm": 0.6370813134862812, "learning_rate": 1.0709301626651414e-06, "loss": 0.2359, "step": 22975 }, { "epoch": 1.5883857587279642, "grad_norm": 0.7990035809091515, "learning_rate": 1.0705839969539388e-06, "loss": 0.3353, "step": 22976 }, { "epoch": 1.5884548911164882, "grad_norm": 1.8637009003566025, "learning_rate": 1.0702378804909525e-06, "loss": 0.3667, "step": 22977 }, { "epoch": 1.5885240235050122, "grad_norm": 0.791563804275054, "learning_rate": 1.0698918132805225e-06, "loss": 0.2346, "step": 22978 }, { "epoch": 1.5885931558935362, "grad_norm": 0.7192466692109863, "learning_rate": 1.0695457953269838e-06, "loss": 0.2299, "step": 22979 }, { "epoch": 1.5886622882820602, "grad_norm": 0.7300891592818322, "learning_rate": 1.0691998266346742e-06, "loss": 0.3333, "step": 22980 }, { "epoch": 1.5887314206705843, "grad_norm": 0.713324756750965, "learning_rate": 1.06885390720793e-06, "loss": 0.2972, "step": 22981 }, { "epoch": 1.5888005530591083, "grad_norm": 0.5980092292145058, "learning_rate": 1.0685080370510886e-06, "loss": 0.2477, "step": 22982 }, { "epoch": 1.5888696854476323, "grad_norm": 0.6652467432163647, "learning_rate": 1.0681622161684807e-06, "loss": 0.3131, "step": 22983 }, { "epoch": 1.5889388178361563, "grad_norm": 0.6873125832480456, "learning_rate": 1.0678164445644447e-06, "loss": 0.3206, "step": 22984 }, { "epoch": 1.5890079502246803, "grad_norm": 0.6238527390876253, "learning_rate": 1.0674707222433107e-06, "loss": 0.1971, "step": 22985 }, { "epoch": 1.5890770826132044, "grad_norm": 0.678884342907944, "learning_rate": 1.0671250492094143e-06, "loss": 0.2787, "step": 22986 }, { "epoch": 1.5891462150017284, "grad_norm": 0.8174826892893923, "learning_rate": 1.0667794254670854e-06, "loss": 0.3302, "step": 22987 }, { "epoch": 1.5892153473902524, "grad_norm": 0.6871750276082913, "learning_rate": 1.0664338510206578e-06, "loss": 0.263, "step": 22988 }, { "epoch": 1.5892844797787764, "grad_norm": 0.60156911568078, "learning_rate": 1.066088325874462e-06, "loss": 0.2211, "step": 22989 }, { "epoch": 1.5893536121673004, "grad_norm": 0.6590840643606335, "learning_rate": 1.0657428500328298e-06, "loss": 0.2769, "step": 22990 }, { "epoch": 1.5894227445558244, "grad_norm": 0.7149390447924379, "learning_rate": 1.0653974235000892e-06, "loss": 0.3159, "step": 22991 }, { "epoch": 1.5894918769443485, "grad_norm": 0.7851601177036497, "learning_rate": 1.0650520462805709e-06, "loss": 0.2647, "step": 22992 }, { "epoch": 1.5895610093328725, "grad_norm": 1.0312243560390326, "learning_rate": 1.0647067183786019e-06, "loss": 0.2492, "step": 22993 }, { "epoch": 1.5896301417213965, "grad_norm": 0.6827396062594858, "learning_rate": 1.0643614397985131e-06, "loss": 0.362, "step": 22994 }, { "epoch": 1.5896992741099205, "grad_norm": 0.6888791750616337, "learning_rate": 1.0640162105446268e-06, "loss": 0.3394, "step": 22995 }, { "epoch": 1.5897684064984445, "grad_norm": 0.6348554739651987, "learning_rate": 1.063671030621276e-06, "loss": 0.204, "step": 22996 }, { "epoch": 1.5898375388869685, "grad_norm": 0.6978127574888797, "learning_rate": 1.0633259000327833e-06, "loss": 0.2453, "step": 22997 }, { "epoch": 1.5899066712754926, "grad_norm": 0.6772697810154459, "learning_rate": 1.0629808187834762e-06, "loss": 0.3794, "step": 22998 }, { "epoch": 1.5899758036640166, "grad_norm": 0.6784938826977962, "learning_rate": 1.0626357868776771e-06, "loss": 0.2402, "step": 22999 }, { "epoch": 1.5900449360525406, "grad_norm": 0.6933686889585662, "learning_rate": 1.0622908043197133e-06, "loss": 0.2322, "step": 23000 }, { "epoch": 1.5901140684410646, "grad_norm": 0.6749065247463786, "learning_rate": 1.0619458711139058e-06, "loss": 0.2669, "step": 23001 }, { "epoch": 1.5901832008295886, "grad_norm": 0.8687645530659357, "learning_rate": 1.0616009872645794e-06, "loss": 0.2855, "step": 23002 }, { "epoch": 1.5902523332181127, "grad_norm": 0.7392029033759197, "learning_rate": 1.0612561527760557e-06, "loss": 0.2449, "step": 23003 }, { "epoch": 1.5903214656066367, "grad_norm": 0.8681209913003541, "learning_rate": 1.060911367652659e-06, "loss": 0.2542, "step": 23004 }, { "epoch": 1.5903905979951607, "grad_norm": 0.6844941250771541, "learning_rate": 1.0605666318987069e-06, "loss": 0.3779, "step": 23005 }, { "epoch": 1.5904597303836847, "grad_norm": 0.8976990865276339, "learning_rate": 1.060221945518523e-06, "loss": 0.259, "step": 23006 }, { "epoch": 1.5905288627722087, "grad_norm": 0.7552924945336577, "learning_rate": 1.0598773085164255e-06, "loss": 0.2834, "step": 23007 }, { "epoch": 1.5905979951607327, "grad_norm": 0.980884536203727, "learning_rate": 1.0595327208967354e-06, "loss": 0.2624, "step": 23008 }, { "epoch": 1.5906671275492568, "grad_norm": 0.6310061197937794, "learning_rate": 1.059188182663769e-06, "loss": 0.3383, "step": 23009 }, { "epoch": 1.5907362599377808, "grad_norm": 0.6754400589512238, "learning_rate": 1.0588436938218461e-06, "loss": 0.2108, "step": 23010 }, { "epoch": 1.5908053923263048, "grad_norm": 0.6793286905871364, "learning_rate": 1.0584992543752849e-06, "loss": 0.2112, "step": 23011 }, { "epoch": 1.5908745247148288, "grad_norm": 0.7979426761813468, "learning_rate": 1.0581548643284023e-06, "loss": 0.3753, "step": 23012 }, { "epoch": 1.5909436571033528, "grad_norm": 0.741965836596521, "learning_rate": 1.0578105236855142e-06, "loss": 0.2792, "step": 23013 }, { "epoch": 1.5910127894918769, "grad_norm": 0.7544652902407164, "learning_rate": 1.0574662324509344e-06, "loss": 0.3036, "step": 23014 }, { "epoch": 1.5910819218804009, "grad_norm": 0.6872776873660554, "learning_rate": 1.0571219906289797e-06, "loss": 0.2347, "step": 23015 }, { "epoch": 1.591151054268925, "grad_norm": 0.6895369511684412, "learning_rate": 1.0567777982239647e-06, "loss": 0.3768, "step": 23016 }, { "epoch": 1.591220186657449, "grad_norm": 0.7812976631523446, "learning_rate": 1.0564336552402039e-06, "loss": 0.2921, "step": 23017 }, { "epoch": 1.591289319045973, "grad_norm": 0.6658345416000547, "learning_rate": 1.0560895616820088e-06, "loss": 0.2237, "step": 23018 }, { "epoch": 1.591358451434497, "grad_norm": 0.8759208133936107, "learning_rate": 1.0557455175536935e-06, "loss": 0.3404, "step": 23019 }, { "epoch": 1.591427583823021, "grad_norm": 0.7335741405283963, "learning_rate": 1.0554015228595687e-06, "loss": 0.313, "step": 23020 }, { "epoch": 1.591496716211545, "grad_norm": 0.6616519811688183, "learning_rate": 1.0550575776039474e-06, "loss": 0.2415, "step": 23021 }, { "epoch": 1.591565848600069, "grad_norm": 0.6631716184978234, "learning_rate": 1.0547136817911379e-06, "loss": 0.2952, "step": 23022 }, { "epoch": 1.591634980988593, "grad_norm": 0.6670239498473051, "learning_rate": 1.0543698354254517e-06, "loss": 0.3407, "step": 23023 }, { "epoch": 1.591704113377117, "grad_norm": 0.7167706143531603, "learning_rate": 1.0540260385111984e-06, "loss": 0.2613, "step": 23024 }, { "epoch": 1.591773245765641, "grad_norm": 0.6962415037693718, "learning_rate": 1.0536822910526878e-06, "loss": 0.1862, "step": 23025 }, { "epoch": 1.591842378154165, "grad_norm": 0.7768317524654902, "learning_rate": 1.053338593054226e-06, "loss": 0.3345, "step": 23026 }, { "epoch": 1.591911510542689, "grad_norm": 0.7416768300686277, "learning_rate": 1.0529949445201238e-06, "loss": 0.375, "step": 23027 }, { "epoch": 1.591980642931213, "grad_norm": 0.6517173869958128, "learning_rate": 1.0526513454546844e-06, "loss": 0.2212, "step": 23028 }, { "epoch": 1.5920497753197373, "grad_norm": 0.6944221402670809, "learning_rate": 1.0523077958622174e-06, "loss": 0.2289, "step": 23029 }, { "epoch": 1.5921189077082614, "grad_norm": 0.7678107936385633, "learning_rate": 1.0519642957470244e-06, "loss": 0.3663, "step": 23030 }, { "epoch": 1.5921880400967854, "grad_norm": 1.1127313503441483, "learning_rate": 1.0516208451134157e-06, "loss": 0.223, "step": 23031 }, { "epoch": 1.5922571724853094, "grad_norm": 0.6654940183298265, "learning_rate": 1.0512774439656925e-06, "loss": 0.2322, "step": 23032 }, { "epoch": 1.5923263048738334, "grad_norm": 0.7067416780387409, "learning_rate": 1.0509340923081611e-06, "loss": 0.3599, "step": 23033 }, { "epoch": 1.5923954372623574, "grad_norm": 0.7188268556598434, "learning_rate": 1.0505907901451224e-06, "loss": 0.3998, "step": 23034 }, { "epoch": 1.5924645696508815, "grad_norm": 0.7403190616555405, "learning_rate": 1.0502475374808812e-06, "loss": 0.2176, "step": 23035 }, { "epoch": 1.5925337020394055, "grad_norm": 0.6465029074643294, "learning_rate": 1.0499043343197373e-06, "loss": 0.2315, "step": 23036 }, { "epoch": 1.5926028344279295, "grad_norm": 0.7217314572144043, "learning_rate": 1.0495611806659929e-06, "loss": 0.3021, "step": 23037 }, { "epoch": 1.5926719668164535, "grad_norm": 0.7389340769244764, "learning_rate": 1.0492180765239496e-06, "loss": 0.3162, "step": 23038 }, { "epoch": 1.5927410992049775, "grad_norm": 0.6838802151245562, "learning_rate": 1.0488750218979082e-06, "loss": 0.2723, "step": 23039 }, { "epoch": 1.5928102315935015, "grad_norm": 1.0459368766042232, "learning_rate": 1.0485320167921665e-06, "loss": 0.1993, "step": 23040 }, { "epoch": 1.5928793639820256, "grad_norm": 0.7904658596871639, "learning_rate": 1.0481890612110251e-06, "loss": 0.3319, "step": 23041 }, { "epoch": 1.5929484963705496, "grad_norm": 0.8062248519572659, "learning_rate": 1.0478461551587803e-06, "loss": 0.2899, "step": 23042 }, { "epoch": 1.5930176287590736, "grad_norm": 0.5943086678265631, "learning_rate": 1.047503298639732e-06, "loss": 0.204, "step": 23043 }, { "epoch": 1.5930867611475976, "grad_norm": 0.8042886999722931, "learning_rate": 1.0471604916581758e-06, "loss": 0.3154, "step": 23044 }, { "epoch": 1.5931558935361216, "grad_norm": 0.783134630456754, "learning_rate": 1.0468177342184083e-06, "loss": 0.3335, "step": 23045 }, { "epoch": 1.5932250259246457, "grad_norm": 0.7375350245799304, "learning_rate": 1.0464750263247252e-06, "loss": 0.2393, "step": 23046 }, { "epoch": 1.5932941583131697, "grad_norm": 0.7230648963033489, "learning_rate": 1.0461323679814239e-06, "loss": 0.268, "step": 23047 }, { "epoch": 1.5933632907016937, "grad_norm": 0.7033097179532497, "learning_rate": 1.0457897591927962e-06, "loss": 0.3554, "step": 23048 }, { "epoch": 1.593432423090218, "grad_norm": 0.6687323937691656, "learning_rate": 1.045447199963139e-06, "loss": 0.2456, "step": 23049 }, { "epoch": 1.593501555478742, "grad_norm": 0.6134614692945939, "learning_rate": 1.0451046902967421e-06, "loss": 0.2278, "step": 23050 }, { "epoch": 1.593570687867266, "grad_norm": 0.6219281963504659, "learning_rate": 1.0447622301978999e-06, "loss": 0.2571, "step": 23051 }, { "epoch": 1.59363982025579, "grad_norm": 0.7981225228719495, "learning_rate": 1.0444198196709048e-06, "loss": 0.3473, "step": 23052 }, { "epoch": 1.593708952644314, "grad_norm": 0.7226471712117479, "learning_rate": 1.0440774587200496e-06, "loss": 0.2257, "step": 23053 }, { "epoch": 1.593778085032838, "grad_norm": 0.6840974715418238, "learning_rate": 1.0437351473496232e-06, "loss": 0.2511, "step": 23054 }, { "epoch": 1.593847217421362, "grad_norm": 0.6803406581611432, "learning_rate": 1.043392885563917e-06, "loss": 0.293, "step": 23055 }, { "epoch": 1.593916349809886, "grad_norm": 0.6750164869669442, "learning_rate": 1.0430506733672186e-06, "loss": 0.2504, "step": 23056 }, { "epoch": 1.59398548219841, "grad_norm": 0.6114857382441146, "learning_rate": 1.0427085107638202e-06, "loss": 0.2096, "step": 23057 }, { "epoch": 1.594054614586934, "grad_norm": 0.658828487369418, "learning_rate": 1.0423663977580074e-06, "loss": 0.2744, "step": 23058 }, { "epoch": 1.5941237469754581, "grad_norm": 0.7625909296272081, "learning_rate": 1.0420243343540688e-06, "loss": 0.2836, "step": 23059 }, { "epoch": 1.5941928793639821, "grad_norm": 0.7268117941036757, "learning_rate": 1.041682320556292e-06, "loss": 0.2587, "step": 23060 }, { "epoch": 1.5942620117525061, "grad_norm": 0.6556693847911363, "learning_rate": 1.0413403563689645e-06, "loss": 0.2215, "step": 23061 }, { "epoch": 1.5943311441410302, "grad_norm": 0.6410997578984462, "learning_rate": 1.0409984417963698e-06, "loss": 0.3066, "step": 23062 }, { "epoch": 1.5944002765295542, "grad_norm": 1.0911000228464938, "learning_rate": 1.0406565768427957e-06, "loss": 0.2273, "step": 23063 }, { "epoch": 1.5944694089180782, "grad_norm": 0.6847166647157749, "learning_rate": 1.0403147615125247e-06, "loss": 0.2322, "step": 23064 }, { "epoch": 1.5945385413066022, "grad_norm": 0.7596294416572646, "learning_rate": 1.0399729958098427e-06, "loss": 0.215, "step": 23065 }, { "epoch": 1.5946076736951262, "grad_norm": 0.7039155915259686, "learning_rate": 1.0396312797390295e-06, "loss": 0.31, "step": 23066 }, { "epoch": 1.5946768060836503, "grad_norm": 0.6780001415768173, "learning_rate": 1.039289613304374e-06, "loss": 0.2003, "step": 23067 }, { "epoch": 1.5947459384721743, "grad_norm": 0.7141739038599074, "learning_rate": 1.0389479965101529e-06, "loss": 0.1976, "step": 23068 }, { "epoch": 1.5948150708606983, "grad_norm": 0.7325092315090237, "learning_rate": 1.0386064293606512e-06, "loss": 0.322, "step": 23069 }, { "epoch": 1.5948842032492223, "grad_norm": 0.7082535392178777, "learning_rate": 1.038264911860148e-06, "loss": 0.3293, "step": 23070 }, { "epoch": 1.5949533356377463, "grad_norm": 0.7627430788315817, "learning_rate": 1.0379234440129244e-06, "loss": 0.2918, "step": 23071 }, { "epoch": 1.5950224680262703, "grad_norm": 0.7349761059871748, "learning_rate": 1.0375820258232588e-06, "loss": 0.3011, "step": 23072 }, { "epoch": 1.5950916004147944, "grad_norm": 0.6535268293244909, "learning_rate": 1.0372406572954313e-06, "loss": 0.3148, "step": 23073 }, { "epoch": 1.5951607328033184, "grad_norm": 0.6835531996640994, "learning_rate": 1.0368993384337206e-06, "loss": 0.2712, "step": 23074 }, { "epoch": 1.5952298651918424, "grad_norm": 0.6885387937808722, "learning_rate": 1.0365580692424054e-06, "loss": 0.2352, "step": 23075 }, { "epoch": 1.5952989975803664, "grad_norm": 0.7375253440586019, "learning_rate": 1.0362168497257602e-06, "loss": 0.302, "step": 23076 }, { "epoch": 1.5953681299688904, "grad_norm": 0.7337356947618775, "learning_rate": 1.0358756798880642e-06, "loss": 0.3336, "step": 23077 }, { "epoch": 1.5954372623574145, "grad_norm": 0.7791832298429536, "learning_rate": 1.0355345597335915e-06, "loss": 0.259, "step": 23078 }, { "epoch": 1.5955063947459385, "grad_norm": 0.6256330852365827, "learning_rate": 1.0351934892666193e-06, "loss": 0.2321, "step": 23079 }, { "epoch": 1.5955755271344625, "grad_norm": 0.7925502759324132, "learning_rate": 1.0348524684914197e-06, "loss": 0.3494, "step": 23080 }, { "epoch": 1.5956446595229865, "grad_norm": 0.6354255333171642, "learning_rate": 1.0345114974122684e-06, "loss": 0.2349, "step": 23081 }, { "epoch": 1.5957137919115105, "grad_norm": 0.7574680207874305, "learning_rate": 1.0341705760334402e-06, "loss": 0.1954, "step": 23082 }, { "epoch": 1.5957829243000345, "grad_norm": 0.7364833950629395, "learning_rate": 1.033829704359205e-06, "loss": 0.2487, "step": 23083 }, { "epoch": 1.5958520566885586, "grad_norm": 0.6778814369134033, "learning_rate": 1.0334888823938378e-06, "loss": 0.3386, "step": 23084 }, { "epoch": 1.5959211890770826, "grad_norm": 0.7108883059603117, "learning_rate": 1.033148110141608e-06, "loss": 0.204, "step": 23085 }, { "epoch": 1.5959903214656066, "grad_norm": 0.6237953565585518, "learning_rate": 1.0328073876067868e-06, "loss": 0.2193, "step": 23086 }, { "epoch": 1.5960594538541306, "grad_norm": 0.6796232909159245, "learning_rate": 1.0324667147936458e-06, "loss": 0.2975, "step": 23087 }, { "epoch": 1.5961285862426546, "grad_norm": 0.7316893627509177, "learning_rate": 1.0321260917064553e-06, "loss": 0.2758, "step": 23088 }, { "epoch": 1.5961977186311787, "grad_norm": 0.6729815510696615, "learning_rate": 1.0317855183494818e-06, "loss": 0.2804, "step": 23089 }, { "epoch": 1.5962668510197027, "grad_norm": 0.75485030776546, "learning_rate": 1.0314449947269967e-06, "loss": 0.2483, "step": 23090 }, { "epoch": 1.5963359834082267, "grad_norm": 0.7115073713670704, "learning_rate": 1.0311045208432646e-06, "loss": 0.2868, "step": 23091 }, { "epoch": 1.5964051157967507, "grad_norm": 0.6688020394798492, "learning_rate": 1.0307640967025562e-06, "loss": 0.255, "step": 23092 }, { "epoch": 1.5964742481852747, "grad_norm": 0.6517318554462588, "learning_rate": 1.030423722309135e-06, "loss": 0.2023, "step": 23093 }, { "epoch": 1.5965433805737987, "grad_norm": 0.777115650191666, "learning_rate": 1.0300833976672685e-06, "loss": 0.3208, "step": 23094 }, { "epoch": 1.5966125129623228, "grad_norm": 0.721189303648595, "learning_rate": 1.029743122781222e-06, "loss": 0.2612, "step": 23095 }, { "epoch": 1.5966816453508468, "grad_norm": 0.697125050436269, "learning_rate": 1.0294028976552618e-06, "loss": 0.253, "step": 23096 }, { "epoch": 1.5967507777393708, "grad_norm": 0.7843843043426367, "learning_rate": 1.0290627222936489e-06, "loss": 0.2443, "step": 23097 }, { "epoch": 1.5968199101278948, "grad_norm": 0.7193613128662812, "learning_rate": 1.0287225967006498e-06, "loss": 0.3138, "step": 23098 }, { "epoch": 1.5968890425164188, "grad_norm": 0.6850610814746405, "learning_rate": 1.0283825208805243e-06, "loss": 0.2851, "step": 23099 }, { "epoch": 1.5969581749049429, "grad_norm": 0.5970061595381476, "learning_rate": 1.0280424948375372e-06, "loss": 0.2253, "step": 23100 }, { "epoch": 1.5970273072934669, "grad_norm": 0.6602689105703512, "learning_rate": 1.027702518575947e-06, "loss": 0.2294, "step": 23101 }, { "epoch": 1.5970964396819909, "grad_norm": 0.9502478687305299, "learning_rate": 1.0273625921000197e-06, "loss": 0.3469, "step": 23102 }, { "epoch": 1.597165572070515, "grad_norm": 0.6892002233153985, "learning_rate": 1.027022715414011e-06, "loss": 0.2239, "step": 23103 }, { "epoch": 1.597234704459039, "grad_norm": 0.683616828983138, "learning_rate": 1.0266828885221842e-06, "loss": 0.2325, "step": 23104 }, { "epoch": 1.597303836847563, "grad_norm": 0.6950171662745911, "learning_rate": 1.0263431114287948e-06, "loss": 0.2912, "step": 23105 }, { "epoch": 1.597372969236087, "grad_norm": 0.6473154592898657, "learning_rate": 1.0260033841381051e-06, "loss": 0.2509, "step": 23106 }, { "epoch": 1.597442101624611, "grad_norm": 0.6871856618927942, "learning_rate": 1.0256637066543696e-06, "loss": 0.25, "step": 23107 }, { "epoch": 1.597511234013135, "grad_norm": 0.6821659103191401, "learning_rate": 1.025324078981847e-06, "loss": 0.2824, "step": 23108 }, { "epoch": 1.597580366401659, "grad_norm": 0.704748539970754, "learning_rate": 1.024984501124794e-06, "loss": 0.3335, "step": 23109 }, { "epoch": 1.5976494987901833, "grad_norm": 0.6955533752627004, "learning_rate": 1.0246449730874675e-06, "loss": 0.2591, "step": 23110 }, { "epoch": 1.5977186311787073, "grad_norm": 0.771956807012567, "learning_rate": 1.0243054948741205e-06, "loss": 0.2288, "step": 23111 }, { "epoch": 1.5977877635672313, "grad_norm": 0.7068301519211798, "learning_rate": 1.0239660664890112e-06, "loss": 0.2992, "step": 23112 }, { "epoch": 1.5978568959557553, "grad_norm": 0.6942323772272845, "learning_rate": 1.0236266879363898e-06, "loss": 0.3232, "step": 23113 }, { "epoch": 1.5979260283442793, "grad_norm": 0.7171371712682123, "learning_rate": 1.0232873592205134e-06, "loss": 0.2434, "step": 23114 }, { "epoch": 1.5979951607328033, "grad_norm": 1.0644952631469553, "learning_rate": 1.0229480803456315e-06, "loss": 0.2457, "step": 23115 }, { "epoch": 1.5980642931213274, "grad_norm": 0.6912030754511957, "learning_rate": 1.0226088513159981e-06, "loss": 0.3424, "step": 23116 }, { "epoch": 1.5981334255098514, "grad_norm": 0.6529579508550445, "learning_rate": 1.0222696721358654e-06, "loss": 0.2823, "step": 23117 }, { "epoch": 1.5982025578983754, "grad_norm": 0.6545818333060323, "learning_rate": 1.0219305428094843e-06, "loss": 0.2357, "step": 23118 }, { "epoch": 1.5982716902868994, "grad_norm": 0.6657531446977597, "learning_rate": 1.021591463341104e-06, "loss": 0.302, "step": 23119 }, { "epoch": 1.5983408226754234, "grad_norm": 0.6709106832133491, "learning_rate": 1.021252433734976e-06, "loss": 0.2773, "step": 23120 }, { "epoch": 1.5984099550639475, "grad_norm": 0.6480327323688935, "learning_rate": 1.020913453995347e-06, "loss": 0.2338, "step": 23121 }, { "epoch": 1.5984790874524715, "grad_norm": 0.8056471973166041, "learning_rate": 1.0205745241264664e-06, "loss": 0.2447, "step": 23122 }, { "epoch": 1.5985482198409955, "grad_norm": 0.71952544236807, "learning_rate": 1.0202356441325828e-06, "loss": 0.3629, "step": 23123 }, { "epoch": 1.5986173522295195, "grad_norm": 0.7159414308741275, "learning_rate": 1.0198968140179449e-06, "loss": 0.2409, "step": 23124 }, { "epoch": 1.5986864846180435, "grad_norm": 0.631185533536886, "learning_rate": 1.019558033786796e-06, "loss": 0.2241, "step": 23125 }, { "epoch": 1.5987556170065675, "grad_norm": 0.771911312532606, "learning_rate": 1.019219303443385e-06, "loss": 0.2657, "step": 23126 }, { "epoch": 1.5988247493950916, "grad_norm": 0.7361387226873964, "learning_rate": 1.018880622991955e-06, "loss": 0.3486, "step": 23127 }, { "epoch": 1.5988938817836156, "grad_norm": 0.713364924905695, "learning_rate": 1.018541992436753e-06, "loss": 0.2569, "step": 23128 }, { "epoch": 1.5989630141721396, "grad_norm": 0.6627311056183772, "learning_rate": 1.0182034117820205e-06, "loss": 0.2309, "step": 23129 }, { "epoch": 1.5990321465606638, "grad_norm": 1.0103883781029386, "learning_rate": 1.0178648810320024e-06, "loss": 0.3293, "step": 23130 }, { "epoch": 1.5991012789491879, "grad_norm": 0.8422829631432657, "learning_rate": 1.0175264001909414e-06, "loss": 0.2384, "step": 23131 }, { "epoch": 1.5991704113377119, "grad_norm": 0.6514293935468788, "learning_rate": 1.0171879692630815e-06, "loss": 0.2768, "step": 23132 }, { "epoch": 1.599239543726236, "grad_norm": 0.6707533205411059, "learning_rate": 1.0168495882526613e-06, "loss": 0.2486, "step": 23133 }, { "epoch": 1.59930867611476, "grad_norm": 0.6585622823897884, "learning_rate": 1.0165112571639246e-06, "loss": 0.3188, "step": 23134 }, { "epoch": 1.599377808503284, "grad_norm": 0.6596863691002134, "learning_rate": 1.016172976001109e-06, "loss": 0.1988, "step": 23135 }, { "epoch": 1.599446940891808, "grad_norm": 0.7050564658265023, "learning_rate": 1.015834744768457e-06, "loss": 0.2782, "step": 23136 }, { "epoch": 1.599516073280332, "grad_norm": 0.7233987541635425, "learning_rate": 1.015496563470204e-06, "loss": 0.3405, "step": 23137 }, { "epoch": 1.599585205668856, "grad_norm": 0.6575120911527007, "learning_rate": 1.0151584321105933e-06, "loss": 0.2257, "step": 23138 }, { "epoch": 1.59965433805738, "grad_norm": 0.6159457750016899, "learning_rate": 1.0148203506938592e-06, "loss": 0.19, "step": 23139 }, { "epoch": 1.599723470445904, "grad_norm": 0.6589498725519266, "learning_rate": 1.0144823192242415e-06, "loss": 0.2207, "step": 23140 }, { "epoch": 1.599792602834428, "grad_norm": 0.7081860585887753, "learning_rate": 1.014144337705974e-06, "loss": 0.3728, "step": 23141 }, { "epoch": 1.599861735222952, "grad_norm": 1.4738291868532203, "learning_rate": 1.0138064061432956e-06, "loss": 0.1994, "step": 23142 }, { "epoch": 1.599930867611476, "grad_norm": 0.7033278249731368, "learning_rate": 1.013468524540439e-06, "loss": 0.216, "step": 23143 }, { "epoch": 1.6, "grad_norm": 0.7885945736382364, "learning_rate": 1.01313069290164e-06, "loss": 0.2484, "step": 23144 }, { "epoch": 1.600069132388524, "grad_norm": 0.6884991060124167, "learning_rate": 1.012792911231133e-06, "loss": 0.2695, "step": 23145 }, { "epoch": 1.6001382647770481, "grad_norm": 0.6353830835312082, "learning_rate": 1.0124551795331522e-06, "loss": 0.2257, "step": 23146 }, { "epoch": 1.6002073971655721, "grad_norm": 0.6664974515809416, "learning_rate": 1.0121174978119303e-06, "loss": 0.2746, "step": 23147 }, { "epoch": 1.6002765295540962, "grad_norm": 0.6406835652786458, "learning_rate": 1.0117798660716971e-06, "loss": 0.292, "step": 23148 }, { "epoch": 1.6003456619426202, "grad_norm": 0.6711795725949947, "learning_rate": 1.0114422843166877e-06, "loss": 0.193, "step": 23149 }, { "epoch": 1.6004147943311442, "grad_norm": 0.6619230689360954, "learning_rate": 1.0111047525511296e-06, "loss": 0.2004, "step": 23150 }, { "epoch": 1.6004839267196682, "grad_norm": 0.7798942619248778, "learning_rate": 1.0107672707792555e-06, "loss": 0.2871, "step": 23151 }, { "epoch": 1.6005530591081922, "grad_norm": 0.7594051796308642, "learning_rate": 1.0104298390052942e-06, "loss": 0.292, "step": 23152 }, { "epoch": 1.6006221914967163, "grad_norm": 0.6857459554914803, "learning_rate": 1.0100924572334763e-06, "loss": 0.2631, "step": 23153 }, { "epoch": 1.6006913238852403, "grad_norm": 0.741123954090959, "learning_rate": 1.0097551254680276e-06, "loss": 0.2554, "step": 23154 }, { "epoch": 1.6007604562737643, "grad_norm": 0.9211062011331481, "learning_rate": 1.0094178437131796e-06, "loss": 0.3588, "step": 23155 }, { "epoch": 1.6008295886622883, "grad_norm": 0.7380958848258846, "learning_rate": 1.0090806119731556e-06, "loss": 0.2669, "step": 23156 }, { "epoch": 1.6008987210508123, "grad_norm": 0.6206302562758934, "learning_rate": 1.0087434302521843e-06, "loss": 0.222, "step": 23157 }, { "epoch": 1.6009678534393363, "grad_norm": 0.7095309982421881, "learning_rate": 1.0084062985544908e-06, "loss": 0.2434, "step": 23158 }, { "epoch": 1.6010369858278604, "grad_norm": 0.7091930869158672, "learning_rate": 1.0080692168843027e-06, "loss": 0.2831, "step": 23159 }, { "epoch": 1.6011061182163844, "grad_norm": 0.6895280722086624, "learning_rate": 1.007732185245842e-06, "loss": 0.2184, "step": 23160 }, { "epoch": 1.6011752506049084, "grad_norm": 0.705646437679034, "learning_rate": 1.0073952036433348e-06, "loss": 0.2735, "step": 23161 }, { "epoch": 1.6012443829934324, "grad_norm": 0.6983386091312165, "learning_rate": 1.0070582720810023e-06, "loss": 0.3206, "step": 23162 }, { "epoch": 1.6013135153819564, "grad_norm": 0.714808916073314, "learning_rate": 1.0067213905630701e-06, "loss": 0.2881, "step": 23163 }, { "epoch": 1.6013826477704804, "grad_norm": 0.6880692602964913, "learning_rate": 1.0063845590937577e-06, "loss": 0.2408, "step": 23164 }, { "epoch": 1.6014517801590045, "grad_norm": 0.7644573914850955, "learning_rate": 1.0060477776772875e-06, "loss": 0.2373, "step": 23165 }, { "epoch": 1.6015209125475285, "grad_norm": 0.7176709173281722, "learning_rate": 1.0057110463178815e-06, "loss": 0.3475, "step": 23166 }, { "epoch": 1.6015900449360525, "grad_norm": 0.6491912181890384, "learning_rate": 1.0053743650197607e-06, "loss": 0.2117, "step": 23167 }, { "epoch": 1.6016591773245765, "grad_norm": 0.6540816451618588, "learning_rate": 1.0050377337871415e-06, "loss": 0.1984, "step": 23168 }, { "epoch": 1.6017283097131005, "grad_norm": 0.7697830327411933, "learning_rate": 1.0047011526242468e-06, "loss": 0.2526, "step": 23169 }, { "epoch": 1.6017974421016246, "grad_norm": 0.7052718215248415, "learning_rate": 1.0043646215352915e-06, "loss": 0.2925, "step": 23170 }, { "epoch": 1.6018665744901486, "grad_norm": 0.6756415829930301, "learning_rate": 1.0040281405244966e-06, "loss": 0.1969, "step": 23171 }, { "epoch": 1.6019357068786726, "grad_norm": 0.6655854314054488, "learning_rate": 1.0036917095960758e-06, "loss": 0.2335, "step": 23172 }, { "epoch": 1.6020048392671966, "grad_norm": 0.6549212719567855, "learning_rate": 1.0033553287542498e-06, "loss": 0.3265, "step": 23173 }, { "epoch": 1.6020739716557206, "grad_norm": 0.6993774304296659, "learning_rate": 1.0030189980032313e-06, "loss": 0.2638, "step": 23174 }, { "epoch": 1.6021431040442446, "grad_norm": 0.6087275951625358, "learning_rate": 1.0026827173472376e-06, "loss": 0.1962, "step": 23175 }, { "epoch": 1.6022122364327687, "grad_norm": 0.7345645591380263, "learning_rate": 1.0023464867904814e-06, "loss": 0.3432, "step": 23176 }, { "epoch": 1.6022813688212927, "grad_norm": 0.7228666902759973, "learning_rate": 1.0020103063371795e-06, "loss": 0.3442, "step": 23177 }, { "epoch": 1.6023505012098167, "grad_norm": 0.7316709942958977, "learning_rate": 1.0016741759915422e-06, "loss": 0.2514, "step": 23178 }, { "epoch": 1.6024196335983407, "grad_norm": 0.6407816326976489, "learning_rate": 1.0013380957577835e-06, "loss": 0.2303, "step": 23179 }, { "epoch": 1.6024887659868647, "grad_norm": 0.7246025766033815, "learning_rate": 1.001002065640116e-06, "loss": 0.339, "step": 23180 }, { "epoch": 1.6025578983753888, "grad_norm": 0.7421474797969692, "learning_rate": 1.0006660856427524e-06, "loss": 0.2043, "step": 23181 }, { "epoch": 1.6026270307639128, "grad_norm": 0.7237999851991663, "learning_rate": 1.0003301557699013e-06, "loss": 0.2147, "step": 23182 }, { "epoch": 1.6026961631524368, "grad_norm": 0.8166427134439667, "learning_rate": 9.999942760257747e-07, "loss": 0.3148, "step": 23183 }, { "epoch": 1.6027652955409608, "grad_norm": 0.7387148620839913, "learning_rate": 9.996584464145809e-07, "loss": 0.3608, "step": 23184 }, { "epoch": 1.6028344279294848, "grad_norm": 0.6831159047277632, "learning_rate": 9.993226669405303e-07, "loss": 0.2201, "step": 23185 }, { "epoch": 1.6029035603180088, "grad_norm": 0.7400323287097088, "learning_rate": 9.989869376078292e-07, "loss": 0.2128, "step": 23186 }, { "epoch": 1.6029726927065329, "grad_norm": 0.7401947583862952, "learning_rate": 9.98651258420687e-07, "loss": 0.3321, "step": 23187 }, { "epoch": 1.6030418250950569, "grad_norm": 0.9167483054747856, "learning_rate": 9.983156293833102e-07, "loss": 0.2833, "step": 23188 }, { "epoch": 1.603110957483581, "grad_norm": 0.721963728451406, "learning_rate": 9.97980050499907e-07, "loss": 0.2609, "step": 23189 }, { "epoch": 1.603180089872105, "grad_norm": 0.663110177226687, "learning_rate": 9.976445217746811e-07, "loss": 0.2328, "step": 23190 }, { "epoch": 1.6032492222606292, "grad_norm": 0.6728899741595169, "learning_rate": 9.973090432118398e-07, "loss": 0.3455, "step": 23191 }, { "epoch": 1.6033183546491532, "grad_norm": 0.6723661953468227, "learning_rate": 9.96973614815585e-07, "loss": 0.2255, "step": 23192 }, { "epoch": 1.6033874870376772, "grad_norm": 0.6523313989027751, "learning_rate": 9.966382365901223e-07, "loss": 0.2386, "step": 23193 }, { "epoch": 1.6034566194262012, "grad_norm": 0.7146473584989724, "learning_rate": 9.963029085396547e-07, "loss": 0.2847, "step": 23194 }, { "epoch": 1.6035257518147252, "grad_norm": 0.7168634548302825, "learning_rate": 9.959676306683869e-07, "loss": 0.3422, "step": 23195 }, { "epoch": 1.6035948842032492, "grad_norm": 0.7938390899365492, "learning_rate": 9.956324029805182e-07, "loss": 0.2109, "step": 23196 }, { "epoch": 1.6036640165917733, "grad_norm": 0.6573625246683503, "learning_rate": 9.952972254802518e-07, "loss": 0.2864, "step": 23197 }, { "epoch": 1.6037331489802973, "grad_norm": 0.6662714886210686, "learning_rate": 9.949620981717873e-07, "loss": 0.3655, "step": 23198 }, { "epoch": 1.6038022813688213, "grad_norm": 0.7521037735560157, "learning_rate": 9.946270210593267e-07, "loss": 0.2474, "step": 23199 }, { "epoch": 1.6038714137573453, "grad_norm": 0.6787529871931658, "learning_rate": 9.942919941470669e-07, "loss": 0.3025, "step": 23200 }, { "epoch": 1.6039405461458693, "grad_norm": 0.6640783501311871, "learning_rate": 9.939570174392088e-07, "loss": 0.2918, "step": 23201 }, { "epoch": 1.6040096785343934, "grad_norm": 1.372382258175171, "learning_rate": 9.936220909399501e-07, "loss": 0.2811, "step": 23202 }, { "epoch": 1.6040788109229174, "grad_norm": 0.7028163856365847, "learning_rate": 9.932872146534904e-07, "loss": 0.2736, "step": 23203 }, { "epoch": 1.6041479433114414, "grad_norm": 2.4605521448993595, "learning_rate": 9.929523885840236e-07, "loss": 0.233, "step": 23204 }, { "epoch": 1.6042170756999654, "grad_norm": 0.7907262349957406, "learning_rate": 9.926176127357495e-07, "loss": 0.3476, "step": 23205 }, { "epoch": 1.6042862080884894, "grad_norm": 0.8214903285613345, "learning_rate": 9.922828871128604e-07, "loss": 0.3048, "step": 23206 }, { "epoch": 1.6043553404770134, "grad_norm": 0.6847752603865821, "learning_rate": 9.919482117195545e-07, "loss": 0.271, "step": 23207 }, { "epoch": 1.6044244728655375, "grad_norm": 0.7136084332853441, "learning_rate": 9.916135865600223e-07, "loss": 0.2347, "step": 23208 }, { "epoch": 1.6044936052540615, "grad_norm": 0.7077136427616483, "learning_rate": 9.912790116384635e-07, "loss": 0.3147, "step": 23209 }, { "epoch": 1.6045627376425855, "grad_norm": 0.7224444750349871, "learning_rate": 9.909444869590674e-07, "loss": 0.2474, "step": 23210 }, { "epoch": 1.6046318700311095, "grad_norm": 0.8455324877266579, "learning_rate": 9.906100125260282e-07, "loss": 0.2692, "step": 23211 }, { "epoch": 1.6047010024196338, "grad_norm": 0.6991993214568953, "learning_rate": 9.902755883435367e-07, "loss": 0.2864, "step": 23212 }, { "epoch": 1.6047701348081578, "grad_norm": 0.6850828179006639, "learning_rate": 9.89941214415786e-07, "loss": 0.2335, "step": 23213 }, { "epoch": 1.6048392671966818, "grad_norm": 0.6976397330775503, "learning_rate": 9.896068907469647e-07, "loss": 0.2351, "step": 23214 }, { "epoch": 1.6049083995852058, "grad_norm": 0.5810937528730191, "learning_rate": 9.892726173412642e-07, "loss": 0.195, "step": 23215 }, { "epoch": 1.6049775319737298, "grad_norm": 0.6489480055654755, "learning_rate": 9.88938394202875e-07, "loss": 0.375, "step": 23216 }, { "epoch": 1.6050466643622538, "grad_norm": 0.7087586457724154, "learning_rate": 9.886042213359842e-07, "loss": 0.2796, "step": 23217 }, { "epoch": 1.6051157967507779, "grad_norm": 0.5801433074557748, "learning_rate": 9.882700987447818e-07, "loss": 0.2124, "step": 23218 }, { "epoch": 1.6051849291393019, "grad_norm": 1.081793863298701, "learning_rate": 9.879360264334532e-07, "loss": 0.3131, "step": 23219 }, { "epoch": 1.605254061527826, "grad_norm": 0.8966515612350313, "learning_rate": 9.87602004406188e-07, "loss": 0.3295, "step": 23220 }, { "epoch": 1.60532319391635, "grad_norm": 0.7575997198851535, "learning_rate": 9.872680326671697e-07, "loss": 0.2933, "step": 23221 }, { "epoch": 1.605392326304874, "grad_norm": 0.7145242892711273, "learning_rate": 9.869341112205854e-07, "loss": 0.2204, "step": 23222 }, { "epoch": 1.605461458693398, "grad_norm": 0.7642507240779143, "learning_rate": 9.86600240070621e-07, "loss": 0.3439, "step": 23223 }, { "epoch": 1.605530591081922, "grad_norm": 4.742751517644994, "learning_rate": 9.862664192214617e-07, "loss": 0.2606, "step": 23224 }, { "epoch": 1.605599723470446, "grad_norm": 0.6500766360316529, "learning_rate": 9.85932648677288e-07, "loss": 0.2167, "step": 23225 }, { "epoch": 1.60566885585897, "grad_norm": 0.6508716640750914, "learning_rate": 9.855989284422868e-07, "loss": 0.2998, "step": 23226 }, { "epoch": 1.605737988247494, "grad_norm": 0.7172043829741718, "learning_rate": 9.85265258520638e-07, "loss": 0.434, "step": 23227 }, { "epoch": 1.605807120636018, "grad_norm": 0.6650368231983557, "learning_rate": 9.849316389165243e-07, "loss": 0.252, "step": 23228 }, { "epoch": 1.605876253024542, "grad_norm": 0.6178324229510023, "learning_rate": 9.845980696341267e-07, "loss": 0.258, "step": 23229 }, { "epoch": 1.605945385413066, "grad_norm": 0.6820444350231594, "learning_rate": 9.842645506776289e-07, "loss": 0.3378, "step": 23230 }, { "epoch": 1.60601451780159, "grad_norm": 0.6858734549674057, "learning_rate": 9.839310820512066e-07, "loss": 0.2262, "step": 23231 }, { "epoch": 1.6060836501901141, "grad_norm": 0.6340869459788555, "learning_rate": 9.835976637590428e-07, "loss": 0.2134, "step": 23232 }, { "epoch": 1.6061527825786381, "grad_norm": 0.6402767841310594, "learning_rate": 9.83264295805313e-07, "loss": 0.3067, "step": 23233 }, { "epoch": 1.6062219149671622, "grad_norm": 0.7986534609488019, "learning_rate": 9.829309781941987e-07, "loss": 0.3938, "step": 23234 }, { "epoch": 1.6062910473556862, "grad_norm": 0.6767874672219152, "learning_rate": 9.82597710929874e-07, "loss": 0.2216, "step": 23235 }, { "epoch": 1.6063601797442102, "grad_norm": 1.4599239516216669, "learning_rate": 9.822644940165183e-07, "loss": 0.2089, "step": 23236 }, { "epoch": 1.6064293121327342, "grad_norm": 1.1589657651513408, "learning_rate": 9.81931327458307e-07, "loss": 0.3345, "step": 23237 }, { "epoch": 1.6064984445212582, "grad_norm": 0.6759952542962646, "learning_rate": 9.81598211259417e-07, "loss": 0.2243, "step": 23238 }, { "epoch": 1.6065675769097822, "grad_norm": 0.6907470345166843, "learning_rate": 9.812651454240212e-07, "loss": 0.2857, "step": 23239 }, { "epoch": 1.6066367092983063, "grad_norm": 0.7169988195213504, "learning_rate": 9.809321299562964e-07, "loss": 0.2321, "step": 23240 }, { "epoch": 1.6067058416868303, "grad_norm": 0.7283980448732023, "learning_rate": 9.805991648604134e-07, "loss": 0.3509, "step": 23241 }, { "epoch": 1.6067749740753543, "grad_norm": 0.6914103957510781, "learning_rate": 9.802662501405479e-07, "loss": 0.1809, "step": 23242 }, { "epoch": 1.6068441064638783, "grad_norm": 0.6539257536107222, "learning_rate": 9.79933385800871e-07, "loss": 0.2112, "step": 23243 }, { "epoch": 1.6069132388524023, "grad_norm": 0.7140839435396499, "learning_rate": 9.796005718455542e-07, "loss": 0.3349, "step": 23244 }, { "epoch": 1.6069823712409264, "grad_norm": 0.7934205298531221, "learning_rate": 9.792678082787704e-07, "loss": 0.3139, "step": 23245 }, { "epoch": 1.6070515036294504, "grad_norm": 0.8559783283788333, "learning_rate": 9.789350951046899e-07, "loss": 0.2227, "step": 23246 }, { "epoch": 1.6071206360179744, "grad_norm": 1.5934281644815032, "learning_rate": 9.786024323274811e-07, "loss": 0.2596, "step": 23247 }, { "epoch": 1.6071897684064984, "grad_norm": 0.7578469695190458, "learning_rate": 9.782698199513151e-07, "loss": 0.312, "step": 23248 }, { "epoch": 1.6072589007950224, "grad_norm": 0.7206444824992664, "learning_rate": 9.779372579803593e-07, "loss": 0.2512, "step": 23249 }, { "epoch": 1.6073280331835464, "grad_norm": 0.6633680806494036, "learning_rate": 9.776047464187816e-07, "loss": 0.2625, "step": 23250 }, { "epoch": 1.6073971655720705, "grad_norm": 0.6388156152410093, "learning_rate": 9.772722852707506e-07, "loss": 0.2927, "step": 23251 }, { "epoch": 1.6074662979605945, "grad_norm": 0.7490757469866286, "learning_rate": 9.769398745404334e-07, "loss": 0.3268, "step": 23252 }, { "epoch": 1.6075354303491185, "grad_norm": 0.7746019375072672, "learning_rate": 9.766075142319941e-07, "loss": 0.1871, "step": 23253 }, { "epoch": 1.6076045627376425, "grad_norm": 0.695837611377289, "learning_rate": 9.762752043496015e-07, "loss": 0.2467, "step": 23254 }, { "epoch": 1.6076736951261665, "grad_norm": 0.7354452188250289, "learning_rate": 9.759429448974168e-07, "loss": 0.3223, "step": 23255 }, { "epoch": 1.6077428275146906, "grad_norm": 0.688044787164189, "learning_rate": 9.756107358796074e-07, "loss": 0.329, "step": 23256 }, { "epoch": 1.6078119599032146, "grad_norm": 0.6747126732264587, "learning_rate": 9.752785773003343e-07, "loss": 0.2649, "step": 23257 }, { "epoch": 1.6078810922917386, "grad_norm": 0.6805214112832001, "learning_rate": 9.749464691637616e-07, "loss": 0.272, "step": 23258 }, { "epoch": 1.6079502246802626, "grad_norm": 0.69012376893915, "learning_rate": 9.746144114740524e-07, "loss": 0.3278, "step": 23259 }, { "epoch": 1.6080193570687866, "grad_norm": 0.6295130639065857, "learning_rate": 9.742824042353687e-07, "loss": 0.2509, "step": 23260 }, { "epoch": 1.6080884894573106, "grad_norm": 0.720479727383739, "learning_rate": 9.739504474518697e-07, "loss": 0.2765, "step": 23261 }, { "epoch": 1.6081576218458347, "grad_norm": 0.7866741537253229, "learning_rate": 9.736185411277187e-07, "loss": 0.3136, "step": 23262 }, { "epoch": 1.6082267542343587, "grad_norm": 0.6626085883203615, "learning_rate": 9.73286685267072e-07, "loss": 0.2124, "step": 23263 }, { "epoch": 1.6082958866228827, "grad_norm": 0.7728459666648659, "learning_rate": 9.729548798740912e-07, "loss": 0.2002, "step": 23264 }, { "epoch": 1.6083650190114067, "grad_norm": 1.0474066933035668, "learning_rate": 9.72623124952934e-07, "loss": 0.2589, "step": 23265 }, { "epoch": 1.6084341513999307, "grad_norm": 0.7932235832273417, "learning_rate": 9.722914205077605e-07, "loss": 0.3789, "step": 23266 }, { "epoch": 1.6085032837884548, "grad_norm": 0.7049734213701176, "learning_rate": 9.719597665427244e-07, "loss": 0.2434, "step": 23267 }, { "epoch": 1.6085724161769788, "grad_norm": 0.6416423980321552, "learning_rate": 9.716281630619862e-07, "loss": 0.265, "step": 23268 }, { "epoch": 1.6086415485655028, "grad_norm": 0.6898317089107612, "learning_rate": 9.71296610069698e-07, "loss": 0.281, "step": 23269 }, { "epoch": 1.6087106809540268, "grad_norm": 0.6752159264684683, "learning_rate": 9.70965107570019e-07, "loss": 0.2873, "step": 23270 }, { "epoch": 1.6087798133425508, "grad_norm": 0.6289742454482561, "learning_rate": 9.70633655567101e-07, "loss": 0.2091, "step": 23271 }, { "epoch": 1.608848945731075, "grad_norm": 0.6718506516152087, "learning_rate": 9.703022540650991e-07, "loss": 0.2587, "step": 23272 }, { "epoch": 1.608918078119599, "grad_norm": 0.7759146543452251, "learning_rate": 9.699709030681676e-07, "loss": 0.3044, "step": 23273 }, { "epoch": 1.608987210508123, "grad_norm": 0.6882283150341685, "learning_rate": 9.6963960258046e-07, "loss": 0.2566, "step": 23274 }, { "epoch": 1.6090563428966471, "grad_norm": 0.6778508901442559, "learning_rate": 9.69308352606126e-07, "loss": 0.27, "step": 23275 }, { "epoch": 1.6091254752851711, "grad_norm": 0.7304613562967153, "learning_rate": 9.689771531493198e-07, "loss": 0.3397, "step": 23276 }, { "epoch": 1.6091946076736952, "grad_norm": 0.8284691672478638, "learning_rate": 9.686460042141904e-07, "loss": 0.4212, "step": 23277 }, { "epoch": 1.6092637400622192, "grad_norm": 0.744083393887534, "learning_rate": 9.683149058048902e-07, "loss": 0.2244, "step": 23278 }, { "epoch": 1.6093328724507432, "grad_norm": 0.6299651142618665, "learning_rate": 9.679838579255667e-07, "loss": 0.2222, "step": 23279 }, { "epoch": 1.6094020048392672, "grad_norm": 0.6550475547927895, "learning_rate": 9.676528605803698e-07, "loss": 0.3149, "step": 23280 }, { "epoch": 1.6094711372277912, "grad_norm": 0.8517532554286421, "learning_rate": 9.673219137734497e-07, "loss": 0.2753, "step": 23281 }, { "epoch": 1.6095402696163152, "grad_norm": 0.6698846206625223, "learning_rate": 9.669910175089513e-07, "loss": 0.2406, "step": 23282 }, { "epoch": 1.6096094020048393, "grad_norm": 0.677927498083466, "learning_rate": 9.66660171791025e-07, "loss": 0.2977, "step": 23283 }, { "epoch": 1.6096785343933633, "grad_norm": 0.7212189475397007, "learning_rate": 9.663293766238142e-07, "loss": 0.3538, "step": 23284 }, { "epoch": 1.6097476667818873, "grad_norm": 0.7097661202441554, "learning_rate": 9.659986320114661e-07, "loss": 0.2325, "step": 23285 }, { "epoch": 1.6098167991704113, "grad_norm": 0.7016046866927881, "learning_rate": 9.656679379581264e-07, "loss": 0.2347, "step": 23286 }, { "epoch": 1.6098859315589353, "grad_norm": 0.7145435907524047, "learning_rate": 9.6533729446794e-07, "loss": 0.3574, "step": 23287 }, { "epoch": 1.6099550639474594, "grad_norm": 0.6394073732448021, "learning_rate": 9.650067015450497e-07, "loss": 0.249, "step": 23288 }, { "epoch": 1.6100241963359834, "grad_norm": 0.6757812052926704, "learning_rate": 9.64676159193601e-07, "loss": 0.2314, "step": 23289 }, { "epoch": 1.6100933287245074, "grad_norm": 0.6368625558729926, "learning_rate": 9.643456674177342e-07, "loss": 0.2305, "step": 23290 }, { "epoch": 1.6101624611130314, "grad_norm": 0.6834717751473467, "learning_rate": 9.640152262215935e-07, "loss": 0.333, "step": 23291 }, { "epoch": 1.6102315935015554, "grad_norm": 0.6844741912570962, "learning_rate": 9.636848356093182e-07, "loss": 0.2627, "step": 23292 }, { "epoch": 1.6103007258900797, "grad_norm": 0.7111436686629194, "learning_rate": 9.633544955850505e-07, "loss": 0.1896, "step": 23293 }, { "epoch": 1.6103698582786037, "grad_norm": 0.7158533688832781, "learning_rate": 9.6302420615293e-07, "loss": 0.3021, "step": 23294 }, { "epoch": 1.6104389906671277, "grad_norm": 0.6946248817576747, "learning_rate": 9.626939673170982e-07, "loss": 0.2921, "step": 23295 }, { "epoch": 1.6105081230556517, "grad_norm": 0.6731846170079254, "learning_rate": 9.623637790816913e-07, "loss": 0.2193, "step": 23296 }, { "epoch": 1.6105772554441757, "grad_norm": 0.7178898033690266, "learning_rate": 9.620336414508507e-07, "loss": 0.1874, "step": 23297 }, { "epoch": 1.6106463878326998, "grad_norm": 0.7118768452782719, "learning_rate": 9.617035544287107e-07, "loss": 0.3472, "step": 23298 }, { "epoch": 1.6107155202212238, "grad_norm": 0.8033985149254061, "learning_rate": 9.6137351801941e-07, "loss": 0.205, "step": 23299 }, { "epoch": 1.6107846526097478, "grad_norm": 0.7378140247293506, "learning_rate": 9.61043532227085e-07, "loss": 0.2488, "step": 23300 }, { "epoch": 1.6108537849982718, "grad_norm": 0.7089588617958907, "learning_rate": 9.607135970558729e-07, "loss": 0.2925, "step": 23301 }, { "epoch": 1.6109229173867958, "grad_norm": 0.7142835113597916, "learning_rate": 9.603837125099064e-07, "loss": 0.3062, "step": 23302 }, { "epoch": 1.6109920497753198, "grad_norm": 0.6499350897194276, "learning_rate": 9.600538785933217e-07, "loss": 0.2274, "step": 23303 }, { "epoch": 1.6110611821638439, "grad_norm": 0.6771153424438922, "learning_rate": 9.59724095310251e-07, "loss": 0.2529, "step": 23304 }, { "epoch": 1.6111303145523679, "grad_norm": 1.0102038987058801, "learning_rate": 9.5939436266483e-07, "loss": 0.304, "step": 23305 }, { "epoch": 1.611199446940892, "grad_norm": 0.7599089682151912, "learning_rate": 9.59064680661188e-07, "loss": 0.2719, "step": 23306 }, { "epoch": 1.611268579329416, "grad_norm": 0.6053640852590979, "learning_rate": 9.587350493034597e-07, "loss": 0.2413, "step": 23307 }, { "epoch": 1.61133771171794, "grad_norm": 0.7249262044602243, "learning_rate": 9.58405468595775e-07, "loss": 0.2662, "step": 23308 }, { "epoch": 1.611406844106464, "grad_norm": 0.670069318403093, "learning_rate": 9.580759385422666e-07, "loss": 0.3498, "step": 23309 }, { "epoch": 1.611475976494988, "grad_norm": 0.690035786760535, "learning_rate": 9.577464591470619e-07, "loss": 0.2684, "step": 23310 }, { "epoch": 1.611545108883512, "grad_norm": 0.7363055305750137, "learning_rate": 9.574170304142932e-07, "loss": 0.2534, "step": 23311 }, { "epoch": 1.611614241272036, "grad_norm": 0.6709710838939233, "learning_rate": 9.570876523480865e-07, "loss": 0.2938, "step": 23312 }, { "epoch": 1.61168337366056, "grad_norm": 0.6995731088968498, "learning_rate": 9.567583249525719e-07, "loss": 0.2431, "step": 23313 }, { "epoch": 1.611752506049084, "grad_norm": 0.6602056674204022, "learning_rate": 9.56429048231875e-07, "loss": 0.291, "step": 23314 }, { "epoch": 1.611821638437608, "grad_norm": 0.6975112402818139, "learning_rate": 9.56099822190124e-07, "loss": 0.2281, "step": 23315 }, { "epoch": 1.611890770826132, "grad_norm": 0.6712998490022906, "learning_rate": 9.557706468314455e-07, "loss": 0.3417, "step": 23316 }, { "epoch": 1.611959903214656, "grad_norm": 0.7099224234655234, "learning_rate": 9.554415221599656e-07, "loss": 0.2299, "step": 23317 }, { "epoch": 1.6120290356031801, "grad_norm": 0.6732972682278963, "learning_rate": 9.551124481798073e-07, "loss": 0.2762, "step": 23318 }, { "epoch": 1.6120981679917041, "grad_norm": 0.6924706487321374, "learning_rate": 9.547834248950972e-07, "loss": 0.3045, "step": 23319 }, { "epoch": 1.6121673003802282, "grad_norm": 0.6959461014298709, "learning_rate": 9.544544523099563e-07, "loss": 0.2599, "step": 23320 }, { "epoch": 1.6122364327687522, "grad_norm": 0.6271410487214245, "learning_rate": 9.541255304285097e-07, "loss": 0.2482, "step": 23321 }, { "epoch": 1.6123055651572762, "grad_norm": 0.7282399192259703, "learning_rate": 9.537966592548791e-07, "loss": 0.2057, "step": 23322 }, { "epoch": 1.6123746975458002, "grad_norm": 0.738653013537984, "learning_rate": 9.534678387931879e-07, "loss": 0.3227, "step": 23323 }, { "epoch": 1.6124438299343242, "grad_norm": 0.6735612676561, "learning_rate": 9.531390690475545e-07, "loss": 0.2454, "step": 23324 }, { "epoch": 1.6125129623228482, "grad_norm": 0.6473347280573232, "learning_rate": 9.528103500221025e-07, "loss": 0.2596, "step": 23325 }, { "epoch": 1.6125820947113723, "grad_norm": 0.6908222160524632, "learning_rate": 9.524816817209487e-07, "loss": 0.2876, "step": 23326 }, { "epoch": 1.6126512270998963, "grad_norm": 0.7444318091015454, "learning_rate": 9.521530641482157e-07, "loss": 0.3288, "step": 23327 }, { "epoch": 1.6127203594884203, "grad_norm": 0.703061989896108, "learning_rate": 9.518244973080187e-07, "loss": 0.2343, "step": 23328 }, { "epoch": 1.6127894918769443, "grad_norm": 0.6571382644046209, "learning_rate": 9.514959812044772e-07, "loss": 0.2619, "step": 23329 }, { "epoch": 1.6128586242654683, "grad_norm": 1.6230029782781634, "learning_rate": 9.511675158417088e-07, "loss": 0.3689, "step": 23330 }, { "epoch": 1.6129277566539924, "grad_norm": 0.681531215290728, "learning_rate": 9.508391012238316e-07, "loss": 0.2332, "step": 23331 }, { "epoch": 1.6129968890425164, "grad_norm": 0.6428128317925446, "learning_rate": 9.505107373549588e-07, "loss": 0.1772, "step": 23332 }, { "epoch": 1.6130660214310404, "grad_norm": 0.694164174785478, "learning_rate": 9.501824242392083e-07, "loss": 0.2855, "step": 23333 }, { "epoch": 1.6131351538195644, "grad_norm": 0.6858611248468534, "learning_rate": 9.498541618806928e-07, "loss": 0.3269, "step": 23334 }, { "epoch": 1.6132042862080884, "grad_norm": 0.6783370706361559, "learning_rate": 9.495259502835281e-07, "loss": 0.2229, "step": 23335 }, { "epoch": 1.6132734185966124, "grad_norm": 0.7712126117915652, "learning_rate": 9.491977894518262e-07, "loss": 0.225, "step": 23336 }, { "epoch": 1.6133425509851365, "grad_norm": 0.649695183805263, "learning_rate": 9.488696793897029e-07, "loss": 0.2983, "step": 23337 }, { "epoch": 1.6134116833736605, "grad_norm": 0.7148013265761487, "learning_rate": 9.485416201012675e-07, "loss": 0.3038, "step": 23338 }, { "epoch": 1.6134808157621845, "grad_norm": 0.6728955314589824, "learning_rate": 9.482136115906338e-07, "loss": 0.2178, "step": 23339 }, { "epoch": 1.6135499481507085, "grad_norm": 0.7103986230647087, "learning_rate": 9.478856538619102e-07, "loss": 0.2485, "step": 23340 }, { "epoch": 1.6136190805392325, "grad_norm": 0.6971504244994096, "learning_rate": 9.475577469192104e-07, "loss": 0.3195, "step": 23341 }, { "epoch": 1.6136882129277565, "grad_norm": 1.0116396900264455, "learning_rate": 9.472298907666405e-07, "loss": 0.1984, "step": 23342 }, { "epoch": 1.6137573453162806, "grad_norm": 0.6260763668088681, "learning_rate": 9.469020854083116e-07, "loss": 0.2181, "step": 23343 }, { "epoch": 1.6138264777048046, "grad_norm": 0.6759343949385925, "learning_rate": 9.465743308483322e-07, "loss": 0.3053, "step": 23344 }, { "epoch": 1.6138956100933286, "grad_norm": 0.7418381580347749, "learning_rate": 9.46246627090811e-07, "loss": 0.312, "step": 23345 }, { "epoch": 1.6139647424818526, "grad_norm": 0.8223147128630204, "learning_rate": 9.459189741398545e-07, "loss": 0.2615, "step": 23346 }, { "epoch": 1.6140338748703766, "grad_norm": 0.67240946782981, "learning_rate": 9.455913719995669e-07, "loss": 0.2565, "step": 23347 }, { "epoch": 1.6141030072589007, "grad_norm": 0.6610822845694729, "learning_rate": 9.452638206740577e-07, "loss": 0.2642, "step": 23348 }, { "epoch": 1.6141721396474247, "grad_norm": 0.757388242584255, "learning_rate": 9.449363201674294e-07, "loss": 0.1996, "step": 23349 }, { "epoch": 1.6142412720359487, "grad_norm": 0.6535498706615829, "learning_rate": 9.446088704837874e-07, "loss": 0.265, "step": 23350 }, { "epoch": 1.6143104044244727, "grad_norm": 0.7015068654545331, "learning_rate": 9.442814716272359e-07, "loss": 0.2732, "step": 23351 }, { "epoch": 1.6143795368129967, "grad_norm": 0.7574209305069254, "learning_rate": 9.439541236018801e-07, "loss": 0.3585, "step": 23352 }, { "epoch": 1.6144486692015207, "grad_norm": 0.7724786446415244, "learning_rate": 9.436268264118192e-07, "loss": 0.2031, "step": 23353 }, { "epoch": 1.614517801590045, "grad_norm": 0.7532626047466278, "learning_rate": 9.432995800611588e-07, "loss": 0.3066, "step": 23354 }, { "epoch": 1.614586933978569, "grad_norm": 0.7893264302712792, "learning_rate": 9.429723845539973e-07, "loss": 0.3568, "step": 23355 }, { "epoch": 1.614656066367093, "grad_norm": 0.9275325395014076, "learning_rate": 9.426452398944369e-07, "loss": 0.2542, "step": 23356 }, { "epoch": 1.614725198755617, "grad_norm": 1.0341291281505633, "learning_rate": 9.423181460865777e-07, "loss": 0.2219, "step": 23357 }, { "epoch": 1.614794331144141, "grad_norm": 0.9123364090108697, "learning_rate": 9.419911031345208e-07, "loss": 0.2651, "step": 23358 }, { "epoch": 1.614863463532665, "grad_norm": 0.7058163388165816, "learning_rate": 9.416641110423619e-07, "loss": 0.333, "step": 23359 }, { "epoch": 1.614932595921189, "grad_norm": 0.7110018043498068, "learning_rate": 9.413371698142026e-07, "loss": 0.2284, "step": 23360 }, { "epoch": 1.6150017283097131, "grad_norm": 0.7788416735622455, "learning_rate": 9.41010279454137e-07, "loss": 0.3108, "step": 23361 }, { "epoch": 1.6150708606982371, "grad_norm": 0.68461025957992, "learning_rate": 9.406834399662657e-07, "loss": 0.3104, "step": 23362 }, { "epoch": 1.6151399930867611, "grad_norm": 0.7140735997801073, "learning_rate": 9.403566513546819e-07, "loss": 0.3031, "step": 23363 }, { "epoch": 1.6152091254752852, "grad_norm": 0.6924864999878186, "learning_rate": 9.40029913623483e-07, "loss": 0.2561, "step": 23364 }, { "epoch": 1.6152782578638092, "grad_norm": 0.6925058320605006, "learning_rate": 9.397032267767637e-07, "loss": 0.2662, "step": 23365 }, { "epoch": 1.6153473902523332, "grad_norm": 0.6886076526017522, "learning_rate": 9.393765908186192e-07, "loss": 0.3314, "step": 23366 }, { "epoch": 1.6154165226408572, "grad_norm": 0.7173452839550087, "learning_rate": 9.39050005753142e-07, "loss": 0.2229, "step": 23367 }, { "epoch": 1.6154856550293812, "grad_norm": 0.6297766540610096, "learning_rate": 9.387234715844268e-07, "loss": 0.2466, "step": 23368 }, { "epoch": 1.6155547874179053, "grad_norm": 0.6880561437246949, "learning_rate": 9.383969883165644e-07, "loss": 0.3143, "step": 23369 }, { "epoch": 1.6156239198064293, "grad_norm": 0.714326543734851, "learning_rate": 9.380705559536469e-07, "loss": 0.3059, "step": 23370 }, { "epoch": 1.6156930521949533, "grad_norm": 0.639857582418582, "learning_rate": 9.377441744997668e-07, "loss": 0.2064, "step": 23371 }, { "epoch": 1.6157621845834773, "grad_norm": 0.7671738991359952, "learning_rate": 9.374178439590148e-07, "loss": 0.2363, "step": 23372 }, { "epoch": 1.6158313169720013, "grad_norm": 0.7694559167034446, "learning_rate": 9.370915643354788e-07, "loss": 0.3165, "step": 23373 }, { "epoch": 1.6159004493605256, "grad_norm": 4.331521355338464, "learning_rate": 9.367653356332513e-07, "loss": 0.2287, "step": 23374 }, { "epoch": 1.6159695817490496, "grad_norm": 0.7117797238507442, "learning_rate": 9.364391578564174e-07, "loss": 0.2152, "step": 23375 }, { "epoch": 1.6160387141375736, "grad_norm": 0.7349700760531086, "learning_rate": 9.361130310090683e-07, "loss": 0.2548, "step": 23376 }, { "epoch": 1.6161078465260976, "grad_norm": 0.7827408078814284, "learning_rate": 9.357869550952886e-07, "loss": 0.3267, "step": 23377 }, { "epoch": 1.6161769789146216, "grad_norm": 0.6560356964222357, "learning_rate": 9.35460930119167e-07, "loss": 0.2295, "step": 23378 }, { "epoch": 1.6162461113031457, "grad_norm": 0.6368284485125817, "learning_rate": 9.351349560847883e-07, "loss": 0.2204, "step": 23379 }, { "epoch": 1.6163152436916697, "grad_norm": 0.6370899310670782, "learning_rate": 9.348090329962406e-07, "loss": 0.3294, "step": 23380 }, { "epoch": 1.6163843760801937, "grad_norm": 0.6937762965361909, "learning_rate": 9.34483160857606e-07, "loss": 0.2869, "step": 23381 }, { "epoch": 1.6164535084687177, "grad_norm": 0.6601552408416921, "learning_rate": 9.341573396729709e-07, "loss": 0.284, "step": 23382 }, { "epoch": 1.6165226408572417, "grad_norm": 0.6658549346161958, "learning_rate": 9.338315694464161e-07, "loss": 0.2459, "step": 23383 }, { "epoch": 1.6165917732457658, "grad_norm": 0.7590786610029016, "learning_rate": 9.335058501820277e-07, "loss": 0.3355, "step": 23384 }, { "epoch": 1.6166609056342898, "grad_norm": 0.7744306130293971, "learning_rate": 9.331801818838854e-07, "loss": 0.2473, "step": 23385 }, { "epoch": 1.6167300380228138, "grad_norm": 0.5780150930249554, "learning_rate": 9.328545645560715e-07, "loss": 0.1858, "step": 23386 }, { "epoch": 1.6167991704113378, "grad_norm": 0.6843962189377056, "learning_rate": 9.325289982026681e-07, "loss": 0.3551, "step": 23387 }, { "epoch": 1.6168683027998618, "grad_norm": 0.7616907330775434, "learning_rate": 9.322034828277554e-07, "loss": 0.2796, "step": 23388 }, { "epoch": 1.6169374351883858, "grad_norm": 0.6225904883433804, "learning_rate": 9.31878018435412e-07, "loss": 0.2318, "step": 23389 }, { "epoch": 1.6170065675769099, "grad_norm": 0.796252776989589, "learning_rate": 9.315526050297191e-07, "loss": 0.2726, "step": 23390 }, { "epoch": 1.6170756999654339, "grad_norm": 0.6597470129202102, "learning_rate": 9.312272426147523e-07, "loss": 0.3065, "step": 23391 }, { "epoch": 1.617144832353958, "grad_norm": 0.7429610367929204, "learning_rate": 9.309019311945911e-07, "loss": 0.2817, "step": 23392 }, { "epoch": 1.617213964742482, "grad_norm": 0.6814812976159677, "learning_rate": 9.305766707733127e-07, "loss": 0.2527, "step": 23393 }, { "epoch": 1.617283097131006, "grad_norm": 0.6507441179711637, "learning_rate": 9.302514613549946e-07, "loss": 0.3307, "step": 23394 }, { "epoch": 1.61735222951953, "grad_norm": 0.7301637555170024, "learning_rate": 9.2992630294371e-07, "loss": 0.3313, "step": 23395 }, { "epoch": 1.617421361908054, "grad_norm": 0.663533568908347, "learning_rate": 9.29601195543538e-07, "loss": 0.2551, "step": 23396 }, { "epoch": 1.617490494296578, "grad_norm": 0.6644541728014131, "learning_rate": 9.292761391585492e-07, "loss": 0.2191, "step": 23397 }, { "epoch": 1.617559626685102, "grad_norm": 0.6806547533277711, "learning_rate": 9.289511337928208e-07, "loss": 0.3426, "step": 23398 }, { "epoch": 1.617628759073626, "grad_norm": 0.8346903588434584, "learning_rate": 9.286261794504231e-07, "loss": 0.2895, "step": 23399 }, { "epoch": 1.61769789146215, "grad_norm": 0.7016076527642106, "learning_rate": 9.283012761354315e-07, "loss": 0.2885, "step": 23400 }, { "epoch": 1.617767023850674, "grad_norm": 0.7089378277156863, "learning_rate": 9.279764238519162e-07, "loss": 0.2276, "step": 23401 }, { "epoch": 1.617836156239198, "grad_norm": 0.7124648397196052, "learning_rate": 9.276516226039517e-07, "loss": 0.3482, "step": 23402 }, { "epoch": 1.617905288627722, "grad_norm": 0.6937878514073139, "learning_rate": 9.273268723956047e-07, "loss": 0.2336, "step": 23403 }, { "epoch": 1.6179744210162461, "grad_norm": 0.7573056586381317, "learning_rate": 9.270021732309492e-07, "loss": 0.2502, "step": 23404 }, { "epoch": 1.6180435534047701, "grad_norm": 0.6601064838593695, "learning_rate": 9.266775251140508e-07, "loss": 0.2945, "step": 23405 }, { "epoch": 1.6181126857932941, "grad_norm": 0.7993009828929356, "learning_rate": 9.263529280489813e-07, "loss": 0.2689, "step": 23406 }, { "epoch": 1.6181818181818182, "grad_norm": 0.6420025197487976, "learning_rate": 9.260283820398075e-07, "loss": 0.2265, "step": 23407 }, { "epoch": 1.6182509505703422, "grad_norm": 0.6758714621386949, "learning_rate": 9.257038870905993e-07, "loss": 0.2516, "step": 23408 }, { "epoch": 1.6183200829588662, "grad_norm": 0.7628778610969353, "learning_rate": 9.253794432054202e-07, "loss": 0.3323, "step": 23409 }, { "epoch": 1.6183892153473902, "grad_norm": 0.6598389759070589, "learning_rate": 9.250550503883404e-07, "loss": 0.2296, "step": 23410 }, { "epoch": 1.6184583477359142, "grad_norm": 0.7285816135155755, "learning_rate": 9.247307086434215e-07, "loss": 0.2274, "step": 23411 }, { "epoch": 1.6185274801244383, "grad_norm": 0.7153561928368498, "learning_rate": 9.244064179747319e-07, "loss": 0.2648, "step": 23412 }, { "epoch": 1.6185966125129623, "grad_norm": 0.9206273029203607, "learning_rate": 9.240821783863335e-07, "loss": 0.2847, "step": 23413 }, { "epoch": 1.6186657449014863, "grad_norm": 0.6504709778454169, "learning_rate": 9.237579898822912e-07, "loss": 0.2338, "step": 23414 }, { "epoch": 1.6187348772900103, "grad_norm": 0.6714970894482143, "learning_rate": 9.234338524666692e-07, "loss": 0.2333, "step": 23415 }, { "epoch": 1.6188040096785343, "grad_norm": 0.7237392823877973, "learning_rate": 9.231097661435279e-07, "loss": 0.3581, "step": 23416 }, { "epoch": 1.6188731420670583, "grad_norm": 0.681367125955051, "learning_rate": 9.227857309169314e-07, "loss": 0.2027, "step": 23417 }, { "epoch": 1.6189422744555824, "grad_norm": 0.6425607327129781, "learning_rate": 9.224617467909381e-07, "loss": 0.268, "step": 23418 }, { "epoch": 1.6190114068441064, "grad_norm": 0.7995876861595006, "learning_rate": 9.221378137696114e-07, "loss": 0.2796, "step": 23419 }, { "epoch": 1.6190805392326304, "grad_norm": 0.7182712845258363, "learning_rate": 9.21813931857009e-07, "loss": 0.301, "step": 23420 }, { "epoch": 1.6191496716211544, "grad_norm": 0.7983430672301266, "learning_rate": 9.214901010571908e-07, "loss": 0.2555, "step": 23421 }, { "epoch": 1.6192188040096784, "grad_norm": 0.7083273493938753, "learning_rate": 9.211663213742155e-07, "loss": 0.2773, "step": 23422 }, { "epoch": 1.6192879363982025, "grad_norm": 0.7559178097911818, "learning_rate": 9.208425928121434e-07, "loss": 0.3385, "step": 23423 }, { "epoch": 1.6193570687867265, "grad_norm": 0.6861829760797243, "learning_rate": 9.205189153750277e-07, "loss": 0.2049, "step": 23424 }, { "epoch": 1.6194262011752505, "grad_norm": 0.6985552421404552, "learning_rate": 9.201952890669291e-07, "loss": 0.2186, "step": 23425 }, { "epoch": 1.6194953335637745, "grad_norm": 0.736698646527973, "learning_rate": 9.198717138919006e-07, "loss": 0.2729, "step": 23426 }, { "epoch": 1.6195644659522985, "grad_norm": 0.7359442016284988, "learning_rate": 9.195481898539988e-07, "loss": 0.3439, "step": 23427 }, { "epoch": 1.6196335983408225, "grad_norm": 0.6893392245720011, "learning_rate": 9.192247169572782e-07, "loss": 0.2114, "step": 23428 }, { "epoch": 1.6197027307293466, "grad_norm": 0.6829230055766133, "learning_rate": 9.189012952057946e-07, "loss": 0.2519, "step": 23429 }, { "epoch": 1.6197718631178706, "grad_norm": 0.6998554899777835, "learning_rate": 9.185779246035992e-07, "loss": 0.3139, "step": 23430 }, { "epoch": 1.6198409955063946, "grad_norm": 0.6904537660671912, "learning_rate": 9.182546051547469e-07, "loss": 0.2234, "step": 23431 }, { "epoch": 1.6199101278949186, "grad_norm": 0.6015215741738295, "learning_rate": 9.17931336863288e-07, "loss": 0.2159, "step": 23432 }, { "epoch": 1.6199792602834426, "grad_norm": 0.7337429953067351, "learning_rate": 9.176081197332759e-07, "loss": 0.2732, "step": 23433 }, { "epoch": 1.6200483926719667, "grad_norm": 0.7195434715668474, "learning_rate": 9.172849537687595e-07, "loss": 0.3601, "step": 23434 }, { "epoch": 1.620117525060491, "grad_norm": 0.7432598277604707, "learning_rate": 9.169618389737905e-07, "loss": 0.2235, "step": 23435 }, { "epoch": 1.620186657449015, "grad_norm": 1.2779959383503807, "learning_rate": 9.166387753524181e-07, "loss": 0.2137, "step": 23436 }, { "epoch": 1.620255789837539, "grad_norm": 0.722846914160671, "learning_rate": 9.163157629086927e-07, "loss": 0.3415, "step": 23437 }, { "epoch": 1.620324922226063, "grad_norm": 0.7184416602955549, "learning_rate": 9.159928016466607e-07, "loss": 0.2708, "step": 23438 }, { "epoch": 1.620394054614587, "grad_norm": 0.6829560938333205, "learning_rate": 9.156698915703716e-07, "loss": 0.2125, "step": 23439 }, { "epoch": 1.620463187003111, "grad_norm": 0.6514410679820037, "learning_rate": 9.153470326838709e-07, "loss": 0.2362, "step": 23440 }, { "epoch": 1.620532319391635, "grad_norm": 0.6827733638124117, "learning_rate": 9.150242249912051e-07, "loss": 0.3092, "step": 23441 }, { "epoch": 1.620601451780159, "grad_norm": 0.7097424693146186, "learning_rate": 9.147014684964212e-07, "loss": 0.2257, "step": 23442 }, { "epoch": 1.620670584168683, "grad_norm": 0.6815507139339946, "learning_rate": 9.143787632035645e-07, "loss": 0.2491, "step": 23443 }, { "epoch": 1.620739716557207, "grad_norm": 0.7026051554354997, "learning_rate": 9.140561091166784e-07, "loss": 0.2879, "step": 23444 }, { "epoch": 1.620808848945731, "grad_norm": 0.729734649144232, "learning_rate": 9.137335062398079e-07, "loss": 0.2881, "step": 23445 }, { "epoch": 1.620877981334255, "grad_norm": 0.673071075417811, "learning_rate": 9.134109545769942e-07, "loss": 0.2167, "step": 23446 }, { "epoch": 1.620947113722779, "grad_norm": 0.6812864137410988, "learning_rate": 9.130884541322832e-07, "loss": 0.2027, "step": 23447 }, { "epoch": 1.6210162461113031, "grad_norm": 0.7133403252044017, "learning_rate": 9.127660049097137e-07, "loss": 0.3574, "step": 23448 }, { "epoch": 1.6210853784998271, "grad_norm": 0.7559484233741023, "learning_rate": 9.12443606913328e-07, "loss": 0.2667, "step": 23449 }, { "epoch": 1.6211545108883512, "grad_norm": 0.6901788832263128, "learning_rate": 9.121212601471674e-07, "loss": 0.2218, "step": 23450 }, { "epoch": 1.6212236432768752, "grad_norm": 0.6813428732983412, "learning_rate": 9.117989646152725e-07, "loss": 0.2955, "step": 23451 }, { "epoch": 1.6212927756653992, "grad_norm": 3.2315657545134946, "learning_rate": 9.114767203216812e-07, "loss": 0.3236, "step": 23452 }, { "epoch": 1.6213619080539232, "grad_norm": 0.6602441130789256, "learning_rate": 9.111545272704337e-07, "loss": 0.204, "step": 23453 }, { "epoch": 1.6214310404424472, "grad_norm": 1.2094284570952827, "learning_rate": 9.108323854655659e-07, "loss": 0.2417, "step": 23454 }, { "epoch": 1.6215001728309715, "grad_norm": 0.6974383225438782, "learning_rate": 9.105102949111183e-07, "loss": 0.3603, "step": 23455 }, { "epoch": 1.6215693052194955, "grad_norm": 0.7313127000400828, "learning_rate": 9.101882556111252e-07, "loss": 0.2936, "step": 23456 }, { "epoch": 1.6216384376080195, "grad_norm": 0.5680008821178321, "learning_rate": 9.09866267569623e-07, "loss": 0.2076, "step": 23457 }, { "epoch": 1.6217075699965435, "grad_norm": 0.7384446318953883, "learning_rate": 9.095443307906487e-07, "loss": 0.2549, "step": 23458 }, { "epoch": 1.6217767023850675, "grad_norm": 0.6714682258657121, "learning_rate": 9.092224452782372e-07, "loss": 0.3213, "step": 23459 }, { "epoch": 1.6218458347735916, "grad_norm": 0.6772938978162056, "learning_rate": 9.089006110364207e-07, "loss": 0.2173, "step": 23460 }, { "epoch": 1.6219149671621156, "grad_norm": 0.767759065372568, "learning_rate": 9.085788280692359e-07, "loss": 0.3159, "step": 23461 }, { "epoch": 1.6219840995506396, "grad_norm": 0.697064882158649, "learning_rate": 9.082570963807125e-07, "loss": 0.3076, "step": 23462 }, { "epoch": 1.6220532319391636, "grad_norm": 0.7075382154529002, "learning_rate": 9.079354159748844e-07, "loss": 0.2792, "step": 23463 }, { "epoch": 1.6221223643276876, "grad_norm": 0.6748838888852288, "learning_rate": 9.076137868557833e-07, "loss": 0.2195, "step": 23464 }, { "epoch": 1.6221914967162117, "grad_norm": 0.678280571330746, "learning_rate": 9.072922090274411e-07, "loss": 0.2325, "step": 23465 }, { "epoch": 1.6222606291047357, "grad_norm": 0.7082717211209896, "learning_rate": 9.069706824938862e-07, "loss": 0.3621, "step": 23466 }, { "epoch": 1.6223297614932597, "grad_norm": 1.129475048192664, "learning_rate": 9.066492072591509e-07, "loss": 0.2307, "step": 23467 }, { "epoch": 1.6223988938817837, "grad_norm": 0.6889137545375035, "learning_rate": 9.063277833272616e-07, "loss": 0.2983, "step": 23468 }, { "epoch": 1.6224680262703077, "grad_norm": 0.6848749880078031, "learning_rate": 9.060064107022492e-07, "loss": 0.305, "step": 23469 }, { "epoch": 1.6225371586588317, "grad_norm": 0.6900361139026101, "learning_rate": 9.056850893881392e-07, "loss": 0.3322, "step": 23470 }, { "epoch": 1.6226062910473558, "grad_norm": 0.6538700275507691, "learning_rate": 9.053638193889596e-07, "loss": 0.2203, "step": 23471 }, { "epoch": 1.6226754234358798, "grad_norm": 0.6834117375925076, "learning_rate": 9.050426007087376e-07, "loss": 0.2661, "step": 23472 }, { "epoch": 1.6227445558244038, "grad_norm": 0.7503486618186017, "learning_rate": 9.047214333515003e-07, "loss": 0.3608, "step": 23473 }, { "epoch": 1.6228136882129278, "grad_norm": 0.6709606424686497, "learning_rate": 9.044003173212701e-07, "loss": 0.2481, "step": 23474 }, { "epoch": 1.6228828206014518, "grad_norm": 0.6816940335158557, "learning_rate": 9.040792526220743e-07, "loss": 0.2449, "step": 23475 }, { "epoch": 1.6229519529899759, "grad_norm": 1.2231570986561302, "learning_rate": 9.037582392579342e-07, "loss": 0.3039, "step": 23476 }, { "epoch": 1.6230210853784999, "grad_norm": 0.7080340776041874, "learning_rate": 9.034372772328743e-07, "loss": 0.3004, "step": 23477 }, { "epoch": 1.623090217767024, "grad_norm": 0.6565909551408159, "learning_rate": 9.031163665509174e-07, "loss": 0.2559, "step": 23478 }, { "epoch": 1.623159350155548, "grad_norm": 0.7351682824029019, "learning_rate": 9.027955072160871e-07, "loss": 0.2418, "step": 23479 }, { "epoch": 1.623228482544072, "grad_norm": 0.7827924253061814, "learning_rate": 9.024746992324035e-07, "loss": 0.3562, "step": 23480 }, { "epoch": 1.623297614932596, "grad_norm": 0.7253865969111947, "learning_rate": 9.021539426038861e-07, "loss": 0.303, "step": 23481 }, { "epoch": 1.62336674732112, "grad_norm": 0.6911904391974272, "learning_rate": 9.018332373345567e-07, "loss": 0.2129, "step": 23482 }, { "epoch": 1.623435879709644, "grad_norm": 0.773714172489064, "learning_rate": 9.015125834284333e-07, "loss": 0.2769, "step": 23483 }, { "epoch": 1.623505012098168, "grad_norm": 0.6755475500821952, "learning_rate": 9.011919808895359e-07, "loss": 0.3186, "step": 23484 }, { "epoch": 1.623574144486692, "grad_norm": 0.7045937725183771, "learning_rate": 9.008714297218818e-07, "loss": 0.2386, "step": 23485 }, { "epoch": 1.623643276875216, "grad_norm": 1.0323805712959466, "learning_rate": 9.005509299294906e-07, "loss": 0.2401, "step": 23486 }, { "epoch": 1.62371240926374, "grad_norm": 0.6838183631301381, "learning_rate": 9.002304815163765e-07, "loss": 0.3033, "step": 23487 }, { "epoch": 1.623781541652264, "grad_norm": 0.7798013262799075, "learning_rate": 8.999100844865583e-07, "loss": 0.2603, "step": 23488 }, { "epoch": 1.623850674040788, "grad_norm": 0.6448800235783935, "learning_rate": 8.995897388440489e-07, "loss": 0.2284, "step": 23489 }, { "epoch": 1.623919806429312, "grad_norm": 0.7407694634179791, "learning_rate": 8.992694445928657e-07, "loss": 0.2661, "step": 23490 }, { "epoch": 1.6239889388178361, "grad_norm": 0.6736061192642183, "learning_rate": 8.989492017370205e-07, "loss": 0.2967, "step": 23491 }, { "epoch": 1.6240580712063601, "grad_norm": 0.8059223079237297, "learning_rate": 8.98629010280529e-07, "loss": 0.2029, "step": 23492 }, { "epoch": 1.6241272035948842, "grad_norm": 0.6679888863762056, "learning_rate": 8.983088702274029e-07, "loss": 0.2349, "step": 23493 }, { "epoch": 1.6241963359834082, "grad_norm": 0.6891003529554551, "learning_rate": 8.979887815816574e-07, "loss": 0.3063, "step": 23494 }, { "epoch": 1.6242654683719322, "grad_norm": 0.7419556336246402, "learning_rate": 8.976687443473004e-07, "loss": 0.3241, "step": 23495 }, { "epoch": 1.6243346007604562, "grad_norm": 0.6635111537136169, "learning_rate": 8.973487585283458e-07, "loss": 0.207, "step": 23496 }, { "epoch": 1.6244037331489802, "grad_norm": 0.7764579968917161, "learning_rate": 8.970288241288022e-07, "loss": 0.2567, "step": 23497 }, { "epoch": 1.6244728655375043, "grad_norm": 0.6879318780921441, "learning_rate": 8.967089411526803e-07, "loss": 0.328, "step": 23498 }, { "epoch": 1.6245419979260283, "grad_norm": 0.699019055886961, "learning_rate": 8.963891096039884e-07, "loss": 0.283, "step": 23499 }, { "epoch": 1.6246111303145523, "grad_norm": 0.642389054074624, "learning_rate": 8.960693294867379e-07, "loss": 0.252, "step": 23500 }, { "epoch": 1.6246802627030763, "grad_norm": 0.7166722763758282, "learning_rate": 8.957496008049326e-07, "loss": 0.3349, "step": 23501 }, { "epoch": 1.6247493950916003, "grad_norm": 0.6932963983705989, "learning_rate": 8.954299235625836e-07, "loss": 0.3465, "step": 23502 }, { "epoch": 1.6248185274801243, "grad_norm": 0.7074819252247324, "learning_rate": 8.951102977636939e-07, "loss": 0.2216, "step": 23503 }, { "epoch": 1.6248876598686484, "grad_norm": 0.6995919248946874, "learning_rate": 8.947907234122727e-07, "loss": 0.237, "step": 23504 }, { "epoch": 1.6249567922571724, "grad_norm": 0.8613671284790066, "learning_rate": 8.944712005123218e-07, "loss": 0.3323, "step": 23505 }, { "epoch": 1.6250259246456964, "grad_norm": 0.6639013857167277, "learning_rate": 8.941517290678487e-07, "loss": 0.2834, "step": 23506 }, { "epoch": 1.6250950570342204, "grad_norm": 0.7012624655558427, "learning_rate": 8.93832309082856e-07, "loss": 0.2041, "step": 23507 }, { "epoch": 1.6251641894227444, "grad_norm": 0.6901580064300523, "learning_rate": 8.935129405613486e-07, "loss": 0.2557, "step": 23508 }, { "epoch": 1.6252333218112684, "grad_norm": 0.7528105986818611, "learning_rate": 8.931936235073268e-07, "loss": 0.3459, "step": 23509 }, { "epoch": 1.6253024541997925, "grad_norm": 0.8998941809686803, "learning_rate": 8.928743579247956e-07, "loss": 0.2722, "step": 23510 }, { "epoch": 1.6253715865883165, "grad_norm": 0.6391975004531929, "learning_rate": 8.925551438177532e-07, "loss": 0.2428, "step": 23511 }, { "epoch": 1.6254407189768405, "grad_norm": 0.6992068406010287, "learning_rate": 8.922359811902026e-07, "loss": 0.3437, "step": 23512 }, { "epoch": 1.6255098513653645, "grad_norm": 0.7219514248972616, "learning_rate": 8.919168700461428e-07, "loss": 0.329, "step": 23513 }, { "epoch": 1.6255789837538885, "grad_norm": 0.768072200456257, "learning_rate": 8.915978103895756e-07, "loss": 0.2267, "step": 23514 }, { "epoch": 1.6256481161424126, "grad_norm": 0.7251179598395127, "learning_rate": 8.912788022244961e-07, "loss": 0.2758, "step": 23515 }, { "epoch": 1.6257172485309368, "grad_norm": 0.7100614442886671, "learning_rate": 8.909598455549063e-07, "loss": 0.3908, "step": 23516 }, { "epoch": 1.6257863809194608, "grad_norm": 1.117509065029144, "learning_rate": 8.906409403848005e-07, "loss": 0.2437, "step": 23517 }, { "epoch": 1.6258555133079848, "grad_norm": 0.6502307906568122, "learning_rate": 8.903220867181783e-07, "loss": 0.2788, "step": 23518 }, { "epoch": 1.6259246456965089, "grad_norm": 0.7033149936602238, "learning_rate": 8.900032845590334e-07, "loss": 0.3232, "step": 23519 }, { "epoch": 1.6259937780850329, "grad_norm": 0.7378220620418612, "learning_rate": 8.896845339113625e-07, "loss": 0.3035, "step": 23520 }, { "epoch": 1.6260629104735569, "grad_norm": 0.6618289441400035, "learning_rate": 8.893658347791612e-07, "loss": 0.2499, "step": 23521 }, { "epoch": 1.626132042862081, "grad_norm": 0.6578043672635403, "learning_rate": 8.890471871664247e-07, "loss": 0.2382, "step": 23522 }, { "epoch": 1.626201175250605, "grad_norm": 2.4826029929620996, "learning_rate": 8.887285910771438e-07, "loss": 0.3251, "step": 23523 }, { "epoch": 1.626270307639129, "grad_norm": 0.7067016417478976, "learning_rate": 8.884100465153145e-07, "loss": 0.2166, "step": 23524 }, { "epoch": 1.626339440027653, "grad_norm": 0.6921342773606787, "learning_rate": 8.880915534849265e-07, "loss": 0.2341, "step": 23525 }, { "epoch": 1.626408572416177, "grad_norm": 0.6757534606794967, "learning_rate": 8.877731119899741e-07, "loss": 0.3092, "step": 23526 }, { "epoch": 1.626477704804701, "grad_norm": 0.6862477796953872, "learning_rate": 8.874547220344459e-07, "loss": 0.3165, "step": 23527 }, { "epoch": 1.626546837193225, "grad_norm": 0.8597001237782158, "learning_rate": 8.871363836223334e-07, "loss": 0.2501, "step": 23528 }, { "epoch": 1.626615969581749, "grad_norm": 0.7804144839132443, "learning_rate": 8.868180967576267e-07, "loss": 0.218, "step": 23529 }, { "epoch": 1.626685101970273, "grad_norm": 0.6874040093588337, "learning_rate": 8.864998614443166e-07, "loss": 0.3371, "step": 23530 }, { "epoch": 1.626754234358797, "grad_norm": 1.0253572450594322, "learning_rate": 8.86181677686388e-07, "loss": 0.261, "step": 23531 }, { "epoch": 1.626823366747321, "grad_norm": 0.6230541405014954, "learning_rate": 8.858635454878323e-07, "loss": 0.2215, "step": 23532 }, { "epoch": 1.626892499135845, "grad_norm": 0.6672752147165847, "learning_rate": 8.855454648526335e-07, "loss": 0.2597, "step": 23533 }, { "epoch": 1.6269616315243691, "grad_norm": 0.7646400065070647, "learning_rate": 8.852274357847795e-07, "loss": 0.2731, "step": 23534 }, { "epoch": 1.6270307639128931, "grad_norm": 0.7083629988999804, "learning_rate": 8.849094582882573e-07, "loss": 0.2448, "step": 23535 }, { "epoch": 1.6270998963014172, "grad_norm": 0.5968706975700966, "learning_rate": 8.845915323670517e-07, "loss": 0.1901, "step": 23536 }, { "epoch": 1.6271690286899414, "grad_norm": 0.6777456923137991, "learning_rate": 8.84273658025146e-07, "loss": 0.3414, "step": 23537 }, { "epoch": 1.6272381610784654, "grad_norm": 0.6684381391881544, "learning_rate": 8.839558352665268e-07, "loss": 0.2464, "step": 23538 }, { "epoch": 1.6273072934669894, "grad_norm": 0.6858326239087075, "learning_rate": 8.836380640951741e-07, "loss": 0.2513, "step": 23539 }, { "epoch": 1.6273764258555135, "grad_norm": 0.6837648640105265, "learning_rate": 8.833203445150739e-07, "loss": 0.2631, "step": 23540 }, { "epoch": 1.6274455582440375, "grad_norm": 0.8421029867141787, "learning_rate": 8.830026765302052e-07, "loss": 0.3912, "step": 23541 }, { "epoch": 1.6275146906325615, "grad_norm": 0.7068775728906209, "learning_rate": 8.826850601445508e-07, "loss": 0.2495, "step": 23542 }, { "epoch": 1.6275838230210855, "grad_norm": 0.644286141069737, "learning_rate": 8.823674953620914e-07, "loss": 0.2452, "step": 23543 }, { "epoch": 1.6276529554096095, "grad_norm": 0.6188267056057482, "learning_rate": 8.820499821868084e-07, "loss": 0.2748, "step": 23544 }, { "epoch": 1.6277220877981335, "grad_norm": 0.7368468435240813, "learning_rate": 8.817325206226802e-07, "loss": 0.285, "step": 23545 }, { "epoch": 1.6277912201866576, "grad_norm": 0.7340804873278254, "learning_rate": 8.814151106736841e-07, "loss": 0.1905, "step": 23546 }, { "epoch": 1.6278603525751816, "grad_norm": 0.6752167790655268, "learning_rate": 8.810977523438014e-07, "loss": 0.2398, "step": 23547 }, { "epoch": 1.6279294849637056, "grad_norm": 0.6950344581505069, "learning_rate": 8.807804456370045e-07, "loss": 0.4002, "step": 23548 }, { "epoch": 1.6279986173522296, "grad_norm": 0.8009502904921122, "learning_rate": 8.804631905572769e-07, "loss": 0.2341, "step": 23549 }, { "epoch": 1.6280677497407536, "grad_norm": 0.7018051514982869, "learning_rate": 8.801459871085899e-07, "loss": 0.2171, "step": 23550 }, { "epoch": 1.6281368821292777, "grad_norm": 0.8665163913460343, "learning_rate": 8.79828835294922e-07, "loss": 0.2852, "step": 23551 }, { "epoch": 1.6282060145178017, "grad_norm": 0.6797363855592881, "learning_rate": 8.795117351202459e-07, "loss": 0.3017, "step": 23552 }, { "epoch": 1.6282751469063257, "grad_norm": 0.6195849608743401, "learning_rate": 8.791946865885382e-07, "loss": 0.2182, "step": 23553 }, { "epoch": 1.6283442792948497, "grad_norm": 0.7895708961265906, "learning_rate": 8.788776897037693e-07, "loss": 0.3012, "step": 23554 }, { "epoch": 1.6284134116833737, "grad_norm": 0.7558093396385401, "learning_rate": 8.785607444699146e-07, "loss": 0.3552, "step": 23555 }, { "epoch": 1.6284825440718977, "grad_norm": 0.6999799316637858, "learning_rate": 8.782438508909457e-07, "loss": 0.2408, "step": 23556 }, { "epoch": 1.6285516764604218, "grad_norm": 0.6869116829133989, "learning_rate": 8.779270089708358e-07, "loss": 0.2726, "step": 23557 }, { "epoch": 1.6286208088489458, "grad_norm": 0.7094654454026083, "learning_rate": 8.776102187135538e-07, "loss": 0.261, "step": 23558 }, { "epoch": 1.6286899412374698, "grad_norm": 0.7151305497053472, "learning_rate": 8.77293480123072e-07, "loss": 0.3128, "step": 23559 }, { "epoch": 1.6287590736259938, "grad_norm": 0.7184731469275903, "learning_rate": 8.769767932033579e-07, "loss": 0.2253, "step": 23560 }, { "epoch": 1.6288282060145178, "grad_norm": 0.7696513034498736, "learning_rate": 8.766601579583827e-07, "loss": 0.2099, "step": 23561 }, { "epoch": 1.6288973384030419, "grad_norm": 0.6751838828649157, "learning_rate": 8.763435743921128e-07, "loss": 0.3127, "step": 23562 }, { "epoch": 1.6289664707915659, "grad_norm": 1.0916391416082893, "learning_rate": 8.760270425085177e-07, "loss": 0.2748, "step": 23563 }, { "epoch": 1.6290356031800899, "grad_norm": 0.6676322239374838, "learning_rate": 8.757105623115636e-07, "loss": 0.2198, "step": 23564 }, { "epoch": 1.629104735568614, "grad_norm": 0.7131429256287994, "learning_rate": 8.753941338052191e-07, "loss": 0.2237, "step": 23565 }, { "epoch": 1.629173867957138, "grad_norm": 0.7169312698732611, "learning_rate": 8.750777569934471e-07, "loss": 0.3734, "step": 23566 }, { "epoch": 1.629243000345662, "grad_norm": 0.6983293409832155, "learning_rate": 8.74761431880215e-07, "loss": 0.2098, "step": 23567 }, { "epoch": 1.629312132734186, "grad_norm": 0.6691322293391851, "learning_rate": 8.744451584694858e-07, "loss": 0.239, "step": 23568 }, { "epoch": 1.62938126512271, "grad_norm": 0.7111979640429918, "learning_rate": 8.741289367652239e-07, "loss": 0.3645, "step": 23569 }, { "epoch": 1.629450397511234, "grad_norm": 0.6996153815911599, "learning_rate": 8.738127667713931e-07, "loss": 0.292, "step": 23570 }, { "epoch": 1.629519529899758, "grad_norm": 0.6981661471505626, "learning_rate": 8.734966484919566e-07, "loss": 0.2156, "step": 23571 }, { "epoch": 1.629588662288282, "grad_norm": 0.7856763715395871, "learning_rate": 8.731805819308747e-07, "loss": 0.3062, "step": 23572 }, { "epoch": 1.629657794676806, "grad_norm": 0.6756779872318125, "learning_rate": 8.728645670921104e-07, "loss": 0.2925, "step": 23573 }, { "epoch": 1.62972692706533, "grad_norm": 0.6511634325138385, "learning_rate": 8.725486039796227e-07, "loss": 0.2234, "step": 23574 }, { "epoch": 1.629796059453854, "grad_norm": 0.7534910410625685, "learning_rate": 8.722326925973734e-07, "loss": 0.269, "step": 23575 }, { "epoch": 1.629865191842378, "grad_norm": 0.7501965347897207, "learning_rate": 8.719168329493199e-07, "loss": 0.3309, "step": 23576 }, { "epoch": 1.6299343242309021, "grad_norm": 0.7057197017724212, "learning_rate": 8.716010250394219e-07, "loss": 0.3572, "step": 23577 }, { "epoch": 1.6300034566194261, "grad_norm": 0.6995375824933517, "learning_rate": 8.712852688716378e-07, "loss": 0.2167, "step": 23578 }, { "epoch": 1.6300725890079502, "grad_norm": 0.6818466436708728, "learning_rate": 8.709695644499255e-07, "loss": 0.2326, "step": 23579 }, { "epoch": 1.6301417213964742, "grad_norm": 0.770506023229752, "learning_rate": 8.706539117782403e-07, "loss": 0.3015, "step": 23580 }, { "epoch": 1.6302108537849982, "grad_norm": 0.7101353368218012, "learning_rate": 8.703383108605407e-07, "loss": 0.2628, "step": 23581 }, { "epoch": 1.6302799861735222, "grad_norm": 0.6428374353836614, "learning_rate": 8.70022761700779e-07, "loss": 0.2254, "step": 23582 }, { "epoch": 1.6303491185620462, "grad_norm": 0.6957575931942143, "learning_rate": 8.69707264302912e-07, "loss": 0.2909, "step": 23583 }, { "epoch": 1.6304182509505702, "grad_norm": 0.9001933243089285, "learning_rate": 8.693918186708932e-07, "loss": 0.2935, "step": 23584 }, { "epoch": 1.6304873833390943, "grad_norm": 0.7448183911179608, "learning_rate": 8.69076424808678e-07, "loss": 0.1695, "step": 23585 }, { "epoch": 1.6305565157276183, "grad_norm": 0.8004532950842367, "learning_rate": 8.68761082720217e-07, "loss": 0.2721, "step": 23586 }, { "epoch": 1.6306256481161423, "grad_norm": 0.6716158639388752, "learning_rate": 8.684457924094641e-07, "loss": 0.3324, "step": 23587 }, { "epoch": 1.6306947805046663, "grad_norm": 0.7052579911476977, "learning_rate": 8.681305538803692e-07, "loss": 0.2685, "step": 23588 }, { "epoch": 1.6307639128931903, "grad_norm": 0.6106281463967322, "learning_rate": 8.67815367136885e-07, "loss": 0.19, "step": 23589 }, { "epoch": 1.6308330452817144, "grad_norm": 0.7660046084119402, "learning_rate": 8.675002321829595e-07, "loss": 0.2882, "step": 23590 }, { "epoch": 1.6309021776702384, "grad_norm": 1.0576414173354032, "learning_rate": 8.671851490225442e-07, "loss": 0.3553, "step": 23591 }, { "epoch": 1.6309713100587624, "grad_norm": 0.7020312251301546, "learning_rate": 8.668701176595879e-07, "loss": 0.237, "step": 23592 }, { "epoch": 1.6310404424472864, "grad_norm": 0.7790695721320262, "learning_rate": 8.665551380980397e-07, "loss": 0.2139, "step": 23593 }, { "epoch": 1.6311095748358104, "grad_norm": 0.724039450251308, "learning_rate": 8.662402103418455e-07, "loss": 0.3145, "step": 23594 }, { "epoch": 1.6311787072243344, "grad_norm": 0.659163478071367, "learning_rate": 8.659253343949541e-07, "loss": 0.2672, "step": 23595 }, { "epoch": 1.6312478396128585, "grad_norm": 0.7027161135699286, "learning_rate": 8.656105102613099e-07, "loss": 0.2422, "step": 23596 }, { "epoch": 1.6313169720013825, "grad_norm": 0.6928162340330077, "learning_rate": 8.652957379448612e-07, "loss": 0.2668, "step": 23597 }, { "epoch": 1.6313861043899067, "grad_norm": 0.7709482735458142, "learning_rate": 8.649810174495499e-07, "loss": 0.3636, "step": 23598 }, { "epoch": 1.6314552367784307, "grad_norm": 0.7266926272259284, "learning_rate": 8.646663487793228e-07, "loss": 0.2709, "step": 23599 }, { "epoch": 1.6315243691669548, "grad_norm": 0.6271618069784024, "learning_rate": 8.643517319381234e-07, "loss": 0.1808, "step": 23600 }, { "epoch": 1.6315935015554788, "grad_norm": 0.6934233916960753, "learning_rate": 8.640371669298953e-07, "loss": 0.2789, "step": 23601 }, { "epoch": 1.6316626339440028, "grad_norm": 0.6859593374106623, "learning_rate": 8.637226537585796e-07, "loss": 0.3784, "step": 23602 }, { "epoch": 1.6317317663325268, "grad_norm": 0.6980440804348087, "learning_rate": 8.634081924281201e-07, "loss": 0.2527, "step": 23603 }, { "epoch": 1.6318008987210508, "grad_norm": 0.6506326248592667, "learning_rate": 8.630937829424552e-07, "loss": 0.2532, "step": 23604 }, { "epoch": 1.6318700311095748, "grad_norm": 0.6592044703902435, "learning_rate": 8.627794253055271e-07, "loss": 0.331, "step": 23605 }, { "epoch": 1.6319391634980989, "grad_norm": 0.6943038307874261, "learning_rate": 8.624651195212764e-07, "loss": 0.2448, "step": 23606 }, { "epoch": 1.6320082958866229, "grad_norm": 0.6674748218080779, "learning_rate": 8.621508655936422e-07, "loss": 0.2313, "step": 23607 }, { "epoch": 1.632077428275147, "grad_norm": 0.6274124765257193, "learning_rate": 8.618366635265618e-07, "loss": 0.224, "step": 23608 }, { "epoch": 1.632146560663671, "grad_norm": 0.8687181994206821, "learning_rate": 8.615225133239746e-07, "loss": 0.3058, "step": 23609 }, { "epoch": 1.632215693052195, "grad_norm": 0.6960902211944259, "learning_rate": 8.612084149898165e-07, "loss": 0.2235, "step": 23610 }, { "epoch": 1.632284825440719, "grad_norm": 0.697450409164256, "learning_rate": 8.608943685280264e-07, "loss": 0.2524, "step": 23611 }, { "epoch": 1.632353957829243, "grad_norm": 0.6482725063079793, "learning_rate": 8.605803739425367e-07, "loss": 0.316, "step": 23612 }, { "epoch": 1.632423090217767, "grad_norm": 0.7392952428889314, "learning_rate": 8.602664312372855e-07, "loss": 0.2403, "step": 23613 }, { "epoch": 1.632492222606291, "grad_norm": 0.7852748362027602, "learning_rate": 8.599525404162079e-07, "loss": 0.2228, "step": 23614 }, { "epoch": 1.632561354994815, "grad_norm": 0.6997815679029344, "learning_rate": 8.596387014832358e-07, "loss": 0.2399, "step": 23615 }, { "epoch": 1.632630487383339, "grad_norm": 0.6925184910822866, "learning_rate": 8.593249144423055e-07, "loss": 0.3168, "step": 23616 }, { "epoch": 1.632699619771863, "grad_norm": 0.6874759591910515, "learning_rate": 8.59011179297346e-07, "loss": 0.2208, "step": 23617 }, { "epoch": 1.6327687521603873, "grad_norm": 0.6676718248654923, "learning_rate": 8.586974960522931e-07, "loss": 0.2263, "step": 23618 }, { "epoch": 1.6328378845489113, "grad_norm": 0.6987192530164004, "learning_rate": 8.583838647110742e-07, "loss": 0.3451, "step": 23619 }, { "epoch": 1.6329070169374353, "grad_norm": 0.6956263339300117, "learning_rate": 8.580702852776246e-07, "loss": 0.3079, "step": 23620 }, { "epoch": 1.6329761493259594, "grad_norm": 0.739734534692847, "learning_rate": 8.577567577558716e-07, "loss": 0.1929, "step": 23621 }, { "epoch": 1.6330452817144834, "grad_norm": 0.6861566003830321, "learning_rate": 8.574432821497463e-07, "loss": 0.2793, "step": 23622 }, { "epoch": 1.6331144141030074, "grad_norm": 0.7326769515976929, "learning_rate": 8.571298584631759e-07, "loss": 0.367, "step": 23623 }, { "epoch": 1.6331835464915314, "grad_norm": 0.7473825150367097, "learning_rate": 8.568164867000905e-07, "loss": 0.2547, "step": 23624 }, { "epoch": 1.6332526788800554, "grad_norm": 0.6256661097117607, "learning_rate": 8.565031668644153e-07, "loss": 0.2223, "step": 23625 }, { "epoch": 1.6333218112685794, "grad_norm": 0.6725686403560676, "learning_rate": 8.561898989600786e-07, "loss": 0.3008, "step": 23626 }, { "epoch": 1.6333909436571035, "grad_norm": 0.8584798424196859, "learning_rate": 8.558766829910065e-07, "loss": 0.3391, "step": 23627 }, { "epoch": 1.6334600760456275, "grad_norm": 0.6655194614556947, "learning_rate": 8.555635189611256e-07, "loss": 0.2698, "step": 23628 }, { "epoch": 1.6335292084341515, "grad_norm": 0.7194372981015127, "learning_rate": 8.55250406874359e-07, "loss": 0.243, "step": 23629 }, { "epoch": 1.6335983408226755, "grad_norm": 0.6526831280484303, "learning_rate": 8.549373467346334e-07, "loss": 0.3044, "step": 23630 }, { "epoch": 1.6336674732111995, "grad_norm": 0.8890049400326965, "learning_rate": 8.546243385458691e-07, "loss": 0.3005, "step": 23631 }, { "epoch": 1.6337366055997236, "grad_norm": 0.6486368091151893, "learning_rate": 8.543113823119925e-07, "loss": 0.2239, "step": 23632 }, { "epoch": 1.6338057379882476, "grad_norm": 0.6823970988618576, "learning_rate": 8.539984780369231e-07, "loss": 0.2333, "step": 23633 }, { "epoch": 1.6338748703767716, "grad_norm": 0.7298101577551148, "learning_rate": 8.53685625724584e-07, "loss": 0.2962, "step": 23634 }, { "epoch": 1.6339440027652956, "grad_norm": 0.7632743245620679, "learning_rate": 8.533728253788964e-07, "loss": 0.2173, "step": 23635 }, { "epoch": 1.6340131351538196, "grad_norm": 0.6906054848155322, "learning_rate": 8.530600770037811e-07, "loss": 0.245, "step": 23636 }, { "epoch": 1.6340822675423436, "grad_norm": 0.703250511629593, "learning_rate": 8.527473806031566e-07, "loss": 0.3507, "step": 23637 }, { "epoch": 1.6341513999308677, "grad_norm": 0.7096031833704742, "learning_rate": 8.524347361809432e-07, "loss": 0.3284, "step": 23638 }, { "epoch": 1.6342205323193917, "grad_norm": 0.636617744641024, "learning_rate": 8.521221437410581e-07, "loss": 0.2313, "step": 23639 }, { "epoch": 1.6342896647079157, "grad_norm": 0.7303891572702317, "learning_rate": 8.51809603287419e-07, "loss": 0.2726, "step": 23640 }, { "epoch": 1.6343587970964397, "grad_norm": 0.779012897339385, "learning_rate": 8.514971148239442e-07, "loss": 0.3568, "step": 23641 }, { "epoch": 1.6344279294849637, "grad_norm": 0.7052911960215626, "learning_rate": 8.511846783545513e-07, "loss": 0.2536, "step": 23642 }, { "epoch": 1.6344970618734878, "grad_norm": 0.635059706233571, "learning_rate": 8.508722938831526e-07, "loss": 0.2164, "step": 23643 }, { "epoch": 1.6345661942620118, "grad_norm": 0.7302843598365433, "learning_rate": 8.505599614136667e-07, "loss": 0.3301, "step": 23644 }, { "epoch": 1.6346353266505358, "grad_norm": 0.7221120810921824, "learning_rate": 8.50247680950006e-07, "loss": 0.3075, "step": 23645 }, { "epoch": 1.6347044590390598, "grad_norm": 0.65759549235738, "learning_rate": 8.499354524960857e-07, "loss": 0.2087, "step": 23646 }, { "epoch": 1.6347735914275838, "grad_norm": 0.8616329720457219, "learning_rate": 8.496232760558171e-07, "loss": 0.2595, "step": 23647 }, { "epoch": 1.6348427238161078, "grad_norm": 0.7300177083409587, "learning_rate": 8.493111516331143e-07, "loss": 0.3876, "step": 23648 }, { "epoch": 1.6349118562046319, "grad_norm": 0.680524258846963, "learning_rate": 8.489990792318886e-07, "loss": 0.2464, "step": 23649 }, { "epoch": 1.6349809885931559, "grad_norm": 0.7037217547879726, "learning_rate": 8.486870588560536e-07, "loss": 0.2316, "step": 23650 }, { "epoch": 1.63505012098168, "grad_norm": 0.878264283105558, "learning_rate": 8.483750905095162e-07, "loss": 0.2542, "step": 23651 }, { "epoch": 1.635119253370204, "grad_norm": 0.7362525124900937, "learning_rate": 8.480631741961898e-07, "loss": 0.3115, "step": 23652 }, { "epoch": 1.635188385758728, "grad_norm": 0.726386019822113, "learning_rate": 8.477513099199802e-07, "loss": 0.2736, "step": 23653 }, { "epoch": 1.635257518147252, "grad_norm": 0.6290050310378241, "learning_rate": 8.474394976847994e-07, "loss": 0.2164, "step": 23654 }, { "epoch": 1.635326650535776, "grad_norm": 0.6450459397307478, "learning_rate": 8.471277374945513e-07, "loss": 0.3757, "step": 23655 }, { "epoch": 1.6353957829243, "grad_norm": 0.6929524678066253, "learning_rate": 8.468160293531486e-07, "loss": 0.2321, "step": 23656 }, { "epoch": 1.635464915312824, "grad_norm": 0.6693512040238555, "learning_rate": 8.465043732644939e-07, "loss": 0.204, "step": 23657 }, { "epoch": 1.635534047701348, "grad_norm": 0.6527493421018524, "learning_rate": 8.461927692324961e-07, "loss": 0.2709, "step": 23658 }, { "epoch": 1.635603180089872, "grad_norm": 0.7166017331027168, "learning_rate": 8.458812172610581e-07, "loss": 0.3601, "step": 23659 }, { "epoch": 1.635672312478396, "grad_norm": 0.8932716558192387, "learning_rate": 8.455697173540866e-07, "loss": 0.266, "step": 23660 }, { "epoch": 1.63574144486692, "grad_norm": 0.6878236284744057, "learning_rate": 8.452582695154832e-07, "loss": 0.2023, "step": 23661 }, { "epoch": 1.635810577255444, "grad_norm": 0.7627510405845019, "learning_rate": 8.449468737491534e-07, "loss": 0.3104, "step": 23662 }, { "epoch": 1.6358797096439681, "grad_norm": 0.8358071790727691, "learning_rate": 8.446355300589992e-07, "loss": 0.3188, "step": 23663 }, { "epoch": 1.6359488420324921, "grad_norm": 0.6959773603453807, "learning_rate": 8.443242384489248e-07, "loss": 0.2662, "step": 23664 }, { "epoch": 1.6360179744210162, "grad_norm": 0.7169914081012952, "learning_rate": 8.440129989228279e-07, "loss": 0.229, "step": 23665 }, { "epoch": 1.6360871068095402, "grad_norm": 0.8803549762346221, "learning_rate": 8.437018114846135e-07, "loss": 0.3359, "step": 23666 }, { "epoch": 1.6361562391980642, "grad_norm": 0.6887725975650248, "learning_rate": 8.433906761381778e-07, "loss": 0.3136, "step": 23667 }, { "epoch": 1.6362253715865882, "grad_norm": 0.6479986696529496, "learning_rate": 8.430795928874236e-07, "loss": 0.1828, "step": 23668 }, { "epoch": 1.6362945039751122, "grad_norm": 0.7271028617042998, "learning_rate": 8.427685617362469e-07, "loss": 0.3074, "step": 23669 }, { "epoch": 1.6363636363636362, "grad_norm": 0.6644073070653176, "learning_rate": 8.424575826885478e-07, "loss": 0.2912, "step": 23670 }, { "epoch": 1.6364327687521603, "grad_norm": 0.6509540295405328, "learning_rate": 8.421466557482233e-07, "loss": 0.2286, "step": 23671 }, { "epoch": 1.6365019011406843, "grad_norm": 0.6682450606922664, "learning_rate": 8.418357809191713e-07, "loss": 0.2174, "step": 23672 }, { "epoch": 1.6365710335292083, "grad_norm": 0.7294397715997448, "learning_rate": 8.415249582052865e-07, "loss": 0.3574, "step": 23673 }, { "epoch": 1.6366401659177323, "grad_norm": 1.071363752335688, "learning_rate": 8.412141876104657e-07, "loss": 0.2525, "step": 23674 }, { "epoch": 1.6367092983062563, "grad_norm": 0.674543745312812, "learning_rate": 8.409034691386026e-07, "loss": 0.258, "step": 23675 }, { "epoch": 1.6367784306947804, "grad_norm": 0.744483193535237, "learning_rate": 8.405928027935928e-07, "loss": 0.2924, "step": 23676 }, { "epoch": 1.6368475630833044, "grad_norm": 0.7311923243186564, "learning_rate": 8.402821885793288e-07, "loss": 0.3634, "step": 23677 }, { "epoch": 1.6369166954718284, "grad_norm": 0.8134753068148013, "learning_rate": 8.399716264997054e-07, "loss": 0.2288, "step": 23678 }, { "epoch": 1.6369858278603526, "grad_norm": 0.6217007870755958, "learning_rate": 8.396611165586138e-07, "loss": 0.2001, "step": 23679 }, { "epoch": 1.6370549602488766, "grad_norm": 0.7542026520882831, "learning_rate": 8.393506587599448e-07, "loss": 0.3006, "step": 23680 }, { "epoch": 1.6371240926374007, "grad_norm": 0.7567855342168688, "learning_rate": 8.390402531075908e-07, "loss": 0.2776, "step": 23681 }, { "epoch": 1.6371932250259247, "grad_norm": 0.7347634268666454, "learning_rate": 8.387298996054411e-07, "loss": 0.2735, "step": 23682 }, { "epoch": 1.6372623574144487, "grad_norm": 0.6807073815953245, "learning_rate": 8.384195982573856e-07, "loss": 0.24, "step": 23683 }, { "epoch": 1.6373314898029727, "grad_norm": 0.7169036319828967, "learning_rate": 8.381093490673142e-07, "loss": 0.2878, "step": 23684 }, { "epoch": 1.6374006221914967, "grad_norm": 0.6997872261871175, "learning_rate": 8.37799152039116e-07, "loss": 0.2464, "step": 23685 }, { "epoch": 1.6374697545800208, "grad_norm": 0.7032060987460659, "learning_rate": 8.374890071766761e-07, "loss": 0.2495, "step": 23686 }, { "epoch": 1.6375388869685448, "grad_norm": 0.7107114046576972, "learning_rate": 8.371789144838844e-07, "loss": 0.3127, "step": 23687 }, { "epoch": 1.6376080193570688, "grad_norm": 0.733527102523886, "learning_rate": 8.368688739646252e-07, "loss": 0.2934, "step": 23688 }, { "epoch": 1.6376771517455928, "grad_norm": 1.0189904864011645, "learning_rate": 8.365588856227863e-07, "loss": 0.1801, "step": 23689 }, { "epoch": 1.6377462841341168, "grad_norm": 0.77898149714295, "learning_rate": 8.362489494622494e-07, "loss": 0.2762, "step": 23690 }, { "epoch": 1.6378154165226408, "grad_norm": 0.6924018424059695, "learning_rate": 8.359390654869037e-07, "loss": 0.3092, "step": 23691 }, { "epoch": 1.6378845489111649, "grad_norm": 0.8022504557125779, "learning_rate": 8.356292337006294e-07, "loss": 0.2352, "step": 23692 }, { "epoch": 1.6379536812996889, "grad_norm": 0.6429361112647825, "learning_rate": 8.353194541073123e-07, "loss": 0.2708, "step": 23693 }, { "epoch": 1.638022813688213, "grad_norm": 0.910104473500975, "learning_rate": 8.350097267108326e-07, "loss": 0.3538, "step": 23694 }, { "epoch": 1.638091946076737, "grad_norm": 0.6967769573338823, "learning_rate": 8.347000515150744e-07, "loss": 0.3294, "step": 23695 }, { "epoch": 1.638161078465261, "grad_norm": 0.837791859129694, "learning_rate": 8.343904285239163e-07, "loss": 0.2807, "step": 23696 }, { "epoch": 1.638230210853785, "grad_norm": 0.810678296641527, "learning_rate": 8.340808577412401e-07, "loss": 0.2314, "step": 23697 }, { "epoch": 1.638299343242309, "grad_norm": 0.7763258824680653, "learning_rate": 8.337713391709268e-07, "loss": 0.3264, "step": 23698 }, { "epoch": 1.6383684756308332, "grad_norm": 0.7460829409926832, "learning_rate": 8.334618728168553e-07, "loss": 0.2951, "step": 23699 }, { "epoch": 1.6384376080193572, "grad_norm": 0.664780586244562, "learning_rate": 8.331524586829027e-07, "loss": 0.2527, "step": 23700 }, { "epoch": 1.6385067404078812, "grad_norm": 0.6688868418466585, "learning_rate": 8.32843096772949e-07, "loss": 0.2929, "step": 23701 }, { "epoch": 1.6385758727964053, "grad_norm": 0.7041807782114422, "learning_rate": 8.325337870908695e-07, "loss": 0.2487, "step": 23702 }, { "epoch": 1.6386450051849293, "grad_norm": 0.6897731063316455, "learning_rate": 8.322245296405429e-07, "loss": 0.2334, "step": 23703 }, { "epoch": 1.6387141375734533, "grad_norm": 0.6800939250825154, "learning_rate": 8.319153244258427e-07, "loss": 0.2232, "step": 23704 }, { "epoch": 1.6387832699619773, "grad_norm": 0.6565944271559055, "learning_rate": 8.316061714506457e-07, "loss": 0.2766, "step": 23705 }, { "epoch": 1.6388524023505013, "grad_norm": 0.6981850882128648, "learning_rate": 8.312970707188267e-07, "loss": 0.2775, "step": 23706 }, { "epoch": 1.6389215347390254, "grad_norm": 0.7160359495179386, "learning_rate": 8.309880222342609e-07, "loss": 0.2325, "step": 23707 }, { "epoch": 1.6389906671275494, "grad_norm": 0.662177516598983, "learning_rate": 8.306790260008185e-07, "loss": 0.2789, "step": 23708 }, { "epoch": 1.6390597995160734, "grad_norm": 0.7530858707987288, "learning_rate": 8.303700820223759e-07, "loss": 0.3538, "step": 23709 }, { "epoch": 1.6391289319045974, "grad_norm": 0.6553616299812359, "learning_rate": 8.300611903028016e-07, "loss": 0.2436, "step": 23710 }, { "epoch": 1.6391980642931214, "grad_norm": 0.6874124498094283, "learning_rate": 8.297523508459687e-07, "loss": 0.2859, "step": 23711 }, { "epoch": 1.6392671966816454, "grad_norm": 0.7482203475061733, "learning_rate": 8.294435636557486e-07, "loss": 0.3826, "step": 23712 }, { "epoch": 1.6393363290701695, "grad_norm": 0.7270576526862427, "learning_rate": 8.291348287360112e-07, "loss": 0.2814, "step": 23713 }, { "epoch": 1.6394054614586935, "grad_norm": 0.7843357680624797, "learning_rate": 8.288261460906244e-07, "loss": 0.2324, "step": 23714 }, { "epoch": 1.6394745938472175, "grad_norm": 0.7171565837122449, "learning_rate": 8.285175157234593e-07, "loss": 0.2345, "step": 23715 }, { "epoch": 1.6395437262357415, "grad_norm": 0.6778687938852246, "learning_rate": 8.282089376383817e-07, "loss": 0.3654, "step": 23716 }, { "epoch": 1.6396128586242655, "grad_norm": 0.6616572736100814, "learning_rate": 8.279004118392614e-07, "loss": 0.2379, "step": 23717 }, { "epoch": 1.6396819910127896, "grad_norm": 0.6235429225719072, "learning_rate": 8.27591938329963e-07, "loss": 0.2183, "step": 23718 }, { "epoch": 1.6397511234013136, "grad_norm": 0.7264156296723255, "learning_rate": 8.272835171143534e-07, "loss": 0.3092, "step": 23719 }, { "epoch": 1.6398202557898376, "grad_norm": 0.7336514026660618, "learning_rate": 8.269751481962979e-07, "loss": 0.2858, "step": 23720 }, { "epoch": 1.6398893881783616, "grad_norm": 1.2022798982276928, "learning_rate": 8.266668315796638e-07, "loss": 0.252, "step": 23721 }, { "epoch": 1.6399585205668856, "grad_norm": 0.7088314290885042, "learning_rate": 8.263585672683116e-07, "loss": 0.2209, "step": 23722 }, { "epoch": 1.6400276529554096, "grad_norm": 0.737615262334326, "learning_rate": 8.260503552661081e-07, "loss": 0.3575, "step": 23723 }, { "epoch": 1.6400967853439337, "grad_norm": 0.7082274105806942, "learning_rate": 8.257421955769135e-07, "loss": 0.2256, "step": 23724 }, { "epoch": 1.6401659177324577, "grad_norm": 0.6313559957161813, "learning_rate": 8.25434088204592e-07, "loss": 0.2152, "step": 23725 }, { "epoch": 1.6402350501209817, "grad_norm": 0.7277866923592189, "learning_rate": 8.25126033153002e-07, "loss": 0.3051, "step": 23726 }, { "epoch": 1.6403041825095057, "grad_norm": 0.7385468793471951, "learning_rate": 8.248180304260095e-07, "loss": 0.3386, "step": 23727 }, { "epoch": 1.6403733148980297, "grad_norm": 0.7281609311428894, "learning_rate": 8.245100800274703e-07, "loss": 0.248, "step": 23728 }, { "epoch": 1.6404424472865538, "grad_norm": 0.7122829219358844, "learning_rate": 8.242021819612478e-07, "loss": 0.2474, "step": 23729 }, { "epoch": 1.6405115796750778, "grad_norm": 0.8346727149843046, "learning_rate": 8.238943362311969e-07, "loss": 0.3703, "step": 23730 }, { "epoch": 1.6405807120636018, "grad_norm": 0.7606406353521596, "learning_rate": 8.235865428411793e-07, "loss": 0.3247, "step": 23731 }, { "epoch": 1.6406498444521258, "grad_norm": 0.683129949205764, "learning_rate": 8.232788017950505e-07, "loss": 0.2457, "step": 23732 }, { "epoch": 1.6407189768406498, "grad_norm": 0.6908050904468995, "learning_rate": 8.229711130966678e-07, "loss": 0.264, "step": 23733 }, { "epoch": 1.6407881092291738, "grad_norm": 0.735026382833245, "learning_rate": 8.226634767498882e-07, "loss": 0.3383, "step": 23734 }, { "epoch": 1.6408572416176979, "grad_norm": 0.7309914818162452, "learning_rate": 8.223558927585679e-07, "loss": 0.2371, "step": 23735 }, { "epoch": 1.6409263740062219, "grad_norm": 0.6835580481213712, "learning_rate": 8.220483611265601e-07, "loss": 0.3295, "step": 23736 }, { "epoch": 1.640995506394746, "grad_norm": 0.6636191298997011, "learning_rate": 8.217408818577216e-07, "loss": 0.309, "step": 23737 }, { "epoch": 1.64106463878327, "grad_norm": 0.7062724820045909, "learning_rate": 8.214334549559034e-07, "loss": 0.2844, "step": 23738 }, { "epoch": 1.641133771171794, "grad_norm": 0.6817516185316324, "learning_rate": 8.21126080424961e-07, "loss": 0.2479, "step": 23739 }, { "epoch": 1.641202903560318, "grad_norm": 0.7941624593680681, "learning_rate": 8.208187582687444e-07, "loss": 0.2778, "step": 23740 }, { "epoch": 1.641272035948842, "grad_norm": 0.6946550210986921, "learning_rate": 8.205114884911069e-07, "loss": 0.333, "step": 23741 }, { "epoch": 1.641341168337366, "grad_norm": 0.731979770519901, "learning_rate": 8.202042710958991e-07, "loss": 0.2518, "step": 23742 }, { "epoch": 1.64141030072589, "grad_norm": 0.6346827623811029, "learning_rate": 8.198971060869726e-07, "loss": 0.2334, "step": 23743 }, { "epoch": 1.641479433114414, "grad_norm": 1.173195555476155, "learning_rate": 8.19589993468175e-07, "loss": 0.326, "step": 23744 }, { "epoch": 1.641548565502938, "grad_norm": 0.715245845685088, "learning_rate": 8.192829332433577e-07, "loss": 0.2628, "step": 23745 }, { "epoch": 1.641617697891462, "grad_norm": 0.7479846996850638, "learning_rate": 8.18975925416367e-07, "loss": 0.2266, "step": 23746 }, { "epoch": 1.641686830279986, "grad_norm": 0.821841565401526, "learning_rate": 8.186689699910516e-07, "loss": 0.2586, "step": 23747 }, { "epoch": 1.64175596266851, "grad_norm": 0.7552085457209244, "learning_rate": 8.183620669712605e-07, "loss": 0.3695, "step": 23748 }, { "epoch": 1.6418250950570341, "grad_norm": 0.7417288911063864, "learning_rate": 8.180552163608363e-07, "loss": 0.2477, "step": 23749 }, { "epoch": 1.6418942274455581, "grad_norm": 0.7167689798273668, "learning_rate": 8.177484181636286e-07, "loss": 0.2165, "step": 23750 }, { "epoch": 1.6419633598340821, "grad_norm": 0.7450854978774141, "learning_rate": 8.1744167238348e-07, "loss": 0.3225, "step": 23751 }, { "epoch": 1.6420324922226062, "grad_norm": 0.724447686021763, "learning_rate": 8.171349790242367e-07, "loss": 0.3135, "step": 23752 }, { "epoch": 1.6421016246111302, "grad_norm": 0.6423178542740904, "learning_rate": 8.168283380897408e-07, "loss": 0.2451, "step": 23753 }, { "epoch": 1.6421707569996542, "grad_norm": 0.6577939282910255, "learning_rate": 8.165217495838363e-07, "loss": 0.2591, "step": 23754 }, { "epoch": 1.6422398893881782, "grad_norm": 0.6773784696444767, "learning_rate": 8.162152135103657e-07, "loss": 0.3214, "step": 23755 }, { "epoch": 1.6423090217767022, "grad_norm": 0.7283537579457743, "learning_rate": 8.159087298731727e-07, "loss": 0.2572, "step": 23756 }, { "epoch": 1.6423781541652263, "grad_norm": 0.6587771503106926, "learning_rate": 8.156022986760958e-07, "loss": 0.2198, "step": 23757 }, { "epoch": 1.6424472865537503, "grad_norm": 0.7028640313577588, "learning_rate": 8.152959199229777e-07, "loss": 0.2728, "step": 23758 }, { "epoch": 1.6425164189422743, "grad_norm": 0.7053254713198026, "learning_rate": 8.149895936176561e-07, "loss": 0.3041, "step": 23759 }, { "epoch": 1.6425855513307985, "grad_norm": 0.5939016380289378, "learning_rate": 8.146833197639725e-07, "loss": 0.1936, "step": 23760 }, { "epoch": 1.6426546837193226, "grad_norm": 0.7041468052850925, "learning_rate": 8.143770983657623e-07, "loss": 0.2262, "step": 23761 }, { "epoch": 1.6427238161078466, "grad_norm": 0.6862659167571568, "learning_rate": 8.140709294268678e-07, "loss": 0.3475, "step": 23762 }, { "epoch": 1.6427929484963706, "grad_norm": 0.7936510687820654, "learning_rate": 8.137648129511228e-07, "loss": 0.2817, "step": 23763 }, { "epoch": 1.6428620808848946, "grad_norm": 0.6330515790049782, "learning_rate": 8.134587489423667e-07, "loss": 0.1931, "step": 23764 }, { "epoch": 1.6429312132734186, "grad_norm": 0.8008654966932159, "learning_rate": 8.131527374044329e-07, "loss": 0.2757, "step": 23765 }, { "epoch": 1.6430003456619426, "grad_norm": 0.6657353091286234, "learning_rate": 8.128467783411587e-07, "loss": 0.3088, "step": 23766 }, { "epoch": 1.6430694780504667, "grad_norm": 0.8090789800959675, "learning_rate": 8.125408717563766e-07, "loss": 0.2427, "step": 23767 }, { "epoch": 1.6431386104389907, "grad_norm": 0.6776137312846665, "learning_rate": 8.122350176539217e-07, "loss": 0.2553, "step": 23768 }, { "epoch": 1.6432077428275147, "grad_norm": 0.7378630525848919, "learning_rate": 8.119292160376274e-07, "loss": 0.2864, "step": 23769 }, { "epoch": 1.6432768752160387, "grad_norm": 0.7813920760850518, "learning_rate": 8.116234669113282e-07, "loss": 0.3278, "step": 23770 }, { "epoch": 1.6433460076045627, "grad_norm": 0.6656461678313988, "learning_rate": 8.113177702788527e-07, "loss": 0.1991, "step": 23771 }, { "epoch": 1.6434151399930867, "grad_norm": 0.7154375721115581, "learning_rate": 8.110121261440351e-07, "loss": 0.2418, "step": 23772 }, { "epoch": 1.6434842723816108, "grad_norm": 0.6883469290879388, "learning_rate": 8.107065345107035e-07, "loss": 0.305, "step": 23773 }, { "epoch": 1.6435534047701348, "grad_norm": 0.7516614018123325, "learning_rate": 8.104009953826913e-07, "loss": 0.2623, "step": 23774 }, { "epoch": 1.6436225371586588, "grad_norm": 0.6902876182534616, "learning_rate": 8.10095508763824e-07, "loss": 0.2819, "step": 23775 }, { "epoch": 1.6436916695471828, "grad_norm": 0.7389447124841388, "learning_rate": 8.097900746579329e-07, "loss": 0.2917, "step": 23776 }, { "epoch": 1.6437608019357068, "grad_norm": 0.7488742487387053, "learning_rate": 8.094846930688449e-07, "loss": 0.3218, "step": 23777 }, { "epoch": 1.6438299343242309, "grad_norm": 0.8001153329315148, "learning_rate": 8.091793640003892e-07, "loss": 0.2006, "step": 23778 }, { "epoch": 1.6438990667127549, "grad_norm": 0.7072231205193835, "learning_rate": 8.088740874563905e-07, "loss": 0.2723, "step": 23779 }, { "epoch": 1.6439681991012791, "grad_norm": 0.7316814151211322, "learning_rate": 8.085688634406768e-07, "loss": 0.333, "step": 23780 }, { "epoch": 1.6440373314898031, "grad_norm": 0.7589979264285286, "learning_rate": 8.082636919570714e-07, "loss": 0.2463, "step": 23781 }, { "epoch": 1.6441064638783272, "grad_norm": 0.7302760527790154, "learning_rate": 8.079585730093997e-07, "loss": 0.2503, "step": 23782 }, { "epoch": 1.6441755962668512, "grad_norm": 0.8305373073122385, "learning_rate": 8.076535066014863e-07, "loss": 0.2544, "step": 23783 }, { "epoch": 1.6442447286553752, "grad_norm": 0.7494191193658429, "learning_rate": 8.073484927371561e-07, "loss": 0.3585, "step": 23784 }, { "epoch": 1.6443138610438992, "grad_norm": 0.7568245370536844, "learning_rate": 8.07043531420229e-07, "loss": 0.2393, "step": 23785 }, { "epoch": 1.6443829934324232, "grad_norm": 0.6709125696218573, "learning_rate": 8.067386226545298e-07, "loss": 0.2447, "step": 23786 }, { "epoch": 1.6444521258209472, "grad_norm": 0.7035003617093856, "learning_rate": 8.064337664438776e-07, "loss": 0.3576, "step": 23787 }, { "epoch": 1.6445212582094713, "grad_norm": 0.9051499369650919, "learning_rate": 8.06128962792096e-07, "loss": 0.2965, "step": 23788 }, { "epoch": 1.6445903905979953, "grad_norm": 0.7302378183155871, "learning_rate": 8.058242117030018e-07, "loss": 0.2771, "step": 23789 }, { "epoch": 1.6446595229865193, "grad_norm": 1.3419181348913338, "learning_rate": 8.055195131804161e-07, "loss": 0.2434, "step": 23790 }, { "epoch": 1.6447286553750433, "grad_norm": 0.7179550361640946, "learning_rate": 8.052148672281585e-07, "loss": 0.3774, "step": 23791 }, { "epoch": 1.6447977877635673, "grad_norm": 0.7575634068874483, "learning_rate": 8.049102738500475e-07, "loss": 0.2465, "step": 23792 }, { "epoch": 1.6448669201520913, "grad_norm": 0.7806944899701348, "learning_rate": 8.046057330498985e-07, "loss": 0.2589, "step": 23793 }, { "epoch": 1.6449360525406154, "grad_norm": 0.6971039293401151, "learning_rate": 8.043012448315312e-07, "loss": 0.3082, "step": 23794 }, { "epoch": 1.6450051849291394, "grad_norm": 0.7434748392966147, "learning_rate": 8.039968091987587e-07, "loss": 0.3236, "step": 23795 }, { "epoch": 1.6450743173176634, "grad_norm": 0.682448731055335, "learning_rate": 8.036924261553991e-07, "loss": 0.2641, "step": 23796 }, { "epoch": 1.6451434497061874, "grad_norm": 0.6633277089700857, "learning_rate": 8.033880957052642e-07, "loss": 0.2155, "step": 23797 }, { "epoch": 1.6452125820947114, "grad_norm": 0.7221537564168573, "learning_rate": 8.030838178521727e-07, "loss": 0.373, "step": 23798 }, { "epoch": 1.6452817144832355, "grad_norm": 0.7302861324684945, "learning_rate": 8.027795925999343e-07, "loss": 0.2721, "step": 23799 }, { "epoch": 1.6453508468717595, "grad_norm": 1.119940879470973, "learning_rate": 8.024754199523649e-07, "loss": 0.2338, "step": 23800 }, { "epoch": 1.6454199792602835, "grad_norm": 0.7027290722409449, "learning_rate": 8.021712999132741e-07, "loss": 0.2725, "step": 23801 }, { "epoch": 1.6454891116488075, "grad_norm": 0.7035318512668138, "learning_rate": 8.018672324864763e-07, "loss": 0.3781, "step": 23802 }, { "epoch": 1.6455582440373315, "grad_norm": 0.6530522784463484, "learning_rate": 8.015632176757793e-07, "loss": 0.2389, "step": 23803 }, { "epoch": 1.6456273764258555, "grad_norm": 0.6542155234971585, "learning_rate": 8.012592554849946e-07, "loss": 0.2794, "step": 23804 }, { "epoch": 1.6456965088143796, "grad_norm": 0.9959109142340575, "learning_rate": 8.009553459179325e-07, "loss": 0.3296, "step": 23805 }, { "epoch": 1.6457656412029036, "grad_norm": 0.7671714011700137, "learning_rate": 8.006514889784023e-07, "loss": 0.2187, "step": 23806 }, { "epoch": 1.6458347735914276, "grad_norm": 0.603023495923497, "learning_rate": 8.003476846702107e-07, "loss": 0.2324, "step": 23807 }, { "epoch": 1.6459039059799516, "grad_norm": 0.765161248415817, "learning_rate": 8.000439329971676e-07, "loss": 0.2593, "step": 23808 }, { "epoch": 1.6459730383684756, "grad_norm": 0.7087389698598775, "learning_rate": 7.997402339630777e-07, "loss": 0.3333, "step": 23809 }, { "epoch": 1.6460421707569997, "grad_norm": 0.6573305207056558, "learning_rate": 7.994365875717491e-07, "loss": 0.2178, "step": 23810 }, { "epoch": 1.6461113031455237, "grad_norm": 0.6638079056806852, "learning_rate": 7.991329938269854e-07, "loss": 0.235, "step": 23811 }, { "epoch": 1.6461804355340477, "grad_norm": 0.64577869193343, "learning_rate": 7.98829452732593e-07, "loss": 0.263, "step": 23812 }, { "epoch": 1.6462495679225717, "grad_norm": 0.697011709459281, "learning_rate": 7.98525964292377e-07, "loss": 0.2508, "step": 23813 }, { "epoch": 1.6463187003110957, "grad_norm": 0.7174347831866723, "learning_rate": 7.982225285101397e-07, "loss": 0.2574, "step": 23814 }, { "epoch": 1.6463878326996197, "grad_norm": 0.7396907796733102, "learning_rate": 7.97919145389685e-07, "loss": 0.2552, "step": 23815 }, { "epoch": 1.6464569650881438, "grad_norm": 0.8377168827524238, "learning_rate": 7.97615814934814e-07, "loss": 0.3582, "step": 23816 }, { "epoch": 1.6465260974766678, "grad_norm": 0.6640217600587295, "learning_rate": 7.97312537149329e-07, "loss": 0.2062, "step": 23817 }, { "epoch": 1.6465952298651918, "grad_norm": 0.7410557427858837, "learning_rate": 7.97009312037032e-07, "loss": 0.2511, "step": 23818 }, { "epoch": 1.6466643622537158, "grad_norm": 0.718829799490214, "learning_rate": 7.967061396017234e-07, "loss": 0.3004, "step": 23819 }, { "epoch": 1.6467334946422398, "grad_norm": 0.7574736087890755, "learning_rate": 7.964030198472006e-07, "loss": 0.2785, "step": 23820 }, { "epoch": 1.6468026270307639, "grad_norm": 0.7299949702004337, "learning_rate": 7.960999527772662e-07, "loss": 0.291, "step": 23821 }, { "epoch": 1.6468717594192879, "grad_norm": 0.6930859089553265, "learning_rate": 7.957969383957153e-07, "loss": 0.293, "step": 23822 }, { "epoch": 1.646940891807812, "grad_norm": 0.7134045607996677, "learning_rate": 7.95493976706348e-07, "loss": 0.3777, "step": 23823 }, { "epoch": 1.647010024196336, "grad_norm": 0.7738997578163617, "learning_rate": 7.951910677129593e-07, "loss": 0.2236, "step": 23824 }, { "epoch": 1.64707915658486, "grad_norm": 0.6949503153126834, "learning_rate": 7.948882114193474e-07, "loss": 0.2781, "step": 23825 }, { "epoch": 1.647148288973384, "grad_norm": 0.7302171879265439, "learning_rate": 7.945854078293063e-07, "loss": 0.3176, "step": 23826 }, { "epoch": 1.647217421361908, "grad_norm": 0.6684295779821948, "learning_rate": 7.942826569466344e-07, "loss": 0.2854, "step": 23827 }, { "epoch": 1.647286553750432, "grad_norm": 0.6789797505062996, "learning_rate": 7.939799587751224e-07, "loss": 0.216, "step": 23828 }, { "epoch": 1.647355686138956, "grad_norm": 0.6102390949493549, "learning_rate": 7.936773133185665e-07, "loss": 0.2104, "step": 23829 }, { "epoch": 1.64742481852748, "grad_norm": 0.8661782027521171, "learning_rate": 7.93374720580758e-07, "loss": 0.3259, "step": 23830 }, { "epoch": 1.647493950916004, "grad_norm": 0.7662614028372882, "learning_rate": 7.930721805654918e-07, "loss": 0.2744, "step": 23831 }, { "epoch": 1.647563083304528, "grad_norm": 0.6373007060522556, "learning_rate": 7.927696932765555e-07, "loss": 0.2396, "step": 23832 }, { "epoch": 1.647632215693052, "grad_norm": 0.7103316388265095, "learning_rate": 7.924672587177451e-07, "loss": 0.2489, "step": 23833 }, { "epoch": 1.647701348081576, "grad_norm": 0.698808694101319, "learning_rate": 7.921648768928481e-07, "loss": 0.2846, "step": 23834 }, { "epoch": 1.6477704804701, "grad_norm": 0.8211052327498233, "learning_rate": 7.918625478056563e-07, "loss": 0.2009, "step": 23835 }, { "epoch": 1.6478396128586241, "grad_norm": 0.7270489702243847, "learning_rate": 7.915602714599568e-07, "loss": 0.2459, "step": 23836 }, { "epoch": 1.6479087452471481, "grad_norm": 0.688526198882757, "learning_rate": 7.912580478595399e-07, "loss": 0.304, "step": 23837 }, { "epoch": 1.6479778776356722, "grad_norm": 0.7952323037878863, "learning_rate": 7.90955877008191e-07, "loss": 0.2706, "step": 23838 }, { "epoch": 1.6480470100241962, "grad_norm": 0.6624890334650833, "learning_rate": 7.906537589096991e-07, "loss": 0.3075, "step": 23839 }, { "epoch": 1.6481161424127202, "grad_norm": 0.7458760633051079, "learning_rate": 7.903516935678507e-07, "loss": 0.3339, "step": 23840 }, { "epoch": 1.6481852748012444, "grad_norm": 0.6907547884174507, "learning_rate": 7.900496809864322e-07, "loss": 0.3195, "step": 23841 }, { "epoch": 1.6482544071897685, "grad_norm": 0.7175404896009056, "learning_rate": 7.897477211692273e-07, "loss": 0.1951, "step": 23842 }, { "epoch": 1.6483235395782925, "grad_norm": 0.7629674477649714, "learning_rate": 7.894458141200218e-07, "loss": 0.1868, "step": 23843 }, { "epoch": 1.6483926719668165, "grad_norm": 0.7370838329705264, "learning_rate": 7.891439598425977e-07, "loss": 0.2972, "step": 23844 }, { "epoch": 1.6484618043553405, "grad_norm": 0.7550953859733895, "learning_rate": 7.888421583407407e-07, "loss": 0.3453, "step": 23845 }, { "epoch": 1.6485309367438645, "grad_norm": 0.7084564706763565, "learning_rate": 7.885404096182314e-07, "loss": 0.2607, "step": 23846 }, { "epoch": 1.6486000691323885, "grad_norm": 0.6824012173082685, "learning_rate": 7.882387136788522e-07, "loss": 0.2458, "step": 23847 }, { "epoch": 1.6486692015209126, "grad_norm": 0.7206082017979792, "learning_rate": 7.879370705263839e-07, "loss": 0.3466, "step": 23848 }, { "epoch": 1.6487383339094366, "grad_norm": 0.8306232123509593, "learning_rate": 7.876354801646097e-07, "loss": 0.3045, "step": 23849 }, { "epoch": 1.6488074662979606, "grad_norm": 0.7010280360704063, "learning_rate": 7.873339425973053e-07, "loss": 0.2402, "step": 23850 }, { "epoch": 1.6488765986864846, "grad_norm": 0.666443719355047, "learning_rate": 7.87032457828254e-07, "loss": 0.2796, "step": 23851 }, { "epoch": 1.6489457310750086, "grad_norm": 1.1316030859220423, "learning_rate": 7.867310258612315e-07, "loss": 0.3135, "step": 23852 }, { "epoch": 1.6490148634635327, "grad_norm": 0.7173960743003566, "learning_rate": 7.86429646700016e-07, "loss": 0.2727, "step": 23853 }, { "epoch": 1.6490839958520567, "grad_norm": 0.716615159581671, "learning_rate": 7.861283203483855e-07, "loss": 0.2614, "step": 23854 }, { "epoch": 1.6491531282405807, "grad_norm": 0.6765687735061503, "learning_rate": 7.858270468101181e-07, "loss": 0.3386, "step": 23855 }, { "epoch": 1.6492222606291047, "grad_norm": 0.7104680940543293, "learning_rate": 7.85525826088987e-07, "loss": 0.2914, "step": 23856 }, { "epoch": 1.6492913930176287, "grad_norm": 0.635288548094637, "learning_rate": 7.852246581887695e-07, "loss": 0.2512, "step": 23857 }, { "epoch": 1.6493605254061527, "grad_norm": 0.7978169132390253, "learning_rate": 7.849235431132379e-07, "loss": 0.2598, "step": 23858 }, { "epoch": 1.6494296577946768, "grad_norm": 0.756470206082246, "learning_rate": 7.84622480866169e-07, "loss": 0.3294, "step": 23859 }, { "epoch": 1.6494987901832008, "grad_norm": 1.0811303732275521, "learning_rate": 7.843214714513337e-07, "loss": 0.2333, "step": 23860 }, { "epoch": 1.6495679225717248, "grad_norm": 0.7319546800461982, "learning_rate": 7.840205148725049e-07, "loss": 0.2013, "step": 23861 }, { "epoch": 1.649637054960249, "grad_norm": 0.7119255593690439, "learning_rate": 7.837196111334555e-07, "loss": 0.3738, "step": 23862 }, { "epoch": 1.649706187348773, "grad_norm": 0.7104218075997554, "learning_rate": 7.834187602379578e-07, "loss": 0.2317, "step": 23863 }, { "epoch": 1.649775319737297, "grad_norm": 0.644477834656495, "learning_rate": 7.8311796218978e-07, "loss": 0.1954, "step": 23864 }, { "epoch": 1.649844452125821, "grad_norm": 0.7384870390778095, "learning_rate": 7.828172169926945e-07, "loss": 0.2457, "step": 23865 }, { "epoch": 1.649913584514345, "grad_norm": 0.6737483805445298, "learning_rate": 7.825165246504673e-07, "loss": 0.3532, "step": 23866 }, { "epoch": 1.6499827169028691, "grad_norm": 0.6801889594174387, "learning_rate": 7.822158851668709e-07, "loss": 0.2582, "step": 23867 }, { "epoch": 1.6500518492913931, "grad_norm": 0.7356799629768183, "learning_rate": 7.819152985456691e-07, "loss": 0.263, "step": 23868 }, { "epoch": 1.6501209816799172, "grad_norm": 0.7150579671858333, "learning_rate": 7.816147647906336e-07, "loss": 0.2846, "step": 23869 }, { "epoch": 1.6501901140684412, "grad_norm": 0.6954598715867446, "learning_rate": 7.813142839055276e-07, "loss": 0.3232, "step": 23870 }, { "epoch": 1.6502592464569652, "grad_norm": 0.7119316578685444, "learning_rate": 7.810138558941199e-07, "loss": 0.2769, "step": 23871 }, { "epoch": 1.6503283788454892, "grad_norm": 0.7117414918469959, "learning_rate": 7.807134807601735e-07, "loss": 0.2304, "step": 23872 }, { "epoch": 1.6503975112340132, "grad_norm": 0.7646651573118479, "learning_rate": 7.804131585074549e-07, "loss": 0.3503, "step": 23873 }, { "epoch": 1.6504666436225373, "grad_norm": 1.7814856291614134, "learning_rate": 7.801128891397258e-07, "loss": 0.2679, "step": 23874 }, { "epoch": 1.6505357760110613, "grad_norm": 0.7696202540146607, "learning_rate": 7.798126726607513e-07, "loss": 0.271, "step": 23875 }, { "epoch": 1.6506049083995853, "grad_norm": 0.6187505611093955, "learning_rate": 7.795125090742933e-07, "loss": 0.2451, "step": 23876 }, { "epoch": 1.6506740407881093, "grad_norm": 0.7166737378614666, "learning_rate": 7.792123983841154e-07, "loss": 0.3377, "step": 23877 }, { "epoch": 1.6507431731766333, "grad_norm": 0.7566193014449025, "learning_rate": 7.78912340593978e-07, "loss": 0.2234, "step": 23878 }, { "epoch": 1.6508123055651573, "grad_norm": 0.6967732931150187, "learning_rate": 7.786123357076402e-07, "loss": 0.2422, "step": 23879 }, { "epoch": 1.6508814379536814, "grad_norm": 0.7314086783867308, "learning_rate": 7.783123837288647e-07, "loss": 0.291, "step": 23880 }, { "epoch": 1.6509505703422054, "grad_norm": 0.7386913393441508, "learning_rate": 7.780124846614084e-07, "loss": 0.2255, "step": 23881 }, { "epoch": 1.6510197027307294, "grad_norm": 0.6107587090800095, "learning_rate": 7.777126385090306e-07, "loss": 0.2012, "step": 23882 }, { "epoch": 1.6510888351192534, "grad_norm": 0.7209217345434352, "learning_rate": 7.774128452754903e-07, "loss": 0.2184, "step": 23883 }, { "epoch": 1.6511579675077774, "grad_norm": 0.7495109806583257, "learning_rate": 7.771131049645459e-07, "loss": 0.3559, "step": 23884 }, { "epoch": 1.6512270998963015, "grad_norm": 0.7940287191986919, "learning_rate": 7.768134175799507e-07, "loss": 0.2826, "step": 23885 }, { "epoch": 1.6512962322848255, "grad_norm": 0.7858539822942594, "learning_rate": 7.765137831254643e-07, "loss": 0.2439, "step": 23886 }, { "epoch": 1.6513653646733495, "grad_norm": 0.7034745294817646, "learning_rate": 7.762142016048396e-07, "loss": 0.3441, "step": 23887 }, { "epoch": 1.6514344970618735, "grad_norm": 0.843156680961387, "learning_rate": 7.75914673021832e-07, "loss": 0.3347, "step": 23888 }, { "epoch": 1.6515036294503975, "grad_norm": 0.7126104539421437, "learning_rate": 7.756151973801956e-07, "loss": 0.2452, "step": 23889 }, { "epoch": 1.6515727618389215, "grad_norm": 0.7677214907772616, "learning_rate": 7.753157746836848e-07, "loss": 0.2567, "step": 23890 }, { "epoch": 1.6516418942274456, "grad_norm": 0.6299361665467473, "learning_rate": 7.750164049360509e-07, "loss": 0.2977, "step": 23891 }, { "epoch": 1.6517110266159696, "grad_norm": 2.3871745785895704, "learning_rate": 7.747170881410476e-07, "loss": 0.2037, "step": 23892 }, { "epoch": 1.6517801590044936, "grad_norm": 0.6817542208241733, "learning_rate": 7.744178243024241e-07, "loss": 0.2514, "step": 23893 }, { "epoch": 1.6518492913930176, "grad_norm": 0.8002379158025201, "learning_rate": 7.741186134239331e-07, "loss": 0.3012, "step": 23894 }, { "epoch": 1.6519184237815416, "grad_norm": 0.7258152260390364, "learning_rate": 7.738194555093231e-07, "loss": 0.2832, "step": 23895 }, { "epoch": 1.6519875561700657, "grad_norm": 0.7287429681434244, "learning_rate": 7.73520350562344e-07, "loss": 0.2494, "step": 23896 }, { "epoch": 1.6520566885585897, "grad_norm": 0.7216276521592215, "learning_rate": 7.732212985867448e-07, "loss": 0.3113, "step": 23897 }, { "epoch": 1.6521258209471137, "grad_norm": 0.8838277246014278, "learning_rate": 7.729222995862751e-07, "loss": 0.3096, "step": 23898 }, { "epoch": 1.6521949533356377, "grad_norm": 0.7484037045370314, "learning_rate": 7.726233535646793e-07, "loss": 0.2651, "step": 23899 }, { "epoch": 1.6522640857241617, "grad_norm": 0.6547418828128704, "learning_rate": 7.723244605257069e-07, "loss": 0.2159, "step": 23900 }, { "epoch": 1.6523332181126857, "grad_norm": 0.6936958867897721, "learning_rate": 7.720256204731014e-07, "loss": 0.2646, "step": 23901 }, { "epoch": 1.6524023505012098, "grad_norm": 0.6963637517820169, "learning_rate": 7.71726833410611e-07, "loss": 0.3593, "step": 23902 }, { "epoch": 1.6524714828897338, "grad_norm": 0.6944364674449276, "learning_rate": 7.714280993419771e-07, "loss": 0.2361, "step": 23903 }, { "epoch": 1.6525406152782578, "grad_norm": 0.7031499310268196, "learning_rate": 7.711294182709472e-07, "loss": 0.2338, "step": 23904 }, { "epoch": 1.6526097476667818, "grad_norm": 0.6959339328623023, "learning_rate": 7.70830790201263e-07, "loss": 0.3495, "step": 23905 }, { "epoch": 1.6526788800553058, "grad_norm": 0.8026014534103743, "learning_rate": 7.705322151366684e-07, "loss": 0.3135, "step": 23906 }, { "epoch": 1.6527480124438299, "grad_norm": 0.843642762311846, "learning_rate": 7.702336930809035e-07, "loss": 0.1819, "step": 23907 }, { "epoch": 1.6528171448323539, "grad_norm": 0.6583200588062407, "learning_rate": 7.699352240377117e-07, "loss": 0.2545, "step": 23908 }, { "epoch": 1.6528862772208779, "grad_norm": 0.6780593089533402, "learning_rate": 7.696368080108319e-07, "loss": 0.2995, "step": 23909 }, { "epoch": 1.652955409609402, "grad_norm": 0.6834047793061979, "learning_rate": 7.693384450040059e-07, "loss": 0.2419, "step": 23910 }, { "epoch": 1.653024541997926, "grad_norm": 0.6903942019252006, "learning_rate": 7.69040135020972e-07, "loss": 0.2656, "step": 23911 }, { "epoch": 1.65309367438645, "grad_norm": 0.7021144550599056, "learning_rate": 7.68741878065471e-07, "loss": 0.3721, "step": 23912 }, { "epoch": 1.653162806774974, "grad_norm": 0.718884889505506, "learning_rate": 7.684436741412376e-07, "loss": 0.2589, "step": 23913 }, { "epoch": 1.653231939163498, "grad_norm": 0.6631612935000271, "learning_rate": 7.68145523252013e-07, "loss": 0.2285, "step": 23914 }, { "epoch": 1.653301071552022, "grad_norm": 0.8868369354015649, "learning_rate": 7.678474254015312e-07, "loss": 0.3547, "step": 23915 }, { "epoch": 1.653370203940546, "grad_norm": 0.7104141645874255, "learning_rate": 7.675493805935302e-07, "loss": 0.3194, "step": 23916 }, { "epoch": 1.65343933632907, "grad_norm": 0.6456436677018481, "learning_rate": 7.672513888317435e-07, "loss": 0.2907, "step": 23917 }, { "epoch": 1.653508468717594, "grad_norm": 0.6307446938937256, "learning_rate": 7.669534501199067e-07, "loss": 0.2356, "step": 23918 }, { "epoch": 1.653577601106118, "grad_norm": 0.777956468852127, "learning_rate": 7.666555644617546e-07, "loss": 0.2825, "step": 23919 }, { "epoch": 1.653646733494642, "grad_norm": 0.749246711516436, "learning_rate": 7.66357731861021e-07, "loss": 0.2945, "step": 23920 }, { "epoch": 1.653715865883166, "grad_norm": 0.7693929820299612, "learning_rate": 7.660599523214374e-07, "loss": 0.2207, "step": 23921 }, { "epoch": 1.6537849982716901, "grad_norm": 0.714311894133951, "learning_rate": 7.657622258467367e-07, "loss": 0.2958, "step": 23922 }, { "epoch": 1.6538541306602144, "grad_norm": 0.7092378121850277, "learning_rate": 7.654645524406501e-07, "loss": 0.3446, "step": 23923 }, { "epoch": 1.6539232630487384, "grad_norm": 0.637718719248529, "learning_rate": 7.651669321069077e-07, "loss": 0.2345, "step": 23924 }, { "epoch": 1.6539923954372624, "grad_norm": 0.6638612656874716, "learning_rate": 7.648693648492405e-07, "loss": 0.2255, "step": 23925 }, { "epoch": 1.6540615278257864, "grad_norm": 0.6934986012069536, "learning_rate": 7.645718506713795e-07, "loss": 0.305, "step": 23926 }, { "epoch": 1.6541306602143104, "grad_norm": 0.7276678051539505, "learning_rate": 7.642743895770499e-07, "loss": 0.3943, "step": 23927 }, { "epoch": 1.6541997926028345, "grad_norm": 0.6995743198060231, "learning_rate": 7.639769815699838e-07, "loss": 0.2322, "step": 23928 }, { "epoch": 1.6542689249913585, "grad_norm": 0.7449955745649589, "learning_rate": 7.636796266539054e-07, "loss": 0.2837, "step": 23929 }, { "epoch": 1.6543380573798825, "grad_norm": 0.6915596126750737, "learning_rate": 7.633823248325439e-07, "loss": 0.2912, "step": 23930 }, { "epoch": 1.6544071897684065, "grad_norm": 0.7783563547315814, "learning_rate": 7.630850761096231e-07, "loss": 0.265, "step": 23931 }, { "epoch": 1.6544763221569305, "grad_norm": 0.7177855556263409, "learning_rate": 7.627878804888694e-07, "loss": 0.1986, "step": 23932 }, { "epoch": 1.6545454545454545, "grad_norm": 0.7605872217780615, "learning_rate": 7.624907379740088e-07, "loss": 0.265, "step": 23933 }, { "epoch": 1.6546145869339786, "grad_norm": 0.6759902387309605, "learning_rate": 7.621936485687653e-07, "loss": 0.361, "step": 23934 }, { "epoch": 1.6546837193225026, "grad_norm": 0.6773302107069216, "learning_rate": 7.618966122768606e-07, "loss": 0.2593, "step": 23935 }, { "epoch": 1.6547528517110266, "grad_norm": 0.6653838056745396, "learning_rate": 7.615996291020195e-07, "loss": 0.2549, "step": 23936 }, { "epoch": 1.6548219840995506, "grad_norm": 0.706016460255434, "learning_rate": 7.613026990479627e-07, "loss": 0.3159, "step": 23937 }, { "epoch": 1.6548911164880746, "grad_norm": 0.7234949926219535, "learning_rate": 7.61005822118413e-07, "loss": 0.273, "step": 23938 }, { "epoch": 1.6549602488765986, "grad_norm": 0.6813978756967085, "learning_rate": 7.607089983170885e-07, "loss": 0.2484, "step": 23939 }, { "epoch": 1.6550293812651227, "grad_norm": 0.7167007304745792, "learning_rate": 7.604122276477132e-07, "loss": 0.251, "step": 23940 }, { "epoch": 1.6550985136536467, "grad_norm": 0.7708896134788354, "learning_rate": 7.601155101140039e-07, "loss": 0.3637, "step": 23941 }, { "epoch": 1.6551676460421707, "grad_norm": 0.6580585329752395, "learning_rate": 7.598188457196814e-07, "loss": 0.2385, "step": 23942 }, { "epoch": 1.655236778430695, "grad_norm": 0.7090736564197667, "learning_rate": 7.595222344684616e-07, "loss": 0.2053, "step": 23943 }, { "epoch": 1.655305910819219, "grad_norm": 0.6735388943394025, "learning_rate": 7.592256763640643e-07, "loss": 0.282, "step": 23944 }, { "epoch": 1.655375043207743, "grad_norm": 0.6718385441914815, "learning_rate": 7.589291714102043e-07, "loss": 0.2763, "step": 23945 }, { "epoch": 1.655444175596267, "grad_norm": 0.6794255777517327, "learning_rate": 7.58632719610598e-07, "loss": 0.2302, "step": 23946 }, { "epoch": 1.655513307984791, "grad_norm": 0.71536518792951, "learning_rate": 7.583363209689632e-07, "loss": 0.2723, "step": 23947 }, { "epoch": 1.655582440373315, "grad_norm": 0.8007459635193653, "learning_rate": 7.580399754890116e-07, "loss": 0.3008, "step": 23948 }, { "epoch": 1.655651572761839, "grad_norm": 0.9954074192912659, "learning_rate": 7.5774368317446e-07, "loss": 0.3066, "step": 23949 }, { "epoch": 1.655720705150363, "grad_norm": 0.7201708679891416, "learning_rate": 7.574474440290191e-07, "loss": 0.2327, "step": 23950 }, { "epoch": 1.655789837538887, "grad_norm": 0.7589013078655471, "learning_rate": 7.57151258056405e-07, "loss": 0.3038, "step": 23951 }, { "epoch": 1.655858969927411, "grad_norm": 0.7409982727512522, "learning_rate": 7.568551252603267e-07, "loss": 0.3441, "step": 23952 }, { "epoch": 1.6559281023159351, "grad_norm": 0.6661272019929301, "learning_rate": 7.565590456444977e-07, "loss": 0.2395, "step": 23953 }, { "epoch": 1.6559972347044591, "grad_norm": 0.6985888146118286, "learning_rate": 7.562630192126275e-07, "loss": 0.2035, "step": 23954 }, { "epoch": 1.6560663670929832, "grad_norm": 0.7499431141245491, "learning_rate": 7.559670459684287e-07, "loss": 0.3389, "step": 23955 }, { "epoch": 1.6561354994815072, "grad_norm": 0.6488567711665085, "learning_rate": 7.55671125915608e-07, "loss": 0.2891, "step": 23956 }, { "epoch": 1.6562046318700312, "grad_norm": 0.6360169305249428, "learning_rate": 7.553752590578766e-07, "loss": 0.1765, "step": 23957 }, { "epoch": 1.6562737642585552, "grad_norm": 0.8054816294352682, "learning_rate": 7.550794453989402e-07, "loss": 0.3001, "step": 23958 }, { "epoch": 1.6563428966470792, "grad_norm": 0.7561454702670182, "learning_rate": 7.547836849425077e-07, "loss": 0.358, "step": 23959 }, { "epoch": 1.6564120290356033, "grad_norm": 0.756768879599477, "learning_rate": 7.544879776922854e-07, "loss": 0.2803, "step": 23960 }, { "epoch": 1.6564811614241273, "grad_norm": 0.62048930563789, "learning_rate": 7.541923236519811e-07, "loss": 0.2053, "step": 23961 }, { "epoch": 1.6565502938126513, "grad_norm": 0.6914332040044402, "learning_rate": 7.53896722825298e-07, "loss": 0.2998, "step": 23962 }, { "epoch": 1.6566194262011753, "grad_norm": 0.7269276929088047, "learning_rate": 7.536011752159433e-07, "loss": 0.3043, "step": 23963 }, { "epoch": 1.6566885585896993, "grad_norm": 0.7134246719444889, "learning_rate": 7.533056808276184e-07, "loss": 0.2349, "step": 23964 }, { "epoch": 1.6567576909782233, "grad_norm": 0.7181944335967543, "learning_rate": 7.530102396640299e-07, "loss": 0.2964, "step": 23965 }, { "epoch": 1.6568268233667474, "grad_norm": 0.6836299421910558, "learning_rate": 7.527148517288773e-07, "loss": 0.2957, "step": 23966 }, { "epoch": 1.6568959557552714, "grad_norm": 0.7551332467803084, "learning_rate": 7.524195170258641e-07, "loss": 0.2362, "step": 23967 }, { "epoch": 1.6569650881437954, "grad_norm": 0.6390205369997534, "learning_rate": 7.521242355586928e-07, "loss": 0.1415, "step": 23968 }, { "epoch": 1.6570342205323194, "grad_norm": 0.8095675240125798, "learning_rate": 7.518290073310641e-07, "loss": 0.3386, "step": 23969 }, { "epoch": 1.6571033529208434, "grad_norm": 0.6844331446252564, "learning_rate": 7.51533832346677e-07, "loss": 0.321, "step": 23970 }, { "epoch": 1.6571724853093674, "grad_norm": 0.8840983194519135, "learning_rate": 7.512387106092323e-07, "loss": 0.2223, "step": 23971 }, { "epoch": 1.6572416176978915, "grad_norm": 0.674483473525743, "learning_rate": 7.50943642122427e-07, "loss": 0.1954, "step": 23972 }, { "epoch": 1.6573107500864155, "grad_norm": 0.7192931524992116, "learning_rate": 7.506486268899621e-07, "loss": 0.3158, "step": 23973 }, { "epoch": 1.6573798824749395, "grad_norm": 0.8033628753804489, "learning_rate": 7.503536649155313e-07, "loss": 0.2537, "step": 23974 }, { "epoch": 1.6574490148634635, "grad_norm": 0.6909846661417633, "learning_rate": 7.500587562028338e-07, "loss": 0.2129, "step": 23975 }, { "epoch": 1.6575181472519875, "grad_norm": 1.0795923506827798, "learning_rate": 7.497639007555657e-07, "loss": 0.3383, "step": 23976 }, { "epoch": 1.6575872796405116, "grad_norm": 0.7019828771858769, "learning_rate": 7.494690985774233e-07, "loss": 0.2882, "step": 23977 }, { "epoch": 1.6576564120290356, "grad_norm": 0.8209841629319986, "learning_rate": 7.491743496720993e-07, "loss": 0.2766, "step": 23978 }, { "epoch": 1.6577255444175596, "grad_norm": 0.7514036417062354, "learning_rate": 7.488796540432897e-07, "loss": 0.2402, "step": 23979 }, { "epoch": 1.6577946768060836, "grad_norm": 0.7815815362479602, "learning_rate": 7.485850116946863e-07, "loss": 0.2742, "step": 23980 }, { "epoch": 1.6578638091946076, "grad_norm": 0.6931587770206212, "learning_rate": 7.482904226299825e-07, "loss": 0.2849, "step": 23981 }, { "epoch": 1.6579329415831316, "grad_norm": 0.9872896647581049, "learning_rate": 7.479958868528708e-07, "loss": 0.3285, "step": 23982 }, { "epoch": 1.6580020739716557, "grad_norm": 0.7174988723508801, "learning_rate": 7.477014043670439e-07, "loss": 0.2853, "step": 23983 }, { "epoch": 1.6580712063601797, "grad_norm": 0.7835311377835418, "learning_rate": 7.474069751761903e-07, "loss": 0.3427, "step": 23984 }, { "epoch": 1.6581403387487037, "grad_norm": 0.8116929291041418, "learning_rate": 7.471125992840023e-07, "loss": 0.2472, "step": 23985 }, { "epoch": 1.6582094711372277, "grad_norm": 0.6611967087441188, "learning_rate": 7.468182766941667e-07, "loss": 0.2425, "step": 23986 }, { "epoch": 1.6582786035257517, "grad_norm": 0.7149230192576462, "learning_rate": 7.465240074103752e-07, "loss": 0.3356, "step": 23987 }, { "epoch": 1.6583477359142758, "grad_norm": 0.7020711613702125, "learning_rate": 7.462297914363137e-07, "loss": 0.3364, "step": 23988 }, { "epoch": 1.6584168683027998, "grad_norm": 0.7092762686701554, "learning_rate": 7.459356287756703e-07, "loss": 0.1863, "step": 23989 }, { "epoch": 1.6584860006913238, "grad_norm": 0.7057192947292845, "learning_rate": 7.45641519432132e-07, "loss": 0.2142, "step": 23990 }, { "epoch": 1.6585551330798478, "grad_norm": 0.7167335577164625, "learning_rate": 7.453474634093866e-07, "loss": 0.3308, "step": 23991 }, { "epoch": 1.6586242654683718, "grad_norm": 0.6997356498387299, "learning_rate": 7.450534607111165e-07, "loss": 0.2456, "step": 23992 }, { "epoch": 1.6586933978568958, "grad_norm": 0.9006860517679948, "learning_rate": 7.447595113410088e-07, "loss": 0.1953, "step": 23993 }, { "epoch": 1.6587625302454199, "grad_norm": 0.6958845301649167, "learning_rate": 7.444656153027463e-07, "loss": 0.3271, "step": 23994 }, { "epoch": 1.6588316626339439, "grad_norm": 0.7817217045990814, "learning_rate": 7.441717726000125e-07, "loss": 0.2987, "step": 23995 }, { "epoch": 1.658900795022468, "grad_norm": 0.7171962022196141, "learning_rate": 7.438779832364912e-07, "loss": 0.2499, "step": 23996 }, { "epoch": 1.658969927410992, "grad_norm": 0.7061951364140098, "learning_rate": 7.435842472158644e-07, "loss": 0.2747, "step": 23997 }, { "epoch": 1.659039059799516, "grad_norm": 0.8287502724858916, "learning_rate": 7.432905645418126e-07, "loss": 0.3482, "step": 23998 }, { "epoch": 1.65910819218804, "grad_norm": 0.8596416649693435, "learning_rate": 7.429969352180183e-07, "loss": 0.2769, "step": 23999 }, { "epoch": 1.659177324576564, "grad_norm": 0.6533633139483129, "learning_rate": 7.427033592481586e-07, "loss": 0.2784, "step": 24000 }, { "epoch": 1.659246456965088, "grad_norm": 0.7931908229711011, "learning_rate": 7.424098366359167e-07, "loss": 0.2688, "step": 24001 }, { "epoch": 1.659315589353612, "grad_norm": 0.7175585156855528, "learning_rate": 7.421163673849679e-07, "loss": 0.3011, "step": 24002 }, { "epoch": 1.659384721742136, "grad_norm": 0.7245081831632244, "learning_rate": 7.418229514989922e-07, "loss": 0.2202, "step": 24003 }, { "epoch": 1.6594538541306603, "grad_norm": 1.007587466522601, "learning_rate": 7.415295889816665e-07, "loss": 0.2674, "step": 24004 }, { "epoch": 1.6595229865191843, "grad_norm": 0.9416754226074563, "learning_rate": 7.412362798366691e-07, "loss": 0.325, "step": 24005 }, { "epoch": 1.6595921189077083, "grad_norm": 0.748950717283262, "learning_rate": 7.409430240676735e-07, "loss": 0.2666, "step": 24006 }, { "epoch": 1.6596612512962323, "grad_norm": 0.6713028937036749, "learning_rate": 7.406498216783581e-07, "loss": 0.2606, "step": 24007 }, { "epoch": 1.6597303836847563, "grad_norm": 0.7865606207810142, "learning_rate": 7.403566726723943e-07, "loss": 0.2471, "step": 24008 }, { "epoch": 1.6597995160732804, "grad_norm": 0.6897183501198143, "learning_rate": 7.400635770534598e-07, "loss": 0.3202, "step": 24009 }, { "epoch": 1.6598686484618044, "grad_norm": 0.6555069762644988, "learning_rate": 7.397705348252232e-07, "loss": 0.2394, "step": 24010 }, { "epoch": 1.6599377808503284, "grad_norm": 0.6931794364763838, "learning_rate": 7.394775459913627e-07, "loss": 0.1926, "step": 24011 }, { "epoch": 1.6600069132388524, "grad_norm": 0.6613747844342397, "learning_rate": 7.391846105555478e-07, "loss": 0.3489, "step": 24012 }, { "epoch": 1.6600760456273764, "grad_norm": 0.7269673679505754, "learning_rate": 7.388917285214491e-07, "loss": 0.2501, "step": 24013 }, { "epoch": 1.6601451780159004, "grad_norm": 0.6530162600827155, "learning_rate": 7.385988998927396e-07, "loss": 0.2816, "step": 24014 }, { "epoch": 1.6602143104044245, "grad_norm": 0.631307578731744, "learning_rate": 7.383061246730867e-07, "loss": 0.2245, "step": 24015 }, { "epoch": 1.6602834427929485, "grad_norm": 0.7497066361608267, "learning_rate": 7.380134028661612e-07, "loss": 0.3404, "step": 24016 }, { "epoch": 1.6603525751814725, "grad_norm": 0.7144337443218416, "learning_rate": 7.377207344756315e-07, "loss": 0.2395, "step": 24017 }, { "epoch": 1.6604217075699965, "grad_norm": 0.626429472023221, "learning_rate": 7.37428119505168e-07, "loss": 0.1964, "step": 24018 }, { "epoch": 1.6604908399585205, "grad_norm": 0.6857111113283777, "learning_rate": 7.371355579584344e-07, "loss": 0.2869, "step": 24019 }, { "epoch": 1.6605599723470446, "grad_norm": 0.7311727936446317, "learning_rate": 7.368430498391005e-07, "loss": 0.3819, "step": 24020 }, { "epoch": 1.6606291047355686, "grad_norm": 0.6445577668717863, "learning_rate": 7.365505951508295e-07, "loss": 0.2389, "step": 24021 }, { "epoch": 1.6606982371240926, "grad_norm": 0.7973097074288791, "learning_rate": 7.362581938972895e-07, "loss": 0.3221, "step": 24022 }, { "epoch": 1.6607673695126166, "grad_norm": 0.7146291233241596, "learning_rate": 7.359658460821434e-07, "loss": 0.2829, "step": 24023 }, { "epoch": 1.6608365019011408, "grad_norm": 0.7165842454725347, "learning_rate": 7.356735517090557e-07, "loss": 0.2911, "step": 24024 }, { "epoch": 1.6609056342896649, "grad_norm": 0.6757733630641846, "learning_rate": 7.353813107816898e-07, "loss": 0.2513, "step": 24025 }, { "epoch": 1.6609747666781889, "grad_norm": 0.8477776630658826, "learning_rate": 7.3508912330371e-07, "loss": 0.2894, "step": 24026 }, { "epoch": 1.661043899066713, "grad_norm": 0.6890284698740105, "learning_rate": 7.347969892787754e-07, "loss": 0.3512, "step": 24027 }, { "epoch": 1.661113031455237, "grad_norm": 0.658414949685767, "learning_rate": 7.345049087105505e-07, "loss": 0.2438, "step": 24028 }, { "epoch": 1.661182163843761, "grad_norm": 0.7219513124962686, "learning_rate": 7.342128816026933e-07, "loss": 0.2417, "step": 24029 }, { "epoch": 1.661251296232285, "grad_norm": 0.6466485555536282, "learning_rate": 7.339209079588649e-07, "loss": 0.2709, "step": 24030 }, { "epoch": 1.661320428620809, "grad_norm": 0.7125397696594317, "learning_rate": 7.336289877827246e-07, "loss": 0.2626, "step": 24031 }, { "epoch": 1.661389561009333, "grad_norm": 0.8075499472390575, "learning_rate": 7.333371210779328e-07, "loss": 0.2464, "step": 24032 }, { "epoch": 1.661458693397857, "grad_norm": 0.6644128475983281, "learning_rate": 7.330453078481447e-07, "loss": 0.246, "step": 24033 }, { "epoch": 1.661527825786381, "grad_norm": 0.7974799991677383, "learning_rate": 7.3275354809702e-07, "loss": 0.3838, "step": 24034 }, { "epoch": 1.661596958174905, "grad_norm": 0.7247016375975927, "learning_rate": 7.324618418282131e-07, "loss": 0.1881, "step": 24035 }, { "epoch": 1.661666090563429, "grad_norm": 0.6058337448256915, "learning_rate": 7.321701890453825e-07, "loss": 0.2282, "step": 24036 }, { "epoch": 1.661735222951953, "grad_norm": 0.7246384008503082, "learning_rate": 7.318785897521813e-07, "loss": 0.3443, "step": 24037 }, { "epoch": 1.661804355340477, "grad_norm": 0.7224615890027274, "learning_rate": 7.315870439522648e-07, "loss": 0.3174, "step": 24038 }, { "epoch": 1.6618734877290011, "grad_norm": 0.6824548229431353, "learning_rate": 7.312955516492875e-07, "loss": 0.2456, "step": 24039 }, { "epoch": 1.6619426201175251, "grad_norm": 0.7689704421838305, "learning_rate": 7.310041128469042e-07, "loss": 0.204, "step": 24040 }, { "epoch": 1.6620117525060492, "grad_norm": 1.2187466867513062, "learning_rate": 7.307127275487641e-07, "loss": 0.3833, "step": 24041 }, { "epoch": 1.6620808848945732, "grad_norm": 0.7236109216285307, "learning_rate": 7.304213957585227e-07, "loss": 0.2772, "step": 24042 }, { "epoch": 1.6621500172830972, "grad_norm": 0.6675104599780827, "learning_rate": 7.301301174798287e-07, "loss": 0.2404, "step": 24043 }, { "epoch": 1.6622191496716212, "grad_norm": 0.6805449454283281, "learning_rate": 7.298388927163346e-07, "loss": 0.2852, "step": 24044 }, { "epoch": 1.6622882820601452, "grad_norm": 0.8961179309140764, "learning_rate": 7.295477214716884e-07, "loss": 0.3164, "step": 24045 }, { "epoch": 1.6623574144486692, "grad_norm": 0.8349174758730207, "learning_rate": 7.292566037495402e-07, "loss": 0.2578, "step": 24046 }, { "epoch": 1.6624265468371933, "grad_norm": 0.6616145668697643, "learning_rate": 7.289655395535395e-07, "loss": 0.2297, "step": 24047 }, { "epoch": 1.6624956792257173, "grad_norm": 0.6738688526632174, "learning_rate": 7.286745288873348e-07, "loss": 0.3667, "step": 24048 }, { "epoch": 1.6625648116142413, "grad_norm": 0.7555089876498622, "learning_rate": 7.283835717545711e-07, "loss": 0.2475, "step": 24049 }, { "epoch": 1.6626339440027653, "grad_norm": 1.131329395265435, "learning_rate": 7.280926681588979e-07, "loss": 0.239, "step": 24050 }, { "epoch": 1.6627030763912893, "grad_norm": 0.6666375630439011, "learning_rate": 7.278018181039576e-07, "loss": 0.3001, "step": 24051 }, { "epoch": 1.6627722087798134, "grad_norm": 0.732733102830803, "learning_rate": 7.275110215933978e-07, "loss": 0.303, "step": 24052 }, { "epoch": 1.6628413411683374, "grad_norm": 0.6595042248807623, "learning_rate": 7.27220278630863e-07, "loss": 0.2306, "step": 24053 }, { "epoch": 1.6629104735568614, "grad_norm": 0.7854054092164972, "learning_rate": 7.269295892199979e-07, "loss": 0.2774, "step": 24054 }, { "epoch": 1.6629796059453854, "grad_norm": 0.7134603911084507, "learning_rate": 7.266389533644436e-07, "loss": 0.3098, "step": 24055 }, { "epoch": 1.6630487383339094, "grad_norm": 0.771423303678632, "learning_rate": 7.263483710678449e-07, "loss": 0.2337, "step": 24056 }, { "epoch": 1.6631178707224334, "grad_norm": 0.6580779181927453, "learning_rate": 7.260578423338416e-07, "loss": 0.2196, "step": 24057 }, { "epoch": 1.6631870031109575, "grad_norm": 0.7723293474152361, "learning_rate": 7.25767367166077e-07, "loss": 0.3142, "step": 24058 }, { "epoch": 1.6632561354994815, "grad_norm": 0.7446787254570586, "learning_rate": 7.254769455681898e-07, "loss": 0.382, "step": 24059 }, { "epoch": 1.6633252678880055, "grad_norm": 0.6561185181047523, "learning_rate": 7.251865775438211e-07, "loss": 0.2231, "step": 24060 }, { "epoch": 1.6633944002765295, "grad_norm": 0.6886757394619231, "learning_rate": 7.248962630966094e-07, "loss": 0.2799, "step": 24061 }, { "epoch": 1.6634635326650535, "grad_norm": 0.6610220889093865, "learning_rate": 7.246060022301954e-07, "loss": 0.3129, "step": 24062 }, { "epoch": 1.6635326650535776, "grad_norm": 0.6997609999556034, "learning_rate": 7.243157949482138e-07, "loss": 0.2828, "step": 24063 }, { "epoch": 1.6636017974421016, "grad_norm": 0.7379154932898861, "learning_rate": 7.24025641254305e-07, "loss": 0.2717, "step": 24064 }, { "epoch": 1.6636709298306256, "grad_norm": 0.6723870433153314, "learning_rate": 7.237355411521025e-07, "loss": 0.2567, "step": 24065 }, { "epoch": 1.6637400622191496, "grad_norm": 0.7049750989023968, "learning_rate": 7.234454946452441e-07, "loss": 0.3736, "step": 24066 }, { "epoch": 1.6638091946076736, "grad_norm": 0.7324127973335482, "learning_rate": 7.231555017373637e-07, "loss": 0.1955, "step": 24067 }, { "epoch": 1.6638783269961976, "grad_norm": 0.665564946250074, "learning_rate": 7.228655624320985e-07, "loss": 0.2367, "step": 24068 }, { "epoch": 1.6639474593847217, "grad_norm": 0.7215233826984012, "learning_rate": 7.225756767330794e-07, "loss": 0.2989, "step": 24069 }, { "epoch": 1.6640165917732457, "grad_norm": 0.6698774313055869, "learning_rate": 7.222858446439418e-07, "loss": 0.2916, "step": 24070 }, { "epoch": 1.6640857241617697, "grad_norm": 0.664317640296663, "learning_rate": 7.219960661683162e-07, "loss": 0.2366, "step": 24071 }, { "epoch": 1.6641548565502937, "grad_norm": 0.6455825828804835, "learning_rate": 7.217063413098363e-07, "loss": 0.2749, "step": 24072 }, { "epoch": 1.6642239889388177, "grad_norm": 0.7378949727331258, "learning_rate": 7.214166700721321e-07, "loss": 0.3586, "step": 24073 }, { "epoch": 1.6642931213273418, "grad_norm": 0.7324210472376902, "learning_rate": 7.211270524588338e-07, "loss": 0.2822, "step": 24074 }, { "epoch": 1.6643622537158658, "grad_norm": 0.6881567341610736, "learning_rate": 7.20837488473572e-07, "loss": 0.2459, "step": 24075 }, { "epoch": 1.6644313861043898, "grad_norm": 0.6574351965643104, "learning_rate": 7.20547978119977e-07, "loss": 0.2513, "step": 24076 }, { "epoch": 1.6645005184929138, "grad_norm": 0.7599076014601103, "learning_rate": 7.202585214016755e-07, "loss": 0.2827, "step": 24077 }, { "epoch": 1.6645696508814378, "grad_norm": 0.6618894865256857, "learning_rate": 7.199691183222957e-07, "loss": 0.1884, "step": 24078 }, { "epoch": 1.6646387832699618, "grad_norm": 0.6600698269323285, "learning_rate": 7.196797688854656e-07, "loss": 0.2714, "step": 24079 }, { "epoch": 1.6647079156584859, "grad_norm": 0.6844874279024262, "learning_rate": 7.193904730948098e-07, "loss": 0.3147, "step": 24080 }, { "epoch": 1.6647770480470099, "grad_norm": 0.7888264956874632, "learning_rate": 7.191012309539552e-07, "loss": 0.3097, "step": 24081 }, { "epoch": 1.664846180435534, "grad_norm": 0.5999732664153231, "learning_rate": 7.188120424665274e-07, "loss": 0.2513, "step": 24082 }, { "epoch": 1.664915312824058, "grad_norm": 0.7387182440395964, "learning_rate": 7.185229076361517e-07, "loss": 0.2955, "step": 24083 }, { "epoch": 1.664984445212582, "grad_norm": 0.6940446979375374, "learning_rate": 7.182338264664496e-07, "loss": 0.2946, "step": 24084 }, { "epoch": 1.6650535776011062, "grad_norm": 0.6772238036678591, "learning_rate": 7.17944798961046e-07, "loss": 0.2383, "step": 24085 }, { "epoch": 1.6651227099896302, "grad_norm": 0.6974710356692537, "learning_rate": 7.176558251235622e-07, "loss": 0.3278, "step": 24086 }, { "epoch": 1.6651918423781542, "grad_norm": 0.6916360152362163, "learning_rate": 7.173669049576199e-07, "loss": 0.324, "step": 24087 }, { "epoch": 1.6652609747666782, "grad_norm": 1.1098787486271846, "learning_rate": 7.170780384668408e-07, "loss": 0.2929, "step": 24088 }, { "epoch": 1.6653301071552022, "grad_norm": 0.5922838898551892, "learning_rate": 7.167892256548464e-07, "loss": 0.2162, "step": 24089 }, { "epoch": 1.6653992395437263, "grad_norm": 0.6768170620982042, "learning_rate": 7.165004665252545e-07, "loss": 0.2559, "step": 24090 }, { "epoch": 1.6654683719322503, "grad_norm": 0.7187108779295376, "learning_rate": 7.16211761081686e-07, "loss": 0.327, "step": 24091 }, { "epoch": 1.6655375043207743, "grad_norm": 0.7609074265404677, "learning_rate": 7.159231093277575e-07, "loss": 0.252, "step": 24092 }, { "epoch": 1.6656066367092983, "grad_norm": 0.7180502674740283, "learning_rate": 7.156345112670881e-07, "loss": 0.2337, "step": 24093 }, { "epoch": 1.6656757690978223, "grad_norm": 0.6780868566446315, "learning_rate": 7.153459669032937e-07, "loss": 0.3232, "step": 24094 }, { "epoch": 1.6657449014863464, "grad_norm": 0.7131114401356172, "learning_rate": 7.150574762399909e-07, "loss": 0.3247, "step": 24095 }, { "epoch": 1.6658140338748704, "grad_norm": 0.7288529745782876, "learning_rate": 7.147690392807954e-07, "loss": 0.2123, "step": 24096 }, { "epoch": 1.6658831662633944, "grad_norm": 0.6667232125812494, "learning_rate": 7.144806560293244e-07, "loss": 0.2224, "step": 24097 }, { "epoch": 1.6659522986519184, "grad_norm": 0.7205033065724292, "learning_rate": 7.141923264891898e-07, "loss": 0.3629, "step": 24098 }, { "epoch": 1.6660214310404424, "grad_norm": 0.6430757430250377, "learning_rate": 7.139040506640066e-07, "loss": 0.2439, "step": 24099 }, { "epoch": 1.6660905634289664, "grad_norm": 0.7116875500141924, "learning_rate": 7.136158285573858e-07, "loss": 0.2168, "step": 24100 }, { "epoch": 1.6661596958174905, "grad_norm": 0.7548828938719521, "learning_rate": 7.133276601729417e-07, "loss": 0.2349, "step": 24101 }, { "epoch": 1.6662288282060145, "grad_norm": 0.7008183284112159, "learning_rate": 7.130395455142852e-07, "loss": 0.3137, "step": 24102 }, { "epoch": 1.6662979605945385, "grad_norm": 0.7267719591174723, "learning_rate": 7.127514845850292e-07, "loss": 0.2622, "step": 24103 }, { "epoch": 1.6663670929830625, "grad_norm": 0.7188104165398669, "learning_rate": 7.124634773887806e-07, "loss": 0.2472, "step": 24104 }, { "epoch": 1.6664362253715868, "grad_norm": 0.7456905048298754, "learning_rate": 7.121755239291522e-07, "loss": 0.3404, "step": 24105 }, { "epoch": 1.6665053577601108, "grad_norm": 0.6967602260917076, "learning_rate": 7.118876242097506e-07, "loss": 0.2731, "step": 24106 }, { "epoch": 1.6665744901486348, "grad_norm": 0.6506401373478644, "learning_rate": 7.115997782341865e-07, "loss": 0.2216, "step": 24107 }, { "epoch": 1.6666436225371588, "grad_norm": 0.6772269061117921, "learning_rate": 7.113119860060647e-07, "loss": 0.2735, "step": 24108 }, { "epoch": 1.6667127549256828, "grad_norm": 0.7549951080857548, "learning_rate": 7.110242475289936e-07, "loss": 0.3696, "step": 24109 }, { "epoch": 1.6667818873142068, "grad_norm": 0.7229893640892388, "learning_rate": 7.107365628065799e-07, "loss": 0.2383, "step": 24110 }, { "epoch": 1.6668510197027309, "grad_norm": 0.613214877438574, "learning_rate": 7.104489318424296e-07, "loss": 0.2371, "step": 24111 }, { "epoch": 1.6669201520912549, "grad_norm": 0.6331911052923722, "learning_rate": 7.101613546401459e-07, "loss": 0.3018, "step": 24112 }, { "epoch": 1.666989284479779, "grad_norm": 2.728554690643953, "learning_rate": 7.098738312033355e-07, "loss": 0.2863, "step": 24113 }, { "epoch": 1.667058416868303, "grad_norm": 0.6255463262142114, "learning_rate": 7.095863615355991e-07, "loss": 0.1877, "step": 24114 }, { "epoch": 1.667127549256827, "grad_norm": 0.7036040950746763, "learning_rate": 7.092989456405424e-07, "loss": 0.2336, "step": 24115 }, { "epoch": 1.667196681645351, "grad_norm": 0.7716362832404446, "learning_rate": 7.090115835217648e-07, "loss": 0.3541, "step": 24116 }, { "epoch": 1.667265814033875, "grad_norm": 0.7054729126944042, "learning_rate": 7.087242751828699e-07, "loss": 0.2958, "step": 24117 }, { "epoch": 1.667334946422399, "grad_norm": 0.6743865988491776, "learning_rate": 7.084370206274577e-07, "loss": 0.2531, "step": 24118 }, { "epoch": 1.667404078810923, "grad_norm": 0.7392889969291971, "learning_rate": 7.081498198591302e-07, "loss": 0.316, "step": 24119 }, { "epoch": 1.667473211199447, "grad_norm": 0.711691951011635, "learning_rate": 7.078626728814842e-07, "loss": 0.3042, "step": 24120 }, { "epoch": 1.667542343587971, "grad_norm": 0.8000047802859949, "learning_rate": 7.07575579698121e-07, "loss": 0.2664, "step": 24121 }, { "epoch": 1.667611475976495, "grad_norm": 0.6373557616821481, "learning_rate": 7.072885403126367e-07, "loss": 0.2826, "step": 24122 }, { "epoch": 1.667680608365019, "grad_norm": 0.6850020966280024, "learning_rate": 7.070015547286302e-07, "loss": 0.3448, "step": 24123 }, { "epoch": 1.667749740753543, "grad_norm": 0.7001613551556687, "learning_rate": 7.067146229496979e-07, "loss": 0.2705, "step": 24124 }, { "epoch": 1.6678188731420671, "grad_norm": 0.7964766089086266, "learning_rate": 7.064277449794366e-07, "loss": 0.2136, "step": 24125 }, { "epoch": 1.6678880055305911, "grad_norm": 0.693456251620204, "learning_rate": 7.061409208214409e-07, "loss": 0.3216, "step": 24126 }, { "epoch": 1.6679571379191152, "grad_norm": 0.7102160518064745, "learning_rate": 7.05854150479307e-07, "loss": 0.3389, "step": 24127 }, { "epoch": 1.6680262703076392, "grad_norm": 0.7200279809307282, "learning_rate": 7.055674339566265e-07, "loss": 0.2409, "step": 24128 }, { "epoch": 1.6680954026961632, "grad_norm": 0.6366495317075529, "learning_rate": 7.052807712569959e-07, "loss": 0.243, "step": 24129 }, { "epoch": 1.6681645350846872, "grad_norm": 1.1058559892976099, "learning_rate": 7.049941623840057e-07, "loss": 0.3128, "step": 24130 }, { "epoch": 1.6682336674732112, "grad_norm": 0.7074200699208555, "learning_rate": 7.047076073412485e-07, "loss": 0.2843, "step": 24131 }, { "epoch": 1.6683027998617352, "grad_norm": 0.6301129052609069, "learning_rate": 7.044211061323164e-07, "loss": 0.2153, "step": 24132 }, { "epoch": 1.6683719322502593, "grad_norm": 0.6846040108593165, "learning_rate": 7.041346587608005e-07, "loss": 0.2663, "step": 24133 }, { "epoch": 1.6684410646387833, "grad_norm": 0.669619414629376, "learning_rate": 7.038482652302897e-07, "loss": 0.3538, "step": 24134 }, { "epoch": 1.6685101970273073, "grad_norm": 0.6173487660301893, "learning_rate": 7.035619255443755e-07, "loss": 0.2177, "step": 24135 }, { "epoch": 1.6685793294158313, "grad_norm": 0.7137882129465755, "learning_rate": 7.032756397066437e-07, "loss": 0.2307, "step": 24136 }, { "epoch": 1.6686484618043553, "grad_norm": 0.8850979292163018, "learning_rate": 7.029894077206839e-07, "loss": 0.324, "step": 24137 }, { "epoch": 1.6687175941928793, "grad_norm": 0.7013696970395132, "learning_rate": 7.027032295900832e-07, "loss": 0.2687, "step": 24138 }, { "epoch": 1.6687867265814034, "grad_norm": 0.7567589298239887, "learning_rate": 7.024171053184303e-07, "loss": 0.2548, "step": 24139 }, { "epoch": 1.6688558589699274, "grad_norm": 0.8869532551224344, "learning_rate": 7.021310349093085e-07, "loss": 0.2802, "step": 24140 }, { "epoch": 1.6689249913584514, "grad_norm": 0.727072614350028, "learning_rate": 7.01845018366305e-07, "loss": 0.3936, "step": 24141 }, { "epoch": 1.6689941237469754, "grad_norm": 0.7150368724693261, "learning_rate": 7.01559055693003e-07, "loss": 0.1915, "step": 24142 }, { "epoch": 1.6690632561354994, "grad_norm": 0.6759776251014351, "learning_rate": 7.012731468929885e-07, "loss": 0.2455, "step": 24143 }, { "epoch": 1.6691323885240235, "grad_norm": 0.7397778780124973, "learning_rate": 7.009872919698423e-07, "loss": 0.3437, "step": 24144 }, { "epoch": 1.6692015209125475, "grad_norm": 0.7278316392227944, "learning_rate": 7.007014909271486e-07, "loss": 0.288, "step": 24145 }, { "epoch": 1.6692706533010715, "grad_norm": 0.658292226376268, "learning_rate": 7.0041574376849e-07, "loss": 0.1986, "step": 24146 }, { "epoch": 1.6693397856895955, "grad_norm": 0.6960731965299617, "learning_rate": 7.001300504974467e-07, "loss": 0.2583, "step": 24147 }, { "epoch": 1.6694089180781195, "grad_norm": 0.7656596039075796, "learning_rate": 6.998444111176006e-07, "loss": 0.3261, "step": 24148 }, { "epoch": 1.6694780504666435, "grad_norm": 0.721696620847118, "learning_rate": 6.995588256325292e-07, "loss": 0.2099, "step": 24149 }, { "epoch": 1.6695471828551676, "grad_norm": 0.7382914655430629, "learning_rate": 6.992732940458152e-07, "loss": 0.2346, "step": 24150 }, { "epoch": 1.6696163152436916, "grad_norm": 0.7176033770083389, "learning_rate": 6.989878163610342e-07, "loss": 0.2375, "step": 24151 }, { "epoch": 1.6696854476322156, "grad_norm": 0.7477867558845165, "learning_rate": 6.987023925817649e-07, "loss": 0.3165, "step": 24152 }, { "epoch": 1.6697545800207396, "grad_norm": 0.7055089832645012, "learning_rate": 6.984170227115855e-07, "loss": 0.2482, "step": 24153 }, { "epoch": 1.6698237124092636, "grad_norm": 0.7855059383627395, "learning_rate": 6.981317067540738e-07, "loss": 0.2201, "step": 24154 }, { "epoch": 1.6698928447977877, "grad_norm": 0.8054318057263299, "learning_rate": 6.978464447128025e-07, "loss": 0.346, "step": 24155 }, { "epoch": 1.6699619771863117, "grad_norm": 0.6760907745698634, "learning_rate": 6.975612365913498e-07, "loss": 0.224, "step": 24156 }, { "epoch": 1.6700311095748357, "grad_norm": 0.7533396051309529, "learning_rate": 6.972760823932873e-07, "loss": 0.189, "step": 24157 }, { "epoch": 1.6701002419633597, "grad_norm": 0.6903202958647551, "learning_rate": 6.969909821221909e-07, "loss": 0.3073, "step": 24158 }, { "epoch": 1.6701693743518837, "grad_norm": 0.7188196239303101, "learning_rate": 6.967059357816336e-07, "loss": 0.3222, "step": 24159 }, { "epoch": 1.6702385067404077, "grad_norm": 0.7394121003542451, "learning_rate": 6.96420943375189e-07, "loss": 0.234, "step": 24160 }, { "epoch": 1.6703076391289318, "grad_norm": 0.679934325025737, "learning_rate": 6.961360049064264e-07, "loss": 0.219, "step": 24161 }, { "epoch": 1.6703767715174558, "grad_norm": 0.7877478852623644, "learning_rate": 6.958511203789198e-07, "loss": 0.3823, "step": 24162 }, { "epoch": 1.6704459039059798, "grad_norm": 0.6966666566655974, "learning_rate": 6.955662897962368e-07, "loss": 0.2609, "step": 24163 }, { "epoch": 1.6705150362945038, "grad_norm": 0.7747542691841242, "learning_rate": 6.9528151316195e-07, "loss": 0.2187, "step": 24164 }, { "epoch": 1.6705841686830278, "grad_norm": 0.7069802679416144, "learning_rate": 6.949967904796267e-07, "loss": 0.2166, "step": 24165 }, { "epoch": 1.670653301071552, "grad_norm": 0.7180484424863508, "learning_rate": 6.947121217528352e-07, "loss": 0.3307, "step": 24166 }, { "epoch": 1.670722433460076, "grad_norm": 0.6884671910604978, "learning_rate": 6.944275069851448e-07, "loss": 0.2355, "step": 24167 }, { "epoch": 1.6707915658486001, "grad_norm": 0.7463909485158108, "learning_rate": 6.94142946180123e-07, "loss": 0.2585, "step": 24168 }, { "epoch": 1.6708606982371241, "grad_norm": 0.6322765126086789, "learning_rate": 6.938584393413339e-07, "loss": 0.2711, "step": 24169 }, { "epoch": 1.6709298306256481, "grad_norm": 0.6586758628918391, "learning_rate": 6.935739864723457e-07, "loss": 0.208, "step": 24170 }, { "epoch": 1.6709989630141722, "grad_norm": 0.6708935535885353, "learning_rate": 6.932895875767209e-07, "loss": 0.2002, "step": 24171 }, { "epoch": 1.6710680954026962, "grad_norm": 0.6824038306189893, "learning_rate": 6.930052426580264e-07, "loss": 0.24, "step": 24172 }, { "epoch": 1.6711372277912202, "grad_norm": 0.6618005726136529, "learning_rate": 6.927209517198241e-07, "loss": 0.2733, "step": 24173 }, { "epoch": 1.6712063601797442, "grad_norm": 0.7269396283440178, "learning_rate": 6.924367147656797e-07, "loss": 0.2607, "step": 24174 }, { "epoch": 1.6712754925682682, "grad_norm": 1.1687146915821336, "learning_rate": 6.921525317991523e-07, "loss": 0.2505, "step": 24175 }, { "epoch": 1.6713446249567923, "grad_norm": 0.7164797828570381, "learning_rate": 6.91868402823807e-07, "loss": 0.2335, "step": 24176 }, { "epoch": 1.6714137573453163, "grad_norm": 0.6998105824601626, "learning_rate": 6.915843278432016e-07, "loss": 0.3305, "step": 24177 }, { "epoch": 1.6714828897338403, "grad_norm": 0.6583453463039264, "learning_rate": 6.913003068608987e-07, "loss": 0.2498, "step": 24178 }, { "epoch": 1.6715520221223643, "grad_norm": 0.7109696241167109, "learning_rate": 6.910163398804565e-07, "loss": 0.2276, "step": 24179 }, { "epoch": 1.6716211545108883, "grad_norm": 0.6427339277221097, "learning_rate": 6.907324269054344e-07, "loss": 0.3388, "step": 24180 }, { "epoch": 1.6716902868994123, "grad_norm": 0.7320579744975284, "learning_rate": 6.904485679393918e-07, "loss": 0.2894, "step": 24181 }, { "epoch": 1.6717594192879364, "grad_norm": 0.643685563096463, "learning_rate": 6.901647629858859e-07, "loss": 0.245, "step": 24182 }, { "epoch": 1.6718285516764604, "grad_norm": 0.7321274356887619, "learning_rate": 6.898810120484728e-07, "loss": 0.2715, "step": 24183 }, { "epoch": 1.6718976840649844, "grad_norm": 0.7237591168364413, "learning_rate": 6.895973151307106e-07, "loss": 0.3543, "step": 24184 }, { "epoch": 1.6719668164535084, "grad_norm": 0.8022050851915111, "learning_rate": 6.893136722361526e-07, "loss": 0.2446, "step": 24185 }, { "epoch": 1.6720359488420324, "grad_norm": 0.8485853719628685, "learning_rate": 6.890300833683561e-07, "loss": 0.2219, "step": 24186 }, { "epoch": 1.6721050812305567, "grad_norm": 0.7141110167774076, "learning_rate": 6.887465485308736e-07, "loss": 0.3253, "step": 24187 }, { "epoch": 1.6721742136190807, "grad_norm": 0.7173770397391046, "learning_rate": 6.884630677272586e-07, "loss": 0.2611, "step": 24188 }, { "epoch": 1.6722433460076047, "grad_norm": 0.658145116866096, "learning_rate": 6.881796409610653e-07, "loss": 0.2539, "step": 24189 }, { "epoch": 1.6723124783961287, "grad_norm": 0.7218916858747052, "learning_rate": 6.878962682358459e-07, "loss": 0.2331, "step": 24190 }, { "epoch": 1.6723816107846528, "grad_norm": 0.6472286650269015, "learning_rate": 6.876129495551509e-07, "loss": 0.3627, "step": 24191 }, { "epoch": 1.6724507431731768, "grad_norm": 0.7008837209671112, "learning_rate": 6.873296849225325e-07, "loss": 0.2471, "step": 24192 }, { "epoch": 1.6725198755617008, "grad_norm": 0.6598492129099411, "learning_rate": 6.870464743415394e-07, "loss": 0.2966, "step": 24193 }, { "epoch": 1.6725890079502248, "grad_norm": 0.6791784035909599, "learning_rate": 6.867633178157218e-07, "loss": 0.2637, "step": 24194 }, { "epoch": 1.6726581403387488, "grad_norm": 0.6841027111052048, "learning_rate": 6.864802153486288e-07, "loss": 0.2979, "step": 24195 }, { "epoch": 1.6727272727272728, "grad_norm": 0.6851727639399364, "learning_rate": 6.861971669438094e-07, "loss": 0.2621, "step": 24196 }, { "epoch": 1.6727964051157969, "grad_norm": 0.698444011414264, "learning_rate": 6.859141726048091e-07, "loss": 0.2699, "step": 24197 }, { "epoch": 1.6728655375043209, "grad_norm": 0.7010030116699427, "learning_rate": 6.856312323351771e-07, "loss": 0.3327, "step": 24198 }, { "epoch": 1.672934669892845, "grad_norm": 0.6371629742806765, "learning_rate": 6.853483461384569e-07, "loss": 0.1778, "step": 24199 }, { "epoch": 1.673003802281369, "grad_norm": 0.9612437544330226, "learning_rate": 6.850655140181967e-07, "loss": 0.2389, "step": 24200 }, { "epoch": 1.673072934669893, "grad_norm": 0.7087121980990894, "learning_rate": 6.847827359779391e-07, "loss": 0.2951, "step": 24201 }, { "epoch": 1.673142067058417, "grad_norm": 0.7438592013330524, "learning_rate": 6.845000120212287e-07, "loss": 0.3224, "step": 24202 }, { "epoch": 1.673211199446941, "grad_norm": 0.6865908659617039, "learning_rate": 6.84217342151609e-07, "loss": 0.2323, "step": 24203 }, { "epoch": 1.673280331835465, "grad_norm": 0.7427406367046101, "learning_rate": 6.839347263726243e-07, "loss": 0.2438, "step": 24204 }, { "epoch": 1.673349464223989, "grad_norm": 0.7045107309445092, "learning_rate": 6.836521646878147e-07, "loss": 0.3707, "step": 24205 }, { "epoch": 1.673418596612513, "grad_norm": 0.7805312243860906, "learning_rate": 6.833696571007232e-07, "loss": 0.3156, "step": 24206 }, { "epoch": 1.673487729001037, "grad_norm": 0.6826755960037842, "learning_rate": 6.830872036148884e-07, "loss": 0.1964, "step": 24207 }, { "epoch": 1.673556861389561, "grad_norm": 0.7041462711096078, "learning_rate": 6.828048042338519e-07, "loss": 0.2747, "step": 24208 }, { "epoch": 1.673625993778085, "grad_norm": 0.7005581021544204, "learning_rate": 6.825224589611529e-07, "loss": 0.314, "step": 24209 }, { "epoch": 1.673695126166609, "grad_norm": 0.6624082424894822, "learning_rate": 6.82240167800331e-07, "loss": 0.234, "step": 24210 }, { "epoch": 1.673764258555133, "grad_norm": 0.688010875835698, "learning_rate": 6.819579307549229e-07, "loss": 0.2781, "step": 24211 }, { "epoch": 1.6738333909436571, "grad_norm": 0.8023115912606485, "learning_rate": 6.816757478284652e-07, "loss": 0.3293, "step": 24212 }, { "epoch": 1.6739025233321811, "grad_norm": 0.7177863483714867, "learning_rate": 6.813936190244968e-07, "loss": 0.3057, "step": 24213 }, { "epoch": 1.6739716557207052, "grad_norm": 0.6808452126852311, "learning_rate": 6.811115443465515e-07, "loss": 0.242, "step": 24214 }, { "epoch": 1.6740407881092292, "grad_norm": 0.8206278102423639, "learning_rate": 6.80829523798166e-07, "loss": 0.2704, "step": 24215 }, { "epoch": 1.6741099204977532, "grad_norm": 0.6877389177897408, "learning_rate": 6.805475573828741e-07, "loss": 0.358, "step": 24216 }, { "epoch": 1.6741790528862772, "grad_norm": 0.6651393612647767, "learning_rate": 6.802656451042111e-07, "loss": 0.2335, "step": 24217 }, { "epoch": 1.6742481852748012, "grad_norm": 0.681679091013352, "learning_rate": 6.799837869657083e-07, "loss": 0.2548, "step": 24218 }, { "epoch": 1.6743173176633253, "grad_norm": 0.7319743532283549, "learning_rate": 6.797019829709006e-07, "loss": 0.3603, "step": 24219 }, { "epoch": 1.6743864500518493, "grad_norm": 0.8943062045955513, "learning_rate": 6.794202331233174e-07, "loss": 0.2908, "step": 24220 }, { "epoch": 1.6744555824403733, "grad_norm": 0.7070204854760861, "learning_rate": 6.791385374264919e-07, "loss": 0.2837, "step": 24221 }, { "epoch": 1.6745247148288973, "grad_norm": 0.6669666560626513, "learning_rate": 6.788568958839531e-07, "loss": 0.2375, "step": 24222 }, { "epoch": 1.6745938472174213, "grad_norm": 0.6809597835359072, "learning_rate": 6.78575308499232e-07, "loss": 0.3317, "step": 24223 }, { "epoch": 1.6746629796059453, "grad_norm": 0.7136643960455961, "learning_rate": 6.782937752758572e-07, "loss": 0.2512, "step": 24224 }, { "epoch": 1.6747321119944694, "grad_norm": 0.6708317157736876, "learning_rate": 6.780122962173585e-07, "loss": 0.2448, "step": 24225 }, { "epoch": 1.6748012443829934, "grad_norm": 0.8822158747372628, "learning_rate": 6.777308713272618e-07, "loss": 0.279, "step": 24226 }, { "epoch": 1.6748703767715174, "grad_norm": 0.7517131713291825, "learning_rate": 6.774495006090964e-07, "loss": 0.3363, "step": 24227 }, { "epoch": 1.6749395091600414, "grad_norm": 0.6877989920175281, "learning_rate": 6.771681840663863e-07, "loss": 0.2277, "step": 24228 }, { "epoch": 1.6750086415485654, "grad_norm": 0.7024638224445428, "learning_rate": 6.768869217026585e-07, "loss": 0.2459, "step": 24229 }, { "epoch": 1.6750777739370895, "grad_norm": 0.7391378943015486, "learning_rate": 6.766057135214382e-07, "loss": 0.3548, "step": 24230 }, { "epoch": 1.6751469063256135, "grad_norm": 0.6380333505833176, "learning_rate": 6.763245595262513e-07, "loss": 0.2641, "step": 24231 }, { "epoch": 1.6752160387141375, "grad_norm": 0.6660638556138048, "learning_rate": 6.760434597206183e-07, "loss": 0.1953, "step": 24232 }, { "epoch": 1.6752851711026615, "grad_norm": 0.7679688167165505, "learning_rate": 6.75762414108066e-07, "loss": 0.2973, "step": 24233 }, { "epoch": 1.6753543034911855, "grad_norm": 0.7340239648311935, "learning_rate": 6.75481422692113e-07, "loss": 0.3653, "step": 24234 }, { "epoch": 1.6754234358797095, "grad_norm": 0.6358101006693717, "learning_rate": 6.752004854762845e-07, "loss": 0.1966, "step": 24235 }, { "epoch": 1.6754925682682336, "grad_norm": 0.7573154823088751, "learning_rate": 6.749196024640991e-07, "loss": 0.2676, "step": 24236 }, { "epoch": 1.6755617006567576, "grad_norm": 0.698024085421537, "learning_rate": 6.746387736590776e-07, "loss": 0.3308, "step": 24237 }, { "epoch": 1.6756308330452816, "grad_norm": 0.658588872265891, "learning_rate": 6.743579990647403e-07, "loss": 0.2862, "step": 24238 }, { "epoch": 1.6756999654338056, "grad_norm": 0.8436076653702403, "learning_rate": 6.740772786846067e-07, "loss": 0.2517, "step": 24239 }, { "epoch": 1.6757690978223296, "grad_norm": 0.6533374841286632, "learning_rate": 6.73796612522194e-07, "loss": 0.2644, "step": 24240 }, { "epoch": 1.6758382302108537, "grad_norm": 0.7539866948645789, "learning_rate": 6.73516000581021e-07, "loss": 0.364, "step": 24241 }, { "epoch": 1.6759073625993777, "grad_norm": 0.703738174349086, "learning_rate": 6.732354428646032e-07, "loss": 0.2408, "step": 24242 }, { "epoch": 1.6759764949879017, "grad_norm": 0.6674538883447921, "learning_rate": 6.729549393764573e-07, "loss": 0.2182, "step": 24243 }, { "epoch": 1.6760456273764257, "grad_norm": 0.7105895106474713, "learning_rate": 6.726744901200994e-07, "loss": 0.2941, "step": 24244 }, { "epoch": 1.6761147597649497, "grad_norm": 0.8257194330756179, "learning_rate": 6.723940950990454e-07, "loss": 0.3271, "step": 24245 }, { "epoch": 1.6761838921534737, "grad_norm": 0.6644513529303316, "learning_rate": 6.721137543168077e-07, "loss": 0.2125, "step": 24246 }, { "epoch": 1.6762530245419978, "grad_norm": 0.7001373490362022, "learning_rate": 6.718334677769012e-07, "loss": 0.2645, "step": 24247 }, { "epoch": 1.676322156930522, "grad_norm": 0.698533172367472, "learning_rate": 6.715532354828375e-07, "loss": 0.3449, "step": 24248 }, { "epoch": 1.676391289319046, "grad_norm": 0.6804115703683243, "learning_rate": 6.712730574381305e-07, "loss": 0.2112, "step": 24249 }, { "epoch": 1.67646042170757, "grad_norm": 0.6964334423944545, "learning_rate": 6.709929336462895e-07, "loss": 0.2383, "step": 24250 }, { "epoch": 1.676529554096094, "grad_norm": 0.6698437842914036, "learning_rate": 6.707128641108268e-07, "loss": 0.3106, "step": 24251 }, { "epoch": 1.676598686484618, "grad_norm": 0.6941005988026557, "learning_rate": 6.704328488352524e-07, "loss": 0.3301, "step": 24252 }, { "epoch": 1.676667818873142, "grad_norm": 0.8477857380388757, "learning_rate": 6.701528878230768e-07, "loss": 0.2645, "step": 24253 }, { "epoch": 1.676736951261666, "grad_norm": 0.7368582334183779, "learning_rate": 6.698729810778065e-07, "loss": 0.3331, "step": 24254 }, { "epoch": 1.6768060836501901, "grad_norm": 0.7529175857679873, "learning_rate": 6.695931286029522e-07, "loss": 0.3697, "step": 24255 }, { "epoch": 1.6768752160387141, "grad_norm": 0.7011408711355446, "learning_rate": 6.693133304020194e-07, "loss": 0.2382, "step": 24256 }, { "epoch": 1.6769443484272382, "grad_norm": 0.75663204578709, "learning_rate": 6.690335864785164e-07, "loss": 0.2153, "step": 24257 }, { "epoch": 1.6770134808157622, "grad_norm": 0.6822027716827669, "learning_rate": 6.68753896835947e-07, "loss": 0.2473, "step": 24258 }, { "epoch": 1.6770826132042862, "grad_norm": 0.6940647139384064, "learning_rate": 6.684742614778183e-07, "loss": 0.3707, "step": 24259 }, { "epoch": 1.6771517455928102, "grad_norm": 0.7204089914570684, "learning_rate": 6.681946804076345e-07, "loss": 0.2843, "step": 24260 }, { "epoch": 1.6772208779813342, "grad_norm": 0.6610608817766654, "learning_rate": 6.679151536289013e-07, "loss": 0.2595, "step": 24261 }, { "epoch": 1.6772900103698583, "grad_norm": 0.7278315877921235, "learning_rate": 6.676356811451195e-07, "loss": 0.3427, "step": 24262 }, { "epoch": 1.6773591427583823, "grad_norm": 1.0525297761707886, "learning_rate": 6.67356262959794e-07, "loss": 0.263, "step": 24263 }, { "epoch": 1.6774282751469063, "grad_norm": 0.7645063616593495, "learning_rate": 6.670768990764253e-07, "loss": 0.251, "step": 24264 }, { "epoch": 1.6774974075354303, "grad_norm": 0.7167536502795457, "learning_rate": 6.667975894985146e-07, "loss": 0.2197, "step": 24265 }, { "epoch": 1.6775665399239543, "grad_norm": 0.6982875515307596, "learning_rate": 6.665183342295634e-07, "loss": 0.374, "step": 24266 }, { "epoch": 1.6776356723124783, "grad_norm": 0.6881704059992177, "learning_rate": 6.662391332730722e-07, "loss": 0.2221, "step": 24267 }, { "epoch": 1.6777048047010026, "grad_norm": 0.6440648864794175, "learning_rate": 6.659599866325389e-07, "loss": 0.2448, "step": 24268 }, { "epoch": 1.6777739370895266, "grad_norm": 0.7088565780825745, "learning_rate": 6.656808943114634e-07, "loss": 0.3118, "step": 24269 }, { "epoch": 1.6778430694780506, "grad_norm": 0.7540329638035289, "learning_rate": 6.654018563133424e-07, "loss": 0.3167, "step": 24270 }, { "epoch": 1.6779122018665746, "grad_norm": 0.6781510403309303, "learning_rate": 6.651228726416742e-07, "loss": 0.2248, "step": 24271 }, { "epoch": 1.6779813342550987, "grad_norm": 0.7265702514161861, "learning_rate": 6.648439432999543e-07, "loss": 0.2741, "step": 24272 }, { "epoch": 1.6780504666436227, "grad_norm": 0.711194676931559, "learning_rate": 6.645650682916787e-07, "loss": 0.3385, "step": 24273 }, { "epoch": 1.6781195990321467, "grad_norm": 0.7648457355931759, "learning_rate": 6.642862476203432e-07, "loss": 0.2421, "step": 24274 }, { "epoch": 1.6781887314206707, "grad_norm": 0.6105454748460919, "learning_rate": 6.640074812894432e-07, "loss": 0.2133, "step": 24275 }, { "epoch": 1.6782578638091947, "grad_norm": 0.7198780937460848, "learning_rate": 6.637287693024702e-07, "loss": 0.2521, "step": 24276 }, { "epoch": 1.6783269961977187, "grad_norm": 0.7122839733078529, "learning_rate": 6.634501116629199e-07, "loss": 0.2703, "step": 24277 }, { "epoch": 1.6783961285862428, "grad_norm": 0.6700821299548904, "learning_rate": 6.63171508374284e-07, "loss": 0.2864, "step": 24278 }, { "epoch": 1.6784652609747668, "grad_norm": 0.6640288520091213, "learning_rate": 6.628929594400513e-07, "loss": 0.2558, "step": 24279 }, { "epoch": 1.6785343933632908, "grad_norm": 0.7068643715982094, "learning_rate": 6.626144648637172e-07, "loss": 0.3289, "step": 24280 }, { "epoch": 1.6786035257518148, "grad_norm": 0.6895095486535919, "learning_rate": 6.623360246487698e-07, "loss": 0.2537, "step": 24281 }, { "epoch": 1.6786726581403388, "grad_norm": 0.6498993723060295, "learning_rate": 6.620576387987004e-07, "loss": 0.2384, "step": 24282 }, { "epoch": 1.6787417905288629, "grad_norm": 0.7214114878977124, "learning_rate": 6.617793073169959e-07, "loss": 0.2416, "step": 24283 }, { "epoch": 1.6788109229173869, "grad_norm": 0.7522567094081243, "learning_rate": 6.615010302071473e-07, "loss": 0.4159, "step": 24284 }, { "epoch": 1.678880055305911, "grad_norm": 0.7534430717769559, "learning_rate": 6.612228074726396e-07, "loss": 0.2711, "step": 24285 }, { "epoch": 1.678949187694435, "grad_norm": 0.6269330584116928, "learning_rate": 6.609446391169611e-07, "loss": 0.2083, "step": 24286 }, { "epoch": 1.679018320082959, "grad_norm": 0.7240087928572447, "learning_rate": 6.606665251435978e-07, "loss": 0.3123, "step": 24287 }, { "epoch": 1.679087452471483, "grad_norm": 0.719000351249689, "learning_rate": 6.603884655560372e-07, "loss": 0.3268, "step": 24288 }, { "epoch": 1.679156584860007, "grad_norm": 0.663139749750553, "learning_rate": 6.601104603577613e-07, "loss": 0.2319, "step": 24289 }, { "epoch": 1.679225717248531, "grad_norm": 0.7431095814418726, "learning_rate": 6.598325095522573e-07, "loss": 0.2307, "step": 24290 }, { "epoch": 1.679294849637055, "grad_norm": 0.7302473167397049, "learning_rate": 6.595546131430064e-07, "loss": 0.343, "step": 24291 }, { "epoch": 1.679363982025579, "grad_norm": 0.7102218037020097, "learning_rate": 6.592767711334936e-07, "loss": 0.2771, "step": 24292 }, { "epoch": 1.679433114414103, "grad_norm": 0.713409461092603, "learning_rate": 6.589989835271993e-07, "loss": 0.2019, "step": 24293 }, { "epoch": 1.679502246802627, "grad_norm": 0.7347120938239702, "learning_rate": 6.587212503276053e-07, "loss": 0.2787, "step": 24294 }, { "epoch": 1.679571379191151, "grad_norm": 0.7578178793757341, "learning_rate": 6.584435715381937e-07, "loss": 0.3023, "step": 24295 }, { "epoch": 1.679640511579675, "grad_norm": 0.7247670612425395, "learning_rate": 6.581659471624447e-07, "loss": 0.2434, "step": 24296 }, { "epoch": 1.679709643968199, "grad_norm": 0.755591535785679, "learning_rate": 6.578883772038358e-07, "loss": 0.2413, "step": 24297 }, { "epoch": 1.6797787763567231, "grad_norm": 0.7110890865450705, "learning_rate": 6.576108616658489e-07, "loss": 0.3485, "step": 24298 }, { "epoch": 1.6798479087452471, "grad_norm": 0.674313405054744, "learning_rate": 6.573334005519594e-07, "loss": 0.211, "step": 24299 }, { "epoch": 1.6799170411337712, "grad_norm": 0.6503849844221813, "learning_rate": 6.570559938656457e-07, "loss": 0.2087, "step": 24300 }, { "epoch": 1.6799861735222952, "grad_norm": 0.723992344472839, "learning_rate": 6.56778641610385e-07, "loss": 0.2906, "step": 24301 }, { "epoch": 1.6800553059108192, "grad_norm": 0.7226577412545223, "learning_rate": 6.565013437896539e-07, "loss": 0.3525, "step": 24302 }, { "epoch": 1.6801244382993432, "grad_norm": 0.8326319655635506, "learning_rate": 6.562241004069264e-07, "loss": 0.2275, "step": 24303 }, { "epoch": 1.6801935706878672, "grad_norm": 0.6557272859807329, "learning_rate": 6.559469114656791e-07, "loss": 0.2571, "step": 24304 }, { "epoch": 1.6802627030763913, "grad_norm": 0.863631465011114, "learning_rate": 6.556697769693838e-07, "loss": 0.35, "step": 24305 }, { "epoch": 1.6803318354649153, "grad_norm": 0.7152347318465581, "learning_rate": 6.553926969215157e-07, "loss": 0.2694, "step": 24306 }, { "epoch": 1.6804009678534393, "grad_norm": 0.89352911368361, "learning_rate": 6.551156713255463e-07, "loss": 0.2342, "step": 24307 }, { "epoch": 1.6804701002419633, "grad_norm": 0.6444487462168856, "learning_rate": 6.548387001849477e-07, "loss": 0.238, "step": 24308 }, { "epoch": 1.6805392326304873, "grad_norm": 0.7001320889860629, "learning_rate": 6.54561783503192e-07, "loss": 0.3574, "step": 24309 }, { "epoch": 1.6806083650190113, "grad_norm": 0.7347923056108772, "learning_rate": 6.54284921283751e-07, "loss": 0.2232, "step": 24310 }, { "epoch": 1.6806774974075354, "grad_norm": 0.6101750319954886, "learning_rate": 6.540081135300913e-07, "loss": 0.1924, "step": 24311 }, { "epoch": 1.6807466297960594, "grad_norm": 0.6886360179952608, "learning_rate": 6.537313602456857e-07, "loss": 0.3227, "step": 24312 }, { "epoch": 1.6808157621845834, "grad_norm": 0.8379541198176589, "learning_rate": 6.534546614340003e-07, "loss": 0.2611, "step": 24313 }, { "epoch": 1.6808848945731074, "grad_norm": 0.7153144112322709, "learning_rate": 6.531780170985036e-07, "loss": 0.2607, "step": 24314 }, { "epoch": 1.6809540269616314, "grad_norm": 0.9175135774058752, "learning_rate": 6.529014272426637e-07, "loss": 0.2834, "step": 24315 }, { "epoch": 1.6810231593501554, "grad_norm": 0.694460978481664, "learning_rate": 6.526248918699474e-07, "loss": 0.3155, "step": 24316 }, { "epoch": 1.6810922917386795, "grad_norm": 0.8220994733027517, "learning_rate": 6.523484109838185e-07, "loss": 0.2264, "step": 24317 }, { "epoch": 1.6811614241272035, "grad_norm": 0.6792961992542285, "learning_rate": 6.520719845877455e-07, "loss": 0.2304, "step": 24318 }, { "epoch": 1.6812305565157275, "grad_norm": 0.7463813034359806, "learning_rate": 6.517956126851888e-07, "loss": 0.2891, "step": 24319 }, { "epoch": 1.6812996889042515, "grad_norm": 0.6858777696540301, "learning_rate": 6.515192952796162e-07, "loss": 0.2604, "step": 24320 }, { "epoch": 1.6813688212927755, "grad_norm": 0.6786182507783082, "learning_rate": 6.512430323744878e-07, "loss": 0.2017, "step": 24321 }, { "epoch": 1.6814379536812996, "grad_norm": 0.6472428307112681, "learning_rate": 6.509668239732669e-07, "loss": 0.2343, "step": 24322 }, { "epoch": 1.6815070860698236, "grad_norm": 0.681671567929049, "learning_rate": 6.506906700794158e-07, "loss": 0.3427, "step": 24323 }, { "epoch": 1.6815762184583476, "grad_norm": 0.7421627533047528, "learning_rate": 6.504145706963966e-07, "loss": 0.295, "step": 24324 }, { "epoch": 1.6816453508468716, "grad_norm": 0.6636485469116593, "learning_rate": 6.501385258276677e-07, "loss": 0.2002, "step": 24325 }, { "epoch": 1.6817144832353956, "grad_norm": 0.6902549937248015, "learning_rate": 6.498625354766907e-07, "loss": 0.2815, "step": 24326 }, { "epoch": 1.6817836156239196, "grad_norm": 0.7513318815851441, "learning_rate": 6.495865996469225e-07, "loss": 0.3623, "step": 24327 }, { "epoch": 1.6818527480124437, "grad_norm": 0.677206034111778, "learning_rate": 6.493107183418234e-07, "loss": 0.2195, "step": 24328 }, { "epoch": 1.681921880400968, "grad_norm": 0.7197784805679962, "learning_rate": 6.490348915648497e-07, "loss": 0.2182, "step": 24329 }, { "epoch": 1.681991012789492, "grad_norm": 0.7469690587392832, "learning_rate": 6.487591193194587e-07, "loss": 0.3772, "step": 24330 }, { "epoch": 1.682060145178016, "grad_norm": 0.782952455278557, "learning_rate": 6.484834016091074e-07, "loss": 0.2866, "step": 24331 }, { "epoch": 1.68212927756654, "grad_norm": 0.6382627447062393, "learning_rate": 6.482077384372514e-07, "loss": 0.2156, "step": 24332 }, { "epoch": 1.682198409955064, "grad_norm": 0.7351853503744153, "learning_rate": 6.479321298073448e-07, "loss": 0.2809, "step": 24333 }, { "epoch": 1.682267542343588, "grad_norm": 0.7816467255151592, "learning_rate": 6.476565757228432e-07, "loss": 0.3453, "step": 24334 }, { "epoch": 1.682336674732112, "grad_norm": 0.7373755579836383, "learning_rate": 6.473810761871985e-07, "loss": 0.2534, "step": 24335 }, { "epoch": 1.682405807120636, "grad_norm": 0.5864788625733623, "learning_rate": 6.471056312038643e-07, "loss": 0.1942, "step": 24336 }, { "epoch": 1.68247493950916, "grad_norm": 0.7562004717430472, "learning_rate": 6.468302407762928e-07, "loss": 0.3149, "step": 24337 }, { "epoch": 1.682544071897684, "grad_norm": 0.7418255888337393, "learning_rate": 6.465549049079367e-07, "loss": 0.272, "step": 24338 }, { "epoch": 1.682613204286208, "grad_norm": 0.7285532302060408, "learning_rate": 6.462796236022445e-07, "loss": 0.2115, "step": 24339 }, { "epoch": 1.682682336674732, "grad_norm": 0.7203768953491325, "learning_rate": 6.46004396862669e-07, "loss": 0.2179, "step": 24340 }, { "epoch": 1.6827514690632561, "grad_norm": 0.6741896539066878, "learning_rate": 6.457292246926572e-07, "loss": 0.3211, "step": 24341 }, { "epoch": 1.6828206014517801, "grad_norm": 0.7350003302170681, "learning_rate": 6.454541070956605e-07, "loss": 0.2229, "step": 24342 }, { "epoch": 1.6828897338403042, "grad_norm": 0.6822153297433848, "learning_rate": 6.451790440751238e-07, "loss": 0.2626, "step": 24343 }, { "epoch": 1.6829588662288282, "grad_norm": 0.6866885817892608, "learning_rate": 6.449040356344966e-07, "loss": 0.3254, "step": 24344 }, { "epoch": 1.6830279986173522, "grad_norm": 0.7931190387970762, "learning_rate": 6.446290817772266e-07, "loss": 0.2691, "step": 24345 }, { "epoch": 1.6830971310058762, "grad_norm": 0.757624851414268, "learning_rate": 6.443541825067568e-07, "loss": 0.2539, "step": 24346 }, { "epoch": 1.6831662633944002, "grad_norm": 0.7109801868925232, "learning_rate": 6.440793378265358e-07, "loss": 0.1681, "step": 24347 }, { "epoch": 1.6832353957829242, "grad_norm": 0.712017180147389, "learning_rate": 6.438045477400056e-07, "loss": 0.361, "step": 24348 }, { "epoch": 1.6833045281714485, "grad_norm": 0.7988658746419954, "learning_rate": 6.435298122506128e-07, "loss": 0.2582, "step": 24349 }, { "epoch": 1.6833736605599725, "grad_norm": 0.6968554253159533, "learning_rate": 6.432551313617968e-07, "loss": 0.2476, "step": 24350 }, { "epoch": 1.6834427929484965, "grad_norm": 0.727109335680397, "learning_rate": 6.429805050770055e-07, "loss": 0.3075, "step": 24351 }, { "epoch": 1.6835119253370205, "grad_norm": 0.6693492815323777, "learning_rate": 6.427059333996766e-07, "loss": 0.2646, "step": 24352 }, { "epoch": 1.6835810577255446, "grad_norm": 0.7531112284376255, "learning_rate": 6.42431416333254e-07, "loss": 0.2419, "step": 24353 }, { "epoch": 1.6836501901140686, "grad_norm": 0.6615073233832959, "learning_rate": 6.421569538811762e-07, "loss": 0.2409, "step": 24354 }, { "epoch": 1.6837193225025926, "grad_norm": 0.867657218734035, "learning_rate": 6.418825460468847e-07, "loss": 0.3183, "step": 24355 }, { "epoch": 1.6837884548911166, "grad_norm": 0.8333432873457364, "learning_rate": 6.416081928338174e-07, "loss": 0.2605, "step": 24356 }, { "epoch": 1.6838575872796406, "grad_norm": 0.7481779044113959, "learning_rate": 6.41333894245414e-07, "loss": 0.2302, "step": 24357 }, { "epoch": 1.6839267196681647, "grad_norm": 0.7421454571611018, "learning_rate": 6.410596502851107e-07, "loss": 0.3424, "step": 24358 }, { "epoch": 1.6839958520566887, "grad_norm": 0.7298219176548486, "learning_rate": 6.40785460956348e-07, "loss": 0.334, "step": 24359 }, { "epoch": 1.6840649844452127, "grad_norm": 0.6871639350783544, "learning_rate": 6.405113262625585e-07, "loss": 0.2242, "step": 24360 }, { "epoch": 1.6841341168337367, "grad_norm": 0.6469711959383574, "learning_rate": 6.402372462071804e-07, "loss": 0.2151, "step": 24361 }, { "epoch": 1.6842032492222607, "grad_norm": 0.6625709434910346, "learning_rate": 6.39963220793648e-07, "loss": 0.2983, "step": 24362 }, { "epoch": 1.6842723816107847, "grad_norm": 0.7084212788197191, "learning_rate": 6.396892500253959e-07, "loss": 0.3303, "step": 24363 }, { "epoch": 1.6843415139993088, "grad_norm": 0.6283592588918648, "learning_rate": 6.394153339058573e-07, "loss": 0.2663, "step": 24364 }, { "epoch": 1.6844106463878328, "grad_norm": 0.6244133919794269, "learning_rate": 6.391414724384654e-07, "loss": 0.2101, "step": 24365 }, { "epoch": 1.6844797787763568, "grad_norm": 0.7845942324393465, "learning_rate": 6.388676656266529e-07, "loss": 0.3059, "step": 24366 }, { "epoch": 1.6845489111648808, "grad_norm": 0.8018140898420667, "learning_rate": 6.385939134738522e-07, "loss": 0.2256, "step": 24367 }, { "epoch": 1.6846180435534048, "grad_norm": 0.6792422131435504, "learning_rate": 6.383202159834928e-07, "loss": 0.2296, "step": 24368 }, { "epoch": 1.6846871759419288, "grad_norm": 0.7014962094566892, "learning_rate": 6.380465731590069e-07, "loss": 0.3443, "step": 24369 }, { "epoch": 1.6847563083304529, "grad_norm": 0.7264222209542827, "learning_rate": 6.377729850038217e-07, "loss": 0.3065, "step": 24370 }, { "epoch": 1.6848254407189769, "grad_norm": 0.7025311726930633, "learning_rate": 6.374994515213673e-07, "loss": 0.2385, "step": 24371 }, { "epoch": 1.684894573107501, "grad_norm": 0.6746016119500303, "learning_rate": 6.37225972715072e-07, "loss": 0.1948, "step": 24372 }, { "epoch": 1.684963705496025, "grad_norm": 0.7852357158292278, "learning_rate": 6.369525485883643e-07, "loss": 0.3397, "step": 24373 }, { "epoch": 1.685032837884549, "grad_norm": 0.7243853509526238, "learning_rate": 6.366791791446697e-07, "loss": 0.2358, "step": 24374 }, { "epoch": 1.685101970273073, "grad_norm": 0.6628247939703588, "learning_rate": 6.364058643874155e-07, "loss": 0.2621, "step": 24375 }, { "epoch": 1.685171102661597, "grad_norm": 0.7481945338423783, "learning_rate": 6.361326043200255e-07, "loss": 0.2732, "step": 24376 }, { "epoch": 1.685240235050121, "grad_norm": 0.7428760806073106, "learning_rate": 6.358593989459272e-07, "loss": 0.3533, "step": 24377 }, { "epoch": 1.685309367438645, "grad_norm": 0.7188777386554461, "learning_rate": 6.355862482685421e-07, "loss": 0.2351, "step": 24378 }, { "epoch": 1.685378499827169, "grad_norm": 0.661280637225402, "learning_rate": 6.353131522912942e-07, "loss": 0.2585, "step": 24379 }, { "epoch": 1.685447632215693, "grad_norm": 0.7470553966171087, "learning_rate": 6.350401110176068e-07, "loss": 0.3806, "step": 24380 }, { "epoch": 1.685516764604217, "grad_norm": 0.6508508894439669, "learning_rate": 6.347671244509035e-07, "loss": 0.2835, "step": 24381 }, { "epoch": 1.685585896992741, "grad_norm": 0.7005081767654682, "learning_rate": 6.34494192594603e-07, "loss": 0.2823, "step": 24382 }, { "epoch": 1.685655029381265, "grad_norm": 0.75430754326951, "learning_rate": 6.342213154521282e-07, "loss": 0.294, "step": 24383 }, { "epoch": 1.6857241617697891, "grad_norm": 0.7627556833950487, "learning_rate": 6.339484930268969e-07, "loss": 0.3914, "step": 24384 }, { "epoch": 1.6857932941583131, "grad_norm": 0.7024128603125731, "learning_rate": 6.336757253223314e-07, "loss": 0.2408, "step": 24385 }, { "epoch": 1.6858624265468372, "grad_norm": 0.7154074454173059, "learning_rate": 6.334030123418455e-07, "loss": 0.2763, "step": 24386 }, { "epoch": 1.6859315589353612, "grad_norm": 0.6564359897931051, "learning_rate": 6.331303540888634e-07, "loss": 0.3008, "step": 24387 }, { "epoch": 1.6860006913238852, "grad_norm": 0.7172896394789577, "learning_rate": 6.328577505667977e-07, "loss": 0.2797, "step": 24388 }, { "epoch": 1.6860698237124092, "grad_norm": 0.6365983321393678, "learning_rate": 6.325852017790673e-07, "loss": 0.2348, "step": 24389 }, { "epoch": 1.6861389561009332, "grad_norm": 0.6758038143898135, "learning_rate": 6.323127077290869e-07, "loss": 0.2253, "step": 24390 }, { "epoch": 1.6862080884894572, "grad_norm": 0.6783686637834115, "learning_rate": 6.320402684202731e-07, "loss": 0.3531, "step": 24391 }, { "epoch": 1.6862772208779813, "grad_norm": 1.0847058505218448, "learning_rate": 6.317678838560387e-07, "loss": 0.2892, "step": 24392 }, { "epoch": 1.6863463532665053, "grad_norm": 0.6357999589503548, "learning_rate": 6.314955540397982e-07, "loss": 0.2525, "step": 24393 }, { "epoch": 1.6864154856550293, "grad_norm": 0.679416430981246, "learning_rate": 6.31223278974965e-07, "loss": 0.2955, "step": 24394 }, { "epoch": 1.6864846180435533, "grad_norm": 0.736675338662312, "learning_rate": 6.309510586649525e-07, "loss": 0.2512, "step": 24395 }, { "epoch": 1.6865537504320773, "grad_norm": 0.64783435353802, "learning_rate": 6.306788931131707e-07, "loss": 0.2137, "step": 24396 }, { "epoch": 1.6866228828206014, "grad_norm": 0.7185734410437052, "learning_rate": 6.304067823230325e-07, "loss": 0.2556, "step": 24397 }, { "epoch": 1.6866920152091254, "grad_norm": 0.7175585151186885, "learning_rate": 6.301347262979463e-07, "loss": 0.3866, "step": 24398 }, { "epoch": 1.6867611475976494, "grad_norm": 0.6836596748243811, "learning_rate": 6.298627250413242e-07, "loss": 0.2448, "step": 24399 }, { "epoch": 1.6868302799861734, "grad_norm": 0.628383238698926, "learning_rate": 6.29590778556573e-07, "loss": 0.207, "step": 24400 }, { "epoch": 1.6868994123746974, "grad_norm": 0.7325637121303384, "learning_rate": 6.293188868471017e-07, "loss": 0.2791, "step": 24401 }, { "epoch": 1.6869685447632214, "grad_norm": 0.7473087622116961, "learning_rate": 6.290470499163182e-07, "loss": 0.3239, "step": 24402 }, { "epoch": 1.6870376771517455, "grad_norm": 0.7179925629840763, "learning_rate": 6.287752677676312e-07, "loss": 0.1986, "step": 24403 }, { "epoch": 1.6871068095402695, "grad_norm": 0.6643356422558074, "learning_rate": 6.285035404044443e-07, "loss": 0.2374, "step": 24404 }, { "epoch": 1.6871759419287935, "grad_norm": 0.6764003012583769, "learning_rate": 6.282318678301647e-07, "loss": 0.3676, "step": 24405 }, { "epoch": 1.6872450743173175, "grad_norm": 0.6955189380922651, "learning_rate": 6.279602500481963e-07, "loss": 0.2738, "step": 24406 }, { "epoch": 1.6873142067058415, "grad_norm": 0.7185716514734973, "learning_rate": 6.276886870619442e-07, "loss": 0.2353, "step": 24407 }, { "epoch": 1.6873833390943656, "grad_norm": 0.6693449134130612, "learning_rate": 6.274171788748113e-07, "loss": 0.2547, "step": 24408 }, { "epoch": 1.6874524714828896, "grad_norm": 0.7359667187147371, "learning_rate": 6.271457254902013e-07, "loss": 0.3506, "step": 24409 }, { "epoch": 1.6875216038714138, "grad_norm": 0.819647885324562, "learning_rate": 6.268743269115168e-07, "loss": 0.2383, "step": 24410 }, { "epoch": 1.6875907362599378, "grad_norm": 0.6884334486674643, "learning_rate": 6.266029831421572e-07, "loss": 0.2425, "step": 24411 }, { "epoch": 1.6876598686484618, "grad_norm": 0.7533352039590594, "learning_rate": 6.263316941855252e-07, "loss": 0.3285, "step": 24412 }, { "epoch": 1.6877290010369859, "grad_norm": 0.7712721666541454, "learning_rate": 6.260604600450199e-07, "loss": 0.2672, "step": 24413 }, { "epoch": 1.6877981334255099, "grad_norm": 0.6753906888242336, "learning_rate": 6.257892807240401e-07, "loss": 0.2334, "step": 24414 }, { "epoch": 1.687867265814034, "grad_norm": 0.643217366894203, "learning_rate": 6.255181562259866e-07, "loss": 0.2118, "step": 24415 }, { "epoch": 1.687936398202558, "grad_norm": 0.701291560485945, "learning_rate": 6.252470865542565e-07, "loss": 0.3434, "step": 24416 }, { "epoch": 1.688005530591082, "grad_norm": 0.7131273507438637, "learning_rate": 6.249760717122466e-07, "loss": 0.2331, "step": 24417 }, { "epoch": 1.688074662979606, "grad_norm": 0.6984800438071338, "learning_rate": 6.247051117033548e-07, "loss": 0.2791, "step": 24418 }, { "epoch": 1.68814379536813, "grad_norm": 0.7174275110182601, "learning_rate": 6.244342065309755e-07, "loss": 0.256, "step": 24419 }, { "epoch": 1.688212927756654, "grad_norm": 0.6789333609112922, "learning_rate": 6.241633561985061e-07, "loss": 0.2895, "step": 24420 }, { "epoch": 1.688282060145178, "grad_norm": 0.8696691696355155, "learning_rate": 6.238925607093377e-07, "loss": 0.2396, "step": 24421 }, { "epoch": 1.688351192533702, "grad_norm": 0.6492457121441763, "learning_rate": 6.236218200668687e-07, "loss": 0.2304, "step": 24422 }, { "epoch": 1.688420324922226, "grad_norm": 0.7302312792133538, "learning_rate": 6.233511342744891e-07, "loss": 0.3273, "step": 24423 }, { "epoch": 1.68848945731075, "grad_norm": 0.6941923059776481, "learning_rate": 6.230805033355935e-07, "loss": 0.2389, "step": 24424 }, { "epoch": 1.688558589699274, "grad_norm": 0.5914826956914631, "learning_rate": 6.228099272535715e-07, "loss": 0.1849, "step": 24425 }, { "epoch": 1.688627722087798, "grad_norm": 0.7732531337089866, "learning_rate": 6.225394060318174e-07, "loss": 0.3193, "step": 24426 }, { "epoch": 1.6886968544763221, "grad_norm": 0.716142105280044, "learning_rate": 6.222689396737181e-07, "loss": 0.3099, "step": 24427 }, { "epoch": 1.6887659868648461, "grad_norm": 0.7439833516894206, "learning_rate": 6.219985281826657e-07, "loss": 0.2463, "step": 24428 }, { "epoch": 1.6888351192533702, "grad_norm": 0.7325033371827407, "learning_rate": 6.217281715620487e-07, "loss": 0.2537, "step": 24429 }, { "epoch": 1.6889042516418942, "grad_norm": 0.7457557093666893, "learning_rate": 6.214578698152568e-07, "loss": 0.358, "step": 24430 }, { "epoch": 1.6889733840304184, "grad_norm": 0.7600671175643755, "learning_rate": 6.211876229456759e-07, "loss": 0.26, "step": 24431 }, { "epoch": 1.6890425164189424, "grad_norm": 0.6968830409235348, "learning_rate": 6.209174309566946e-07, "loss": 0.2393, "step": 24432 }, { "epoch": 1.6891116488074664, "grad_norm": 0.7052443636246842, "learning_rate": 6.206472938516972e-07, "loss": 0.2488, "step": 24433 }, { "epoch": 1.6891807811959905, "grad_norm": 0.7690904612288736, "learning_rate": 6.203772116340718e-07, "loss": 0.3526, "step": 24434 }, { "epoch": 1.6892499135845145, "grad_norm": 0.6574844137513237, "learning_rate": 6.201071843072015e-07, "loss": 0.2288, "step": 24435 }, { "epoch": 1.6893190459730385, "grad_norm": 0.6399902293128675, "learning_rate": 6.198372118744711e-07, "loss": 0.2403, "step": 24436 }, { "epoch": 1.6893881783615625, "grad_norm": 0.6916002616230053, "learning_rate": 6.195672943392644e-07, "loss": 0.2865, "step": 24437 }, { "epoch": 1.6894573107500865, "grad_norm": 0.7212424839310966, "learning_rate": 6.192974317049661e-07, "loss": 0.2639, "step": 24438 }, { "epoch": 1.6895264431386106, "grad_norm": 0.7129506545299007, "learning_rate": 6.190276239749549e-07, "loss": 0.2091, "step": 24439 }, { "epoch": 1.6895955755271346, "grad_norm": 0.6951484607893832, "learning_rate": 6.18757871152616e-07, "loss": 0.243, "step": 24440 }, { "epoch": 1.6896647079156586, "grad_norm": 0.6245676841345705, "learning_rate": 6.184881732413267e-07, "loss": 0.3149, "step": 24441 }, { "epoch": 1.6897338403041826, "grad_norm": 0.7463079181212419, "learning_rate": 6.182185302444699e-07, "loss": 0.2927, "step": 24442 }, { "epoch": 1.6898029726927066, "grad_norm": 0.6821652262709854, "learning_rate": 6.179489421654233e-07, "loss": 0.248, "step": 24443 }, { "epoch": 1.6898721050812306, "grad_norm": 0.7239224978509754, "learning_rate": 6.176794090075683e-07, "loss": 0.3234, "step": 24444 }, { "epoch": 1.6899412374697547, "grad_norm": 0.6865117405057096, "learning_rate": 6.174099307742798e-07, "loss": 0.3155, "step": 24445 }, { "epoch": 1.6900103698582787, "grad_norm": 0.7814333887106811, "learning_rate": 6.171405074689379e-07, "loss": 0.2231, "step": 24446 }, { "epoch": 1.6900795022468027, "grad_norm": 0.7394304067510117, "learning_rate": 6.168711390949172e-07, "loss": 0.2818, "step": 24447 }, { "epoch": 1.6901486346353267, "grad_norm": 0.7421284829212295, "learning_rate": 6.166018256555961e-07, "loss": 0.3792, "step": 24448 }, { "epoch": 1.6902177670238507, "grad_norm": 1.3216987100318227, "learning_rate": 6.163325671543469e-07, "loss": 0.2528, "step": 24449 }, { "epoch": 1.6902868994123748, "grad_norm": 0.7545449536847162, "learning_rate": 6.160633635945462e-07, "loss": 0.182, "step": 24450 }, { "epoch": 1.6903560318008988, "grad_norm": 0.7430760818019571, "learning_rate": 6.157942149795682e-07, "loss": 0.2659, "step": 24451 }, { "epoch": 1.6904251641894228, "grad_norm": 0.7186714358789005, "learning_rate": 6.155251213127867e-07, "loss": 0.3101, "step": 24452 }, { "epoch": 1.6904942965779468, "grad_norm": 0.7197996360217571, "learning_rate": 6.152560825975718e-07, "loss": 0.2618, "step": 24453 }, { "epoch": 1.6905634289664708, "grad_norm": 0.7099094420483311, "learning_rate": 6.14987098837298e-07, "loss": 0.2174, "step": 24454 }, { "epoch": 1.6906325613549948, "grad_norm": 0.7325003302258534, "learning_rate": 6.14718170035335e-07, "loss": 0.3763, "step": 24455 }, { "epoch": 1.6907016937435189, "grad_norm": 0.696361025803882, "learning_rate": 6.144492961950543e-07, "loss": 0.2392, "step": 24456 }, { "epoch": 1.6907708261320429, "grad_norm": 0.7746583219847368, "learning_rate": 6.14180477319824e-07, "loss": 0.2089, "step": 24457 }, { "epoch": 1.690839958520567, "grad_norm": 0.676449217834983, "learning_rate": 6.139117134130162e-07, "loss": 0.2609, "step": 24458 }, { "epoch": 1.690909090909091, "grad_norm": 0.7014684527167779, "learning_rate": 6.136430044779968e-07, "loss": 0.3405, "step": 24459 }, { "epoch": 1.690978223297615, "grad_norm": 0.7411918869890531, "learning_rate": 6.13374350518135e-07, "loss": 0.2141, "step": 24460 }, { "epoch": 1.691047355686139, "grad_norm": 0.6630613220176788, "learning_rate": 6.131057515367972e-07, "loss": 0.2359, "step": 24461 }, { "epoch": 1.691116488074663, "grad_norm": 0.6717096330907735, "learning_rate": 6.128372075373507e-07, "loss": 0.3419, "step": 24462 }, { "epoch": 1.691185620463187, "grad_norm": 0.6577929981438612, "learning_rate": 6.125687185231593e-07, "loss": 0.2631, "step": 24463 }, { "epoch": 1.691254752851711, "grad_norm": 0.6619120315841901, "learning_rate": 6.123002844975901e-07, "loss": 0.2033, "step": 24464 }, { "epoch": 1.691323885240235, "grad_norm": 0.7653625356925475, "learning_rate": 6.120319054640056e-07, "loss": 0.245, "step": 24465 }, { "epoch": 1.691393017628759, "grad_norm": 0.6657046199870404, "learning_rate": 6.117635814257722e-07, "loss": 0.4019, "step": 24466 }, { "epoch": 1.691462150017283, "grad_norm": 1.04721338413598, "learning_rate": 6.114953123862499e-07, "loss": 0.2418, "step": 24467 }, { "epoch": 1.691531282405807, "grad_norm": 0.5919638516065722, "learning_rate": 6.112270983488033e-07, "loss": 0.1752, "step": 24468 }, { "epoch": 1.691600414794331, "grad_norm": 0.6735170195932287, "learning_rate": 6.10958939316792e-07, "loss": 0.3248, "step": 24469 }, { "epoch": 1.6916695471828551, "grad_norm": 0.7123886412797372, "learning_rate": 6.106908352935787e-07, "loss": 0.3284, "step": 24470 }, { "epoch": 1.6917386795713791, "grad_norm": 0.6528437502529234, "learning_rate": 6.104227862825218e-07, "loss": 0.2042, "step": 24471 }, { "epoch": 1.6918078119599032, "grad_norm": 0.665065176004666, "learning_rate": 6.101547922869816e-07, "loss": 0.2508, "step": 24472 }, { "epoch": 1.6918769443484272, "grad_norm": 0.7331899519355833, "learning_rate": 6.098868533103169e-07, "loss": 0.3858, "step": 24473 }, { "epoch": 1.6919460767369512, "grad_norm": 0.663197817176624, "learning_rate": 6.096189693558869e-07, "loss": 0.2495, "step": 24474 }, { "epoch": 1.6920152091254752, "grad_norm": 0.8153022279635584, "learning_rate": 6.093511404270475e-07, "loss": 0.2426, "step": 24475 }, { "epoch": 1.6920843415139992, "grad_norm": 0.6866866923047049, "learning_rate": 6.090833665271572e-07, "loss": 0.2934, "step": 24476 }, { "epoch": 1.6921534739025232, "grad_norm": 0.65756424682206, "learning_rate": 6.088156476595696e-07, "loss": 0.3144, "step": 24477 }, { "epoch": 1.6922226062910473, "grad_norm": 0.6305308555528532, "learning_rate": 6.085479838276415e-07, "loss": 0.2084, "step": 24478 }, { "epoch": 1.6922917386795713, "grad_norm": 0.7075326803083921, "learning_rate": 6.082803750347288e-07, "loss": 0.2617, "step": 24479 }, { "epoch": 1.6923608710680953, "grad_norm": 0.7665866680085569, "learning_rate": 6.080128212841829e-07, "loss": 0.3702, "step": 24480 }, { "epoch": 1.6924300034566193, "grad_norm": 0.653684677040364, "learning_rate": 6.077453225793595e-07, "loss": 0.2211, "step": 24481 }, { "epoch": 1.6924991358451433, "grad_norm": 0.6286969723704913, "learning_rate": 6.074778789236091e-07, "loss": 0.2102, "step": 24482 }, { "epoch": 1.6925682682336674, "grad_norm": 0.6125067325223619, "learning_rate": 6.072104903202857e-07, "loss": 0.1985, "step": 24483 }, { "epoch": 1.6926374006221914, "grad_norm": 0.6841326972980117, "learning_rate": 6.069431567727385e-07, "loss": 0.3536, "step": 24484 }, { "epoch": 1.6927065330107154, "grad_norm": 0.6516150545488221, "learning_rate": 6.066758782843191e-07, "loss": 0.2801, "step": 24485 }, { "epoch": 1.6927756653992394, "grad_norm": 0.6384500751733936, "learning_rate": 6.064086548583776e-07, "loss": 0.2366, "step": 24486 }, { "epoch": 1.6928447977877634, "grad_norm": 0.6600319706973442, "learning_rate": 6.061414864982635e-07, "loss": 0.3104, "step": 24487 }, { "epoch": 1.6929139301762874, "grad_norm": 0.6950543277476233, "learning_rate": 6.058743732073241e-07, "loss": 0.3031, "step": 24488 }, { "epoch": 1.6929830625648115, "grad_norm": 0.6420784134570507, "learning_rate": 6.056073149889086e-07, "loss": 0.1991, "step": 24489 }, { "epoch": 1.6930521949533355, "grad_norm": 0.8917556334554427, "learning_rate": 6.053403118463618e-07, "loss": 0.3063, "step": 24490 }, { "epoch": 1.6931213273418597, "grad_norm": 0.7072913184453798, "learning_rate": 6.05073363783033e-07, "loss": 0.3573, "step": 24491 }, { "epoch": 1.6931904597303837, "grad_norm": 0.6938390020513515, "learning_rate": 6.048064708022644e-07, "loss": 0.2297, "step": 24492 }, { "epoch": 1.6932595921189078, "grad_norm": 0.6568556052343587, "learning_rate": 6.045396329074055e-07, "loss": 0.1899, "step": 24493 }, { "epoch": 1.6933287245074318, "grad_norm": 0.7642256168207068, "learning_rate": 6.042728501017964e-07, "loss": 0.2713, "step": 24494 }, { "epoch": 1.6933978568959558, "grad_norm": 0.7291541999950844, "learning_rate": 6.040061223887845e-07, "loss": 0.2577, "step": 24495 }, { "epoch": 1.6934669892844798, "grad_norm": 1.4382380947545306, "learning_rate": 6.037394497717092e-07, "loss": 0.2416, "step": 24496 }, { "epoch": 1.6935361216730038, "grad_norm": 0.7012947536069919, "learning_rate": 6.034728322539157e-07, "loss": 0.2408, "step": 24497 }, { "epoch": 1.6936052540615278, "grad_norm": 0.691828895153031, "learning_rate": 6.032062698387431e-07, "loss": 0.368, "step": 24498 }, { "epoch": 1.6936743864500519, "grad_norm": 0.6782406928810201, "learning_rate": 6.029397625295336e-07, "loss": 0.2566, "step": 24499 }, { "epoch": 1.6937435188385759, "grad_norm": 0.6953862713028562, "learning_rate": 6.026733103296272e-07, "loss": 0.2208, "step": 24500 }, { "epoch": 1.6938126512271, "grad_norm": 0.7004534804220516, "learning_rate": 6.024069132423649e-07, "loss": 0.2851, "step": 24501 }, { "epoch": 1.693881783615624, "grad_norm": 0.7640285227033764, "learning_rate": 6.021405712710826e-07, "loss": 0.3482, "step": 24502 }, { "epoch": 1.693950916004148, "grad_norm": 0.6638375839761512, "learning_rate": 6.018742844191211e-07, "loss": 0.2287, "step": 24503 }, { "epoch": 1.694020048392672, "grad_norm": 0.7039792562799796, "learning_rate": 6.016080526898155e-07, "loss": 0.2663, "step": 24504 }, { "epoch": 1.694089180781196, "grad_norm": 0.7645126401969736, "learning_rate": 6.013418760865048e-07, "loss": 0.3371, "step": 24505 }, { "epoch": 1.69415831316972, "grad_norm": 1.3324663986637857, "learning_rate": 6.010757546125229e-07, "loss": 0.2495, "step": 24506 }, { "epoch": 1.694227445558244, "grad_norm": 0.6383869014178563, "learning_rate": 6.008096882712061e-07, "loss": 0.2801, "step": 24507 }, { "epoch": 1.694296577946768, "grad_norm": 0.6667321542011139, "learning_rate": 6.005436770658896e-07, "loss": 0.2774, "step": 24508 }, { "epoch": 1.694365710335292, "grad_norm": 0.6940752753937063, "learning_rate": 6.002777209999072e-07, "loss": 0.3408, "step": 24509 }, { "epoch": 1.694434842723816, "grad_norm": 0.7326993169533758, "learning_rate": 6.000118200765914e-07, "loss": 0.262, "step": 24510 }, { "epoch": 1.69450397511234, "grad_norm": 0.8706235424630936, "learning_rate": 5.997459742992762e-07, "loss": 0.2546, "step": 24511 }, { "epoch": 1.6945731075008643, "grad_norm": 0.7823845810964378, "learning_rate": 5.994801836712916e-07, "loss": 0.2923, "step": 24512 }, { "epoch": 1.6946422398893883, "grad_norm": 0.8315866057061295, "learning_rate": 5.9921444819597e-07, "loss": 0.2648, "step": 24513 }, { "epoch": 1.6947113722779124, "grad_norm": 0.6192800454842224, "learning_rate": 5.989487678766415e-07, "loss": 0.2249, "step": 24514 }, { "epoch": 1.6947805046664364, "grad_norm": 0.6935162126765348, "learning_rate": 5.986831427166367e-07, "loss": 0.2558, "step": 24515 }, { "epoch": 1.6948496370549604, "grad_norm": 0.8240292231783781, "learning_rate": 5.984175727192837e-07, "loss": 0.3491, "step": 24516 }, { "epoch": 1.6949187694434844, "grad_norm": 0.775682042330296, "learning_rate": 5.98152057887913e-07, "loss": 0.248, "step": 24517 }, { "epoch": 1.6949879018320084, "grad_norm": 0.6558086707336789, "learning_rate": 5.978865982258492e-07, "loss": 0.2064, "step": 24518 }, { "epoch": 1.6950570342205324, "grad_norm": 0.8889175551121493, "learning_rate": 5.976211937364218e-07, "loss": 0.288, "step": 24519 }, { "epoch": 1.6951261666090565, "grad_norm": 0.7636352826170472, "learning_rate": 5.973558444229554e-07, "loss": 0.3166, "step": 24520 }, { "epoch": 1.6951952989975805, "grad_norm": 0.7327839446814121, "learning_rate": 5.970905502887769e-07, "loss": 0.2341, "step": 24521 }, { "epoch": 1.6952644313861045, "grad_norm": 0.6612506946655673, "learning_rate": 5.968253113372108e-07, "loss": 0.256, "step": 24522 }, { "epoch": 1.6953335637746285, "grad_norm": 0.7111480911710265, "learning_rate": 5.965601275715826e-07, "loss": 0.319, "step": 24523 }, { "epoch": 1.6954026961631525, "grad_norm": 0.6879410634695903, "learning_rate": 5.962949989952143e-07, "loss": 0.293, "step": 24524 }, { "epoch": 1.6954718285516766, "grad_norm": 0.6148386815510821, "learning_rate": 5.960299256114299e-07, "loss": 0.2114, "step": 24525 }, { "epoch": 1.6955409609402006, "grad_norm": 0.7070166853062392, "learning_rate": 5.957649074235505e-07, "loss": 0.2727, "step": 24526 }, { "epoch": 1.6956100933287246, "grad_norm": 0.7002554697095155, "learning_rate": 5.954999444348991e-07, "loss": 0.3185, "step": 24527 }, { "epoch": 1.6956792257172486, "grad_norm": 0.7100151591329465, "learning_rate": 5.952350366487935e-07, "loss": 0.2262, "step": 24528 }, { "epoch": 1.6957483581057726, "grad_norm": 0.7713711536382385, "learning_rate": 5.949701840685585e-07, "loss": 0.3052, "step": 24529 }, { "epoch": 1.6958174904942966, "grad_norm": 0.9761617001869717, "learning_rate": 5.9470538669751e-07, "loss": 0.36, "step": 24530 }, { "epoch": 1.6958866228828207, "grad_norm": 0.793352089383267, "learning_rate": 5.944406445389689e-07, "loss": 0.2597, "step": 24531 }, { "epoch": 1.6959557552713447, "grad_norm": 0.6419538950947056, "learning_rate": 5.941759575962508e-07, "loss": 0.1871, "step": 24532 }, { "epoch": 1.6960248876598687, "grad_norm": 0.672301898148881, "learning_rate": 5.939113258726764e-07, "loss": 0.2557, "step": 24533 }, { "epoch": 1.6960940200483927, "grad_norm": 0.7514384648451905, "learning_rate": 5.936467493715586e-07, "loss": 0.3402, "step": 24534 }, { "epoch": 1.6961631524369167, "grad_norm": 0.7097764290685407, "learning_rate": 5.93382228096216e-07, "loss": 0.2766, "step": 24535 }, { "epoch": 1.6962322848254408, "grad_norm": 0.630793184990889, "learning_rate": 5.931177620499628e-07, "loss": 0.2079, "step": 24536 }, { "epoch": 1.6963014172139648, "grad_norm": 0.7210188363336018, "learning_rate": 5.928533512361151e-07, "loss": 0.328, "step": 24537 }, { "epoch": 1.6963705496024888, "grad_norm": 0.7434907870547695, "learning_rate": 5.925889956579845e-07, "loss": 0.2629, "step": 24538 }, { "epoch": 1.6964396819910128, "grad_norm": 0.7696458036955022, "learning_rate": 5.923246953188872e-07, "loss": 0.2601, "step": 24539 }, { "epoch": 1.6965088143795368, "grad_norm": 0.6371083495283563, "learning_rate": 5.920604502221322e-07, "loss": 0.2369, "step": 24540 }, { "epoch": 1.6965779467680608, "grad_norm": 0.6734240853748832, "learning_rate": 5.917962603710348e-07, "loss": 0.4127, "step": 24541 }, { "epoch": 1.6966470791565849, "grad_norm": 0.7060333988098051, "learning_rate": 5.915321257689028e-07, "loss": 0.2068, "step": 24542 }, { "epoch": 1.6967162115451089, "grad_norm": 0.6497502828642316, "learning_rate": 5.912680464190485e-07, "loss": 0.2365, "step": 24543 }, { "epoch": 1.696785343933633, "grad_norm": 0.6674698358133874, "learning_rate": 5.910040223247826e-07, "loss": 0.2422, "step": 24544 }, { "epoch": 1.696854476322157, "grad_norm": 0.8342076829518532, "learning_rate": 5.907400534894125e-07, "loss": 0.2683, "step": 24545 }, { "epoch": 1.696923608710681, "grad_norm": 0.630027867889467, "learning_rate": 5.904761399162479e-07, "loss": 0.209, "step": 24546 }, { "epoch": 1.696992741099205, "grad_norm": 0.6882886103004485, "learning_rate": 5.902122816085942e-07, "loss": 0.2754, "step": 24547 }, { "epoch": 1.697061873487729, "grad_norm": 0.7943464871651337, "learning_rate": 5.899484785697607e-07, "loss": 0.3045, "step": 24548 }, { "epoch": 1.697131005876253, "grad_norm": 1.127666011126283, "learning_rate": 5.896847308030524e-07, "loss": 0.2354, "step": 24549 }, { "epoch": 1.697200138264777, "grad_norm": 0.6396373237326188, "learning_rate": 5.894210383117765e-07, "loss": 0.1995, "step": 24550 }, { "epoch": 1.697269270653301, "grad_norm": 0.8306370614137462, "learning_rate": 5.891574010992362e-07, "loss": 0.2358, "step": 24551 }, { "epoch": 1.697338403041825, "grad_norm": 0.7313002782400433, "learning_rate": 5.888938191687371e-07, "loss": 0.2855, "step": 24552 }, { "epoch": 1.697407535430349, "grad_norm": 0.6864526262564423, "learning_rate": 5.88630292523581e-07, "loss": 0.2433, "step": 24553 }, { "epoch": 1.697476667818873, "grad_norm": 0.6745066234095711, "learning_rate": 5.883668211670729e-07, "loss": 0.2414, "step": 24554 }, { "epoch": 1.697545800207397, "grad_norm": 0.6950282172510711, "learning_rate": 5.881034051025125e-07, "loss": 0.297, "step": 24555 }, { "epoch": 1.6976149325959211, "grad_norm": 0.6890075086193304, "learning_rate": 5.878400443332028e-07, "loss": 0.2541, "step": 24556 }, { "epoch": 1.6976840649844451, "grad_norm": 0.6597162553582417, "learning_rate": 5.875767388624443e-07, "loss": 0.2643, "step": 24557 }, { "epoch": 1.6977531973729691, "grad_norm": 0.7605457780957928, "learning_rate": 5.873134886935383e-07, "loss": 0.3487, "step": 24558 }, { "epoch": 1.6978223297614932, "grad_norm": 0.6893620668950596, "learning_rate": 5.870502938297812e-07, "loss": 0.3089, "step": 24559 }, { "epoch": 1.6978914621500172, "grad_norm": 0.6841364918742502, "learning_rate": 5.867871542744752e-07, "loss": 0.2287, "step": 24560 }, { "epoch": 1.6979605945385412, "grad_norm": 0.6326217751083281, "learning_rate": 5.86524070030915e-07, "loss": 0.2248, "step": 24561 }, { "epoch": 1.6980297269270652, "grad_norm": 0.7105793630174029, "learning_rate": 5.862610411024006e-07, "loss": 0.3693, "step": 24562 }, { "epoch": 1.6980988593155892, "grad_norm": 0.7234074710768648, "learning_rate": 5.85998067492225e-07, "loss": 0.3002, "step": 24563 }, { "epoch": 1.6981679917041133, "grad_norm": 0.7026944563592055, "learning_rate": 5.857351492036889e-07, "loss": 0.1812, "step": 24564 }, { "epoch": 1.6982371240926373, "grad_norm": 0.8432708967352197, "learning_rate": 5.854722862400835e-07, "loss": 0.258, "step": 24565 }, { "epoch": 1.6983062564811613, "grad_norm": 0.8854389353888942, "learning_rate": 5.852094786047063e-07, "loss": 0.3125, "step": 24566 }, { "epoch": 1.6983753888696853, "grad_norm": 0.7428721296474303, "learning_rate": 5.84946726300849e-07, "loss": 0.1892, "step": 24567 }, { "epoch": 1.6984445212582093, "grad_norm": 0.8675450296536764, "learning_rate": 5.846840293318063e-07, "loss": 0.2403, "step": 24568 }, { "epoch": 1.6985136536467333, "grad_norm": 0.6536937299125506, "learning_rate": 5.844213877008687e-07, "loss": 0.3293, "step": 24569 }, { "epoch": 1.6985827860352574, "grad_norm": 0.7264552608465412, "learning_rate": 5.841588014113292e-07, "loss": 0.2965, "step": 24570 }, { "epoch": 1.6986519184237814, "grad_norm": 0.7433526552801423, "learning_rate": 5.838962704664786e-07, "loss": 0.1849, "step": 24571 }, { "epoch": 1.6987210508123054, "grad_norm": 0.7428998143004408, "learning_rate": 5.836337948696086e-07, "loss": 0.2521, "step": 24572 }, { "epoch": 1.6987901832008296, "grad_norm": 0.7110964792015757, "learning_rate": 5.83371374624006e-07, "loss": 0.3453, "step": 24573 }, { "epoch": 1.6988593155893537, "grad_norm": 0.829206171698729, "learning_rate": 5.831090097329629e-07, "loss": 0.2229, "step": 24574 }, { "epoch": 1.6989284479778777, "grad_norm": 0.6361141577380843, "learning_rate": 5.82846700199765e-07, "loss": 0.221, "step": 24575 }, { "epoch": 1.6989975803664017, "grad_norm": 0.7142881865469197, "learning_rate": 5.82584446027702e-07, "loss": 0.2805, "step": 24576 }, { "epoch": 1.6990667127549257, "grad_norm": 0.7246269240583774, "learning_rate": 5.823222472200585e-07, "loss": 0.4034, "step": 24577 }, { "epoch": 1.6991358451434497, "grad_norm": 0.6749856764499284, "learning_rate": 5.820601037801221e-07, "loss": 0.2543, "step": 24578 }, { "epoch": 1.6992049775319737, "grad_norm": 0.6922130581110895, "learning_rate": 5.81798015711178e-07, "loss": 0.228, "step": 24579 }, { "epoch": 1.6992741099204978, "grad_norm": 0.66408498464989, "learning_rate": 5.815359830165124e-07, "loss": 0.3036, "step": 24580 }, { "epoch": 1.6993432423090218, "grad_norm": 0.7175445492431098, "learning_rate": 5.812740056994071e-07, "loss": 0.2348, "step": 24581 }, { "epoch": 1.6994123746975458, "grad_norm": 0.7213181867275055, "learning_rate": 5.810120837631472e-07, "loss": 0.2082, "step": 24582 }, { "epoch": 1.6994815070860698, "grad_norm": 0.7137338263067368, "learning_rate": 5.807502172110141e-07, "loss": 0.281, "step": 24583 }, { "epoch": 1.6995506394745938, "grad_norm": 0.7242354743560419, "learning_rate": 5.804884060462906e-07, "loss": 0.3442, "step": 24584 }, { "epoch": 1.6996197718631179, "grad_norm": 0.640961016662028, "learning_rate": 5.802266502722575e-07, "loss": 0.1865, "step": 24585 }, { "epoch": 1.6996889042516419, "grad_norm": 0.6286685396292252, "learning_rate": 5.799649498921972e-07, "loss": 0.1948, "step": 24586 }, { "epoch": 1.699758036640166, "grad_norm": 0.7147701683070887, "learning_rate": 5.797033049093875e-07, "loss": 0.3136, "step": 24587 }, { "epoch": 1.69982716902869, "grad_norm": 0.784961529191815, "learning_rate": 5.794417153271092e-07, "loss": 0.302, "step": 24588 }, { "epoch": 1.699896301417214, "grad_norm": 0.6974007318364998, "learning_rate": 5.791801811486397e-07, "loss": 0.2633, "step": 24589 }, { "epoch": 1.699965433805738, "grad_norm": 0.7692760482395055, "learning_rate": 5.78918702377258e-07, "loss": 0.2756, "step": 24590 }, { "epoch": 1.700034566194262, "grad_norm": 0.6774059079884358, "learning_rate": 5.786572790162392e-07, "loss": 0.3262, "step": 24591 }, { "epoch": 1.700103698582786, "grad_norm": 0.6275374187360673, "learning_rate": 5.783959110688614e-07, "loss": 0.2383, "step": 24592 }, { "epoch": 1.7001728309713102, "grad_norm": 0.7053374124276485, "learning_rate": 5.781345985384007e-07, "loss": 0.2381, "step": 24593 }, { "epoch": 1.7002419633598342, "grad_norm": 0.6974544018274108, "learning_rate": 5.778733414281318e-07, "loss": 0.2618, "step": 24594 }, { "epoch": 1.7003110957483583, "grad_norm": 0.8119152069405808, "learning_rate": 5.776121397413287e-07, "loss": 0.2828, "step": 24595 }, { "epoch": 1.7003802281368823, "grad_norm": 0.7662688641359003, "learning_rate": 5.77350993481266e-07, "loss": 0.2102, "step": 24596 }, { "epoch": 1.7004493605254063, "grad_norm": 0.8934271789834574, "learning_rate": 5.770899026512144e-07, "loss": 0.2823, "step": 24597 }, { "epoch": 1.7005184929139303, "grad_norm": 0.6924665250230555, "learning_rate": 5.768288672544498e-07, "loss": 0.3412, "step": 24598 }, { "epoch": 1.7005876253024543, "grad_norm": 0.7072973785003873, "learning_rate": 5.765678872942393e-07, "loss": 0.2592, "step": 24599 }, { "epoch": 1.7006567576909783, "grad_norm": 0.7137903468099678, "learning_rate": 5.763069627738588e-07, "loss": 0.2239, "step": 24600 }, { "epoch": 1.7007258900795024, "grad_norm": 0.6281008169822427, "learning_rate": 5.760460936965745e-07, "loss": 0.3101, "step": 24601 }, { "epoch": 1.7007950224680264, "grad_norm": 0.7094204239928071, "learning_rate": 5.757852800656588e-07, "loss": 0.3295, "step": 24602 }, { "epoch": 1.7008641548565504, "grad_norm": 0.7222302563865722, "learning_rate": 5.755245218843786e-07, "loss": 0.2453, "step": 24603 }, { "epoch": 1.7009332872450744, "grad_norm": 0.7289728395045637, "learning_rate": 5.752638191560028e-07, "loss": 0.2739, "step": 24604 }, { "epoch": 1.7010024196335984, "grad_norm": 0.6804280334038897, "learning_rate": 5.750031718837984e-07, "loss": 0.3216, "step": 24605 }, { "epoch": 1.7010715520221225, "grad_norm": 0.6938293565360086, "learning_rate": 5.747425800710327e-07, "loss": 0.2504, "step": 24606 }, { "epoch": 1.7011406844106465, "grad_norm": 0.6774362385991446, "learning_rate": 5.744820437209709e-07, "loss": 0.2432, "step": 24607 }, { "epoch": 1.7012098167991705, "grad_norm": 0.7011322064848303, "learning_rate": 5.742215628368802e-07, "loss": 0.2757, "step": 24608 }, { "epoch": 1.7012789491876945, "grad_norm": 0.7451309194562956, "learning_rate": 5.739611374220244e-07, "loss": 0.3842, "step": 24609 }, { "epoch": 1.7013480815762185, "grad_norm": 0.6375608409977718, "learning_rate": 5.737007674796657e-07, "loss": 0.2106, "step": 24610 }, { "epoch": 1.7014172139647425, "grad_norm": 0.6907323583616992, "learning_rate": 5.734404530130699e-07, "loss": 0.2678, "step": 24611 }, { "epoch": 1.7014863463532666, "grad_norm": 0.7246403747859299, "learning_rate": 5.731801940254977e-07, "loss": 0.3318, "step": 24612 }, { "epoch": 1.7015554787417906, "grad_norm": 0.7087771918184478, "learning_rate": 5.729199905202116e-07, "loss": 0.3002, "step": 24613 }, { "epoch": 1.7016246111303146, "grad_norm": 0.7268435113681758, "learning_rate": 5.726598425004725e-07, "loss": 0.2518, "step": 24614 }, { "epoch": 1.7016937435188386, "grad_norm": 0.6898021108399237, "learning_rate": 5.723997499695428e-07, "loss": 0.2731, "step": 24615 }, { "epoch": 1.7017628759073626, "grad_norm": 1.6087515517151005, "learning_rate": 5.721397129306799e-07, "loss": 0.3605, "step": 24616 }, { "epoch": 1.7018320082958867, "grad_norm": 0.9153003809583775, "learning_rate": 5.718797313871449e-07, "loss": 0.2654, "step": 24617 }, { "epoch": 1.7019011406844107, "grad_norm": 0.905885394181514, "learning_rate": 5.71619805342194e-07, "loss": 0.2186, "step": 24618 }, { "epoch": 1.7019702730729347, "grad_norm": 0.6887905263567333, "learning_rate": 5.713599347990856e-07, "loss": 0.3157, "step": 24619 }, { "epoch": 1.7020394054614587, "grad_norm": 0.7980860587956935, "learning_rate": 5.711001197610777e-07, "loss": 0.2923, "step": 24620 }, { "epoch": 1.7021085378499827, "grad_norm": 0.8485231758003723, "learning_rate": 5.708403602314266e-07, "loss": 0.2199, "step": 24621 }, { "epoch": 1.7021776702385067, "grad_norm": 0.6509203273969763, "learning_rate": 5.70580656213387e-07, "loss": 0.2157, "step": 24622 }, { "epoch": 1.7022468026270308, "grad_norm": 0.687419980111205, "learning_rate": 5.70321007710215e-07, "loss": 0.2963, "step": 24623 }, { "epoch": 1.7023159350155548, "grad_norm": 0.759510558275077, "learning_rate": 5.700614147251626e-07, "loss": 0.2239, "step": 24624 }, { "epoch": 1.7023850674040788, "grad_norm": 1.0789920142162843, "learning_rate": 5.698018772614867e-07, "loss": 0.2549, "step": 24625 }, { "epoch": 1.7024541997926028, "grad_norm": 0.7518599296093472, "learning_rate": 5.695423953224361e-07, "loss": 0.3261, "step": 24626 }, { "epoch": 1.7025233321811268, "grad_norm": 0.7945581836622656, "learning_rate": 5.692829689112655e-07, "loss": 0.2596, "step": 24627 }, { "epoch": 1.7025924645696509, "grad_norm": 0.6339648802171194, "learning_rate": 5.690235980312264e-07, "loss": 0.2028, "step": 24628 }, { "epoch": 1.7026615969581749, "grad_norm": 6.591104294272874, "learning_rate": 5.687642826855699e-07, "loss": 0.2007, "step": 24629 }, { "epoch": 1.702730729346699, "grad_norm": 0.6755496932132349, "learning_rate": 5.685050228775441e-07, "loss": 0.3036, "step": 24630 }, { "epoch": 1.702799861735223, "grad_norm": 0.6743770838695905, "learning_rate": 5.682458186104001e-07, "loss": 0.2486, "step": 24631 }, { "epoch": 1.702868994123747, "grad_norm": 0.6480833426915866, "learning_rate": 5.679866698873854e-07, "loss": 0.2497, "step": 24632 }, { "epoch": 1.702938126512271, "grad_norm": 0.6784413015315828, "learning_rate": 5.67727576711749e-07, "loss": 0.3049, "step": 24633 }, { "epoch": 1.703007258900795, "grad_norm": 0.7110098643786013, "learning_rate": 5.674685390867363e-07, "loss": 0.3245, "step": 24634 }, { "epoch": 1.703076391289319, "grad_norm": 0.7644626809853662, "learning_rate": 5.672095570155973e-07, "loss": 0.2316, "step": 24635 }, { "epoch": 1.703145523677843, "grad_norm": 0.6729680038337147, "learning_rate": 5.669506305015743e-07, "loss": 0.2442, "step": 24636 }, { "epoch": 1.703214656066367, "grad_norm": 0.6797795421189244, "learning_rate": 5.666917595479154e-07, "loss": 0.2967, "step": 24637 }, { "epoch": 1.703283788454891, "grad_norm": 0.7063038679056937, "learning_rate": 5.664329441578625e-07, "loss": 0.2695, "step": 24638 }, { "epoch": 1.703352920843415, "grad_norm": 0.648775967469941, "learning_rate": 5.661741843346613e-07, "loss": 0.259, "step": 24639 }, { "epoch": 1.703422053231939, "grad_norm": 0.683534730552894, "learning_rate": 5.659154800815536e-07, "loss": 0.253, "step": 24640 }, { "epoch": 1.703491185620463, "grad_norm": 0.755061174272461, "learning_rate": 5.656568314017824e-07, "loss": 0.3522, "step": 24641 }, { "epoch": 1.703560318008987, "grad_norm": 0.7028745625278984, "learning_rate": 5.653982382985896e-07, "loss": 0.2056, "step": 24642 }, { "epoch": 1.7036294503975111, "grad_norm": 0.6624174851904797, "learning_rate": 5.651397007752163e-07, "loss": 0.2124, "step": 24643 }, { "epoch": 1.7036985827860351, "grad_norm": 0.694544260367082, "learning_rate": 5.648812188349018e-07, "loss": 0.2618, "step": 24644 }, { "epoch": 1.7037677151745592, "grad_norm": 0.7404125400654398, "learning_rate": 5.646227924808878e-07, "loss": 0.2856, "step": 24645 }, { "epoch": 1.7038368475630832, "grad_norm": 0.6463825858957238, "learning_rate": 5.643644217164101e-07, "loss": 0.1788, "step": 24646 }, { "epoch": 1.7039059799516072, "grad_norm": 0.6940961421970671, "learning_rate": 5.641061065447101e-07, "loss": 0.2643, "step": 24647 }, { "epoch": 1.7039751123401312, "grad_norm": 0.6840686366542723, "learning_rate": 5.638478469690229e-07, "loss": 0.3644, "step": 24648 }, { "epoch": 1.7040442447286552, "grad_norm": 0.722721289081532, "learning_rate": 5.635896429925858e-07, "loss": 0.2568, "step": 24649 }, { "epoch": 1.7041133771171793, "grad_norm": 0.7158944038329607, "learning_rate": 5.633314946186358e-07, "loss": 0.2276, "step": 24650 }, { "epoch": 1.7041825095057033, "grad_norm": 0.7754570622267686, "learning_rate": 5.630734018504086e-07, "loss": 0.3603, "step": 24651 }, { "epoch": 1.7042516418942273, "grad_norm": 1.0239122115976902, "learning_rate": 5.628153646911378e-07, "loss": 0.3492, "step": 24652 }, { "epoch": 1.7043207742827513, "grad_norm": 0.7473965462261759, "learning_rate": 5.625573831440584e-07, "loss": 0.2597, "step": 24653 }, { "epoch": 1.7043899066712755, "grad_norm": 0.7043565976756644, "learning_rate": 5.622994572124024e-07, "loss": 0.2418, "step": 24654 }, { "epoch": 1.7044590390597996, "grad_norm": 0.7726255116645763, "learning_rate": 5.620415868994034e-07, "loss": 0.2914, "step": 24655 }, { "epoch": 1.7045281714483236, "grad_norm": 0.7353016976883872, "learning_rate": 5.617837722082931e-07, "loss": 0.2714, "step": 24656 }, { "epoch": 1.7045973038368476, "grad_norm": 0.5945711195894563, "learning_rate": 5.615260131423039e-07, "loss": 0.1589, "step": 24657 }, { "epoch": 1.7046664362253716, "grad_norm": 0.7067393504548921, "learning_rate": 5.612683097046639e-07, "loss": 0.3238, "step": 24658 }, { "epoch": 1.7047355686138956, "grad_norm": 0.7344574215030059, "learning_rate": 5.610106618986061e-07, "loss": 0.3103, "step": 24659 }, { "epoch": 1.7048047010024197, "grad_norm": 0.6627489090212476, "learning_rate": 5.607530697273566e-07, "loss": 0.2619, "step": 24660 }, { "epoch": 1.7048738333909437, "grad_norm": 0.6628905245725396, "learning_rate": 5.604955331941458e-07, "loss": 0.2474, "step": 24661 }, { "epoch": 1.7049429657794677, "grad_norm": 0.8083880452846268, "learning_rate": 5.602380523022005e-07, "loss": 0.334, "step": 24662 }, { "epoch": 1.7050120981679917, "grad_norm": 0.6614997702813418, "learning_rate": 5.599806270547475e-07, "loss": 0.2474, "step": 24663 }, { "epoch": 1.7050812305565157, "grad_norm": 0.7832134322637925, "learning_rate": 5.597232574550143e-07, "loss": 0.2083, "step": 24664 }, { "epoch": 1.7051503629450397, "grad_norm": 0.6430466023966711, "learning_rate": 5.594659435062266e-07, "loss": 0.2239, "step": 24665 }, { "epoch": 1.7052194953335638, "grad_norm": 0.649600401351397, "learning_rate": 5.59208685211608e-07, "loss": 0.3271, "step": 24666 }, { "epoch": 1.7052886277220878, "grad_norm": 0.6962313454436635, "learning_rate": 5.589514825743847e-07, "loss": 0.2609, "step": 24667 }, { "epoch": 1.7053577601106118, "grad_norm": 0.6762559882719192, "learning_rate": 5.586943355977781e-07, "loss": 0.2313, "step": 24668 }, { "epoch": 1.7054268924991358, "grad_norm": 0.6737688426247698, "learning_rate": 5.584372442850127e-07, "loss": 0.3575, "step": 24669 }, { "epoch": 1.7054960248876598, "grad_norm": 0.7294338315432745, "learning_rate": 5.581802086393085e-07, "loss": 0.2884, "step": 24670 }, { "epoch": 1.7055651572761839, "grad_norm": 0.740193692010744, "learning_rate": 5.579232286638908e-07, "loss": 0.2223, "step": 24671 }, { "epoch": 1.7056342896647079, "grad_norm": 0.7433744087985694, "learning_rate": 5.576663043619768e-07, "loss": 0.2916, "step": 24672 }, { "epoch": 1.7057034220532319, "grad_norm": 0.7887544343731527, "learning_rate": 5.574094357367893e-07, "loss": 0.4188, "step": 24673 }, { "epoch": 1.7057725544417561, "grad_norm": 0.7181736395738313, "learning_rate": 5.571526227915452e-07, "loss": 0.2597, "step": 24674 }, { "epoch": 1.7058416868302801, "grad_norm": 0.7602787772350286, "learning_rate": 5.568958655294654e-07, "loss": 0.3151, "step": 24675 }, { "epoch": 1.7059108192188042, "grad_norm": 0.6942407459206393, "learning_rate": 5.566391639537655e-07, "loss": 0.2973, "step": 24676 }, { "epoch": 1.7059799516073282, "grad_norm": 0.7077601882564201, "learning_rate": 5.563825180676647e-07, "loss": 0.3541, "step": 24677 }, { "epoch": 1.7060490839958522, "grad_norm": 0.6822080011895879, "learning_rate": 5.561259278743797e-07, "loss": 0.2338, "step": 24678 }, { "epoch": 1.7061182163843762, "grad_norm": 0.7075751892388258, "learning_rate": 5.558693933771248e-07, "loss": 0.2801, "step": 24679 }, { "epoch": 1.7061873487729002, "grad_norm": 0.6792134894354481, "learning_rate": 5.556129145791172e-07, "loss": 0.3624, "step": 24680 }, { "epoch": 1.7062564811614243, "grad_norm": 0.7685497898766409, "learning_rate": 5.553564914835697e-07, "loss": 0.2807, "step": 24681 }, { "epoch": 1.7063256135499483, "grad_norm": 0.7224890242028978, "learning_rate": 5.551001240936971e-07, "loss": 0.187, "step": 24682 }, { "epoch": 1.7063947459384723, "grad_norm": 0.655132287452765, "learning_rate": 5.548438124127115e-07, "loss": 0.2783, "step": 24683 }, { "epoch": 1.7064638783269963, "grad_norm": 0.8509956185228713, "learning_rate": 5.545875564438258e-07, "loss": 0.4254, "step": 24684 }, { "epoch": 1.7065330107155203, "grad_norm": 0.6946642955568713, "learning_rate": 5.543313561902525e-07, "loss": 0.2707, "step": 24685 }, { "epoch": 1.7066021431040443, "grad_norm": 0.7027366930253492, "learning_rate": 5.540752116552023e-07, "loss": 0.2566, "step": 24686 }, { "epoch": 1.7066712754925684, "grad_norm": 0.7264145583146563, "learning_rate": 5.538191228418844e-07, "loss": 0.3232, "step": 24687 }, { "epoch": 1.7067404078810924, "grad_norm": 0.6476098888905704, "learning_rate": 5.535630897535105e-07, "loss": 0.2737, "step": 24688 }, { "epoch": 1.7068095402696164, "grad_norm": 0.6351398819531561, "learning_rate": 5.533071123932865e-07, "loss": 0.2214, "step": 24689 }, { "epoch": 1.7068786726581404, "grad_norm": 0.6666630219483288, "learning_rate": 5.530511907644231e-07, "loss": 0.2415, "step": 24690 }, { "epoch": 1.7069478050466644, "grad_norm": 0.720476293499972, "learning_rate": 5.527953248701268e-07, "loss": 0.3073, "step": 24691 }, { "epoch": 1.7070169374351885, "grad_norm": 0.7482161600036814, "learning_rate": 5.525395147136059e-07, "loss": 0.226, "step": 24692 }, { "epoch": 1.7070860698237125, "grad_norm": 0.6923847480165364, "learning_rate": 5.522837602980641e-07, "loss": 0.2138, "step": 24693 }, { "epoch": 1.7071552022122365, "grad_norm": 0.7179928541449593, "learning_rate": 5.520280616267093e-07, "loss": 0.272, "step": 24694 }, { "epoch": 1.7072243346007605, "grad_norm": 0.7265166803615766, "learning_rate": 5.51772418702744e-07, "loss": 0.2782, "step": 24695 }, { "epoch": 1.7072934669892845, "grad_norm": 0.7687912388181585, "learning_rate": 5.51516831529374e-07, "loss": 0.26, "step": 24696 }, { "epoch": 1.7073625993778085, "grad_norm": 0.6851494507351156, "learning_rate": 5.51261300109801e-07, "loss": 0.2004, "step": 24697 }, { "epoch": 1.7074317317663326, "grad_norm": 0.6826307939758804, "learning_rate": 5.510058244472283e-07, "loss": 0.3809, "step": 24698 }, { "epoch": 1.7075008641548566, "grad_norm": 1.1580699443429698, "learning_rate": 5.50750404544858e-07, "loss": 0.2292, "step": 24699 }, { "epoch": 1.7075699965433806, "grad_norm": 0.8176652219555104, "learning_rate": 5.504950404058923e-07, "loss": 0.2134, "step": 24700 }, { "epoch": 1.7076391289319046, "grad_norm": 0.6911058561145259, "learning_rate": 5.502397320335301e-07, "loss": 0.3108, "step": 24701 }, { "epoch": 1.7077082613204286, "grad_norm": 0.7352770799261716, "learning_rate": 5.499844794309727e-07, "loss": 0.2642, "step": 24702 }, { "epoch": 1.7077773937089527, "grad_norm": 0.6384141807884716, "learning_rate": 5.497292826014172e-07, "loss": 0.2519, "step": 24703 }, { "epoch": 1.7078465260974767, "grad_norm": 0.6651236162451195, "learning_rate": 5.494741415480648e-07, "loss": 0.2301, "step": 24704 }, { "epoch": 1.7079156584860007, "grad_norm": 0.6979398891937294, "learning_rate": 5.492190562741101e-07, "loss": 0.3508, "step": 24705 }, { "epoch": 1.7079847908745247, "grad_norm": 0.6897451432883113, "learning_rate": 5.489640267827523e-07, "loss": 0.2382, "step": 24706 }, { "epoch": 1.7080539232630487, "grad_norm": 0.5883845811796256, "learning_rate": 5.487090530771871e-07, "loss": 0.1986, "step": 24707 }, { "epoch": 1.7081230556515727, "grad_norm": 0.6701165533650766, "learning_rate": 5.48454135160611e-07, "loss": 0.2367, "step": 24708 }, { "epoch": 1.7081921880400968, "grad_norm": 0.6874349905554479, "learning_rate": 5.481992730362173e-07, "loss": 0.3175, "step": 24709 }, { "epoch": 1.7082613204286208, "grad_norm": 0.7298357251945442, "learning_rate": 5.47944466707202e-07, "loss": 0.2343, "step": 24710 }, { "epoch": 1.7083304528171448, "grad_norm": 0.6700921567585431, "learning_rate": 5.47689716176757e-07, "loss": 0.2251, "step": 24711 }, { "epoch": 1.7083995852056688, "grad_norm": 0.7330316398056153, "learning_rate": 5.474350214480756e-07, "loss": 0.3387, "step": 24712 }, { "epoch": 1.7084687175941928, "grad_norm": 0.6941201400531062, "learning_rate": 5.471803825243505e-07, "loss": 0.3184, "step": 24713 }, { "epoch": 1.7085378499827168, "grad_norm": 0.6920700138484328, "learning_rate": 5.46925799408774e-07, "loss": 0.238, "step": 24714 }, { "epoch": 1.7086069823712409, "grad_norm": 0.7297044283094161, "learning_rate": 5.466712721045342e-07, "loss": 0.2725, "step": 24715 }, { "epoch": 1.7086761147597649, "grad_norm": 0.6856277839071927, "learning_rate": 5.464168006148241e-07, "loss": 0.3593, "step": 24716 }, { "epoch": 1.708745247148289, "grad_norm": 0.7028464858830931, "learning_rate": 5.461623849428304e-07, "loss": 0.2526, "step": 24717 }, { "epoch": 1.708814379536813, "grad_norm": 0.6550614811463937, "learning_rate": 5.459080250917442e-07, "loss": 0.2448, "step": 24718 }, { "epoch": 1.708883511925337, "grad_norm": 1.3682664311934525, "learning_rate": 5.456537210647511e-07, "loss": 0.342, "step": 24719 }, { "epoch": 1.708952644313861, "grad_norm": 0.6597759593899665, "learning_rate": 5.453994728650397e-07, "loss": 0.3197, "step": 24720 }, { "epoch": 1.709021776702385, "grad_norm": 0.6748078083109535, "learning_rate": 5.451452804957962e-07, "loss": 0.2429, "step": 24721 }, { "epoch": 1.709090909090909, "grad_norm": 0.7908502090368349, "learning_rate": 5.448911439602073e-07, "loss": 0.262, "step": 24722 }, { "epoch": 1.709160041479433, "grad_norm": 0.6429550951406722, "learning_rate": 5.446370632614568e-07, "loss": 0.3057, "step": 24723 }, { "epoch": 1.709229173867957, "grad_norm": 0.656358017502532, "learning_rate": 5.443830384027304e-07, "loss": 0.222, "step": 24724 }, { "epoch": 1.709298306256481, "grad_norm": 0.6343090554171756, "learning_rate": 5.441290693872103e-07, "loss": 0.2281, "step": 24725 }, { "epoch": 1.709367438645005, "grad_norm": 0.9745553472345362, "learning_rate": 5.438751562180805e-07, "loss": 0.2707, "step": 24726 }, { "epoch": 1.709436571033529, "grad_norm": 1.1721642692578234, "learning_rate": 5.436212988985235e-07, "loss": 0.3339, "step": 24727 }, { "epoch": 1.709505703422053, "grad_norm": 0.6644167503862554, "learning_rate": 5.433674974317216e-07, "loss": 0.2455, "step": 24728 }, { "epoch": 1.7095748358105771, "grad_norm": 0.6979095382492436, "learning_rate": 5.431137518208541e-07, "loss": 0.189, "step": 24729 }, { "epoch": 1.7096439681991011, "grad_norm": 0.712713094951407, "learning_rate": 5.428600620691027e-07, "loss": 0.3545, "step": 24730 }, { "epoch": 1.7097131005876252, "grad_norm": 0.7493078152646131, "learning_rate": 5.426064281796455e-07, "loss": 0.1981, "step": 24731 }, { "epoch": 1.7097822329761492, "grad_norm": 0.637424775165038, "learning_rate": 5.42352850155663e-07, "loss": 0.2211, "step": 24732 }, { "epoch": 1.7098513653646732, "grad_norm": 0.6533401794698933, "learning_rate": 5.420993280003317e-07, "loss": 0.2253, "step": 24733 }, { "epoch": 1.7099204977531972, "grad_norm": 0.8488593512284587, "learning_rate": 5.418458617168299e-07, "loss": 0.3269, "step": 24734 }, { "epoch": 1.7099896301417215, "grad_norm": 0.8142800515091886, "learning_rate": 5.415924513083337e-07, "loss": 0.2428, "step": 24735 }, { "epoch": 1.7100587625302455, "grad_norm": 0.6662829574412943, "learning_rate": 5.413390967780213e-07, "loss": 0.2021, "step": 24736 }, { "epoch": 1.7101278949187695, "grad_norm": 0.7374913612017711, "learning_rate": 5.410857981290651e-07, "loss": 0.3492, "step": 24737 }, { "epoch": 1.7101970273072935, "grad_norm": 0.698994687165726, "learning_rate": 5.408325553646426e-07, "loss": 0.2422, "step": 24738 }, { "epoch": 1.7102661596958175, "grad_norm": 0.646994304894017, "learning_rate": 5.405793684879251e-07, "loss": 0.2344, "step": 24739 }, { "epoch": 1.7103352920843415, "grad_norm": 0.706279823955734, "learning_rate": 5.403262375020879e-07, "loss": 0.2342, "step": 24740 }, { "epoch": 1.7104044244728656, "grad_norm": 0.6723041281922894, "learning_rate": 5.400731624103006e-07, "loss": 0.3741, "step": 24741 }, { "epoch": 1.7104735568613896, "grad_norm": 0.6635948879296252, "learning_rate": 5.398201432157396e-07, "loss": 0.2435, "step": 24742 }, { "epoch": 1.7105426892499136, "grad_norm": 0.6530231903075056, "learning_rate": 5.39567179921573e-07, "loss": 0.2396, "step": 24743 }, { "epoch": 1.7106118216384376, "grad_norm": 0.7353118048900593, "learning_rate": 5.393142725309708e-07, "loss": 0.3095, "step": 24744 }, { "epoch": 1.7106809540269616, "grad_norm": 0.6573182959460189, "learning_rate": 5.390614210471052e-07, "loss": 0.2793, "step": 24745 }, { "epoch": 1.7107500864154856, "grad_norm": 0.6282751576526755, "learning_rate": 5.388086254731423e-07, "loss": 0.2307, "step": 24746 }, { "epoch": 1.7108192188040097, "grad_norm": 0.6914186750566447, "learning_rate": 5.385558858122525e-07, "loss": 0.2681, "step": 24747 }, { "epoch": 1.7108883511925337, "grad_norm": 0.6884623857538505, "learning_rate": 5.383032020676027e-07, "loss": 0.3153, "step": 24748 }, { "epoch": 1.7109574835810577, "grad_norm": 0.6794324821531681, "learning_rate": 5.380505742423603e-07, "loss": 0.2895, "step": 24749 }, { "epoch": 1.7110266159695817, "grad_norm": 0.6670778976248537, "learning_rate": 5.377980023396911e-07, "loss": 0.2449, "step": 24750 }, { "epoch": 1.7110957483581057, "grad_norm": 0.6870367901972764, "learning_rate": 5.375454863627616e-07, "loss": 0.2738, "step": 24751 }, { "epoch": 1.7111648807466298, "grad_norm": 0.7354106646411389, "learning_rate": 5.372930263147347e-07, "loss": 0.3261, "step": 24752 }, { "epoch": 1.7112340131351538, "grad_norm": 0.6587163241237008, "learning_rate": 5.370406221987762e-07, "loss": 0.1835, "step": 24753 }, { "epoch": 1.7113031455236778, "grad_norm": 0.6627174380892507, "learning_rate": 5.367882740180486e-07, "loss": 0.2921, "step": 24754 }, { "epoch": 1.7113722779122018, "grad_norm": 0.7329360223760083, "learning_rate": 5.365359817757148e-07, "loss": 0.3721, "step": 24755 }, { "epoch": 1.711441410300726, "grad_norm": 0.7700999664400731, "learning_rate": 5.36283745474937e-07, "loss": 0.2549, "step": 24756 }, { "epoch": 1.71151054268925, "grad_norm": 0.6600123856219959, "learning_rate": 5.36031565118878e-07, "loss": 0.2174, "step": 24757 }, { "epoch": 1.711579675077774, "grad_norm": 0.8280287790008873, "learning_rate": 5.357794407106959e-07, "loss": 0.2135, "step": 24758 }, { "epoch": 1.711648807466298, "grad_norm": 0.704113175489499, "learning_rate": 5.355273722535525e-07, "loss": 0.3485, "step": 24759 }, { "epoch": 1.7117179398548221, "grad_norm": 0.740305278813512, "learning_rate": 5.35275359750605e-07, "loss": 0.3348, "step": 24760 }, { "epoch": 1.7117870722433461, "grad_norm": 0.6569176142782893, "learning_rate": 5.350234032050139e-07, "loss": 0.2467, "step": 24761 }, { "epoch": 1.7118562046318702, "grad_norm": 0.6888590183402107, "learning_rate": 5.347715026199357e-07, "loss": 0.2824, "step": 24762 }, { "epoch": 1.7119253370203942, "grad_norm": 0.693615121090933, "learning_rate": 5.345196579985295e-07, "loss": 0.2646, "step": 24763 }, { "epoch": 1.7119944694089182, "grad_norm": 0.687782439917353, "learning_rate": 5.342678693439496e-07, "loss": 0.2318, "step": 24764 }, { "epoch": 1.7120636017974422, "grad_norm": 0.6889010440930713, "learning_rate": 5.340161366593533e-07, "loss": 0.2809, "step": 24765 }, { "epoch": 1.7121327341859662, "grad_norm": 0.7405622310173929, "learning_rate": 5.337644599478936e-07, "loss": 0.3369, "step": 24766 }, { "epoch": 1.7122018665744902, "grad_norm": 0.8494422432307648, "learning_rate": 5.33512839212727e-07, "loss": 0.2468, "step": 24767 }, { "epoch": 1.7122709989630143, "grad_norm": 0.561663454237393, "learning_rate": 5.332612744570054e-07, "loss": 0.1369, "step": 24768 }, { "epoch": 1.7123401313515383, "grad_norm": 1.1236335175835759, "learning_rate": 5.330097656838823e-07, "loss": 0.2615, "step": 24769 }, { "epoch": 1.7124092637400623, "grad_norm": 0.6762016545402467, "learning_rate": 5.327583128965102e-07, "loss": 0.2917, "step": 24770 }, { "epoch": 1.7124783961285863, "grad_norm": 0.7081325808331584, "learning_rate": 5.325069160980417e-07, "loss": 0.1932, "step": 24771 }, { "epoch": 1.7125475285171103, "grad_norm": 0.7110829849071699, "learning_rate": 5.322555752916253e-07, "loss": 0.2401, "step": 24772 }, { "epoch": 1.7126166609056344, "grad_norm": 1.1660120224503991, "learning_rate": 5.320042904804129e-07, "loss": 0.3712, "step": 24773 }, { "epoch": 1.7126857932941584, "grad_norm": 0.743509714075509, "learning_rate": 5.317530616675526e-07, "loss": 0.2709, "step": 24774 }, { "epoch": 1.7127549256826824, "grad_norm": 0.6917382354609878, "learning_rate": 5.315018888561945e-07, "loss": 0.2001, "step": 24775 }, { "epoch": 1.7128240580712064, "grad_norm": 0.7048077188445141, "learning_rate": 5.312507720494848e-07, "loss": 0.278, "step": 24776 }, { "epoch": 1.7128931904597304, "grad_norm": 0.6739393948922876, "learning_rate": 5.309997112505721e-07, "loss": 0.315, "step": 24777 }, { "epoch": 1.7129623228482544, "grad_norm": 0.7406073009305736, "learning_rate": 5.307487064626027e-07, "loss": 0.2991, "step": 24778 }, { "epoch": 1.7130314552367785, "grad_norm": 0.7170189637281936, "learning_rate": 5.304977576887233e-07, "loss": 0.2743, "step": 24779 }, { "epoch": 1.7131005876253025, "grad_norm": 0.7910762303280396, "learning_rate": 5.302468649320775e-07, "loss": 0.2992, "step": 24780 }, { "epoch": 1.7131697200138265, "grad_norm": 0.6846600428478217, "learning_rate": 5.299960281958117e-07, "loss": 0.2913, "step": 24781 }, { "epoch": 1.7132388524023505, "grad_norm": 0.9023443605011697, "learning_rate": 5.297452474830672e-07, "loss": 0.2213, "step": 24782 }, { "epoch": 1.7133079847908745, "grad_norm": 0.6987991190959432, "learning_rate": 5.294945227969889e-07, "loss": 0.2624, "step": 24783 }, { "epoch": 1.7133771171793986, "grad_norm": 0.699007505046189, "learning_rate": 5.292438541407185e-07, "loss": 0.3312, "step": 24784 }, { "epoch": 1.7134462495679226, "grad_norm": 0.7267006239558671, "learning_rate": 5.289932415173993e-07, "loss": 0.2107, "step": 24785 }, { "epoch": 1.7135153819564466, "grad_norm": 0.7088197841085194, "learning_rate": 5.287426849301697e-07, "loss": 0.2454, "step": 24786 }, { "epoch": 1.7135845143449706, "grad_norm": 0.7101181167280791, "learning_rate": 5.284921843821722e-07, "loss": 0.3182, "step": 24787 }, { "epoch": 1.7136536467334946, "grad_norm": 0.7035728892854706, "learning_rate": 5.282417398765444e-07, "loss": 0.2535, "step": 24788 }, { "epoch": 1.7137227791220186, "grad_norm": 0.8820591054526747, "learning_rate": 5.27991351416427e-07, "loss": 0.2048, "step": 24789 }, { "epoch": 1.7137919115105427, "grad_norm": 0.80701672043456, "learning_rate": 5.277410190049565e-07, "loss": 0.2264, "step": 24790 }, { "epoch": 1.7138610438990667, "grad_norm": 0.7104066717584215, "learning_rate": 5.274907426452713e-07, "loss": 0.3299, "step": 24791 }, { "epoch": 1.7139301762875907, "grad_norm": 0.7266042822773245, "learning_rate": 5.272405223405086e-07, "loss": 0.2573, "step": 24792 }, { "epoch": 1.7139993086761147, "grad_norm": 0.6457823251579476, "learning_rate": 5.269903580938041e-07, "loss": 0.2483, "step": 24793 }, { "epoch": 1.7140684410646387, "grad_norm": 0.7088827875916649, "learning_rate": 5.267402499082925e-07, "loss": 0.2687, "step": 24794 }, { "epoch": 1.7141375734531628, "grad_norm": 0.6914541075886876, "learning_rate": 5.264901977871101e-07, "loss": 0.2814, "step": 24795 }, { "epoch": 1.7142067058416868, "grad_norm": 0.665104197416446, "learning_rate": 5.262402017333884e-07, "loss": 0.2037, "step": 24796 }, { "epoch": 1.7142758382302108, "grad_norm": 0.7301673924954253, "learning_rate": 5.259902617502621e-07, "loss": 0.2525, "step": 24797 }, { "epoch": 1.7143449706187348, "grad_norm": 0.7271983084009492, "learning_rate": 5.257403778408632e-07, "loss": 0.3954, "step": 24798 }, { "epoch": 1.7144141030072588, "grad_norm": 0.7192014017035565, "learning_rate": 5.254905500083257e-07, "loss": 0.2364, "step": 24799 }, { "epoch": 1.7144832353957828, "grad_norm": 0.8388444337329181, "learning_rate": 5.252407782557778e-07, "loss": 0.2227, "step": 24800 }, { "epoch": 1.7145523677843069, "grad_norm": 0.7984249676661181, "learning_rate": 5.249910625863519e-07, "loss": 0.2192, "step": 24801 }, { "epoch": 1.7146215001728309, "grad_norm": 0.7008532591404003, "learning_rate": 5.247414030031761e-07, "loss": 0.3174, "step": 24802 }, { "epoch": 1.714690632561355, "grad_norm": 0.7253117273660166, "learning_rate": 5.244917995093818e-07, "loss": 0.2514, "step": 24803 }, { "epoch": 1.714759764949879, "grad_norm": 0.7064687698785747, "learning_rate": 5.242422521080942e-07, "loss": 0.302, "step": 24804 }, { "epoch": 1.714828897338403, "grad_norm": 0.706481295543912, "learning_rate": 5.239927608024431e-07, "loss": 0.3194, "step": 24805 }, { "epoch": 1.714898029726927, "grad_norm": 0.742244527074974, "learning_rate": 5.237433255955549e-07, "loss": 0.2648, "step": 24806 }, { "epoch": 1.714967162115451, "grad_norm": 0.6633374136074736, "learning_rate": 5.234939464905569e-07, "loss": 0.2012, "step": 24807 }, { "epoch": 1.715036294503975, "grad_norm": 0.6993954519094578, "learning_rate": 5.232446234905725e-07, "loss": 0.3012, "step": 24808 }, { "epoch": 1.715105426892499, "grad_norm": 0.719724646021802, "learning_rate": 5.229953565987289e-07, "loss": 0.3393, "step": 24809 }, { "epoch": 1.715174559281023, "grad_norm": 0.7178242635067317, "learning_rate": 5.227461458181488e-07, "loss": 0.2615, "step": 24810 }, { "epoch": 1.715243691669547, "grad_norm": 0.7191042009960859, "learning_rate": 5.224969911519545e-07, "loss": 0.2696, "step": 24811 }, { "epoch": 1.715312824058071, "grad_norm": 0.6953657203658437, "learning_rate": 5.222478926032698e-07, "loss": 0.3265, "step": 24812 }, { "epoch": 1.715381956446595, "grad_norm": 0.7779236505391521, "learning_rate": 5.219988501752171e-07, "loss": 0.3235, "step": 24813 }, { "epoch": 1.715451088835119, "grad_norm": 0.6638709974786808, "learning_rate": 5.217498638709184e-07, "loss": 0.2596, "step": 24814 }, { "epoch": 1.7155202212236431, "grad_norm": 0.7414170389434196, "learning_rate": 5.215009336934929e-07, "loss": 0.2863, "step": 24815 }, { "epoch": 1.7155893536121671, "grad_norm": 0.7023917855051097, "learning_rate": 5.212520596460613e-07, "loss": 0.4016, "step": 24816 }, { "epoch": 1.7156584860006914, "grad_norm": 0.7078159647549024, "learning_rate": 5.210032417317418e-07, "loss": 0.2688, "step": 24817 }, { "epoch": 1.7157276183892154, "grad_norm": 0.7393639927206135, "learning_rate": 5.207544799536534e-07, "loss": 0.2109, "step": 24818 }, { "epoch": 1.7157967507777394, "grad_norm": 0.7136614486182502, "learning_rate": 5.205057743149138e-07, "loss": 0.3059, "step": 24819 }, { "epoch": 1.7158658831662634, "grad_norm": 0.6909596856336396, "learning_rate": 5.202571248186417e-07, "loss": 0.2934, "step": 24820 }, { "epoch": 1.7159350155547874, "grad_norm": 0.7522098360221608, "learning_rate": 5.200085314679509e-07, "loss": 0.2073, "step": 24821 }, { "epoch": 1.7160041479433115, "grad_norm": 0.724587742275033, "learning_rate": 5.197599942659593e-07, "loss": 0.286, "step": 24822 }, { "epoch": 1.7160732803318355, "grad_norm": 0.697587165491003, "learning_rate": 5.195115132157796e-07, "loss": 0.3352, "step": 24823 }, { "epoch": 1.7161424127203595, "grad_norm": 0.7623895548139219, "learning_rate": 5.192630883205285e-07, "loss": 0.2585, "step": 24824 }, { "epoch": 1.7162115451088835, "grad_norm": 0.6211377749473993, "learning_rate": 5.190147195833172e-07, "loss": 0.2041, "step": 24825 }, { "epoch": 1.7162806774974075, "grad_norm": 0.6731014306730755, "learning_rate": 5.187664070072602e-07, "loss": 0.3056, "step": 24826 }, { "epoch": 1.7163498098859316, "grad_norm": 0.8452371266292681, "learning_rate": 5.185181505954684e-07, "loss": 0.3007, "step": 24827 }, { "epoch": 1.7164189422744556, "grad_norm": 0.6908370779266926, "learning_rate": 5.182699503510552e-07, "loss": 0.2165, "step": 24828 }, { "epoch": 1.7164880746629796, "grad_norm": 0.7249350235802958, "learning_rate": 5.180218062771298e-07, "loss": 0.224, "step": 24829 }, { "epoch": 1.7165572070515036, "grad_norm": 0.7162423846071011, "learning_rate": 5.177737183768028e-07, "loss": 0.335, "step": 24830 }, { "epoch": 1.7166263394400276, "grad_norm": 0.7522476493582888, "learning_rate": 5.17525686653183e-07, "loss": 0.3476, "step": 24831 }, { "epoch": 1.7166954718285516, "grad_norm": 0.7183033569513152, "learning_rate": 5.172777111093791e-07, "loss": 0.241, "step": 24832 }, { "epoch": 1.7167646042170757, "grad_norm": 0.6994485163053528, "learning_rate": 5.170297917484996e-07, "loss": 0.2989, "step": 24833 }, { "epoch": 1.7168337366055997, "grad_norm": 0.659912910996325, "learning_rate": 5.16781928573652e-07, "loss": 0.3342, "step": 24834 }, { "epoch": 1.7169028689941237, "grad_norm": 0.7969802587624972, "learning_rate": 5.165341215879416e-07, "loss": 0.2671, "step": 24835 }, { "epoch": 1.7169720013826477, "grad_norm": 0.7696450277283368, "learning_rate": 5.162863707944754e-07, "loss": 0.204, "step": 24836 }, { "epoch": 1.717041133771172, "grad_norm": 0.6864017165183542, "learning_rate": 5.160386761963576e-07, "loss": 0.3052, "step": 24837 }, { "epoch": 1.717110266159696, "grad_norm": 0.7891132500570254, "learning_rate": 5.157910377966935e-07, "loss": 0.3266, "step": 24838 }, { "epoch": 1.71717939854822, "grad_norm": 0.6701455695600271, "learning_rate": 5.155434555985855e-07, "loss": 0.2415, "step": 24839 }, { "epoch": 1.717248530936744, "grad_norm": 0.7097538542602322, "learning_rate": 5.152959296051374e-07, "loss": 0.2212, "step": 24840 }, { "epoch": 1.717317663325268, "grad_norm": 0.6930569774809027, "learning_rate": 5.150484598194516e-07, "loss": 0.322, "step": 24841 }, { "epoch": 1.717386795713792, "grad_norm": 0.6020600134462115, "learning_rate": 5.148010462446306e-07, "loss": 0.2507, "step": 24842 }, { "epoch": 1.717455928102316, "grad_norm": 0.7087557517277232, "learning_rate": 5.145536888837733e-07, "loss": 0.2329, "step": 24843 }, { "epoch": 1.71752506049084, "grad_norm": 0.65906618136604, "learning_rate": 5.143063877399817e-07, "loss": 0.3087, "step": 24844 }, { "epoch": 1.717594192879364, "grad_norm": 0.7534548042072357, "learning_rate": 5.140591428163538e-07, "loss": 0.3204, "step": 24845 }, { "epoch": 1.7176633252678881, "grad_norm": 0.7415830532120946, "learning_rate": 5.138119541159897e-07, "loss": 0.3068, "step": 24846 }, { "epoch": 1.7177324576564121, "grad_norm": 0.7714458072329884, "learning_rate": 5.135648216419859e-07, "loss": 0.2882, "step": 24847 }, { "epoch": 1.7178015900449362, "grad_norm": 0.6602125732127921, "learning_rate": 5.133177453974403e-07, "loss": 0.3181, "step": 24848 }, { "epoch": 1.7178707224334602, "grad_norm": 0.6801524998401217, "learning_rate": 5.130707253854505e-07, "loss": 0.2888, "step": 24849 }, { "epoch": 1.7179398548219842, "grad_norm": 0.7416941389935066, "learning_rate": 5.128237616091125e-07, "loss": 0.2374, "step": 24850 }, { "epoch": 1.7180089872105082, "grad_norm": 0.7271532231408633, "learning_rate": 5.125768540715204e-07, "loss": 0.2703, "step": 24851 }, { "epoch": 1.7180781195990322, "grad_norm": 0.7407593428472933, "learning_rate": 5.123300027757694e-07, "loss": 0.364, "step": 24852 }, { "epoch": 1.7181472519875562, "grad_norm": 0.8367877793933541, "learning_rate": 5.120832077249533e-07, "loss": 0.2705, "step": 24853 }, { "epoch": 1.7182163843760803, "grad_norm": 0.640283911761981, "learning_rate": 5.118364689221644e-07, "loss": 0.1939, "step": 24854 }, { "epoch": 1.7182855167646043, "grad_norm": 0.720988260668358, "learning_rate": 5.11589786370496e-07, "loss": 0.3343, "step": 24855 }, { "epoch": 1.7183546491531283, "grad_norm": 0.753434139989585, "learning_rate": 5.113431600730407e-07, "loss": 0.222, "step": 24856 }, { "epoch": 1.7184237815416523, "grad_norm": 0.6598008019556608, "learning_rate": 5.110965900328874e-07, "loss": 0.2095, "step": 24857 }, { "epoch": 1.7184929139301763, "grad_norm": 0.6642190139097608, "learning_rate": 5.108500762531288e-07, "loss": 0.2722, "step": 24858 }, { "epoch": 1.7185620463187004, "grad_norm": 0.7034072407177564, "learning_rate": 5.106036187368524e-07, "loss": 0.327, "step": 24859 }, { "epoch": 1.7186311787072244, "grad_norm": 0.6596951778694419, "learning_rate": 5.103572174871485e-07, "loss": 0.2257, "step": 24860 }, { "epoch": 1.7187003110957484, "grad_norm": 0.6645308378545082, "learning_rate": 5.101108725071042e-07, "loss": 0.2369, "step": 24861 }, { "epoch": 1.7187694434842724, "grad_norm": 1.1619305588775084, "learning_rate": 5.098645837998073e-07, "loss": 0.304, "step": 24862 }, { "epoch": 1.7188385758727964, "grad_norm": 0.7919608674408112, "learning_rate": 5.096183513683445e-07, "loss": 0.3175, "step": 24863 }, { "epoch": 1.7189077082613204, "grad_norm": 0.805406042367225, "learning_rate": 5.093721752158037e-07, "loss": 0.2587, "step": 24864 }, { "epoch": 1.7189768406498445, "grad_norm": 0.6844029700319118, "learning_rate": 5.091260553452676e-07, "loss": 0.2387, "step": 24865 }, { "epoch": 1.7190459730383685, "grad_norm": 0.704195321019833, "learning_rate": 5.08879991759823e-07, "loss": 0.3538, "step": 24866 }, { "epoch": 1.7191151054268925, "grad_norm": 0.7069095368062625, "learning_rate": 5.086339844625521e-07, "loss": 0.2373, "step": 24867 }, { "epoch": 1.7191842378154165, "grad_norm": 0.6613763116245294, "learning_rate": 5.083880334565389e-07, "loss": 0.2098, "step": 24868 }, { "epoch": 1.7192533702039405, "grad_norm": 0.7095778835774837, "learning_rate": 5.081421387448665e-07, "loss": 0.3198, "step": 24869 }, { "epoch": 1.7193225025924646, "grad_norm": 0.7687385211343714, "learning_rate": 5.078963003306164e-07, "loss": 0.257, "step": 24870 }, { "epoch": 1.7193916349809886, "grad_norm": 0.7119873072477554, "learning_rate": 5.076505182168695e-07, "loss": 0.2219, "step": 24871 }, { "epoch": 1.7194607673695126, "grad_norm": 0.7650664718878407, "learning_rate": 5.074047924067071e-07, "loss": 0.2913, "step": 24872 }, { "epoch": 1.7195298997580366, "grad_norm": 0.725924038060654, "learning_rate": 5.071591229032069e-07, "loss": 0.3548, "step": 24873 }, { "epoch": 1.7195990321465606, "grad_norm": 0.8370463805501891, "learning_rate": 5.069135097094507e-07, "loss": 0.2386, "step": 24874 }, { "epoch": 1.7196681645350846, "grad_norm": 0.651628566039747, "learning_rate": 5.066679528285145e-07, "loss": 0.2173, "step": 24875 }, { "epoch": 1.7197372969236087, "grad_norm": 0.7603634399688403, "learning_rate": 5.064224522634764e-07, "loss": 0.2783, "step": 24876 }, { "epoch": 1.7198064293121327, "grad_norm": 0.7833180510856076, "learning_rate": 5.061770080174149e-07, "loss": 0.2959, "step": 24877 }, { "epoch": 1.7198755617006567, "grad_norm": 0.7079064831006283, "learning_rate": 5.059316200934039e-07, "loss": 0.2117, "step": 24878 }, { "epoch": 1.7199446940891807, "grad_norm": 0.6993376512965418, "learning_rate": 5.056862884945212e-07, "loss": 0.2341, "step": 24879 }, { "epoch": 1.7200138264777047, "grad_norm": 0.6677825900933604, "learning_rate": 5.054410132238391e-07, "loss": 0.3382, "step": 24880 }, { "epoch": 1.7200829588662288, "grad_norm": 0.7547057856897089, "learning_rate": 5.051957942844343e-07, "loss": 0.2478, "step": 24881 }, { "epoch": 1.7201520912547528, "grad_norm": 0.6545897956321256, "learning_rate": 5.049506316793779e-07, "loss": 0.2595, "step": 24882 }, { "epoch": 1.7202212236432768, "grad_norm": 0.690451525758343, "learning_rate": 5.047055254117433e-07, "loss": 0.285, "step": 24883 }, { "epoch": 1.7202903560318008, "grad_norm": 0.7178824834392346, "learning_rate": 5.044604754846033e-07, "loss": 0.2892, "step": 24884 }, { "epoch": 1.7203594884203248, "grad_norm": 0.6674261214005146, "learning_rate": 5.042154819010292e-07, "loss": 0.2214, "step": 24885 }, { "epoch": 1.7204286208088488, "grad_norm": 0.6686315599337395, "learning_rate": 5.039705446640897e-07, "loss": 0.2328, "step": 24886 }, { "epoch": 1.7204977531973729, "grad_norm": 0.7182341917955102, "learning_rate": 5.037256637768573e-07, "loss": 0.3595, "step": 24887 }, { "epoch": 1.7205668855858969, "grad_norm": 0.7543580547595771, "learning_rate": 5.034808392423985e-07, "loss": 0.2648, "step": 24888 }, { "epoch": 1.720636017974421, "grad_norm": 0.6513634014688502, "learning_rate": 5.032360710637829e-07, "loss": 0.2361, "step": 24889 }, { "epoch": 1.720705150362945, "grad_norm": 0.822357045533863, "learning_rate": 5.029913592440782e-07, "loss": 0.2832, "step": 24890 }, { "epoch": 1.720774282751469, "grad_norm": 0.6877950864323307, "learning_rate": 5.027467037863532e-07, "loss": 0.3392, "step": 24891 }, { "epoch": 1.720843415139993, "grad_norm": 0.6625186360421801, "learning_rate": 5.025021046936712e-07, "loss": 0.2425, "step": 24892 }, { "epoch": 1.720912547528517, "grad_norm": 0.6842019546970547, "learning_rate": 5.022575619691001e-07, "loss": 0.2413, "step": 24893 }, { "epoch": 1.720981679917041, "grad_norm": 0.6961328042997734, "learning_rate": 5.020130756157027e-07, "loss": 0.3238, "step": 24894 }, { "epoch": 1.721050812305565, "grad_norm": 0.7082243537317963, "learning_rate": 5.017686456365456e-07, "loss": 0.3784, "step": 24895 }, { "epoch": 1.721119944694089, "grad_norm": 0.6590876802287041, "learning_rate": 5.015242720346903e-07, "loss": 0.2595, "step": 24896 }, { "epoch": 1.721189077082613, "grad_norm": 0.6785938582428548, "learning_rate": 5.012799548131997e-07, "loss": 0.2971, "step": 24897 }, { "epoch": 1.7212582094711373, "grad_norm": 0.7273499561836502, "learning_rate": 5.010356939751371e-07, "loss": 0.353, "step": 24898 }, { "epoch": 1.7213273418596613, "grad_norm": 0.8281237393703361, "learning_rate": 5.007914895235644e-07, "loss": 0.1944, "step": 24899 }, { "epoch": 1.7213964742481853, "grad_norm": 0.7153108110131314, "learning_rate": 5.005473414615397e-07, "loss": 0.2717, "step": 24900 }, { "epoch": 1.7214656066367093, "grad_norm": 0.8173274274776187, "learning_rate": 5.003032497921257e-07, "loss": 0.3091, "step": 24901 }, { "epoch": 1.7215347390252334, "grad_norm": 0.6645879130164338, "learning_rate": 5.000592145183791e-07, "loss": 0.2914, "step": 24902 }, { "epoch": 1.7216038714137574, "grad_norm": 0.6510830860054607, "learning_rate": 4.9981523564336e-07, "loss": 0.2548, "step": 24903 }, { "epoch": 1.7216730038022814, "grad_norm": 0.7021953657159854, "learning_rate": 4.995713131701263e-07, "loss": 0.2395, "step": 24904 }, { "epoch": 1.7217421361908054, "grad_norm": 0.6930773776849187, "learning_rate": 4.993274471017356e-07, "loss": 0.3303, "step": 24905 }, { "epoch": 1.7218112685793294, "grad_norm": 0.7556143820802856, "learning_rate": 4.990836374412422e-07, "loss": 0.2926, "step": 24906 }, { "epoch": 1.7218804009678534, "grad_norm": 0.7314025470365974, "learning_rate": 4.988398841917047e-07, "loss": 0.2304, "step": 24907 }, { "epoch": 1.7219495333563775, "grad_norm": 0.9013727402512356, "learning_rate": 4.985961873561751e-07, "loss": 0.272, "step": 24908 }, { "epoch": 1.7220186657449015, "grad_norm": 0.6689734931223182, "learning_rate": 4.983525469377105e-07, "loss": 0.3005, "step": 24909 }, { "epoch": 1.7220877981334255, "grad_norm": 1.3421351852907937, "learning_rate": 4.98108962939362e-07, "loss": 0.2161, "step": 24910 }, { "epoch": 1.7221569305219495, "grad_norm": 0.6916929699137017, "learning_rate": 4.978654353641837e-07, "loss": 0.201, "step": 24911 }, { "epoch": 1.7222260629104735, "grad_norm": 0.7723216161437968, "learning_rate": 4.976219642152274e-07, "loss": 0.3541, "step": 24912 }, { "epoch": 1.7222951952989975, "grad_norm": 0.7255172118446185, "learning_rate": 4.973785494955464e-07, "loss": 0.2658, "step": 24913 }, { "epoch": 1.7223643276875216, "grad_norm": 0.6620425346364918, "learning_rate": 4.97135191208189e-07, "loss": 0.2103, "step": 24914 }, { "epoch": 1.7224334600760456, "grad_norm": 0.681703994719337, "learning_rate": 4.968918893562069e-07, "loss": 0.2554, "step": 24915 }, { "epoch": 1.7225025924645696, "grad_norm": 0.6688251758967081, "learning_rate": 4.966486439426482e-07, "loss": 0.3983, "step": 24916 }, { "epoch": 1.7225717248530936, "grad_norm": 0.7687538245250085, "learning_rate": 4.964054549705633e-07, "loss": 0.2706, "step": 24917 }, { "epoch": 1.7226408572416179, "grad_norm": 0.760084337612965, "learning_rate": 4.961623224429973e-07, "loss": 0.2429, "step": 24918 }, { "epoch": 1.7227099896301419, "grad_norm": 0.7689800971057201, "learning_rate": 4.959192463629997e-07, "loss": 0.3362, "step": 24919 }, { "epoch": 1.722779122018666, "grad_norm": 0.7989864758414074, "learning_rate": 4.956762267336162e-07, "loss": 0.2537, "step": 24920 }, { "epoch": 1.72284825440719, "grad_norm": 0.6468175458407892, "learning_rate": 4.95433263557894e-07, "loss": 0.2338, "step": 24921 }, { "epoch": 1.722917386795714, "grad_norm": 0.6724866356291445, "learning_rate": 4.95190356838876e-07, "loss": 0.2583, "step": 24922 }, { "epoch": 1.722986519184238, "grad_norm": 0.7386067635721322, "learning_rate": 4.949475065796083e-07, "loss": 0.3079, "step": 24923 }, { "epoch": 1.723055651572762, "grad_norm": 0.8009122283236526, "learning_rate": 4.947047127831339e-07, "loss": 0.2467, "step": 24924 }, { "epoch": 1.723124783961286, "grad_norm": 0.6769990353282787, "learning_rate": 4.944619754524949e-07, "loss": 0.2246, "step": 24925 }, { "epoch": 1.72319391634981, "grad_norm": 0.950514224319426, "learning_rate": 4.942192945907354e-07, "loss": 0.3089, "step": 24926 }, { "epoch": 1.723263048738334, "grad_norm": 0.7255867342321347, "learning_rate": 4.939766702008963e-07, "loss": 0.3138, "step": 24927 }, { "epoch": 1.723332181126858, "grad_norm": 0.6506488174045697, "learning_rate": 4.937341022860181e-07, "loss": 0.2115, "step": 24928 }, { "epoch": 1.723401313515382, "grad_norm": 0.692045202776894, "learning_rate": 4.934915908491416e-07, "loss": 0.2829, "step": 24929 }, { "epoch": 1.723470445903906, "grad_norm": 0.7000616371227443, "learning_rate": 4.93249135893305e-07, "loss": 0.3685, "step": 24930 }, { "epoch": 1.72353957829243, "grad_norm": 0.7273214105209398, "learning_rate": 4.930067374215486e-07, "loss": 0.2709, "step": 24931 }, { "epoch": 1.7236087106809541, "grad_norm": 0.7167044942510974, "learning_rate": 4.927643954369088e-07, "loss": 0.2796, "step": 24932 }, { "epoch": 1.7236778430694781, "grad_norm": 0.7269258096366367, "learning_rate": 4.925221099424238e-07, "loss": 0.2348, "step": 24933 }, { "epoch": 1.7237469754580022, "grad_norm": 0.6858519467421452, "learning_rate": 4.922798809411299e-07, "loss": 0.2774, "step": 24934 }, { "epoch": 1.7238161078465262, "grad_norm": 1.4570865481042397, "learning_rate": 4.920377084360644e-07, "loss": 0.2507, "step": 24935 }, { "epoch": 1.7238852402350502, "grad_norm": 0.6272882859837079, "learning_rate": 4.917955924302609e-07, "loss": 0.2266, "step": 24936 }, { "epoch": 1.7239543726235742, "grad_norm": 0.7360963790071093, "learning_rate": 4.915535329267551e-07, "loss": 0.2856, "step": 24937 }, { "epoch": 1.7240235050120982, "grad_norm": 0.6995862975546792, "learning_rate": 4.913115299285787e-07, "loss": 0.2803, "step": 24938 }, { "epoch": 1.7240926374006222, "grad_norm": 0.6645463182762549, "learning_rate": 4.910695834387669e-07, "loss": 0.2031, "step": 24939 }, { "epoch": 1.7241617697891463, "grad_norm": 0.6802348559225999, "learning_rate": 4.908276934603507e-07, "loss": 0.1992, "step": 24940 }, { "epoch": 1.7242309021776703, "grad_norm": 0.7106131460715848, "learning_rate": 4.905858599963637e-07, "loss": 0.3499, "step": 24941 }, { "epoch": 1.7243000345661943, "grad_norm": 0.7861117328629905, "learning_rate": 4.903440830498352e-07, "loss": 0.2332, "step": 24942 }, { "epoch": 1.7243691669547183, "grad_norm": 0.6825949616184643, "learning_rate": 4.901023626237949e-07, "loss": 0.2081, "step": 24943 }, { "epoch": 1.7244382993432423, "grad_norm": 0.7270822177879703, "learning_rate": 4.898606987212745e-07, "loss": 0.3698, "step": 24944 }, { "epoch": 1.7245074317317663, "grad_norm": 0.7474153973387564, "learning_rate": 4.896190913453002e-07, "loss": 0.28, "step": 24945 }, { "epoch": 1.7245765641202904, "grad_norm": 0.7849080819831808, "learning_rate": 4.893775404989021e-07, "loss": 0.2598, "step": 24946 }, { "epoch": 1.7246456965088144, "grad_norm": 0.7186220142567389, "learning_rate": 4.891360461851063e-07, "loss": 0.2411, "step": 24947 }, { "epoch": 1.7247148288973384, "grad_norm": 0.7634511030014827, "learning_rate": 4.888946084069412e-07, "loss": 0.365, "step": 24948 }, { "epoch": 1.7247839612858624, "grad_norm": 0.6756016712002874, "learning_rate": 4.88653227167431e-07, "loss": 0.2701, "step": 24949 }, { "epoch": 1.7248530936743864, "grad_norm": 2.451808412647786, "learning_rate": 4.884119024696027e-07, "loss": 0.2381, "step": 24950 }, { "epoch": 1.7249222260629105, "grad_norm": 0.7303577930985455, "learning_rate": 4.881706343164789e-07, "loss": 0.3061, "step": 24951 }, { "epoch": 1.7249913584514345, "grad_norm": 0.754262599906906, "learning_rate": 4.879294227110853e-07, "loss": 0.2736, "step": 24952 }, { "epoch": 1.7250604908399585, "grad_norm": 0.6953628730868162, "learning_rate": 4.876882676564431e-07, "loss": 0.2391, "step": 24953 }, { "epoch": 1.7251296232284825, "grad_norm": 0.6518318046651032, "learning_rate": 4.874471691555765e-07, "loss": 0.233, "step": 24954 }, { "epoch": 1.7251987556170065, "grad_norm": 0.6834121018325857, "learning_rate": 4.872061272115064e-07, "loss": 0.308, "step": 24955 }, { "epoch": 1.7252678880055305, "grad_norm": 1.052790682205839, "learning_rate": 4.869651418272548e-07, "loss": 0.3023, "step": 24956 }, { "epoch": 1.7253370203940546, "grad_norm": 0.716776881420896, "learning_rate": 4.867242130058403e-07, "loss": 0.2848, "step": 24957 }, { "epoch": 1.7254061527825786, "grad_norm": 0.8151717574493826, "learning_rate": 4.864833407502845e-07, "loss": 0.2922, "step": 24958 }, { "epoch": 1.7254752851711026, "grad_norm": 0.7497957465290426, "learning_rate": 4.862425250636043e-07, "loss": 0.3232, "step": 24959 }, { "epoch": 1.7255444175596266, "grad_norm": 0.7088482517176762, "learning_rate": 4.860017659488192e-07, "loss": 0.2096, "step": 24960 }, { "epoch": 1.7256135499481506, "grad_norm": 0.6490564194056002, "learning_rate": 4.85761063408946e-07, "loss": 0.2027, "step": 24961 }, { "epoch": 1.7256826823366747, "grad_norm": 0.8365714404718043, "learning_rate": 4.85520417447003e-07, "loss": 0.37, "step": 24962 }, { "epoch": 1.7257518147251987, "grad_norm": 0.8298503569784829, "learning_rate": 4.852798280660042e-07, "loss": 0.2953, "step": 24963 }, { "epoch": 1.7258209471137227, "grad_norm": 0.6660599502007714, "learning_rate": 4.850392952689665e-07, "loss": 0.2095, "step": 24964 }, { "epoch": 1.7258900795022467, "grad_norm": 0.7538977619142896, "learning_rate": 4.847988190589032e-07, "loss": 0.2795, "step": 24965 }, { "epoch": 1.7259592118907707, "grad_norm": 0.7740003263890649, "learning_rate": 4.845583994388297e-07, "loss": 0.2913, "step": 24966 }, { "epoch": 1.7260283442792947, "grad_norm": 0.6783402129084104, "learning_rate": 4.843180364117583e-07, "loss": 0.2306, "step": 24967 }, { "epoch": 1.7260974766678188, "grad_norm": 0.7183865085839544, "learning_rate": 4.840777299807009e-07, "loss": 0.2675, "step": 24968 }, { "epoch": 1.7261666090563428, "grad_norm": 0.6492630832907975, "learning_rate": 4.838374801486706e-07, "loss": 0.2913, "step": 24969 }, { "epoch": 1.7262357414448668, "grad_norm": 0.6608855265684065, "learning_rate": 4.835972869186789e-07, "loss": 0.2194, "step": 24970 }, { "epoch": 1.7263048738333908, "grad_norm": 0.6785954514509762, "learning_rate": 4.83357150293734e-07, "loss": 0.2588, "step": 24971 }, { "epoch": 1.7263740062219148, "grad_norm": 0.708320013830472, "learning_rate": 4.831170702768484e-07, "loss": 0.2964, "step": 24972 }, { "epoch": 1.7264431386104389, "grad_norm": 0.7041115730005674, "learning_rate": 4.828770468710281e-07, "loss": 0.3583, "step": 24973 }, { "epoch": 1.7265122709989629, "grad_norm": 0.6921683045367586, "learning_rate": 4.826370800792834e-07, "loss": 0.2276, "step": 24974 }, { "epoch": 1.726581403387487, "grad_norm": 0.677758353306513, "learning_rate": 4.82397169904621e-07, "loss": 0.2628, "step": 24975 }, { "epoch": 1.726650535776011, "grad_norm": 0.7285480825686989, "learning_rate": 4.821573163500487e-07, "loss": 0.3137, "step": 24976 }, { "epoch": 1.726719668164535, "grad_norm": 0.6919925208190443, "learning_rate": 4.819175194185716e-07, "loss": 0.3233, "step": 24977 }, { "epoch": 1.726788800553059, "grad_norm": 0.7468385120084248, "learning_rate": 4.816777791131961e-07, "loss": 0.2456, "step": 24978 }, { "epoch": 1.7268579329415832, "grad_norm": 0.6987137070373841, "learning_rate": 4.814380954369257e-07, "loss": 0.2534, "step": 24979 }, { "epoch": 1.7269270653301072, "grad_norm": 0.6788357223400213, "learning_rate": 4.811984683927657e-07, "loss": 0.2989, "step": 24980 }, { "epoch": 1.7269961977186312, "grad_norm": 0.7513793676375129, "learning_rate": 4.809588979837176e-07, "loss": 0.2555, "step": 24981 }, { "epoch": 1.7270653301071552, "grad_norm": 0.7602794551561526, "learning_rate": 4.807193842127855e-07, "loss": 0.2351, "step": 24982 }, { "epoch": 1.7271344624956793, "grad_norm": 0.6264953983273743, "learning_rate": 4.804799270829707e-07, "loss": 0.2377, "step": 24983 }, { "epoch": 1.7272035948842033, "grad_norm": 0.8613101744675481, "learning_rate": 4.80240526597276e-07, "loss": 0.322, "step": 24984 }, { "epoch": 1.7272727272727273, "grad_norm": 0.6763385089021078, "learning_rate": 4.800011827586992e-07, "loss": 0.224, "step": 24985 }, { "epoch": 1.7273418596612513, "grad_norm": 0.8080815838129383, "learning_rate": 4.797618955702421e-07, "loss": 0.2356, "step": 24986 }, { "epoch": 1.7274109920497753, "grad_norm": 0.7068785502074167, "learning_rate": 4.795226650349016e-07, "loss": 0.2816, "step": 24987 }, { "epoch": 1.7274801244382993, "grad_norm": 0.7589216312796216, "learning_rate": 4.792834911556788e-07, "loss": 0.3422, "step": 24988 }, { "epoch": 1.7275492568268234, "grad_norm": 0.6899150539121528, "learning_rate": 4.790443739355688e-07, "loss": 0.2435, "step": 24989 }, { "epoch": 1.7276183892153474, "grad_norm": 0.6979895437469715, "learning_rate": 4.788053133775695e-07, "loss": 0.2593, "step": 24990 }, { "epoch": 1.7276875216038714, "grad_norm": 0.7184592669428492, "learning_rate": 4.78566309484677e-07, "loss": 0.3552, "step": 24991 }, { "epoch": 1.7277566539923954, "grad_norm": 0.6776090716895417, "learning_rate": 4.783273622598883e-07, "loss": 0.2785, "step": 24992 }, { "epoch": 1.7278257863809194, "grad_norm": 0.8966161921279494, "learning_rate": 4.780884717061956e-07, "loss": 0.2234, "step": 24993 }, { "epoch": 1.7278949187694435, "grad_norm": 0.729727568749636, "learning_rate": 4.778496378265957e-07, "loss": 0.3433, "step": 24994 }, { "epoch": 1.7279640511579675, "grad_norm": 0.7199135352014812, "learning_rate": 4.776108606240787e-07, "loss": 0.4303, "step": 24995 }, { "epoch": 1.7280331835464915, "grad_norm": 0.6711750958752395, "learning_rate": 4.773721401016395e-07, "loss": 0.226, "step": 24996 }, { "epoch": 1.7281023159350155, "grad_norm": 0.7134060015390731, "learning_rate": 4.771334762622693e-07, "loss": 0.254, "step": 24997 }, { "epoch": 1.7281714483235395, "grad_norm": 0.7420644399943114, "learning_rate": 4.768948691089604e-07, "loss": 0.3419, "step": 24998 }, { "epoch": 1.7282405807120638, "grad_norm": 0.7440961560329177, "learning_rate": 4.766563186447015e-07, "loss": 0.272, "step": 24999 }, { "epoch": 1.7283097131005878, "grad_norm": 0.6093853781424199, "learning_rate": 4.764178248724849e-07, "loss": 0.2075, "step": 25000 }, { "epoch": 1.7283788454891118, "grad_norm": 0.732643087602201, "learning_rate": 4.7617938779529694e-07, "loss": 0.3252, "step": 25001 }, { "epoch": 1.7284479778776358, "grad_norm": 0.7250919054720537, "learning_rate": 4.7594100741612803e-07, "loss": 0.3114, "step": 25002 }, { "epoch": 1.7285171102661598, "grad_norm": 0.7468258620657294, "learning_rate": 4.757026837379641e-07, "loss": 0.1959, "step": 25003 }, { "epoch": 1.7285862426546839, "grad_norm": 0.6384036467849803, "learning_rate": 4.754644167637928e-07, "loss": 0.2298, "step": 25004 }, { "epoch": 1.7286553750432079, "grad_norm": 0.8485826173710389, "learning_rate": 4.7522620649660123e-07, "loss": 0.3304, "step": 25005 }, { "epoch": 1.728724507431732, "grad_norm": 0.7601884999070777, "learning_rate": 4.7498805293937465e-07, "loss": 0.2909, "step": 25006 }, { "epoch": 1.728793639820256, "grad_norm": 0.8256079209353879, "learning_rate": 4.7474995609509743e-07, "loss": 0.2248, "step": 25007 }, { "epoch": 1.72886277220878, "grad_norm": 0.7319934049266871, "learning_rate": 4.7451191596675384e-07, "loss": 0.2617, "step": 25008 }, { "epoch": 1.728931904597304, "grad_norm": 0.6716088951095106, "learning_rate": 4.7427393255732814e-07, "loss": 0.3109, "step": 25009 }, { "epoch": 1.729001036985828, "grad_norm": 0.6359774063419327, "learning_rate": 4.740360058697996e-07, "loss": 0.2167, "step": 25010 }, { "epoch": 1.729070169374352, "grad_norm": 0.7059709726368429, "learning_rate": 4.7379813590715527e-07, "loss": 0.2409, "step": 25011 }, { "epoch": 1.729139301762876, "grad_norm": 0.6858038048193863, "learning_rate": 4.7356032267237286e-07, "loss": 0.312, "step": 25012 }, { "epoch": 1.7292084341514, "grad_norm": 0.9249502943719204, "learning_rate": 4.733225661684349e-07, "loss": 0.2469, "step": 25013 }, { "epoch": 1.729277566539924, "grad_norm": 0.611245863951694, "learning_rate": 4.7308486639831896e-07, "loss": 0.1982, "step": 25014 }, { "epoch": 1.729346698928448, "grad_norm": 0.7514735548610698, "learning_rate": 4.7284722336500724e-07, "loss": 0.2411, "step": 25015 }, { "epoch": 1.729415831316972, "grad_norm": 0.6671460726311882, "learning_rate": 4.72609637071475e-07, "loss": 0.2918, "step": 25016 }, { "epoch": 1.729484963705496, "grad_norm": 0.7039535904447557, "learning_rate": 4.723721075207016e-07, "loss": 0.2519, "step": 25017 }, { "epoch": 1.72955409609402, "grad_norm": 0.7478001217778995, "learning_rate": 4.7213463471566413e-07, "loss": 0.1974, "step": 25018 }, { "epoch": 1.7296232284825441, "grad_norm": 0.7085444477911171, "learning_rate": 4.7189721865933956e-07, "loss": 0.3236, "step": 25019 }, { "epoch": 1.7296923608710681, "grad_norm": 0.7128817095949578, "learning_rate": 4.7165985935470113e-07, "loss": 0.335, "step": 25020 }, { "epoch": 1.7297614932595922, "grad_norm": 0.7334355752694386, "learning_rate": 4.71422556804727e-07, "loss": 0.279, "step": 25021 }, { "epoch": 1.7298306256481162, "grad_norm": 0.6521330357405589, "learning_rate": 4.711853110123882e-07, "loss": 0.2221, "step": 25022 }, { "epoch": 1.7298997580366402, "grad_norm": 0.7253736160216444, "learning_rate": 4.7094812198066e-07, "loss": 0.3502, "step": 25023 }, { "epoch": 1.7299688904251642, "grad_norm": 0.7331089819551594, "learning_rate": 4.7071098971251395e-07, "loss": 0.2024, "step": 25024 }, { "epoch": 1.7300380228136882, "grad_norm": 0.8987738316752268, "learning_rate": 4.7047391421092323e-07, "loss": 0.2442, "step": 25025 }, { "epoch": 1.7301071552022123, "grad_norm": 0.7126271591586792, "learning_rate": 4.702368954788583e-07, "loss": 0.2722, "step": 25026 }, { "epoch": 1.7301762875907363, "grad_norm": 0.8057971490614121, "learning_rate": 4.699999335192912e-07, "loss": 0.3207, "step": 25027 }, { "epoch": 1.7302454199792603, "grad_norm": 0.8904953573869435, "learning_rate": 4.697630283351895e-07, "loss": 0.2791, "step": 25028 }, { "epoch": 1.7303145523677843, "grad_norm": 0.6835050198722383, "learning_rate": 4.6952617992952475e-07, "loss": 0.302, "step": 25029 }, { "epoch": 1.7303836847563083, "grad_norm": 0.7478845516886184, "learning_rate": 4.6928938830526405e-07, "loss": 0.337, "step": 25030 }, { "epoch": 1.7304528171448323, "grad_norm": 0.7274717622210392, "learning_rate": 4.6905265346537496e-07, "loss": 0.2761, "step": 25031 }, { "epoch": 1.7305219495333564, "grad_norm": 0.6525958823220197, "learning_rate": 4.6881597541282464e-07, "loss": 0.2497, "step": 25032 }, { "epoch": 1.7305910819218804, "grad_norm": 0.6884691544074353, "learning_rate": 4.685793541505812e-07, "loss": 0.2549, "step": 25033 }, { "epoch": 1.7306602143104044, "grad_norm": 0.7803734135067674, "learning_rate": 4.6834278968160787e-07, "loss": 0.3704, "step": 25034 }, { "epoch": 1.7307293466989284, "grad_norm": 0.7171806142465494, "learning_rate": 4.6810628200887165e-07, "loss": 0.2597, "step": 25035 }, { "epoch": 1.7307984790874524, "grad_norm": 0.950771574703291, "learning_rate": 4.6786983113533415e-07, "loss": 0.2235, "step": 25036 }, { "epoch": 1.7308676114759765, "grad_norm": 0.6713026687671122, "learning_rate": 4.6763343706396183e-07, "loss": 0.3325, "step": 25037 }, { "epoch": 1.7309367438645005, "grad_norm": 0.7514582395327746, "learning_rate": 4.6739709979771506e-07, "loss": 0.2825, "step": 25038 }, { "epoch": 1.7310058762530245, "grad_norm": 0.6685702296073439, "learning_rate": 4.6716081933955657e-07, "loss": 0.2054, "step": 25039 }, { "epoch": 1.7310750086415485, "grad_norm": 0.6671343168017594, "learning_rate": 4.669245956924484e-07, "loss": 0.2479, "step": 25040 }, { "epoch": 1.7311441410300725, "grad_norm": 0.6734565587258021, "learning_rate": 4.6668842885935153e-07, "loss": 0.3562, "step": 25041 }, { "epoch": 1.7312132734185965, "grad_norm": 0.6936565624095375, "learning_rate": 4.6645231884322404e-07, "loss": 0.1967, "step": 25042 }, { "epoch": 1.7312824058071206, "grad_norm": 0.6895207095520735, "learning_rate": 4.6621626564702703e-07, "loss": 0.2158, "step": 25043 }, { "epoch": 1.7313515381956446, "grad_norm": 0.7524401449294619, "learning_rate": 4.659802692737181e-07, "loss": 0.3296, "step": 25044 }, { "epoch": 1.7314206705841686, "grad_norm": 0.7041902315898726, "learning_rate": 4.657443297262543e-07, "loss": 0.2798, "step": 25045 }, { "epoch": 1.7314898029726926, "grad_norm": 0.6798305485808345, "learning_rate": 4.655084470075938e-07, "loss": 0.2288, "step": 25046 }, { "epoch": 1.7315589353612166, "grad_norm": 0.6722702899979113, "learning_rate": 4.6527262112069427e-07, "loss": 0.2397, "step": 25047 }, { "epoch": 1.7316280677497407, "grad_norm": 0.724613650541852, "learning_rate": 4.6503685206850835e-07, "loss": 0.382, "step": 25048 }, { "epoch": 1.7316972001382647, "grad_norm": 0.7339356800827523, "learning_rate": 4.648011398539937e-07, "loss": 0.2393, "step": 25049 }, { "epoch": 1.7317663325267887, "grad_norm": 0.6587466393722845, "learning_rate": 4.6456548448010173e-07, "loss": 0.2384, "step": 25050 }, { "epoch": 1.7318354649153127, "grad_norm": 0.6586304233600567, "learning_rate": 4.643298859497891e-07, "loss": 0.2465, "step": 25051 }, { "epoch": 1.7319045973038367, "grad_norm": 0.7299453000604752, "learning_rate": 4.640943442660062e-07, "loss": 0.3586, "step": 25052 }, { "epoch": 1.7319737296923607, "grad_norm": 0.7535311290458612, "learning_rate": 4.638588594317056e-07, "loss": 0.2273, "step": 25053 }, { "epoch": 1.7320428620808848, "grad_norm": 0.7560796594137352, "learning_rate": 4.636234314498395e-07, "loss": 0.2572, "step": 25054 }, { "epoch": 1.7321119944694088, "grad_norm": 0.7235236934740149, "learning_rate": 4.6338806032335925e-07, "loss": 0.3286, "step": 25055 }, { "epoch": 1.7321811268579328, "grad_norm": 0.7453948235391744, "learning_rate": 4.631527460552121e-07, "loss": 0.2093, "step": 25056 }, { "epoch": 1.7322502592464568, "grad_norm": 0.6847820442073493, "learning_rate": 4.6291748864835063e-07, "loss": 0.2808, "step": 25057 }, { "epoch": 1.7323193916349808, "grad_norm": 0.7660169473783708, "learning_rate": 4.6268228810572025e-07, "loss": 0.299, "step": 25058 }, { "epoch": 1.7323885240235048, "grad_norm": 0.7444450457942798, "learning_rate": 4.6244714443027085e-07, "loss": 0.3247, "step": 25059 }, { "epoch": 1.732457656412029, "grad_norm": 0.6817217665310471, "learning_rate": 4.622120576249484e-07, "loss": 0.2333, "step": 25060 }, { "epoch": 1.732526788800553, "grad_norm": 0.6967089372537674, "learning_rate": 4.619770276926994e-07, "loss": 0.2555, "step": 25061 }, { "epoch": 1.7325959211890771, "grad_norm": 0.6831549428294167, "learning_rate": 4.617420546364704e-07, "loss": 0.3428, "step": 25062 }, { "epoch": 1.7326650535776011, "grad_norm": 0.6998791702588727, "learning_rate": 4.6150713845920626e-07, "loss": 0.276, "step": 25063 }, { "epoch": 1.7327341859661252, "grad_norm": 0.7290885999115018, "learning_rate": 4.612722791638502e-07, "loss": 0.2245, "step": 25064 }, { "epoch": 1.7328033183546492, "grad_norm": 0.7574164651910583, "learning_rate": 4.6103747675334767e-07, "loss": 0.2699, "step": 25065 }, { "epoch": 1.7328724507431732, "grad_norm": 0.7391485577003122, "learning_rate": 4.6080273123063845e-07, "loss": 0.3437, "step": 25066 }, { "epoch": 1.7329415831316972, "grad_norm": 0.6639145614670902, "learning_rate": 4.6056804259866694e-07, "loss": 0.2264, "step": 25067 }, { "epoch": 1.7330107155202212, "grad_norm": 0.9712747000686119, "learning_rate": 4.6033341086037406e-07, "loss": 0.3103, "step": 25068 }, { "epoch": 1.7330798479087453, "grad_norm": 0.6892335773709013, "learning_rate": 4.6009883601870144e-07, "loss": 0.3127, "step": 25069 }, { "epoch": 1.7331489802972693, "grad_norm": 0.7688568396771968, "learning_rate": 4.5986431807658717e-07, "loss": 0.2649, "step": 25070 }, { "epoch": 1.7332181126857933, "grad_norm": 0.7077769405445413, "learning_rate": 4.5962985703697225e-07, "loss": 0.3055, "step": 25071 }, { "epoch": 1.7332872450743173, "grad_norm": 0.6400098681290118, "learning_rate": 4.593954529027933e-07, "loss": 0.2655, "step": 25072 }, { "epoch": 1.7333563774628413, "grad_norm": 0.7026914719584776, "learning_rate": 4.591611056769901e-07, "loss": 0.3705, "step": 25073 }, { "epoch": 1.7334255098513653, "grad_norm": 0.8151341728906669, "learning_rate": 4.5892681536249874e-07, "loss": 0.2267, "step": 25074 }, { "epoch": 1.7334946422398894, "grad_norm": 0.8345826819618479, "learning_rate": 4.586925819622551e-07, "loss": 0.2322, "step": 25075 }, { "epoch": 1.7335637746284134, "grad_norm": 0.778141839607765, "learning_rate": 4.5845840547919685e-07, "loss": 0.3075, "step": 25076 }, { "epoch": 1.7336329070169374, "grad_norm": 0.8009851976758668, "learning_rate": 4.582242859162572e-07, "loss": 0.337, "step": 25077 }, { "epoch": 1.7337020394054614, "grad_norm": 0.9983255055291774, "learning_rate": 4.5799022327637113e-07, "loss": 0.2764, "step": 25078 }, { "epoch": 1.7337711717939854, "grad_norm": 0.76599937355357, "learning_rate": 4.577562175624717e-07, "loss": 0.2283, "step": 25079 }, { "epoch": 1.7338403041825095, "grad_norm": 0.7021941660243239, "learning_rate": 4.575222687774927e-07, "loss": 0.332, "step": 25080 }, { "epoch": 1.7339094365710337, "grad_norm": 0.7248545498327458, "learning_rate": 4.572883769243641e-07, "loss": 0.3344, "step": 25081 }, { "epoch": 1.7339785689595577, "grad_norm": 0.7061017456258452, "learning_rate": 4.5705454200602017e-07, "loss": 0.2426, "step": 25082 }, { "epoch": 1.7340477013480817, "grad_norm": 0.7403098582408474, "learning_rate": 4.5682076402538966e-07, "loss": 0.3081, "step": 25083 }, { "epoch": 1.7341168337366057, "grad_norm": 1.2807909877085628, "learning_rate": 4.565870429854047e-07, "loss": 0.3946, "step": 25084 }, { "epoch": 1.7341859661251298, "grad_norm": 0.6715972054720163, "learning_rate": 4.563533788889912e-07, "loss": 0.229, "step": 25085 }, { "epoch": 1.7342550985136538, "grad_norm": 0.8050473596442616, "learning_rate": 4.561197717390808e-07, "loss": 0.2719, "step": 25086 }, { "epoch": 1.7343242309021778, "grad_norm": 0.7238304322475942, "learning_rate": 4.558862215385995e-07, "loss": 0.3747, "step": 25087 }, { "epoch": 1.7343933632907018, "grad_norm": 0.7055607979196652, "learning_rate": 4.556527282904749e-07, "loss": 0.2498, "step": 25088 }, { "epoch": 1.7344624956792258, "grad_norm": 0.6114342835107485, "learning_rate": 4.5541929199763356e-07, "loss": 0.1865, "step": 25089 }, { "epoch": 1.7345316280677499, "grad_norm": 0.6886571507371361, "learning_rate": 4.551859126630026e-07, "loss": 0.2847, "step": 25090 }, { "epoch": 1.7346007604562739, "grad_norm": 0.7436982313385063, "learning_rate": 4.549525902895041e-07, "loss": 0.3473, "step": 25091 }, { "epoch": 1.734669892844798, "grad_norm": 0.6929096975686371, "learning_rate": 4.547193248800652e-07, "loss": 0.1858, "step": 25092 }, { "epoch": 1.734739025233322, "grad_norm": 0.7193553490074523, "learning_rate": 4.5448611643760685e-07, "loss": 0.2205, "step": 25093 }, { "epoch": 1.734808157621846, "grad_norm": 0.7008194749123486, "learning_rate": 4.5425296496505453e-07, "loss": 0.3248, "step": 25094 }, { "epoch": 1.73487729001037, "grad_norm": 0.6884737701793111, "learning_rate": 4.5401987046532813e-07, "loss": 0.3189, "step": 25095 }, { "epoch": 1.734946422398894, "grad_norm": 0.6531534532925467, "learning_rate": 4.5378683294134973e-07, "loss": 0.2484, "step": 25096 }, { "epoch": 1.735015554787418, "grad_norm": 0.6679569496111991, "learning_rate": 4.5355385239604034e-07, "loss": 0.2337, "step": 25097 }, { "epoch": 1.735084687175942, "grad_norm": 0.8847942264397769, "learning_rate": 4.533209288323209e-07, "loss": 0.3381, "step": 25098 }, { "epoch": 1.735153819564466, "grad_norm": 0.741130026254915, "learning_rate": 4.5308806225310864e-07, "loss": 0.2506, "step": 25099 }, { "epoch": 1.73522295195299, "grad_norm": 0.6983526306357004, "learning_rate": 4.5285525266132444e-07, "loss": 0.2145, "step": 25100 }, { "epoch": 1.735292084341514, "grad_norm": 0.6435476658031973, "learning_rate": 4.5262250005988327e-07, "loss": 0.2592, "step": 25101 }, { "epoch": 1.735361216730038, "grad_norm": 0.7100869688852695, "learning_rate": 4.5238980445170443e-07, "loss": 0.3623, "step": 25102 }, { "epoch": 1.735430349118562, "grad_norm": 0.7289025942917282, "learning_rate": 4.521571658397039e-07, "loss": 0.2515, "step": 25103 }, { "epoch": 1.735499481507086, "grad_norm": 0.6883374782313603, "learning_rate": 4.519245842267983e-07, "loss": 0.233, "step": 25104 }, { "epoch": 1.7355686138956101, "grad_norm": 0.7197521355301763, "learning_rate": 4.5169205961590026e-07, "loss": 0.3197, "step": 25105 }, { "epoch": 1.7356377462841341, "grad_norm": 0.7590582810367864, "learning_rate": 4.5145959200992685e-07, "loss": 0.2644, "step": 25106 }, { "epoch": 1.7357068786726582, "grad_norm": 0.620719967014245, "learning_rate": 4.5122718141178913e-07, "loss": 0.231, "step": 25107 }, { "epoch": 1.7357760110611822, "grad_norm": 0.7179691402480814, "learning_rate": 4.5099482782440195e-07, "loss": 0.2301, "step": 25108 }, { "epoch": 1.7358451434497062, "grad_norm": 0.7029745491629771, "learning_rate": 4.5076253125067524e-07, "loss": 0.3655, "step": 25109 }, { "epoch": 1.7359142758382302, "grad_norm": 0.7691600945460685, "learning_rate": 4.5053029169352225e-07, "loss": 0.239, "step": 25110 }, { "epoch": 1.7359834082267542, "grad_norm": 0.7064554736519687, "learning_rate": 4.502981091558528e-07, "loss": 0.2222, "step": 25111 }, { "epoch": 1.7360525406152783, "grad_norm": 0.6548886636350641, "learning_rate": 4.500659836405785e-07, "loss": 0.3197, "step": 25112 }, { "epoch": 1.7361216730038023, "grad_norm": 0.6848902005302578, "learning_rate": 4.498339151506059e-07, "loss": 0.2802, "step": 25113 }, { "epoch": 1.7361908053923263, "grad_norm": 0.6851336611122776, "learning_rate": 4.496019036888466e-07, "loss": 0.2453, "step": 25114 }, { "epoch": 1.7362599377808503, "grad_norm": 0.818112774172798, "learning_rate": 4.4936994925820543e-07, "loss": 0.2691, "step": 25115 }, { "epoch": 1.7363290701693743, "grad_norm": 0.659288034236178, "learning_rate": 4.491380518615923e-07, "loss": 0.3309, "step": 25116 }, { "epoch": 1.7363982025578983, "grad_norm": 0.7676791041713097, "learning_rate": 4.489062115019105e-07, "loss": 0.2615, "step": 25117 }, { "epoch": 1.7364673349464224, "grad_norm": 0.7192731638968474, "learning_rate": 4.486744281820693e-07, "loss": 0.2278, "step": 25118 }, { "epoch": 1.7365364673349464, "grad_norm": 0.707364972053524, "learning_rate": 4.4844270190497086e-07, "loss": 0.294, "step": 25119 }, { "epoch": 1.7366055997234704, "grad_norm": 0.7269367445324396, "learning_rate": 4.4821103267352116e-07, "loss": 0.3152, "step": 25120 }, { "epoch": 1.7366747321119944, "grad_norm": 0.7404011482156428, "learning_rate": 4.4797942049062296e-07, "loss": 0.2393, "step": 25121 }, { "epoch": 1.7367438645005184, "grad_norm": 0.73529599859887, "learning_rate": 4.4774786535917935e-07, "loss": 0.2106, "step": 25122 }, { "epoch": 1.7368129968890424, "grad_norm": 0.6707736563882762, "learning_rate": 4.47516367282092e-07, "loss": 0.307, "step": 25123 }, { "epoch": 1.7368821292775665, "grad_norm": 0.7924616788741627, "learning_rate": 4.4728492626226305e-07, "loss": 0.2363, "step": 25124 }, { "epoch": 1.7369512616660905, "grad_norm": 0.729382797776497, "learning_rate": 4.4705354230259236e-07, "loss": 0.2931, "step": 25125 }, { "epoch": 1.7370203940546145, "grad_norm": 0.7108500065994393, "learning_rate": 4.468222154059815e-07, "loss": 0.2328, "step": 25126 }, { "epoch": 1.7370895264431385, "grad_norm": 0.6446416853041383, "learning_rate": 4.4659094557532756e-07, "loss": 0.2739, "step": 25127 }, { "epoch": 1.7371586588316625, "grad_norm": 0.6840479480299778, "learning_rate": 4.4635973281353106e-07, "loss": 0.2362, "step": 25128 }, { "epoch": 1.7372277912201866, "grad_norm": 0.6610895766520652, "learning_rate": 4.461285771234886e-07, "loss": 0.2375, "step": 25129 }, { "epoch": 1.7372969236087106, "grad_norm": 0.8194662928516224, "learning_rate": 4.4589747850809827e-07, "loss": 0.3342, "step": 25130 }, { "epoch": 1.7373660559972346, "grad_norm": 0.6842484291396743, "learning_rate": 4.4566643697025457e-07, "loss": 0.2302, "step": 25131 }, { "epoch": 1.7374351883857586, "grad_norm": 0.7272159491520724, "learning_rate": 4.454354525128551e-07, "loss": 0.233, "step": 25132 }, { "epoch": 1.7375043207742826, "grad_norm": 0.6873196734620959, "learning_rate": 4.4520452513879375e-07, "loss": 0.2933, "step": 25133 }, { "epoch": 1.7375734531628066, "grad_norm": 0.7301368197840448, "learning_rate": 4.449736548509664e-07, "loss": 0.3209, "step": 25134 }, { "epoch": 1.7376425855513307, "grad_norm": 0.709773052663104, "learning_rate": 4.447428416522648e-07, "loss": 0.2434, "step": 25135 }, { "epoch": 1.7377117179398547, "grad_norm": 0.6610972645594643, "learning_rate": 4.4451208554558313e-07, "loss": 0.2281, "step": 25136 }, { "epoch": 1.7377808503283787, "grad_norm": 0.6908821293007806, "learning_rate": 4.442813865338119e-07, "loss": 0.2851, "step": 25137 }, { "epoch": 1.7378499827169027, "grad_norm": 0.69076547059848, "learning_rate": 4.440507446198439e-07, "loss": 0.25, "step": 25138 }, { "epoch": 1.7379191151054267, "grad_norm": 0.6977698174106264, "learning_rate": 4.438201598065689e-07, "loss": 0.2015, "step": 25139 }, { "epoch": 1.7379882474939508, "grad_norm": 0.7953916635533455, "learning_rate": 4.43589632096878e-07, "loss": 0.2289, "step": 25140 }, { "epoch": 1.7380573798824748, "grad_norm": 1.080672892436749, "learning_rate": 4.4335916149366054e-07, "loss": 0.3864, "step": 25141 }, { "epoch": 1.738126512270999, "grad_norm": 0.6306917249883648, "learning_rate": 4.431287479998025e-07, "loss": 0.2364, "step": 25142 }, { "epoch": 1.738195644659523, "grad_norm": 0.6614862151645972, "learning_rate": 4.4289839161819505e-07, "loss": 0.2295, "step": 25143 }, { "epoch": 1.738264777048047, "grad_norm": 0.7851123666360147, "learning_rate": 4.426680923517229e-07, "loss": 0.2683, "step": 25144 }, { "epoch": 1.738333909436571, "grad_norm": 0.7271672877822, "learning_rate": 4.424378502032728e-07, "loss": 0.2982, "step": 25145 }, { "epoch": 1.738403041825095, "grad_norm": 0.7733471780817592, "learning_rate": 4.422076651757312e-07, "loss": 0.2832, "step": 25146 }, { "epoch": 1.738472174213619, "grad_norm": 0.7137889963081885, "learning_rate": 4.4197753727198366e-07, "loss": 0.261, "step": 25147 }, { "epoch": 1.7385413066021431, "grad_norm": 0.8174808650793272, "learning_rate": 4.4174746649491284e-07, "loss": 0.3841, "step": 25148 }, { "epoch": 1.7386104389906671, "grad_norm": 0.6840887640546308, "learning_rate": 4.4151745284740365e-07, "loss": 0.2025, "step": 25149 }, { "epoch": 1.7386795713791912, "grad_norm": 0.7659326580491413, "learning_rate": 4.4128749633233716e-07, "loss": 0.225, "step": 25150 }, { "epoch": 1.7387487037677152, "grad_norm": 0.7195401476847552, "learning_rate": 4.410575969525976e-07, "loss": 0.2887, "step": 25151 }, { "epoch": 1.7388178361562392, "grad_norm": 0.71549995676822, "learning_rate": 4.408277547110634e-07, "loss": 0.272, "step": 25152 }, { "epoch": 1.7388869685447632, "grad_norm": 0.764893274800972, "learning_rate": 4.4059796961061886e-07, "loss": 0.2236, "step": 25153 }, { "epoch": 1.7389561009332872, "grad_norm": 0.7860101448831983, "learning_rate": 4.4036824165414104e-07, "loss": 0.2575, "step": 25154 }, { "epoch": 1.7390252333218112, "grad_norm": 0.7390135811713388, "learning_rate": 4.401385708445116e-07, "loss": 0.2973, "step": 25155 }, { "epoch": 1.7390943657103353, "grad_norm": 0.6932192545620527, "learning_rate": 4.399089571846066e-07, "loss": 0.2403, "step": 25156 }, { "epoch": 1.7391634980988593, "grad_norm": 0.6647178854008449, "learning_rate": 4.3967940067730583e-07, "loss": 0.1954, "step": 25157 }, { "epoch": 1.7392326304873833, "grad_norm": 0.716561751523148, "learning_rate": 4.3944990132548495e-07, "loss": 0.279, "step": 25158 }, { "epoch": 1.7393017628759073, "grad_norm": 0.6605642209836862, "learning_rate": 4.3922045913202036e-07, "loss": 0.3647, "step": 25159 }, { "epoch": 1.7393708952644313, "grad_norm": 0.68716455402896, "learning_rate": 4.389910740997888e-07, "loss": 0.2069, "step": 25160 }, { "epoch": 1.7394400276529554, "grad_norm": 0.8583273574902668, "learning_rate": 4.387617462316651e-07, "loss": 0.2473, "step": 25161 }, { "epoch": 1.7395091600414796, "grad_norm": 0.6789287875303921, "learning_rate": 4.385324755305226e-07, "loss": 0.2865, "step": 25162 }, { "epoch": 1.7395782924300036, "grad_norm": 0.6977157763107273, "learning_rate": 4.383032619992356e-07, "loss": 0.3211, "step": 25163 }, { "epoch": 1.7396474248185276, "grad_norm": 1.0386390076417398, "learning_rate": 4.380741056406757e-07, "loss": 0.2447, "step": 25164 }, { "epoch": 1.7397165572070517, "grad_norm": 0.6313544022580385, "learning_rate": 4.3784500645771724e-07, "loss": 0.2339, "step": 25165 }, { "epoch": 1.7397856895955757, "grad_norm": 0.7726623437283107, "learning_rate": 4.3761596445322853e-07, "loss": 0.373, "step": 25166 }, { "epoch": 1.7398548219840997, "grad_norm": 0.6126652480624573, "learning_rate": 4.373869796300817e-07, "loss": 0.2078, "step": 25167 }, { "epoch": 1.7399239543726237, "grad_norm": 0.653899831385861, "learning_rate": 4.3715805199114724e-07, "loss": 0.2072, "step": 25168 }, { "epoch": 1.7399930867611477, "grad_norm": 0.7006480855962591, "learning_rate": 4.369291815392945e-07, "loss": 0.3386, "step": 25169 }, { "epoch": 1.7400622191496717, "grad_norm": 0.7329343393776797, "learning_rate": 4.3670036827739016e-07, "loss": 0.273, "step": 25170 }, { "epoch": 1.7401313515381958, "grad_norm": 0.7259665362849137, "learning_rate": 4.364716122083046e-07, "loss": 0.2422, "step": 25171 }, { "epoch": 1.7402004839267198, "grad_norm": 0.7332187303797624, "learning_rate": 4.362429133349022e-07, "loss": 0.2957, "step": 25172 }, { "epoch": 1.7402696163152438, "grad_norm": 0.6499178635763138, "learning_rate": 4.360142716600502e-07, "loss": 0.3003, "step": 25173 }, { "epoch": 1.7403387487037678, "grad_norm": 0.7122886255948696, "learning_rate": 4.3578568718661505e-07, "loss": 0.2634, "step": 25174 }, { "epoch": 1.7404078810922918, "grad_norm": 0.638543025677638, "learning_rate": 4.355571599174618e-07, "loss": 0.2335, "step": 25175 }, { "epoch": 1.7404770134808158, "grad_norm": 0.7372176178437763, "learning_rate": 4.3532868985545264e-07, "loss": 0.288, "step": 25176 }, { "epoch": 1.7405461458693399, "grad_norm": 0.7314766646552548, "learning_rate": 4.351002770034535e-07, "loss": 0.3126, "step": 25177 }, { "epoch": 1.7406152782578639, "grad_norm": 0.6486536726948394, "learning_rate": 4.3487192136432545e-07, "loss": 0.2007, "step": 25178 }, { "epoch": 1.740684410646388, "grad_norm": 0.7428313273115201, "learning_rate": 4.346436229409312e-07, "loss": 0.2116, "step": 25179 }, { "epoch": 1.740753543034912, "grad_norm": 1.4299219830549283, "learning_rate": 4.344153817361313e-07, "loss": 0.2872, "step": 25180 }, { "epoch": 1.740822675423436, "grad_norm": 0.8153262707548428, "learning_rate": 4.341871977527873e-07, "loss": 0.3012, "step": 25181 }, { "epoch": 1.74089180781196, "grad_norm": 0.6507592066488201, "learning_rate": 4.3395907099375855e-07, "loss": 0.2397, "step": 25182 }, { "epoch": 1.740960940200484, "grad_norm": 0.7519495225902806, "learning_rate": 4.3373100146190505e-07, "loss": 0.2692, "step": 25183 }, { "epoch": 1.741030072589008, "grad_norm": 0.7827846326285949, "learning_rate": 4.3350298916008393e-07, "loss": 0.2853, "step": 25184 }, { "epoch": 1.741099204977532, "grad_norm": 0.8131408510602869, "learning_rate": 4.33275034091154e-07, "loss": 0.32, "step": 25185 }, { "epoch": 1.741168337366056, "grad_norm": 0.7066856392838061, "learning_rate": 4.3304713625797137e-07, "loss": 0.1916, "step": 25186 }, { "epoch": 1.74123746975458, "grad_norm": 0.8039320083519725, "learning_rate": 4.3281929566339363e-07, "loss": 0.3295, "step": 25187 }, { "epoch": 1.741306602143104, "grad_norm": 0.8800160778557309, "learning_rate": 4.3259151231027364e-07, "loss": 0.2844, "step": 25188 }, { "epoch": 1.741375734531628, "grad_norm": 0.7076412721466132, "learning_rate": 4.3236378620147013e-07, "loss": 0.2408, "step": 25189 }, { "epoch": 1.741444866920152, "grad_norm": 0.7268374940318798, "learning_rate": 4.3213611733983417e-07, "loss": 0.2596, "step": 25190 }, { "epoch": 1.7415139993086761, "grad_norm": 0.6897170445259074, "learning_rate": 4.3190850572822073e-07, "loss": 0.3464, "step": 25191 }, { "epoch": 1.7415831316972001, "grad_norm": 0.7777586761726402, "learning_rate": 4.31680951369482e-07, "loss": 0.2268, "step": 25192 }, { "epoch": 1.7416522640857242, "grad_norm": 0.6838385459646105, "learning_rate": 4.314534542664706e-07, "loss": 0.2224, "step": 25193 }, { "epoch": 1.7417213964742482, "grad_norm": 0.7131875022447843, "learning_rate": 4.3122601442203594e-07, "loss": 0.3487, "step": 25194 }, { "epoch": 1.7417905288627722, "grad_norm": 0.647131745007762, "learning_rate": 4.309986318390302e-07, "loss": 0.2555, "step": 25195 }, { "epoch": 1.7418596612512962, "grad_norm": 0.7918942679608855, "learning_rate": 4.3077130652030286e-07, "loss": 0.2257, "step": 25196 }, { "epoch": 1.7419287936398202, "grad_norm": 0.6725428858379582, "learning_rate": 4.3054403846870375e-07, "loss": 0.2283, "step": 25197 }, { "epoch": 1.7419979260283442, "grad_norm": 0.7982792352081013, "learning_rate": 4.303168276870795e-07, "loss": 0.3676, "step": 25198 }, { "epoch": 1.7420670584168683, "grad_norm": 0.7097136734179612, "learning_rate": 4.3008967417828004e-07, "loss": 0.2631, "step": 25199 }, { "epoch": 1.7421361908053923, "grad_norm": 0.7752477223183644, "learning_rate": 4.2986257794515043e-07, "loss": 0.2746, "step": 25200 }, { "epoch": 1.7422053231939163, "grad_norm": 0.7189395540638063, "learning_rate": 4.296355389905382e-07, "loss": 0.2853, "step": 25201 }, { "epoch": 1.7422744555824403, "grad_norm": 0.6878601076773708, "learning_rate": 4.2940855731728736e-07, "loss": 0.2693, "step": 25202 }, { "epoch": 1.7423435879709643, "grad_norm": 0.6290801636729364, "learning_rate": 4.291816329282439e-07, "loss": 0.25, "step": 25203 }, { "epoch": 1.7424127203594884, "grad_norm": 0.6959489297754071, "learning_rate": 4.2895476582625164e-07, "loss": 0.2217, "step": 25204 }, { "epoch": 1.7424818527480124, "grad_norm": 0.6670438167338829, "learning_rate": 4.28727956014155e-07, "loss": 0.336, "step": 25205 }, { "epoch": 1.7425509851365364, "grad_norm": 1.4450754936737613, "learning_rate": 4.2850120349479443e-07, "loss": 0.2792, "step": 25206 }, { "epoch": 1.7426201175250604, "grad_norm": 0.7312693602613872, "learning_rate": 4.2827450827101434e-07, "loss": 0.2184, "step": 25207 }, { "epoch": 1.7426892499135844, "grad_norm": 0.7296765042126204, "learning_rate": 4.2804787034565363e-07, "loss": 0.2921, "step": 25208 }, { "epoch": 1.7427583823021084, "grad_norm": 0.7001576839337391, "learning_rate": 4.278212897215539e-07, "loss": 0.3246, "step": 25209 }, { "epoch": 1.7428275146906325, "grad_norm": 0.7233877214610415, "learning_rate": 4.275947664015556e-07, "loss": 0.2111, "step": 25210 }, { "epoch": 1.7428966470791565, "grad_norm": 0.7285916800810784, "learning_rate": 4.2736830038849597e-07, "loss": 0.233, "step": 25211 }, { "epoch": 1.7429657794676805, "grad_norm": 0.6564871949864887, "learning_rate": 4.271418916852155e-07, "loss": 0.2894, "step": 25212 }, { "epoch": 1.7430349118562045, "grad_norm": 0.7193907277836279, "learning_rate": 4.269155402945502e-07, "loss": 0.2916, "step": 25213 }, { "epoch": 1.7431040442447285, "grad_norm": 0.6565266758025292, "learning_rate": 4.2668924621933837e-07, "loss": 0.2532, "step": 25214 }, { "epoch": 1.7431731766332526, "grad_norm": 0.748130309443486, "learning_rate": 4.264630094624145e-07, "loss": 0.2707, "step": 25215 }, { "epoch": 1.7432423090217766, "grad_norm": 0.7073713800428466, "learning_rate": 4.2623683002661454e-07, "loss": 0.3264, "step": 25216 }, { "epoch": 1.7433114414103006, "grad_norm": 0.8457973712350977, "learning_rate": 4.2601070791477405e-07, "loss": 0.2316, "step": 25217 }, { "epoch": 1.7433805737988246, "grad_norm": 1.101641312499142, "learning_rate": 4.25784643129728e-07, "loss": 0.2027, "step": 25218 }, { "epoch": 1.7434497061873486, "grad_norm": 0.7233720703653658, "learning_rate": 4.2555863567430687e-07, "loss": 0.3475, "step": 25219 }, { "epoch": 1.7435188385758726, "grad_norm": 0.7203878891006851, "learning_rate": 4.2533268555134567e-07, "loss": 0.3096, "step": 25220 }, { "epoch": 1.7435879709643967, "grad_norm": 0.7245868630091115, "learning_rate": 4.2510679276367483e-07, "loss": 0.2247, "step": 25221 }, { "epoch": 1.7436571033529207, "grad_norm": 0.8277080017748698, "learning_rate": 4.2488095731412657e-07, "loss": 0.2243, "step": 25222 }, { "epoch": 1.743726235741445, "grad_norm": 0.8369283330767339, "learning_rate": 4.2465517920552977e-07, "loss": 0.365, "step": 25223 }, { "epoch": 1.743795368129969, "grad_norm": 0.7013765666471827, "learning_rate": 4.244294584407166e-07, "loss": 0.2495, "step": 25224 }, { "epoch": 1.743864500518493, "grad_norm": 0.6860132016491141, "learning_rate": 4.242037950225136e-07, "loss": 0.2719, "step": 25225 }, { "epoch": 1.743933632907017, "grad_norm": 0.7152955541002929, "learning_rate": 4.2397818895375143e-07, "loss": 0.3639, "step": 25226 }, { "epoch": 1.744002765295541, "grad_norm": 0.7434927723488309, "learning_rate": 4.2375264023725495e-07, "loss": 0.3194, "step": 25227 }, { "epoch": 1.744071897684065, "grad_norm": 0.6766006933635362, "learning_rate": 4.2352714887585357e-07, "loss": 0.2357, "step": 25228 }, { "epoch": 1.744141030072589, "grad_norm": 0.6998905618376855, "learning_rate": 4.233017148723717e-07, "loss": 0.2272, "step": 25229 }, { "epoch": 1.744210162461113, "grad_norm": 0.7126712117295398, "learning_rate": 4.2307633822963487e-07, "loss": 0.3281, "step": 25230 }, { "epoch": 1.744279294849637, "grad_norm": 0.7164698844391837, "learning_rate": 4.2285101895046863e-07, "loss": 0.2321, "step": 25231 }, { "epoch": 1.744348427238161, "grad_norm": 0.6398271642128126, "learning_rate": 4.226257570376974e-07, "loss": 0.2475, "step": 25232 }, { "epoch": 1.744417559626685, "grad_norm": 0.6634748583525635, "learning_rate": 4.224005524941427e-07, "loss": 0.2862, "step": 25233 }, { "epoch": 1.7444866920152091, "grad_norm": 0.7744285002791443, "learning_rate": 4.22175405322629e-07, "loss": 0.2798, "step": 25234 }, { "epoch": 1.7445558244037331, "grad_norm": 0.6614142595175392, "learning_rate": 4.219503155259763e-07, "loss": 0.2347, "step": 25235 }, { "epoch": 1.7446249567922572, "grad_norm": 0.6866397788705545, "learning_rate": 4.217252831070068e-07, "loss": 0.2579, "step": 25236 }, { "epoch": 1.7446940891807812, "grad_norm": 0.6724720107436503, "learning_rate": 4.215003080685404e-07, "loss": 0.3411, "step": 25237 }, { "epoch": 1.7447632215693052, "grad_norm": 0.6777714381438398, "learning_rate": 4.2127539041339647e-07, "loss": 0.3094, "step": 25238 }, { "epoch": 1.7448323539578292, "grad_norm": 0.6298297546534947, "learning_rate": 4.2105053014439455e-07, "loss": 0.2495, "step": 25239 }, { "epoch": 1.7449014863463532, "grad_norm": 3.7608589812746955, "learning_rate": 4.208257272643534e-07, "loss": 0.2028, "step": 25240 }, { "epoch": 1.7449706187348772, "grad_norm": 0.7502515217853196, "learning_rate": 4.2060098177608923e-07, "loss": 0.3268, "step": 25241 }, { "epoch": 1.7450397511234013, "grad_norm": 0.8666075246451057, "learning_rate": 4.2037629368242015e-07, "loss": 0.2123, "step": 25242 }, { "epoch": 1.7451088835119255, "grad_norm": 0.7045355363952953, "learning_rate": 4.201516629861607e-07, "loss": 0.1988, "step": 25243 }, { "epoch": 1.7451780159004495, "grad_norm": 0.7277086252339127, "learning_rate": 4.199270896901275e-07, "loss": 0.2554, "step": 25244 }, { "epoch": 1.7452471482889735, "grad_norm": 0.7211379591295234, "learning_rate": 4.197025737971344e-07, "loss": 0.2692, "step": 25245 }, { "epoch": 1.7453162806774976, "grad_norm": 0.9570899665029632, "learning_rate": 4.1947811530999636e-07, "loss": 0.2437, "step": 25246 }, { "epoch": 1.7453854130660216, "grad_norm": 0.670598256678208, "learning_rate": 4.1925371423152505e-07, "loss": 0.3343, "step": 25247 }, { "epoch": 1.7454545454545456, "grad_norm": 0.7346572689560873, "learning_rate": 4.190293705645343e-07, "loss": 0.3318, "step": 25248 }, { "epoch": 1.7455236778430696, "grad_norm": 0.6977734263526089, "learning_rate": 4.188050843118352e-07, "loss": 0.2436, "step": 25249 }, { "epoch": 1.7455928102315936, "grad_norm": 0.6662562166694647, "learning_rate": 4.1858085547623883e-07, "loss": 0.245, "step": 25250 }, { "epoch": 1.7456619426201176, "grad_norm": 0.7342909018902108, "learning_rate": 4.1835668406055516e-07, "loss": 0.3049, "step": 25251 }, { "epoch": 1.7457310750086417, "grad_norm": 0.8115922736100513, "learning_rate": 4.1813257006759424e-07, "loss": 0.3777, "step": 25252 }, { "epoch": 1.7458002073971657, "grad_norm": 0.6978948687878709, "learning_rate": 4.1790851350016424e-07, "loss": 0.2367, "step": 25253 }, { "epoch": 1.7458693397856897, "grad_norm": 0.6482524499892928, "learning_rate": 4.1768451436107525e-07, "loss": 0.2104, "step": 25254 }, { "epoch": 1.7459384721742137, "grad_norm": 0.782058886054963, "learning_rate": 4.174605726531328e-07, "loss": 0.3492, "step": 25255 }, { "epoch": 1.7460076045627377, "grad_norm": 0.7593508559688762, "learning_rate": 4.1723668837914455e-07, "loss": 0.2663, "step": 25256 }, { "epoch": 1.7460767369512618, "grad_norm": 0.7039659049095384, "learning_rate": 4.1701286154191554e-07, "loss": 0.2437, "step": 25257 }, { "epoch": 1.7461458693397858, "grad_norm": 0.6779312210265803, "learning_rate": 4.167890921442519e-07, "loss": 0.2547, "step": 25258 }, { "epoch": 1.7462150017283098, "grad_norm": 0.8033306907810388, "learning_rate": 4.1656538018895685e-07, "loss": 0.3003, "step": 25259 }, { "epoch": 1.7462841341168338, "grad_norm": 0.7164717482058709, "learning_rate": 4.1634172567883655e-07, "loss": 0.2721, "step": 25260 }, { "epoch": 1.7463532665053578, "grad_norm": 0.6347727056978166, "learning_rate": 4.161181286166921e-07, "loss": 0.2272, "step": 25261 }, { "epoch": 1.7464223988938818, "grad_norm": 0.6578250661035636, "learning_rate": 4.1589458900532733e-07, "loss": 0.2684, "step": 25262 }, { "epoch": 1.7464915312824059, "grad_norm": 0.697984696863009, "learning_rate": 4.1567110684754276e-07, "loss": 0.2523, "step": 25263 }, { "epoch": 1.7465606636709299, "grad_norm": 0.6632588021882138, "learning_rate": 4.1544768214614006e-07, "loss": 0.219, "step": 25264 }, { "epoch": 1.746629796059454, "grad_norm": 0.6420423914189045, "learning_rate": 4.152243149039187e-07, "loss": 0.2349, "step": 25265 }, { "epoch": 1.746698928447978, "grad_norm": 0.6933476474575522, "learning_rate": 4.15001005123678e-07, "loss": 0.3832, "step": 25266 }, { "epoch": 1.746768060836502, "grad_norm": 0.7507862302907321, "learning_rate": 4.147777528082181e-07, "loss": 0.2781, "step": 25267 }, { "epoch": 1.746837193225026, "grad_norm": 0.6406258394538962, "learning_rate": 4.145545579603372e-07, "loss": 0.2441, "step": 25268 }, { "epoch": 1.74690632561355, "grad_norm": 0.706141386336606, "learning_rate": 4.143314205828308e-07, "loss": 0.2814, "step": 25269 }, { "epoch": 1.746975458002074, "grad_norm": 0.7336689578980494, "learning_rate": 4.141083406784974e-07, "loss": 0.306, "step": 25270 }, { "epoch": 1.747044590390598, "grad_norm": 0.6489052584272127, "learning_rate": 4.1388531825013123e-07, "loss": 0.2168, "step": 25271 }, { "epoch": 1.747113722779122, "grad_norm": 0.7578999271023086, "learning_rate": 4.1366235330052906e-07, "loss": 0.2573, "step": 25272 }, { "epoch": 1.747182855167646, "grad_norm": 0.6908044542381144, "learning_rate": 4.1343944583248365e-07, "loss": 0.3299, "step": 25273 }, { "epoch": 1.74725198755617, "grad_norm": 0.7452920267479465, "learning_rate": 4.1321659584878993e-07, "loss": 0.2434, "step": 25274 }, { "epoch": 1.747321119944694, "grad_norm": 0.7215768348353108, "learning_rate": 4.1299380335224127e-07, "loss": 0.2109, "step": 25275 }, { "epoch": 1.747390252333218, "grad_norm": 0.7095679722053686, "learning_rate": 4.1277106834562874e-07, "loss": 0.295, "step": 25276 }, { "epoch": 1.7474593847217421, "grad_norm": 0.6650667632264171, "learning_rate": 4.125483908317457e-07, "loss": 0.2569, "step": 25277 }, { "epoch": 1.7475285171102661, "grad_norm": 0.6483868240572283, "learning_rate": 4.12325770813381e-07, "loss": 0.1895, "step": 25278 }, { "epoch": 1.7475976494987902, "grad_norm": 0.9249450087875853, "learning_rate": 4.1210320829332516e-07, "loss": 0.2252, "step": 25279 }, { "epoch": 1.7476667818873142, "grad_norm": 0.790245238879962, "learning_rate": 4.118807032743688e-07, "loss": 0.3806, "step": 25280 }, { "epoch": 1.7477359142758382, "grad_norm": 0.7763263790070645, "learning_rate": 4.116582557593002e-07, "loss": 0.2537, "step": 25281 }, { "epoch": 1.7478050466643622, "grad_norm": 0.7640745274196198, "learning_rate": 4.1143586575090653e-07, "loss": 0.2087, "step": 25282 }, { "epoch": 1.7478741790528862, "grad_norm": 0.7658292067719773, "learning_rate": 4.112135332519768e-07, "loss": 0.2434, "step": 25283 }, { "epoch": 1.7479433114414102, "grad_norm": 0.7536743091982637, "learning_rate": 4.109912582652953e-07, "loss": 0.3713, "step": 25284 }, { "epoch": 1.7480124438299343, "grad_norm": 0.6567012153431717, "learning_rate": 4.107690407936493e-07, "loss": 0.2282, "step": 25285 }, { "epoch": 1.7480815762184583, "grad_norm": 0.6643989039055826, "learning_rate": 4.105468808398233e-07, "loss": 0.2208, "step": 25286 }, { "epoch": 1.7481507086069823, "grad_norm": 0.7586283443234438, "learning_rate": 4.1032477840660165e-07, "loss": 0.3212, "step": 25287 }, { "epoch": 1.7482198409955063, "grad_norm": 0.6891019017031061, "learning_rate": 4.101027334967683e-07, "loss": 0.3386, "step": 25288 }, { "epoch": 1.7482889733840303, "grad_norm": 0.674355556548064, "learning_rate": 4.0988074611310714e-07, "loss": 0.2515, "step": 25289 }, { "epoch": 1.7483581057725543, "grad_norm": 0.7283930227082375, "learning_rate": 4.096588162583981e-07, "loss": 0.2881, "step": 25290 }, { "epoch": 1.7484272381610784, "grad_norm": 0.6725848897885123, "learning_rate": 4.094369439354251e-07, "loss": 0.3491, "step": 25291 }, { "epoch": 1.7484963705496024, "grad_norm": 0.6918377564661397, "learning_rate": 4.09215129146967e-07, "loss": 0.2387, "step": 25292 }, { "epoch": 1.7485655029381264, "grad_norm": 0.7320436655838695, "learning_rate": 4.089933718958056e-07, "loss": 0.2444, "step": 25293 }, { "epoch": 1.7486346353266504, "grad_norm": 0.8449307768645461, "learning_rate": 4.0877167218471735e-07, "loss": 0.306, "step": 25294 }, { "epoch": 1.7487037677151744, "grad_norm": 0.7264068769832521, "learning_rate": 4.085500300164841e-07, "loss": 0.3441, "step": 25295 }, { "epoch": 1.7487729001036985, "grad_norm": 0.7015642539037897, "learning_rate": 4.0832844539388237e-07, "loss": 0.2685, "step": 25296 }, { "epoch": 1.7488420324922225, "grad_norm": 0.6649807232694297, "learning_rate": 4.081069183196895e-07, "loss": 0.2179, "step": 25297 }, { "epoch": 1.7489111648807465, "grad_norm": 0.7422563000245512, "learning_rate": 4.078854487966816e-07, "loss": 0.292, "step": 25298 }, { "epoch": 1.7489802972692705, "grad_norm": 0.6854862495834255, "learning_rate": 4.076640368276347e-07, "loss": 0.2433, "step": 25299 }, { "epoch": 1.7490494296577945, "grad_norm": 0.6352213783498917, "learning_rate": 4.0744268241532327e-07, "loss": 0.2158, "step": 25300 }, { "epoch": 1.7491185620463185, "grad_norm": 0.6881303730615075, "learning_rate": 4.072213855625223e-07, "loss": 0.2888, "step": 25301 }, { "epoch": 1.7491876944348426, "grad_norm": 0.7076085438146585, "learning_rate": 4.070001462720047e-07, "loss": 0.3444, "step": 25302 }, { "epoch": 1.7492568268233666, "grad_norm": 0.6299187872575985, "learning_rate": 4.067789645465448e-07, "loss": 0.1972, "step": 25303 }, { "epoch": 1.7493259592118908, "grad_norm": 0.7131620712128108, "learning_rate": 4.0655784038891254e-07, "loss": 0.3026, "step": 25304 }, { "epoch": 1.7493950916004148, "grad_norm": 0.693744400116112, "learning_rate": 4.063367738018814e-07, "loss": 0.3183, "step": 25305 }, { "epoch": 1.7494642239889389, "grad_norm": 0.7092165422185743, "learning_rate": 4.0611576478821966e-07, "loss": 0.2958, "step": 25306 }, { "epoch": 1.7495333563774629, "grad_norm": 0.7572322727564486, "learning_rate": 4.0589481335070015e-07, "loss": 0.2395, "step": 25307 }, { "epoch": 1.749602488765987, "grad_norm": 0.7674918701182405, "learning_rate": 4.0567391949208944e-07, "loss": 0.2727, "step": 25308 }, { "epoch": 1.749671621154511, "grad_norm": 0.6876378716463636, "learning_rate": 4.054530832151576e-07, "loss": 0.3416, "step": 25309 }, { "epoch": 1.749740753543035, "grad_norm": 0.7504170347654239, "learning_rate": 4.052323045226714e-07, "loss": 0.2571, "step": 25310 }, { "epoch": 1.749809885931559, "grad_norm": 0.6535172759048946, "learning_rate": 4.050115834174001e-07, "loss": 0.1859, "step": 25311 }, { "epoch": 1.749879018320083, "grad_norm": 0.7191859409201671, "learning_rate": 4.0479091990210715e-07, "loss": 0.3213, "step": 25312 }, { "epoch": 1.749948150708607, "grad_norm": 0.689585239896343, "learning_rate": 4.0457031397956036e-07, "loss": 0.3016, "step": 25313 }, { "epoch": 1.750017283097131, "grad_norm": 0.678200370740747, "learning_rate": 4.0434976565252304e-07, "loss": 0.2847, "step": 25314 }, { "epoch": 1.750086415485655, "grad_norm": 0.7249868116523832, "learning_rate": 4.041292749237602e-07, "loss": 0.247, "step": 25315 }, { "epoch": 1.750155547874179, "grad_norm": 0.6824093700636197, "learning_rate": 4.0390884179603463e-07, "loss": 0.3799, "step": 25316 }, { "epoch": 1.750224680262703, "grad_norm": 0.65637471929163, "learning_rate": 4.036884662721113e-07, "loss": 0.2489, "step": 25317 }, { "epoch": 1.750293812651227, "grad_norm": 0.6084129485088237, "learning_rate": 4.034681483547492e-07, "loss": 0.2284, "step": 25318 }, { "epoch": 1.750362945039751, "grad_norm": 0.7669679810251336, "learning_rate": 4.032478880467122e-07, "loss": 0.2878, "step": 25319 }, { "epoch": 1.7504320774282751, "grad_norm": 0.7753759334313577, "learning_rate": 4.030276853507581e-07, "loss": 0.2988, "step": 25320 }, { "epoch": 1.7505012098167991, "grad_norm": 0.7622108986366415, "learning_rate": 4.028075402696496e-07, "loss": 0.1847, "step": 25321 }, { "epoch": 1.7505703422053231, "grad_norm": 1.184534229218848, "learning_rate": 4.025874528061441e-07, "loss": 0.2296, "step": 25322 }, { "epoch": 1.7506394745938472, "grad_norm": 0.6962214512383327, "learning_rate": 4.023674229629998e-07, "loss": 0.349, "step": 25323 }, { "epoch": 1.7507086069823714, "grad_norm": 0.6691527320494104, "learning_rate": 4.021474507429751e-07, "loss": 0.2302, "step": 25324 }, { "epoch": 1.7507777393708954, "grad_norm": 0.7000568194596594, "learning_rate": 4.0192753614882783e-07, "loss": 0.1984, "step": 25325 }, { "epoch": 1.7508468717594194, "grad_norm": 0.7278432650984876, "learning_rate": 4.0170767918331245e-07, "loss": 0.3006, "step": 25326 }, { "epoch": 1.7509160041479435, "grad_norm": 0.7194259649028506, "learning_rate": 4.014878798491861e-07, "loss": 0.3175, "step": 25327 }, { "epoch": 1.7509851365364675, "grad_norm": 0.6570769809606914, "learning_rate": 4.012681381492023e-07, "loss": 0.2432, "step": 25328 }, { "epoch": 1.7510542689249915, "grad_norm": 0.750054394738803, "learning_rate": 4.0104845408611595e-07, "loss": 0.2503, "step": 25329 }, { "epoch": 1.7511234013135155, "grad_norm": 0.728880096839228, "learning_rate": 4.008288276626782e-07, "loss": 0.3561, "step": 25330 }, { "epoch": 1.7511925337020395, "grad_norm": 0.7621757206385752, "learning_rate": 4.006092588816457e-07, "loss": 0.2719, "step": 25331 }, { "epoch": 1.7512616660905636, "grad_norm": 0.7766666788717148, "learning_rate": 4.003897477457669e-07, "loss": 0.2022, "step": 25332 }, { "epoch": 1.7513307984790876, "grad_norm": 0.8705185628382811, "learning_rate": 4.001702942577951e-07, "loss": 0.2615, "step": 25333 }, { "epoch": 1.7513999308676116, "grad_norm": 0.7054248910854175, "learning_rate": 3.999508984204792e-07, "loss": 0.3749, "step": 25334 }, { "epoch": 1.7514690632561356, "grad_norm": 0.7504377630512908, "learning_rate": 3.997315602365709e-07, "loss": 0.2101, "step": 25335 }, { "epoch": 1.7515381956446596, "grad_norm": 0.6794728551891371, "learning_rate": 3.9951227970881635e-07, "loss": 0.2143, "step": 25336 }, { "epoch": 1.7516073280331836, "grad_norm": 0.7074759478629585, "learning_rate": 3.992930568399661e-07, "loss": 0.3658, "step": 25337 }, { "epoch": 1.7516764604217077, "grad_norm": 0.7012564667089888, "learning_rate": 3.9907389163276635e-07, "loss": 0.2716, "step": 25338 }, { "epoch": 1.7517455928102317, "grad_norm": 0.7843834920102031, "learning_rate": 3.988547840899659e-07, "loss": 0.2318, "step": 25339 }, { "epoch": 1.7518147251987557, "grad_norm": 0.7292254928233012, "learning_rate": 3.9863573421430877e-07, "loss": 0.276, "step": 25340 }, { "epoch": 1.7518838575872797, "grad_norm": 0.7381330722879867, "learning_rate": 3.984167420085422e-07, "loss": 0.2995, "step": 25341 }, { "epoch": 1.7519529899758037, "grad_norm": 0.6844745759558771, "learning_rate": 3.9819780747540946e-07, "loss": 0.2617, "step": 25342 }, { "epoch": 1.7520221223643277, "grad_norm": 0.7180004728903129, "learning_rate": 3.979789306176546e-07, "loss": 0.2572, "step": 25343 }, { "epoch": 1.7520912547528518, "grad_norm": 0.8075393617007092, "learning_rate": 3.9776011143802087e-07, "loss": 0.2818, "step": 25344 }, { "epoch": 1.7521603871413758, "grad_norm": 0.670390473757793, "learning_rate": 3.9754134993925056e-07, "loss": 0.2425, "step": 25345 }, { "epoch": 1.7522295195298998, "grad_norm": 0.6707141292399371, "learning_rate": 3.9732264612408755e-07, "loss": 0.2314, "step": 25346 }, { "epoch": 1.7522986519184238, "grad_norm": 0.6671085686884383, "learning_rate": 3.9710399999527025e-07, "loss": 0.1994, "step": 25347 }, { "epoch": 1.7523677843069478, "grad_norm": 0.7054014820776956, "learning_rate": 3.9688541155554094e-07, "loss": 0.2846, "step": 25348 }, { "epoch": 1.7524369166954719, "grad_norm": 0.725113761702954, "learning_rate": 3.9666688080763737e-07, "loss": 0.2553, "step": 25349 }, { "epoch": 1.7525060490839959, "grad_norm": 0.6599731301900177, "learning_rate": 3.964484077542996e-07, "loss": 0.2405, "step": 25350 }, { "epoch": 1.75257518147252, "grad_norm": 0.7462581399579589, "learning_rate": 3.9622999239826596e-07, "loss": 0.3059, "step": 25351 }, { "epoch": 1.752644313861044, "grad_norm": 0.7444448483918739, "learning_rate": 3.960116347422738e-07, "loss": 0.2763, "step": 25352 }, { "epoch": 1.752713446249568, "grad_norm": 0.672621105158447, "learning_rate": 3.9579333478905915e-07, "loss": 0.2234, "step": 25353 }, { "epoch": 1.752782578638092, "grad_norm": 0.6231496839698065, "learning_rate": 3.955750925413593e-07, "loss": 0.2258, "step": 25354 }, { "epoch": 1.752851711026616, "grad_norm": 0.7456954521401109, "learning_rate": 3.9535690800190765e-07, "loss": 0.3199, "step": 25355 }, { "epoch": 1.75292084341514, "grad_norm": 0.6426636881400012, "learning_rate": 3.9513878117344085e-07, "loss": 0.2441, "step": 25356 }, { "epoch": 1.752989975803664, "grad_norm": 0.6413521617382925, "learning_rate": 3.9492071205869063e-07, "loss": 0.184, "step": 25357 }, { "epoch": 1.753059108192188, "grad_norm": 0.7288466739247076, "learning_rate": 3.9470270066039094e-07, "loss": 0.2968, "step": 25358 }, { "epoch": 1.753128240580712, "grad_norm": 0.7159901016556773, "learning_rate": 3.9448474698127516e-07, "loss": 0.347, "step": 25359 }, { "epoch": 1.753197372969236, "grad_norm": 0.7453599036328045, "learning_rate": 3.942668510240744e-07, "loss": 0.2949, "step": 25360 }, { "epoch": 1.75326650535776, "grad_norm": 0.6400538635475086, "learning_rate": 3.9404901279151865e-07, "loss": 0.2134, "step": 25361 }, { "epoch": 1.753335637746284, "grad_norm": 0.8738976230074462, "learning_rate": 3.938312322863397e-07, "loss": 0.3264, "step": 25362 }, { "epoch": 1.753404770134808, "grad_norm": 0.7122155174688338, "learning_rate": 3.936135095112653e-07, "loss": 0.3009, "step": 25363 }, { "epoch": 1.7534739025233321, "grad_norm": 0.6809637844248215, "learning_rate": 3.933958444690261e-07, "loss": 0.2193, "step": 25364 }, { "epoch": 1.7535430349118561, "grad_norm": 0.677899993580126, "learning_rate": 3.9317823716234714e-07, "loss": 0.2112, "step": 25365 }, { "epoch": 1.7536121673003802, "grad_norm": 1.0853825827355597, "learning_rate": 3.9296068759395956e-07, "loss": 0.391, "step": 25366 }, { "epoch": 1.7536812996889042, "grad_norm": 0.7689402866453517, "learning_rate": 3.9274319576658727e-07, "loss": 0.1824, "step": 25367 }, { "epoch": 1.7537504320774282, "grad_norm": 0.6896833178784532, "learning_rate": 3.9252576168295807e-07, "loss": 0.2247, "step": 25368 }, { "epoch": 1.7538195644659522, "grad_norm": 1.2850036007105492, "learning_rate": 3.9230838534579486e-07, "loss": 0.3105, "step": 25369 }, { "epoch": 1.7538886968544762, "grad_norm": 0.6987901568787862, "learning_rate": 3.9209106675782425e-07, "loss": 0.301, "step": 25370 }, { "epoch": 1.7539578292430003, "grad_norm": 0.6837514263624987, "learning_rate": 3.9187380592176805e-07, "loss": 0.2128, "step": 25371 }, { "epoch": 1.7540269616315243, "grad_norm": 0.646167294323014, "learning_rate": 3.9165660284035013e-07, "loss": 0.2166, "step": 25372 }, { "epoch": 1.7540960940200483, "grad_norm": 0.6809793428654559, "learning_rate": 3.9143945751629274e-07, "loss": 0.3594, "step": 25373 }, { "epoch": 1.7541652264085723, "grad_norm": 0.7299676127938983, "learning_rate": 3.9122236995231824e-07, "loss": 0.2187, "step": 25374 }, { "epoch": 1.7542343587970963, "grad_norm": 0.7772397035452477, "learning_rate": 3.91005340151146e-07, "loss": 0.1906, "step": 25375 }, { "epoch": 1.7543034911856203, "grad_norm": 0.6392659577915735, "learning_rate": 3.9078836811549726e-07, "loss": 0.235, "step": 25376 }, { "epoch": 1.7543726235741444, "grad_norm": 0.7621062128028412, "learning_rate": 3.905714538480898e-07, "loss": 0.3068, "step": 25377 }, { "epoch": 1.7544417559626684, "grad_norm": 0.7031681990105111, "learning_rate": 3.903545973516448e-07, "loss": 0.2225, "step": 25378 }, { "epoch": 1.7545108883511924, "grad_norm": 0.6920200342836148, "learning_rate": 3.9013779862887737e-07, "loss": 0.2384, "step": 25379 }, { "epoch": 1.7545800207397164, "grad_norm": 0.7729484030046166, "learning_rate": 3.8992105768250575e-07, "loss": 0.2992, "step": 25380 }, { "epoch": 1.7546491531282404, "grad_norm": 0.6971698252571218, "learning_rate": 3.897043745152468e-07, "loss": 0.2689, "step": 25381 }, { "epoch": 1.7547182855167645, "grad_norm": 0.9058851300435269, "learning_rate": 3.8948774912981713e-07, "loss": 0.2493, "step": 25382 }, { "epoch": 1.7547874179052885, "grad_norm": 0.708794435701856, "learning_rate": 3.892711815289296e-07, "loss": 0.2717, "step": 25383 }, { "epoch": 1.7548565502938125, "grad_norm": 0.84433183933745, "learning_rate": 3.8905467171530043e-07, "loss": 0.3709, "step": 25384 }, { "epoch": 1.7549256826823367, "grad_norm": 0.6263394038103218, "learning_rate": 3.8883821969164124e-07, "loss": 0.2184, "step": 25385 }, { "epoch": 1.7549948150708607, "grad_norm": 0.6684625250494731, "learning_rate": 3.8862182546066665e-07, "loss": 0.2289, "step": 25386 }, { "epoch": 1.7550639474593848, "grad_norm": 0.6853950902652084, "learning_rate": 3.884054890250877e-07, "loss": 0.2981, "step": 25387 }, { "epoch": 1.7551330798479088, "grad_norm": 0.8646702104569512, "learning_rate": 3.8818921038761724e-07, "loss": 0.2744, "step": 25388 }, { "epoch": 1.7552022122364328, "grad_norm": 0.7314668248045789, "learning_rate": 3.8797298955096374e-07, "loss": 0.2539, "step": 25389 }, { "epoch": 1.7552713446249568, "grad_norm": 0.7396068699352952, "learning_rate": 3.877568265178394e-07, "loss": 0.2516, "step": 25390 }, { "epoch": 1.7553404770134808, "grad_norm": 0.7445032944419309, "learning_rate": 3.87540721290951e-07, "loss": 0.3844, "step": 25391 }, { "epoch": 1.7554096094020049, "grad_norm": 0.6946679745870988, "learning_rate": 3.8732467387300966e-07, "loss": 0.1855, "step": 25392 }, { "epoch": 1.7554787417905289, "grad_norm": 0.632944368800806, "learning_rate": 3.87108684266721e-07, "loss": 0.1781, "step": 25393 }, { "epoch": 1.755547874179053, "grad_norm": 0.6817629862867508, "learning_rate": 3.8689275247479285e-07, "loss": 0.291, "step": 25394 }, { "epoch": 1.755617006567577, "grad_norm": 0.6927747718971546, "learning_rate": 3.866768784999314e-07, "loss": 0.2882, "step": 25395 }, { "epoch": 1.755686138956101, "grad_norm": 0.7365875220737996, "learning_rate": 3.864610623448434e-07, "loss": 0.2381, "step": 25396 }, { "epoch": 1.755755271344625, "grad_norm": 0.7205646980797971, "learning_rate": 3.8624530401223213e-07, "loss": 0.2301, "step": 25397 }, { "epoch": 1.755824403733149, "grad_norm": 0.7776023771226872, "learning_rate": 3.860296035048028e-07, "loss": 0.3483, "step": 25398 }, { "epoch": 1.755893536121673, "grad_norm": 0.92540898893533, "learning_rate": 3.858139608252581e-07, "loss": 0.2116, "step": 25399 }, { "epoch": 1.755962668510197, "grad_norm": 0.7280565742533458, "learning_rate": 3.8559837597630155e-07, "loss": 0.2852, "step": 25400 }, { "epoch": 1.756031800898721, "grad_norm": 0.7114534690803928, "learning_rate": 3.853828489606326e-07, "loss": 0.2612, "step": 25401 }, { "epoch": 1.756100933287245, "grad_norm": 0.7171580225529476, "learning_rate": 3.851673797809563e-07, "loss": 0.3224, "step": 25402 }, { "epoch": 1.756170065675769, "grad_norm": 0.763783367680792, "learning_rate": 3.849519684399705e-07, "loss": 0.2232, "step": 25403 }, { "epoch": 1.756239198064293, "grad_norm": 0.6699228602204466, "learning_rate": 3.8473661494037697e-07, "loss": 0.2437, "step": 25404 }, { "epoch": 1.756308330452817, "grad_norm": 0.9020038617455053, "learning_rate": 3.8452131928487293e-07, "loss": 0.3146, "step": 25405 }, { "epoch": 1.7563774628413413, "grad_norm": 0.7082545341821658, "learning_rate": 3.843060814761579e-07, "loss": 0.2565, "step": 25406 }, { "epoch": 1.7564465952298653, "grad_norm": 0.7036632098840399, "learning_rate": 3.840909015169286e-07, "loss": 0.2714, "step": 25407 }, { "epoch": 1.7565157276183894, "grad_norm": 0.6685372596649998, "learning_rate": 3.8387577940988185e-07, "loss": 0.2459, "step": 25408 }, { "epoch": 1.7565848600069134, "grad_norm": 0.7021793088578252, "learning_rate": 3.83660715157716e-07, "loss": 0.2957, "step": 25409 }, { "epoch": 1.7566539923954374, "grad_norm": 0.6860498159549143, "learning_rate": 3.834457087631233e-07, "loss": 0.2071, "step": 25410 }, { "epoch": 1.7567231247839614, "grad_norm": 0.7392761962543324, "learning_rate": 3.83230760228801e-07, "loss": 0.2289, "step": 25411 }, { "epoch": 1.7567922571724854, "grad_norm": 0.7534342361320335, "learning_rate": 3.8301586955744154e-07, "loss": 0.3514, "step": 25412 }, { "epoch": 1.7568613895610095, "grad_norm": 0.7474298039045527, "learning_rate": 3.828010367517393e-07, "loss": 0.2922, "step": 25413 }, { "epoch": 1.7569305219495335, "grad_norm": 0.6753030652574871, "learning_rate": 3.825862618143855e-07, "loss": 0.19, "step": 25414 }, { "epoch": 1.7569996543380575, "grad_norm": 0.6973698302687472, "learning_rate": 3.82371544748073e-07, "loss": 0.2294, "step": 25415 }, { "epoch": 1.7570687867265815, "grad_norm": 0.7434778686706643, "learning_rate": 3.8215688555549247e-07, "loss": 0.4356, "step": 25416 }, { "epoch": 1.7571379191151055, "grad_norm": 0.8073644925663127, "learning_rate": 3.8194228423933553e-07, "loss": 0.3152, "step": 25417 }, { "epoch": 1.7572070515036295, "grad_norm": 0.631778725725556, "learning_rate": 3.8172774080229015e-07, "loss": 0.2285, "step": 25418 }, { "epoch": 1.7572761838921536, "grad_norm": 0.7163197493787921, "learning_rate": 3.815132552470463e-07, "loss": 0.3192, "step": 25419 }, { "epoch": 1.7573453162806776, "grad_norm": 0.7141223820438649, "learning_rate": 3.8129882757629076e-07, "loss": 0.3204, "step": 25420 }, { "epoch": 1.7574144486692016, "grad_norm": 1.0071524238505087, "learning_rate": 3.8108445779271197e-07, "loss": 0.252, "step": 25421 }, { "epoch": 1.7574835810577256, "grad_norm": 0.7446716777287591, "learning_rate": 3.808701458989972e-07, "loss": 0.2238, "step": 25422 }, { "epoch": 1.7575527134462496, "grad_norm": 0.7601648159515784, "learning_rate": 3.8065589189783205e-07, "loss": 0.3336, "step": 25423 }, { "epoch": 1.7576218458347737, "grad_norm": 0.7449682709639501, "learning_rate": 3.804416957919016e-07, "loss": 0.2837, "step": 25424 }, { "epoch": 1.7576909782232977, "grad_norm": 0.664146272312978, "learning_rate": 3.80227557583891e-07, "loss": 0.2465, "step": 25425 }, { "epoch": 1.7577601106118217, "grad_norm": 0.6836112875672532, "learning_rate": 3.80013477276483e-07, "loss": 0.2352, "step": 25426 }, { "epoch": 1.7578292430003457, "grad_norm": 0.6775915535750321, "learning_rate": 3.797994548723616e-07, "loss": 0.2944, "step": 25427 }, { "epoch": 1.7578983753888697, "grad_norm": 0.7525822081254105, "learning_rate": 3.7958549037420854e-07, "loss": 0.2201, "step": 25428 }, { "epoch": 1.7579675077773937, "grad_norm": 0.661135697072433, "learning_rate": 3.7937158378470563e-07, "loss": 0.2265, "step": 25429 }, { "epoch": 1.7580366401659178, "grad_norm": 0.7425017516561496, "learning_rate": 3.791577351065345e-07, "loss": 0.3273, "step": 25430 }, { "epoch": 1.7581057725544418, "grad_norm": 0.7589703314563444, "learning_rate": 3.7894394434237537e-07, "loss": 0.2829, "step": 25431 }, { "epoch": 1.7581749049429658, "grad_norm": 0.7054750766392367, "learning_rate": 3.78730211494906e-07, "loss": 0.2295, "step": 25432 }, { "epoch": 1.7582440373314898, "grad_norm": 0.662103692210048, "learning_rate": 3.785165365668081e-07, "loss": 0.2807, "step": 25433 }, { "epoch": 1.7583131697200138, "grad_norm": 0.8650239533753413, "learning_rate": 3.783029195607568e-07, "loss": 0.3219, "step": 25434 }, { "epoch": 1.7583823021085379, "grad_norm": 0.7855364166801415, "learning_rate": 3.780893604794311e-07, "loss": 0.2204, "step": 25435 }, { "epoch": 1.7584514344970619, "grad_norm": 0.6902370697219176, "learning_rate": 3.778758593255066e-07, "loss": 0.1958, "step": 25436 }, { "epoch": 1.758520566885586, "grad_norm": 0.702563947315538, "learning_rate": 3.7766241610165956e-07, "loss": 0.3113, "step": 25437 }, { "epoch": 1.75858969927411, "grad_norm": 0.7254669609067145, "learning_rate": 3.774490308105655e-07, "loss": 0.2462, "step": 25438 }, { "epoch": 1.758658831662634, "grad_norm": 0.7540921254067031, "learning_rate": 3.7723570345489903e-07, "loss": 0.2373, "step": 25439 }, { "epoch": 1.758727964051158, "grad_norm": 0.6952622946643782, "learning_rate": 3.770224340373324e-07, "loss": 0.2429, "step": 25440 }, { "epoch": 1.758797096439682, "grad_norm": 0.7173865792300781, "learning_rate": 3.7680922256054076e-07, "loss": 0.3845, "step": 25441 }, { "epoch": 1.758866228828206, "grad_norm": 0.8032168300227533, "learning_rate": 3.765960690271936e-07, "loss": 0.2476, "step": 25442 }, { "epoch": 1.75893536121673, "grad_norm": 0.6273787465889479, "learning_rate": 3.7638297343996434e-07, "loss": 0.1962, "step": 25443 }, { "epoch": 1.759004493605254, "grad_norm": 0.677482925407081, "learning_rate": 3.7616993580152307e-07, "loss": 0.3115, "step": 25444 }, { "epoch": 1.759073625993778, "grad_norm": 0.7358623277404261, "learning_rate": 3.75956956114541e-07, "loss": 0.2847, "step": 25445 }, { "epoch": 1.759142758382302, "grad_norm": 0.671104704970535, "learning_rate": 3.7574403438168593e-07, "loss": 0.246, "step": 25446 }, { "epoch": 1.759211890770826, "grad_norm": 0.6517749728912838, "learning_rate": 3.755311706056275e-07, "loss": 0.2275, "step": 25447 }, { "epoch": 1.75928102315935, "grad_norm": 0.7332416415091005, "learning_rate": 3.7531836478903236e-07, "loss": 0.2742, "step": 25448 }, { "epoch": 1.759350155547874, "grad_norm": 0.7364637344662653, "learning_rate": 3.751056169345696e-07, "loss": 0.2678, "step": 25449 }, { "epoch": 1.7594192879363981, "grad_norm": 0.766081377276228, "learning_rate": 3.748929270449031e-07, "loss": 0.2021, "step": 25450 }, { "epoch": 1.7594884203249221, "grad_norm": 0.773965154885411, "learning_rate": 3.746802951227002e-07, "loss": 0.3618, "step": 25451 }, { "epoch": 1.7595575527134462, "grad_norm": 0.6828276358666512, "learning_rate": 3.7446772117062545e-07, "loss": 0.3099, "step": 25452 }, { "epoch": 1.7596266851019702, "grad_norm": 0.7302396785634975, "learning_rate": 3.7425520519134394e-07, "loss": 0.2203, "step": 25453 }, { "epoch": 1.7596958174904942, "grad_norm": 0.6106709287085769, "learning_rate": 3.740427471875174e-07, "loss": 0.2188, "step": 25454 }, { "epoch": 1.7597649498790182, "grad_norm": 0.7419425274753159, "learning_rate": 3.73830347161811e-07, "loss": 0.3351, "step": 25455 }, { "epoch": 1.7598340822675422, "grad_norm": 0.7668601938962945, "learning_rate": 3.736180051168842e-07, "loss": 0.2493, "step": 25456 }, { "epoch": 1.7599032146560663, "grad_norm": 1.0640071987818238, "learning_rate": 3.7340572105539987e-07, "loss": 0.2152, "step": 25457 }, { "epoch": 1.7599723470445903, "grad_norm": 0.7546479882927948, "learning_rate": 3.7319349498001753e-07, "loss": 0.2528, "step": 25458 }, { "epoch": 1.7600414794331143, "grad_norm": 0.7138988220049429, "learning_rate": 3.729813268933996e-07, "loss": 0.2944, "step": 25459 }, { "epoch": 1.7601106118216383, "grad_norm": 0.69321933577668, "learning_rate": 3.727692167982022e-07, "loss": 0.2138, "step": 25460 }, { "epoch": 1.7601797442101623, "grad_norm": 0.6914471567989113, "learning_rate": 3.725571646970855e-07, "loss": 0.2192, "step": 25461 }, { "epoch": 1.7602488765986863, "grad_norm": 0.7485222843746107, "learning_rate": 3.7234517059270667e-07, "loss": 0.3291, "step": 25462 }, { "epoch": 1.7603180089872104, "grad_norm": 0.7472023315800091, "learning_rate": 3.7213323448772265e-07, "loss": 0.2483, "step": 25463 }, { "epoch": 1.7603871413757344, "grad_norm": 0.6754828857015829, "learning_rate": 3.7192135638478955e-07, "loss": 0.2206, "step": 25464 }, { "epoch": 1.7604562737642584, "grad_norm": 0.6966840690360275, "learning_rate": 3.7170953628656305e-07, "loss": 0.2291, "step": 25465 }, { "epoch": 1.7605254061527824, "grad_norm": 0.8017401889892668, "learning_rate": 3.714977741956977e-07, "loss": 0.3802, "step": 25466 }, { "epoch": 1.7605945385413067, "grad_norm": 0.6735741836983506, "learning_rate": 3.7128607011484853e-07, "loss": 0.2626, "step": 25467 }, { "epoch": 1.7606636709298307, "grad_norm": 0.6282999818380722, "learning_rate": 3.710744240466674e-07, "loss": 0.2642, "step": 25468 }, { "epoch": 1.7607328033183547, "grad_norm": 0.8036890553103357, "learning_rate": 3.708628359938088e-07, "loss": 0.3694, "step": 25469 }, { "epoch": 1.7608019357068787, "grad_norm": 0.766295286125541, "learning_rate": 3.706513059589223e-07, "loss": 0.3034, "step": 25470 }, { "epoch": 1.7608710680954027, "grad_norm": 0.6632413047548203, "learning_rate": 3.7043983394466133e-07, "loss": 0.2632, "step": 25471 }, { "epoch": 1.7609402004839267, "grad_norm": 0.688112521331818, "learning_rate": 3.7022841995367374e-07, "loss": 0.2278, "step": 25472 }, { "epoch": 1.7610093328724508, "grad_norm": 0.6977670147697781, "learning_rate": 3.700170639886119e-07, "loss": 0.3481, "step": 25473 }, { "epoch": 1.7610784652609748, "grad_norm": 0.7269128076074817, "learning_rate": 3.698057660521237e-07, "loss": 0.2755, "step": 25474 }, { "epoch": 1.7611475976494988, "grad_norm": 0.6863438969659702, "learning_rate": 3.6959452614685695e-07, "loss": 0.2197, "step": 25475 }, { "epoch": 1.7612167300380228, "grad_norm": 0.7618257382588957, "learning_rate": 3.693833442754596e-07, "loss": 0.2589, "step": 25476 }, { "epoch": 1.7612858624265468, "grad_norm": 0.8716681078626304, "learning_rate": 3.6917222044057786e-07, "loss": 0.2982, "step": 25477 }, { "epoch": 1.7613549948150709, "grad_norm": 0.6554452759986419, "learning_rate": 3.6896115464485793e-07, "loss": 0.2322, "step": 25478 }, { "epoch": 1.7614241272035949, "grad_norm": 0.6828295116019805, "learning_rate": 3.68750146890946e-07, "loss": 0.2562, "step": 25479 }, { "epoch": 1.7614932595921189, "grad_norm": 0.779681425099004, "learning_rate": 3.685391971814867e-07, "loss": 0.3443, "step": 25480 }, { "epoch": 1.761562391980643, "grad_norm": 0.6795026725472525, "learning_rate": 3.6832830551912226e-07, "loss": 0.2464, "step": 25481 }, { "epoch": 1.761631524369167, "grad_norm": 0.7038442006808869, "learning_rate": 3.681174719064984e-07, "loss": 0.2422, "step": 25482 }, { "epoch": 1.761700656757691, "grad_norm": 0.6596386642338923, "learning_rate": 3.679066963462546e-07, "loss": 0.2358, "step": 25483 }, { "epoch": 1.761769789146215, "grad_norm": 0.7127746272011463, "learning_rate": 3.67695978841035e-07, "loss": 0.3522, "step": 25484 }, { "epoch": 1.761838921534739, "grad_norm": 0.6707477497384784, "learning_rate": 3.674853193934791e-07, "loss": 0.2026, "step": 25485 }, { "epoch": 1.761908053923263, "grad_norm": 0.8462766430853961, "learning_rate": 3.672747180062275e-07, "loss": 0.2244, "step": 25486 }, { "epoch": 1.7619771863117872, "grad_norm": 0.7640967370149628, "learning_rate": 3.6706417468191923e-07, "loss": 0.3136, "step": 25487 }, { "epoch": 1.7620463187003113, "grad_norm": 0.7404909641844232, "learning_rate": 3.66853689423195e-07, "loss": 0.2628, "step": 25488 }, { "epoch": 1.7621154510888353, "grad_norm": 0.7050659317919241, "learning_rate": 3.66643262232691e-07, "loss": 0.2954, "step": 25489 }, { "epoch": 1.7621845834773593, "grad_norm": 0.736648924715587, "learning_rate": 3.6643289311304564e-07, "loss": 0.2979, "step": 25490 }, { "epoch": 1.7622537158658833, "grad_norm": 0.6486985989279419, "learning_rate": 3.662225820668941e-07, "loss": 0.311, "step": 25491 }, { "epoch": 1.7623228482544073, "grad_norm": 0.7127424573560974, "learning_rate": 3.660123290968731e-07, "loss": 0.2439, "step": 25492 }, { "epoch": 1.7623919806429313, "grad_norm": 0.6944301127095568, "learning_rate": 3.658021342056184e-07, "loss": 0.2078, "step": 25493 }, { "epoch": 1.7624611130314554, "grad_norm": 0.7273257503042495, "learning_rate": 3.655919973957639e-07, "loss": 0.2881, "step": 25494 }, { "epoch": 1.7625302454199794, "grad_norm": 0.7171963922759154, "learning_rate": 3.653819186699431e-07, "loss": 0.2404, "step": 25495 }, { "epoch": 1.7625993778085034, "grad_norm": 0.7268330346128345, "learning_rate": 3.6517189803078947e-07, "loss": 0.2459, "step": 25496 }, { "epoch": 1.7626685101970274, "grad_norm": 0.7043387673680201, "learning_rate": 3.649619354809342e-07, "loss": 0.2496, "step": 25497 }, { "epoch": 1.7627376425855514, "grad_norm": 0.7239612069443124, "learning_rate": 3.6475203102301017e-07, "loss": 0.316, "step": 25498 }, { "epoch": 1.7628067749740755, "grad_norm": 0.7767581278862176, "learning_rate": 3.6454218465964643e-07, "loss": 0.3361, "step": 25499 }, { "epoch": 1.7628759073625995, "grad_norm": 0.6859104543130178, "learning_rate": 3.6433239639347475e-07, "loss": 0.2425, "step": 25500 }, { "epoch": 1.7629450397511235, "grad_norm": 0.7063837179359759, "learning_rate": 3.64122666227123e-07, "loss": 0.2639, "step": 25501 }, { "epoch": 1.7630141721396475, "grad_norm": 0.7253125865669952, "learning_rate": 3.639129941632219e-07, "loss": 0.333, "step": 25502 }, { "epoch": 1.7630833045281715, "grad_norm": 0.747793868436209, "learning_rate": 3.637033802043971e-07, "loss": 0.2384, "step": 25503 }, { "epoch": 1.7631524369166955, "grad_norm": 0.6939105251782997, "learning_rate": 3.6349382435327704e-07, "loss": 0.2757, "step": 25504 }, { "epoch": 1.7632215693052196, "grad_norm": 0.6842196118492909, "learning_rate": 3.632843266124869e-07, "loss": 0.372, "step": 25505 }, { "epoch": 1.7632907016937436, "grad_norm": 0.7789707087634258, "learning_rate": 3.6307488698465454e-07, "loss": 0.2947, "step": 25506 }, { "epoch": 1.7633598340822676, "grad_norm": 1.7793555849950753, "learning_rate": 3.628655054724023e-07, "loss": 0.2076, "step": 25507 }, { "epoch": 1.7634289664707916, "grad_norm": 0.6780604633380308, "learning_rate": 3.6265618207835527e-07, "loss": 0.273, "step": 25508 }, { "epoch": 1.7634980988593156, "grad_norm": 0.6625149914412275, "learning_rate": 3.6244691680513754e-07, "loss": 0.3204, "step": 25509 }, { "epoch": 1.7635672312478397, "grad_norm": 0.6946076631838367, "learning_rate": 3.622377096553725e-07, "loss": 0.2474, "step": 25510 }, { "epoch": 1.7636363636363637, "grad_norm": 0.8309708690316717, "learning_rate": 3.620285606316809e-07, "loss": 0.2274, "step": 25511 }, { "epoch": 1.7637054960248877, "grad_norm": 0.73413296182636, "learning_rate": 3.618194697366845e-07, "loss": 0.3275, "step": 25512 }, { "epoch": 1.7637746284134117, "grad_norm": 0.6777409439528258, "learning_rate": 3.6161043697300333e-07, "loss": 0.2889, "step": 25513 }, { "epoch": 1.7638437608019357, "grad_norm": 0.7032727403740713, "learning_rate": 3.6140146234325825e-07, "loss": 0.2324, "step": 25514 }, { "epoch": 1.7639128931904597, "grad_norm": 0.7386308139693593, "learning_rate": 3.61192545850067e-07, "loss": 0.2405, "step": 25515 }, { "epoch": 1.7639820255789838, "grad_norm": 0.6333681333305066, "learning_rate": 3.609836874960504e-07, "loss": 0.2844, "step": 25516 }, { "epoch": 1.7640511579675078, "grad_norm": 0.7590447387462087, "learning_rate": 3.607748872838235e-07, "loss": 0.2448, "step": 25517 }, { "epoch": 1.7641202903560318, "grad_norm": 0.6867133696033216, "learning_rate": 3.605661452160053e-07, "loss": 0.2016, "step": 25518 }, { "epoch": 1.7641894227445558, "grad_norm": 0.7408636194756445, "learning_rate": 3.6035746129520986e-07, "loss": 0.2992, "step": 25519 }, { "epoch": 1.7642585551330798, "grad_norm": 0.7310815364353436, "learning_rate": 3.601488355240551e-07, "loss": 0.309, "step": 25520 }, { "epoch": 1.7643276875216038, "grad_norm": 0.6767343876827598, "learning_rate": 3.599402679051533e-07, "loss": 0.2116, "step": 25521 }, { "epoch": 1.7643968199101279, "grad_norm": 0.7035122127495922, "learning_rate": 3.5973175844111975e-07, "loss": 0.2573, "step": 25522 }, { "epoch": 1.7644659522986519, "grad_norm": 0.7887970204856459, "learning_rate": 3.595233071345677e-07, "loss": 0.3189, "step": 25523 }, { "epoch": 1.764535084687176, "grad_norm": 0.6953533909945951, "learning_rate": 3.593149139881108e-07, "loss": 0.2191, "step": 25524 }, { "epoch": 1.7646042170757, "grad_norm": 0.6404062458005324, "learning_rate": 3.5910657900435854e-07, "loss": 0.2059, "step": 25525 }, { "epoch": 1.764673349464224, "grad_norm": 0.7259251028863504, "learning_rate": 3.5889830218592437e-07, "loss": 0.3386, "step": 25526 }, { "epoch": 1.764742481852748, "grad_norm": 0.6886598997255209, "learning_rate": 3.586900835354162e-07, "loss": 0.3198, "step": 25527 }, { "epoch": 1.764811614241272, "grad_norm": 0.6462819812954195, "learning_rate": 3.5848192305544594e-07, "loss": 0.215, "step": 25528 }, { "epoch": 1.764880746629796, "grad_norm": 1.0214748935583196, "learning_rate": 3.582738207486208e-07, "loss": 0.2337, "step": 25529 }, { "epoch": 1.76494987901832, "grad_norm": 0.7181900781096209, "learning_rate": 3.5806577661755103e-07, "loss": 0.3145, "step": 25530 }, { "epoch": 1.765019011406844, "grad_norm": 0.7160875415406901, "learning_rate": 3.578577906648417e-07, "loss": 0.3065, "step": 25531 }, { "epoch": 1.765088143795368, "grad_norm": 0.6825809089364941, "learning_rate": 3.576498628931019e-07, "loss": 0.2296, "step": 25532 }, { "epoch": 1.765157276183892, "grad_norm": 0.71394128303512, "learning_rate": 3.5744199330493555e-07, "loss": 0.2923, "step": 25533 }, { "epoch": 1.765226408572416, "grad_norm": 0.7324818222067316, "learning_rate": 3.5723418190294957e-07, "loss": 0.3248, "step": 25534 }, { "epoch": 1.76529554096094, "grad_norm": 0.7668434391792311, "learning_rate": 3.570264286897468e-07, "loss": 0.2394, "step": 25535 }, { "epoch": 1.7653646733494641, "grad_norm": 0.714418219122625, "learning_rate": 3.5681873366793185e-07, "loss": 0.2047, "step": 25536 }, { "epoch": 1.7654338057379881, "grad_norm": 0.7355264953565551, "learning_rate": 3.566110968401082e-07, "loss": 0.36, "step": 25537 }, { "epoch": 1.7655029381265122, "grad_norm": 0.7156305177486365, "learning_rate": 3.5640351820887874e-07, "loss": 0.2477, "step": 25538 }, { "epoch": 1.7655720705150362, "grad_norm": 0.6993357406664085, "learning_rate": 3.561959977768437e-07, "loss": 0.1968, "step": 25539 }, { "epoch": 1.7656412029035602, "grad_norm": 0.6640733495763266, "learning_rate": 3.5598853554660527e-07, "loss": 0.2378, "step": 25540 }, { "epoch": 1.7657103352920842, "grad_norm": 0.6743995560993077, "learning_rate": 3.5578113152076263e-07, "loss": 0.2989, "step": 25541 }, { "epoch": 1.7657794676806082, "grad_norm": 0.7832228256305952, "learning_rate": 3.5557378570191483e-07, "loss": 0.2499, "step": 25542 }, { "epoch": 1.7658486000691322, "grad_norm": 0.7145602991132953, "learning_rate": 3.553664980926613e-07, "loss": 0.1987, "step": 25543 }, { "epoch": 1.7659177324576563, "grad_norm": 0.6710975366808196, "learning_rate": 3.5515926869560014e-07, "loss": 0.2693, "step": 25544 }, { "epoch": 1.7659868648461803, "grad_norm": 0.6660427062071355, "learning_rate": 3.5495209751332915e-07, "loss": 0.3702, "step": 25545 }, { "epoch": 1.7660559972347043, "grad_norm": 0.7344902101212425, "learning_rate": 3.5474498454844355e-07, "loss": 0.2512, "step": 25546 }, { "epoch": 1.7661251296232283, "grad_norm": 0.69027812478332, "learning_rate": 3.545379298035401e-07, "loss": 0.2445, "step": 25547 }, { "epoch": 1.7661942620117526, "grad_norm": 0.7414937096501227, "learning_rate": 3.5433093328121337e-07, "loss": 0.3115, "step": 25548 }, { "epoch": 1.7662633944002766, "grad_norm": 0.7630622835037851, "learning_rate": 3.5412399498405745e-07, "loss": 0.2211, "step": 25549 }, { "epoch": 1.7663325267888006, "grad_norm": 0.8129545262809983, "learning_rate": 3.539171149146658e-07, "loss": 0.2292, "step": 25550 }, { "epoch": 1.7664016591773246, "grad_norm": 0.6897103494448463, "learning_rate": 3.5371029307563365e-07, "loss": 0.3096, "step": 25551 }, { "epoch": 1.7664707915658486, "grad_norm": 0.7263187666557093, "learning_rate": 3.5350352946954993e-07, "loss": 0.3297, "step": 25552 }, { "epoch": 1.7665399239543726, "grad_norm": 0.7010565013736654, "learning_rate": 3.5329682409900813e-07, "loss": 0.2443, "step": 25553 }, { "epoch": 1.7666090563428967, "grad_norm": 0.6588992384258966, "learning_rate": 3.530901769665973e-07, "loss": 0.2168, "step": 25554 }, { "epoch": 1.7666781887314207, "grad_norm": 0.7270605651565936, "learning_rate": 3.528835880749093e-07, "loss": 0.3807, "step": 25555 }, { "epoch": 1.7667473211199447, "grad_norm": 0.707633207812354, "learning_rate": 3.5267705742653204e-07, "loss": 0.2795, "step": 25556 }, { "epoch": 1.7668164535084687, "grad_norm": 0.6611907669824613, "learning_rate": 3.5247058502405397e-07, "loss": 0.1958, "step": 25557 }, { "epoch": 1.7668855858969927, "grad_norm": 0.6713909578116529, "learning_rate": 3.52264170870063e-07, "loss": 0.2433, "step": 25558 }, { "epoch": 1.7669547182855168, "grad_norm": 0.6543167500319715, "learning_rate": 3.520578149671472e-07, "loss": 0.3057, "step": 25559 }, { "epoch": 1.7670238506740408, "grad_norm": 0.7424356477057914, "learning_rate": 3.518515173178916e-07, "loss": 0.2806, "step": 25560 }, { "epoch": 1.7670929830625648, "grad_norm": 0.7400234391753505, "learning_rate": 3.51645277924883e-07, "loss": 0.2459, "step": 25561 }, { "epoch": 1.7671621154510888, "grad_norm": 0.8981766624747723, "learning_rate": 3.5143909679070444e-07, "loss": 0.3267, "step": 25562 }, { "epoch": 1.7672312478396128, "grad_norm": 0.6847284793558999, "learning_rate": 3.51232973917941e-07, "loss": 0.2566, "step": 25563 }, { "epoch": 1.7673003802281368, "grad_norm": 0.9566334831933337, "learning_rate": 3.510269093091767e-07, "loss": 0.2186, "step": 25564 }, { "epoch": 1.7673695126166609, "grad_norm": 0.6595878590359469, "learning_rate": 3.50820902966994e-07, "loss": 0.2386, "step": 25565 }, { "epoch": 1.7674386450051849, "grad_norm": 0.6436655454414534, "learning_rate": 3.5061495489397357e-07, "loss": 0.3344, "step": 25566 }, { "epoch": 1.767507777393709, "grad_norm": 0.7066363337741535, "learning_rate": 3.5040906509269835e-07, "loss": 0.2372, "step": 25567 }, { "epoch": 1.7675769097822331, "grad_norm": 0.6790846568733723, "learning_rate": 3.5020323356574736e-07, "loss": 0.2379, "step": 25568 }, { "epoch": 1.7676460421707572, "grad_norm": 0.709197600836698, "learning_rate": 3.499974603157014e-07, "loss": 0.3654, "step": 25569 }, { "epoch": 1.7677151745592812, "grad_norm": 0.7001332797053594, "learning_rate": 3.497917453451383e-07, "loss": 0.3635, "step": 25570 }, { "epoch": 1.7677843069478052, "grad_norm": 0.7380502856663331, "learning_rate": 3.4958608865663714e-07, "loss": 0.1868, "step": 25571 }, { "epoch": 1.7678534393363292, "grad_norm": 0.7538660943861437, "learning_rate": 3.4938049025277533e-07, "loss": 0.2344, "step": 25572 }, { "epoch": 1.7679225717248532, "grad_norm": 0.6781926892843757, "learning_rate": 3.491749501361302e-07, "loss": 0.283, "step": 25573 }, { "epoch": 1.7679917041133772, "grad_norm": 0.8456127303881733, "learning_rate": 3.4896946830927646e-07, "loss": 0.253, "step": 25574 }, { "epoch": 1.7680608365019013, "grad_norm": 0.6994130275199222, "learning_rate": 3.487640447747914e-07, "loss": 0.2283, "step": 25575 }, { "epoch": 1.7681299688904253, "grad_norm": 0.7204669448842643, "learning_rate": 3.4855867953524744e-07, "loss": 0.3216, "step": 25576 }, { "epoch": 1.7681991012789493, "grad_norm": 0.7311787504772246, "learning_rate": 3.4835337259322087e-07, "loss": 0.2932, "step": 25577 }, { "epoch": 1.7682682336674733, "grad_norm": 0.7081410606259325, "learning_rate": 3.481481239512824e-07, "loss": 0.2268, "step": 25578 }, { "epoch": 1.7683373660559973, "grad_norm": 0.6414474893491636, "learning_rate": 3.479429336120055e-07, "loss": 0.2326, "step": 25579 }, { "epoch": 1.7684064984445214, "grad_norm": 1.209741068517857, "learning_rate": 3.47737801577962e-07, "loss": 0.3356, "step": 25580 }, { "epoch": 1.7684756308330454, "grad_norm": 0.7124448184563648, "learning_rate": 3.4753272785172374e-07, "loss": 0.2296, "step": 25581 }, { "epoch": 1.7685447632215694, "grad_norm": 0.6783960102883705, "learning_rate": 3.4732771243585926e-07, "loss": 0.2343, "step": 25582 }, { "epoch": 1.7686138956100934, "grad_norm": 0.6640428223755251, "learning_rate": 3.4712275533293926e-07, "loss": 0.2302, "step": 25583 }, { "epoch": 1.7686830279986174, "grad_norm": 0.7087879319203072, "learning_rate": 3.4691785654553167e-07, "loss": 0.3327, "step": 25584 }, { "epoch": 1.7687521603871414, "grad_norm": 0.6696448758533745, "learning_rate": 3.46713016076205e-07, "loss": 0.2462, "step": 25585 }, { "epoch": 1.7688212927756655, "grad_norm": 0.6579859054627542, "learning_rate": 3.4650823392752606e-07, "loss": 0.2359, "step": 25586 }, { "epoch": 1.7688904251641895, "grad_norm": 0.6780645825058517, "learning_rate": 3.4630351010206344e-07, "loss": 0.3198, "step": 25587 }, { "epoch": 1.7689595575527135, "grad_norm": 0.7317427407878758, "learning_rate": 3.4609884460237997e-07, "loss": 0.2645, "step": 25588 }, { "epoch": 1.7690286899412375, "grad_norm": 0.8014078129960768, "learning_rate": 3.458942374310431e-07, "loss": 0.2481, "step": 25589 }, { "epoch": 1.7690978223297615, "grad_norm": 0.7469832547464416, "learning_rate": 3.456896885906158e-07, "loss": 0.2466, "step": 25590 }, { "epoch": 1.7691669547182856, "grad_norm": 0.678964023123759, "learning_rate": 3.4548519808366267e-07, "loss": 0.3776, "step": 25591 }, { "epoch": 1.7692360871068096, "grad_norm": 0.7035356126940728, "learning_rate": 3.452807659127461e-07, "loss": 0.2738, "step": 25592 }, { "epoch": 1.7693052194953336, "grad_norm": 0.7261335783180333, "learning_rate": 3.4507639208042785e-07, "loss": 0.2469, "step": 25593 }, { "epoch": 1.7693743518838576, "grad_norm": 0.7003141003321688, "learning_rate": 3.448720765892699e-07, "loss": 0.3237, "step": 25594 }, { "epoch": 1.7694434842723816, "grad_norm": 0.7626796667700773, "learning_rate": 3.4466781944183404e-07, "loss": 0.3057, "step": 25595 }, { "epoch": 1.7695126166609056, "grad_norm": 0.6661425135306943, "learning_rate": 3.4446362064067873e-07, "loss": 0.1986, "step": 25596 }, { "epoch": 1.7695817490494297, "grad_norm": 0.6928832614530763, "learning_rate": 3.442594801883642e-07, "loss": 0.211, "step": 25597 }, { "epoch": 1.7696508814379537, "grad_norm": 0.6896456409072814, "learning_rate": 3.440553980874478e-07, "loss": 0.3516, "step": 25598 }, { "epoch": 1.7697200138264777, "grad_norm": 0.8421915816562228, "learning_rate": 3.438513743404881e-07, "loss": 0.2544, "step": 25599 }, { "epoch": 1.7697891462150017, "grad_norm": 0.6785178443915426, "learning_rate": 3.436474089500419e-07, "loss": 0.2062, "step": 25600 }, { "epoch": 1.7698582786035257, "grad_norm": 0.7122902241115646, "learning_rate": 3.434435019186666e-07, "loss": 0.2636, "step": 25601 }, { "epoch": 1.7699274109920498, "grad_norm": 0.7846590773744193, "learning_rate": 3.4323965324891684e-07, "loss": 0.3083, "step": 25602 }, { "epoch": 1.7699965433805738, "grad_norm": 0.9130588782398946, "learning_rate": 3.430358629433478e-07, "loss": 0.227, "step": 25603 }, { "epoch": 1.7700656757690978, "grad_norm": 0.711603939223691, "learning_rate": 3.4283213100451295e-07, "loss": 0.2775, "step": 25604 }, { "epoch": 1.7701348081576218, "grad_norm": 0.7537744028401735, "learning_rate": 3.42628457434967e-07, "loss": 0.3528, "step": 25605 }, { "epoch": 1.7702039405461458, "grad_norm": 0.765054643391007, "learning_rate": 3.4242484223726113e-07, "loss": 0.2431, "step": 25606 }, { "epoch": 1.7702730729346698, "grad_norm": 0.8111447775661571, "learning_rate": 3.422212854139484e-07, "loss": 0.1969, "step": 25607 }, { "epoch": 1.7703422053231939, "grad_norm": 0.7202436612890191, "learning_rate": 3.4201778696758003e-07, "loss": 0.2954, "step": 25608 }, { "epoch": 1.7704113377117179, "grad_norm": 1.068509817785729, "learning_rate": 3.4181434690070514e-07, "loss": 0.2993, "step": 25609 }, { "epoch": 1.770480470100242, "grad_norm": 0.7198482816844464, "learning_rate": 3.4161096521587556e-07, "loss": 0.2628, "step": 25610 }, { "epoch": 1.770549602488766, "grad_norm": 0.7173508243593217, "learning_rate": 3.414076419156387e-07, "loss": 0.2329, "step": 25611 }, { "epoch": 1.77061873487729, "grad_norm": 0.7603478980004427, "learning_rate": 3.412043770025442e-07, "loss": 0.3353, "step": 25612 }, { "epoch": 1.770687867265814, "grad_norm": 0.8031151528556615, "learning_rate": 3.4100117047913774e-07, "loss": 0.2722, "step": 25613 }, { "epoch": 1.770756999654338, "grad_norm": 0.6781264368034886, "learning_rate": 3.4079802234796733e-07, "loss": 0.2237, "step": 25614 }, { "epoch": 1.770826132042862, "grad_norm": 0.7990084196227829, "learning_rate": 3.405949326115793e-07, "loss": 0.2566, "step": 25615 }, { "epoch": 1.770895264431386, "grad_norm": 0.7426621994165032, "learning_rate": 3.403919012725188e-07, "loss": 0.3052, "step": 25616 }, { "epoch": 1.77096439681991, "grad_norm": 0.636153647514001, "learning_rate": 3.401889283333304e-07, "loss": 0.2105, "step": 25617 }, { "epoch": 1.771033529208434, "grad_norm": 0.6407113935258743, "learning_rate": 3.3998601379655826e-07, "loss": 0.2167, "step": 25618 }, { "epoch": 1.771102661596958, "grad_norm": 0.6591601983402727, "learning_rate": 3.397831576647448e-07, "loss": 0.2914, "step": 25619 }, { "epoch": 1.771171793985482, "grad_norm": 0.9783138029492899, "learning_rate": 3.3958035994043236e-07, "loss": 0.2976, "step": 25620 }, { "epoch": 1.771240926374006, "grad_norm": 0.6806397150058636, "learning_rate": 3.39377620626164e-07, "loss": 0.3094, "step": 25621 }, { "epoch": 1.7713100587625301, "grad_norm": 0.7181922139954966, "learning_rate": 3.391749397244798e-07, "loss": 0.1968, "step": 25622 }, { "epoch": 1.7713791911510541, "grad_norm": 0.6869413218745284, "learning_rate": 3.3897231723792e-07, "loss": 0.3221, "step": 25623 }, { "epoch": 1.7714483235395782, "grad_norm": 0.7991475844287109, "learning_rate": 3.3876975316902485e-07, "loss": 0.2075, "step": 25624 }, { "epoch": 1.7715174559281022, "grad_norm": 0.6274228892903786, "learning_rate": 3.3856724752033166e-07, "loss": 0.2054, "step": 25625 }, { "epoch": 1.7715865883166262, "grad_norm": 0.8794456817214099, "learning_rate": 3.383648002943801e-07, "loss": 0.2649, "step": 25626 }, { "epoch": 1.7716557207051502, "grad_norm": 0.7306880039597025, "learning_rate": 3.38162411493706e-07, "loss": 0.2782, "step": 25627 }, { "epoch": 1.7717248530936742, "grad_norm": 0.8003358600358553, "learning_rate": 3.379600811208467e-07, "loss": 0.238, "step": 25628 }, { "epoch": 1.7717939854821985, "grad_norm": 0.8621290186904284, "learning_rate": 3.377578091783379e-07, "loss": 0.2337, "step": 25629 }, { "epoch": 1.7718631178707225, "grad_norm": 0.7137226882158059, "learning_rate": 3.3755559566871543e-07, "loss": 0.3209, "step": 25630 }, { "epoch": 1.7719322502592465, "grad_norm": 0.8348084678725751, "learning_rate": 3.373534405945128e-07, "loss": 0.2521, "step": 25631 }, { "epoch": 1.7720013826477705, "grad_norm": 0.8429561367333206, "learning_rate": 3.371513439582641e-07, "loss": 0.2032, "step": 25632 }, { "epoch": 1.7720705150362945, "grad_norm": 0.7581010808849948, "learning_rate": 3.3694930576250174e-07, "loss": 0.2908, "step": 25633 }, { "epoch": 1.7721396474248186, "grad_norm": 0.8108521261509367, "learning_rate": 3.367473260097581e-07, "loss": 0.2988, "step": 25634 }, { "epoch": 1.7722087798133426, "grad_norm": 0.6919912860929893, "learning_rate": 3.365454047025646e-07, "loss": 0.2214, "step": 25635 }, { "epoch": 1.7722779122018666, "grad_norm": 0.624609009692131, "learning_rate": 3.3634354184345295e-07, "loss": 0.2049, "step": 25636 }, { "epoch": 1.7723470445903906, "grad_norm": 0.7383387013997763, "learning_rate": 3.3614173743495183e-07, "loss": 0.3238, "step": 25637 }, { "epoch": 1.7724161769789146, "grad_norm": 0.7405436316760599, "learning_rate": 3.359399914795919e-07, "loss": 0.3082, "step": 25638 }, { "epoch": 1.7724853093674386, "grad_norm": 0.7268984488949117, "learning_rate": 3.357383039798995e-07, "loss": 0.2259, "step": 25639 }, { "epoch": 1.7725544417559627, "grad_norm": 0.72076210661569, "learning_rate": 3.3553667493840483e-07, "loss": 0.294, "step": 25640 }, { "epoch": 1.7726235741444867, "grad_norm": 0.6846732483703621, "learning_rate": 3.3533510435763304e-07, "loss": 0.3558, "step": 25641 }, { "epoch": 1.7726927065330107, "grad_norm": 0.8120457091728348, "learning_rate": 3.351335922401111e-07, "loss": 0.2717, "step": 25642 }, { "epoch": 1.7727618389215347, "grad_norm": 0.6585694738651361, "learning_rate": 3.349321385883647e-07, "loss": 0.1954, "step": 25643 }, { "epoch": 1.7728309713100587, "grad_norm": 0.7777085014938057, "learning_rate": 3.3473074340491963e-07, "loss": 0.3093, "step": 25644 }, { "epoch": 1.7729001036985828, "grad_norm": 0.721371134083659, "learning_rate": 3.3452940669229886e-07, "loss": 0.299, "step": 25645 }, { "epoch": 1.7729692360871068, "grad_norm": 0.7471493784885664, "learning_rate": 3.3432812845302587e-07, "loss": 0.2564, "step": 25646 }, { "epoch": 1.7730383684756308, "grad_norm": 0.6911222951559867, "learning_rate": 3.3412690868962373e-07, "loss": 0.2513, "step": 25647 }, { "epoch": 1.7731075008641548, "grad_norm": 0.6644123632747723, "learning_rate": 3.3392574740461425e-07, "loss": 0.3346, "step": 25648 }, { "epoch": 1.773176633252679, "grad_norm": 0.7211265074385319, "learning_rate": 3.337246446005182e-07, "loss": 0.2823, "step": 25649 }, { "epoch": 1.773245765641203, "grad_norm": 0.6627741998119924, "learning_rate": 3.3352360027985584e-07, "loss": 0.2047, "step": 25650 }, { "epoch": 1.773314898029727, "grad_norm": 0.7338994932364445, "learning_rate": 3.333226144451479e-07, "loss": 0.3383, "step": 25651 }, { "epoch": 1.773384030418251, "grad_norm": 0.7339311394797462, "learning_rate": 3.331216870989135e-07, "loss": 0.2838, "step": 25652 }, { "epoch": 1.7734531628067751, "grad_norm": 0.7215977320157476, "learning_rate": 3.3292081824366993e-07, "loss": 0.2064, "step": 25653 }, { "epoch": 1.7735222951952991, "grad_norm": 0.6899839248614262, "learning_rate": 3.3272000788193537e-07, "loss": 0.2078, "step": 25654 }, { "epoch": 1.7735914275838232, "grad_norm": 0.6801048014332927, "learning_rate": 3.3251925601622604e-07, "loss": 0.2993, "step": 25655 }, { "epoch": 1.7736605599723472, "grad_norm": 0.772309206106187, "learning_rate": 3.323185626490577e-07, "loss": 0.2581, "step": 25656 }, { "epoch": 1.7737296923608712, "grad_norm": 0.6945425999070874, "learning_rate": 3.3211792778294726e-07, "loss": 0.192, "step": 25657 }, { "epoch": 1.7737988247493952, "grad_norm": 0.8036585362376831, "learning_rate": 3.319173514204083e-07, "loss": 0.3206, "step": 25658 }, { "epoch": 1.7738679571379192, "grad_norm": 0.7509033961269402, "learning_rate": 3.317168335639548e-07, "loss": 0.3176, "step": 25659 }, { "epoch": 1.7739370895264432, "grad_norm": 0.6954214752446192, "learning_rate": 3.3151637421609983e-07, "loss": 0.1952, "step": 25660 }, { "epoch": 1.7740062219149673, "grad_norm": 0.7096432242699182, "learning_rate": 3.313159733793553e-07, "loss": 0.2392, "step": 25661 }, { "epoch": 1.7740753543034913, "grad_norm": 0.7994419622293812, "learning_rate": 3.3111563105623467e-07, "loss": 0.3387, "step": 25662 }, { "epoch": 1.7741444866920153, "grad_norm": 0.7467203605491339, "learning_rate": 3.3091534724924655e-07, "loss": 0.3048, "step": 25663 }, { "epoch": 1.7742136190805393, "grad_norm": 0.7774714267160573, "learning_rate": 3.307151219609017e-07, "loss": 0.262, "step": 25664 }, { "epoch": 1.7742827514690633, "grad_norm": 0.6874220814029162, "learning_rate": 3.305149551937109e-07, "loss": 0.2677, "step": 25665 }, { "epoch": 1.7743518838575874, "grad_norm": 0.687217344480402, "learning_rate": 3.3031484695018213e-07, "loss": 0.394, "step": 25666 }, { "epoch": 1.7744210162461114, "grad_norm": 0.7784302691127876, "learning_rate": 3.301147972328228e-07, "loss": 0.2864, "step": 25667 }, { "epoch": 1.7744901486346354, "grad_norm": 0.7251378584959568, "learning_rate": 3.299148060441415e-07, "loss": 0.2401, "step": 25668 }, { "epoch": 1.7745592810231594, "grad_norm": 0.7481536361179194, "learning_rate": 3.2971487338664344e-07, "loss": 0.3065, "step": 25669 }, { "epoch": 1.7746284134116834, "grad_norm": 0.6764587327990043, "learning_rate": 3.295149992628349e-07, "loss": 0.2418, "step": 25670 }, { "epoch": 1.7746975458002074, "grad_norm": 0.6855381475114647, "learning_rate": 3.2931518367522066e-07, "loss": 0.2176, "step": 25671 }, { "epoch": 1.7747666781887315, "grad_norm": 0.6586433774924867, "learning_rate": 3.291154266263064e-07, "loss": 0.2536, "step": 25672 }, { "epoch": 1.7748358105772555, "grad_norm": 0.6864465495735417, "learning_rate": 3.2891572811859454e-07, "loss": 0.3372, "step": 25673 }, { "epoch": 1.7749049429657795, "grad_norm": 0.7204352437155911, "learning_rate": 3.2871608815458753e-07, "loss": 0.2422, "step": 25674 }, { "epoch": 1.7749740753543035, "grad_norm": 0.6703068115240195, "learning_rate": 3.285165067367885e-07, "loss": 0.2372, "step": 25675 }, { "epoch": 1.7750432077428275, "grad_norm": 0.9108988492875079, "learning_rate": 3.2831698386769804e-07, "loss": 0.3009, "step": 25676 }, { "epoch": 1.7751123401313516, "grad_norm": 0.6655940337785248, "learning_rate": 3.281175195498171e-07, "loss": 0.3428, "step": 25677 }, { "epoch": 1.7751814725198756, "grad_norm": 0.6701564676487685, "learning_rate": 3.2791811378564574e-07, "loss": 0.2339, "step": 25678 }, { "epoch": 1.7752506049083996, "grad_norm": 0.7165706279302844, "learning_rate": 3.2771876657768377e-07, "loss": 0.2521, "step": 25679 }, { "epoch": 1.7753197372969236, "grad_norm": 0.7153213743248354, "learning_rate": 3.2751947792842796e-07, "loss": 0.3359, "step": 25680 }, { "epoch": 1.7753888696854476, "grad_norm": 0.8341861057713451, "learning_rate": 3.2732024784037807e-07, "loss": 0.2787, "step": 25681 }, { "epoch": 1.7754580020739716, "grad_norm": 0.673830637197102, "learning_rate": 3.2712107631602986e-07, "loss": 0.2367, "step": 25682 }, { "epoch": 1.7755271344624957, "grad_norm": 0.7339778879185928, "learning_rate": 3.269219633578802e-07, "loss": 0.2468, "step": 25683 }, { "epoch": 1.7755962668510197, "grad_norm": 0.7586423291481219, "learning_rate": 3.2672290896842316e-07, "loss": 0.3349, "step": 25684 }, { "epoch": 1.7756653992395437, "grad_norm": 0.7249437696675622, "learning_rate": 3.2652391315015464e-07, "loss": 0.1956, "step": 25685 }, { "epoch": 1.7757345316280677, "grad_norm": 0.6989799573907415, "learning_rate": 3.2632497590556924e-07, "loss": 0.2928, "step": 25686 }, { "epoch": 1.7758036640165917, "grad_norm": 0.6940498008683186, "learning_rate": 3.261260972371599e-07, "loss": 0.2886, "step": 25687 }, { "epoch": 1.7758727964051157, "grad_norm": 0.8791643005459063, "learning_rate": 3.2592727714741865e-07, "loss": 0.2926, "step": 25688 }, { "epoch": 1.7759419287936398, "grad_norm": 0.8059445855414109, "learning_rate": 3.2572851563883845e-07, "loss": 0.2682, "step": 25689 }, { "epoch": 1.7760110611821638, "grad_norm": 0.6485246781366866, "learning_rate": 3.2552981271390894e-07, "loss": 0.2071, "step": 25690 }, { "epoch": 1.7760801935706878, "grad_norm": 0.7197838102822519, "learning_rate": 3.253311683751209e-07, "loss": 0.3287, "step": 25691 }, { "epoch": 1.7761493259592118, "grad_norm": 0.7152105584270264, "learning_rate": 3.251325826249646e-07, "loss": 0.2528, "step": 25692 }, { "epoch": 1.7762184583477358, "grad_norm": 0.6739721680147047, "learning_rate": 3.249340554659297e-07, "loss": 0.1927, "step": 25693 }, { "epoch": 1.7762875907362599, "grad_norm": 0.7469550550301053, "learning_rate": 3.24735586900502e-07, "loss": 0.2876, "step": 25694 }, { "epoch": 1.7763567231247839, "grad_norm": 0.7036353559525274, "learning_rate": 3.245371769311717e-07, "loss": 0.2484, "step": 25695 }, { "epoch": 1.776425855513308, "grad_norm": 0.7001666849339747, "learning_rate": 3.2433882556042297e-07, "loss": 0.2423, "step": 25696 }, { "epoch": 1.776494987901832, "grad_norm": 0.7190494885145535, "learning_rate": 3.241405327907443e-07, "loss": 0.261, "step": 25697 }, { "epoch": 1.776564120290356, "grad_norm": 0.7205436754113254, "learning_rate": 3.2394229862461824e-07, "loss": 0.397, "step": 25698 }, { "epoch": 1.77663325267888, "grad_norm": 0.7312679841118334, "learning_rate": 3.237441230645311e-07, "loss": 0.2462, "step": 25699 }, { "epoch": 1.776702385067404, "grad_norm": 0.7400334857417273, "learning_rate": 3.2354600611296585e-07, "loss": 0.2239, "step": 25700 }, { "epoch": 1.776771517455928, "grad_norm": 0.7026990175269892, "learning_rate": 3.2334794777240666e-07, "loss": 0.2458, "step": 25701 }, { "epoch": 1.776840649844452, "grad_norm": 0.7523896712811525, "learning_rate": 3.2314994804533485e-07, "loss": 0.3641, "step": 25702 }, { "epoch": 1.776909782232976, "grad_norm": 0.6880152238779125, "learning_rate": 3.2295200693423233e-07, "loss": 0.2054, "step": 25703 }, { "epoch": 1.7769789146215, "grad_norm": 0.7275646675433662, "learning_rate": 3.227541244415794e-07, "loss": 0.2517, "step": 25704 }, { "epoch": 1.777048047010024, "grad_norm": 0.6933909405980506, "learning_rate": 3.2255630056985674e-07, "loss": 0.3366, "step": 25705 }, { "epoch": 1.777117179398548, "grad_norm": 0.7714836421630673, "learning_rate": 3.223585353215436e-07, "loss": 0.2654, "step": 25706 }, { "epoch": 1.777186311787072, "grad_norm": 0.6857728434030763, "learning_rate": 3.2216082869911903e-07, "loss": 0.2432, "step": 25707 }, { "epoch": 1.777255444175596, "grad_norm": 0.7727326296219204, "learning_rate": 3.2196318070505996e-07, "loss": 0.26, "step": 25708 }, { "epoch": 1.7773245765641201, "grad_norm": 1.4687348297992218, "learning_rate": 3.217655913418449e-07, "loss": 0.287, "step": 25709 }, { "epoch": 1.7773937089526444, "grad_norm": 0.6833730038744419, "learning_rate": 3.215680606119487e-07, "loss": 0.2092, "step": 25710 }, { "epoch": 1.7774628413411684, "grad_norm": 0.7304954415153312, "learning_rate": 3.213705885178481e-07, "loss": 0.2116, "step": 25711 }, { "epoch": 1.7775319737296924, "grad_norm": 0.7381373520412592, "learning_rate": 3.211731750620173e-07, "loss": 0.3254, "step": 25712 }, { "epoch": 1.7776011061182164, "grad_norm": 1.931167936140816, "learning_rate": 3.2097582024693153e-07, "loss": 0.3036, "step": 25713 }, { "epoch": 1.7776702385067404, "grad_norm": 0.703962887015141, "learning_rate": 3.2077852407506325e-07, "loss": 0.2339, "step": 25714 }, { "epoch": 1.7777393708952645, "grad_norm": 0.6859051464203714, "learning_rate": 3.2058128654888665e-07, "loss": 0.2363, "step": 25715 }, { "epoch": 1.7778085032837885, "grad_norm": 0.8243508017669035, "learning_rate": 3.203841076708719e-07, "loss": 0.3045, "step": 25716 }, { "epoch": 1.7778776356723125, "grad_norm": 0.683754703617189, "learning_rate": 3.201869874434921e-07, "loss": 0.2162, "step": 25717 }, { "epoch": 1.7779467680608365, "grad_norm": 0.6815604702730347, "learning_rate": 3.199899258692157e-07, "loss": 0.252, "step": 25718 }, { "epoch": 1.7780159004493605, "grad_norm": 0.8166747232599934, "learning_rate": 3.197929229505153e-07, "loss": 0.3166, "step": 25719 }, { "epoch": 1.7780850328378845, "grad_norm": 0.7704294137532768, "learning_rate": 3.1959597868985725e-07, "loss": 0.301, "step": 25720 }, { "epoch": 1.7781541652264086, "grad_norm": 0.7374655624563182, "learning_rate": 3.1939909308971064e-07, "loss": 0.1757, "step": 25721 }, { "epoch": 1.7782232976149326, "grad_norm": 0.6955863200921193, "learning_rate": 3.1920226615254403e-07, "loss": 0.1867, "step": 25722 }, { "epoch": 1.7782924300034566, "grad_norm": 0.7790223428282079, "learning_rate": 3.1900549788082434e-07, "loss": 0.3756, "step": 25723 }, { "epoch": 1.7783615623919806, "grad_norm": 0.7550457949046004, "learning_rate": 3.1880878827701633e-07, "loss": 0.2681, "step": 25724 }, { "epoch": 1.7784306947805046, "grad_norm": 0.7241905040187594, "learning_rate": 3.186121373435869e-07, "loss": 0.2555, "step": 25725 }, { "epoch": 1.7784998271690287, "grad_norm": 0.7043433552691211, "learning_rate": 3.1841554508299954e-07, "loss": 0.2893, "step": 25726 }, { "epoch": 1.7785689595575527, "grad_norm": 0.7586002808616517, "learning_rate": 3.182190114977185e-07, "loss": 0.3347, "step": 25727 }, { "epoch": 1.7786380919460767, "grad_norm": 1.0959772423700593, "learning_rate": 3.1802253659020734e-07, "loss": 0.2925, "step": 25728 }, { "epoch": 1.7787072243346007, "grad_norm": 0.6658265742681421, "learning_rate": 3.1782612036292846e-07, "loss": 0.2336, "step": 25729 }, { "epoch": 1.7787763567231247, "grad_norm": 0.6995522076487752, "learning_rate": 3.1762976281834336e-07, "loss": 0.3381, "step": 25730 }, { "epoch": 1.778845489111649, "grad_norm": 0.7106325021903804, "learning_rate": 3.1743346395891384e-07, "loss": 0.2261, "step": 25731 }, { "epoch": 1.778914621500173, "grad_norm": 0.6494550685831109, "learning_rate": 3.172372237870985e-07, "loss": 0.242, "step": 25732 }, { "epoch": 1.778983753888697, "grad_norm": 0.6946123222875112, "learning_rate": 3.1704104230535824e-07, "loss": 0.3057, "step": 25733 }, { "epoch": 1.779052886277221, "grad_norm": 0.6992456027907202, "learning_rate": 3.1684491951615104e-07, "loss": 0.329, "step": 25734 }, { "epoch": 1.779122018665745, "grad_norm": 0.7438677462223441, "learning_rate": 3.166488554219349e-07, "loss": 0.2581, "step": 25735 }, { "epoch": 1.779191151054269, "grad_norm": 0.6775545671408689, "learning_rate": 3.164528500251679e-07, "loss": 0.202, "step": 25736 }, { "epoch": 1.779260283442793, "grad_norm": 0.7349122814740254, "learning_rate": 3.1625690332830694e-07, "loss": 0.3897, "step": 25737 }, { "epoch": 1.779329415831317, "grad_norm": 0.7370219361511998, "learning_rate": 3.1606101533380675e-07, "loss": 0.3031, "step": 25738 }, { "epoch": 1.7793985482198411, "grad_norm": 0.6552835907872825, "learning_rate": 3.158651860441231e-07, "loss": 0.2477, "step": 25739 }, { "epoch": 1.7794676806083651, "grad_norm": 0.6861964525696229, "learning_rate": 3.1566941546171016e-07, "loss": 0.2706, "step": 25740 }, { "epoch": 1.7795368129968892, "grad_norm": 0.6608827290959091, "learning_rate": 3.154737035890204e-07, "loss": 0.3092, "step": 25741 }, { "epoch": 1.7796059453854132, "grad_norm": 0.7087775557939103, "learning_rate": 3.152780504285091e-07, "loss": 0.2494, "step": 25742 }, { "epoch": 1.7796750777739372, "grad_norm": 0.6774770630256032, "learning_rate": 3.1508245598262656e-07, "loss": 0.2528, "step": 25743 }, { "epoch": 1.7797442101624612, "grad_norm": 0.7085270000237254, "learning_rate": 3.1488692025382516e-07, "loss": 0.3773, "step": 25744 }, { "epoch": 1.7798133425509852, "grad_norm": 0.6787044658295994, "learning_rate": 3.1469144324455526e-07, "loss": 0.2898, "step": 25745 }, { "epoch": 1.7798824749395092, "grad_norm": 0.7169103226295398, "learning_rate": 3.144960249572671e-07, "loss": 0.1914, "step": 25746 }, { "epoch": 1.7799516073280333, "grad_norm": 0.647640432315654, "learning_rate": 3.1430066539440864e-07, "loss": 0.2285, "step": 25747 }, { "epoch": 1.7800207397165573, "grad_norm": 0.6987775400624846, "learning_rate": 3.141053645584297e-07, "loss": 0.3506, "step": 25748 }, { "epoch": 1.7800898721050813, "grad_norm": 0.7441504642114373, "learning_rate": 3.139101224517777e-07, "loss": 0.2203, "step": 25749 }, { "epoch": 1.7801590044936053, "grad_norm": 0.7442383512762617, "learning_rate": 3.137149390769001e-07, "loss": 0.2562, "step": 25750 }, { "epoch": 1.7802281368821293, "grad_norm": 0.6511964789571443, "learning_rate": 3.135198144362422e-07, "loss": 0.251, "step": 25751 }, { "epoch": 1.7802972692706533, "grad_norm": 0.774813568111919, "learning_rate": 3.133247485322505e-07, "loss": 0.2825, "step": 25752 }, { "epoch": 1.7803664016591774, "grad_norm": 0.6437391062553446, "learning_rate": 3.131297413673684e-07, "loss": 0.2237, "step": 25753 }, { "epoch": 1.7804355340477014, "grad_norm": 0.6519365094009936, "learning_rate": 3.1293479294404236e-07, "loss": 0.2885, "step": 25754 }, { "epoch": 1.7805046664362254, "grad_norm": 0.8290461378233654, "learning_rate": 3.127399032647127e-07, "loss": 0.3631, "step": 25755 }, { "epoch": 1.7805737988247494, "grad_norm": 0.705210466810137, "learning_rate": 3.125450723318241e-07, "loss": 0.2538, "step": 25756 }, { "epoch": 1.7806429312132734, "grad_norm": 0.8610979760459034, "learning_rate": 3.123503001478179e-07, "loss": 0.242, "step": 25757 }, { "epoch": 1.7807120636017975, "grad_norm": 0.7223046193373003, "learning_rate": 3.1215558671513545e-07, "loss": 0.281, "step": 25758 }, { "epoch": 1.7807811959903215, "grad_norm": 0.7244088048821673, "learning_rate": 3.119609320362166e-07, "loss": 0.2907, "step": 25759 }, { "epoch": 1.7808503283788455, "grad_norm": 0.7553556775445596, "learning_rate": 3.1176633611350203e-07, "loss": 0.2556, "step": 25760 }, { "epoch": 1.7809194607673695, "grad_norm": 0.6507386073718662, "learning_rate": 3.115717989494288e-07, "loss": 0.2219, "step": 25761 }, { "epoch": 1.7809885931558935, "grad_norm": 0.7648062347049792, "learning_rate": 3.1137732054643656e-07, "loss": 0.3466, "step": 25762 }, { "epoch": 1.7810577255444175, "grad_norm": 0.7737105621012911, "learning_rate": 3.111829009069622e-07, "loss": 0.3064, "step": 25763 }, { "epoch": 1.7811268579329416, "grad_norm": 0.6577570642391168, "learning_rate": 3.1098854003344335e-07, "loss": 0.2343, "step": 25764 }, { "epoch": 1.7811959903214656, "grad_norm": 0.6963781552594805, "learning_rate": 3.1079423792831466e-07, "loss": 0.2421, "step": 25765 }, { "epoch": 1.7812651227099896, "grad_norm": 0.899488604010166, "learning_rate": 3.1059999459401247e-07, "loss": 0.3524, "step": 25766 }, { "epoch": 1.7813342550985136, "grad_norm": 0.7476084151031964, "learning_rate": 3.104058100329699e-07, "loss": 0.2562, "step": 25767 }, { "epoch": 1.7814033874870376, "grad_norm": 0.6838989605752843, "learning_rate": 3.102116842476227e-07, "loss": 0.2009, "step": 25768 }, { "epoch": 1.7814725198755617, "grad_norm": 0.7420414976903842, "learning_rate": 3.100176172404018e-07, "loss": 0.2514, "step": 25769 }, { "epoch": 1.7815416522640857, "grad_norm": 0.8093804695187434, "learning_rate": 3.098236090137402e-07, "loss": 0.2931, "step": 25770 }, { "epoch": 1.7816107846526097, "grad_norm": 0.6262549364866193, "learning_rate": 3.0962965957006983e-07, "loss": 0.1996, "step": 25771 }, { "epoch": 1.7816799170411337, "grad_norm": 0.7562853722548015, "learning_rate": 3.094357689118221e-07, "loss": 0.2422, "step": 25772 }, { "epoch": 1.7817490494296577, "grad_norm": 0.6271744418101418, "learning_rate": 3.092419370414257e-07, "loss": 0.322, "step": 25773 }, { "epoch": 1.7818181818181817, "grad_norm": 0.6514715305469589, "learning_rate": 3.0904816396131134e-07, "loss": 0.2492, "step": 25774 }, { "epoch": 1.7818873142067058, "grad_norm": 0.6755570076333816, "learning_rate": 3.08854449673906e-07, "loss": 0.241, "step": 25775 }, { "epoch": 1.7819564465952298, "grad_norm": 0.6714033756640412, "learning_rate": 3.086607941816383e-07, "loss": 0.245, "step": 25776 }, { "epoch": 1.7820255789837538, "grad_norm": 0.7457065086851361, "learning_rate": 3.0846719748693574e-07, "loss": 0.3311, "step": 25777 }, { "epoch": 1.7820947113722778, "grad_norm": 0.7059508641947619, "learning_rate": 3.0827365959222477e-07, "loss": 0.2629, "step": 25778 }, { "epoch": 1.7821638437608018, "grad_norm": 1.1211773196438344, "learning_rate": 3.0808018049993e-07, "loss": 0.2552, "step": 25779 }, { "epoch": 1.7822329761493259, "grad_norm": 0.7125060429211524, "learning_rate": 3.07886760212478e-07, "loss": 0.3003, "step": 25780 }, { "epoch": 1.7823021085378499, "grad_norm": 0.8358439010645196, "learning_rate": 3.0769339873229166e-07, "loss": 0.3061, "step": 25781 }, { "epoch": 1.782371240926374, "grad_norm": 0.6422225454414004, "learning_rate": 3.075000960617952e-07, "loss": 0.2141, "step": 25782 }, { "epoch": 1.782440373314898, "grad_norm": 0.7292606734197242, "learning_rate": 3.0730685220341007e-07, "loss": 0.248, "step": 25783 }, { "epoch": 1.782509505703422, "grad_norm": 0.990131298700288, "learning_rate": 3.071136671595587e-07, "loss": 0.3199, "step": 25784 }, { "epoch": 1.782578638091946, "grad_norm": 0.7490666418254347, "learning_rate": 3.069205409326631e-07, "loss": 0.2533, "step": 25785 }, { "epoch": 1.78264777048047, "grad_norm": 0.6989678871936235, "learning_rate": 3.067274735251441e-07, "loss": 0.2475, "step": 25786 }, { "epoch": 1.782716902868994, "grad_norm": 0.743718603310624, "learning_rate": 3.0653446493941976e-07, "loss": 0.3019, "step": 25787 }, { "epoch": 1.782786035257518, "grad_norm": 0.6523599240417882, "learning_rate": 3.0634151517791087e-07, "loss": 0.2332, "step": 25788 }, { "epoch": 1.782855167646042, "grad_norm": 0.7479517172902125, "learning_rate": 3.0614862424303446e-07, "loss": 0.2102, "step": 25789 }, { "epoch": 1.782924300034566, "grad_norm": 0.6896040205107202, "learning_rate": 3.059557921372092e-07, "loss": 0.2644, "step": 25790 }, { "epoch": 1.78299343242309, "grad_norm": 0.634317061054413, "learning_rate": 3.057630188628502e-07, "loss": 0.3286, "step": 25791 }, { "epoch": 1.7830625648116143, "grad_norm": 0.7120059166992984, "learning_rate": 3.0557030442237403e-07, "loss": 0.254, "step": 25792 }, { "epoch": 1.7831316972001383, "grad_norm": 0.6613443517864309, "learning_rate": 3.0537764881819697e-07, "loss": 0.2503, "step": 25793 }, { "epoch": 1.7832008295886623, "grad_norm": 0.6902460482373162, "learning_rate": 3.051850520527344e-07, "loss": 0.2789, "step": 25794 }, { "epoch": 1.7832699619771863, "grad_norm": 0.7041315166909752, "learning_rate": 3.049925141283977e-07, "loss": 0.2374, "step": 25795 }, { "epoch": 1.7833390943657104, "grad_norm": 0.7052478217296316, "learning_rate": 3.048000350476016e-07, "loss": 0.2321, "step": 25796 }, { "epoch": 1.7834082267542344, "grad_norm": 0.6698028127020221, "learning_rate": 3.04607614812758e-07, "loss": 0.2599, "step": 25797 }, { "epoch": 1.7834773591427584, "grad_norm": 0.6722110340065279, "learning_rate": 3.0441525342627844e-07, "loss": 0.3604, "step": 25798 }, { "epoch": 1.7835464915312824, "grad_norm": 0.7912810099405966, "learning_rate": 3.0422295089057363e-07, "loss": 0.2712, "step": 25799 }, { "epoch": 1.7836156239198064, "grad_norm": 0.6758836282209333, "learning_rate": 3.0403070720805504e-07, "loss": 0.2296, "step": 25800 }, { "epoch": 1.7836847563083305, "grad_norm": 0.7023314402153675, "learning_rate": 3.0383852238113077e-07, "loss": 0.299, "step": 25801 }, { "epoch": 1.7837538886968545, "grad_norm": 0.7236355972789489, "learning_rate": 3.036463964122105e-07, "loss": 0.3376, "step": 25802 }, { "epoch": 1.7838230210853785, "grad_norm": 0.7434051108735977, "learning_rate": 3.0345432930370124e-07, "loss": 0.2745, "step": 25803 }, { "epoch": 1.7838921534739025, "grad_norm": 0.6385854844969969, "learning_rate": 3.03262321058011e-07, "loss": 0.2232, "step": 25804 }, { "epoch": 1.7839612858624265, "grad_norm": 0.7038539816729454, "learning_rate": 3.0307037167754514e-07, "loss": 0.3303, "step": 25805 }, { "epoch": 1.7840304182509505, "grad_norm": 0.6821500674520746, "learning_rate": 3.028784811647101e-07, "loss": 0.2343, "step": 25806 }, { "epoch": 1.7840995506394746, "grad_norm": 0.8753200091595171, "learning_rate": 3.026866495219116e-07, "loss": 0.1958, "step": 25807 }, { "epoch": 1.7841686830279986, "grad_norm": 0.6841190733981167, "learning_rate": 3.0249487675155286e-07, "loss": 0.2519, "step": 25808 }, { "epoch": 1.7842378154165226, "grad_norm": 0.7872462826132176, "learning_rate": 3.02303162856038e-07, "loss": 0.3135, "step": 25809 }, { "epoch": 1.7843069478050466, "grad_norm": 0.7056180308822646, "learning_rate": 3.02111507837769e-07, "loss": 0.2296, "step": 25810 }, { "epoch": 1.7843760801935706, "grad_norm": 0.697793954989878, "learning_rate": 3.0191991169914945e-07, "loss": 0.231, "step": 25811 }, { "epoch": 1.7844452125820949, "grad_norm": 0.7822021897696343, "learning_rate": 3.0172837444257807e-07, "loss": 0.3133, "step": 25812 }, { "epoch": 1.784514344970619, "grad_norm": 1.4771678966882127, "learning_rate": 3.015368960704584e-07, "loss": 0.2652, "step": 25813 }, { "epoch": 1.784583477359143, "grad_norm": 0.6361893146271375, "learning_rate": 3.013454765851881e-07, "loss": 0.1996, "step": 25814 }, { "epoch": 1.784652609747667, "grad_norm": 0.7153892812237744, "learning_rate": 3.0115411598916843e-07, "loss": 0.2604, "step": 25815 }, { "epoch": 1.784721742136191, "grad_norm": 0.7658224198328871, "learning_rate": 3.0096281428479535e-07, "loss": 0.3485, "step": 25816 }, { "epoch": 1.784790874524715, "grad_norm": 0.7015267553224461, "learning_rate": 3.0077157147446857e-07, "loss": 0.2217, "step": 25817 }, { "epoch": 1.784860006913239, "grad_norm": 0.6927498817358904, "learning_rate": 3.0058038756058285e-07, "loss": 0.3024, "step": 25818 }, { "epoch": 1.784929139301763, "grad_norm": 0.7155184601574882, "learning_rate": 3.003892625455357e-07, "loss": 0.2765, "step": 25819 }, { "epoch": 1.784998271690287, "grad_norm": 0.7631165398347332, "learning_rate": 3.0019819643172244e-07, "loss": 0.3394, "step": 25820 }, { "epoch": 1.785067404078811, "grad_norm": 0.7901235655139972, "learning_rate": 3.0000718922153783e-07, "loss": 0.2346, "step": 25821 }, { "epoch": 1.785136536467335, "grad_norm": 0.698601427838567, "learning_rate": 2.998162409173755e-07, "loss": 0.2554, "step": 25822 }, { "epoch": 1.785205668855859, "grad_norm": 0.8772168489107472, "learning_rate": 2.996253515216291e-07, "loss": 0.3532, "step": 25823 }, { "epoch": 1.785274801244383, "grad_norm": 0.7538580228328388, "learning_rate": 2.9943452103669004e-07, "loss": 0.244, "step": 25824 }, { "epoch": 1.785343933632907, "grad_norm": 0.7441602961742136, "learning_rate": 2.992437494649514e-07, "loss": 0.2518, "step": 25825 }, { "epoch": 1.7854130660214311, "grad_norm": 0.7089057682177167, "learning_rate": 2.9905303680880293e-07, "loss": 0.2911, "step": 25826 }, { "epoch": 1.7854821984099551, "grad_norm": 0.7506621496561747, "learning_rate": 2.9886238307063554e-07, "loss": 0.3309, "step": 25827 }, { "epoch": 1.7855513307984792, "grad_norm": 0.6715557970423327, "learning_rate": 2.9867178825283837e-07, "loss": 0.195, "step": 25828 }, { "epoch": 1.7856204631870032, "grad_norm": 0.6659024430497654, "learning_rate": 2.984812523578012e-07, "loss": 0.3042, "step": 25829 }, { "epoch": 1.7856895955755272, "grad_norm": 0.7682753119197168, "learning_rate": 2.982907753879105e-07, "loss": 0.3428, "step": 25830 }, { "epoch": 1.7857587279640512, "grad_norm": 0.7649920743005063, "learning_rate": 2.9810035734555543e-07, "loss": 0.2984, "step": 25831 }, { "epoch": 1.7858278603525752, "grad_norm": 0.6859175461526503, "learning_rate": 2.979099982331202e-07, "loss": 0.2424, "step": 25832 }, { "epoch": 1.7858969927410993, "grad_norm": 0.803729137120951, "learning_rate": 2.977196980529923e-07, "loss": 0.332, "step": 25833 }, { "epoch": 1.7859661251296233, "grad_norm": 0.7330868701042875, "learning_rate": 2.9752945680755653e-07, "loss": 0.3359, "step": 25834 }, { "epoch": 1.7860352575181473, "grad_norm": 0.7232874059506887, "learning_rate": 2.9733927449919707e-07, "loss": 0.2427, "step": 25835 }, { "epoch": 1.7861043899066713, "grad_norm": 0.6843491960070754, "learning_rate": 2.971491511302976e-07, "loss": 0.2234, "step": 25836 }, { "epoch": 1.7861735222951953, "grad_norm": 0.7326488901096654, "learning_rate": 2.9695908670324127e-07, "loss": 0.3369, "step": 25837 }, { "epoch": 1.7862426546837193, "grad_norm": 0.7777466408843743, "learning_rate": 2.967690812204094e-07, "loss": 0.2602, "step": 25838 }, { "epoch": 1.7863117870722434, "grad_norm": 0.6621681938427373, "learning_rate": 2.96579134684184e-07, "loss": 0.2068, "step": 25839 }, { "epoch": 1.7863809194607674, "grad_norm": 0.7513108605892249, "learning_rate": 2.963892470969448e-07, "loss": 0.271, "step": 25840 }, { "epoch": 1.7864500518492914, "grad_norm": 0.7179825620349423, "learning_rate": 2.9619941846107283e-07, "loss": 0.3408, "step": 25841 }, { "epoch": 1.7865191842378154, "grad_norm": 0.6525093020909054, "learning_rate": 2.9600964877894654e-07, "loss": 0.1987, "step": 25842 }, { "epoch": 1.7865883166263394, "grad_norm": 0.694449449565322, "learning_rate": 2.9581993805294586e-07, "loss": 0.2634, "step": 25843 }, { "epoch": 1.7866574490148635, "grad_norm": 3.3587911274104343, "learning_rate": 2.95630286285446e-07, "loss": 0.3186, "step": 25844 }, { "epoch": 1.7867265814033875, "grad_norm": 0.7164829521720795, "learning_rate": 2.954406934788262e-07, "loss": 0.2911, "step": 25845 }, { "epoch": 1.7867957137919115, "grad_norm": 0.7075545333182578, "learning_rate": 2.952511596354607e-07, "loss": 0.1884, "step": 25846 }, { "epoch": 1.7868648461804355, "grad_norm": 0.7222710961280091, "learning_rate": 2.95061684757727e-07, "loss": 0.252, "step": 25847 }, { "epoch": 1.7869339785689595, "grad_norm": 0.7946989163833588, "learning_rate": 2.9487226884799715e-07, "loss": 0.3461, "step": 25848 }, { "epoch": 1.7870031109574835, "grad_norm": 0.6829270026723951, "learning_rate": 2.9468291190864804e-07, "loss": 0.2104, "step": 25849 }, { "epoch": 1.7870722433460076, "grad_norm": 0.6633809910304481, "learning_rate": 2.944936139420512e-07, "loss": 0.1945, "step": 25850 }, { "epoch": 1.7871413757345316, "grad_norm": 0.7523521005691877, "learning_rate": 2.9430437495057973e-07, "loss": 0.2345, "step": 25851 }, { "epoch": 1.7872105081230556, "grad_norm": 0.7756573848561995, "learning_rate": 2.94115194936605e-07, "loss": 0.3061, "step": 25852 }, { "epoch": 1.7872796405115796, "grad_norm": 0.7025913750434845, "learning_rate": 2.939260739024985e-07, "loss": 0.2618, "step": 25853 }, { "epoch": 1.7873487729001036, "grad_norm": 0.7481777223064209, "learning_rate": 2.9373701185062997e-07, "loss": 0.2425, "step": 25854 }, { "epoch": 1.7874179052886277, "grad_norm": 0.6713602638810827, "learning_rate": 2.935480087833692e-07, "loss": 0.3229, "step": 25855 }, { "epoch": 1.7874870376771517, "grad_norm": 0.7088672343165339, "learning_rate": 2.933590647030854e-07, "loss": 0.2797, "step": 25856 }, { "epoch": 1.7875561700656757, "grad_norm": 0.6505254432321884, "learning_rate": 2.9317017961214667e-07, "loss": 0.2324, "step": 25857 }, { "epoch": 1.7876253024541997, "grad_norm": 0.7777542731360599, "learning_rate": 2.9298135351291943e-07, "loss": 0.291, "step": 25858 }, { "epoch": 1.7876944348427237, "grad_norm": 2.966719977654462, "learning_rate": 2.927925864077713e-07, "loss": 0.3003, "step": 25859 }, { "epoch": 1.7877635672312477, "grad_norm": 0.762914976501354, "learning_rate": 2.9260387829906753e-07, "loss": 0.2439, "step": 25860 }, { "epoch": 1.7878326996197718, "grad_norm": 0.730375677587184, "learning_rate": 2.9241522918917356e-07, "loss": 0.2256, "step": 25861 }, { "epoch": 1.7879018320082958, "grad_norm": 0.758172467998272, "learning_rate": 2.9222663908045347e-07, "loss": 0.3427, "step": 25862 }, { "epoch": 1.7879709643968198, "grad_norm": 0.7348460403771211, "learning_rate": 2.9203810797527043e-07, "loss": 0.3248, "step": 25863 }, { "epoch": 1.7880400967853438, "grad_norm": 0.6266261115028454, "learning_rate": 2.9184963587598867e-07, "loss": 0.167, "step": 25864 }, { "epoch": 1.7881092291738678, "grad_norm": 0.7207595693117531, "learning_rate": 2.916612227849702e-07, "loss": 0.2311, "step": 25865 }, { "epoch": 1.7881783615623918, "grad_norm": 0.6890806565153219, "learning_rate": 2.914728687045748e-07, "loss": 0.3514, "step": 25866 }, { "epoch": 1.7882474939509159, "grad_norm": 0.656052410460814, "learning_rate": 2.91284573637165e-07, "loss": 0.2614, "step": 25867 }, { "epoch": 1.7883166263394399, "grad_norm": 0.6728460332110284, "learning_rate": 2.9109633758509945e-07, "loss": 0.2321, "step": 25868 }, { "epoch": 1.788385758727964, "grad_norm": 0.7238382604735277, "learning_rate": 2.90908160550738e-07, "loss": 0.2924, "step": 25869 }, { "epoch": 1.788454891116488, "grad_norm": 0.7586868762224452, "learning_rate": 2.907200425364387e-07, "loss": 0.2986, "step": 25870 }, { "epoch": 1.788524023505012, "grad_norm": 0.7118421844454144, "learning_rate": 2.905319835445608e-07, "loss": 0.3009, "step": 25871 }, { "epoch": 1.788593155893536, "grad_norm": 1.2457337503756227, "learning_rate": 2.903439835774591e-07, "loss": 0.2455, "step": 25872 }, { "epoch": 1.7886622882820602, "grad_norm": 0.7495948274122901, "learning_rate": 2.9015604263749165e-07, "loss": 0.3848, "step": 25873 }, { "epoch": 1.7887314206705842, "grad_norm": 0.7334501495174176, "learning_rate": 2.899681607270127e-07, "loss": 0.2091, "step": 25874 }, { "epoch": 1.7888005530591082, "grad_norm": 0.6510241376391922, "learning_rate": 2.8978033784837713e-07, "loss": 0.1952, "step": 25875 }, { "epoch": 1.7888696854476323, "grad_norm": 0.7246321108085702, "learning_rate": 2.8959257400393903e-07, "loss": 0.2919, "step": 25876 }, { "epoch": 1.7889388178361563, "grad_norm": 3.6481439689249755, "learning_rate": 2.894048691960516e-07, "loss": 0.3094, "step": 25877 }, { "epoch": 1.7890079502246803, "grad_norm": 0.6974396809744264, "learning_rate": 2.89217223427069e-07, "loss": 0.255, "step": 25878 }, { "epoch": 1.7890770826132043, "grad_norm": 0.7332249994530997, "learning_rate": 2.8902963669934057e-07, "loss": 0.2441, "step": 25879 }, { "epoch": 1.7891462150017283, "grad_norm": 0.7044540826527703, "learning_rate": 2.888421090152199e-07, "loss": 0.3764, "step": 25880 }, { "epoch": 1.7892153473902523, "grad_norm": 0.7850146454835902, "learning_rate": 2.886546403770546e-07, "loss": 0.2704, "step": 25881 }, { "epoch": 1.7892844797787764, "grad_norm": 0.7203676899318375, "learning_rate": 2.8846723078719663e-07, "loss": 0.2288, "step": 25882 }, { "epoch": 1.7893536121673004, "grad_norm": 0.7282409211278154, "learning_rate": 2.882798802479925e-07, "loss": 0.2977, "step": 25883 }, { "epoch": 1.7894227445558244, "grad_norm": 0.7118668996000256, "learning_rate": 2.8809258876179247e-07, "loss": 0.3218, "step": 25884 }, { "epoch": 1.7894918769443484, "grad_norm": 0.7839724736762762, "learning_rate": 2.8790535633094306e-07, "loss": 0.2716, "step": 25885 }, { "epoch": 1.7895610093328724, "grad_norm": 0.7926975766581517, "learning_rate": 2.877181829577913e-07, "loss": 0.2649, "step": 25886 }, { "epoch": 1.7896301417213965, "grad_norm": 0.6948539688573413, "learning_rate": 2.8753106864468196e-07, "loss": 0.3262, "step": 25887 }, { "epoch": 1.7896992741099205, "grad_norm": 0.845820412179754, "learning_rate": 2.8734401339396146e-07, "loss": 0.2757, "step": 25888 }, { "epoch": 1.7897684064984445, "grad_norm": 0.7366464741732468, "learning_rate": 2.8715701720797297e-07, "loss": 0.2799, "step": 25889 }, { "epoch": 1.7898375388869685, "grad_norm": 0.6617684564267978, "learning_rate": 2.869700800890607e-07, "loss": 0.2336, "step": 25890 }, { "epoch": 1.7899066712754925, "grad_norm": 0.7320965789576739, "learning_rate": 2.8678320203956835e-07, "loss": 0.3304, "step": 25891 }, { "epoch": 1.7899758036640165, "grad_norm": 0.6777073646563289, "learning_rate": 2.8659638306183737e-07, "loss": 0.2454, "step": 25892 }, { "epoch": 1.7900449360525408, "grad_norm": 0.6764314910440863, "learning_rate": 2.864096231582092e-07, "loss": 0.233, "step": 25893 }, { "epoch": 1.7901140684410648, "grad_norm": 0.7454749858709846, "learning_rate": 2.8622292233102475e-07, "loss": 0.3155, "step": 25894 }, { "epoch": 1.7901832008295888, "grad_norm": 0.7371818971352928, "learning_rate": 2.860362805826239e-07, "loss": 0.2677, "step": 25895 }, { "epoch": 1.7902523332181128, "grad_norm": 0.8361002702421753, "learning_rate": 2.858496979153458e-07, "loss": 0.3008, "step": 25896 }, { "epoch": 1.7903214656066369, "grad_norm": 0.7892278530346861, "learning_rate": 2.856631743315286e-07, "loss": 0.2777, "step": 25897 }, { "epoch": 1.7903905979951609, "grad_norm": 0.7469876475115473, "learning_rate": 2.854767098335104e-07, "loss": 0.3484, "step": 25898 }, { "epoch": 1.7904597303836849, "grad_norm": 0.6896849454338956, "learning_rate": 2.8529030442362826e-07, "loss": 0.2157, "step": 25899 }, { "epoch": 1.790528862772209, "grad_norm": 0.6353907164127283, "learning_rate": 2.8510395810421866e-07, "loss": 0.2397, "step": 25900 }, { "epoch": 1.790597995160733, "grad_norm": 0.6783261460682014, "learning_rate": 2.8491767087761635e-07, "loss": 0.218, "step": 25901 }, { "epoch": 1.790667127549257, "grad_norm": 0.6936760375720872, "learning_rate": 2.847314427461573e-07, "loss": 0.2837, "step": 25902 }, { "epoch": 1.790736259937781, "grad_norm": 0.644928512086728, "learning_rate": 2.8454527371217455e-07, "loss": 0.1895, "step": 25903 }, { "epoch": 1.790805392326305, "grad_norm": 0.7650853182584536, "learning_rate": 2.843591637780013e-07, "loss": 0.2719, "step": 25904 }, { "epoch": 1.790874524714829, "grad_norm": 0.6975240092434811, "learning_rate": 2.8417311294597015e-07, "loss": 0.3476, "step": 25905 }, { "epoch": 1.790943657103353, "grad_norm": 1.9572103538137582, "learning_rate": 2.839871212184148e-07, "loss": 0.2505, "step": 25906 }, { "epoch": 1.791012789491877, "grad_norm": 0.6679744083281139, "learning_rate": 2.838011885976633e-07, "loss": 0.2016, "step": 25907 }, { "epoch": 1.791081921880401, "grad_norm": 0.7617759152360125, "learning_rate": 2.836153150860488e-07, "loss": 0.332, "step": 25908 }, { "epoch": 1.791151054268925, "grad_norm": 0.752637125770984, "learning_rate": 2.8342950068589835e-07, "loss": 0.3186, "step": 25909 }, { "epoch": 1.791220186657449, "grad_norm": 0.941240265501988, "learning_rate": 2.832437453995429e-07, "loss": 0.2531, "step": 25910 }, { "epoch": 1.791289319045973, "grad_norm": 0.704209230057742, "learning_rate": 2.8305804922930893e-07, "loss": 0.2324, "step": 25911 }, { "epoch": 1.7913584514344971, "grad_norm": 0.791147533713392, "learning_rate": 2.8287241217752504e-07, "loss": 0.2768, "step": 25912 }, { "epoch": 1.7914275838230211, "grad_norm": 0.6979503212522293, "learning_rate": 2.8268683424651666e-07, "loss": 0.2794, "step": 25913 }, { "epoch": 1.7914967162115452, "grad_norm": 0.6605999599389637, "learning_rate": 2.8250131543861134e-07, "loss": 0.1823, "step": 25914 }, { "epoch": 1.7915658486000692, "grad_norm": 0.7112691556728236, "learning_rate": 2.8231585575613287e-07, "loss": 0.2606, "step": 25915 }, { "epoch": 1.7916349809885932, "grad_norm": 0.7024306785409016, "learning_rate": 2.821304552014065e-07, "loss": 0.3097, "step": 25916 }, { "epoch": 1.7917041133771172, "grad_norm": 0.6602995116877062, "learning_rate": 2.819451137767548e-07, "loss": 0.2483, "step": 25917 }, { "epoch": 1.7917732457656412, "grad_norm": 0.962442964667054, "learning_rate": 2.817598314845016e-07, "loss": 0.2135, "step": 25918 }, { "epoch": 1.7918423781541652, "grad_norm": 0.6561942912434894, "learning_rate": 2.8157460832696826e-07, "loss": 0.294, "step": 25919 }, { "epoch": 1.7919115105426893, "grad_norm": 0.6630849194623875, "learning_rate": 2.8138944430647743e-07, "loss": 0.2621, "step": 25920 }, { "epoch": 1.7919806429312133, "grad_norm": 0.790783548728619, "learning_rate": 2.8120433942534884e-07, "loss": 0.2774, "step": 25921 }, { "epoch": 1.7920497753197373, "grad_norm": 0.7472286371886757, "learning_rate": 2.8101929368590355e-07, "loss": 0.2705, "step": 25922 }, { "epoch": 1.7921189077082613, "grad_norm": 0.716532234520232, "learning_rate": 2.80834307090459e-07, "loss": 0.2992, "step": 25923 }, { "epoch": 1.7921880400967853, "grad_norm": 0.7248533932757959, "learning_rate": 2.8064937964133564e-07, "loss": 0.2772, "step": 25924 }, { "epoch": 1.7922571724853094, "grad_norm": 0.7183527538201294, "learning_rate": 2.8046451134084994e-07, "loss": 0.2123, "step": 25925 }, { "epoch": 1.7923263048738334, "grad_norm": 0.7413329844123794, "learning_rate": 2.802797021913184e-07, "loss": 0.258, "step": 25926 }, { "epoch": 1.7923954372623574, "grad_norm": 0.9081301040625495, "learning_rate": 2.8009495219505854e-07, "loss": 0.3031, "step": 25927 }, { "epoch": 1.7924645696508814, "grad_norm": 0.6573892808678659, "learning_rate": 2.7991026135438637e-07, "loss": 0.2095, "step": 25928 }, { "epoch": 1.7925337020394054, "grad_norm": 0.7135755805138702, "learning_rate": 2.7972562967161444e-07, "loss": 0.2744, "step": 25929 }, { "epoch": 1.7926028344279294, "grad_norm": 0.7757417303163049, "learning_rate": 2.795410571490592e-07, "loss": 0.3299, "step": 25930 }, { "epoch": 1.7926719668164535, "grad_norm": 0.8877134738489332, "learning_rate": 2.7935654378903167e-07, "loss": 0.2638, "step": 25931 }, { "epoch": 1.7927410992049775, "grad_norm": 0.6654897883209467, "learning_rate": 2.791720895938466e-07, "loss": 0.2266, "step": 25932 }, { "epoch": 1.7928102315935015, "grad_norm": 0.7299288311214669, "learning_rate": 2.7898769456581323e-07, "loss": 0.2849, "step": 25933 }, { "epoch": 1.7928793639820255, "grad_norm": 0.713341549782829, "learning_rate": 2.7880335870724476e-07, "loss": 0.3523, "step": 25934 }, { "epoch": 1.7929484963705495, "grad_norm": 0.6482363900268464, "learning_rate": 2.786190820204504e-07, "loss": 0.1895, "step": 25935 }, { "epoch": 1.7930176287590736, "grad_norm": 0.684167820406529, "learning_rate": 2.7843486450774114e-07, "loss": 0.2327, "step": 25936 }, { "epoch": 1.7930867611475976, "grad_norm": 0.7925656835919946, "learning_rate": 2.7825070617142455e-07, "loss": 0.3107, "step": 25937 }, { "epoch": 1.7931558935361216, "grad_norm": 0.7534206327266202, "learning_rate": 2.7806660701380936e-07, "loss": 0.33, "step": 25938 }, { "epoch": 1.7932250259246456, "grad_norm": 0.7222487037014529, "learning_rate": 2.778825670372015e-07, "loss": 0.2415, "step": 25939 }, { "epoch": 1.7932941583131696, "grad_norm": 0.8537956564979754, "learning_rate": 2.7769858624390913e-07, "loss": 0.2717, "step": 25940 }, { "epoch": 1.7933632907016936, "grad_norm": 0.7247011208727492, "learning_rate": 2.775146646362381e-07, "loss": 0.3523, "step": 25941 }, { "epoch": 1.7934324230902177, "grad_norm": 0.7228625760684317, "learning_rate": 2.773308022164928e-07, "loss": 0.2835, "step": 25942 }, { "epoch": 1.7935015554787417, "grad_norm": 0.7364296075118997, "learning_rate": 2.7714699898697796e-07, "loss": 0.1886, "step": 25943 }, { "epoch": 1.7935706878672657, "grad_norm": 0.6894577395430925, "learning_rate": 2.7696325494999675e-07, "loss": 0.2794, "step": 25944 }, { "epoch": 1.7936398202557897, "grad_norm": 1.170948038359011, "learning_rate": 2.767795701078535e-07, "loss": 0.2579, "step": 25945 }, { "epoch": 1.7937089526443137, "grad_norm": 0.6909653945809817, "learning_rate": 2.7659594446284853e-07, "loss": 0.2373, "step": 25946 }, { "epoch": 1.7937780850328378, "grad_norm": 0.7164572626816348, "learning_rate": 2.7641237801728394e-07, "loss": 0.3066, "step": 25947 }, { "epoch": 1.7938472174213618, "grad_norm": 0.7149532516128914, "learning_rate": 2.7622887077346063e-07, "loss": 0.3794, "step": 25948 }, { "epoch": 1.7939163498098858, "grad_norm": 0.7577521673842101, "learning_rate": 2.760454227336795e-07, "loss": 0.2991, "step": 25949 }, { "epoch": 1.7939854821984098, "grad_norm": 0.6635492939591673, "learning_rate": 2.758620339002377e-07, "loss": 0.2069, "step": 25950 }, { "epoch": 1.7940546145869338, "grad_norm": 0.7894118312515016, "learning_rate": 2.7567870427543553e-07, "loss": 0.2843, "step": 25951 }, { "epoch": 1.7941237469754578, "grad_norm": 0.7126900072980482, "learning_rate": 2.75495433861569e-07, "loss": 0.3041, "step": 25952 }, { "epoch": 1.7941928793639819, "grad_norm": 0.6750250120463851, "learning_rate": 2.753122226609367e-07, "loss": 0.2077, "step": 25953 }, { "epoch": 1.794262011752506, "grad_norm": 0.6955511839633917, "learning_rate": 2.751290706758325e-07, "loss": 0.2169, "step": 25954 }, { "epoch": 1.7943311441410301, "grad_norm": 0.6817117721328926, "learning_rate": 2.7494597790855504e-07, "loss": 0.3295, "step": 25955 }, { "epoch": 1.7944002765295541, "grad_norm": 0.9201689120423853, "learning_rate": 2.747629443613969e-07, "loss": 0.256, "step": 25956 }, { "epoch": 1.7944694089180782, "grad_norm": 0.7180364276734638, "learning_rate": 2.745799700366536e-07, "loss": 0.2385, "step": 25957 }, { "epoch": 1.7945385413066022, "grad_norm": 0.6925838487988529, "learning_rate": 2.743970549366165e-07, "loss": 0.2145, "step": 25958 }, { "epoch": 1.7946076736951262, "grad_norm": 0.7806396030479148, "learning_rate": 2.7421419906357994e-07, "loss": 0.3434, "step": 25959 }, { "epoch": 1.7946768060836502, "grad_norm": 0.9027551419035852, "learning_rate": 2.740314024198343e-07, "loss": 0.2441, "step": 25960 }, { "epoch": 1.7947459384721742, "grad_norm": 0.7077786888092625, "learning_rate": 2.738486650076705e-07, "loss": 0.2366, "step": 25961 }, { "epoch": 1.7948150708606982, "grad_norm": 0.774069582415767, "learning_rate": 2.736659868293801e-07, "loss": 0.3495, "step": 25962 }, { "epoch": 1.7948842032492223, "grad_norm": 0.7592128080393711, "learning_rate": 2.7348336788725285e-07, "loss": 0.237, "step": 25963 }, { "epoch": 1.7949533356377463, "grad_norm": 0.6991587833763269, "learning_rate": 2.733008081835753e-07, "loss": 0.2147, "step": 25964 }, { "epoch": 1.7950224680262703, "grad_norm": 0.6972308651051011, "learning_rate": 2.731183077206384e-07, "loss": 0.2464, "step": 25965 }, { "epoch": 1.7950916004147943, "grad_norm": 0.6993477237557455, "learning_rate": 2.72935866500727e-07, "loss": 0.281, "step": 25966 }, { "epoch": 1.7951607328033183, "grad_norm": 0.6930410008502081, "learning_rate": 2.7275348452612926e-07, "loss": 0.2136, "step": 25967 }, { "epoch": 1.7952298651918424, "grad_norm": 0.6360700607223108, "learning_rate": 2.7257116179913e-07, "loss": 0.2346, "step": 25968 }, { "epoch": 1.7952989975803664, "grad_norm": 0.7791944854510834, "learning_rate": 2.7238889832201465e-07, "loss": 0.2639, "step": 25969 }, { "epoch": 1.7953681299688904, "grad_norm": 0.7562588850879282, "learning_rate": 2.7220669409706743e-07, "loss": 0.3206, "step": 25970 }, { "epoch": 1.7954372623574144, "grad_norm": 0.6543439185508495, "learning_rate": 2.7202454912657325e-07, "loss": 0.2356, "step": 25971 }, { "epoch": 1.7955063947459384, "grad_norm": 0.6470749008274262, "learning_rate": 2.7184246341281307e-07, "loss": 0.251, "step": 25972 }, { "epoch": 1.7955755271344624, "grad_norm": 0.6950141649139985, "learning_rate": 2.7166043695807053e-07, "loss": 0.3057, "step": 25973 }, { "epoch": 1.7956446595229865, "grad_norm": 0.7319569723161764, "learning_rate": 2.7147846976462553e-07, "loss": 0.2231, "step": 25974 }, { "epoch": 1.7957137919115107, "grad_norm": 0.6916666043856271, "learning_rate": 2.7129656183475904e-07, "loss": 0.2425, "step": 25975 }, { "epoch": 1.7957829243000347, "grad_norm": 0.9990055909652991, "learning_rate": 2.71114713170752e-07, "loss": 0.2744, "step": 25976 }, { "epoch": 1.7958520566885587, "grad_norm": 0.7324684536635244, "learning_rate": 2.7093292377488365e-07, "loss": 0.3743, "step": 25977 }, { "epoch": 1.7959211890770828, "grad_norm": 0.6776858742679882, "learning_rate": 2.707511936494311e-07, "loss": 0.2403, "step": 25978 }, { "epoch": 1.7959903214656068, "grad_norm": 0.6707139425490726, "learning_rate": 2.7056952279667317e-07, "loss": 0.2281, "step": 25979 }, { "epoch": 1.7960594538541308, "grad_norm": 0.7043679231863954, "learning_rate": 2.703879112188851e-07, "loss": 0.3183, "step": 25980 }, { "epoch": 1.7961285862426548, "grad_norm": 0.7347912238441622, "learning_rate": 2.7020635891834513e-07, "loss": 0.2403, "step": 25981 }, { "epoch": 1.7961977186311788, "grad_norm": 0.6976460416742966, "learning_rate": 2.700248658973276e-07, "loss": 0.2209, "step": 25982 }, { "epoch": 1.7962668510197028, "grad_norm": 0.6574355775065688, "learning_rate": 2.698434321581067e-07, "loss": 0.2658, "step": 25983 }, { "epoch": 1.7963359834082269, "grad_norm": 0.6744758478599482, "learning_rate": 2.696620577029574e-07, "loss": 0.3219, "step": 25984 }, { "epoch": 1.7964051157967509, "grad_norm": 0.7279740637101524, "learning_rate": 2.694807425341528e-07, "loss": 0.255, "step": 25985 }, { "epoch": 1.796474248185275, "grad_norm": 0.7681204304946648, "learning_rate": 2.6929948665396497e-07, "loss": 0.1918, "step": 25986 }, { "epoch": 1.796543380573799, "grad_norm": 0.7590667935430288, "learning_rate": 2.6911829006466605e-07, "loss": 0.3633, "step": 25987 }, { "epoch": 1.796612512962323, "grad_norm": 0.8301993578983395, "learning_rate": 2.689371527685264e-07, "loss": 0.3015, "step": 25988 }, { "epoch": 1.796681645350847, "grad_norm": 0.7304457465257788, "learning_rate": 2.68756074767817e-07, "loss": 0.2839, "step": 25989 }, { "epoch": 1.796750777739371, "grad_norm": 0.7153556217248608, "learning_rate": 2.685750560648054e-07, "loss": 0.2653, "step": 25990 }, { "epoch": 1.796819910127895, "grad_norm": 0.6981873575779048, "learning_rate": 2.683940966617637e-07, "loss": 0.3606, "step": 25991 }, { "epoch": 1.796889042516419, "grad_norm": 0.7036210287670045, "learning_rate": 2.682131965609569e-07, "loss": 0.1919, "step": 25992 }, { "epoch": 1.796958174904943, "grad_norm": 0.63469072928886, "learning_rate": 2.6803235576465414e-07, "loss": 0.2125, "step": 25993 }, { "epoch": 1.797027307293467, "grad_norm": 0.6870800156116703, "learning_rate": 2.6785157427512085e-07, "loss": 0.3025, "step": 25994 }, { "epoch": 1.797096439681991, "grad_norm": 0.7813557752666619, "learning_rate": 2.676708520946236e-07, "loss": 0.2757, "step": 25995 }, { "epoch": 1.797165572070515, "grad_norm": 0.7033016164254782, "learning_rate": 2.6749018922542667e-07, "loss": 0.2572, "step": 25996 }, { "epoch": 1.797234704459039, "grad_norm": 0.6913011718270005, "learning_rate": 2.673095856697944e-07, "loss": 0.2412, "step": 25997 }, { "epoch": 1.7973038368475631, "grad_norm": 0.6835070341240327, "learning_rate": 2.671290414299904e-07, "loss": 0.328, "step": 25998 }, { "epoch": 1.7973729692360871, "grad_norm": 0.7724276837234302, "learning_rate": 2.669485565082791e-07, "loss": 0.3072, "step": 25999 }, { "epoch": 1.7974421016246112, "grad_norm": 0.7034389969614337, "learning_rate": 2.6676813090691976e-07, "loss": 0.2565, "step": 26000 }, { "epoch": 1.7975112340131352, "grad_norm": 0.7987199397684696, "learning_rate": 2.6658776462817615e-07, "loss": 0.2749, "step": 26001 }, { "epoch": 1.7975803664016592, "grad_norm": 0.7428390993066865, "learning_rate": 2.66407457674307e-07, "loss": 0.2816, "step": 26002 }, { "epoch": 1.7976494987901832, "grad_norm": 0.7141418846024518, "learning_rate": 2.6622721004757325e-07, "loss": 0.2386, "step": 26003 }, { "epoch": 1.7977186311787072, "grad_norm": 0.6441334402842966, "learning_rate": 2.660470217502331e-07, "loss": 0.2532, "step": 26004 }, { "epoch": 1.7977877635672312, "grad_norm": 0.6757210996849885, "learning_rate": 2.6586689278454535e-07, "loss": 0.2893, "step": 26005 }, { "epoch": 1.7978568959557553, "grad_norm": 0.692585499663227, "learning_rate": 2.656868231527687e-07, "loss": 0.2674, "step": 26006 }, { "epoch": 1.7979260283442793, "grad_norm": 1.0407677608452577, "learning_rate": 2.655068128571581e-07, "loss": 0.2256, "step": 26007 }, { "epoch": 1.7979951607328033, "grad_norm": 0.6622712036992374, "learning_rate": 2.6532686189997103e-07, "loss": 0.2055, "step": 26008 }, { "epoch": 1.7980642931213273, "grad_norm": 0.7917100405348857, "learning_rate": 2.6514697028346194e-07, "loss": 0.3313, "step": 26009 }, { "epoch": 1.7981334255098513, "grad_norm": 0.7730962860862295, "learning_rate": 2.649671380098856e-07, "loss": 0.2535, "step": 26010 }, { "epoch": 1.7982025578983754, "grad_norm": 0.8126964959208725, "learning_rate": 2.6478736508149585e-07, "loss": 0.3229, "step": 26011 }, { "epoch": 1.7982716902868994, "grad_norm": 0.7793581156348984, "learning_rate": 2.6460765150054744e-07, "loss": 0.3394, "step": 26012 }, { "epoch": 1.7983408226754234, "grad_norm": 0.7298311495599972, "learning_rate": 2.6442799726929034e-07, "loss": 0.2822, "step": 26013 }, { "epoch": 1.7984099550639474, "grad_norm": 0.6394445158305541, "learning_rate": 2.642484023899783e-07, "loss": 0.2331, "step": 26014 }, { "epoch": 1.7984790874524714, "grad_norm": 0.6921087167022845, "learning_rate": 2.6406886686486e-07, "loss": 0.235, "step": 26015 }, { "epoch": 1.7985482198409954, "grad_norm": 0.6952881461650747, "learning_rate": 2.638893906961881e-07, "loss": 0.3586, "step": 26016 }, { "epoch": 1.7986173522295195, "grad_norm": 0.6844101964931903, "learning_rate": 2.637099738862098e-07, "loss": 0.2285, "step": 26017 }, { "epoch": 1.7986864846180435, "grad_norm": 0.693040484858001, "learning_rate": 2.635306164371748e-07, "loss": 0.2596, "step": 26018 }, { "epoch": 1.7987556170065675, "grad_norm": 0.6680665018007915, "learning_rate": 2.633513183513303e-07, "loss": 0.3111, "step": 26019 }, { "epoch": 1.7988247493950915, "grad_norm": 0.7048914142789233, "learning_rate": 2.6317207963092564e-07, "loss": 0.2465, "step": 26020 }, { "epoch": 1.7988938817836155, "grad_norm": 0.6572429204559975, "learning_rate": 2.6299290027820446e-07, "loss": 0.241, "step": 26021 }, { "epoch": 1.7989630141721396, "grad_norm": 0.783766077776744, "learning_rate": 2.628137802954145e-07, "loss": 0.2663, "step": 26022 }, { "epoch": 1.7990321465606636, "grad_norm": 0.6805044203571675, "learning_rate": 2.626347196847995e-07, "loss": 0.3091, "step": 26023 }, { "epoch": 1.7991012789491876, "grad_norm": 0.7249425450887377, "learning_rate": 2.624557184486043e-07, "loss": 0.2875, "step": 26024 }, { "epoch": 1.7991704113377116, "grad_norm": 0.7085197551532298, "learning_rate": 2.62276776589071e-07, "loss": 0.216, "step": 26025 }, { "epoch": 1.7992395437262356, "grad_norm": 0.6858862260362798, "learning_rate": 2.620978941084451e-07, "loss": 0.2808, "step": 26026 }, { "epoch": 1.7993086761147596, "grad_norm": 0.779686663058916, "learning_rate": 2.6191907100896587e-07, "loss": 0.3452, "step": 26027 }, { "epoch": 1.7993778085032837, "grad_norm": 0.6423914119920329, "learning_rate": 2.617403072928765e-07, "loss": 0.2215, "step": 26028 }, { "epoch": 1.7994469408918077, "grad_norm": 0.688554945641841, "learning_rate": 2.6156160296241573e-07, "loss": 0.245, "step": 26029 }, { "epoch": 1.7995160732803317, "grad_norm": 0.730887515523313, "learning_rate": 2.613829580198246e-07, "loss": 0.3597, "step": 26030 }, { "epoch": 1.7995852056688557, "grad_norm": 0.7082577789691438, "learning_rate": 2.6120437246734134e-07, "loss": 0.2398, "step": 26031 }, { "epoch": 1.7996543380573797, "grad_norm": 0.7154127058185459, "learning_rate": 2.6102584630720463e-07, "loss": 0.2728, "step": 26032 }, { "epoch": 1.7997234704459038, "grad_norm": 0.7191159993676058, "learning_rate": 2.608473795416516e-07, "loss": 0.3066, "step": 26033 }, { "epoch": 1.7997926028344278, "grad_norm": 0.7894750221743453, "learning_rate": 2.6066897217291997e-07, "loss": 0.3176, "step": 26034 }, { "epoch": 1.799861735222952, "grad_norm": 0.6532073982299473, "learning_rate": 2.6049062420324456e-07, "loss": 0.2537, "step": 26035 }, { "epoch": 1.799930867611476, "grad_norm": 0.6658458805824917, "learning_rate": 2.603123356348619e-07, "loss": 0.215, "step": 26036 }, { "epoch": 1.8, "grad_norm": 0.6807879429006474, "learning_rate": 2.601341064700047e-07, "loss": 0.3292, "step": 26037 }, { "epoch": 1.800069132388524, "grad_norm": 0.7091498454105769, "learning_rate": 2.5995593671090833e-07, "loss": 0.2472, "step": 26038 }, { "epoch": 1.800138264777048, "grad_norm": 0.6945194486228141, "learning_rate": 2.59777826359805e-07, "loss": 0.2379, "step": 26039 }, { "epoch": 1.800207397165572, "grad_norm": 0.699618525672136, "learning_rate": 2.5959977541892725e-07, "loss": 0.2206, "step": 26040 }, { "epoch": 1.8002765295540961, "grad_norm": 0.7368848349030624, "learning_rate": 2.5942178389050663e-07, "loss": 0.403, "step": 26041 }, { "epoch": 1.8003456619426201, "grad_norm": 1.0232313883821695, "learning_rate": 2.592438517767748e-07, "loss": 0.2419, "step": 26042 }, { "epoch": 1.8004147943311442, "grad_norm": 0.7160188696583011, "learning_rate": 2.5906597907996047e-07, "loss": 0.2671, "step": 26043 }, { "epoch": 1.8004839267196682, "grad_norm": 0.7682333316972056, "learning_rate": 2.58888165802294e-07, "loss": 0.289, "step": 26044 }, { "epoch": 1.8005530591081922, "grad_norm": 0.6982584612910318, "learning_rate": 2.587104119460027e-07, "loss": 0.2517, "step": 26045 }, { "epoch": 1.8006221914967162, "grad_norm": 0.6556398061319033, "learning_rate": 2.5853271751331566e-07, "loss": 0.2298, "step": 26046 }, { "epoch": 1.8006913238852402, "grad_norm": 0.6078552323329967, "learning_rate": 2.5835508250645904e-07, "loss": 0.2148, "step": 26047 }, { "epoch": 1.8007604562737642, "grad_norm": 0.6312599018370523, "learning_rate": 2.5817750692766043e-07, "loss": 0.2799, "step": 26048 }, { "epoch": 1.8008295886622883, "grad_norm": 0.698856477002601, "learning_rate": 2.579999907791436e-07, "loss": 0.2634, "step": 26049 }, { "epoch": 1.8008987210508123, "grad_norm": 0.6377622728415308, "learning_rate": 2.578225340631352e-07, "loss": 0.1857, "step": 26050 }, { "epoch": 1.8009678534393363, "grad_norm": 0.7216857737163604, "learning_rate": 2.5764513678185833e-07, "loss": 0.3004, "step": 26051 }, { "epoch": 1.8010369858278603, "grad_norm": 0.7382836942922895, "learning_rate": 2.574677989375368e-07, "loss": 0.3183, "step": 26052 }, { "epoch": 1.8011061182163843, "grad_norm": 0.6587818534583035, "learning_rate": 2.572905205323922e-07, "loss": 0.2291, "step": 26053 }, { "epoch": 1.8011752506049084, "grad_norm": 0.7411092755984029, "learning_rate": 2.571133015686478e-07, "loss": 0.2085, "step": 26054 }, { "epoch": 1.8012443829934324, "grad_norm": 0.8710393669756534, "learning_rate": 2.5693614204852335e-07, "loss": 0.3804, "step": 26055 }, { "epoch": 1.8013135153819566, "grad_norm": 1.9655928910334708, "learning_rate": 2.567590419742411e-07, "loss": 0.2784, "step": 26056 }, { "epoch": 1.8013826477704806, "grad_norm": 0.6560711399707796, "learning_rate": 2.5658200134801915e-07, "loss": 0.1985, "step": 26057 }, { "epoch": 1.8014517801590046, "grad_norm": 0.6354449577867715, "learning_rate": 2.564050201720769e-07, "loss": 0.2699, "step": 26058 }, { "epoch": 1.8015209125475287, "grad_norm": 0.6725271332503714, "learning_rate": 2.5622809844863205e-07, "loss": 0.3083, "step": 26059 }, { "epoch": 1.8015900449360527, "grad_norm": 0.8032904534030859, "learning_rate": 2.560512361799028e-07, "loss": 0.2896, "step": 26060 }, { "epoch": 1.8016591773245767, "grad_norm": 0.6870490466944148, "learning_rate": 2.55874433368104e-07, "loss": 0.1891, "step": 26061 }, { "epoch": 1.8017283097131007, "grad_norm": 0.6909011001496473, "learning_rate": 2.556976900154545e-07, "loss": 0.3552, "step": 26062 }, { "epoch": 1.8017974421016247, "grad_norm": 0.7241981554008614, "learning_rate": 2.55521006124167e-07, "loss": 0.2406, "step": 26063 }, { "epoch": 1.8018665744901488, "grad_norm": 0.7477443342738246, "learning_rate": 2.5534438169645793e-07, "loss": 0.2387, "step": 26064 }, { "epoch": 1.8019357068786728, "grad_norm": 0.807846298964488, "learning_rate": 2.551678167345384e-07, "loss": 0.3273, "step": 26065 }, { "epoch": 1.8020048392671968, "grad_norm": 0.7001640683636587, "learning_rate": 2.549913112406238e-07, "loss": 0.3878, "step": 26066 }, { "epoch": 1.8020739716557208, "grad_norm": 0.7135875976409587, "learning_rate": 2.5481486521692467e-07, "loss": 0.2614, "step": 26067 }, { "epoch": 1.8021431040442448, "grad_norm": 0.7982610851155969, "learning_rate": 2.5463847866565306e-07, "loss": 0.2482, "step": 26068 }, { "epoch": 1.8022122364327688, "grad_norm": 0.6883958816969284, "learning_rate": 2.5446215158901944e-07, "loss": 0.266, "step": 26069 }, { "epoch": 1.8022813688212929, "grad_norm": 0.7876904564295863, "learning_rate": 2.5428588398923483e-07, "loss": 0.2765, "step": 26070 }, { "epoch": 1.8023505012098169, "grad_norm": 0.7072781872062204, "learning_rate": 2.541096758685063e-07, "loss": 0.1907, "step": 26071 }, { "epoch": 1.802419633598341, "grad_norm": 0.63499399389672, "learning_rate": 2.5393352722904494e-07, "loss": 0.2285, "step": 26072 }, { "epoch": 1.802488765986865, "grad_norm": 0.6950981659215384, "learning_rate": 2.5375743807305674e-07, "loss": 0.3384, "step": 26073 }, { "epoch": 1.802557898375389, "grad_norm": 0.9578369957433565, "learning_rate": 2.535814084027488e-07, "loss": 0.228, "step": 26074 }, { "epoch": 1.802627030763913, "grad_norm": 0.6600828998869738, "learning_rate": 2.5340543822032715e-07, "loss": 0.2159, "step": 26075 }, { "epoch": 1.802696163152437, "grad_norm": 0.6798157565992131, "learning_rate": 2.532295275279972e-07, "loss": 0.3056, "step": 26076 }, { "epoch": 1.802765295540961, "grad_norm": 0.7271759916794598, "learning_rate": 2.5305367632796565e-07, "loss": 0.2797, "step": 26077 }, { "epoch": 1.802834427929485, "grad_norm": 0.7672222226422303, "learning_rate": 2.528778846224339e-07, "loss": 0.2203, "step": 26078 }, { "epoch": 1.802903560318009, "grad_norm": 0.6670012754391794, "learning_rate": 2.52702152413607e-07, "loss": 0.2327, "step": 26079 }, { "epoch": 1.802972692706533, "grad_norm": 0.7763127827010705, "learning_rate": 2.5252647970368584e-07, "loss": 0.3047, "step": 26080 }, { "epoch": 1.803041825095057, "grad_norm": 4.032446283948727, "learning_rate": 2.5235086649487315e-07, "loss": 0.2875, "step": 26081 }, { "epoch": 1.803110957483581, "grad_norm": 0.6810730936971543, "learning_rate": 2.5217531278936945e-07, "loss": 0.2542, "step": 26082 }, { "epoch": 1.803180089872105, "grad_norm": 0.7384978432092789, "learning_rate": 2.519998185893763e-07, "loss": 0.2833, "step": 26083 }, { "epoch": 1.8032492222606291, "grad_norm": 0.7358117638802886, "learning_rate": 2.5182438389709186e-07, "loss": 0.3517, "step": 26084 }, { "epoch": 1.8033183546491531, "grad_norm": 0.6771076215575003, "learning_rate": 2.5164900871471556e-07, "loss": 0.2127, "step": 26085 }, { "epoch": 1.8033874870376772, "grad_norm": 0.7055455561590003, "learning_rate": 2.5147369304444447e-07, "loss": 0.2051, "step": 26086 }, { "epoch": 1.8034566194262012, "grad_norm": 0.6649032552084236, "learning_rate": 2.5129843688847745e-07, "loss": 0.3403, "step": 26087 }, { "epoch": 1.8035257518147252, "grad_norm": 0.6842776567929415, "learning_rate": 2.5112324024900936e-07, "loss": 0.2648, "step": 26088 }, { "epoch": 1.8035948842032492, "grad_norm": 0.6636023888276544, "learning_rate": 2.5094810312823624e-07, "loss": 0.2495, "step": 26089 }, { "epoch": 1.8036640165917732, "grad_norm": 0.7003334220238577, "learning_rate": 2.5077302552835414e-07, "loss": 0.2638, "step": 26090 }, { "epoch": 1.8037331489802972, "grad_norm": 0.7128143323450007, "learning_rate": 2.505980074515568e-07, "loss": 0.3499, "step": 26091 }, { "epoch": 1.8038022813688213, "grad_norm": 0.8756833273525592, "learning_rate": 2.5042304890003755e-07, "loss": 0.2455, "step": 26092 }, { "epoch": 1.8038714137573453, "grad_norm": 1.1370424096590244, "learning_rate": 2.502481498759901e-07, "loss": 0.2096, "step": 26093 }, { "epoch": 1.8039405461458693, "grad_norm": 0.669114862396524, "learning_rate": 2.5007331038160497e-07, "loss": 0.3154, "step": 26094 }, { "epoch": 1.8040096785343933, "grad_norm": 0.7082037814714892, "learning_rate": 2.498985304190754e-07, "loss": 0.297, "step": 26095 }, { "epoch": 1.8040788109229173, "grad_norm": 0.6684474597197748, "learning_rate": 2.4972380999058856e-07, "loss": 0.214, "step": 26096 }, { "epoch": 1.8041479433114413, "grad_norm": 0.6528503684873985, "learning_rate": 2.495491490983387e-07, "loss": 0.262, "step": 26097 }, { "epoch": 1.8042170756999654, "grad_norm": 0.7036862188644404, "learning_rate": 2.493745477445114e-07, "loss": 0.3464, "step": 26098 }, { "epoch": 1.8042862080884894, "grad_norm": 0.840824364520053, "learning_rate": 2.492000059312977e-07, "loss": 0.24, "step": 26099 }, { "epoch": 1.8043553404770134, "grad_norm": 0.6545395618551428, "learning_rate": 2.490255236608824e-07, "loss": 0.2233, "step": 26100 }, { "epoch": 1.8044244728655374, "grad_norm": 0.752246891470308, "learning_rate": 2.488511009354544e-07, "loss": 0.2829, "step": 26101 }, { "epoch": 1.8044936052540614, "grad_norm": 0.7046966372847254, "learning_rate": 2.486767377571986e-07, "loss": 0.2963, "step": 26102 }, { "epoch": 1.8045627376425855, "grad_norm": 0.7465362272633185, "learning_rate": 2.485024341283004e-07, "loss": 0.3233, "step": 26103 }, { "epoch": 1.8046318700311095, "grad_norm": 0.6713406948060081, "learning_rate": 2.4832819005094543e-07, "loss": 0.2068, "step": 26104 }, { "epoch": 1.8047010024196335, "grad_norm": 0.7421028549101302, "learning_rate": 2.4815400552731685e-07, "loss": 0.3768, "step": 26105 }, { "epoch": 1.8047701348081575, "grad_norm": 0.7003128902227271, "learning_rate": 2.479798805595973e-07, "loss": 0.2733, "step": 26106 }, { "epoch": 1.8048392671966815, "grad_norm": 0.6815214690016574, "learning_rate": 2.478058151499707e-07, "loss": 0.2424, "step": 26107 }, { "epoch": 1.8049083995852055, "grad_norm": 0.727917079994915, "learning_rate": 2.476318093006164e-07, "loss": 0.2859, "step": 26108 }, { "epoch": 1.8049775319737296, "grad_norm": 0.7049952776123716, "learning_rate": 2.47457863013717e-07, "loss": 0.3948, "step": 26109 }, { "epoch": 1.8050466643622536, "grad_norm": 0.689608442995413, "learning_rate": 2.472839762914514e-07, "loss": 0.2101, "step": 26110 }, { "epoch": 1.8051157967507776, "grad_norm": 0.6476833044851781, "learning_rate": 2.471101491359995e-07, "loss": 0.1769, "step": 26111 }, { "epoch": 1.8051849291393016, "grad_norm": 0.7868599719454774, "learning_rate": 2.469363815495401e-07, "loss": 0.3209, "step": 26112 }, { "epoch": 1.8052540615278256, "grad_norm": 0.9552972574388792, "learning_rate": 2.467626735342521e-07, "loss": 0.2603, "step": 26113 }, { "epoch": 1.8053231939163497, "grad_norm": 0.7031543646660905, "learning_rate": 2.465890250923103e-07, "loss": 0.1965, "step": 26114 }, { "epoch": 1.8053923263048737, "grad_norm": 0.8777732500976336, "learning_rate": 2.464154362258925e-07, "loss": 0.2726, "step": 26115 }, { "epoch": 1.8054614586933977, "grad_norm": 0.7271228958696486, "learning_rate": 2.462419069371741e-07, "loss": 0.3365, "step": 26116 }, { "epoch": 1.805530591081922, "grad_norm": 1.2395271304335345, "learning_rate": 2.460684372283295e-07, "loss": 0.2805, "step": 26117 }, { "epoch": 1.805599723470446, "grad_norm": 0.7247966486397011, "learning_rate": 2.458950271015331e-07, "loss": 0.2404, "step": 26118 }, { "epoch": 1.80566885585897, "grad_norm": 0.6857611424953074, "learning_rate": 2.4572167655895985e-07, "loss": 0.3014, "step": 26119 }, { "epoch": 1.805737988247494, "grad_norm": 0.7386860337431649, "learning_rate": 2.455483856027796e-07, "loss": 0.332, "step": 26120 }, { "epoch": 1.805807120636018, "grad_norm": 0.6283301080694325, "learning_rate": 2.453751542351668e-07, "loss": 0.2347, "step": 26121 }, { "epoch": 1.805876253024542, "grad_norm": 0.7192849359317135, "learning_rate": 2.452019824582902e-07, "loss": 0.2385, "step": 26122 }, { "epoch": 1.805945385413066, "grad_norm": 0.7033135738594275, "learning_rate": 2.450288702743225e-07, "loss": 0.3576, "step": 26123 }, { "epoch": 1.80601451780159, "grad_norm": 0.742215664644622, "learning_rate": 2.4485581768543156e-07, "loss": 0.2599, "step": 26124 }, { "epoch": 1.806083650190114, "grad_norm": 0.7001630888383695, "learning_rate": 2.4468282469378713e-07, "loss": 0.3482, "step": 26125 }, { "epoch": 1.806152782578638, "grad_norm": 0.7042563053893598, "learning_rate": 2.44509891301557e-07, "loss": 0.2916, "step": 26126 }, { "epoch": 1.8062219149671621, "grad_norm": 0.6971201547415582, "learning_rate": 2.443370175109094e-07, "loss": 0.3335, "step": 26127 }, { "epoch": 1.8062910473556861, "grad_norm": 0.7313516514580383, "learning_rate": 2.4416420332401047e-07, "loss": 0.2613, "step": 26128 }, { "epoch": 1.8063601797442101, "grad_norm": 0.7240542503184477, "learning_rate": 2.4399144874302616e-07, "loss": 0.2419, "step": 26129 }, { "epoch": 1.8064293121327342, "grad_norm": 0.7049209945337668, "learning_rate": 2.4381875377012087e-07, "loss": 0.3314, "step": 26130 }, { "epoch": 1.8064984445212582, "grad_norm": 0.739889898882747, "learning_rate": 2.436461184074607e-07, "loss": 0.2834, "step": 26131 }, { "epoch": 1.8065675769097822, "grad_norm": 0.8092240642924927, "learning_rate": 2.434735426572066e-07, "loss": 0.2285, "step": 26132 }, { "epoch": 1.8066367092983062, "grad_norm": 0.6631740532356883, "learning_rate": 2.433010265215247e-07, "loss": 0.2721, "step": 26133 }, { "epoch": 1.8067058416868302, "grad_norm": 0.7492878693405429, "learning_rate": 2.431285700025754e-07, "loss": 0.3232, "step": 26134 }, { "epoch": 1.8067749740753543, "grad_norm": 0.6615919357687137, "learning_rate": 2.4295617310252094e-07, "loss": 0.2269, "step": 26135 }, { "epoch": 1.8068441064638783, "grad_norm": 0.7170290154868989, "learning_rate": 2.4278383582352074e-07, "loss": 0.2435, "step": 26136 }, { "epoch": 1.8069132388524025, "grad_norm": 1.5158944057756276, "learning_rate": 2.426115581677363e-07, "loss": 0.3355, "step": 26137 }, { "epoch": 1.8069823712409265, "grad_norm": 0.7107286845064782, "learning_rate": 2.4243934013732483e-07, "loss": 0.3046, "step": 26138 }, { "epoch": 1.8070515036294506, "grad_norm": 0.747249317609931, "learning_rate": 2.4226718173444687e-07, "loss": 0.2251, "step": 26139 }, { "epoch": 1.8071206360179746, "grad_norm": 0.7620462619576948, "learning_rate": 2.420950829612595e-07, "loss": 0.2361, "step": 26140 }, { "epoch": 1.8071897684064986, "grad_norm": 0.7065928150518295, "learning_rate": 2.4192304381991827e-07, "loss": 0.3259, "step": 26141 }, { "epoch": 1.8072589007950226, "grad_norm": 0.7589478004252067, "learning_rate": 2.41751064312582e-07, "loss": 0.2696, "step": 26142 }, { "epoch": 1.8073280331835466, "grad_norm": 0.6934604788551034, "learning_rate": 2.415791444414034e-07, "loss": 0.2711, "step": 26143 }, { "epoch": 1.8073971655720706, "grad_norm": 1.0006596498499678, "learning_rate": 2.4140728420853965e-07, "loss": 0.284, "step": 26144 }, { "epoch": 1.8074662979605947, "grad_norm": 0.7184449003065755, "learning_rate": 2.4123548361614233e-07, "loss": 0.3344, "step": 26145 }, { "epoch": 1.8075354303491187, "grad_norm": 2.619975978870857, "learning_rate": 2.410637426663659e-07, "loss": 0.2368, "step": 26146 }, { "epoch": 1.8076045627376427, "grad_norm": 0.7294827553215175, "learning_rate": 2.40892061361363e-07, "loss": 0.2379, "step": 26147 }, { "epoch": 1.8076736951261667, "grad_norm": 0.9505376705419382, "learning_rate": 2.4072043970328585e-07, "loss": 0.4009, "step": 26148 }, { "epoch": 1.8077428275146907, "grad_norm": 0.7474276439430853, "learning_rate": 2.405488776942838e-07, "loss": 0.2183, "step": 26149 }, { "epoch": 1.8078119599032147, "grad_norm": 0.6817615936886257, "learning_rate": 2.403773753365085e-07, "loss": 0.2443, "step": 26150 }, { "epoch": 1.8078810922917388, "grad_norm": 0.7132557911655959, "learning_rate": 2.4020593263210825e-07, "loss": 0.3039, "step": 26151 }, { "epoch": 1.8079502246802628, "grad_norm": 0.7694309923983533, "learning_rate": 2.4003454958323237e-07, "loss": 0.3481, "step": 26152 }, { "epoch": 1.8080193570687868, "grad_norm": 1.0383592471106813, "learning_rate": 2.398632261920292e-07, "loss": 0.2189, "step": 26153 }, { "epoch": 1.8080884894573108, "grad_norm": 0.6671595787088435, "learning_rate": 2.3969196246064587e-07, "loss": 0.2549, "step": 26154 }, { "epoch": 1.8081576218458348, "grad_norm": 0.8741062040352604, "learning_rate": 2.395207583912279e-07, "loss": 0.3181, "step": 26155 }, { "epoch": 1.8082267542343589, "grad_norm": 0.7229365779755198, "learning_rate": 2.3934961398592307e-07, "loss": 0.2519, "step": 26156 }, { "epoch": 1.8082958866228829, "grad_norm": 0.6895795206958697, "learning_rate": 2.39178529246874e-07, "loss": 0.2833, "step": 26157 }, { "epoch": 1.808365019011407, "grad_norm": 0.7502578175659254, "learning_rate": 2.3900750417622686e-07, "loss": 0.2553, "step": 26158 }, { "epoch": 1.808434151399931, "grad_norm": 0.6979895091446617, "learning_rate": 2.3883653877612324e-07, "loss": 0.2866, "step": 26159 }, { "epoch": 1.808503283788455, "grad_norm": 0.8014047779268908, "learning_rate": 2.3866563304870693e-07, "loss": 0.2714, "step": 26160 }, { "epoch": 1.808572416176979, "grad_norm": 0.719718562100475, "learning_rate": 2.384947869961202e-07, "loss": 0.2702, "step": 26161 }, { "epoch": 1.808641548565503, "grad_norm": 0.8045005497287906, "learning_rate": 2.3832400062050453e-07, "loss": 0.3045, "step": 26162 }, { "epoch": 1.808710680954027, "grad_norm": 0.7533251244595013, "learning_rate": 2.3815327392399946e-07, "loss": 0.271, "step": 26163 }, { "epoch": 1.808779813342551, "grad_norm": 0.8499879634870406, "learning_rate": 2.3798260690874596e-07, "loss": 0.2271, "step": 26164 }, { "epoch": 1.808848945731075, "grad_norm": 0.7082819163005433, "learning_rate": 2.3781199957688127e-07, "loss": 0.2782, "step": 26165 }, { "epoch": 1.808918078119599, "grad_norm": 0.7426154903449173, "learning_rate": 2.376414519305459e-07, "loss": 0.3525, "step": 26166 }, { "epoch": 1.808987210508123, "grad_norm": 0.6425770475002668, "learning_rate": 2.3747096397187475e-07, "loss": 0.2093, "step": 26167 }, { "epoch": 1.809056342896647, "grad_norm": 0.6751445952602894, "learning_rate": 2.3730053570300616e-07, "loss": 0.247, "step": 26168 }, { "epoch": 1.809125475285171, "grad_norm": 0.641742642746079, "learning_rate": 2.3713016712607617e-07, "loss": 0.2871, "step": 26169 }, { "epoch": 1.809194607673695, "grad_norm": 0.7168830967095571, "learning_rate": 2.369598582432203e-07, "loss": 0.2891, "step": 26170 }, { "epoch": 1.8092637400622191, "grad_norm": 0.7672571485191735, "learning_rate": 2.3678960905657188e-07, "loss": 0.264, "step": 26171 }, { "epoch": 1.8093328724507431, "grad_norm": 0.7791572800433285, "learning_rate": 2.3661941956826585e-07, "loss": 0.2369, "step": 26172 }, { "epoch": 1.8094020048392672, "grad_norm": 0.7057766156482909, "learning_rate": 2.3644928978043436e-07, "loss": 0.347, "step": 26173 }, { "epoch": 1.8094711372277912, "grad_norm": 0.7441568083620964, "learning_rate": 2.3627921969521017e-07, "loss": 0.2296, "step": 26174 }, { "epoch": 1.8095402696163152, "grad_norm": 0.6168430865635123, "learning_rate": 2.3610920931472492e-07, "loss": 0.1992, "step": 26175 }, { "epoch": 1.8096094020048392, "grad_norm": 0.7017495385005235, "learning_rate": 2.3593925864110967e-07, "loss": 0.2213, "step": 26176 }, { "epoch": 1.8096785343933632, "grad_norm": 0.7007637542377189, "learning_rate": 2.3576936767649272e-07, "loss": 0.2775, "step": 26177 }, { "epoch": 1.8097476667818873, "grad_norm": 0.8137189141565511, "learning_rate": 2.3559953642300572e-07, "loss": 0.2527, "step": 26178 }, { "epoch": 1.8098167991704113, "grad_norm": 0.717775160016388, "learning_rate": 2.354297648827758e-07, "loss": 0.2296, "step": 26179 }, { "epoch": 1.8098859315589353, "grad_norm": 0.704447931647379, "learning_rate": 2.3526005305793132e-07, "loss": 0.3431, "step": 26180 }, { "epoch": 1.8099550639474593, "grad_norm": 0.7067787798219898, "learning_rate": 2.350904009505983e-07, "loss": 0.2671, "step": 26181 }, { "epoch": 1.8100241963359833, "grad_norm": 0.7103224444468831, "learning_rate": 2.3492080856290346e-07, "loss": 0.2905, "step": 26182 }, { "epoch": 1.8100933287245073, "grad_norm": 0.753704275443752, "learning_rate": 2.3475127589697277e-07, "loss": 0.2518, "step": 26183 }, { "epoch": 1.8101624611130314, "grad_norm": 0.6954272543529885, "learning_rate": 2.3458180295493181e-07, "loss": 0.3062, "step": 26184 }, { "epoch": 1.8102315935015554, "grad_norm": 3.279143300799826, "learning_rate": 2.344123897389028e-07, "loss": 0.2518, "step": 26185 }, { "epoch": 1.8103007258900794, "grad_norm": 0.7002264059309491, "learning_rate": 2.3424303625101119e-07, "loss": 0.2462, "step": 26186 }, { "epoch": 1.8103698582786034, "grad_norm": 0.7029474826395277, "learning_rate": 2.34073742493377e-07, "loss": 0.3602, "step": 26187 }, { "epoch": 1.8104389906671274, "grad_norm": 0.7452306455416915, "learning_rate": 2.3390450846812297e-07, "loss": 0.2455, "step": 26188 }, { "epoch": 1.8105081230556515, "grad_norm": 0.6429151544427442, "learning_rate": 2.337353341773707e-07, "loss": 0.2169, "step": 26189 }, { "epoch": 1.8105772554441755, "grad_norm": 0.7037734752015101, "learning_rate": 2.3356621962324133e-07, "loss": 0.2671, "step": 26190 }, { "epoch": 1.8106463878326995, "grad_norm": 0.6838302043474017, "learning_rate": 2.33397164807852e-07, "loss": 0.3578, "step": 26191 }, { "epoch": 1.8107155202212235, "grad_norm": 0.7419576878843037, "learning_rate": 2.3322816973332386e-07, "loss": 0.2073, "step": 26192 }, { "epoch": 1.8107846526097475, "grad_norm": 0.6118501350915637, "learning_rate": 2.3305923440177348e-07, "loss": 0.1755, "step": 26193 }, { "epoch": 1.8108537849982715, "grad_norm": 0.7297532877337295, "learning_rate": 2.3289035881531918e-07, "loss": 0.2925, "step": 26194 }, { "epoch": 1.8109229173867956, "grad_norm": 0.7168618605874564, "learning_rate": 2.3272154297607597e-07, "loss": 0.2942, "step": 26195 }, { "epoch": 1.8109920497753196, "grad_norm": 0.790609538126091, "learning_rate": 2.3255278688616102e-07, "loss": 0.2322, "step": 26196 }, { "epoch": 1.8110611821638436, "grad_norm": 0.7610415519129853, "learning_rate": 2.3238409054768818e-07, "loss": 0.2831, "step": 26197 }, { "epoch": 1.8111303145523678, "grad_norm": 0.673341496387256, "learning_rate": 2.3221545396277412e-07, "loss": 0.3098, "step": 26198 }, { "epoch": 1.8111994469408919, "grad_norm": 0.7129376493444727, "learning_rate": 2.3204687713352992e-07, "loss": 0.2688, "step": 26199 }, { "epoch": 1.8112685793294159, "grad_norm": 0.7327825713221394, "learning_rate": 2.3187836006206998e-07, "loss": 0.1948, "step": 26200 }, { "epoch": 1.81133771171794, "grad_norm": 0.6689045119046677, "learning_rate": 2.3170990275050488e-07, "loss": 0.2916, "step": 26201 }, { "epoch": 1.811406844106464, "grad_norm": 0.6933867962543445, "learning_rate": 2.3154150520094732e-07, "loss": 0.3552, "step": 26202 }, { "epoch": 1.811475976494988, "grad_norm": 0.7414467624901284, "learning_rate": 2.313731674155062e-07, "loss": 0.2175, "step": 26203 }, { "epoch": 1.811545108883512, "grad_norm": 0.7066829621944783, "learning_rate": 2.3120488939629371e-07, "loss": 0.2164, "step": 26204 }, { "epoch": 1.811614241272036, "grad_norm": 0.7172971924214326, "learning_rate": 2.3103667114541761e-07, "loss": 0.3274, "step": 26205 }, { "epoch": 1.81168337366056, "grad_norm": 0.7474515622628672, "learning_rate": 2.3086851266498567e-07, "loss": 0.2729, "step": 26206 }, { "epoch": 1.811752506049084, "grad_norm": 0.7149947838385063, "learning_rate": 2.3070041395710674e-07, "loss": 0.2291, "step": 26207 }, { "epoch": 1.811821638437608, "grad_norm": 0.7249404357290813, "learning_rate": 2.3053237502388636e-07, "loss": 0.2609, "step": 26208 }, { "epoch": 1.811890770826132, "grad_norm": 0.6216186918757611, "learning_rate": 2.3036439586743065e-07, "loss": 0.3241, "step": 26209 }, { "epoch": 1.811959903214656, "grad_norm": 0.6819809294707984, "learning_rate": 2.3019647648984623e-07, "loss": 0.228, "step": 26210 }, { "epoch": 1.81202903560318, "grad_norm": 0.9334569944883953, "learning_rate": 2.3002861689323696e-07, "loss": 0.2114, "step": 26211 }, { "epoch": 1.812098167991704, "grad_norm": 0.7341135299311793, "learning_rate": 2.2986081707970564e-07, "loss": 0.3307, "step": 26212 }, { "epoch": 1.812167300380228, "grad_norm": 1.1773528575107879, "learning_rate": 2.2969307705135778e-07, "loss": 0.3076, "step": 26213 }, { "epoch": 1.8122364327687521, "grad_norm": 0.8196253967850925, "learning_rate": 2.2952539681029283e-07, "loss": 0.2243, "step": 26214 }, { "epoch": 1.8123055651572761, "grad_norm": 0.8008522123875951, "learning_rate": 2.2935777635861468e-07, "loss": 0.2773, "step": 26215 }, { "epoch": 1.8123746975458002, "grad_norm": 0.6758053570023967, "learning_rate": 2.291902156984227e-07, "loss": 0.2934, "step": 26216 }, { "epoch": 1.8124438299343242, "grad_norm": 0.7301089102260693, "learning_rate": 2.2902271483181692e-07, "loss": 0.221, "step": 26217 }, { "epoch": 1.8125129623228484, "grad_norm": 0.7603044657875144, "learning_rate": 2.2885527376089788e-07, "loss": 0.1757, "step": 26218 }, { "epoch": 1.8125820947113724, "grad_norm": 0.7217622564571674, "learning_rate": 2.2868789248776392e-07, "loss": 0.3183, "step": 26219 }, { "epoch": 1.8126512270998965, "grad_norm": 0.8517002776325645, "learning_rate": 2.2852057101451165e-07, "loss": 0.3051, "step": 26220 }, { "epoch": 1.8127203594884205, "grad_norm": 0.6834697464534161, "learning_rate": 2.2835330934323995e-07, "loss": 0.2041, "step": 26221 }, { "epoch": 1.8127894918769445, "grad_norm": 0.6516486409692567, "learning_rate": 2.2818610747604275e-07, "loss": 0.214, "step": 26222 }, { "epoch": 1.8128586242654685, "grad_norm": 0.6739619957539542, "learning_rate": 2.2801896541501777e-07, "loss": 0.3076, "step": 26223 }, { "epoch": 1.8129277566539925, "grad_norm": 0.7148848742494424, "learning_rate": 2.2785188316225837e-07, "loss": 0.2114, "step": 26224 }, { "epoch": 1.8129968890425165, "grad_norm": 0.6852560178399036, "learning_rate": 2.2768486071986062e-07, "loss": 0.213, "step": 26225 }, { "epoch": 1.8130660214310406, "grad_norm": 0.7414730833240933, "learning_rate": 2.2751789808991566e-07, "loss": 0.2959, "step": 26226 }, { "epoch": 1.8131351538195646, "grad_norm": 1.4729686294845077, "learning_rate": 2.2735099527451786e-07, "loss": 0.2744, "step": 26227 }, { "epoch": 1.8132042862080886, "grad_norm": 0.7153603977456422, "learning_rate": 2.271841522757573e-07, "loss": 0.2172, "step": 26228 }, { "epoch": 1.8132734185966126, "grad_norm": 0.7204950738930161, "learning_rate": 2.2701736909572615e-07, "loss": 0.2039, "step": 26229 }, { "epoch": 1.8133425509851366, "grad_norm": 0.6875205767022395, "learning_rate": 2.268506457365144e-07, "loss": 0.268, "step": 26230 }, { "epoch": 1.8134116833736607, "grad_norm": 0.7549269267623676, "learning_rate": 2.266839822002115e-07, "loss": 0.2428, "step": 26231 }, { "epoch": 1.8134808157621847, "grad_norm": 0.7369610004762296, "learning_rate": 2.2651737848890687e-07, "loss": 0.2092, "step": 26232 }, { "epoch": 1.8135499481507087, "grad_norm": 0.7082008879095089, "learning_rate": 2.2635083460468832e-07, "loss": 0.2645, "step": 26233 }, { "epoch": 1.8136190805392327, "grad_norm": 0.6977610018063463, "learning_rate": 2.2618435054964305e-07, "loss": 0.3313, "step": 26234 }, { "epoch": 1.8136882129277567, "grad_norm": 0.8217496752275094, "learning_rate": 2.260179263258583e-07, "loss": 0.2216, "step": 26235 }, { "epoch": 1.8137573453162807, "grad_norm": 0.708791189882003, "learning_rate": 2.258515619354179e-07, "loss": 0.2515, "step": 26236 }, { "epoch": 1.8138264777048048, "grad_norm": 0.7207148584935579, "learning_rate": 2.256852573804097e-07, "loss": 0.3579, "step": 26237 }, { "epoch": 1.8138956100933288, "grad_norm": 0.7065877345369608, "learning_rate": 2.2551901266291587e-07, "loss": 0.2341, "step": 26238 }, { "epoch": 1.8139647424818528, "grad_norm": 0.6733382399513392, "learning_rate": 2.2535282778502032e-07, "loss": 0.2257, "step": 26239 }, { "epoch": 1.8140338748703768, "grad_norm": 0.6802068488297388, "learning_rate": 2.2518670274880695e-07, "loss": 0.207, "step": 26240 }, { "epoch": 1.8141030072589008, "grad_norm": 0.7596738641504077, "learning_rate": 2.2502063755635739e-07, "loss": 0.3539, "step": 26241 }, { "epoch": 1.8141721396474249, "grad_norm": 0.7956138585218813, "learning_rate": 2.2485463220975278e-07, "loss": 0.2727, "step": 26242 }, { "epoch": 1.8142412720359489, "grad_norm": 0.8854375042079466, "learning_rate": 2.2468868671107368e-07, "loss": 0.2439, "step": 26243 }, { "epoch": 1.8143104044244729, "grad_norm": 0.8512182500670553, "learning_rate": 2.2452280106239954e-07, "loss": 0.3032, "step": 26244 }, { "epoch": 1.814379536812997, "grad_norm": 0.7680075829349152, "learning_rate": 2.243569752658098e-07, "loss": 0.3278, "step": 26245 }, { "epoch": 1.814448669201521, "grad_norm": 0.7059507636052006, "learning_rate": 2.2419120932338335e-07, "loss": 0.2391, "step": 26246 }, { "epoch": 1.814517801590045, "grad_norm": 0.7211996354856536, "learning_rate": 2.240255032371974e-07, "loss": 0.2351, "step": 26247 }, { "epoch": 1.814586933978569, "grad_norm": 0.7334307146832737, "learning_rate": 2.2385985700932811e-07, "loss": 0.3373, "step": 26248 }, { "epoch": 1.814656066367093, "grad_norm": 0.7432159484644885, "learning_rate": 2.2369427064185324e-07, "loss": 0.229, "step": 26249 }, { "epoch": 1.814725198755617, "grad_norm": 0.666675855642267, "learning_rate": 2.2352874413684554e-07, "loss": 0.1996, "step": 26250 }, { "epoch": 1.814794331144141, "grad_norm": 0.7609224449570793, "learning_rate": 2.233632774963823e-07, "loss": 0.3059, "step": 26251 }, { "epoch": 1.814863463532665, "grad_norm": 0.7365778912674028, "learning_rate": 2.231978707225352e-07, "loss": 0.3068, "step": 26252 }, { "epoch": 1.814932595921189, "grad_norm": 0.6640262471184751, "learning_rate": 2.2303252381737806e-07, "loss": 0.1952, "step": 26253 }, { "epoch": 1.815001728309713, "grad_norm": 0.6537737540196044, "learning_rate": 2.2286723678298372e-07, "loss": 0.2443, "step": 26254 }, { "epoch": 1.815070860698237, "grad_norm": 0.6973955633185114, "learning_rate": 2.2270200962142385e-07, "loss": 0.3028, "step": 26255 }, { "epoch": 1.815139993086761, "grad_norm": 0.6583654179363557, "learning_rate": 2.2253684233476792e-07, "loss": 0.2704, "step": 26256 }, { "epoch": 1.8152091254752851, "grad_norm": 0.6892948364005967, "learning_rate": 2.2237173492508758e-07, "loss": 0.2215, "step": 26257 }, { "epoch": 1.8152782578638091, "grad_norm": 0.64234553405303, "learning_rate": 2.2220668739445118e-07, "loss": 0.2551, "step": 26258 }, { "epoch": 1.8153473902523332, "grad_norm": 0.7180559931997663, "learning_rate": 2.2204169974492762e-07, "loss": 0.3162, "step": 26259 }, { "epoch": 1.8154165226408572, "grad_norm": 0.7232061277463318, "learning_rate": 2.2187677197858414e-07, "loss": 0.2211, "step": 26260 }, { "epoch": 1.8154856550293812, "grad_norm": 0.9362219811400152, "learning_rate": 2.2171190409748967e-07, "loss": 0.2138, "step": 26261 }, { "epoch": 1.8155547874179052, "grad_norm": 0.6721319414956155, "learning_rate": 2.2154709610370806e-07, "loss": 0.3275, "step": 26262 }, { "epoch": 1.8156239198064292, "grad_norm": 0.8179279518577721, "learning_rate": 2.2138234799930714e-07, "loss": 0.3529, "step": 26263 }, { "epoch": 1.8156930521949532, "grad_norm": 0.6458516510200142, "learning_rate": 2.2121765978634968e-07, "loss": 0.2357, "step": 26264 }, { "epoch": 1.8157621845834773, "grad_norm": 0.776812094379799, "learning_rate": 2.210530314669018e-07, "loss": 0.2876, "step": 26265 }, { "epoch": 1.8158313169720013, "grad_norm": 0.6728777594710471, "learning_rate": 2.208884630430247e-07, "loss": 0.3142, "step": 26266 }, { "epoch": 1.8159004493605253, "grad_norm": 0.8358561037525689, "learning_rate": 2.207239545167822e-07, "loss": 0.2381, "step": 26267 }, { "epoch": 1.8159695817490493, "grad_norm": 0.679522046668212, "learning_rate": 2.20559505890236e-07, "loss": 0.2483, "step": 26268 }, { "epoch": 1.8160387141375733, "grad_norm": 0.6908367981396398, "learning_rate": 2.203951171654478e-07, "loss": 0.2769, "step": 26269 }, { "epoch": 1.8161078465260974, "grad_norm": 0.7779777915865437, "learning_rate": 2.2023078834447653e-07, "loss": 0.3222, "step": 26270 }, { "epoch": 1.8161769789146214, "grad_norm": 0.7279438006344443, "learning_rate": 2.2006651942938274e-07, "loss": 0.2189, "step": 26271 }, { "epoch": 1.8162461113031454, "grad_norm": 0.7416300541549519, "learning_rate": 2.1990231042222532e-07, "loss": 0.225, "step": 26272 }, { "epoch": 1.8163152436916694, "grad_norm": 0.70870109751252, "learning_rate": 2.1973816132506153e-07, "loss": 0.3872, "step": 26273 }, { "epoch": 1.8163843760801934, "grad_norm": 0.7290652250587483, "learning_rate": 2.1957407213994864e-07, "loss": 0.2682, "step": 26274 }, { "epoch": 1.8164535084687174, "grad_norm": 0.670286736934611, "learning_rate": 2.1941004286894386e-07, "loss": 0.1813, "step": 26275 }, { "epoch": 1.8165226408572415, "grad_norm": 0.6854915727770231, "learning_rate": 2.1924607351410332e-07, "loss": 0.2609, "step": 26276 }, { "epoch": 1.8165917732457655, "grad_norm": 0.789210398369985, "learning_rate": 2.190821640774815e-07, "loss": 0.3259, "step": 26277 }, { "epoch": 1.8166609056342895, "grad_norm": 0.7646011275578035, "learning_rate": 2.189183145611329e-07, "loss": 0.2446, "step": 26278 }, { "epoch": 1.8167300380228137, "grad_norm": 0.6634773470389346, "learning_rate": 2.187545249671108e-07, "loss": 0.2256, "step": 26279 }, { "epoch": 1.8167991704113378, "grad_norm": 0.6637037747651396, "learning_rate": 2.1859079529746751e-07, "loss": 0.3159, "step": 26280 }, { "epoch": 1.8168683027998618, "grad_norm": 0.7141739903746281, "learning_rate": 2.1842712555425638e-07, "loss": 0.1979, "step": 26281 }, { "epoch": 1.8169374351883858, "grad_norm": 0.61771855855477, "learning_rate": 2.1826351573952908e-07, "loss": 0.1701, "step": 26282 }, { "epoch": 1.8170065675769098, "grad_norm": 0.7274210256281995, "learning_rate": 2.1809996585533398e-07, "loss": 0.2933, "step": 26283 }, { "epoch": 1.8170756999654338, "grad_norm": 0.7556333047023058, "learning_rate": 2.179364759037228e-07, "loss": 0.3012, "step": 26284 }, { "epoch": 1.8171448323539579, "grad_norm": 0.6750812375635017, "learning_rate": 2.177730458867433e-07, "loss": 0.2392, "step": 26285 }, { "epoch": 1.8172139647424819, "grad_norm": 0.5776307453685818, "learning_rate": 2.1760967580644555e-07, "loss": 0.1718, "step": 26286 }, { "epoch": 1.8172830971310059, "grad_norm": 0.6852138740540642, "learning_rate": 2.1744636566487454e-07, "loss": 0.2721, "step": 26287 }, { "epoch": 1.81735222951953, "grad_norm": 0.7289052404459215, "learning_rate": 2.172831154640792e-07, "loss": 0.273, "step": 26288 }, { "epoch": 1.817421361908054, "grad_norm": 0.7099724288368674, "learning_rate": 2.1711992520610458e-07, "loss": 0.2552, "step": 26289 }, { "epoch": 1.817490494296578, "grad_norm": 0.723329760400494, "learning_rate": 2.1695679489299682e-07, "loss": 0.2369, "step": 26290 }, { "epoch": 1.817559626685102, "grad_norm": 0.808782634421098, "learning_rate": 2.1679372452679926e-07, "loss": 0.3294, "step": 26291 }, { "epoch": 1.817628759073626, "grad_norm": 0.7889050095392196, "learning_rate": 2.1663071410955693e-07, "loss": 0.2265, "step": 26292 }, { "epoch": 1.81769789146215, "grad_norm": 0.6446137597901727, "learning_rate": 2.1646776364331212e-07, "loss": 0.2188, "step": 26293 }, { "epoch": 1.817767023850674, "grad_norm": 0.7223664844904555, "learning_rate": 2.1630487313010706e-07, "loss": 0.2743, "step": 26294 }, { "epoch": 1.817836156239198, "grad_norm": 0.7201250307274889, "learning_rate": 2.161420425719829e-07, "loss": 0.3069, "step": 26295 }, { "epoch": 1.817905288627722, "grad_norm": 0.636745719758364, "learning_rate": 2.1597927197098246e-07, "loss": 0.2009, "step": 26296 }, { "epoch": 1.817974421016246, "grad_norm": 0.7237624499620823, "learning_rate": 2.1581656132914353e-07, "loss": 0.2285, "step": 26297 }, { "epoch": 1.81804355340477, "grad_norm": 0.7210991910966474, "learning_rate": 2.156539106485067e-07, "loss": 0.2811, "step": 26298 }, { "epoch": 1.818112685793294, "grad_norm": 0.7717416233598109, "learning_rate": 2.1549131993110983e-07, "loss": 0.2944, "step": 26299 }, { "epoch": 1.8181818181818183, "grad_norm": 0.6695671529927048, "learning_rate": 2.1532878917899125e-07, "loss": 0.2158, "step": 26300 }, { "epoch": 1.8182509505703424, "grad_norm": 0.6888974016711603, "learning_rate": 2.151663183941871e-07, "loss": 0.2869, "step": 26301 }, { "epoch": 1.8183200829588664, "grad_norm": 0.6698027724114239, "learning_rate": 2.1500390757873414e-07, "loss": 0.3114, "step": 26302 }, { "epoch": 1.8183892153473904, "grad_norm": 1.0512814579504903, "learning_rate": 2.1484155673466788e-07, "loss": 0.2894, "step": 26303 }, { "epoch": 1.8184583477359144, "grad_norm": 0.6722619428723499, "learning_rate": 2.14679265864024e-07, "loss": 0.2352, "step": 26304 }, { "epoch": 1.8185274801244384, "grad_norm": 0.6613528951731742, "learning_rate": 2.1451703496883526e-07, "loss": 0.291, "step": 26305 }, { "epoch": 1.8185966125129625, "grad_norm": 0.752434471081517, "learning_rate": 2.1435486405113615e-07, "loss": 0.2928, "step": 26306 }, { "epoch": 1.8186657449014865, "grad_norm": 0.6780812673441158, "learning_rate": 2.141927531129573e-07, "loss": 0.2517, "step": 26307 }, { "epoch": 1.8187348772900105, "grad_norm": 0.7115924573835186, "learning_rate": 2.1403070215633258e-07, "loss": 0.3216, "step": 26308 }, { "epoch": 1.8188040096785345, "grad_norm": 0.7492796555796377, "learning_rate": 2.1386871118329156e-07, "loss": 0.3702, "step": 26309 }, { "epoch": 1.8188731420670585, "grad_norm": 0.6727394120715744, "learning_rate": 2.1370678019586477e-07, "loss": 0.222, "step": 26310 }, { "epoch": 1.8189422744555825, "grad_norm": 0.6101685129234539, "learning_rate": 2.1354490919608173e-07, "loss": 0.1902, "step": 26311 }, { "epoch": 1.8190114068441066, "grad_norm": 0.9843743877237855, "learning_rate": 2.1338309818597247e-07, "loss": 0.3023, "step": 26312 }, { "epoch": 1.8190805392326306, "grad_norm": 0.7799127833375628, "learning_rate": 2.1322134716756314e-07, "loss": 0.2503, "step": 26313 }, { "epoch": 1.8191496716211546, "grad_norm": 0.6074844218266073, "learning_rate": 2.130596561428827e-07, "loss": 0.2231, "step": 26314 }, { "epoch": 1.8192188040096786, "grad_norm": 0.6827132931150999, "learning_rate": 2.1289802511395618e-07, "loss": 0.2567, "step": 26315 }, { "epoch": 1.8192879363982026, "grad_norm": 0.658624000759805, "learning_rate": 2.1273645408280974e-07, "loss": 0.3606, "step": 26316 }, { "epoch": 1.8193570687867266, "grad_norm": 0.738959947813929, "learning_rate": 2.1257494305146897e-07, "loss": 0.2596, "step": 26317 }, { "epoch": 1.8194262011752507, "grad_norm": 0.6755650134927007, "learning_rate": 2.1241349202195782e-07, "loss": 0.2599, "step": 26318 }, { "epoch": 1.8194953335637747, "grad_norm": 0.747522908843562, "learning_rate": 2.122521009962991e-07, "loss": 0.3055, "step": 26319 }, { "epoch": 1.8195644659522987, "grad_norm": 0.6217430154120194, "learning_rate": 2.1209076997651734e-07, "loss": 0.2998, "step": 26320 }, { "epoch": 1.8196335983408227, "grad_norm": 0.729354967855386, "learning_rate": 2.1192949896463256e-07, "loss": 0.2358, "step": 26321 }, { "epoch": 1.8197027307293467, "grad_norm": 0.7128002805578391, "learning_rate": 2.1176828796266703e-07, "loss": 0.2245, "step": 26322 }, { "epoch": 1.8197718631178708, "grad_norm": 0.6871578728615741, "learning_rate": 2.116071369726408e-07, "loss": 0.3489, "step": 26323 }, { "epoch": 1.8198409955063948, "grad_norm": 0.7150742027738325, "learning_rate": 2.1144604599657392e-07, "loss": 0.2331, "step": 26324 }, { "epoch": 1.8199101278949188, "grad_norm": 0.6683102031196582, "learning_rate": 2.112850150364848e-07, "loss": 0.2125, "step": 26325 }, { "epoch": 1.8199792602834428, "grad_norm": 0.6929357380836516, "learning_rate": 2.111240440943929e-07, "loss": 0.2468, "step": 26326 }, { "epoch": 1.8200483926719668, "grad_norm": 0.6945136019161666, "learning_rate": 2.1096313317231497e-07, "loss": 0.3539, "step": 26327 }, { "epoch": 1.8201175250604908, "grad_norm": 0.8186229188966567, "learning_rate": 2.1080228227226773e-07, "loss": 0.2739, "step": 26328 }, { "epoch": 1.8201866574490149, "grad_norm": 0.6930461786801774, "learning_rate": 2.1064149139626678e-07, "loss": 0.2543, "step": 26329 }, { "epoch": 1.8202557898375389, "grad_norm": 0.6627719743477345, "learning_rate": 2.1048076054632772e-07, "loss": 0.3321, "step": 26330 }, { "epoch": 1.820324922226063, "grad_norm": 0.7512921360108172, "learning_rate": 2.1032008972446505e-07, "loss": 0.2449, "step": 26331 }, { "epoch": 1.820394054614587, "grad_norm": 0.6808499303302371, "learning_rate": 2.101594789326933e-07, "loss": 0.1893, "step": 26332 }, { "epoch": 1.820463187003111, "grad_norm": 0.6727744872545175, "learning_rate": 2.0999892817302358e-07, "loss": 0.2539, "step": 26333 }, { "epoch": 1.820532319391635, "grad_norm": 0.6917525175120214, "learning_rate": 2.098384374474699e-07, "loss": 0.2736, "step": 26334 }, { "epoch": 1.820601451780159, "grad_norm": 0.7215940773965562, "learning_rate": 2.0967800675804285e-07, "loss": 0.2333, "step": 26335 }, { "epoch": 1.820670584168683, "grad_norm": 0.6741962399553781, "learning_rate": 2.0951763610675303e-07, "loss": 0.1846, "step": 26336 }, { "epoch": 1.820739716557207, "grad_norm": 0.6828667724919996, "learning_rate": 2.093573254956105e-07, "loss": 0.3995, "step": 26337 }, { "epoch": 1.820808848945731, "grad_norm": 0.6986160996995684, "learning_rate": 2.091970749266248e-07, "loss": 0.3218, "step": 26338 }, { "epoch": 1.820877981334255, "grad_norm": 0.6244491051043953, "learning_rate": 2.0903688440180482e-07, "loss": 0.2286, "step": 26339 }, { "epoch": 1.820947113722779, "grad_norm": 0.7206718901491708, "learning_rate": 2.0887675392315675e-07, "loss": 0.2799, "step": 26340 }, { "epoch": 1.821016246111303, "grad_norm": 0.658467855233176, "learning_rate": 2.0871668349268904e-07, "loss": 0.3174, "step": 26341 }, { "epoch": 1.821085378499827, "grad_norm": 0.7672042788374691, "learning_rate": 2.0855667311240668e-07, "loss": 0.24, "step": 26342 }, { "epoch": 1.8211545108883511, "grad_norm": 0.7898541016469223, "learning_rate": 2.0839672278431644e-07, "loss": 0.2283, "step": 26343 }, { "epoch": 1.8212236432768751, "grad_norm": 0.7018335208750831, "learning_rate": 2.082368325104217e-07, "loss": 0.305, "step": 26344 }, { "epoch": 1.8212927756653992, "grad_norm": 0.8649739475837723, "learning_rate": 2.08077002292727e-07, "loss": 0.2697, "step": 26345 }, { "epoch": 1.8213619080539232, "grad_norm": 0.7490322115710002, "learning_rate": 2.0791723213323512e-07, "loss": 0.2098, "step": 26346 }, { "epoch": 1.8214310404424472, "grad_norm": 0.7182219186334341, "learning_rate": 2.0775752203395005e-07, "loss": 0.1986, "step": 26347 }, { "epoch": 1.8215001728309712, "grad_norm": 0.8961424883063827, "learning_rate": 2.0759787199687131e-07, "loss": 0.3217, "step": 26348 }, { "epoch": 1.8215693052194952, "grad_norm": 0.7172466267550227, "learning_rate": 2.0743828202400118e-07, "loss": 0.273, "step": 26349 }, { "epoch": 1.8216384376080192, "grad_norm": 0.6573187100002095, "learning_rate": 2.0727875211733917e-07, "loss": 0.213, "step": 26350 }, { "epoch": 1.8217075699965433, "grad_norm": 0.7126324548699602, "learning_rate": 2.0711928227888534e-07, "loss": 0.2629, "step": 26351 }, { "epoch": 1.8217767023850673, "grad_norm": 0.6804754051277258, "learning_rate": 2.0695987251063753e-07, "loss": 0.3295, "step": 26352 }, { "epoch": 1.8218458347735913, "grad_norm": 0.8141679854495724, "learning_rate": 2.0680052281459528e-07, "loss": 0.229, "step": 26353 }, { "epoch": 1.8219149671621153, "grad_norm": 0.794790587282455, "learning_rate": 2.0664123319275365e-07, "loss": 0.2261, "step": 26354 }, { "epoch": 1.8219840995506393, "grad_norm": 0.8002625604084305, "learning_rate": 2.0648200364711045e-07, "loss": 0.3736, "step": 26355 }, { "epoch": 1.8220532319391634, "grad_norm": 0.7499739538325745, "learning_rate": 2.0632283417966082e-07, "loss": 0.2588, "step": 26356 }, { "epoch": 1.8221223643276874, "grad_norm": 0.6930638868495537, "learning_rate": 2.0616372479239977e-07, "loss": 0.234, "step": 26357 }, { "epoch": 1.8221914967162114, "grad_norm": 0.7450774612329564, "learning_rate": 2.0600467548732128e-07, "loss": 0.2861, "step": 26358 }, { "epoch": 1.8222606291047354, "grad_norm": 0.7288298041427425, "learning_rate": 2.0584568626641876e-07, "loss": 0.3261, "step": 26359 }, { "epoch": 1.8223297614932594, "grad_norm": 0.7506886077814859, "learning_rate": 2.0568675713168506e-07, "loss": 0.2507, "step": 26360 }, { "epoch": 1.8223988938817837, "grad_norm": 0.9214895286612573, "learning_rate": 2.055278880851125e-07, "loss": 0.1836, "step": 26361 }, { "epoch": 1.8224680262703077, "grad_norm": 0.6905277159169272, "learning_rate": 2.0536907912869163e-07, "loss": 0.3207, "step": 26362 }, { "epoch": 1.8225371586588317, "grad_norm": 1.2289931062814539, "learning_rate": 2.0521033026441317e-07, "loss": 0.2403, "step": 26363 }, { "epoch": 1.8226062910473557, "grad_norm": 0.6419640998470707, "learning_rate": 2.0505164149426548e-07, "loss": 0.1993, "step": 26364 }, { "epoch": 1.8226754234358797, "grad_norm": 0.7351126209842012, "learning_rate": 2.048930128202392e-07, "loss": 0.2759, "step": 26365 }, { "epoch": 1.8227445558244038, "grad_norm": 0.7771487073554113, "learning_rate": 2.0473444424432166e-07, "loss": 0.3417, "step": 26366 }, { "epoch": 1.8228136882129278, "grad_norm": 0.7357455315309848, "learning_rate": 2.0457593576850065e-07, "loss": 0.2203, "step": 26367 }, { "epoch": 1.8228828206014518, "grad_norm": 0.714266315928766, "learning_rate": 2.0441748739476187e-07, "loss": 0.3017, "step": 26368 }, { "epoch": 1.8229519529899758, "grad_norm": 0.7107334545767013, "learning_rate": 2.0425909912509257e-07, "loss": 0.2749, "step": 26369 }, { "epoch": 1.8230210853784998, "grad_norm": 1.316676370617056, "learning_rate": 2.0410077096147617e-07, "loss": 0.2939, "step": 26370 }, { "epoch": 1.8230902177670238, "grad_norm": 0.6937668492035033, "learning_rate": 2.039425029058989e-07, "loss": 0.2269, "step": 26371 }, { "epoch": 1.8231593501555479, "grad_norm": 0.7064227044761717, "learning_rate": 2.0378429496034303e-07, "loss": 0.2424, "step": 26372 }, { "epoch": 1.8232284825440719, "grad_norm": 0.6678540950404523, "learning_rate": 2.036261471267914e-07, "loss": 0.3125, "step": 26373 }, { "epoch": 1.823297614932596, "grad_norm": 0.7282922055295515, "learning_rate": 2.0346805940722637e-07, "loss": 0.2488, "step": 26374 }, { "epoch": 1.82336674732112, "grad_norm": 0.700413538062993, "learning_rate": 2.0331003180363018e-07, "loss": 0.2205, "step": 26375 }, { "epoch": 1.823435879709644, "grad_norm": 0.6913230674689341, "learning_rate": 2.0315206431798184e-07, "loss": 0.2545, "step": 26376 }, { "epoch": 1.823505012098168, "grad_norm": 0.6786806038501119, "learning_rate": 2.029941569522631e-07, "loss": 0.2957, "step": 26377 }, { "epoch": 1.823574144486692, "grad_norm": 0.7753623597062443, "learning_rate": 2.0283630970845125e-07, "loss": 0.2345, "step": 26378 }, { "epoch": 1.823643276875216, "grad_norm": 0.7131063088277619, "learning_rate": 2.0267852258852582e-07, "loss": 0.2234, "step": 26379 }, { "epoch": 1.82371240926374, "grad_norm": 0.7041776846272111, "learning_rate": 2.0252079559446304e-07, "loss": 0.3571, "step": 26380 }, { "epoch": 1.8237815416522642, "grad_norm": 0.7510259109014418, "learning_rate": 2.0236312872824072e-07, "loss": 0.2657, "step": 26381 }, { "epoch": 1.8238506740407883, "grad_norm": 0.67940198386043, "learning_rate": 2.0220552199183507e-07, "loss": 0.2101, "step": 26382 }, { "epoch": 1.8239198064293123, "grad_norm": 0.6765067284923054, "learning_rate": 2.0204797538722177e-07, "loss": 0.2551, "step": 26383 }, { "epoch": 1.8239889388178363, "grad_norm": 0.7441288811320438, "learning_rate": 2.018904889163742e-07, "loss": 0.3712, "step": 26384 }, { "epoch": 1.8240580712063603, "grad_norm": 0.7416755580712775, "learning_rate": 2.017330625812669e-07, "loss": 0.2424, "step": 26385 }, { "epoch": 1.8241272035948843, "grad_norm": 0.6077630240484287, "learning_rate": 2.015756963838722e-07, "loss": 0.1867, "step": 26386 }, { "epoch": 1.8241963359834084, "grad_norm": 0.7019610520835299, "learning_rate": 2.014183903261635e-07, "loss": 0.3623, "step": 26387 }, { "epoch": 1.8242654683719324, "grad_norm": 0.7340997941046989, "learning_rate": 2.0126114441011147e-07, "loss": 0.2518, "step": 26388 }, { "epoch": 1.8243346007604564, "grad_norm": 0.7062780347299811, "learning_rate": 2.0110395863768783e-07, "loss": 0.2452, "step": 26389 }, { "epoch": 1.8244037331489804, "grad_norm": 0.8085191122890041, "learning_rate": 2.009468330108616e-07, "loss": 0.2162, "step": 26390 }, { "epoch": 1.8244728655375044, "grad_norm": 0.7273654490586018, "learning_rate": 2.007897675316034e-07, "loss": 0.3419, "step": 26391 }, { "epoch": 1.8245419979260284, "grad_norm": 0.6550309869734199, "learning_rate": 2.0063276220187943e-07, "loss": 0.2086, "step": 26392 }, { "epoch": 1.8246111303145525, "grad_norm": 0.6785195586560218, "learning_rate": 2.0047581702366036e-07, "loss": 0.2292, "step": 26393 }, { "epoch": 1.8246802627030765, "grad_norm": 0.7295211767736195, "learning_rate": 2.0031893199891072e-07, "loss": 0.3178, "step": 26394 }, { "epoch": 1.8247493950916005, "grad_norm": 0.9122738701232821, "learning_rate": 2.0016210712959838e-07, "loss": 0.3043, "step": 26395 }, { "epoch": 1.8248185274801245, "grad_norm": 0.7425182967738957, "learning_rate": 2.0000534241768787e-07, "loss": 0.1933, "step": 26396 }, { "epoch": 1.8248876598686485, "grad_norm": 0.6900679848842324, "learning_rate": 1.9984863786514485e-07, "loss": 0.2275, "step": 26397 }, { "epoch": 1.8249567922571726, "grad_norm": 0.8131958716761013, "learning_rate": 1.9969199347393276e-07, "loss": 0.2996, "step": 26398 }, { "epoch": 1.8250259246456966, "grad_norm": 0.6644433882003246, "learning_rate": 1.9953540924601555e-07, "loss": 0.2583, "step": 26399 }, { "epoch": 1.8250950570342206, "grad_norm": 0.7939913017055296, "learning_rate": 1.9937888518335448e-07, "loss": 0.2573, "step": 26400 }, { "epoch": 1.8251641894227446, "grad_norm": 0.6463013434068833, "learning_rate": 1.992224212879118e-07, "loss": 0.2997, "step": 26401 }, { "epoch": 1.8252333218112686, "grad_norm": 0.7083914900895574, "learning_rate": 1.9906601756164878e-07, "loss": 0.3087, "step": 26402 }, { "epoch": 1.8253024541997926, "grad_norm": 0.721270277321573, "learning_rate": 1.9890967400652605e-07, "loss": 0.2671, "step": 26403 }, { "epoch": 1.8253715865883167, "grad_norm": 0.5898866112327216, "learning_rate": 1.9875339062450206e-07, "loss": 0.1966, "step": 26404 }, { "epoch": 1.8254407189768407, "grad_norm": 0.7440879266349387, "learning_rate": 1.9859716741753688e-07, "loss": 0.3323, "step": 26405 }, { "epoch": 1.8255098513653647, "grad_norm": 0.7000931607653081, "learning_rate": 1.9844100438758784e-07, "loss": 0.2802, "step": 26406 }, { "epoch": 1.8255789837538887, "grad_norm": 0.7071197327994211, "learning_rate": 1.9828490153661116e-07, "loss": 0.2231, "step": 26407 }, { "epoch": 1.8256481161424127, "grad_norm": 0.649188491341601, "learning_rate": 1.9812885886656473e-07, "loss": 0.2495, "step": 26408 }, { "epoch": 1.8257172485309368, "grad_norm": 0.7315704900840386, "learning_rate": 1.979728763794031e-07, "loss": 0.3749, "step": 26409 }, { "epoch": 1.8257863809194608, "grad_norm": 1.0461401895115763, "learning_rate": 1.9781695407708246e-07, "loss": 0.1962, "step": 26410 }, { "epoch": 1.8258555133079848, "grad_norm": 0.6665449893129466, "learning_rate": 1.9766109196155625e-07, "loss": 0.1904, "step": 26411 }, { "epoch": 1.8259246456965088, "grad_norm": 0.6796071896873042, "learning_rate": 1.9750529003477847e-07, "loss": 0.2737, "step": 26412 }, { "epoch": 1.8259937780850328, "grad_norm": 0.662109093440715, "learning_rate": 1.973495482987009e-07, "loss": 0.2556, "step": 26413 }, { "epoch": 1.8260629104735568, "grad_norm": 0.7348109614556683, "learning_rate": 1.971938667552764e-07, "loss": 0.2553, "step": 26414 }, { "epoch": 1.8261320428620809, "grad_norm": 0.7429903243456395, "learning_rate": 1.9703824540645565e-07, "loss": 0.2481, "step": 26415 }, { "epoch": 1.8262011752506049, "grad_norm": 0.6833809465406581, "learning_rate": 1.9688268425418876e-07, "loss": 0.2985, "step": 26416 }, { "epoch": 1.826270307639129, "grad_norm": 0.7718152633254256, "learning_rate": 1.967271833004264e-07, "loss": 0.2615, "step": 26417 }, { "epoch": 1.826339440027653, "grad_norm": 0.649957499285885, "learning_rate": 1.9657174254711696e-07, "loss": 0.2275, "step": 26418 }, { "epoch": 1.826408572416177, "grad_norm": 0.6902338193181928, "learning_rate": 1.9641636199620838e-07, "loss": 0.2941, "step": 26419 }, { "epoch": 1.826477704804701, "grad_norm": 0.7088714463327168, "learning_rate": 1.962610416496491e-07, "loss": 0.3423, "step": 26420 }, { "epoch": 1.826546837193225, "grad_norm": 0.6489838768589254, "learning_rate": 1.961057815093842e-07, "loss": 0.2287, "step": 26421 }, { "epoch": 1.826615969581749, "grad_norm": 0.6972486860167099, "learning_rate": 1.959505815773599e-07, "loss": 0.2584, "step": 26422 }, { "epoch": 1.826685101970273, "grad_norm": 0.8706361254609147, "learning_rate": 1.9579544185552247e-07, "loss": 0.3466, "step": 26423 }, { "epoch": 1.826754234358797, "grad_norm": 0.7184347270170399, "learning_rate": 1.956403623458164e-07, "loss": 0.2753, "step": 26424 }, { "epoch": 1.826823366747321, "grad_norm": 0.6550121544684431, "learning_rate": 1.954853430501835e-07, "loss": 0.2038, "step": 26425 }, { "epoch": 1.826892499135845, "grad_norm": 0.7749714595699595, "learning_rate": 1.9533038397056836e-07, "loss": 0.3028, "step": 26426 }, { "epoch": 1.826961631524369, "grad_norm": 0.7198624682384489, "learning_rate": 1.951754851089127e-07, "loss": 0.3139, "step": 26427 }, { "epoch": 1.827030763912893, "grad_norm": 0.7030400811739226, "learning_rate": 1.950206464671578e-07, "loss": 0.241, "step": 26428 }, { "epoch": 1.8270998963014171, "grad_norm": 0.6856027154957824, "learning_rate": 1.9486586804724373e-07, "loss": 0.2999, "step": 26429 }, { "epoch": 1.8271690286899411, "grad_norm": 0.7640180107728343, "learning_rate": 1.947111498511106e-07, "loss": 0.3286, "step": 26430 }, { "epoch": 1.8272381610784652, "grad_norm": 0.6984072762055701, "learning_rate": 1.9455649188069803e-07, "loss": 0.2705, "step": 26431 }, { "epoch": 1.8273072934669892, "grad_norm": 0.7123509866830802, "learning_rate": 1.944018941379444e-07, "loss": 0.218, "step": 26432 }, { "epoch": 1.8273764258555132, "grad_norm": 0.7185161212315553, "learning_rate": 1.9424735662478712e-07, "loss": 0.2311, "step": 26433 }, { "epoch": 1.8274455582440372, "grad_norm": 0.6902106896152476, "learning_rate": 1.940928793431629e-07, "loss": 0.3681, "step": 26434 }, { "epoch": 1.8275146906325612, "grad_norm": 0.7126081306233292, "learning_rate": 1.9393846229500745e-07, "loss": 0.2783, "step": 26435 }, { "epoch": 1.8275838230210852, "grad_norm": 0.6724993164837856, "learning_rate": 1.93784105482257e-07, "loss": 0.2498, "step": 26436 }, { "epoch": 1.8276529554096093, "grad_norm": 0.7375607262382949, "learning_rate": 1.9362980890684558e-07, "loss": 0.354, "step": 26437 }, { "epoch": 1.8277220877981333, "grad_norm": 0.7328553742101002, "learning_rate": 1.934755725707077e-07, "loss": 0.2802, "step": 26438 }, { "epoch": 1.8277912201866573, "grad_norm": 0.8420980805960304, "learning_rate": 1.933213964757752e-07, "loss": 0.2287, "step": 26439 }, { "epoch": 1.8278603525751813, "grad_norm": 0.6849907474851176, "learning_rate": 1.931672806239815e-07, "loss": 0.2672, "step": 26440 }, { "epoch": 1.8279294849637053, "grad_norm": 0.7565676819917581, "learning_rate": 1.9301322501725727e-07, "loss": 0.3949, "step": 26441 }, { "epoch": 1.8279986173522296, "grad_norm": 0.9099451177771074, "learning_rate": 1.9285922965753434e-07, "loss": 0.2182, "step": 26442 }, { "epoch": 1.8280677497407536, "grad_norm": 0.6903506151481745, "learning_rate": 1.927052945467417e-07, "loss": 0.2493, "step": 26443 }, { "epoch": 1.8281368821292776, "grad_norm": 0.7222999085041494, "learning_rate": 1.9255141968680946e-07, "loss": 0.3711, "step": 26444 }, { "epoch": 1.8282060145178016, "grad_norm": 0.7051249972286745, "learning_rate": 1.9239760507966554e-07, "loss": 0.3369, "step": 26445 }, { "epoch": 1.8282751469063256, "grad_norm": 0.6995808405477176, "learning_rate": 1.9224385072723895e-07, "loss": 0.2245, "step": 26446 }, { "epoch": 1.8283442792948497, "grad_norm": 0.7532338106110247, "learning_rate": 1.9209015663145536e-07, "loss": 0.2449, "step": 26447 }, { "epoch": 1.8284134116833737, "grad_norm": 0.72711293728636, "learning_rate": 1.9193652279424214e-07, "loss": 0.2732, "step": 26448 }, { "epoch": 1.8284825440718977, "grad_norm": 0.7594177425308731, "learning_rate": 1.9178294921752327e-07, "loss": 0.2418, "step": 26449 }, { "epoch": 1.8285516764604217, "grad_norm": 0.6967436146479583, "learning_rate": 1.9162943590322502e-07, "loss": 0.1998, "step": 26450 }, { "epoch": 1.8286208088489457, "grad_norm": 0.6951122923200181, "learning_rate": 1.9147598285327084e-07, "loss": 0.3173, "step": 26451 }, { "epoch": 1.8286899412374698, "grad_norm": 0.7536627682896542, "learning_rate": 1.9132259006958364e-07, "loss": 0.3455, "step": 26452 }, { "epoch": 1.8287590736259938, "grad_norm": 0.7418637302739579, "learning_rate": 1.9116925755408633e-07, "loss": 0.2309, "step": 26453 }, { "epoch": 1.8288282060145178, "grad_norm": 0.8315027093895676, "learning_rate": 1.9101598530870126e-07, "loss": 0.2374, "step": 26454 }, { "epoch": 1.8288973384030418, "grad_norm": 0.8136548870033178, "learning_rate": 1.9086277333534853e-07, "loss": 0.3717, "step": 26455 }, { "epoch": 1.8289664707915658, "grad_norm": 0.7174172109290141, "learning_rate": 1.9070962163594886e-07, "loss": 0.2607, "step": 26456 }, { "epoch": 1.8290356031800898, "grad_norm": 0.7494968248007263, "learning_rate": 1.9055653021242072e-07, "loss": 0.2299, "step": 26457 }, { "epoch": 1.8291047355686139, "grad_norm": 0.7169907799348311, "learning_rate": 1.904034990666842e-07, "loss": 0.2816, "step": 26458 }, { "epoch": 1.8291738679571379, "grad_norm": 0.7254244702145235, "learning_rate": 1.9025052820065616e-07, "loss": 0.3879, "step": 26459 }, { "epoch": 1.829243000345662, "grad_norm": 0.7342328630234546, "learning_rate": 1.90097617616255e-07, "loss": 0.236, "step": 26460 }, { "epoch": 1.829312132734186, "grad_norm": 0.7048856884190243, "learning_rate": 1.8994476731539645e-07, "loss": 0.2664, "step": 26461 }, { "epoch": 1.8293812651227102, "grad_norm": 0.7382768126520339, "learning_rate": 1.8979197729999676e-07, "loss": 0.3434, "step": 26462 }, { "epoch": 1.8294503975112342, "grad_norm": 0.7266939552767144, "learning_rate": 1.896392475719694e-07, "loss": 0.2794, "step": 26463 }, { "epoch": 1.8295195298997582, "grad_norm": 0.8537090307249805, "learning_rate": 1.8948657813323057e-07, "loss": 0.2435, "step": 26464 }, { "epoch": 1.8295886622882822, "grad_norm": 0.7605052966908898, "learning_rate": 1.8933396898569212e-07, "loss": 0.2725, "step": 26465 }, { "epoch": 1.8296577946768062, "grad_norm": 0.7384574132398275, "learning_rate": 1.891814201312675e-07, "loss": 0.4455, "step": 26466 }, { "epoch": 1.8297269270653302, "grad_norm": 0.7850833609795672, "learning_rate": 1.8902893157186797e-07, "loss": 0.2917, "step": 26467 }, { "epoch": 1.8297960594538543, "grad_norm": 0.770463756804472, "learning_rate": 1.8887650330940645e-07, "loss": 0.2594, "step": 26468 }, { "epoch": 1.8298651918423783, "grad_norm": 0.7639854581950413, "learning_rate": 1.8872413534579083e-07, "loss": 0.2812, "step": 26469 }, { "epoch": 1.8299343242309023, "grad_norm": 0.654345939865541, "learning_rate": 1.8857182768293292e-07, "loss": 0.2987, "step": 26470 }, { "epoch": 1.8300034566194263, "grad_norm": 0.7236756559200472, "learning_rate": 1.884195803227412e-07, "loss": 0.3157, "step": 26471 }, { "epoch": 1.8300725890079503, "grad_norm": 0.717699190482762, "learning_rate": 1.8826739326712195e-07, "loss": 0.2743, "step": 26472 }, { "epoch": 1.8301417213964744, "grad_norm": 0.6641169356828373, "learning_rate": 1.8811526651798473e-07, "loss": 0.3199, "step": 26473 }, { "epoch": 1.8302108537849984, "grad_norm": 0.8309915809243673, "learning_rate": 1.8796320007723523e-07, "loss": 0.22, "step": 26474 }, { "epoch": 1.8302799861735224, "grad_norm": 0.696588601181931, "learning_rate": 1.8781119394678026e-07, "loss": 0.2413, "step": 26475 }, { "epoch": 1.8303491185620464, "grad_norm": 0.6658188420835545, "learning_rate": 1.8765924812852388e-07, "loss": 0.2752, "step": 26476 }, { "epoch": 1.8304182509505704, "grad_norm": 0.673474925099111, "learning_rate": 1.875073626243712e-07, "loss": 0.2835, "step": 26477 }, { "epoch": 1.8304873833390944, "grad_norm": 0.6903649921241385, "learning_rate": 1.8735553743622458e-07, "loss": 0.2447, "step": 26478 }, { "epoch": 1.8305565157276185, "grad_norm": 0.7776779542518567, "learning_rate": 1.8720377256598753e-07, "loss": 0.2386, "step": 26479 }, { "epoch": 1.8306256481161425, "grad_norm": 0.7288555511656449, "learning_rate": 1.8705206801556297e-07, "loss": 0.3229, "step": 26480 }, { "epoch": 1.8306947805046665, "grad_norm": 0.8608681874480166, "learning_rate": 1.8690042378685214e-07, "loss": 0.2674, "step": 26481 }, { "epoch": 1.8307639128931905, "grad_norm": 0.6419416584475766, "learning_rate": 1.8674883988175463e-07, "loss": 0.1737, "step": 26482 }, { "epoch": 1.8308330452817145, "grad_norm": 0.664646787700362, "learning_rate": 1.8659731630217114e-07, "loss": 0.2756, "step": 26483 }, { "epoch": 1.8309021776702386, "grad_norm": 0.7550550386378003, "learning_rate": 1.8644585304999964e-07, "loss": 0.3411, "step": 26484 }, { "epoch": 1.8309713100587626, "grad_norm": 0.6587345169222907, "learning_rate": 1.8629445012713965e-07, "loss": 0.243, "step": 26485 }, { "epoch": 1.8310404424472866, "grad_norm": 0.6385196384830693, "learning_rate": 1.8614310753548803e-07, "loss": 0.2324, "step": 26486 }, { "epoch": 1.8311095748358106, "grad_norm": 0.7611181364775436, "learning_rate": 1.8599182527694215e-07, "loss": 0.375, "step": 26487 }, { "epoch": 1.8311787072243346, "grad_norm": 0.6527872174976174, "learning_rate": 1.8584060335339715e-07, "loss": 0.2548, "step": 26488 }, { "epoch": 1.8312478396128586, "grad_norm": 0.6281752878194415, "learning_rate": 1.8568944176674987e-07, "loss": 0.2317, "step": 26489 }, { "epoch": 1.8313169720013827, "grad_norm": 0.7101231056402336, "learning_rate": 1.855383405188932e-07, "loss": 0.3258, "step": 26490 }, { "epoch": 1.8313861043899067, "grad_norm": 0.6990560647891407, "learning_rate": 1.8538729961172231e-07, "loss": 0.3409, "step": 26491 }, { "epoch": 1.8314552367784307, "grad_norm": 0.7416667553556386, "learning_rate": 1.8523631904712902e-07, "loss": 0.2314, "step": 26492 }, { "epoch": 1.8315243691669547, "grad_norm": 0.6635495825625268, "learning_rate": 1.8508539882700626e-07, "loss": 0.3076, "step": 26493 }, { "epoch": 1.8315935015554787, "grad_norm": 0.825502690968243, "learning_rate": 1.8493453895324532e-07, "loss": 0.3124, "step": 26494 }, { "epoch": 1.8316626339440027, "grad_norm": 0.7606672618604071, "learning_rate": 1.8478373942773743e-07, "loss": 0.3268, "step": 26495 }, { "epoch": 1.8317317663325268, "grad_norm": 0.7015643798139998, "learning_rate": 1.8463300025237162e-07, "loss": 0.2213, "step": 26496 }, { "epoch": 1.8318008987210508, "grad_norm": 0.791640316593841, "learning_rate": 1.8448232142903865e-07, "loss": 0.3019, "step": 26497 }, { "epoch": 1.8318700311095748, "grad_norm": 0.6983973996693391, "learning_rate": 1.8433170295962532e-07, "loss": 0.3846, "step": 26498 }, { "epoch": 1.8319391634980988, "grad_norm": 0.794622825144537, "learning_rate": 1.8418114484602068e-07, "loss": 0.2671, "step": 26499 }, { "epoch": 1.8320082958866228, "grad_norm": 0.595393405329585, "learning_rate": 1.8403064709011097e-07, "loss": 0.1688, "step": 26500 }, { "epoch": 1.8320774282751469, "grad_norm": 0.6941551148063412, "learning_rate": 1.8388020969378194e-07, "loss": 0.3044, "step": 26501 }, { "epoch": 1.8321465606636709, "grad_norm": 0.7078337846719525, "learning_rate": 1.8372983265892042e-07, "loss": 0.3599, "step": 26502 }, { "epoch": 1.832215693052195, "grad_norm": 0.7881936274520549, "learning_rate": 1.8357951598741041e-07, "loss": 0.2926, "step": 26503 }, { "epoch": 1.832284825440719, "grad_norm": 0.735352039623831, "learning_rate": 1.8342925968113546e-07, "loss": 0.2524, "step": 26504 }, { "epoch": 1.832353957829243, "grad_norm": 0.710215062665001, "learning_rate": 1.8327906374198014e-07, "loss": 0.3378, "step": 26505 }, { "epoch": 1.832423090217767, "grad_norm": 0.6654919648432566, "learning_rate": 1.8312892817182516e-07, "loss": 0.2723, "step": 26506 }, { "epoch": 1.832492222606291, "grad_norm": 0.7311640352237007, "learning_rate": 1.829788529725529e-07, "loss": 0.2471, "step": 26507 }, { "epoch": 1.832561354994815, "grad_norm": 0.6998736726171467, "learning_rate": 1.8282883814604412e-07, "loss": 0.2538, "step": 26508 }, { "epoch": 1.832630487383339, "grad_norm": 0.6974002904361737, "learning_rate": 1.8267888369418007e-07, "loss": 0.3463, "step": 26509 }, { "epoch": 1.832699619771863, "grad_norm": 0.6062742593019964, "learning_rate": 1.825289896188387e-07, "loss": 0.2186, "step": 26510 }, { "epoch": 1.832768752160387, "grad_norm": 0.7170563881369207, "learning_rate": 1.8237915592190015e-07, "loss": 0.2694, "step": 26511 }, { "epoch": 1.832837884548911, "grad_norm": 0.7078877901549545, "learning_rate": 1.822293826052407e-07, "loss": 0.36, "step": 26512 }, { "epoch": 1.832907016937435, "grad_norm": 0.750063182725333, "learning_rate": 1.820796696707383e-07, "loss": 0.2577, "step": 26513 }, { "epoch": 1.832976149325959, "grad_norm": 0.8137017571292968, "learning_rate": 1.8193001712026925e-07, "loss": 0.2282, "step": 26514 }, { "epoch": 1.833045281714483, "grad_norm": 0.6617290728707009, "learning_rate": 1.8178042495570924e-07, "loss": 0.2453, "step": 26515 }, { "epoch": 1.8331144141030071, "grad_norm": 0.6958705445469747, "learning_rate": 1.816308931789329e-07, "loss": 0.3365, "step": 26516 }, { "epoch": 1.8331835464915311, "grad_norm": 0.6826395455649822, "learning_rate": 1.8148142179181537e-07, "loss": 0.2958, "step": 26517 }, { "epoch": 1.8332526788800552, "grad_norm": 0.6792706714840826, "learning_rate": 1.8133201079622797e-07, "loss": 0.2281, "step": 26518 }, { "epoch": 1.8333218112685792, "grad_norm": 0.7293648121779182, "learning_rate": 1.8118266019404585e-07, "loss": 0.2748, "step": 26519 }, { "epoch": 1.8333909436571032, "grad_norm": 0.6981161456117577, "learning_rate": 1.810333699871386e-07, "loss": 0.2575, "step": 26520 }, { "epoch": 1.8334600760456272, "grad_norm": 0.6677503453593328, "learning_rate": 1.8088414017737865e-07, "loss": 0.2009, "step": 26521 }, { "epoch": 1.8335292084341512, "grad_norm": 0.6454293480116632, "learning_rate": 1.807349707666356e-07, "loss": 0.2232, "step": 26522 }, { "epoch": 1.8335983408226755, "grad_norm": 0.7241450699381351, "learning_rate": 1.8058586175677904e-07, "loss": 0.3997, "step": 26523 }, { "epoch": 1.8336674732111995, "grad_norm": 0.6862450589008364, "learning_rate": 1.8043681314967864e-07, "loss": 0.2255, "step": 26524 }, { "epoch": 1.8337366055997235, "grad_norm": 0.6810631598262065, "learning_rate": 1.8028782494720177e-07, "loss": 0.2386, "step": 26525 }, { "epoch": 1.8338057379882475, "grad_norm": 0.768694575358714, "learning_rate": 1.801388971512158e-07, "loss": 0.2529, "step": 26526 }, { "epoch": 1.8338748703767715, "grad_norm": 0.7302843205715059, "learning_rate": 1.799900297635876e-07, "loss": 0.3593, "step": 26527 }, { "epoch": 1.8339440027652956, "grad_norm": 0.7413966674429365, "learning_rate": 1.7984122278618233e-07, "loss": 0.2054, "step": 26528 }, { "epoch": 1.8340131351538196, "grad_norm": 0.6448832023108093, "learning_rate": 1.7969247622086516e-07, "loss": 0.2453, "step": 26529 }, { "epoch": 1.8340822675423436, "grad_norm": 0.6728383006227397, "learning_rate": 1.7954379006950074e-07, "loss": 0.2883, "step": 26530 }, { "epoch": 1.8341513999308676, "grad_norm": 0.7764280806513175, "learning_rate": 1.7939516433395255e-07, "loss": 0.2776, "step": 26531 }, { "epoch": 1.8342205323193916, "grad_norm": 0.7345359825242626, "learning_rate": 1.7924659901608298e-07, "loss": 0.2414, "step": 26532 }, { "epoch": 1.8342896647079157, "grad_norm": 0.7103872098700437, "learning_rate": 1.7909809411775503e-07, "loss": 0.2621, "step": 26533 }, { "epoch": 1.8343587970964397, "grad_norm": 0.7442176193455717, "learning_rate": 1.7894964964082828e-07, "loss": 0.3144, "step": 26534 }, { "epoch": 1.8344279294849637, "grad_norm": 0.6657615312752084, "learning_rate": 1.7880126558716514e-07, "loss": 0.2142, "step": 26535 }, { "epoch": 1.8344970618734877, "grad_norm": 0.7056495642438608, "learning_rate": 1.7865294195862359e-07, "loss": 0.2228, "step": 26536 }, { "epoch": 1.8345661942620117, "grad_norm": 0.7242765778048948, "learning_rate": 1.7850467875706324e-07, "loss": 0.3646, "step": 26537 }, { "epoch": 1.8346353266505357, "grad_norm": 0.7212853365833081, "learning_rate": 1.783564759843426e-07, "loss": 0.2814, "step": 26538 }, { "epoch": 1.8347044590390598, "grad_norm": 0.8277414980943101, "learning_rate": 1.7820833364231905e-07, "loss": 0.2503, "step": 26539 }, { "epoch": 1.8347735914275838, "grad_norm": 0.6880884174821053, "learning_rate": 1.7806025173284947e-07, "loss": 0.2915, "step": 26540 }, { "epoch": 1.8348427238161078, "grad_norm": 0.7087964855072739, "learning_rate": 1.7791223025778847e-07, "loss": 0.3251, "step": 26541 }, { "epoch": 1.8349118562046318, "grad_norm": 0.7576735114292137, "learning_rate": 1.7776426921899349e-07, "loss": 0.2307, "step": 26542 }, { "epoch": 1.834980988593156, "grad_norm": 0.6677187900451255, "learning_rate": 1.7761636861831632e-07, "loss": 0.2681, "step": 26543 }, { "epoch": 1.83505012098168, "grad_norm": 2.217885237294138, "learning_rate": 1.7746852845761275e-07, "loss": 0.3078, "step": 26544 }, { "epoch": 1.835119253370204, "grad_norm": 0.7798888951551821, "learning_rate": 1.7732074873873517e-07, "loss": 0.2962, "step": 26545 }, { "epoch": 1.8351883857587281, "grad_norm": 0.7656490402246645, "learning_rate": 1.7717302946353544e-07, "loss": 0.2403, "step": 26546 }, { "epoch": 1.8352575181472521, "grad_norm": 0.6354327069828855, "learning_rate": 1.7702537063386483e-07, "loss": 0.2129, "step": 26547 }, { "epoch": 1.8353266505357761, "grad_norm": 0.77851444949061, "learning_rate": 1.7687777225157521e-07, "loss": 0.3693, "step": 26548 }, { "epoch": 1.8353957829243002, "grad_norm": 0.703787942019026, "learning_rate": 1.76730234318514e-07, "loss": 0.2795, "step": 26549 }, { "epoch": 1.8354649153128242, "grad_norm": 0.7141717403734518, "learning_rate": 1.7658275683653247e-07, "loss": 0.2559, "step": 26550 }, { "epoch": 1.8355340477013482, "grad_norm": 0.6799723618747713, "learning_rate": 1.7643533980747806e-07, "loss": 0.1951, "step": 26551 }, { "epoch": 1.8356031800898722, "grad_norm": 0.7997092291458923, "learning_rate": 1.762879832331993e-07, "loss": 0.3177, "step": 26552 }, { "epoch": 1.8356723124783962, "grad_norm": 0.6700812405344754, "learning_rate": 1.7614068711554188e-07, "loss": 0.2062, "step": 26553 }, { "epoch": 1.8357414448669203, "grad_norm": 0.699119644694574, "learning_rate": 1.759934514563527e-07, "loss": 0.2416, "step": 26554 }, { "epoch": 1.8358105772554443, "grad_norm": 0.697629208792044, "learning_rate": 1.758462762574764e-07, "loss": 0.3154, "step": 26555 }, { "epoch": 1.8358797096439683, "grad_norm": 0.7054361674022532, "learning_rate": 1.7569916152075818e-07, "loss": 0.2715, "step": 26556 }, { "epoch": 1.8359488420324923, "grad_norm": 1.140540212530544, "learning_rate": 1.755521072480415e-07, "loss": 0.2739, "step": 26557 }, { "epoch": 1.8360179744210163, "grad_norm": 0.7882234523186835, "learning_rate": 1.754051134411694e-07, "loss": 0.27, "step": 26558 }, { "epoch": 1.8360871068095403, "grad_norm": 0.9547897257222885, "learning_rate": 1.7525818010198426e-07, "loss": 0.356, "step": 26559 }, { "epoch": 1.8361562391980644, "grad_norm": 0.6603208645203617, "learning_rate": 1.7511130723232794e-07, "loss": 0.2365, "step": 26560 }, { "epoch": 1.8362253715865884, "grad_norm": 0.8469473082817321, "learning_rate": 1.7496449483404065e-07, "loss": 0.247, "step": 26561 }, { "epoch": 1.8362945039751124, "grad_norm": 0.7973073155131133, "learning_rate": 1.748177429089637e-07, "loss": 0.3857, "step": 26562 }, { "epoch": 1.8363636363636364, "grad_norm": 0.7444261357802757, "learning_rate": 1.7467105145893448e-07, "loss": 0.2447, "step": 26563 }, { "epoch": 1.8364327687521604, "grad_norm": 0.7332867687075081, "learning_rate": 1.7452442048579266e-07, "loss": 0.2509, "step": 26564 }, { "epoch": 1.8365019011406845, "grad_norm": 0.7631447499566542, "learning_rate": 1.7437784999137563e-07, "loss": 0.2323, "step": 26565 }, { "epoch": 1.8365710335292085, "grad_norm": 0.7303852242750317, "learning_rate": 1.7423133997752084e-07, "loss": 0.3478, "step": 26566 }, { "epoch": 1.8366401659177325, "grad_norm": 0.7894751488975158, "learning_rate": 1.7408489044606403e-07, "loss": 0.2279, "step": 26567 }, { "epoch": 1.8367092983062565, "grad_norm": 0.7787390361043899, "learning_rate": 1.739385013988415e-07, "loss": 0.2516, "step": 26568 }, { "epoch": 1.8367784306947805, "grad_norm": 0.685955322135402, "learning_rate": 1.737921728376868e-07, "loss": 0.3037, "step": 26569 }, { "epoch": 1.8368475630833045, "grad_norm": 0.7366525968242629, "learning_rate": 1.7364590476443454e-07, "loss": 0.3159, "step": 26570 }, { "epoch": 1.8369166954718286, "grad_norm": 0.6976579278818765, "learning_rate": 1.7349969718091774e-07, "loss": 0.237, "step": 26571 }, { "epoch": 1.8369858278603526, "grad_norm": 0.6933275521604172, "learning_rate": 1.733535500889688e-07, "loss": 0.2233, "step": 26572 }, { "epoch": 1.8370549602488766, "grad_norm": 0.8787544127805916, "learning_rate": 1.7320746349041962e-07, "loss": 0.3201, "step": 26573 }, { "epoch": 1.8371240926374006, "grad_norm": 0.73404983208137, "learning_rate": 1.73061437387102e-07, "loss": 0.2932, "step": 26574 }, { "epoch": 1.8371932250259246, "grad_norm": 0.6674547383625818, "learning_rate": 1.72915471780844e-07, "loss": 0.2221, "step": 26575 }, { "epoch": 1.8372623574144487, "grad_norm": 0.6893358293799609, "learning_rate": 1.7276956667347745e-07, "loss": 0.3025, "step": 26576 }, { "epoch": 1.8373314898029727, "grad_norm": 0.7169059883715978, "learning_rate": 1.7262372206682864e-07, "loss": 0.2975, "step": 26577 }, { "epoch": 1.8374006221914967, "grad_norm": 0.8017830620655523, "learning_rate": 1.7247793796272727e-07, "loss": 0.2258, "step": 26578 }, { "epoch": 1.8374697545800207, "grad_norm": 0.6585165123625811, "learning_rate": 1.7233221436299909e-07, "loss": 0.2352, "step": 26579 }, { "epoch": 1.8375388869685447, "grad_norm": 0.661379686321623, "learning_rate": 1.7218655126947204e-07, "loss": 0.2851, "step": 26580 }, { "epoch": 1.8376080193570687, "grad_norm": 0.9416186884871062, "learning_rate": 1.7204094868397081e-07, "loss": 0.2089, "step": 26581 }, { "epoch": 1.8376771517455928, "grad_norm": 0.8489301148039133, "learning_rate": 1.718954066083206e-07, "loss": 0.2506, "step": 26582 }, { "epoch": 1.8377462841341168, "grad_norm": 0.656141673141952, "learning_rate": 1.7174992504434496e-07, "loss": 0.2343, "step": 26583 }, { "epoch": 1.8378154165226408, "grad_norm": 0.8358178657248505, "learning_rate": 1.7160450399386797e-07, "loss": 0.2944, "step": 26584 }, { "epoch": 1.8378845489111648, "grad_norm": 0.6932741942837416, "learning_rate": 1.714591434587115e-07, "loss": 0.2053, "step": 26585 }, { "epoch": 1.8379536812996888, "grad_norm": 0.6909705991494435, "learning_rate": 1.7131384344069746e-07, "loss": 0.2274, "step": 26586 }, { "epoch": 1.8380228136882129, "grad_norm": 0.6652904095055142, "learning_rate": 1.7116860394164714e-07, "loss": 0.2715, "step": 26587 }, { "epoch": 1.8380919460767369, "grad_norm": 0.7353892624920492, "learning_rate": 1.7102342496338185e-07, "loss": 0.2728, "step": 26588 }, { "epoch": 1.838161078465261, "grad_norm": 0.6655308837608612, "learning_rate": 1.7087830650771965e-07, "loss": 0.2766, "step": 26589 }, { "epoch": 1.838230210853785, "grad_norm": 0.7531040807454077, "learning_rate": 1.7073324857648067e-07, "loss": 0.2813, "step": 26590 }, { "epoch": 1.838299343242309, "grad_norm": 0.7015960815707348, "learning_rate": 1.7058825117148127e-07, "loss": 0.3338, "step": 26591 }, { "epoch": 1.838368475630833, "grad_norm": 0.7076874573230441, "learning_rate": 1.7044331429454053e-07, "loss": 0.2498, "step": 26592 }, { "epoch": 1.838437608019357, "grad_norm": 0.7568477307943696, "learning_rate": 1.7029843794747314e-07, "loss": 0.2752, "step": 26593 }, { "epoch": 1.838506740407881, "grad_norm": 0.716072671720799, "learning_rate": 1.7015362213209596e-07, "loss": 0.3269, "step": 26594 }, { "epoch": 1.838575872796405, "grad_norm": 0.7119327658764375, "learning_rate": 1.700088668502242e-07, "loss": 0.2889, "step": 26595 }, { "epoch": 1.838645005184929, "grad_norm": 0.7206748518053464, "learning_rate": 1.6986417210367202e-07, "loss": 0.2336, "step": 26596 }, { "epoch": 1.838714137573453, "grad_norm": 0.6961422065499109, "learning_rate": 1.6971953789425234e-07, "loss": 0.2571, "step": 26597 }, { "epoch": 1.838783269961977, "grad_norm": 0.6895622889602031, "learning_rate": 1.6957496422377873e-07, "loss": 0.3341, "step": 26598 }, { "epoch": 1.838852402350501, "grad_norm": 0.7475041802557768, "learning_rate": 1.6943045109406197e-07, "loss": 0.3158, "step": 26599 }, { "epoch": 1.838921534739025, "grad_norm": 0.6937507254688267, "learning_rate": 1.6928599850691395e-07, "loss": 0.2184, "step": 26600 }, { "epoch": 1.838990667127549, "grad_norm": 0.7401105401918603, "learning_rate": 1.6914160646414546e-07, "loss": 0.3016, "step": 26601 }, { "epoch": 1.8390597995160731, "grad_norm": 0.6900433797454889, "learning_rate": 1.6899727496756614e-07, "loss": 0.2984, "step": 26602 }, { "epoch": 1.8391289319045971, "grad_norm": 0.6575237871224358, "learning_rate": 1.6885300401898397e-07, "loss": 0.2273, "step": 26603 }, { "epoch": 1.8391980642931214, "grad_norm": 0.7666114994008062, "learning_rate": 1.6870879362020864e-07, "loss": 0.2453, "step": 26604 }, { "epoch": 1.8392671966816454, "grad_norm": 0.7720051550607332, "learning_rate": 1.6856464377304703e-07, "loss": 0.3491, "step": 26605 }, { "epoch": 1.8393363290701694, "grad_norm": 0.7398619024988924, "learning_rate": 1.6842055447930439e-07, "loss": 0.2829, "step": 26606 }, { "epoch": 1.8394054614586934, "grad_norm": 0.6723655909107192, "learning_rate": 1.682765257407881e-07, "loss": 0.2365, "step": 26607 }, { "epoch": 1.8394745938472175, "grad_norm": 0.7313652646798203, "learning_rate": 1.6813255755930292e-07, "loss": 0.2679, "step": 26608 }, { "epoch": 1.8395437262357415, "grad_norm": 0.6876934890096272, "learning_rate": 1.67988649936654e-07, "loss": 0.2902, "step": 26609 }, { "epoch": 1.8396128586242655, "grad_norm": 0.634851197489642, "learning_rate": 1.6784480287464323e-07, "loss": 0.1943, "step": 26610 }, { "epoch": 1.8396819910127895, "grad_norm": 0.7641305507957946, "learning_rate": 1.677010163750753e-07, "loss": 0.2489, "step": 26611 }, { "epoch": 1.8397511234013135, "grad_norm": 0.7196128257645854, "learning_rate": 1.67557290439751e-07, "loss": 0.3591, "step": 26612 }, { "epoch": 1.8398202557898375, "grad_norm": 0.7719975103205947, "learning_rate": 1.6741362507047275e-07, "loss": 0.3404, "step": 26613 }, { "epoch": 1.8398893881783616, "grad_norm": 0.7645461558918205, "learning_rate": 1.6727002026903917e-07, "loss": 0.2603, "step": 26614 }, { "epoch": 1.8399585205668856, "grad_norm": 0.7719983455100251, "learning_rate": 1.671264760372532e-07, "loss": 0.2993, "step": 26615 }, { "epoch": 1.8400276529554096, "grad_norm": 0.7238206163857359, "learning_rate": 1.669829923769112e-07, "loss": 0.3374, "step": 26616 }, { "epoch": 1.8400967853439336, "grad_norm": 0.7317706375831756, "learning_rate": 1.6683956928981283e-07, "loss": 0.231, "step": 26617 }, { "epoch": 1.8401659177324576, "grad_norm": 0.668858348813057, "learning_rate": 1.66696206777755e-07, "loss": 0.2068, "step": 26618 }, { "epoch": 1.8402350501209817, "grad_norm": 0.7295960899528499, "learning_rate": 1.6655290484253572e-07, "loss": 0.3095, "step": 26619 }, { "epoch": 1.8403041825095057, "grad_norm": 0.8825102674227949, "learning_rate": 1.6640966348594912e-07, "loss": 0.2909, "step": 26620 }, { "epoch": 1.8403733148980297, "grad_norm": 0.7217339654041143, "learning_rate": 1.662664827097915e-07, "loss": 0.2019, "step": 26621 }, { "epoch": 1.8404424472865537, "grad_norm": 0.7071859129791455, "learning_rate": 1.661233625158576e-07, "loss": 0.2574, "step": 26622 }, { "epoch": 1.8405115796750777, "grad_norm": 0.715089826266132, "learning_rate": 1.659803029059409e-07, "loss": 0.318, "step": 26623 }, { "epoch": 1.8405807120636017, "grad_norm": 0.720644772553101, "learning_rate": 1.658373038818345e-07, "loss": 0.2629, "step": 26624 }, { "epoch": 1.840649844452126, "grad_norm": 0.6027489417609613, "learning_rate": 1.656943654453308e-07, "loss": 0.2234, "step": 26625 }, { "epoch": 1.84071897684065, "grad_norm": 0.754495215389454, "learning_rate": 1.6555148759822058e-07, "loss": 0.3201, "step": 26626 }, { "epoch": 1.840788109229174, "grad_norm": 0.7297111565371921, "learning_rate": 1.6540867034229524e-07, "loss": 0.345, "step": 26627 }, { "epoch": 1.840857241617698, "grad_norm": 0.7423029930238554, "learning_rate": 1.6526591367934387e-07, "loss": 0.278, "step": 26628 }, { "epoch": 1.840926374006222, "grad_norm": 0.8296792482214583, "learning_rate": 1.6512321761115667e-07, "loss": 0.257, "step": 26629 }, { "epoch": 1.840995506394746, "grad_norm": 0.7106208588652138, "learning_rate": 1.6498058213952174e-07, "loss": 0.2917, "step": 26630 }, { "epoch": 1.84106463878327, "grad_norm": 0.8244794076476143, "learning_rate": 1.6483800726622702e-07, "loss": 0.3422, "step": 26631 }, { "epoch": 1.841133771171794, "grad_norm": 2.0177657965503997, "learning_rate": 1.6469549299305888e-07, "loss": 0.2335, "step": 26632 }, { "epoch": 1.8412029035603181, "grad_norm": 0.7637268820144111, "learning_rate": 1.6455303932180422e-07, "loss": 0.2741, "step": 26633 }, { "epoch": 1.8412720359488421, "grad_norm": 0.7139571692532195, "learning_rate": 1.644106462542472e-07, "loss": 0.3674, "step": 26634 }, { "epoch": 1.8413411683373662, "grad_norm": 0.8787473848335287, "learning_rate": 1.6426831379217356e-07, "loss": 0.232, "step": 26635 }, { "epoch": 1.8414103007258902, "grad_norm": 0.6335760205881337, "learning_rate": 1.6412604193736693e-07, "loss": 0.1917, "step": 26636 }, { "epoch": 1.8414794331144142, "grad_norm": 1.36671003530858, "learning_rate": 1.6398383069161028e-07, "loss": 0.364, "step": 26637 }, { "epoch": 1.8415485655029382, "grad_norm": 0.773810532653682, "learning_rate": 1.6384168005668611e-07, "loss": 0.226, "step": 26638 }, { "epoch": 1.8416176978914622, "grad_norm": 0.7318396118555728, "learning_rate": 1.636995900343763e-07, "loss": 0.1864, "step": 26639 }, { "epoch": 1.8416868302799863, "grad_norm": 0.7227856278890024, "learning_rate": 1.635575606264611e-07, "loss": 0.2469, "step": 26640 }, { "epoch": 1.8417559626685103, "grad_norm": 0.754948936758926, "learning_rate": 1.6341559183472078e-07, "loss": 0.3618, "step": 26641 }, { "epoch": 1.8418250950570343, "grad_norm": 0.8876466143934457, "learning_rate": 1.6327368366093443e-07, "loss": 0.2388, "step": 26642 }, { "epoch": 1.8418942274455583, "grad_norm": 0.6722333768914304, "learning_rate": 1.631318361068812e-07, "loss": 0.2724, "step": 26643 }, { "epoch": 1.8419633598340823, "grad_norm": 0.7041499008451546, "learning_rate": 1.629900491743386e-07, "loss": 0.3967, "step": 26644 }, { "epoch": 1.8420324922226063, "grad_norm": 0.7164239495213992, "learning_rate": 1.6284832286508402e-07, "loss": 0.3044, "step": 26645 }, { "epoch": 1.8421016246111304, "grad_norm": 0.6838814150896668, "learning_rate": 1.6270665718089328e-07, "loss": 0.2238, "step": 26646 }, { "epoch": 1.8421707569996544, "grad_norm": 0.7269036830892996, "learning_rate": 1.6256505212354222e-07, "loss": 0.336, "step": 26647 }, { "epoch": 1.8422398893881784, "grad_norm": 0.7063360289613357, "learning_rate": 1.6242350769480496e-07, "loss": 0.3436, "step": 26648 }, { "epoch": 1.8423090217767024, "grad_norm": 0.8966673871664311, "learning_rate": 1.622820238964562e-07, "loss": 0.2524, "step": 26649 }, { "epoch": 1.8423781541652264, "grad_norm": 0.6646430933924571, "learning_rate": 1.6214060073026837e-07, "loss": 0.2228, "step": 26650 }, { "epoch": 1.8424472865537505, "grad_norm": 0.7183483954055455, "learning_rate": 1.6199923819801567e-07, "loss": 0.2828, "step": 26651 }, { "epoch": 1.8425164189422745, "grad_norm": 0.7384610107240689, "learning_rate": 1.6185793630146773e-07, "loss": 0.3276, "step": 26652 }, { "epoch": 1.8425855513307985, "grad_norm": 0.6856623857296217, "learning_rate": 1.6171669504239762e-07, "loss": 0.2124, "step": 26653 }, { "epoch": 1.8426546837193225, "grad_norm": 0.6787440132831982, "learning_rate": 1.6157551442257336e-07, "loss": 0.2712, "step": 26654 }, { "epoch": 1.8427238161078465, "grad_norm": 0.7570750429683704, "learning_rate": 1.614343944437663e-07, "loss": 0.3721, "step": 26655 }, { "epoch": 1.8427929484963705, "grad_norm": 0.7891704074085998, "learning_rate": 1.6129333510774393e-07, "loss": 0.328, "step": 26656 }, { "epoch": 1.8428620808848946, "grad_norm": 0.6509259235757925, "learning_rate": 1.6115233641627425e-07, "loss": 0.2253, "step": 26657 }, { "epoch": 1.8429312132734186, "grad_norm": 0.7192077932716479, "learning_rate": 1.6101139837112479e-07, "loss": 0.2848, "step": 26658 }, { "epoch": 1.8430003456619426, "grad_norm": 0.7239159204323409, "learning_rate": 1.6087052097406242e-07, "loss": 0.3881, "step": 26659 }, { "epoch": 1.8430694780504666, "grad_norm": 0.772549818323667, "learning_rate": 1.6072970422685186e-07, "loss": 0.2762, "step": 26660 }, { "epoch": 1.8431386104389906, "grad_norm": 0.8347244621170054, "learning_rate": 1.605889481312589e-07, "loss": 0.241, "step": 26661 }, { "epoch": 1.8432077428275147, "grad_norm": 0.7601605314840091, "learning_rate": 1.6044825268904663e-07, "loss": 0.3323, "step": 26662 }, { "epoch": 1.8432768752160387, "grad_norm": 0.6366318850838534, "learning_rate": 1.6030761790197912e-07, "loss": 0.2624, "step": 26663 }, { "epoch": 1.8433460076045627, "grad_norm": 0.6373595868577219, "learning_rate": 1.6016704377181836e-07, "loss": 0.2101, "step": 26664 }, { "epoch": 1.8434151399930867, "grad_norm": 0.6801452242040382, "learning_rate": 1.6002653030032678e-07, "loss": 0.2811, "step": 26665 }, { "epoch": 1.8434842723816107, "grad_norm": 0.7563608468489359, "learning_rate": 1.5988607748926576e-07, "loss": 0.326, "step": 26666 }, { "epoch": 1.8435534047701347, "grad_norm": 1.097753379082161, "learning_rate": 1.5974568534039504e-07, "loss": 0.2141, "step": 26667 }, { "epoch": 1.8436225371586588, "grad_norm": 0.7490285977028585, "learning_rate": 1.5960535385547425e-07, "loss": 0.2716, "step": 26668 }, { "epoch": 1.8436916695471828, "grad_norm": 0.7084820954462994, "learning_rate": 1.5946508303626263e-07, "loss": 0.3095, "step": 26669 }, { "epoch": 1.8437608019357068, "grad_norm": 0.7631463087068382, "learning_rate": 1.5932487288451703e-07, "loss": 0.3131, "step": 26670 }, { "epoch": 1.8438299343242308, "grad_norm": 0.6862052947226668, "learning_rate": 1.591847234019961e-07, "loss": 0.1979, "step": 26671 }, { "epoch": 1.8438990667127548, "grad_norm": 0.7482087546969731, "learning_rate": 1.5904463459045616e-07, "loss": 0.2931, "step": 26672 }, { "epoch": 1.8439681991012788, "grad_norm": 0.7086070995529712, "learning_rate": 1.5890460645165195e-07, "loss": 0.3657, "step": 26673 }, { "epoch": 1.8440373314898029, "grad_norm": 0.8011749149617939, "learning_rate": 1.5876463898733984e-07, "loss": 0.2544, "step": 26674 }, { "epoch": 1.8441064638783269, "grad_norm": 0.6730991419689939, "learning_rate": 1.5862473219927288e-07, "loss": 0.2113, "step": 26675 }, { "epoch": 1.844175596266851, "grad_norm": 0.8362987318169836, "learning_rate": 1.5848488608920577e-07, "loss": 0.2188, "step": 26676 }, { "epoch": 1.844244728655375, "grad_norm": 0.7367478596321556, "learning_rate": 1.5834510065888985e-07, "loss": 0.3001, "step": 26677 }, { "epoch": 1.844313861043899, "grad_norm": 0.7926057442583648, "learning_rate": 1.5820537591007768e-07, "loss": 0.2521, "step": 26678 }, { "epoch": 1.844382993432423, "grad_norm": 0.7967209646942598, "learning_rate": 1.5806571184452114e-07, "loss": 0.2602, "step": 26679 }, { "epoch": 1.844452125820947, "grad_norm": 0.8683265887132122, "learning_rate": 1.5792610846396995e-07, "loss": 0.3171, "step": 26680 }, { "epoch": 1.844521258209471, "grad_norm": 0.7250107846710317, "learning_rate": 1.5778656577017325e-07, "loss": 0.2554, "step": 26681 }, { "epoch": 1.844590390597995, "grad_norm": 0.7154376960532784, "learning_rate": 1.5764708376488191e-07, "loss": 0.2481, "step": 26682 }, { "epoch": 1.844659522986519, "grad_norm": 0.7168344847826819, "learning_rate": 1.5750766244984173e-07, "loss": 0.2637, "step": 26683 }, { "epoch": 1.844728655375043, "grad_norm": 0.7640630974436614, "learning_rate": 1.5736830182680185e-07, "loss": 0.3125, "step": 26684 }, { "epoch": 1.844797787763567, "grad_norm": 0.7123637295202644, "learning_rate": 1.57229001897507e-07, "loss": 0.2435, "step": 26685 }, { "epoch": 1.8448669201520913, "grad_norm": 0.7138873569385863, "learning_rate": 1.570897626637058e-07, "loss": 0.2044, "step": 26686 }, { "epoch": 1.8449360525406153, "grad_norm": 0.6792789861265798, "learning_rate": 1.5695058412714125e-07, "loss": 0.3155, "step": 26687 }, { "epoch": 1.8450051849291393, "grad_norm": 0.6698144054248834, "learning_rate": 1.5681146628955867e-07, "loss": 0.2837, "step": 26688 }, { "epoch": 1.8450743173176634, "grad_norm": 0.6269910028488966, "learning_rate": 1.5667240915270055e-07, "loss": 0.2016, "step": 26689 }, { "epoch": 1.8451434497061874, "grad_norm": 0.7988935200331462, "learning_rate": 1.5653341271831102e-07, "loss": 0.2955, "step": 26690 }, { "epoch": 1.8452125820947114, "grad_norm": 0.7322739458481679, "learning_rate": 1.5639447698813148e-07, "loss": 0.3628, "step": 26691 }, { "epoch": 1.8452817144832354, "grad_norm": 0.7404566316597159, "learning_rate": 1.5625560196390333e-07, "loss": 0.2394, "step": 26692 }, { "epoch": 1.8453508468717594, "grad_norm": 0.6178137505229934, "learning_rate": 1.561167876473668e-07, "loss": 0.1898, "step": 26693 }, { "epoch": 1.8454199792602834, "grad_norm": 0.6860426309461561, "learning_rate": 1.559780340402628e-07, "loss": 0.3332, "step": 26694 }, { "epoch": 1.8454891116488075, "grad_norm": 0.7258639432174744, "learning_rate": 1.558393411443293e-07, "loss": 0.3103, "step": 26695 }, { "epoch": 1.8455582440373315, "grad_norm": 0.865509376276194, "learning_rate": 1.5570070896130497e-07, "loss": 0.2377, "step": 26696 }, { "epoch": 1.8456273764258555, "grad_norm": 0.7074824799951263, "learning_rate": 1.555621374929267e-07, "loss": 0.2178, "step": 26697 }, { "epoch": 1.8456965088143795, "grad_norm": 0.708769960569854, "learning_rate": 1.554236267409326e-07, "loss": 0.2875, "step": 26698 }, { "epoch": 1.8457656412029035, "grad_norm": 0.7595058221792511, "learning_rate": 1.552851767070568e-07, "loss": 0.2437, "step": 26699 }, { "epoch": 1.8458347735914276, "grad_norm": 0.6639114075610107, "learning_rate": 1.5514678739303623e-07, "loss": 0.2058, "step": 26700 }, { "epoch": 1.8459039059799516, "grad_norm": 0.7052269017257227, "learning_rate": 1.55008458800604e-07, "loss": 0.261, "step": 26701 }, { "epoch": 1.8459730383684756, "grad_norm": 0.691753848846748, "learning_rate": 1.5487019093149535e-07, "loss": 0.3753, "step": 26702 }, { "epoch": 1.8460421707569996, "grad_norm": 0.7994347968498263, "learning_rate": 1.547319837874417e-07, "loss": 0.2318, "step": 26703 }, { "epoch": 1.8461113031455236, "grad_norm": 0.769212168731476, "learning_rate": 1.5459383737017664e-07, "loss": 0.2117, "step": 26704 }, { "epoch": 1.8461804355340476, "grad_norm": 0.7348839492335801, "learning_rate": 1.544557516814299e-07, "loss": 0.3365, "step": 26705 }, { "epoch": 1.8462495679225719, "grad_norm": 0.6259771774942167, "learning_rate": 1.5431772672293288e-07, "loss": 0.2396, "step": 26706 }, { "epoch": 1.846318700311096, "grad_norm": 0.687727684891226, "learning_rate": 1.5417976249641587e-07, "loss": 0.238, "step": 26707 }, { "epoch": 1.84638783269962, "grad_norm": 0.6710608187547412, "learning_rate": 1.5404185900360856e-07, "loss": 0.2554, "step": 26708 }, { "epoch": 1.846456965088144, "grad_norm": 0.6982436434926053, "learning_rate": 1.5390401624623796e-07, "loss": 0.3387, "step": 26709 }, { "epoch": 1.846526097476668, "grad_norm": 0.7356589426284115, "learning_rate": 1.5376623422603265e-07, "loss": 0.2522, "step": 26710 }, { "epoch": 1.846595229865192, "grad_norm": 0.7034154830229702, "learning_rate": 1.5362851294471849e-07, "loss": 0.2914, "step": 26711 }, { "epoch": 1.846664362253716, "grad_norm": 0.7144879035259907, "learning_rate": 1.5349085240402294e-07, "loss": 0.3128, "step": 26712 }, { "epoch": 1.84673349464224, "grad_norm": 0.7122599068788338, "learning_rate": 1.5335325260567024e-07, "loss": 0.3065, "step": 26713 }, { "epoch": 1.846802627030764, "grad_norm": 0.7431412102469371, "learning_rate": 1.5321571355138455e-07, "loss": 0.2547, "step": 26714 }, { "epoch": 1.846871759419288, "grad_norm": 0.772050519062446, "learning_rate": 1.530782352428911e-07, "loss": 0.2634, "step": 26715 }, { "epoch": 1.846940891807812, "grad_norm": 0.6590439985387698, "learning_rate": 1.529408176819125e-07, "loss": 0.3135, "step": 26716 }, { "epoch": 1.847010024196336, "grad_norm": 0.7739473043566991, "learning_rate": 1.5280346087017007e-07, "loss": 0.2544, "step": 26717 }, { "epoch": 1.84707915658486, "grad_norm": 0.761153634728952, "learning_rate": 1.526661648093869e-07, "loss": 0.2693, "step": 26718 }, { "epoch": 1.8471482889733841, "grad_norm": 1.0673522381351794, "learning_rate": 1.5252892950128218e-07, "loss": 0.2972, "step": 26719 }, { "epoch": 1.8472174213619081, "grad_norm": 0.741324668270764, "learning_rate": 1.523917549475773e-07, "loss": 0.2901, "step": 26720 }, { "epoch": 1.8472865537504322, "grad_norm": 0.6632703510897977, "learning_rate": 1.522546411499898e-07, "loss": 0.2021, "step": 26721 }, { "epoch": 1.8473556861389562, "grad_norm": 0.7586311982369321, "learning_rate": 1.521175881102399e-07, "loss": 0.2603, "step": 26722 }, { "epoch": 1.8474248185274802, "grad_norm": 0.9875118319503086, "learning_rate": 1.5198059583004465e-07, "loss": 0.3472, "step": 26723 }, { "epoch": 1.8474939509160042, "grad_norm": 0.7040879761867677, "learning_rate": 1.5184366431112096e-07, "loss": 0.2747, "step": 26724 }, { "epoch": 1.8475630833045282, "grad_norm": 0.680072717865847, "learning_rate": 1.517067935551847e-07, "loss": 0.2343, "step": 26725 }, { "epoch": 1.8476322156930522, "grad_norm": 0.698457419006534, "learning_rate": 1.5156998356395225e-07, "loss": 0.2644, "step": 26726 }, { "epoch": 1.8477013480815763, "grad_norm": 0.6601907581413772, "learning_rate": 1.514332343391367e-07, "loss": 0.309, "step": 26727 }, { "epoch": 1.8477704804701003, "grad_norm": 0.6237092704287382, "learning_rate": 1.5129654588245335e-07, "loss": 0.2304, "step": 26728 }, { "epoch": 1.8478396128586243, "grad_norm": 0.8069586479008887, "learning_rate": 1.5115991819561472e-07, "loss": 0.2949, "step": 26729 }, { "epoch": 1.8479087452471483, "grad_norm": 0.6846848820720293, "learning_rate": 1.5102335128033386e-07, "loss": 0.3257, "step": 26730 }, { "epoch": 1.8479778776356723, "grad_norm": 0.7478376407669158, "learning_rate": 1.5088684513832163e-07, "loss": 0.2765, "step": 26731 }, { "epoch": 1.8480470100241964, "grad_norm": 0.7122064523974101, "learning_rate": 1.5075039977128948e-07, "loss": 0.2178, "step": 26732 }, { "epoch": 1.8481161424127204, "grad_norm": 0.6804628720552212, "learning_rate": 1.5061401518094654e-07, "loss": 0.2324, "step": 26733 }, { "epoch": 1.8481852748012444, "grad_norm": 0.7584612503480228, "learning_rate": 1.504776913690037e-07, "loss": 0.3355, "step": 26734 }, { "epoch": 1.8482544071897684, "grad_norm": 0.7135619448833374, "learning_rate": 1.5034142833716792e-07, "loss": 0.2647, "step": 26735 }, { "epoch": 1.8483235395782924, "grad_norm": 0.8665148686572129, "learning_rate": 1.5020522608714726e-07, "loss": 0.1999, "step": 26736 }, { "epoch": 1.8483926719668164, "grad_norm": 0.6953060333183209, "learning_rate": 1.500690846206504e-07, "loss": 0.3149, "step": 26737 }, { "epoch": 1.8484618043553405, "grad_norm": 0.7066076760496988, "learning_rate": 1.4993300393938205e-07, "loss": 0.2746, "step": 26738 }, { "epoch": 1.8485309367438645, "grad_norm": 0.7194761776073373, "learning_rate": 1.4979698404504807e-07, "loss": 0.2117, "step": 26739 }, { "epoch": 1.8486000691323885, "grad_norm": 0.6519420578116425, "learning_rate": 1.496610249393532e-07, "loss": 0.1947, "step": 26740 }, { "epoch": 1.8486692015209125, "grad_norm": 0.6976458089321614, "learning_rate": 1.4952512662400165e-07, "loss": 0.355, "step": 26741 }, { "epoch": 1.8487383339094365, "grad_norm": 0.6885561615921595, "learning_rate": 1.493892891006965e-07, "loss": 0.2358, "step": 26742 }, { "epoch": 1.8488074662979606, "grad_norm": 0.6727977932980487, "learning_rate": 1.4925351237114027e-07, "loss": 0.2217, "step": 26743 }, { "epoch": 1.8488765986864846, "grad_norm": 0.7851040528261957, "learning_rate": 1.4911779643703493e-07, "loss": 0.2653, "step": 26744 }, { "epoch": 1.8489457310750086, "grad_norm": 0.6613842726981685, "learning_rate": 1.4898214130008137e-07, "loss": 0.2896, "step": 26745 }, { "epoch": 1.8490148634635326, "grad_norm": 0.7711098833905126, "learning_rate": 1.4884654696197932e-07, "loss": 0.2207, "step": 26746 }, { "epoch": 1.8490839958520566, "grad_norm": 0.7261769016558508, "learning_rate": 1.4871101342442905e-07, "loss": 0.2613, "step": 26747 }, { "epoch": 1.8491531282405806, "grad_norm": 0.6371491211333103, "learning_rate": 1.4857554068912816e-07, "loss": 0.311, "step": 26748 }, { "epoch": 1.8492222606291047, "grad_norm": 0.7273509023718705, "learning_rate": 1.484401287577747e-07, "loss": 0.262, "step": 26749 }, { "epoch": 1.8492913930176287, "grad_norm": 0.6839538825920197, "learning_rate": 1.4830477763206675e-07, "loss": 0.2438, "step": 26750 }, { "epoch": 1.8493605254061527, "grad_norm": 0.7198289213988409, "learning_rate": 1.481694873137002e-07, "loss": 0.281, "step": 26751 }, { "epoch": 1.8494296577946767, "grad_norm": 0.777250906548487, "learning_rate": 1.4803425780437086e-07, "loss": 0.3417, "step": 26752 }, { "epoch": 1.8494987901832007, "grad_norm": 0.7322021884422526, "learning_rate": 1.4789908910577356e-07, "loss": 0.245, "step": 26753 }, { "epoch": 1.8495679225717248, "grad_norm": 0.7423497225951873, "learning_rate": 1.4776398121960135e-07, "loss": 0.2848, "step": 26754 }, { "epoch": 1.8496370549602488, "grad_norm": 0.7305429146216412, "learning_rate": 1.47628934147549e-07, "loss": 0.3327, "step": 26755 }, { "epoch": 1.8497061873487728, "grad_norm": 0.6580305507148198, "learning_rate": 1.474939478913079e-07, "loss": 0.2135, "step": 26756 }, { "epoch": 1.8497753197372968, "grad_norm": 0.8013876835392012, "learning_rate": 1.4735902245257118e-07, "loss": 0.2042, "step": 26757 }, { "epoch": 1.8498444521258208, "grad_norm": 0.8283196795997969, "learning_rate": 1.4722415783302912e-07, "loss": 0.2927, "step": 26758 }, { "epoch": 1.8499135845143448, "grad_norm": 0.7114973780778859, "learning_rate": 1.4708935403437208e-07, "loss": 0.2888, "step": 26759 }, { "epoch": 1.8499827169028689, "grad_norm": 0.6444017920674576, "learning_rate": 1.4695461105828923e-07, "loss": 0.213, "step": 26760 }, { "epoch": 1.8500518492913929, "grad_norm": 0.8737224167610755, "learning_rate": 1.4681992890647035e-07, "loss": 0.17, "step": 26761 }, { "epoch": 1.850120981679917, "grad_norm": 0.7841888008923359, "learning_rate": 1.4668530758060185e-07, "loss": 0.329, "step": 26762 }, { "epoch": 1.850190114068441, "grad_norm": 0.7769646710314165, "learning_rate": 1.4655074708237237e-07, "loss": 0.266, "step": 26763 }, { "epoch": 1.850259246456965, "grad_norm": 0.6816564427976936, "learning_rate": 1.4641624741346783e-07, "loss": 0.2502, "step": 26764 }, { "epoch": 1.850328378845489, "grad_norm": 0.7582284165290776, "learning_rate": 1.4628180857557406e-07, "loss": 0.351, "step": 26765 }, { "epoch": 1.850397511234013, "grad_norm": 0.6735462718421734, "learning_rate": 1.4614743057037583e-07, "loss": 0.3478, "step": 26766 }, { "epoch": 1.8504666436225372, "grad_norm": 0.7370145589299275, "learning_rate": 1.4601311339955792e-07, "loss": 0.2567, "step": 26767 }, { "epoch": 1.8505357760110612, "grad_norm": 0.7880760998625296, "learning_rate": 1.4587885706480288e-07, "loss": 0.2498, "step": 26768 }, { "epoch": 1.8506049083995852, "grad_norm": 0.7224211385976693, "learning_rate": 1.4574466156779432e-07, "loss": 0.2717, "step": 26769 }, { "epoch": 1.8506740407881093, "grad_norm": 0.6733215309501986, "learning_rate": 1.4561052691021316e-07, "loss": 0.2703, "step": 26770 }, { "epoch": 1.8507431731766333, "grad_norm": 0.7170433315213476, "learning_rate": 1.4547645309374082e-07, "loss": 0.2561, "step": 26771 }, { "epoch": 1.8508123055651573, "grad_norm": 0.7300653131792325, "learning_rate": 1.4534244012005817e-07, "loss": 0.2702, "step": 26772 }, { "epoch": 1.8508814379536813, "grad_norm": 0.7712849092582017, "learning_rate": 1.452084879908444e-07, "loss": 0.3067, "step": 26773 }, { "epoch": 1.8509505703422053, "grad_norm": 0.7138552925291273, "learning_rate": 1.4507459670777823e-07, "loss": 0.2407, "step": 26774 }, { "epoch": 1.8510197027307294, "grad_norm": 0.835432772038695, "learning_rate": 1.4494076627253883e-07, "loss": 0.2161, "step": 26775 }, { "epoch": 1.8510888351192534, "grad_norm": 0.6900892279290681, "learning_rate": 1.448069966868021e-07, "loss": 0.2687, "step": 26776 }, { "epoch": 1.8511579675077774, "grad_norm": 0.7683700823870253, "learning_rate": 1.44673287952245e-07, "loss": 0.3685, "step": 26777 }, { "epoch": 1.8512270998963014, "grad_norm": 0.7188399953602852, "learning_rate": 1.4453964007054343e-07, "loss": 0.2321, "step": 26778 }, { "epoch": 1.8512962322848254, "grad_norm": 0.7418854061694767, "learning_rate": 1.4440605304337328e-07, "loss": 0.2847, "step": 26779 }, { "epoch": 1.8513653646733494, "grad_norm": 0.7612438345255657, "learning_rate": 1.442725268724071e-07, "loss": 0.3151, "step": 26780 }, { "epoch": 1.8514344970618735, "grad_norm": 0.6941583627428657, "learning_rate": 1.4413906155932023e-07, "loss": 0.2548, "step": 26781 }, { "epoch": 1.8515036294503975, "grad_norm": 0.6590991749423558, "learning_rate": 1.4400565710578406e-07, "loss": 0.2142, "step": 26782 }, { "epoch": 1.8515727618389215, "grad_norm": 0.656788828696007, "learning_rate": 1.438723135134712e-07, "loss": 0.2873, "step": 26783 }, { "epoch": 1.8516418942274455, "grad_norm": 0.7810108169549252, "learning_rate": 1.437390307840525e-07, "loss": 0.3199, "step": 26784 }, { "epoch": 1.8517110266159695, "grad_norm": 0.8554127237182969, "learning_rate": 1.4360580891919884e-07, "loss": 0.3436, "step": 26785 }, { "epoch": 1.8517801590044936, "grad_norm": 0.6584918256959597, "learning_rate": 1.4347264792057946e-07, "loss": 0.2186, "step": 26786 }, { "epoch": 1.8518492913930178, "grad_norm": 0.7262197640285045, "learning_rate": 1.433395477898647e-07, "loss": 0.3635, "step": 26787 }, { "epoch": 1.8519184237815418, "grad_norm": 0.8107686293231472, "learning_rate": 1.4320650852872044e-07, "loss": 0.3438, "step": 26788 }, { "epoch": 1.8519875561700658, "grad_norm": 0.6938565942321625, "learning_rate": 1.4307353013881585e-07, "loss": 0.2515, "step": 26789 }, { "epoch": 1.8520566885585898, "grad_norm": 0.6770827821147107, "learning_rate": 1.4294061262181637e-07, "loss": 0.2594, "step": 26790 }, { "epoch": 1.8521258209471139, "grad_norm": 0.7145030181665433, "learning_rate": 1.4280775597938944e-07, "loss": 0.4273, "step": 26791 }, { "epoch": 1.8521949533356379, "grad_norm": 0.7098362092458194, "learning_rate": 1.426749602131977e-07, "loss": 0.2889, "step": 26792 }, { "epoch": 1.852264085724162, "grad_norm": 0.6747739710287021, "learning_rate": 1.4254222532490869e-07, "loss": 0.2267, "step": 26793 }, { "epoch": 1.852333218112686, "grad_norm": 0.7392606496962844, "learning_rate": 1.424095513161833e-07, "loss": 0.3202, "step": 26794 }, { "epoch": 1.85240235050121, "grad_norm": 0.7002470289390339, "learning_rate": 1.4227693818868626e-07, "loss": 0.2652, "step": 26795 }, { "epoch": 1.852471482889734, "grad_norm": 0.8865174536632965, "learning_rate": 1.42144385944078e-07, "loss": 0.2145, "step": 26796 }, { "epoch": 1.852540615278258, "grad_norm": 0.7007982226026012, "learning_rate": 1.4201189458402154e-07, "loss": 0.2577, "step": 26797 }, { "epoch": 1.852609747666782, "grad_norm": 0.7132304946889868, "learning_rate": 1.4187946411017563e-07, "loss": 0.3359, "step": 26798 }, { "epoch": 1.852678880055306, "grad_norm": 0.7024098491284021, "learning_rate": 1.417470945242011e-07, "loss": 0.2349, "step": 26799 }, { "epoch": 1.85274801244383, "grad_norm": 0.649621765191301, "learning_rate": 1.4161478582775668e-07, "loss": 0.2534, "step": 26800 }, { "epoch": 1.852817144832354, "grad_norm": 0.8522919725420435, "learning_rate": 1.4148253802250155e-07, "loss": 0.3171, "step": 26801 }, { "epoch": 1.852886277220878, "grad_norm": 0.6845408384319434, "learning_rate": 1.4135035111009166e-07, "loss": 0.3487, "step": 26802 }, { "epoch": 1.852955409609402, "grad_norm": 0.7155852316456661, "learning_rate": 1.4121822509218507e-07, "loss": 0.2313, "step": 26803 }, { "epoch": 1.853024541997926, "grad_norm": 0.6219585448024739, "learning_rate": 1.4108615997043717e-07, "loss": 0.2146, "step": 26804 }, { "epoch": 1.8530936743864501, "grad_norm": 0.7743065011372844, "learning_rate": 1.4095415574650272e-07, "loss": 0.3322, "step": 26805 }, { "epoch": 1.8531628067749741, "grad_norm": 0.7677985162047725, "learning_rate": 1.4082221242203598e-07, "loss": 0.2806, "step": 26806 }, { "epoch": 1.8532319391634982, "grad_norm": 15.716223411682389, "learning_rate": 1.406903299986917e-07, "loss": 0.2015, "step": 26807 }, { "epoch": 1.8533010715520222, "grad_norm": 1.1538767689375073, "learning_rate": 1.4055850847812302e-07, "loss": 0.2676, "step": 26808 }, { "epoch": 1.8533702039405462, "grad_norm": 0.6481780037099175, "learning_rate": 1.404267478619803e-07, "loss": 0.2838, "step": 26809 }, { "epoch": 1.8534393363290702, "grad_norm": 0.7410347109187327, "learning_rate": 1.4029504815191664e-07, "loss": 0.2237, "step": 26810 }, { "epoch": 1.8535084687175942, "grad_norm": 0.6451720413135268, "learning_rate": 1.4016340934958184e-07, "loss": 0.2441, "step": 26811 }, { "epoch": 1.8535776011061182, "grad_norm": 0.7117227557024296, "learning_rate": 1.4003183145662568e-07, "loss": 0.2794, "step": 26812 }, { "epoch": 1.8536467334946423, "grad_norm": 0.7514884234143083, "learning_rate": 1.3990031447469744e-07, "loss": 0.3332, "step": 26813 }, { "epoch": 1.8537158658831663, "grad_norm": 0.6407311685407139, "learning_rate": 1.397688584054463e-07, "loss": 0.3018, "step": 26814 }, { "epoch": 1.8537849982716903, "grad_norm": 0.685532696145374, "learning_rate": 1.3963746325051874e-07, "loss": 0.2752, "step": 26815 }, { "epoch": 1.8538541306602143, "grad_norm": 0.8522753739386227, "learning_rate": 1.3950612901156178e-07, "loss": 0.3133, "step": 26816 }, { "epoch": 1.8539232630487383, "grad_norm": 3.7070971904708188, "learning_rate": 1.3937485569022136e-07, "loss": 0.2166, "step": 26817 }, { "epoch": 1.8539923954372624, "grad_norm": 0.7733801285725445, "learning_rate": 1.3924364328814333e-07, "loss": 0.2679, "step": 26818 }, { "epoch": 1.8540615278257864, "grad_norm": 0.7298055794120925, "learning_rate": 1.391124918069714e-07, "loss": 0.3432, "step": 26819 }, { "epoch": 1.8541306602143104, "grad_norm": 0.7549103543138452, "learning_rate": 1.389814012483498e-07, "loss": 0.3155, "step": 26820 }, { "epoch": 1.8541997926028344, "grad_norm": 0.7350557742593589, "learning_rate": 1.3885037161392167e-07, "loss": 0.229, "step": 26821 }, { "epoch": 1.8542689249913584, "grad_norm": 0.6932346446644291, "learning_rate": 1.3871940290532903e-07, "loss": 0.2258, "step": 26822 }, { "epoch": 1.8543380573798824, "grad_norm": 0.7321197738265335, "learning_rate": 1.385884951242128e-07, "loss": 0.3381, "step": 26823 }, { "epoch": 1.8544071897684065, "grad_norm": 0.6506994866337819, "learning_rate": 1.3845764827221498e-07, "loss": 0.2539, "step": 26824 }, { "epoch": 1.8544763221569305, "grad_norm": 0.7070348163355501, "learning_rate": 1.3832686235097426e-07, "loss": 0.2379, "step": 26825 }, { "epoch": 1.8545454545454545, "grad_norm": 0.7042993223841032, "learning_rate": 1.38196137362131e-07, "loss": 0.2918, "step": 26826 }, { "epoch": 1.8546145869339785, "grad_norm": 0.7085157241669765, "learning_rate": 1.3806547330732167e-07, "loss": 0.3497, "step": 26827 }, { "epoch": 1.8546837193225025, "grad_norm": 0.6807300803270774, "learning_rate": 1.3793487018818608e-07, "loss": 0.2701, "step": 26828 }, { "epoch": 1.8547528517110266, "grad_norm": 0.7202140057384611, "learning_rate": 1.3780432800635956e-07, "loss": 0.2347, "step": 26829 }, { "epoch": 1.8548219840995506, "grad_norm": 0.6970005472217758, "learning_rate": 1.3767384676347917e-07, "loss": 0.3694, "step": 26830 }, { "epoch": 1.8548911164880746, "grad_norm": 0.7685695804336483, "learning_rate": 1.3754342646117967e-07, "loss": 0.2568, "step": 26831 }, { "epoch": 1.8549602488765986, "grad_norm": 0.7026601622906057, "learning_rate": 1.3741306710109647e-07, "loss": 0.2187, "step": 26832 }, { "epoch": 1.8550293812651226, "grad_norm": 0.7105612207990363, "learning_rate": 1.3728276868486211e-07, "loss": 0.2782, "step": 26833 }, { "epoch": 1.8550985136536466, "grad_norm": 0.7020619129937598, "learning_rate": 1.3715253121411032e-07, "loss": 0.3219, "step": 26834 }, { "epoch": 1.8551676460421707, "grad_norm": 0.6846857660881577, "learning_rate": 1.370223546904731e-07, "loss": 0.2292, "step": 26835 }, { "epoch": 1.8552367784306947, "grad_norm": 0.685244689618925, "learning_rate": 1.3689223911558357e-07, "loss": 0.2176, "step": 26836 }, { "epoch": 1.8553059108192187, "grad_norm": 0.7366635130793258, "learning_rate": 1.3676218449106992e-07, "loss": 0.3152, "step": 26837 }, { "epoch": 1.8553750432077427, "grad_norm": 0.844368504212402, "learning_rate": 1.3663219081856415e-07, "loss": 0.2782, "step": 26838 }, { "epoch": 1.8554441755962667, "grad_norm": 0.7386030104754696, "learning_rate": 1.3650225809969498e-07, "loss": 0.2121, "step": 26839 }, { "epoch": 1.8555133079847907, "grad_norm": 0.6900893416823483, "learning_rate": 1.3637238633609052e-07, "loss": 0.2569, "step": 26840 }, { "epoch": 1.8555824403733148, "grad_norm": 0.7000327432321932, "learning_rate": 1.3624257552937836e-07, "loss": 0.338, "step": 26841 }, { "epoch": 1.8556515727618388, "grad_norm": 0.7711585343292096, "learning_rate": 1.3611282568118556e-07, "loss": 0.2646, "step": 26842 }, { "epoch": 1.8557207051503628, "grad_norm": 0.6352096954878934, "learning_rate": 1.3598313679313857e-07, "loss": 0.2037, "step": 26843 }, { "epoch": 1.8557898375388868, "grad_norm": 0.8456418996513847, "learning_rate": 1.3585350886686332e-07, "loss": 0.3763, "step": 26844 }, { "epoch": 1.8558589699274108, "grad_norm": 0.686291804757773, "learning_rate": 1.3572394190398353e-07, "loss": 0.3158, "step": 26845 }, { "epoch": 1.8559281023159349, "grad_norm": 0.6409028259808515, "learning_rate": 1.3559443590612398e-07, "loss": 0.2168, "step": 26846 }, { "epoch": 1.8559972347044589, "grad_norm": 0.6984015214500775, "learning_rate": 1.3546499087490617e-07, "loss": 0.2871, "step": 26847 }, { "epoch": 1.8560663670929831, "grad_norm": 0.9091636263670403, "learning_rate": 1.3533560681195434e-07, "loss": 0.3518, "step": 26848 }, { "epoch": 1.8561354994815071, "grad_norm": 0.8026327984259471, "learning_rate": 1.3520628371888887e-07, "loss": 0.2282, "step": 26849 }, { "epoch": 1.8562046318700312, "grad_norm": 0.7321182529154857, "learning_rate": 1.3507702159733127e-07, "loss": 0.2401, "step": 26850 }, { "epoch": 1.8562737642585552, "grad_norm": 0.7345068908429793, "learning_rate": 1.3494782044890132e-07, "loss": 0.2888, "step": 26851 }, { "epoch": 1.8563428966470792, "grad_norm": 0.7801154559214244, "learning_rate": 1.3481868027521828e-07, "loss": 0.2998, "step": 26852 }, { "epoch": 1.8564120290356032, "grad_norm": 0.7130719054048466, "learning_rate": 1.3468960107790086e-07, "loss": 0.2137, "step": 26853 }, { "epoch": 1.8564811614241272, "grad_norm": 0.6865517753323923, "learning_rate": 1.345605828585672e-07, "loss": 0.2667, "step": 26854 }, { "epoch": 1.8565502938126512, "grad_norm": 0.7440567104181484, "learning_rate": 1.3443162561883328e-07, "loss": 0.3774, "step": 26855 }, { "epoch": 1.8566194262011753, "grad_norm": 0.683560639814498, "learning_rate": 1.343027293603155e-07, "loss": 0.2519, "step": 26856 }, { "epoch": 1.8566885585896993, "grad_norm": 0.7215923202882562, "learning_rate": 1.3417389408463043e-07, "loss": 0.2018, "step": 26857 }, { "epoch": 1.8567576909782233, "grad_norm": 0.7326958547024935, "learning_rate": 1.340451197933923e-07, "loss": 0.2756, "step": 26858 }, { "epoch": 1.8568268233667473, "grad_norm": 0.7292702040759221, "learning_rate": 1.3391640648821424e-07, "loss": 0.3297, "step": 26859 }, { "epoch": 1.8568959557552713, "grad_norm": 0.7180253607527114, "learning_rate": 1.3378775417071111e-07, "loss": 0.261, "step": 26860 }, { "epoch": 1.8569650881437954, "grad_norm": 0.6670981348941664, "learning_rate": 1.336591628424938e-07, "loss": 0.2769, "step": 26861 }, { "epoch": 1.8570342205323194, "grad_norm": 0.653582026510294, "learning_rate": 1.335306325051744e-07, "loss": 0.3161, "step": 26862 }, { "epoch": 1.8571033529208434, "grad_norm": 0.8612768682675512, "learning_rate": 1.3340216316036324e-07, "loss": 0.2149, "step": 26863 }, { "epoch": 1.8571724853093674, "grad_norm": 0.6687016063186407, "learning_rate": 1.3327375480967243e-07, "loss": 0.2323, "step": 26864 }, { "epoch": 1.8572416176978914, "grad_norm": 0.7893395629247445, "learning_rate": 1.331454074547095e-07, "loss": 0.237, "step": 26865 }, { "epoch": 1.8573107500864154, "grad_norm": 0.685438489123669, "learning_rate": 1.3301712109708376e-07, "loss": 0.3678, "step": 26866 }, { "epoch": 1.8573798824749395, "grad_norm": 0.7151006621822807, "learning_rate": 1.3288889573840225e-07, "loss": 0.2344, "step": 26867 }, { "epoch": 1.8574490148634637, "grad_norm": 0.7035034764199677, "learning_rate": 1.3276073138027368e-07, "loss": 0.2039, "step": 26868 }, { "epoch": 1.8575181472519877, "grad_norm": 0.693148677726639, "learning_rate": 1.3263262802430288e-07, "loss": 0.3116, "step": 26869 }, { "epoch": 1.8575872796405117, "grad_norm": 0.7966182222511723, "learning_rate": 1.3250458567209524e-07, "loss": 0.2923, "step": 26870 }, { "epoch": 1.8576564120290358, "grad_norm": 0.6750370444156552, "learning_rate": 1.3237660432525722e-07, "loss": 0.2276, "step": 26871 }, { "epoch": 1.8577255444175598, "grad_norm": 0.7942795832285159, "learning_rate": 1.3224868398539093e-07, "loss": 0.3034, "step": 26872 }, { "epoch": 1.8577946768060838, "grad_norm": 0.7502918752190417, "learning_rate": 1.321208246541017e-07, "loss": 0.3158, "step": 26873 }, { "epoch": 1.8578638091946078, "grad_norm": 2.487663394328661, "learning_rate": 1.3199302633298938e-07, "loss": 0.2266, "step": 26874 }, { "epoch": 1.8579329415831318, "grad_norm": 0.662828591771267, "learning_rate": 1.3186528902365825e-07, "loss": 0.21, "step": 26875 }, { "epoch": 1.8580020739716558, "grad_norm": 0.6895939869416379, "learning_rate": 1.3173761272770758e-07, "loss": 0.2934, "step": 26876 }, { "epoch": 1.8580712063601799, "grad_norm": 0.8747699874115606, "learning_rate": 1.3160999744673774e-07, "loss": 0.3322, "step": 26877 }, { "epoch": 1.8581403387487039, "grad_norm": 0.9602627877682723, "learning_rate": 1.3148244318234915e-07, "loss": 0.2637, "step": 26878 }, { "epoch": 1.858209471137228, "grad_norm": 0.7152471805730356, "learning_rate": 1.3135494993613994e-07, "loss": 0.3306, "step": 26879 }, { "epoch": 1.858278603525752, "grad_norm": 0.7018785167614461, "learning_rate": 1.3122751770970777e-07, "loss": 0.3605, "step": 26880 }, { "epoch": 1.858347735914276, "grad_norm": 1.3221257961533317, "learning_rate": 1.311001465046502e-07, "loss": 0.2398, "step": 26881 }, { "epoch": 1.8584168683028, "grad_norm": 0.7759046810636228, "learning_rate": 1.3097283632256264e-07, "loss": 0.2672, "step": 26882 }, { "epoch": 1.858486000691324, "grad_norm": 0.6708988748488919, "learning_rate": 1.3084558716504158e-07, "loss": 0.2722, "step": 26883 }, { "epoch": 1.858555133079848, "grad_norm": 0.7298278708505802, "learning_rate": 1.3071839903368188e-07, "loss": 0.3176, "step": 26884 }, { "epoch": 1.858624265468372, "grad_norm": 0.7193647898722499, "learning_rate": 1.3059127193007782e-07, "loss": 0.2074, "step": 26885 }, { "epoch": 1.858693397856896, "grad_norm": 0.6813800837499366, "learning_rate": 1.3046420585582197e-07, "loss": 0.2429, "step": 26886 }, { "epoch": 1.85876253024542, "grad_norm": 0.7827095998699393, "learning_rate": 1.3033720081250756e-07, "loss": 0.3551, "step": 26887 }, { "epoch": 1.858831662633944, "grad_norm": 0.7667417000800443, "learning_rate": 1.302102568017255e-07, "loss": 0.2961, "step": 26888 }, { "epoch": 1.858900795022468, "grad_norm": 0.7022459544186667, "learning_rate": 1.3008337382506786e-07, "loss": 0.2296, "step": 26889 }, { "epoch": 1.858969927410992, "grad_norm": 0.6927308106064541, "learning_rate": 1.2995655188412393e-07, "loss": 0.24, "step": 26890 }, { "epoch": 1.8590390597995161, "grad_norm": 0.70380409223373, "learning_rate": 1.2982979098048354e-07, "loss": 0.3846, "step": 26891 }, { "epoch": 1.8591081921880401, "grad_norm": 0.7041354832101798, "learning_rate": 1.2970309111573598e-07, "loss": 0.2045, "step": 26892 }, { "epoch": 1.8591773245765641, "grad_norm": 0.6404008007382763, "learning_rate": 1.2957645229146887e-07, "loss": 0.2703, "step": 26893 }, { "epoch": 1.8592464569650882, "grad_norm": 0.6829543863365173, "learning_rate": 1.2944987450926872e-07, "loss": 0.2939, "step": 26894 }, { "epoch": 1.8593155893536122, "grad_norm": 0.7894982590889816, "learning_rate": 1.2932335777072258e-07, "loss": 0.2876, "step": 26895 }, { "epoch": 1.8593847217421362, "grad_norm": 0.7480303967783956, "learning_rate": 1.291969020774164e-07, "loss": 0.2244, "step": 26896 }, { "epoch": 1.8594538541306602, "grad_norm": 1.1642739323880285, "learning_rate": 1.2907050743093451e-07, "loss": 0.2517, "step": 26897 }, { "epoch": 1.8595229865191842, "grad_norm": 0.7142421415972809, "learning_rate": 1.2894417383286117e-07, "loss": 0.356, "step": 26898 }, { "epoch": 1.8595921189077083, "grad_norm": 0.6931433424470219, "learning_rate": 1.2881790128477956e-07, "loss": 0.23, "step": 26899 }, { "epoch": 1.8596612512962323, "grad_norm": 0.7278304022541233, "learning_rate": 1.2869168978827285e-07, "loss": 0.236, "step": 26900 }, { "epoch": 1.8597303836847563, "grad_norm": 0.6737696385024945, "learning_rate": 1.2856553934492255e-07, "loss": 0.2926, "step": 26901 }, { "epoch": 1.8597995160732803, "grad_norm": 0.6797747114193462, "learning_rate": 1.2843944995630964e-07, "loss": 0.318, "step": 26902 }, { "epoch": 1.8598686484618043, "grad_norm": 0.6901433163684656, "learning_rate": 1.2831342162401506e-07, "loss": 0.2006, "step": 26903 }, { "epoch": 1.8599377808503283, "grad_norm": 0.7116219757535973, "learning_rate": 1.2818745434961755e-07, "loss": 0.2581, "step": 26904 }, { "epoch": 1.8600069132388524, "grad_norm": 0.779558680437331, "learning_rate": 1.2806154813469584e-07, "loss": 0.3542, "step": 26905 }, { "epoch": 1.8600760456273764, "grad_norm": 1.08484509398487, "learning_rate": 1.2793570298082815e-07, "loss": 0.2934, "step": 26906 }, { "epoch": 1.8601451780159004, "grad_norm": 0.9880940162315994, "learning_rate": 1.278099188895926e-07, "loss": 0.2334, "step": 26907 }, { "epoch": 1.8602143104044244, "grad_norm": 0.6676891072053952, "learning_rate": 1.2768419586256408e-07, "loss": 0.2458, "step": 26908 }, { "epoch": 1.8602834427929484, "grad_norm": 0.6980534217034202, "learning_rate": 1.2755853390131967e-07, "loss": 0.3118, "step": 26909 }, { "epoch": 1.8603525751814725, "grad_norm": 0.6650880174058273, "learning_rate": 1.2743293300743364e-07, "loss": 0.2368, "step": 26910 }, { "epoch": 1.8604217075699965, "grad_norm": 0.7185011708031355, "learning_rate": 1.2730739318248032e-07, "loss": 0.239, "step": 26911 }, { "epoch": 1.8604908399585205, "grad_norm": 0.6692358835413633, "learning_rate": 1.2718191442803284e-07, "loss": 0.3214, "step": 26912 }, { "epoch": 1.8605599723470445, "grad_norm": 0.8084626873096143, "learning_rate": 1.2705649674566446e-07, "loss": 0.2869, "step": 26913 }, { "epoch": 1.8606291047355685, "grad_norm": 0.7500357931071532, "learning_rate": 1.2693114013694608e-07, "loss": 0.2289, "step": 26914 }, { "epoch": 1.8606982371240925, "grad_norm": 0.7552989910488169, "learning_rate": 1.2680584460345035e-07, "loss": 0.2814, "step": 26915 }, { "epoch": 1.8607673695126166, "grad_norm": 0.7542123167038832, "learning_rate": 1.266806101467466e-07, "loss": 0.3576, "step": 26916 }, { "epoch": 1.8608365019011406, "grad_norm": 0.6670526656932375, "learning_rate": 1.2655543676840466e-07, "loss": 0.2228, "step": 26917 }, { "epoch": 1.8609056342896646, "grad_norm": 0.6611115299301191, "learning_rate": 1.264303244699927e-07, "loss": 0.2164, "step": 26918 }, { "epoch": 1.8609747666781886, "grad_norm": 0.6696880304879862, "learning_rate": 1.263052732530795e-07, "loss": 0.3483, "step": 26919 }, { "epoch": 1.8610438990667126, "grad_norm": 0.8088415421766941, "learning_rate": 1.2618028311923215e-07, "loss": 0.3006, "step": 26920 }, { "epoch": 1.8611130314552367, "grad_norm": 0.7012554185801351, "learning_rate": 1.260553540700177e-07, "loss": 0.2517, "step": 26921 }, { "epoch": 1.8611821638437607, "grad_norm": 0.7546323847452007, "learning_rate": 1.25930486107001e-07, "loss": 0.2277, "step": 26922 }, { "epoch": 1.8612512962322847, "grad_norm": 0.7708150443889553, "learning_rate": 1.25805679231748e-07, "loss": 0.3001, "step": 26923 }, { "epoch": 1.8613204286208087, "grad_norm": 0.6765068594770673, "learning_rate": 1.25680933445822e-07, "loss": 0.2737, "step": 26924 }, { "epoch": 1.8613895610093327, "grad_norm": 0.6896084973671227, "learning_rate": 1.255562487507872e-07, "loss": 0.2212, "step": 26925 }, { "epoch": 1.8614586933978567, "grad_norm": 0.7040135991529456, "learning_rate": 1.2543162514820572e-07, "loss": 0.2393, "step": 26926 }, { "epoch": 1.8615278257863808, "grad_norm": 0.6686219491396532, "learning_rate": 1.2530706263963964e-07, "loss": 0.2936, "step": 26927 }, { "epoch": 1.8615969581749048, "grad_norm": 0.6897892420730519, "learning_rate": 1.2518256122664996e-07, "loss": 0.2133, "step": 26928 }, { "epoch": 1.861666090563429, "grad_norm": 0.6911949245676412, "learning_rate": 1.2505812091079816e-07, "loss": 0.2427, "step": 26929 }, { "epoch": 1.861735222951953, "grad_norm": 0.7759070073341178, "learning_rate": 1.2493374169364247e-07, "loss": 0.3427, "step": 26930 }, { "epoch": 1.861804355340477, "grad_norm": 0.6670861337535121, "learning_rate": 1.248094235767433e-07, "loss": 0.2481, "step": 26931 }, { "epoch": 1.861873487729001, "grad_norm": 0.8541925742487153, "learning_rate": 1.2468516656165664e-07, "loss": 0.2003, "step": 26932 }, { "epoch": 1.861942620117525, "grad_norm": 0.698979540224769, "learning_rate": 1.2456097064994177e-07, "loss": 0.2844, "step": 26933 }, { "epoch": 1.8620117525060491, "grad_norm": 0.7540241705766942, "learning_rate": 1.2443683584315415e-07, "loss": 0.3401, "step": 26934 }, { "epoch": 1.8620808848945731, "grad_norm": 0.6965852469447404, "learning_rate": 1.2431276214285027e-07, "loss": 0.278, "step": 26935 }, { "epoch": 1.8621500172830971, "grad_norm": 0.7284143402830571, "learning_rate": 1.2418874955058446e-07, "loss": 0.2509, "step": 26936 }, { "epoch": 1.8622191496716212, "grad_norm": 0.7286733279482547, "learning_rate": 1.2406479806791216e-07, "loss": 0.3436, "step": 26937 }, { "epoch": 1.8622882820601452, "grad_norm": 0.7784834992134042, "learning_rate": 1.23940907696386e-07, "loss": 0.2852, "step": 26938 }, { "epoch": 1.8623574144486692, "grad_norm": 0.7266004852984045, "learning_rate": 1.2381707843755808e-07, "loss": 0.238, "step": 26939 }, { "epoch": 1.8624265468371932, "grad_norm": 0.7317536342757424, "learning_rate": 1.2369331029298158e-07, "loss": 0.2538, "step": 26940 }, { "epoch": 1.8624956792257172, "grad_norm": 0.7396959846042898, "learning_rate": 1.2356960326420697e-07, "loss": 0.3097, "step": 26941 }, { "epoch": 1.8625648116142413, "grad_norm": 0.6954125106546835, "learning_rate": 1.234459573527852e-07, "loss": 0.2384, "step": 26942 }, { "epoch": 1.8626339440027653, "grad_norm": 0.7020543733720316, "learning_rate": 1.2332237256026613e-07, "loss": 0.2014, "step": 26943 }, { "epoch": 1.8627030763912893, "grad_norm": 0.7591748990784225, "learning_rate": 1.23198848888198e-07, "loss": 0.2577, "step": 26944 }, { "epoch": 1.8627722087798133, "grad_norm": 0.7121385562247576, "learning_rate": 1.23075386338129e-07, "loss": 0.313, "step": 26945 }, { "epoch": 1.8628413411683373, "grad_norm": 0.7373146610048695, "learning_rate": 1.2295198491160732e-07, "loss": 0.2153, "step": 26946 }, { "epoch": 1.8629104735568613, "grad_norm": 0.8494556042034537, "learning_rate": 1.228286446101784e-07, "loss": 0.2676, "step": 26947 }, { "epoch": 1.8629796059453854, "grad_norm": 0.6800906038598161, "learning_rate": 1.2270536543538935e-07, "loss": 0.3419, "step": 26948 }, { "epoch": 1.8630487383339094, "grad_norm": 0.7205933309536222, "learning_rate": 1.2258214738878392e-07, "loss": 0.2497, "step": 26949 }, { "epoch": 1.8631178707224336, "grad_norm": 1.2780996812972691, "learning_rate": 1.2245899047190756e-07, "loss": 0.2119, "step": 26950 }, { "epoch": 1.8631870031109576, "grad_norm": 0.8364013339214447, "learning_rate": 1.2233589468630347e-07, "loss": 0.2685, "step": 26951 }, { "epoch": 1.8632561354994817, "grad_norm": 0.7039586155250014, "learning_rate": 1.222128600335143e-07, "loss": 0.3287, "step": 26952 }, { "epoch": 1.8633252678880057, "grad_norm": 0.7279079888298687, "learning_rate": 1.2208988651508212e-07, "loss": 0.2551, "step": 26953 }, { "epoch": 1.8633944002765297, "grad_norm": 0.664894935584019, "learning_rate": 1.2196697413254799e-07, "loss": 0.2306, "step": 26954 }, { "epoch": 1.8634635326650537, "grad_norm": 0.6976780927406826, "learning_rate": 1.2184412288745228e-07, "loss": 0.3133, "step": 26955 }, { "epoch": 1.8635326650535777, "grad_norm": 0.7297738510841831, "learning_rate": 1.21721332781336e-07, "loss": 0.2267, "step": 26956 }, { "epoch": 1.8636017974421017, "grad_norm": 0.7924104002008168, "learning_rate": 1.215986038157363e-07, "loss": 0.2634, "step": 26957 }, { "epoch": 1.8636709298306258, "grad_norm": 0.7038650228184881, "learning_rate": 1.2147593599219298e-07, "loss": 0.3, "step": 26958 }, { "epoch": 1.8637400622191498, "grad_norm": 0.6767084099848337, "learning_rate": 1.213533293122421e-07, "loss": 0.3446, "step": 26959 }, { "epoch": 1.8638091946076738, "grad_norm": 0.7591625032810476, "learning_rate": 1.2123078377742125e-07, "loss": 0.2263, "step": 26960 }, { "epoch": 1.8638783269961978, "grad_norm": 0.711875929877323, "learning_rate": 1.2110829938926537e-07, "loss": 0.243, "step": 26961 }, { "epoch": 1.8639474593847218, "grad_norm": 0.7574139514824024, "learning_rate": 1.2098587614931046e-07, "loss": 0.3525, "step": 26962 }, { "epoch": 1.8640165917732459, "grad_norm": 0.6991467084751566, "learning_rate": 1.208635140590908e-07, "loss": 0.2707, "step": 26963 }, { "epoch": 1.8640857241617699, "grad_norm": 0.7629807450924556, "learning_rate": 1.2074121312013965e-07, "loss": 0.2252, "step": 26964 }, { "epoch": 1.864154856550294, "grad_norm": 0.7343347369676895, "learning_rate": 1.2061897333399018e-07, "loss": 0.2699, "step": 26965 }, { "epoch": 1.864223988938818, "grad_norm": 0.688299934797349, "learning_rate": 1.2049679470217403e-07, "loss": 0.2864, "step": 26966 }, { "epoch": 1.864293121327342, "grad_norm": 0.7533680059725609, "learning_rate": 1.2037467722622266e-07, "loss": 0.2389, "step": 26967 }, { "epoch": 1.864362253715866, "grad_norm": 0.6486480114446211, "learning_rate": 1.202526209076671e-07, "loss": 0.1778, "step": 26968 }, { "epoch": 1.86443138610439, "grad_norm": 0.7481359886242, "learning_rate": 1.2013062574803613e-07, "loss": 0.3849, "step": 26969 }, { "epoch": 1.864500518492914, "grad_norm": 0.7601642964970425, "learning_rate": 1.200086917488591e-07, "loss": 0.2947, "step": 26970 }, { "epoch": 1.864569650881438, "grad_norm": 0.6596465389334006, "learning_rate": 1.198868189116642e-07, "loss": 0.2138, "step": 26971 }, { "epoch": 1.864638783269962, "grad_norm": 0.6795744559785422, "learning_rate": 1.1976500723798025e-07, "loss": 0.2137, "step": 26972 }, { "epoch": 1.864707915658486, "grad_norm": 0.6905742077459657, "learning_rate": 1.1964325672933154e-07, "loss": 0.3281, "step": 26973 }, { "epoch": 1.86477704804701, "grad_norm": 0.9146890357707023, "learning_rate": 1.195215673872463e-07, "loss": 0.242, "step": 26974 }, { "epoch": 1.864846180435534, "grad_norm": 0.8280982810722315, "learning_rate": 1.1939993921324778e-07, "loss": 0.2417, "step": 26975 }, { "epoch": 1.864915312824058, "grad_norm": 0.7017843786110601, "learning_rate": 1.1927837220886085e-07, "loss": 0.2402, "step": 26976 }, { "epoch": 1.864984445212582, "grad_norm": 0.6760836985695632, "learning_rate": 1.1915686637560985e-07, "loss": 0.2941, "step": 26977 }, { "epoch": 1.8650535776011061, "grad_norm": 0.7619634491059497, "learning_rate": 1.1903542171501803e-07, "loss": 0.2688, "step": 26978 }, { "epoch": 1.8651227099896301, "grad_norm": 0.649785111234889, "learning_rate": 1.1891403822860581e-07, "loss": 0.2291, "step": 26979 }, { "epoch": 1.8651918423781542, "grad_norm": 0.6905990582047196, "learning_rate": 1.1879271591789643e-07, "loss": 0.3757, "step": 26980 }, { "epoch": 1.8652609747666782, "grad_norm": 0.6753040768849178, "learning_rate": 1.1867145478440867e-07, "loss": 0.2865, "step": 26981 }, { "epoch": 1.8653301071552022, "grad_norm": 0.6210721303560015, "learning_rate": 1.1855025482966354e-07, "loss": 0.2145, "step": 26982 }, { "epoch": 1.8653992395437262, "grad_norm": 0.6525251991039899, "learning_rate": 1.1842911605517927e-07, "loss": 0.2247, "step": 26983 }, { "epoch": 1.8654683719322502, "grad_norm": 0.7049734683324207, "learning_rate": 1.1830803846247407e-07, "loss": 0.2877, "step": 26984 }, { "epoch": 1.8655375043207743, "grad_norm": 0.7408092969002704, "learning_rate": 1.181870220530662e-07, "loss": 0.2096, "step": 26985 }, { "epoch": 1.8656066367092983, "grad_norm": 0.7492897902199039, "learning_rate": 1.1806606682847221e-07, "loss": 0.2113, "step": 26986 }, { "epoch": 1.8656757690978223, "grad_norm": 0.7230589352444539, "learning_rate": 1.1794517279020812e-07, "loss": 0.3358, "step": 26987 }, { "epoch": 1.8657449014863463, "grad_norm": 0.8414929532264369, "learning_rate": 1.1782433993978881e-07, "loss": 0.2343, "step": 26988 }, { "epoch": 1.8658140338748703, "grad_norm": 0.6793615555851507, "learning_rate": 1.1770356827872809e-07, "loss": 0.2197, "step": 26989 }, { "epoch": 1.8658831662633943, "grad_norm": 0.7462203363223582, "learning_rate": 1.1758285780854029e-07, "loss": 0.2646, "step": 26990 }, { "epoch": 1.8659522986519184, "grad_norm": 0.6927057696974372, "learning_rate": 1.1746220853073864e-07, "loss": 0.3252, "step": 26991 }, { "epoch": 1.8660214310404424, "grad_norm": 0.7308026381925387, "learning_rate": 1.1734162044683528e-07, "loss": 0.2573, "step": 26992 }, { "epoch": 1.8660905634289664, "grad_norm": 0.6304943622899805, "learning_rate": 1.1722109355834066e-07, "loss": 0.1806, "step": 26993 }, { "epoch": 1.8661596958174904, "grad_norm": 0.6604127865578503, "learning_rate": 1.171006278667669e-07, "loss": 0.2717, "step": 26994 }, { "epoch": 1.8662288282060144, "grad_norm": 0.7356199183371829, "learning_rate": 1.1698022337362169e-07, "loss": 0.3151, "step": 26995 }, { "epoch": 1.8662979605945385, "grad_norm": 0.5967553921253209, "learning_rate": 1.1685988008041604e-07, "loss": 0.218, "step": 26996 }, { "epoch": 1.8663670929830625, "grad_norm": 0.7635434904626962, "learning_rate": 1.1673959798865709e-07, "loss": 0.2846, "step": 26997 }, { "epoch": 1.8664362253715865, "grad_norm": 0.6826903653464301, "learning_rate": 1.1661937709985249e-07, "loss": 0.342, "step": 26998 }, { "epoch": 1.8665053577601105, "grad_norm": 0.7617165789101793, "learning_rate": 1.164992174155094e-07, "loss": 0.3004, "step": 26999 }, { "epoch": 1.8665744901486345, "grad_norm": 0.649107017042889, "learning_rate": 1.1637911893713383e-07, "loss": 0.2629, "step": 27000 }, { "epoch": 1.8666436225371585, "grad_norm": 0.7256609524013782, "learning_rate": 1.1625908166623012e-07, "loss": 0.2324, "step": 27001 }, { "epoch": 1.8667127549256826, "grad_norm": 0.7401943101797291, "learning_rate": 1.161391056043043e-07, "loss": 0.3154, "step": 27002 }, { "epoch": 1.8667818873142066, "grad_norm": 1.3563171155306972, "learning_rate": 1.1601919075285906e-07, "loss": 0.1859, "step": 27003 }, { "epoch": 1.8668510197027306, "grad_norm": 0.6240562032183968, "learning_rate": 1.1589933711339707e-07, "loss": 0.2293, "step": 27004 }, { "epoch": 1.8669201520912546, "grad_norm": 0.6879424824687734, "learning_rate": 1.1577954468742048e-07, "loss": 0.3327, "step": 27005 }, { "epoch": 1.8669892844797786, "grad_norm": 0.8037379631960704, "learning_rate": 1.156598134764314e-07, "loss": 0.2843, "step": 27006 }, { "epoch": 1.8670584168683027, "grad_norm": 0.6027029318908731, "learning_rate": 1.1554014348193032e-07, "loss": 0.2137, "step": 27007 }, { "epoch": 1.8671275492568267, "grad_norm": 1.0650255095276295, "learning_rate": 1.1542053470541658e-07, "loss": 0.2626, "step": 27008 }, { "epoch": 1.8671966816453507, "grad_norm": 0.7032058282207918, "learning_rate": 1.1530098714838956e-07, "loss": 0.3429, "step": 27009 }, { "epoch": 1.8672658140338747, "grad_norm": 0.7903373773715971, "learning_rate": 1.1518150081234747e-07, "loss": 0.2453, "step": 27010 }, { "epoch": 1.867334946422399, "grad_norm": 0.7052280065023976, "learning_rate": 1.1506207569878747e-07, "loss": 0.2611, "step": 27011 }, { "epoch": 1.867404078810923, "grad_norm": 0.7347613617843948, "learning_rate": 1.1494271180920724e-07, "loss": 0.3615, "step": 27012 }, { "epoch": 1.867473211199447, "grad_norm": 0.7793118368800088, "learning_rate": 1.1482340914510226e-07, "loss": 0.265, "step": 27013 }, { "epoch": 1.867542343587971, "grad_norm": 0.6686451770490308, "learning_rate": 1.14704167707968e-07, "loss": 0.2336, "step": 27014 }, { "epoch": 1.867611475976495, "grad_norm": 0.9221808803371191, "learning_rate": 1.145849874992988e-07, "loss": 0.2471, "step": 27015 }, { "epoch": 1.867680608365019, "grad_norm": 0.784155997794902, "learning_rate": 1.1446586852058794e-07, "loss": 0.3726, "step": 27016 }, { "epoch": 1.867749740753543, "grad_norm": 0.7050851743940074, "learning_rate": 1.1434681077332976e-07, "loss": 0.2587, "step": 27017 }, { "epoch": 1.867818873142067, "grad_norm": 0.6635102083338471, "learning_rate": 1.1422781425901474e-07, "loss": 0.2082, "step": 27018 }, { "epoch": 1.867888005530591, "grad_norm": 0.6984795145821255, "learning_rate": 1.1410887897913503e-07, "loss": 0.3491, "step": 27019 }, { "epoch": 1.867957137919115, "grad_norm": 0.7626590512970244, "learning_rate": 1.1399000493518109e-07, "loss": 0.2912, "step": 27020 }, { "epoch": 1.8680262703076391, "grad_norm": 0.6610208070603207, "learning_rate": 1.1387119212864339e-07, "loss": 0.275, "step": 27021 }, { "epoch": 1.8680954026961631, "grad_norm": 0.8820369640923881, "learning_rate": 1.137524405610102e-07, "loss": 0.2186, "step": 27022 }, { "epoch": 1.8681645350846872, "grad_norm": 0.7381737744693166, "learning_rate": 1.1363375023377088e-07, "loss": 0.3186, "step": 27023 }, { "epoch": 1.8682336674732112, "grad_norm": 0.7255271221436028, "learning_rate": 1.1351512114841201e-07, "loss": 0.3222, "step": 27024 }, { "epoch": 1.8683027998617352, "grad_norm": 0.6741949832737844, "learning_rate": 1.1339655330642019e-07, "loss": 0.2074, "step": 27025 }, { "epoch": 1.8683719322502592, "grad_norm": 0.9137882940570432, "learning_rate": 1.1327804670928256e-07, "loss": 0.2869, "step": 27026 }, { "epoch": 1.8684410646387832, "grad_norm": 0.710268147830463, "learning_rate": 1.1315960135848403e-07, "loss": 0.3043, "step": 27027 }, { "epoch": 1.8685101970273073, "grad_norm": 0.6813753006277001, "learning_rate": 1.1304121725550899e-07, "loss": 0.2218, "step": 27028 }, { "epoch": 1.8685793294158313, "grad_norm": 0.6327423839527685, "learning_rate": 1.1292289440184123e-07, "loss": 0.224, "step": 27029 }, { "epoch": 1.8686484618043553, "grad_norm": 0.6626508683452224, "learning_rate": 1.1280463279896293e-07, "loss": 0.3205, "step": 27030 }, { "epoch": 1.8687175941928795, "grad_norm": 0.6945058036587959, "learning_rate": 1.1268643244835786e-07, "loss": 0.2556, "step": 27031 }, { "epoch": 1.8687867265814035, "grad_norm": 0.6442292101813086, "learning_rate": 1.1256829335150599e-07, "loss": 0.1812, "step": 27032 }, { "epoch": 1.8688558589699276, "grad_norm": 0.7023972307086134, "learning_rate": 1.1245021550988833e-07, "loss": 0.2632, "step": 27033 }, { "epoch": 1.8689249913584516, "grad_norm": 0.7546041198414819, "learning_rate": 1.1233219892498537e-07, "loss": 0.3925, "step": 27034 }, { "epoch": 1.8689941237469756, "grad_norm": 0.6976881480645284, "learning_rate": 1.1221424359827592e-07, "loss": 0.2282, "step": 27035 }, { "epoch": 1.8690632561354996, "grad_norm": 0.641120137757045, "learning_rate": 1.1209634953123771e-07, "loss": 0.2351, "step": 27036 }, { "epoch": 1.8691323885240236, "grad_norm": 0.7206409825062002, "learning_rate": 1.1197851672534954e-07, "loss": 0.3325, "step": 27037 }, { "epoch": 1.8692015209125477, "grad_norm": 1.0702814593605816, "learning_rate": 1.1186074518208745e-07, "loss": 0.2356, "step": 27038 }, { "epoch": 1.8692706533010717, "grad_norm": 0.6979600927735713, "learning_rate": 1.1174303490292804e-07, "loss": 0.2919, "step": 27039 }, { "epoch": 1.8693397856895957, "grad_norm": 0.8202335931122486, "learning_rate": 1.1162538588934569e-07, "loss": 0.2626, "step": 27040 }, { "epoch": 1.8694089180781197, "grad_norm": 0.7130306415971035, "learning_rate": 1.1150779814281532e-07, "loss": 0.4048, "step": 27041 }, { "epoch": 1.8694780504666437, "grad_norm": 0.7022502174471933, "learning_rate": 1.1139027166481076e-07, "loss": 0.2304, "step": 27042 }, { "epoch": 1.8695471828551677, "grad_norm": 0.6791640208871659, "learning_rate": 1.1127280645680527e-07, "loss": 0.2638, "step": 27043 }, { "epoch": 1.8696163152436918, "grad_norm": 0.7254665721930172, "learning_rate": 1.1115540252027046e-07, "loss": 0.3341, "step": 27044 }, { "epoch": 1.8696854476322158, "grad_norm": 0.8185620925078811, "learning_rate": 1.1103805985667903e-07, "loss": 0.2544, "step": 27045 }, { "epoch": 1.8697545800207398, "grad_norm": 1.040809983889439, "learning_rate": 1.1092077846749982e-07, "loss": 0.2661, "step": 27046 }, { "epoch": 1.8698237124092638, "grad_norm": 0.6886224215138504, "learning_rate": 1.1080355835420386e-07, "loss": 0.25, "step": 27047 }, { "epoch": 1.8698928447977878, "grad_norm": 0.6761483586552163, "learning_rate": 1.1068639951826054e-07, "loss": 0.2929, "step": 27048 }, { "epoch": 1.8699619771863119, "grad_norm": 0.7327759777988924, "learning_rate": 1.1056930196113813e-07, "loss": 0.3202, "step": 27049 }, { "epoch": 1.8700311095748359, "grad_norm": 0.6058909760438004, "learning_rate": 1.1045226568430322e-07, "loss": 0.1765, "step": 27050 }, { "epoch": 1.8701002419633599, "grad_norm": 0.649720422103707, "learning_rate": 1.1033529068922356e-07, "loss": 0.2444, "step": 27051 }, { "epoch": 1.870169374351884, "grad_norm": 0.6818299498032231, "learning_rate": 1.1021837697736515e-07, "loss": 0.3514, "step": 27052 }, { "epoch": 1.870238506740408, "grad_norm": 0.7114965711342346, "learning_rate": 1.1010152455019352e-07, "loss": 0.1948, "step": 27053 }, { "epoch": 1.870307639128932, "grad_norm": 0.6848551684163038, "learning_rate": 1.0998473340917193e-07, "loss": 0.2916, "step": 27054 }, { "epoch": 1.870376771517456, "grad_norm": 0.6927685738081526, "learning_rate": 1.0986800355576588e-07, "loss": 0.3458, "step": 27055 }, { "epoch": 1.87044590390598, "grad_norm": 0.7998971610351926, "learning_rate": 1.0975133499143697e-07, "loss": 0.254, "step": 27056 }, { "epoch": 1.870515036294504, "grad_norm": 0.7151281751137879, "learning_rate": 1.096347277176485e-07, "loss": 0.232, "step": 27057 }, { "epoch": 1.870584168683028, "grad_norm": 0.6918985492107841, "learning_rate": 1.0951818173586148e-07, "loss": 0.2354, "step": 27058 }, { "epoch": 1.870653301071552, "grad_norm": 0.942400889862168, "learning_rate": 1.0940169704753644e-07, "loss": 0.3659, "step": 27059 }, { "epoch": 1.870722433460076, "grad_norm": 0.6537798067724406, "learning_rate": 1.0928527365413333e-07, "loss": 0.2149, "step": 27060 }, { "epoch": 1.8707915658486, "grad_norm": 0.6752012846224174, "learning_rate": 1.0916891155711151e-07, "loss": 0.2318, "step": 27061 }, { "epoch": 1.870860698237124, "grad_norm": 0.919115748131483, "learning_rate": 1.0905261075792928e-07, "loss": 0.3002, "step": 27062 }, { "epoch": 1.870929830625648, "grad_norm": 0.7547422983739385, "learning_rate": 1.0893637125804491e-07, "loss": 0.2698, "step": 27063 }, { "epoch": 1.8709989630141721, "grad_norm": 0.6441787900912311, "learning_rate": 1.088201930589139e-07, "loss": 0.2217, "step": 27064 }, { "epoch": 1.8710680954026961, "grad_norm": 0.6743307925086248, "learning_rate": 1.0870407616199341e-07, "loss": 0.2395, "step": 27065 }, { "epoch": 1.8711372277912202, "grad_norm": 0.6897405215258077, "learning_rate": 1.0858802056873785e-07, "loss": 0.3654, "step": 27066 }, { "epoch": 1.8712063601797442, "grad_norm": 0.6630179944651934, "learning_rate": 1.0847202628060327e-07, "loss": 0.2667, "step": 27067 }, { "epoch": 1.8712754925682682, "grad_norm": 0.6104438548588085, "learning_rate": 1.0835609329904185e-07, "loss": 0.19, "step": 27068 }, { "epoch": 1.8713446249567922, "grad_norm": 0.6443884234908954, "learning_rate": 1.082402216255074e-07, "loss": 0.2881, "step": 27069 }, { "epoch": 1.8714137573453162, "grad_norm": 0.7460796440590011, "learning_rate": 1.0812441126145212e-07, "loss": 0.3172, "step": 27070 }, { "epoch": 1.8714828897338402, "grad_norm": 0.7637108474576193, "learning_rate": 1.0800866220832707e-07, "loss": 0.2235, "step": 27071 }, { "epoch": 1.8715520221223643, "grad_norm": 0.8215330955022089, "learning_rate": 1.078929744675844e-07, "loss": 0.2728, "step": 27072 }, { "epoch": 1.8716211545108883, "grad_norm": 0.7802238961155677, "learning_rate": 1.0777734804067185e-07, "loss": 0.3478, "step": 27073 }, { "epoch": 1.8716902868994123, "grad_norm": 2.7696376001594816, "learning_rate": 1.0766178292903994e-07, "loss": 0.2381, "step": 27074 }, { "epoch": 1.8717594192879363, "grad_norm": 0.7330168131159331, "learning_rate": 1.0754627913413695e-07, "loss": 0.2607, "step": 27075 }, { "epoch": 1.8718285516764603, "grad_norm": 0.6858127980039542, "learning_rate": 1.0743083665741006e-07, "loss": 0.2952, "step": 27076 }, { "epoch": 1.8718976840649844, "grad_norm": 0.7166225956786703, "learning_rate": 1.0731545550030642e-07, "loss": 0.2878, "step": 27077 }, { "epoch": 1.8719668164535084, "grad_norm": 0.8001796089269262, "learning_rate": 1.0720013566427268e-07, "loss": 0.2801, "step": 27078 }, { "epoch": 1.8720359488420324, "grad_norm": 0.7266709319748996, "learning_rate": 1.0708487715075322e-07, "loss": 0.2418, "step": 27079 }, { "epoch": 1.8721050812305564, "grad_norm": 0.6906862067359933, "learning_rate": 1.06969679961193e-07, "loss": 0.3767, "step": 27080 }, { "epoch": 1.8721742136190804, "grad_norm": 0.6664922960392965, "learning_rate": 1.0685454409703588e-07, "loss": 0.2274, "step": 27081 }, { "epoch": 1.8722433460076044, "grad_norm": 0.6506915731359246, "learning_rate": 1.0673946955972514e-07, "loss": 0.2268, "step": 27082 }, { "epoch": 1.8723124783961285, "grad_norm": 0.7243827069649132, "learning_rate": 1.066244563507024e-07, "loss": 0.2269, "step": 27083 }, { "epoch": 1.8723816107846525, "grad_norm": 0.9279529235584102, "learning_rate": 1.0650950447140984e-07, "loss": 0.3075, "step": 27084 }, { "epoch": 1.8724507431731765, "grad_norm": 0.662926032377069, "learning_rate": 1.0639461392328743e-07, "loss": 0.2049, "step": 27085 }, { "epoch": 1.8725198755617005, "grad_norm": 1.0440088417149163, "learning_rate": 1.0627978470777567e-07, "loss": 0.2474, "step": 27086 }, { "epoch": 1.8725890079502245, "grad_norm": 0.662701640398429, "learning_rate": 1.0616501682631341e-07, "loss": 0.3053, "step": 27087 }, { "epoch": 1.8726581403387486, "grad_norm": 0.6732116194456547, "learning_rate": 1.0605031028033952e-07, "loss": 0.2333, "step": 27088 }, { "epoch": 1.8727272727272726, "grad_norm": 0.7224342947958833, "learning_rate": 1.0593566507129116e-07, "loss": 0.2846, "step": 27089 }, { "epoch": 1.8727964051157966, "grad_norm": 0.7472008652289535, "learning_rate": 1.0582108120060552e-07, "loss": 0.2324, "step": 27090 }, { "epoch": 1.8728655375043206, "grad_norm": 0.7622131861883241, "learning_rate": 1.0570655866971813e-07, "loss": 0.3418, "step": 27091 }, { "epoch": 1.8729346698928449, "grad_norm": 0.7014246392447867, "learning_rate": 1.055920974800656e-07, "loss": 0.2606, "step": 27092 }, { "epoch": 1.8730038022813689, "grad_norm": 0.6673119450604561, "learning_rate": 1.0547769763308068e-07, "loss": 0.232, "step": 27093 }, { "epoch": 1.8730729346698929, "grad_norm": 0.9074898729055948, "learning_rate": 1.0536335913019891e-07, "loss": 0.3697, "step": 27094 }, { "epoch": 1.873142067058417, "grad_norm": 0.7350212474956748, "learning_rate": 1.0524908197285245e-07, "loss": 0.2985, "step": 27095 }, { "epoch": 1.873211199446941, "grad_norm": 0.8657354797292066, "learning_rate": 1.0513486616247292e-07, "loss": 0.2354, "step": 27096 }, { "epoch": 1.873280331835465, "grad_norm": 0.6524193337484939, "learning_rate": 1.050207117004931e-07, "loss": 0.2355, "step": 27097 }, { "epoch": 1.873349464223989, "grad_norm": 0.6737041444149896, "learning_rate": 1.049066185883435e-07, "loss": 0.351, "step": 27098 }, { "epoch": 1.873418596612513, "grad_norm": 0.7666061494672952, "learning_rate": 1.0479258682745353e-07, "loss": 0.2789, "step": 27099 }, { "epoch": 1.873487729001037, "grad_norm": 0.7000374915173948, "learning_rate": 1.0467861641925259e-07, "loss": 0.2239, "step": 27100 }, { "epoch": 1.873556861389561, "grad_norm": 0.7187526454865626, "learning_rate": 1.04564707365169e-07, "loss": 0.2911, "step": 27101 }, { "epoch": 1.873625993778085, "grad_norm": 0.680461778054412, "learning_rate": 1.0445085966663104e-07, "loss": 0.2676, "step": 27102 }, { "epoch": 1.873695126166609, "grad_norm": 0.6606881674018956, "learning_rate": 1.0433707332506426e-07, "loss": 0.236, "step": 27103 }, { "epoch": 1.873764258555133, "grad_norm": 0.6937953704497742, "learning_rate": 1.0422334834189585e-07, "loss": 0.2197, "step": 27104 }, { "epoch": 1.873833390943657, "grad_norm": 0.7538707932451562, "learning_rate": 1.0410968471855076e-07, "loss": 0.3264, "step": 27105 }, { "epoch": 1.873902523332181, "grad_norm": 0.7035865335271249, "learning_rate": 1.0399608245645398e-07, "loss": 0.2484, "step": 27106 }, { "epoch": 1.8739716557207051, "grad_norm": 0.759891939630217, "learning_rate": 1.0388254155702882e-07, "loss": 0.2459, "step": 27107 }, { "epoch": 1.8740407881092291, "grad_norm": 0.9833311447461667, "learning_rate": 1.0376906202169857e-07, "loss": 0.2354, "step": 27108 }, { "epoch": 1.8741099204977532, "grad_norm": 0.709133461927805, "learning_rate": 1.0365564385188543e-07, "loss": 0.3518, "step": 27109 }, { "epoch": 1.8741790528862772, "grad_norm": 0.7527467668953292, "learning_rate": 1.0354228704901103e-07, "loss": 0.2918, "step": 27110 }, { "epoch": 1.8742481852748012, "grad_norm": 0.9208973464101977, "learning_rate": 1.0342899161449538e-07, "loss": 0.2333, "step": 27111 }, { "epoch": 1.8743173176633254, "grad_norm": 0.7418620648652079, "learning_rate": 1.0331575754975954e-07, "loss": 0.3896, "step": 27112 }, { "epoch": 1.8743864500518495, "grad_norm": 0.7165198534572895, "learning_rate": 1.0320258485622126e-07, "loss": 0.3189, "step": 27113 }, { "epoch": 1.8744555824403735, "grad_norm": 0.7125560541697552, "learning_rate": 1.0308947353530108e-07, "loss": 0.1994, "step": 27114 }, { "epoch": 1.8745247148288975, "grad_norm": 0.765919200726967, "learning_rate": 1.0297642358841453e-07, "loss": 0.2694, "step": 27115 }, { "epoch": 1.8745938472174215, "grad_norm": 0.7108268070558917, "learning_rate": 1.0286343501697993e-07, "loss": 0.2869, "step": 27116 }, { "epoch": 1.8746629796059455, "grad_norm": 0.8219336202292281, "learning_rate": 1.027505078224128e-07, "loss": 0.2449, "step": 27117 }, { "epoch": 1.8747321119944695, "grad_norm": 0.8707090711056567, "learning_rate": 1.0263764200612813e-07, "loss": 0.2316, "step": 27118 }, { "epoch": 1.8748012443829936, "grad_norm": 0.703736214166991, "learning_rate": 1.0252483756954091e-07, "loss": 0.3762, "step": 27119 }, { "epoch": 1.8748703767715176, "grad_norm": 0.7488032940899327, "learning_rate": 1.0241209451406497e-07, "loss": 0.3258, "step": 27120 }, { "epoch": 1.8749395091600416, "grad_norm": 0.7868184860023419, "learning_rate": 1.0229941284111366e-07, "loss": 0.2458, "step": 27121 }, { "epoch": 1.8750086415485656, "grad_norm": 0.6950443421819191, "learning_rate": 1.0218679255209863e-07, "loss": 0.2607, "step": 27122 }, { "epoch": 1.8750777739370896, "grad_norm": 0.6704768513671767, "learning_rate": 1.020742336484315e-07, "loss": 0.325, "step": 27123 }, { "epoch": 1.8751469063256136, "grad_norm": 0.742677207730967, "learning_rate": 1.0196173613152338e-07, "loss": 0.243, "step": 27124 }, { "epoch": 1.8752160387141377, "grad_norm": 0.6360445217890642, "learning_rate": 1.018493000027837e-07, "loss": 0.2147, "step": 27125 }, { "epoch": 1.8752851711026617, "grad_norm": 0.7251747740947606, "learning_rate": 1.0173692526362189e-07, "loss": 0.3294, "step": 27126 }, { "epoch": 1.8753543034911857, "grad_norm": 0.7010616306644899, "learning_rate": 1.0162461191544625e-07, "loss": 0.3782, "step": 27127 }, { "epoch": 1.8754234358797097, "grad_norm": 0.7106869567315818, "learning_rate": 1.0151235995966514e-07, "loss": 0.2221, "step": 27128 }, { "epoch": 1.8754925682682337, "grad_norm": 0.60522258548918, "learning_rate": 1.014001693976846e-07, "loss": 0.2168, "step": 27129 }, { "epoch": 1.8755617006567578, "grad_norm": 0.746592485069658, "learning_rate": 1.0128804023091076e-07, "loss": 0.3165, "step": 27130 }, { "epoch": 1.8756308330452818, "grad_norm": 0.7484501625816736, "learning_rate": 1.0117597246074917e-07, "loss": 0.248, "step": 27131 }, { "epoch": 1.8756999654338058, "grad_norm": 0.6623244079932692, "learning_rate": 1.010639660886048e-07, "loss": 0.2583, "step": 27132 }, { "epoch": 1.8757690978223298, "grad_norm": 0.7051934349221587, "learning_rate": 1.0095202111588043e-07, "loss": 0.336, "step": 27133 }, { "epoch": 1.8758382302108538, "grad_norm": 1.926156243799021, "learning_rate": 1.0084013754398048e-07, "loss": 0.3762, "step": 27134 }, { "epoch": 1.8759073625993778, "grad_norm": 0.7448945639393273, "learning_rate": 1.0072831537430605e-07, "loss": 0.2491, "step": 27135 }, { "epoch": 1.8759764949879019, "grad_norm": 0.7407275145082773, "learning_rate": 1.0061655460825937e-07, "loss": 0.2757, "step": 27136 }, { "epoch": 1.8760456273764259, "grad_norm": 0.723807299469276, "learning_rate": 1.0050485524724097e-07, "loss": 0.3502, "step": 27137 }, { "epoch": 1.87611475976495, "grad_norm": 0.7257955290228049, "learning_rate": 1.0039321729264972e-07, "loss": 0.2676, "step": 27138 }, { "epoch": 1.876183892153474, "grad_norm": 0.678712320950413, "learning_rate": 1.002816407458862e-07, "loss": 0.2038, "step": 27139 }, { "epoch": 1.876253024541998, "grad_norm": 0.7488165873060296, "learning_rate": 1.0017012560834872e-07, "loss": 0.2644, "step": 27140 }, { "epoch": 1.876322156930522, "grad_norm": 0.6962754768020505, "learning_rate": 1.0005867188143448e-07, "loss": 0.3538, "step": 27141 }, { "epoch": 1.876391289319046, "grad_norm": 0.6971398628726212, "learning_rate": 9.994727956654015e-08, "loss": 0.2588, "step": 27142 }, { "epoch": 1.87646042170757, "grad_norm": 1.3449416062472694, "learning_rate": 9.983594866506242e-08, "loss": 0.1956, "step": 27143 }, { "epoch": 1.876529554096094, "grad_norm": 0.7460785095926694, "learning_rate": 9.972467917839623e-08, "loss": 0.3205, "step": 27144 }, { "epoch": 1.876598686484618, "grad_norm": 0.7191465331239285, "learning_rate": 9.961347110793662e-08, "loss": 0.3438, "step": 27145 }, { "epoch": 1.876667818873142, "grad_norm": 0.861663884721256, "learning_rate": 9.950232445507635e-08, "loss": 0.2116, "step": 27146 }, { "epoch": 1.876736951261666, "grad_norm": 0.8591814539843253, "learning_rate": 9.93912392212093e-08, "loss": 0.2417, "step": 27147 }, { "epoch": 1.87680608365019, "grad_norm": 0.8284355998257854, "learning_rate": 9.928021540772714e-08, "loss": 0.3253, "step": 27148 }, { "epoch": 1.876875216038714, "grad_norm": 0.9745875032626569, "learning_rate": 9.916925301602264e-08, "loss": 0.2341, "step": 27149 }, { "epoch": 1.8769443484272381, "grad_norm": 0.653601472269338, "learning_rate": 9.905835204748527e-08, "loss": 0.1964, "step": 27150 }, { "epoch": 1.8770134808157621, "grad_norm": 0.6800691380240517, "learning_rate": 9.894751250350554e-08, "loss": 0.3045, "step": 27151 }, { "epoch": 1.8770826132042862, "grad_norm": 0.8261318912267109, "learning_rate": 9.88367343854718e-08, "loss": 0.3521, "step": 27152 }, { "epoch": 1.8771517455928102, "grad_norm": 0.6431912654399017, "learning_rate": 9.872601769477352e-08, "loss": 0.2326, "step": 27153 }, { "epoch": 1.8772208779813342, "grad_norm": 0.7035374061324157, "learning_rate": 9.861536243279735e-08, "loss": 0.2559, "step": 27154 }, { "epoch": 1.8772900103698582, "grad_norm": 0.6474790640516096, "learning_rate": 9.850476860093161e-08, "loss": 0.2965, "step": 27155 }, { "epoch": 1.8773591427583822, "grad_norm": 0.6807172334613213, "learning_rate": 9.839423620056076e-08, "loss": 0.2586, "step": 27156 }, { "epoch": 1.8774282751469062, "grad_norm": 0.6793833989873787, "learning_rate": 9.828376523307148e-08, "loss": 0.2431, "step": 27157 }, { "epoch": 1.8774974075354303, "grad_norm": 0.7523889182686226, "learning_rate": 9.81733556998471e-08, "loss": 0.2816, "step": 27158 }, { "epoch": 1.8775665399239543, "grad_norm": 0.7487606309202643, "learning_rate": 9.806300760227261e-08, "loss": 0.2679, "step": 27159 }, { "epoch": 1.8776356723124783, "grad_norm": 0.6814667810945046, "learning_rate": 9.795272094173025e-08, "loss": 0.2736, "step": 27160 }, { "epoch": 1.8777048047010023, "grad_norm": 0.6680488594423387, "learning_rate": 9.784249571960169e-08, "loss": 0.2279, "step": 27161 }, { "epoch": 1.8777739370895263, "grad_norm": 0.7369437432580671, "learning_rate": 9.773233193726972e-08, "loss": 0.3308, "step": 27162 }, { "epoch": 1.8778430694780504, "grad_norm": 0.6969318355427424, "learning_rate": 9.762222959611489e-08, "loss": 0.264, "step": 27163 }, { "epoch": 1.8779122018665744, "grad_norm": 0.814666887622565, "learning_rate": 9.751218869751611e-08, "loss": 0.2698, "step": 27164 }, { "epoch": 1.8779813342550984, "grad_norm": 0.7193044000461857, "learning_rate": 9.740220924285338e-08, "loss": 0.2738, "step": 27165 }, { "epoch": 1.8780504666436224, "grad_norm": 0.7337455305963249, "learning_rate": 9.729229123350503e-08, "loss": 0.3669, "step": 27166 }, { "epoch": 1.8781195990321464, "grad_norm": 0.7035102395891677, "learning_rate": 9.718243467084831e-08, "loss": 0.2909, "step": 27167 }, { "epoch": 1.8781887314206704, "grad_norm": 0.7615916457117261, "learning_rate": 9.707263955625989e-08, "loss": 0.2162, "step": 27168 }, { "epoch": 1.8782578638091945, "grad_norm": 0.7366729135640306, "learning_rate": 9.696290589111701e-08, "loss": 0.3063, "step": 27169 }, { "epoch": 1.8783269961977185, "grad_norm": 0.7085242023263929, "learning_rate": 9.685323367679411e-08, "loss": 0.2989, "step": 27170 }, { "epoch": 1.8783961285862425, "grad_norm": 0.7715345063646785, "learning_rate": 9.674362291466566e-08, "loss": 0.2219, "step": 27171 }, { "epoch": 1.8784652609747665, "grad_norm": 0.7119742166164106, "learning_rate": 9.663407360610555e-08, "loss": 0.3044, "step": 27172 }, { "epoch": 1.8785343933632908, "grad_norm": 0.7426284427400532, "learning_rate": 9.652458575248769e-08, "loss": 0.3641, "step": 27173 }, { "epoch": 1.8786035257518148, "grad_norm": 1.4720977371173194, "learning_rate": 9.641515935518265e-08, "loss": 0.2509, "step": 27174 }, { "epoch": 1.8786726581403388, "grad_norm": 0.6623994706933943, "learning_rate": 9.630579441556265e-08, "loss": 0.2089, "step": 27175 }, { "epoch": 1.8787417905288628, "grad_norm": 0.6572958887714598, "learning_rate": 9.619649093499883e-08, "loss": 0.2595, "step": 27176 }, { "epoch": 1.8788109229173868, "grad_norm": 0.7352669852295799, "learning_rate": 9.608724891486065e-08, "loss": 0.3135, "step": 27177 }, { "epoch": 1.8788800553059108, "grad_norm": 0.7068240086831985, "learning_rate": 9.597806835651757e-08, "loss": 0.1847, "step": 27178 }, { "epoch": 1.8789491876944349, "grad_norm": 0.9606830624989938, "learning_rate": 9.586894926133794e-08, "loss": 0.293, "step": 27179 }, { "epoch": 1.8790183200829589, "grad_norm": 0.6929252638384976, "learning_rate": 9.575989163068844e-08, "loss": 0.3464, "step": 27180 }, { "epoch": 1.879087452471483, "grad_norm": 0.7135243936920975, "learning_rate": 9.565089546593742e-08, "loss": 0.2639, "step": 27181 }, { "epoch": 1.879156584860007, "grad_norm": 0.7016623161113271, "learning_rate": 9.55419607684499e-08, "loss": 0.259, "step": 27182 }, { "epoch": 1.879225717248531, "grad_norm": 0.6783328575427523, "learning_rate": 9.543308753959146e-08, "loss": 0.2511, "step": 27183 }, { "epoch": 1.879294849637055, "grad_norm": 0.8363264185687621, "learning_rate": 9.532427578072657e-08, "loss": 0.3245, "step": 27184 }, { "epoch": 1.879363982025579, "grad_norm": 0.6957870293234792, "learning_rate": 9.521552549321966e-08, "loss": 0.2669, "step": 27185 }, { "epoch": 1.879433114414103, "grad_norm": 0.6867738663269947, "learning_rate": 9.510683667843302e-08, "loss": 0.234, "step": 27186 }, { "epoch": 1.879502246802627, "grad_norm": 0.6478525016044245, "learning_rate": 9.499820933772885e-08, "loss": 0.3051, "step": 27187 }, { "epoch": 1.879571379191151, "grad_norm": 0.8035268499706127, "learning_rate": 9.488964347246832e-08, "loss": 0.2857, "step": 27188 }, { "epoch": 1.879640511579675, "grad_norm": 0.6573575656928782, "learning_rate": 9.478113908401309e-08, "loss": 0.2485, "step": 27189 }, { "epoch": 1.879709643968199, "grad_norm": 0.7801918944284472, "learning_rate": 9.46726961737221e-08, "loss": 0.2693, "step": 27190 }, { "epoch": 1.879778776356723, "grad_norm": 0.8195944449006201, "learning_rate": 9.456431474295535e-08, "loss": 0.3123, "step": 27191 }, { "epoch": 1.879847908745247, "grad_norm": 1.0014481461842226, "learning_rate": 9.445599479307066e-08, "loss": 0.308, "step": 27192 }, { "epoch": 1.8799170411337713, "grad_norm": 0.6878688692326971, "learning_rate": 9.434773632542582e-08, "loss": 0.202, "step": 27193 }, { "epoch": 1.8799861735222954, "grad_norm": 0.7786294578415828, "learning_rate": 9.423953934137753e-08, "loss": 0.3049, "step": 27194 }, { "epoch": 1.8800553059108194, "grad_norm": 0.8293225304693004, "learning_rate": 9.413140384228192e-08, "loss": 0.3112, "step": 27195 }, { "epoch": 1.8801244382993434, "grad_norm": 0.6923449013202057, "learning_rate": 9.402332982949403e-08, "loss": 0.2617, "step": 27196 }, { "epoch": 1.8801935706878674, "grad_norm": 0.7738276900552549, "learning_rate": 9.391531730436886e-08, "loss": 0.2368, "step": 27197 }, { "epoch": 1.8802627030763914, "grad_norm": 0.6790708036888852, "learning_rate": 9.380736626825926e-08, "loss": 0.3568, "step": 27198 }, { "epoch": 1.8803318354649154, "grad_norm": 0.6981838863045908, "learning_rate": 9.369947672251966e-08, "loss": 0.2468, "step": 27199 }, { "epoch": 1.8804009678534395, "grad_norm": 0.5787640505383959, "learning_rate": 9.359164866850068e-08, "loss": 0.1729, "step": 27200 }, { "epoch": 1.8804701002419635, "grad_norm": 0.672092265719262, "learning_rate": 9.34838821075551e-08, "loss": 0.2792, "step": 27201 }, { "epoch": 1.8805392326304875, "grad_norm": 0.7969418348706753, "learning_rate": 9.337617704103297e-08, "loss": 0.2795, "step": 27202 }, { "epoch": 1.8806083650190115, "grad_norm": 0.7137069975571584, "learning_rate": 9.326853347028376e-08, "loss": 0.2354, "step": 27203 }, { "epoch": 1.8806774974075355, "grad_norm": 0.6943389695531973, "learning_rate": 9.316095139665749e-08, "loss": 0.2724, "step": 27204 }, { "epoch": 1.8807466297960596, "grad_norm": 1.2259999631118168, "learning_rate": 9.305343082150198e-08, "loss": 0.3821, "step": 27205 }, { "epoch": 1.8808157621845836, "grad_norm": 0.7414905624305814, "learning_rate": 9.294597174616504e-08, "loss": 0.2709, "step": 27206 }, { "epoch": 1.8808848945731076, "grad_norm": 0.6287473695608375, "learning_rate": 9.283857417199282e-08, "loss": 0.2206, "step": 27207 }, { "epoch": 1.8809540269616316, "grad_norm": 1.1060954484526855, "learning_rate": 9.273123810033202e-08, "loss": 0.2333, "step": 27208 }, { "epoch": 1.8810231593501556, "grad_norm": 0.7786951677745874, "learning_rate": 9.262396353252768e-08, "loss": 0.3374, "step": 27209 }, { "epoch": 1.8810922917386796, "grad_norm": 0.6718724550877868, "learning_rate": 9.251675046992426e-08, "loss": 0.2353, "step": 27210 }, { "epoch": 1.8811614241272037, "grad_norm": 0.7103602937668726, "learning_rate": 9.240959891386569e-08, "loss": 0.2328, "step": 27211 }, { "epoch": 1.8812305565157277, "grad_norm": 0.7073487761338731, "learning_rate": 9.230250886569481e-08, "loss": 0.3043, "step": 27212 }, { "epoch": 1.8812996889042517, "grad_norm": 0.7157304714575397, "learning_rate": 9.219548032675385e-08, "loss": 0.26, "step": 27213 }, { "epoch": 1.8813688212927757, "grad_norm": 0.7233372740433927, "learning_rate": 9.208851329838453e-08, "loss": 0.22, "step": 27214 }, { "epoch": 1.8814379536812997, "grad_norm": 0.6816921194278011, "learning_rate": 9.198160778192633e-08, "loss": 0.2616, "step": 27215 }, { "epoch": 1.8815070860698238, "grad_norm": 0.7431115412526442, "learning_rate": 9.187476377872095e-08, "loss": 0.3548, "step": 27216 }, { "epoch": 1.8815762184583478, "grad_norm": 0.6993537788490172, "learning_rate": 9.176798129010566e-08, "loss": 0.2417, "step": 27217 }, { "epoch": 1.8816453508468718, "grad_norm": 0.6313881479705563, "learning_rate": 9.16612603174194e-08, "loss": 0.2328, "step": 27218 }, { "epoch": 1.8817144832353958, "grad_norm": 0.770748017634292, "learning_rate": 9.155460086200052e-08, "loss": 0.2889, "step": 27219 }, { "epoch": 1.8817836156239198, "grad_norm": 0.720938550198911, "learning_rate": 9.144800292518519e-08, "loss": 0.3496, "step": 27220 }, { "epoch": 1.8818527480124438, "grad_norm": 0.7124937955033932, "learning_rate": 9.134146650830899e-08, "loss": 0.2449, "step": 27221 }, { "epoch": 1.8819218804009679, "grad_norm": 0.709544003517884, "learning_rate": 9.12349916127081e-08, "loss": 0.2231, "step": 27222 }, { "epoch": 1.8819910127894919, "grad_norm": 0.6005453158288906, "learning_rate": 9.112857823971644e-08, "loss": 0.3293, "step": 27223 }, { "epoch": 1.882060145178016, "grad_norm": 0.7167457803842935, "learning_rate": 9.102222639066738e-08, "loss": 0.2296, "step": 27224 }, { "epoch": 1.88212927756654, "grad_norm": 0.7034400736804108, "learning_rate": 9.09159360668943e-08, "loss": 0.2397, "step": 27225 }, { "epoch": 1.882198409955064, "grad_norm": 0.6848868751369153, "learning_rate": 9.080970726973005e-08, "loss": 0.2656, "step": 27226 }, { "epoch": 1.882267542343588, "grad_norm": 2.5875576874438693, "learning_rate": 9.070354000050408e-08, "loss": 0.3068, "step": 27227 }, { "epoch": 1.882336674732112, "grad_norm": 0.7488450291575204, "learning_rate": 9.059743426054923e-08, "loss": 0.241, "step": 27228 }, { "epoch": 1.882405807120636, "grad_norm": 0.7409103227453111, "learning_rate": 9.049139005119389e-08, "loss": 0.2519, "step": 27229 }, { "epoch": 1.88247493950916, "grad_norm": 0.7521603667775392, "learning_rate": 9.03854073737681e-08, "loss": 0.3076, "step": 27230 }, { "epoch": 1.882544071897684, "grad_norm": 0.7327299176766094, "learning_rate": 9.027948622959859e-08, "loss": 0.2974, "step": 27231 }, { "epoch": 1.882613204286208, "grad_norm": 0.6966225257996108, "learning_rate": 9.017362662001428e-08, "loss": 0.2547, "step": 27232 }, { "epoch": 1.882682336674732, "grad_norm": 0.6986312110081837, "learning_rate": 9.006782854634188e-08, "loss": 0.2463, "step": 27233 }, { "epoch": 1.882751469063256, "grad_norm": 0.7162431365855952, "learning_rate": 8.996209200990701e-08, "loss": 0.3708, "step": 27234 }, { "epoch": 1.88282060145178, "grad_norm": 0.9185512377957716, "learning_rate": 8.985641701203529e-08, "loss": 0.2412, "step": 27235 }, { "epoch": 1.8828897338403041, "grad_norm": 0.5756501233344845, "learning_rate": 8.975080355405064e-08, "loss": 0.1957, "step": 27236 }, { "epoch": 1.8829588662288281, "grad_norm": 0.7166910839671639, "learning_rate": 8.964525163727644e-08, "loss": 0.3224, "step": 27237 }, { "epoch": 1.8830279986173522, "grad_norm": 0.6844173932419502, "learning_rate": 8.95397612630361e-08, "loss": 0.2872, "step": 27238 }, { "epoch": 1.8830971310058762, "grad_norm": 0.6739282721504015, "learning_rate": 8.943433243265187e-08, "loss": 0.257, "step": 27239 }, { "epoch": 1.8831662633944002, "grad_norm": 0.6738354360657767, "learning_rate": 8.932896514744548e-08, "loss": 0.2382, "step": 27240 }, { "epoch": 1.8832353957829242, "grad_norm": 0.6843203347866951, "learning_rate": 8.9223659408737e-08, "loss": 0.3072, "step": 27241 }, { "epoch": 1.8833045281714482, "grad_norm": 0.6847009304528974, "learning_rate": 8.911841521784592e-08, "loss": 0.2529, "step": 27242 }, { "epoch": 1.8833736605599722, "grad_norm": 0.6903985639258813, "learning_rate": 8.901323257609174e-08, "loss": 0.2256, "step": 27243 }, { "epoch": 1.8834427929484963, "grad_norm": 0.8595233087089095, "learning_rate": 8.890811148479283e-08, "loss": 0.3333, "step": 27244 }, { "epoch": 1.8835119253370203, "grad_norm": 0.7341699434573629, "learning_rate": 8.88030519452665e-08, "loss": 0.2805, "step": 27245 }, { "epoch": 1.8835810577255443, "grad_norm": 0.6868781009665148, "learning_rate": 8.869805395882891e-08, "loss": 0.2148, "step": 27246 }, { "epoch": 1.8836501901140683, "grad_norm": 0.8977964604477566, "learning_rate": 8.859311752679678e-08, "loss": 0.2243, "step": 27247 }, { "epoch": 1.8837193225025923, "grad_norm": 0.755792607532563, "learning_rate": 8.848824265048572e-08, "loss": 0.3923, "step": 27248 }, { "epoch": 1.8837884548911163, "grad_norm": 0.7068433560707333, "learning_rate": 8.838342933120914e-08, "loss": 0.2989, "step": 27249 }, { "epoch": 1.8838575872796404, "grad_norm": 0.645242243335503, "learning_rate": 8.827867757028097e-08, "loss": 0.2054, "step": 27250 }, { "epoch": 1.8839267196681644, "grad_norm": 0.7447770824148097, "learning_rate": 8.817398736901461e-08, "loss": 0.2892, "step": 27251 }, { "epoch": 1.8839958520566884, "grad_norm": 0.7607221682648827, "learning_rate": 8.80693587287218e-08, "loss": 0.2896, "step": 27252 }, { "epoch": 1.8840649844452124, "grad_norm": 0.7729256386412865, "learning_rate": 8.796479165071315e-08, "loss": 0.2598, "step": 27253 }, { "epoch": 1.8841341168337367, "grad_norm": 0.5818717666850601, "learning_rate": 8.786028613630037e-08, "loss": 0.2004, "step": 27254 }, { "epoch": 1.8842032492222607, "grad_norm": 0.7069121707172935, "learning_rate": 8.775584218679245e-08, "loss": 0.3113, "step": 27255 }, { "epoch": 1.8842723816107847, "grad_norm": 0.6982980966827307, "learning_rate": 8.765145980349943e-08, "loss": 0.2857, "step": 27256 }, { "epoch": 1.8843415139993087, "grad_norm": 0.649582451395579, "learning_rate": 8.75471389877286e-08, "loss": 0.2032, "step": 27257 }, { "epoch": 1.8844106463878327, "grad_norm": 0.641061739142194, "learning_rate": 8.74428797407878e-08, "loss": 0.2374, "step": 27258 }, { "epoch": 1.8844797787763568, "grad_norm": 0.6458701870977207, "learning_rate": 8.733868206398321e-08, "loss": 0.3125, "step": 27259 }, { "epoch": 1.8845489111648808, "grad_norm": 0.6975073788444608, "learning_rate": 8.723454595862157e-08, "loss": 0.2191, "step": 27260 }, { "epoch": 1.8846180435534048, "grad_norm": 0.672948047537549, "learning_rate": 8.713047142600739e-08, "loss": 0.216, "step": 27261 }, { "epoch": 1.8846871759419288, "grad_norm": 0.9076945869564241, "learning_rate": 8.702645846744573e-08, "loss": 0.3461, "step": 27262 }, { "epoch": 1.8847563083304528, "grad_norm": 0.7174798471457374, "learning_rate": 8.692250708423944e-08, "loss": 0.2673, "step": 27263 }, { "epoch": 1.8848254407189768, "grad_norm": 0.7575648610289125, "learning_rate": 8.681861727769192e-08, "loss": 0.2136, "step": 27264 }, { "epoch": 1.8848945731075009, "grad_norm": 0.7313851812548136, "learning_rate": 8.67147890491049e-08, "loss": 0.2618, "step": 27265 }, { "epoch": 1.8849637054960249, "grad_norm": 0.7557346813884085, "learning_rate": 8.66110223997807e-08, "loss": 0.3872, "step": 27266 }, { "epoch": 1.885032837884549, "grad_norm": 0.6773380545616532, "learning_rate": 8.650731733101825e-08, "loss": 0.2486, "step": 27267 }, { "epoch": 1.885101970273073, "grad_norm": 0.6782246093834934, "learning_rate": 8.640367384411763e-08, "loss": 0.2277, "step": 27268 }, { "epoch": 1.885171102661597, "grad_norm": 0.6677869107075772, "learning_rate": 8.630009194037891e-08, "loss": 0.298, "step": 27269 }, { "epoch": 1.885240235050121, "grad_norm": 0.8231335733354305, "learning_rate": 8.61965716210994e-08, "loss": 0.2494, "step": 27270 }, { "epoch": 1.885309367438645, "grad_norm": 0.6227716534568227, "learning_rate": 8.609311288757749e-08, "loss": 0.1959, "step": 27271 }, { "epoch": 1.885378499827169, "grad_norm": 0.7390257001188025, "learning_rate": 8.598971574110826e-08, "loss": 0.2442, "step": 27272 }, { "epoch": 1.885447632215693, "grad_norm": 0.6972493192371585, "learning_rate": 8.588638018298901e-08, "loss": 0.3599, "step": 27273 }, { "epoch": 1.885516764604217, "grad_norm": 0.6831521066909618, "learning_rate": 8.578310621451368e-08, "loss": 0.2778, "step": 27274 }, { "epoch": 1.8855858969927413, "grad_norm": 0.657878086599435, "learning_rate": 8.567989383697794e-08, "loss": 0.2214, "step": 27275 }, { "epoch": 1.8856550293812653, "grad_norm": 0.6960692894344807, "learning_rate": 8.557674305167407e-08, "loss": 0.2669, "step": 27276 }, { "epoch": 1.8857241617697893, "grad_norm": 0.8003519441893121, "learning_rate": 8.547365385989659e-08, "loss": 0.3399, "step": 27277 }, { "epoch": 1.8857932941583133, "grad_norm": 0.7043372611343494, "learning_rate": 8.537062626293557e-08, "loss": 0.2284, "step": 27278 }, { "epoch": 1.8858624265468373, "grad_norm": 0.7033209970878279, "learning_rate": 8.526766026208388e-08, "loss": 0.2503, "step": 27279 }, { "epoch": 1.8859315589353614, "grad_norm": 0.7209562257374325, "learning_rate": 8.516475585863049e-08, "loss": 0.3424, "step": 27280 }, { "epoch": 1.8860006913238854, "grad_norm": 0.7660022934491841, "learning_rate": 8.506191305386657e-08, "loss": 0.304, "step": 27281 }, { "epoch": 1.8860698237124094, "grad_norm": 0.725294649375115, "learning_rate": 8.495913184908e-08, "loss": 0.2888, "step": 27282 }, { "epoch": 1.8861389561009334, "grad_norm": 0.7752420816423087, "learning_rate": 8.485641224556029e-08, "loss": 0.2889, "step": 27283 }, { "epoch": 1.8862080884894574, "grad_norm": 0.7067410834539432, "learning_rate": 8.475375424459309e-08, "loss": 0.3152, "step": 27284 }, { "epoch": 1.8862772208779814, "grad_norm": 0.7441639695954503, "learning_rate": 8.465115784746681e-08, "loss": 0.2828, "step": 27285 }, { "epoch": 1.8863463532665055, "grad_norm": 0.708818165666497, "learning_rate": 8.454862305546596e-08, "loss": 0.2578, "step": 27286 }, { "epoch": 1.8864154856550295, "grad_norm": 0.7076237263537246, "learning_rate": 8.444614986987621e-08, "loss": 0.2759, "step": 27287 }, { "epoch": 1.8864846180435535, "grad_norm": 0.7167184383496565, "learning_rate": 8.43437382919815e-08, "loss": 0.2595, "step": 27288 }, { "epoch": 1.8865537504320775, "grad_norm": 0.6559098980480639, "learning_rate": 8.424138832306583e-08, "loss": 0.1982, "step": 27289 }, { "epoch": 1.8866228828206015, "grad_norm": 0.7734088535788926, "learning_rate": 8.41390999644115e-08, "loss": 0.2397, "step": 27290 }, { "epoch": 1.8866920152091256, "grad_norm": 0.633975826257249, "learning_rate": 8.403687321730137e-08, "loss": 0.3333, "step": 27291 }, { "epoch": 1.8867611475976496, "grad_norm": 0.6990001473941531, "learning_rate": 8.393470808301552e-08, "loss": 0.2467, "step": 27292 }, { "epoch": 1.8868302799861736, "grad_norm": 0.685695689384828, "learning_rate": 8.383260456283571e-08, "loss": 0.2015, "step": 27293 }, { "epoch": 1.8868994123746976, "grad_norm": 0.6892812292034961, "learning_rate": 8.373056265804036e-08, "loss": 0.3285, "step": 27294 }, { "epoch": 1.8869685447632216, "grad_norm": 0.8995140673293684, "learning_rate": 8.362858236990901e-08, "loss": 0.2921, "step": 27295 }, { "epoch": 1.8870376771517456, "grad_norm": 0.696410116978164, "learning_rate": 8.352666369971952e-08, "loss": 0.2084, "step": 27296 }, { "epoch": 1.8871068095402697, "grad_norm": 0.6837906463098133, "learning_rate": 8.342480664874974e-08, "loss": 0.2339, "step": 27297 }, { "epoch": 1.8871759419287937, "grad_norm": 0.7168894958356989, "learning_rate": 8.33230112182759e-08, "loss": 0.3749, "step": 27298 }, { "epoch": 1.8872450743173177, "grad_norm": 0.8122707841730056, "learning_rate": 8.322127740957419e-08, "loss": 0.2922, "step": 27299 }, { "epoch": 1.8873142067058417, "grad_norm": 0.7329173204729821, "learning_rate": 8.311960522391916e-08, "loss": 0.2065, "step": 27300 }, { "epoch": 1.8873833390943657, "grad_norm": 0.7262199572893288, "learning_rate": 8.301799466258532e-08, "loss": 0.273, "step": 27301 }, { "epoch": 1.8874524714828897, "grad_norm": 0.7580100858595568, "learning_rate": 8.291644572684609e-08, "loss": 0.2881, "step": 27302 }, { "epoch": 1.8875216038714138, "grad_norm": 0.5832234141521805, "learning_rate": 8.281495841797437e-08, "loss": 0.1719, "step": 27303 }, { "epoch": 1.8875907362599378, "grad_norm": 1.1524459290855653, "learning_rate": 8.271353273724136e-08, "loss": 0.2046, "step": 27304 }, { "epoch": 1.8876598686484618, "grad_norm": 0.721256600261805, "learning_rate": 8.261216868591992e-08, "loss": 0.384, "step": 27305 }, { "epoch": 1.8877290010369858, "grad_norm": 0.7349267652763338, "learning_rate": 8.251086626527904e-08, "loss": 0.2663, "step": 27306 }, { "epoch": 1.8877981334255098, "grad_norm": 0.7008286609879326, "learning_rate": 8.240962547658937e-08, "loss": 0.204, "step": 27307 }, { "epoch": 1.8878672658140339, "grad_norm": 0.6984965174153039, "learning_rate": 8.230844632111823e-08, "loss": 0.2641, "step": 27308 }, { "epoch": 1.8879363982025579, "grad_norm": 0.7503166237042787, "learning_rate": 8.220732880013571e-08, "loss": 0.3579, "step": 27309 }, { "epoch": 1.888005530591082, "grad_norm": 0.7161209750710809, "learning_rate": 8.210627291490692e-08, "loss": 0.2307, "step": 27310 }, { "epoch": 1.888074662979606, "grad_norm": 0.6674264343974489, "learning_rate": 8.200527866670083e-08, "loss": 0.238, "step": 27311 }, { "epoch": 1.88814379536813, "grad_norm": 0.7704273698088123, "learning_rate": 8.190434605678088e-08, "loss": 0.3844, "step": 27312 }, { "epoch": 1.888212927756654, "grad_norm": 0.7307135137689491, "learning_rate": 8.180347508641439e-08, "loss": 0.273, "step": 27313 }, { "epoch": 1.888282060145178, "grad_norm": 0.6238225556312059, "learning_rate": 8.17026657568637e-08, "loss": 0.2133, "step": 27314 }, { "epoch": 1.888351192533702, "grad_norm": 0.6430641390168795, "learning_rate": 8.160191806939333e-08, "loss": 0.2486, "step": 27315 }, { "epoch": 1.888420324922226, "grad_norm": 0.8061813434120837, "learning_rate": 8.150123202526561e-08, "loss": 0.3472, "step": 27316 }, { "epoch": 1.88848945731075, "grad_norm": 0.7184259543448414, "learning_rate": 8.140060762574176e-08, "loss": 0.2869, "step": 27317 }, { "epoch": 1.888558589699274, "grad_norm": 0.63443374280641, "learning_rate": 8.13000448720841e-08, "loss": 0.1895, "step": 27318 }, { "epoch": 1.888627722087798, "grad_norm": 0.6491070179220758, "learning_rate": 8.119954376555328e-08, "loss": 0.2818, "step": 27319 }, { "epoch": 1.888696854476322, "grad_norm": 0.7055198700832562, "learning_rate": 8.109910430740719e-08, "loss": 0.3146, "step": 27320 }, { "epoch": 1.888765986864846, "grad_norm": 0.7045433143067621, "learning_rate": 8.09987264989065e-08, "loss": 0.2634, "step": 27321 }, { "epoch": 1.88883511925337, "grad_norm": 0.6916235921121552, "learning_rate": 8.089841034130796e-08, "loss": 0.2725, "step": 27322 }, { "epoch": 1.8889042516418941, "grad_norm": 0.7090474910503191, "learning_rate": 8.079815583586948e-08, "loss": 0.3179, "step": 27323 }, { "epoch": 1.8889733840304181, "grad_norm": 0.6941669139930626, "learning_rate": 8.069796298384725e-08, "loss": 0.2114, "step": 27324 }, { "epoch": 1.8890425164189422, "grad_norm": 0.7275078233013037, "learning_rate": 8.059783178649693e-08, "loss": 0.2319, "step": 27325 }, { "epoch": 1.8891116488074662, "grad_norm": 0.6929642663752187, "learning_rate": 8.049776224507422e-08, "loss": 0.2535, "step": 27326 }, { "epoch": 1.8891807811959902, "grad_norm": 0.7757929557622351, "learning_rate": 8.039775436083253e-08, "loss": 0.3172, "step": 27327 }, { "epoch": 1.8892499135845142, "grad_norm": 0.6575492499187676, "learning_rate": 8.029780813502585e-08, "loss": 0.2295, "step": 27328 }, { "epoch": 1.8893190459730382, "grad_norm": 0.7651196530405302, "learning_rate": 8.019792356890654e-08, "loss": 0.2895, "step": 27329 }, { "epoch": 1.8893881783615623, "grad_norm": 0.7923211161069225, "learning_rate": 8.00981006637258e-08, "loss": 0.33, "step": 27330 }, { "epoch": 1.8894573107500863, "grad_norm": 0.7721610268180642, "learning_rate": 7.999833942073598e-08, "loss": 0.267, "step": 27331 }, { "epoch": 1.8895264431386103, "grad_norm": 0.6757174119472131, "learning_rate": 7.989863984118663e-08, "loss": 0.2249, "step": 27332 }, { "epoch": 1.8895955755271343, "grad_norm": 0.6198525721587881, "learning_rate": 7.979900192632783e-08, "loss": 0.2295, "step": 27333 }, { "epoch": 1.8896647079156583, "grad_norm": 0.7215197457512087, "learning_rate": 7.969942567740752e-08, "loss": 0.3393, "step": 27334 }, { "epoch": 1.8897338403041823, "grad_norm": 0.7605151695174445, "learning_rate": 7.959991109567467e-08, "loss": 0.301, "step": 27335 }, { "epoch": 1.8898029726927066, "grad_norm": 0.69545836539493, "learning_rate": 7.950045818237606e-08, "loss": 0.2497, "step": 27336 }, { "epoch": 1.8898721050812306, "grad_norm": 0.7100496274126321, "learning_rate": 7.940106693875793e-08, "loss": 0.3051, "step": 27337 }, { "epoch": 1.8899412374697546, "grad_norm": 0.7474865000375321, "learning_rate": 7.93017373660665e-08, "loss": 0.3247, "step": 27338 }, { "epoch": 1.8900103698582786, "grad_norm": 0.6003405535114936, "learning_rate": 7.920246946554633e-08, "loss": 0.2426, "step": 27339 }, { "epoch": 1.8900795022468027, "grad_norm": 0.717796760954057, "learning_rate": 7.910326323844142e-08, "loss": 0.2092, "step": 27340 }, { "epoch": 1.8901486346353267, "grad_norm": 0.7664972393436661, "learning_rate": 7.900411868599577e-08, "loss": 0.3334, "step": 27341 }, { "epoch": 1.8902177670238507, "grad_norm": 0.714336577952946, "learning_rate": 7.890503580945175e-08, "loss": 0.2879, "step": 27342 }, { "epoch": 1.8902868994123747, "grad_norm": 0.6692948682676988, "learning_rate": 7.880601461005e-08, "loss": 0.2462, "step": 27343 }, { "epoch": 1.8903560318008987, "grad_norm": 0.7682964092490508, "learning_rate": 7.870705508903343e-08, "loss": 0.3483, "step": 27344 }, { "epoch": 1.8904251641894227, "grad_norm": 0.7691000511629502, "learning_rate": 7.860815724764104e-08, "loss": 0.3093, "step": 27345 }, { "epoch": 1.8904942965779468, "grad_norm": 0.710072416702577, "learning_rate": 7.85093210871135e-08, "loss": 0.2152, "step": 27346 }, { "epoch": 1.8905634289664708, "grad_norm": 0.7086607194535275, "learning_rate": 7.841054660868819e-08, "loss": 0.2323, "step": 27347 }, { "epoch": 1.8906325613549948, "grad_norm": 0.7061855376027817, "learning_rate": 7.831183381360407e-08, "loss": 0.322, "step": 27348 }, { "epoch": 1.8907016937435188, "grad_norm": 0.7163313411415165, "learning_rate": 7.821318270309742e-08, "loss": 0.2345, "step": 27349 }, { "epoch": 1.8907708261320428, "grad_norm": 1.1121065752833672, "learning_rate": 7.811459327840554e-08, "loss": 0.2607, "step": 27350 }, { "epoch": 1.8908399585205669, "grad_norm": 0.7316799438414107, "learning_rate": 7.801606554076358e-08, "loss": 0.3085, "step": 27351 }, { "epoch": 1.8909090909090909, "grad_norm": 0.7730919639544118, "learning_rate": 7.791759949140665e-08, "loss": 0.3237, "step": 27352 }, { "epoch": 1.890978223297615, "grad_norm": 0.6978238437526448, "learning_rate": 7.781919513156876e-08, "loss": 0.2425, "step": 27353 }, { "epoch": 1.891047355686139, "grad_norm": 0.7155661060131477, "learning_rate": 7.77208524624834e-08, "loss": 0.259, "step": 27354 }, { "epoch": 1.891116488074663, "grad_norm": 0.7492608318024813, "learning_rate": 7.762257148538289e-08, "loss": 0.3274, "step": 27355 }, { "epoch": 1.8911856204631872, "grad_norm": 0.6564196215369255, "learning_rate": 7.752435220149956e-08, "loss": 0.185, "step": 27356 }, { "epoch": 1.8912547528517112, "grad_norm": 0.726210291478087, "learning_rate": 7.742619461206303e-08, "loss": 0.2146, "step": 27357 }, { "epoch": 1.8913238852402352, "grad_norm": 0.7747651245080811, "learning_rate": 7.732809871830505e-08, "loss": 0.2359, "step": 27358 }, { "epoch": 1.8913930176287592, "grad_norm": 0.7450544708985928, "learning_rate": 7.72300645214541e-08, "loss": 0.3201, "step": 27359 }, { "epoch": 1.8914621500172832, "grad_norm": 0.7565001097047099, "learning_rate": 7.713209202273919e-08, "loss": 0.2958, "step": 27360 }, { "epoch": 1.8915312824058073, "grad_norm": 0.6639204597330417, "learning_rate": 7.703418122338823e-08, "loss": 0.2197, "step": 27361 }, { "epoch": 1.8916004147943313, "grad_norm": 0.6868635396123186, "learning_rate": 7.693633212462858e-08, "loss": 0.2815, "step": 27362 }, { "epoch": 1.8916695471828553, "grad_norm": 0.751882759392402, "learning_rate": 7.683854472768649e-08, "loss": 0.3307, "step": 27363 }, { "epoch": 1.8917386795713793, "grad_norm": 0.7252961808883119, "learning_rate": 7.67408190337876e-08, "loss": 0.1917, "step": 27364 }, { "epoch": 1.8918078119599033, "grad_norm": 0.7074394100340159, "learning_rate": 7.664315504415599e-08, "loss": 0.2548, "step": 27365 }, { "epoch": 1.8918769443484273, "grad_norm": 0.7045974164711993, "learning_rate": 7.654555276001619e-08, "loss": 0.298, "step": 27366 }, { "epoch": 1.8919460767369514, "grad_norm": 0.7067511582642626, "learning_rate": 7.644801218259223e-08, "loss": 0.2879, "step": 27367 }, { "epoch": 1.8920152091254754, "grad_norm": 0.6558928148284522, "learning_rate": 7.635053331310593e-08, "loss": 0.2065, "step": 27368 }, { "epoch": 1.8920843415139994, "grad_norm": 0.6987216883135735, "learning_rate": 7.625311615277852e-08, "loss": 0.3178, "step": 27369 }, { "epoch": 1.8921534739025234, "grad_norm": 0.738888010032573, "learning_rate": 7.615576070283236e-08, "loss": 0.3228, "step": 27370 }, { "epoch": 1.8922226062910474, "grad_norm": 0.7834923772216383, "learning_rate": 7.605846696448594e-08, "loss": 0.2259, "step": 27371 }, { "epoch": 1.8922917386795715, "grad_norm": 0.7328014801461932, "learning_rate": 7.59612349389599e-08, "loss": 0.2198, "step": 27372 }, { "epoch": 1.8923608710680955, "grad_norm": 0.7514981592699947, "learning_rate": 7.58640646274722e-08, "loss": 0.3623, "step": 27373 }, { "epoch": 1.8924300034566195, "grad_norm": 0.8028793685965114, "learning_rate": 7.576695603124074e-08, "loss": 0.3234, "step": 27374 }, { "epoch": 1.8924991358451435, "grad_norm": 0.6904160577939665, "learning_rate": 7.566990915148343e-08, "loss": 0.2723, "step": 27375 }, { "epoch": 1.8925682682336675, "grad_norm": 0.698860943296337, "learning_rate": 7.55729239894154e-08, "loss": 0.3092, "step": 27376 }, { "epoch": 1.8926374006221915, "grad_norm": 0.6703581932489804, "learning_rate": 7.547600054625293e-08, "loss": 0.3134, "step": 27377 }, { "epoch": 1.8927065330107156, "grad_norm": 0.66300676421704, "learning_rate": 7.537913882321057e-08, "loss": 0.1969, "step": 27378 }, { "epoch": 1.8927756653992396, "grad_norm": 0.8997645530578878, "learning_rate": 7.528233882150238e-08, "loss": 0.203, "step": 27379 }, { "epoch": 1.8928447977877636, "grad_norm": 0.731751176463332, "learning_rate": 7.51856005423418e-08, "loss": 0.3814, "step": 27380 }, { "epoch": 1.8929139301762876, "grad_norm": 0.7003618731525253, "learning_rate": 7.508892398694012e-08, "loss": 0.2701, "step": 27381 }, { "epoch": 1.8929830625648116, "grad_norm": 0.6896250838895426, "learning_rate": 7.499230915651079e-08, "loss": 0.2088, "step": 27382 }, { "epoch": 1.8930521949533357, "grad_norm": 0.6800622547591685, "learning_rate": 7.48957560522634e-08, "loss": 0.2351, "step": 27383 }, { "epoch": 1.8931213273418597, "grad_norm": 0.7314258714886956, "learning_rate": 7.479926467540865e-08, "loss": 0.3447, "step": 27384 }, { "epoch": 1.8931904597303837, "grad_norm": 0.6375610041934362, "learning_rate": 7.470283502715559e-08, "loss": 0.2017, "step": 27385 }, { "epoch": 1.8932595921189077, "grad_norm": 0.7275053885328171, "learning_rate": 7.460646710871267e-08, "loss": 0.2629, "step": 27386 }, { "epoch": 1.8933287245074317, "grad_norm": 0.7191838199039974, "learning_rate": 7.451016092128838e-08, "loss": 0.3248, "step": 27387 }, { "epoch": 1.8933978568959557, "grad_norm": 0.7328821338855225, "learning_rate": 7.441391646608842e-08, "loss": 0.4028, "step": 27388 }, { "epoch": 1.8934669892844798, "grad_norm": 0.6699505040512056, "learning_rate": 7.431773374432071e-08, "loss": 0.2216, "step": 27389 }, { "epoch": 1.8935361216730038, "grad_norm": 0.8319650701174974, "learning_rate": 7.422161275718987e-08, "loss": 0.2422, "step": 27390 }, { "epoch": 1.8936052540615278, "grad_norm": 0.7025506153443987, "learning_rate": 7.412555350590045e-08, "loss": 0.3262, "step": 27391 }, { "epoch": 1.8936743864500518, "grad_norm": 0.7507887425760107, "learning_rate": 7.402955599165707e-08, "loss": 0.2776, "step": 27392 }, { "epoch": 1.8937435188385758, "grad_norm": 0.662162970420045, "learning_rate": 7.393362021566153e-08, "loss": 0.2424, "step": 27393 }, { "epoch": 1.8938126512270999, "grad_norm": 0.7382315806371983, "learning_rate": 7.383774617911787e-08, "loss": 0.3071, "step": 27394 }, { "epoch": 1.8938817836156239, "grad_norm": 0.7548225859592873, "learning_rate": 7.374193388322682e-08, "loss": 0.306, "step": 27395 }, { "epoch": 1.8939509160041479, "grad_norm": 0.7353279125149692, "learning_rate": 7.364618332918904e-08, "loss": 0.2607, "step": 27396 }, { "epoch": 1.894020048392672, "grad_norm": 0.6969215798997617, "learning_rate": 7.355049451820472e-08, "loss": 0.2738, "step": 27397 }, { "epoch": 1.894089180781196, "grad_norm": 0.7282352136345326, "learning_rate": 7.345486745147401e-08, "loss": 0.3632, "step": 27398 }, { "epoch": 1.89415831316972, "grad_norm": 0.756364875848105, "learning_rate": 7.335930213019427e-08, "loss": 0.2155, "step": 27399 }, { "epoch": 1.894227445558244, "grad_norm": 0.6767516027403688, "learning_rate": 7.326379855556398e-08, "loss": 0.2008, "step": 27400 }, { "epoch": 1.894296577946768, "grad_norm": 0.6887980572108698, "learning_rate": 7.316835672877942e-08, "loss": 0.277, "step": 27401 }, { "epoch": 1.894365710335292, "grad_norm": 0.6948768487532887, "learning_rate": 7.307297665103686e-08, "loss": 0.2556, "step": 27402 }, { "epoch": 1.894434842723816, "grad_norm": 0.7248848019195295, "learning_rate": 7.29776583235331e-08, "loss": 0.2409, "step": 27403 }, { "epoch": 1.89450397511234, "grad_norm": 0.695756813395929, "learning_rate": 7.288240174746109e-08, "loss": 0.2499, "step": 27404 }, { "epoch": 1.894573107500864, "grad_norm": 0.706864715670665, "learning_rate": 7.278720692401541e-08, "loss": 0.3139, "step": 27405 }, { "epoch": 1.894642239889388, "grad_norm": 0.7326250335796479, "learning_rate": 7.269207385438903e-08, "loss": 0.2688, "step": 27406 }, { "epoch": 1.894711372277912, "grad_norm": 0.6393177407663189, "learning_rate": 7.25970025397743e-08, "loss": 0.2329, "step": 27407 }, { "epoch": 1.894780504666436, "grad_norm": 1.0471220052324273, "learning_rate": 7.250199298136306e-08, "loss": 0.2647, "step": 27408 }, { "epoch": 1.8948496370549601, "grad_norm": 0.6903727122794957, "learning_rate": 7.240704518034548e-08, "loss": 0.3103, "step": 27409 }, { "epoch": 1.8949187694434841, "grad_norm": 0.7076641538738182, "learning_rate": 7.231215913791168e-08, "loss": 0.2264, "step": 27410 }, { "epoch": 1.8949879018320082, "grad_norm": 0.6356001264141771, "learning_rate": 7.221733485525184e-08, "loss": 0.1888, "step": 27411 }, { "epoch": 1.8950570342205322, "grad_norm": 0.7828070500678505, "learning_rate": 7.212257233355336e-08, "loss": 0.3609, "step": 27412 }, { "epoch": 1.8951261666090562, "grad_norm": 1.0019876087171466, "learning_rate": 7.202787157400416e-08, "loss": 0.2833, "step": 27413 }, { "epoch": 1.8951952989975802, "grad_norm": 0.6990765419105803, "learning_rate": 7.193323257779105e-08, "loss": 0.1987, "step": 27414 }, { "epoch": 1.8952644313861042, "grad_norm": 0.6553095906258927, "learning_rate": 7.183865534610091e-08, "loss": 0.1904, "step": 27415 }, { "epoch": 1.8953335637746282, "grad_norm": 0.7320731119475852, "learning_rate": 7.174413988011775e-08, "loss": 0.3271, "step": 27416 }, { "epoch": 1.8954026961631525, "grad_norm": 0.6805275211223064, "learning_rate": 7.164968618102785e-08, "loss": 0.2774, "step": 27417 }, { "epoch": 1.8954718285516765, "grad_norm": 0.6913518861432565, "learning_rate": 7.155529425001307e-08, "loss": 0.2337, "step": 27418 }, { "epoch": 1.8955409609402005, "grad_norm": 0.7224709381170795, "learning_rate": 7.146096408825854e-08, "loss": 0.2918, "step": 27419 }, { "epoch": 1.8956100933287245, "grad_norm": 0.6844606189732825, "learning_rate": 7.136669569694444e-08, "loss": 0.2312, "step": 27420 }, { "epoch": 1.8956792257172486, "grad_norm": 0.7088562199827166, "learning_rate": 7.127248907725425e-08, "loss": 0.2382, "step": 27421 }, { "epoch": 1.8957483581057726, "grad_norm": 0.6417610371833329, "learning_rate": 7.117834423036707e-08, "loss": 0.2102, "step": 27422 }, { "epoch": 1.8958174904942966, "grad_norm": 0.6720116446419765, "learning_rate": 7.108426115746358e-08, "loss": 0.305, "step": 27423 }, { "epoch": 1.8958866228828206, "grad_norm": 0.6542253638637959, "learning_rate": 7.099023985972286e-08, "loss": 0.2443, "step": 27424 }, { "epoch": 1.8959557552713446, "grad_norm": 0.6171966516322108, "learning_rate": 7.089628033832341e-08, "loss": 0.2041, "step": 27425 }, { "epoch": 1.8960248876598687, "grad_norm": 0.7956025867226438, "learning_rate": 7.080238259444205e-08, "loss": 0.2555, "step": 27426 }, { "epoch": 1.8960940200483927, "grad_norm": 0.6951367872207604, "learning_rate": 7.07085466292573e-08, "loss": 0.3208, "step": 27427 }, { "epoch": 1.8961631524369167, "grad_norm": 0.8126525590115368, "learning_rate": 7.061477244394321e-08, "loss": 0.2562, "step": 27428 }, { "epoch": 1.8962322848254407, "grad_norm": 0.7873968453575229, "learning_rate": 7.052106003967719e-08, "loss": 0.2802, "step": 27429 }, { "epoch": 1.8963014172139647, "grad_norm": 0.7222373312390269, "learning_rate": 7.04274094176316e-08, "loss": 0.3659, "step": 27430 }, { "epoch": 1.8963705496024887, "grad_norm": 0.6963616845359103, "learning_rate": 7.033382057898164e-08, "loss": 0.2733, "step": 27431 }, { "epoch": 1.8964396819910128, "grad_norm": 0.6347525674922414, "learning_rate": 7.024029352489969e-08, "loss": 0.2238, "step": 27432 }, { "epoch": 1.8965088143795368, "grad_norm": 0.8143810298839564, "learning_rate": 7.014682825655816e-08, "loss": 0.245, "step": 27433 }, { "epoch": 1.8965779467680608, "grad_norm": 0.757876508838305, "learning_rate": 7.005342477512889e-08, "loss": 0.3969, "step": 27434 }, { "epoch": 1.8966470791565848, "grad_norm": 0.7318750205316792, "learning_rate": 6.996008308178148e-08, "loss": 0.2858, "step": 27435 }, { "epoch": 1.8967162115451088, "grad_norm": 0.7768937879662555, "learning_rate": 6.986680317768668e-08, "loss": 0.2461, "step": 27436 }, { "epoch": 1.896785343933633, "grad_norm": 0.6988648041773559, "learning_rate": 6.9773585064013e-08, "loss": 0.2959, "step": 27437 }, { "epoch": 1.896854476322157, "grad_norm": 0.7361668011837752, "learning_rate": 6.968042874192893e-08, "loss": 0.2822, "step": 27438 }, { "epoch": 1.896923608710681, "grad_norm": 0.6906957199959802, "learning_rate": 6.958733421260244e-08, "loss": 0.2144, "step": 27439 }, { "epoch": 1.8969927410992051, "grad_norm": 0.7427407130179252, "learning_rate": 6.949430147720038e-08, "loss": 0.3187, "step": 27440 }, { "epoch": 1.8970618734877291, "grad_norm": 0.7043281038125172, "learning_rate": 6.940133053688791e-08, "loss": 0.3358, "step": 27441 }, { "epoch": 1.8971310058762532, "grad_norm": 0.8372757556425277, "learning_rate": 6.930842139283078e-08, "loss": 0.2313, "step": 27442 }, { "epoch": 1.8972001382647772, "grad_norm": 0.761497383830942, "learning_rate": 6.921557404619361e-08, "loss": 0.2392, "step": 27443 }, { "epoch": 1.8972692706533012, "grad_norm": 0.6670485499302619, "learning_rate": 6.912278849813936e-08, "loss": 0.2924, "step": 27444 }, { "epoch": 1.8973384030418252, "grad_norm": 0.7406774503948218, "learning_rate": 6.903006474983153e-08, "loss": 0.3082, "step": 27445 }, { "epoch": 1.8974075354303492, "grad_norm": 0.6512932693394293, "learning_rate": 6.8937402802432e-08, "loss": 0.2536, "step": 27446 }, { "epoch": 1.8974766678188733, "grad_norm": 0.7659208587067564, "learning_rate": 6.884480265710258e-08, "loss": 0.263, "step": 27447 }, { "epoch": 1.8975458002073973, "grad_norm": 0.7429669491429259, "learning_rate": 6.875226431500348e-08, "loss": 0.3648, "step": 27448 }, { "epoch": 1.8976149325959213, "grad_norm": 0.7797807653774539, "learning_rate": 6.865978777729487e-08, "loss": 0.2727, "step": 27449 }, { "epoch": 1.8976840649844453, "grad_norm": 0.7361949514387937, "learning_rate": 6.856737304513472e-08, "loss": 0.296, "step": 27450 }, { "epoch": 1.8977531973729693, "grad_norm": 0.7275623133614426, "learning_rate": 6.847502011968266e-08, "loss": 0.2445, "step": 27451 }, { "epoch": 1.8978223297614933, "grad_norm": 0.738387084374026, "learning_rate": 6.838272900209442e-08, "loss": 0.345, "step": 27452 }, { "epoch": 1.8978914621500174, "grad_norm": 0.8012355928890045, "learning_rate": 6.829049969352908e-08, "loss": 0.2629, "step": 27453 }, { "epoch": 1.8979605945385414, "grad_norm": 0.6798438818573476, "learning_rate": 6.819833219514072e-08, "loss": 0.2569, "step": 27454 }, { "epoch": 1.8980297269270654, "grad_norm": 1.0659568664636712, "learning_rate": 6.810622650808508e-08, "loss": 0.2973, "step": 27455 }, { "epoch": 1.8980988593155894, "grad_norm": 0.85036423357908, "learning_rate": 6.801418263351677e-08, "loss": 0.305, "step": 27456 }, { "epoch": 1.8981679917041134, "grad_norm": 0.6867202753321016, "learning_rate": 6.792220057258936e-08, "loss": 0.1926, "step": 27457 }, { "epoch": 1.8982371240926375, "grad_norm": 0.775174968794953, "learning_rate": 6.783028032645467e-08, "loss": 0.3009, "step": 27458 }, { "epoch": 1.8983062564811615, "grad_norm": 0.631075462849129, "learning_rate": 6.773842189626622e-08, "loss": 0.3071, "step": 27459 }, { "epoch": 1.8983753888696855, "grad_norm": 0.6661243226026092, "learning_rate": 6.764662528317423e-08, "loss": 0.2098, "step": 27460 }, { "epoch": 1.8984445212582095, "grad_norm": 0.7380658633377656, "learning_rate": 6.755489048832998e-08, "loss": 0.2161, "step": 27461 }, { "epoch": 1.8985136536467335, "grad_norm": 0.7280943603105201, "learning_rate": 6.746321751288255e-08, "loss": 0.2877, "step": 27462 }, { "epoch": 1.8985827860352575, "grad_norm": 0.7026651538452353, "learning_rate": 6.737160635798156e-08, "loss": 0.3372, "step": 27463 }, { "epoch": 1.8986519184237816, "grad_norm": 0.6437183617634529, "learning_rate": 6.728005702477447e-08, "loss": 0.1944, "step": 27464 }, { "epoch": 1.8987210508123056, "grad_norm": 0.7615632378269866, "learning_rate": 6.718856951440922e-08, "loss": 0.2932, "step": 27465 }, { "epoch": 1.8987901832008296, "grad_norm": 0.7223673602720275, "learning_rate": 6.709714382803212e-08, "loss": 0.3782, "step": 27466 }, { "epoch": 1.8988593155893536, "grad_norm": 0.7335102093219187, "learning_rate": 6.700577996678892e-08, "loss": 0.21, "step": 27467 }, { "epoch": 1.8989284479778776, "grad_norm": 0.6233905237635893, "learning_rate": 6.691447793182538e-08, "loss": 0.2079, "step": 27468 }, { "epoch": 1.8989975803664016, "grad_norm": 0.716663346621055, "learning_rate": 6.682323772428501e-08, "loss": 0.346, "step": 27469 }, { "epoch": 1.8990667127549257, "grad_norm": 0.6958048710954597, "learning_rate": 6.673205934531246e-08, "loss": 0.3405, "step": 27470 }, { "epoch": 1.8991358451434497, "grad_norm": 0.6727763972588013, "learning_rate": 6.664094279604905e-08, "loss": 0.2041, "step": 27471 }, { "epoch": 1.8992049775319737, "grad_norm": 0.6865091028781355, "learning_rate": 6.654988807763717e-08, "loss": 0.2316, "step": 27472 }, { "epoch": 1.8992741099204977, "grad_norm": 0.7165303962429099, "learning_rate": 6.645889519121874e-08, "loss": 0.2772, "step": 27473 }, { "epoch": 1.8993432423090217, "grad_norm": 0.9325473918018687, "learning_rate": 6.636796413793389e-08, "loss": 0.22, "step": 27474 }, { "epoch": 1.8994123746975458, "grad_norm": 0.7050951350058177, "learning_rate": 6.627709491892176e-08, "loss": 0.2479, "step": 27475 }, { "epoch": 1.8994815070860698, "grad_norm": 0.6505112177325272, "learning_rate": 6.618628753532197e-08, "loss": 0.2847, "step": 27476 }, { "epoch": 1.8995506394745938, "grad_norm": 0.7002635372685337, "learning_rate": 6.609554198827195e-08, "loss": 0.3107, "step": 27477 }, { "epoch": 1.8996197718631178, "grad_norm": 0.8937883768924983, "learning_rate": 6.600485827891023e-08, "loss": 0.2715, "step": 27478 }, { "epoch": 1.8996889042516418, "grad_norm": 0.6516386726345915, "learning_rate": 6.591423640837147e-08, "loss": 0.2475, "step": 27479 }, { "epoch": 1.8997580366401658, "grad_norm": 0.7254154618549055, "learning_rate": 6.582367637779308e-08, "loss": 0.2997, "step": 27480 }, { "epoch": 1.8998271690286899, "grad_norm": 0.7344618293782876, "learning_rate": 6.57331781883086e-08, "loss": 0.2534, "step": 27481 }, { "epoch": 1.8998963014172139, "grad_norm": 0.6551766038070983, "learning_rate": 6.564274184105435e-08, "loss": 0.2893, "step": 27482 }, { "epoch": 1.899965433805738, "grad_norm": 0.6463593812432687, "learning_rate": 6.555236733716163e-08, "loss": 0.2573, "step": 27483 }, { "epoch": 1.900034566194262, "grad_norm": 0.7931402476778051, "learning_rate": 6.546205467776512e-08, "loss": 0.3778, "step": 27484 }, { "epoch": 1.900103698582786, "grad_norm": 0.7212549684140509, "learning_rate": 6.537180386399445e-08, "loss": 0.2679, "step": 27485 }, { "epoch": 1.90017283097131, "grad_norm": 0.819613644243199, "learning_rate": 6.528161489698315e-08, "loss": 0.2915, "step": 27486 }, { "epoch": 1.900241963359834, "grad_norm": 0.7054092686867841, "learning_rate": 6.519148777785922e-08, "loss": 0.2842, "step": 27487 }, { "epoch": 1.900311095748358, "grad_norm": 0.7421513386842183, "learning_rate": 6.510142250775397e-08, "loss": 0.2361, "step": 27488 }, { "epoch": 1.900380228136882, "grad_norm": 2.643082166965259, "learning_rate": 6.50114190877954e-08, "loss": 0.1757, "step": 27489 }, { "epoch": 1.900449360525406, "grad_norm": 0.7503419594452888, "learning_rate": 6.49214775191126e-08, "loss": 0.3118, "step": 27490 }, { "epoch": 1.90051849291393, "grad_norm": 0.7077238261739384, "learning_rate": 6.483159780283133e-08, "loss": 0.2976, "step": 27491 }, { "epoch": 1.900587625302454, "grad_norm": 0.7747095945655583, "learning_rate": 6.474177994007957e-08, "loss": 0.2589, "step": 27492 }, { "epoch": 1.900656757690978, "grad_norm": 0.6284167542890374, "learning_rate": 6.465202393198145e-08, "loss": 0.2154, "step": 27493 }, { "epoch": 1.900725890079502, "grad_norm": 0.7483566506571461, "learning_rate": 6.456232977966271e-08, "loss": 0.2941, "step": 27494 }, { "epoch": 1.9007950224680261, "grad_norm": 0.6713744831125282, "learning_rate": 6.447269748424745e-08, "loss": 0.2726, "step": 27495 }, { "epoch": 1.9008641548565501, "grad_norm": 0.6121464045144913, "learning_rate": 6.438312704685922e-08, "loss": 0.2117, "step": 27496 }, { "epoch": 1.9009332872450742, "grad_norm": 0.677973334181006, "learning_rate": 6.429361846862047e-08, "loss": 0.2238, "step": 27497 }, { "epoch": 1.9010024196335984, "grad_norm": 1.2595061762808561, "learning_rate": 6.420417175065363e-08, "loss": 0.3318, "step": 27498 }, { "epoch": 1.9010715520221224, "grad_norm": 0.7686427277526826, "learning_rate": 6.411478689407835e-08, "loss": 0.3183, "step": 27499 }, { "epoch": 1.9011406844106464, "grad_norm": 1.026849954235865, "learning_rate": 6.402546390001652e-08, "loss": 0.2125, "step": 27500 }, { "epoch": 1.9012098167991704, "grad_norm": 0.9579010545499317, "learning_rate": 6.393620276958612e-08, "loss": 0.2614, "step": 27501 }, { "epoch": 1.9012789491876945, "grad_norm": 0.7272851053026304, "learning_rate": 6.384700350390682e-08, "loss": 0.3007, "step": 27502 }, { "epoch": 1.9013480815762185, "grad_norm": 0.6901665200125314, "learning_rate": 6.375786610409663e-08, "loss": 0.2411, "step": 27503 }, { "epoch": 1.9014172139647425, "grad_norm": 0.7662944737279263, "learning_rate": 6.36687905712724e-08, "loss": 0.2057, "step": 27504 }, { "epoch": 1.9014863463532665, "grad_norm": 0.733285459000695, "learning_rate": 6.357977690655048e-08, "loss": 0.3575, "step": 27505 }, { "epoch": 1.9015554787417905, "grad_norm": 0.6870072746237991, "learning_rate": 6.349082511104665e-08, "loss": 0.2553, "step": 27506 }, { "epoch": 1.9016246111303146, "grad_norm": 0.6177602302740893, "learning_rate": 6.340193518587556e-08, "loss": 0.2243, "step": 27507 }, { "epoch": 1.9016937435188386, "grad_norm": 0.7574988699401075, "learning_rate": 6.331310713215132e-08, "loss": 0.2795, "step": 27508 }, { "epoch": 1.9017628759073626, "grad_norm": 0.7072842968766018, "learning_rate": 6.32243409509875e-08, "loss": 0.3343, "step": 27509 }, { "epoch": 1.9018320082958866, "grad_norm": 0.7423348976701332, "learning_rate": 6.313563664349709e-08, "loss": 0.2477, "step": 27510 }, { "epoch": 1.9019011406844106, "grad_norm": 0.6775364854850506, "learning_rate": 6.304699421079086e-08, "loss": 0.2123, "step": 27511 }, { "epoch": 1.9019702730729346, "grad_norm": 0.7181744070972603, "learning_rate": 6.295841365398014e-08, "loss": 0.3207, "step": 27512 }, { "epoch": 1.9020394054614587, "grad_norm": 1.0563897174901635, "learning_rate": 6.286989497417517e-08, "loss": 0.306, "step": 27513 }, { "epoch": 1.9021085378499827, "grad_norm": 0.7484992839657371, "learning_rate": 6.27814381724856e-08, "loss": 0.2489, "step": 27514 }, { "epoch": 1.9021776702385067, "grad_norm": 0.7382116286694396, "learning_rate": 6.269304325001946e-08, "loss": 0.2669, "step": 27515 }, { "epoch": 1.9022468026270307, "grad_norm": 0.6707011351898956, "learning_rate": 6.260471020788472e-08, "loss": 0.3505, "step": 27516 }, { "epoch": 1.9023159350155547, "grad_norm": 0.7211778094305163, "learning_rate": 6.251643904718885e-08, "loss": 0.2369, "step": 27517 }, { "epoch": 1.9023850674040788, "grad_norm": 0.6302807290942519, "learning_rate": 6.242822976903873e-08, "loss": 0.2607, "step": 27518 }, { "epoch": 1.902454199792603, "grad_norm": 0.7005402631709196, "learning_rate": 6.234008237453904e-08, "loss": 0.3268, "step": 27519 }, { "epoch": 1.902523332181127, "grad_norm": 0.7192235227874862, "learning_rate": 6.2251996864795e-08, "loss": 0.2747, "step": 27520 }, { "epoch": 1.902592464569651, "grad_norm": 0.8065022762780394, "learning_rate": 6.216397324090961e-08, "loss": 0.2452, "step": 27521 }, { "epoch": 1.902661596958175, "grad_norm": 0.6709547310430906, "learning_rate": 6.207601150398756e-08, "loss": 0.2174, "step": 27522 }, { "epoch": 1.902730729346699, "grad_norm": 0.6328294392751637, "learning_rate": 6.198811165512963e-08, "loss": 0.2828, "step": 27523 }, { "epoch": 1.902799861735223, "grad_norm": 0.7506614811307473, "learning_rate": 6.190027369543938e-08, "loss": 0.3355, "step": 27524 }, { "epoch": 1.902868994123747, "grad_norm": 0.8378415901698933, "learning_rate": 6.181249762601648e-08, "loss": 0.2308, "step": 27525 }, { "epoch": 1.9029381265122711, "grad_norm": 0.751774952178748, "learning_rate": 6.172478344796174e-08, "loss": 0.306, "step": 27526 }, { "epoch": 1.9030072589007951, "grad_norm": 0.9024352695157258, "learning_rate": 6.163713116237313e-08, "loss": 0.3099, "step": 27527 }, { "epoch": 1.9030763912893192, "grad_norm": 0.8018402226289196, "learning_rate": 6.154954077035091e-08, "loss": 0.2738, "step": 27528 }, { "epoch": 1.9031455236778432, "grad_norm": 0.6540968934364046, "learning_rate": 6.146201227299198e-08, "loss": 0.2045, "step": 27529 }, { "epoch": 1.9032146560663672, "grad_norm": 0.7534241367313216, "learning_rate": 6.137454567139322e-08, "loss": 0.3371, "step": 27530 }, { "epoch": 1.9032837884548912, "grad_norm": 0.6991151826093219, "learning_rate": 6.128714096665157e-08, "loss": 0.2962, "step": 27531 }, { "epoch": 1.9033529208434152, "grad_norm": 0.7461083751310226, "learning_rate": 6.119979815986221e-08, "loss": 0.2633, "step": 27532 }, { "epoch": 1.9034220532319392, "grad_norm": 0.7007250167514277, "learning_rate": 6.111251725211931e-08, "loss": 0.2444, "step": 27533 }, { "epoch": 1.9034911856204633, "grad_norm": 0.7086527101102599, "learning_rate": 6.102529824451753e-08, "loss": 0.3382, "step": 27534 }, { "epoch": 1.9035603180089873, "grad_norm": 0.659909446300062, "learning_rate": 6.093814113814933e-08, "loss": 0.2025, "step": 27535 }, { "epoch": 1.9036294503975113, "grad_norm": 0.700255574102434, "learning_rate": 6.085104593410718e-08, "loss": 0.2252, "step": 27536 }, { "epoch": 1.9036985827860353, "grad_norm": 0.6959295163984878, "learning_rate": 6.076401263348298e-08, "loss": 0.3262, "step": 27537 }, { "epoch": 1.9037677151745593, "grad_norm": 0.7320907203504996, "learning_rate": 6.067704123736751e-08, "loss": 0.323, "step": 27538 }, { "epoch": 1.9038368475630834, "grad_norm": 0.7235952541725613, "learning_rate": 6.059013174685103e-08, "loss": 0.2125, "step": 27539 }, { "epoch": 1.9039059799516074, "grad_norm": 0.6911603775187706, "learning_rate": 6.050328416302209e-08, "loss": 0.2812, "step": 27540 }, { "epoch": 1.9039751123401314, "grad_norm": 0.7201318127315973, "learning_rate": 6.041649848696984e-08, "loss": 0.3488, "step": 27541 }, { "epoch": 1.9040442447286554, "grad_norm": 0.6760193717684848, "learning_rate": 6.032977471978119e-08, "loss": 0.2721, "step": 27542 }, { "epoch": 1.9041133771171794, "grad_norm": 0.6763867656969843, "learning_rate": 6.024311286254413e-08, "loss": 0.2167, "step": 27543 }, { "epoch": 1.9041825095057034, "grad_norm": 0.7240752219143721, "learning_rate": 6.015651291634395e-08, "loss": 0.2824, "step": 27544 }, { "epoch": 1.9042516418942275, "grad_norm": 0.7238306652103179, "learning_rate": 6.00699748822664e-08, "loss": 0.3166, "step": 27545 }, { "epoch": 1.9043207742827515, "grad_norm": 0.7848209261170803, "learning_rate": 5.998349876139619e-08, "loss": 0.2469, "step": 27546 }, { "epoch": 1.9043899066712755, "grad_norm": 0.7227217086519935, "learning_rate": 5.989708455481691e-08, "loss": 0.3269, "step": 27547 }, { "epoch": 1.9044590390597995, "grad_norm": 0.775684350114359, "learning_rate": 5.981073226361155e-08, "loss": 0.3506, "step": 27548 }, { "epoch": 1.9045281714483235, "grad_norm": 0.7408665314247145, "learning_rate": 5.972444188886262e-08, "loss": 0.2509, "step": 27549 }, { "epoch": 1.9045973038368476, "grad_norm": 0.6600564596557018, "learning_rate": 5.963821343165089e-08, "loss": 0.237, "step": 27550 }, { "epoch": 1.9046664362253716, "grad_norm": 0.677042220710313, "learning_rate": 5.955204689305827e-08, "loss": 0.2435, "step": 27551 }, { "epoch": 1.9047355686138956, "grad_norm": 0.720815461695354, "learning_rate": 5.946594227416391e-08, "loss": 0.3014, "step": 27552 }, { "epoch": 1.9048047010024196, "grad_norm": 0.7068436122958272, "learning_rate": 5.9379899576047504e-08, "loss": 0.215, "step": 27553 }, { "epoch": 1.9048738333909436, "grad_norm": 0.6586966270390355, "learning_rate": 5.929391879978707e-08, "loss": 0.2911, "step": 27554 }, { "epoch": 1.9049429657794676, "grad_norm": 0.753459480557752, "learning_rate": 5.9207999946460096e-08, "loss": 0.3555, "step": 27555 }, { "epoch": 1.9050120981679917, "grad_norm": 0.8632285016490026, "learning_rate": 5.912214301714347e-08, "loss": 0.2273, "step": 27556 }, { "epoch": 1.9050812305565157, "grad_norm": 0.6302714960129169, "learning_rate": 5.903634801291358e-08, "loss": 0.193, "step": 27557 }, { "epoch": 1.9051503629450397, "grad_norm": 0.9713813876852359, "learning_rate": 5.895061493484511e-08, "loss": 0.234, "step": 27558 }, { "epoch": 1.9052194953335637, "grad_norm": 0.740026484961676, "learning_rate": 5.8864943784013864e-08, "loss": 0.3393, "step": 27559 }, { "epoch": 1.9052886277220877, "grad_norm": 0.7427836885900171, "learning_rate": 5.877933456149176e-08, "loss": 0.2314, "step": 27560 }, { "epoch": 1.9053577601106118, "grad_norm": 0.658286760425158, "learning_rate": 5.869378726835351e-08, "loss": 0.2745, "step": 27561 }, { "epoch": 1.9054268924991358, "grad_norm": 0.6767440889202931, "learning_rate": 5.8608301905669906e-08, "loss": 0.2779, "step": 27562 }, { "epoch": 1.9054960248876598, "grad_norm": 0.7313502521466696, "learning_rate": 5.852287847451344e-08, "loss": 0.2916, "step": 27563 }, { "epoch": 1.9055651572761838, "grad_norm": 0.6599577212245724, "learning_rate": 5.843751697595379e-08, "loss": 0.1983, "step": 27564 }, { "epoch": 1.9056342896647078, "grad_norm": 0.6904017023962539, "learning_rate": 5.835221741106123e-08, "loss": 0.2136, "step": 27565 }, { "epoch": 1.9057034220532318, "grad_norm": 0.6409116457907004, "learning_rate": 5.82669797809049e-08, "loss": 0.2798, "step": 27566 }, { "epoch": 1.9057725544417559, "grad_norm": 0.6548764886423256, "learning_rate": 5.818180408655338e-08, "loss": 0.2833, "step": 27567 }, { "epoch": 1.9058416868302799, "grad_norm": 0.7058047243683194, "learning_rate": 5.809669032907361e-08, "loss": 0.1754, "step": 27568 }, { "epoch": 1.905910819218804, "grad_norm": 0.7484507148575951, "learning_rate": 5.801163850953306e-08, "loss": 0.3286, "step": 27569 }, { "epoch": 1.905979951607328, "grad_norm": 0.6861707237989515, "learning_rate": 5.792664862899644e-08, "loss": 0.2966, "step": 27570 }, { "epoch": 1.906049083995852, "grad_norm": 0.7050888265316803, "learning_rate": 5.784172068853067e-08, "loss": 0.2165, "step": 27571 }, { "epoch": 1.906118216384376, "grad_norm": 0.7138210609783059, "learning_rate": 5.775685468919823e-08, "loss": 0.2234, "step": 27572 }, { "epoch": 1.9061873487729, "grad_norm": 0.6187653557067039, "learning_rate": 5.767205063206438e-08, "loss": 0.3229, "step": 27573 }, { "epoch": 1.906256481161424, "grad_norm": 0.6823113414459779, "learning_rate": 5.758730851819106e-08, "loss": 0.2602, "step": 27574 }, { "epoch": 1.906325613549948, "grad_norm": 0.733763433186403, "learning_rate": 5.7502628348641285e-08, "loss": 0.2086, "step": 27575 }, { "epoch": 1.906394745938472, "grad_norm": 0.7943388525624862, "learning_rate": 5.7418010124475345e-08, "loss": 0.3047, "step": 27576 }, { "epoch": 1.906463878326996, "grad_norm": 0.7246357555097356, "learning_rate": 5.7333453846754594e-08, "loss": 0.2891, "step": 27577 }, { "epoch": 1.90653301071552, "grad_norm": 0.6749276791558078, "learning_rate": 5.724895951653764e-08, "loss": 0.2363, "step": 27578 }, { "epoch": 1.9066021431040443, "grad_norm": 0.6731433141250416, "learning_rate": 5.716452713488474e-08, "loss": 0.2358, "step": 27579 }, { "epoch": 1.9066712754925683, "grad_norm": 0.7419470068746639, "learning_rate": 5.7080156702853384e-08, "loss": 0.3008, "step": 27580 }, { "epoch": 1.9067404078810923, "grad_norm": 0.723769322036331, "learning_rate": 5.699584822150162e-08, "loss": 0.2183, "step": 27581 }, { "epoch": 1.9068095402696164, "grad_norm": 0.6583066482043292, "learning_rate": 5.691160169188581e-08, "loss": 0.2099, "step": 27582 }, { "epoch": 1.9068786726581404, "grad_norm": 0.7174563565092719, "learning_rate": 5.6827417115061234e-08, "loss": 0.2365, "step": 27583 }, { "epoch": 1.9069478050466644, "grad_norm": 0.7537919927656781, "learning_rate": 5.674329449208371e-08, "loss": 0.333, "step": 27584 }, { "epoch": 1.9070169374351884, "grad_norm": 0.6602445592525058, "learning_rate": 5.6659233824007954e-08, "loss": 0.2355, "step": 27585 }, { "epoch": 1.9070860698237124, "grad_norm": 0.7292109099522103, "learning_rate": 5.657523511188589e-08, "loss": 0.2088, "step": 27586 }, { "epoch": 1.9071552022122364, "grad_norm": 0.7009862897432833, "learning_rate": 5.6491298356771695e-08, "loss": 0.3051, "step": 27587 }, { "epoch": 1.9072243346007605, "grad_norm": 0.7221334937877283, "learning_rate": 5.6407423559716734e-08, "loss": 0.2383, "step": 27588 }, { "epoch": 1.9072934669892845, "grad_norm": 0.6881040155483725, "learning_rate": 5.632361072177295e-08, "loss": 0.2224, "step": 27589 }, { "epoch": 1.9073625993778085, "grad_norm": 0.6233128667090554, "learning_rate": 5.623985984399005e-08, "loss": 0.2105, "step": 27590 }, { "epoch": 1.9074317317663325, "grad_norm": 0.695005417130791, "learning_rate": 5.615617092741832e-08, "loss": 0.3407, "step": 27591 }, { "epoch": 1.9075008641548565, "grad_norm": 0.7377881488459075, "learning_rate": 5.6072543973106355e-08, "loss": 0.2386, "step": 27592 }, { "epoch": 1.9075699965433806, "grad_norm": 0.6222083584153966, "learning_rate": 5.59889789821022e-08, "loss": 0.1988, "step": 27593 }, { "epoch": 1.9076391289319046, "grad_norm": 0.7757963009406535, "learning_rate": 5.590547595545226e-08, "loss": 0.308, "step": 27594 }, { "epoch": 1.9077082613204286, "grad_norm": 0.7621308404862203, "learning_rate": 5.5822034894205125e-08, "loss": 0.2628, "step": 27595 }, { "epoch": 1.9077773937089526, "grad_norm": 0.7478968687397913, "learning_rate": 5.573865579940496e-08, "loss": 0.2256, "step": 27596 }, { "epoch": 1.9078465260974766, "grad_norm": 0.6063203908842151, "learning_rate": 5.56553386720976e-08, "loss": 0.2242, "step": 27597 }, { "epoch": 1.9079156584860006, "grad_norm": 0.906380989258299, "learning_rate": 5.5572083513326655e-08, "loss": 0.3521, "step": 27598 }, { "epoch": 1.9079847908745247, "grad_norm": 0.804841765359499, "learning_rate": 5.548889032413629e-08, "loss": 0.2765, "step": 27599 }, { "epoch": 1.908053923263049, "grad_norm": 0.6719388993926602, "learning_rate": 5.540575910556845e-08, "loss": 0.2299, "step": 27600 }, { "epoch": 1.908123055651573, "grad_norm": 0.7999698278898947, "learning_rate": 5.532268985866562e-08, "loss": 0.299, "step": 27601 }, { "epoch": 1.908192188040097, "grad_norm": 0.7133465314751174, "learning_rate": 5.523968258446866e-08, "loss": 0.3268, "step": 27602 }, { "epoch": 1.908261320428621, "grad_norm": 0.6650650485310087, "learning_rate": 5.5156737284017824e-08, "loss": 0.2091, "step": 27603 }, { "epoch": 1.908330452817145, "grad_norm": 0.6326699080335336, "learning_rate": 5.507385395835341e-08, "loss": 0.2875, "step": 27604 }, { "epoch": 1.908399585205669, "grad_norm": 0.673653584029752, "learning_rate": 5.4991032608512906e-08, "loss": 0.2956, "step": 27605 }, { "epoch": 1.908468717594193, "grad_norm": 0.7365433308547235, "learning_rate": 5.490827323553549e-08, "loss": 0.2942, "step": 27606 }, { "epoch": 1.908537849982717, "grad_norm": 0.6177195848938082, "learning_rate": 5.4825575840457554e-08, "loss": 0.1984, "step": 27607 }, { "epoch": 1.908606982371241, "grad_norm": 0.786843559914124, "learning_rate": 5.474294042431605e-08, "loss": 0.263, "step": 27608 }, { "epoch": 1.908676114759765, "grad_norm": 0.7569788879755988, "learning_rate": 5.466036698814625e-08, "loss": 0.2974, "step": 27609 }, { "epoch": 1.908745247148289, "grad_norm": 0.6890249635075495, "learning_rate": 5.4577855532984005e-08, "loss": 0.2261, "step": 27610 }, { "epoch": 1.908814379536813, "grad_norm": 0.7070939619724672, "learning_rate": 5.449540605986292e-08, "loss": 0.2525, "step": 27611 }, { "epoch": 1.9088835119253371, "grad_norm": 0.6973259233892429, "learning_rate": 5.441301856981607e-08, "loss": 0.3019, "step": 27612 }, { "epoch": 1.9089526443138611, "grad_norm": 0.7334986826045761, "learning_rate": 5.433069306387595e-08, "loss": 0.2995, "step": 27613 }, { "epoch": 1.9090217767023852, "grad_norm": 0.6747987980217549, "learning_rate": 5.424842954307507e-08, "loss": 0.2308, "step": 27614 }, { "epoch": 1.9090909090909092, "grad_norm": 0.6840082643259783, "learning_rate": 5.416622800844373e-08, "loss": 0.2503, "step": 27615 }, { "epoch": 1.9091600414794332, "grad_norm": 0.9085507952401541, "learning_rate": 5.408408846101276e-08, "loss": 0.348, "step": 27616 }, { "epoch": 1.9092291738679572, "grad_norm": 0.7471988505534918, "learning_rate": 5.4002010901810786e-08, "loss": 0.2273, "step": 27617 }, { "epoch": 1.9092983062564812, "grad_norm": 0.6644351685049013, "learning_rate": 5.391999533186809e-08, "loss": 0.1629, "step": 27618 }, { "epoch": 1.9093674386450052, "grad_norm": 0.755608765085129, "learning_rate": 5.3838041752210526e-08, "loss": 0.3441, "step": 27619 }, { "epoch": 1.9094365710335293, "grad_norm": 0.7570781128392365, "learning_rate": 5.375615016386726e-08, "loss": 0.3915, "step": 27620 }, { "epoch": 1.9095057034220533, "grad_norm": 0.6840990729658825, "learning_rate": 5.3674320567863035e-08, "loss": 0.2017, "step": 27621 }, { "epoch": 1.9095748358105773, "grad_norm": 0.6999741751161468, "learning_rate": 5.35925529652237e-08, "loss": 0.2859, "step": 27622 }, { "epoch": 1.9096439681991013, "grad_norm": 0.7292473613020358, "learning_rate": 5.351084735697509e-08, "loss": 0.3731, "step": 27623 }, { "epoch": 1.9097131005876253, "grad_norm": 0.7530352385380362, "learning_rate": 5.342920374414085e-08, "loss": 0.2429, "step": 27624 }, { "epoch": 1.9097822329761494, "grad_norm": 0.7151222644541847, "learning_rate": 5.334762212774347e-08, "loss": 0.2237, "step": 27625 }, { "epoch": 1.9098513653646734, "grad_norm": 0.6800092230394923, "learning_rate": 5.326610250880604e-08, "loss": 0.2835, "step": 27626 }, { "epoch": 1.9099204977531974, "grad_norm": 0.7184229532838995, "learning_rate": 5.318464488834996e-08, "loss": 0.2963, "step": 27627 }, { "epoch": 1.9099896301417214, "grad_norm": 0.7288381296440364, "learning_rate": 5.310324926739718e-08, "loss": 0.2449, "step": 27628 }, { "epoch": 1.9100587625302454, "grad_norm": 0.7073421585609982, "learning_rate": 5.3021915646965796e-08, "loss": 0.2533, "step": 27629 }, { "epoch": 1.9101278949187694, "grad_norm": 0.7244524595817005, "learning_rate": 5.294064402807775e-08, "loss": 0.3144, "step": 27630 }, { "epoch": 1.9101970273072935, "grad_norm": 0.7255292704972127, "learning_rate": 5.2859434411749455e-08, "loss": 0.2961, "step": 27631 }, { "epoch": 1.9102661596958175, "grad_norm": 0.68731313137893, "learning_rate": 5.27782867990001e-08, "loss": 0.2705, "step": 27632 }, { "epoch": 1.9103352920843415, "grad_norm": 0.7140153877845715, "learning_rate": 5.2697201190845536e-08, "loss": 0.2891, "step": 27633 }, { "epoch": 1.9104044244728655, "grad_norm": 0.7165026587067652, "learning_rate": 5.2616177588303285e-08, "loss": 0.3377, "step": 27634 }, { "epoch": 1.9104735568613895, "grad_norm": 0.7676274917546517, "learning_rate": 5.253521599238809e-08, "loss": 0.2713, "step": 27635 }, { "epoch": 1.9105426892499136, "grad_norm": 0.6283594700818436, "learning_rate": 5.245431640411469e-08, "loss": 0.1892, "step": 27636 }, { "epoch": 1.9106118216384376, "grad_norm": 0.6763860950787728, "learning_rate": 5.2373478824496725e-08, "loss": 0.2776, "step": 27637 }, { "epoch": 1.9106809540269616, "grad_norm": 0.6996080644671473, "learning_rate": 5.229270325454838e-08, "loss": 0.2519, "step": 27638 }, { "epoch": 1.9107500864154856, "grad_norm": 0.779160030247049, "learning_rate": 5.2211989695281626e-08, "loss": 0.2657, "step": 27639 }, { "epoch": 1.9108192188040096, "grad_norm": 0.6825212174966699, "learning_rate": 5.213133814770732e-08, "loss": 0.242, "step": 27640 }, { "epoch": 1.9108883511925336, "grad_norm": 0.721892035488095, "learning_rate": 5.2050748612836874e-08, "loss": 0.3814, "step": 27641 }, { "epoch": 1.9109574835810577, "grad_norm": 0.6668285797824195, "learning_rate": 5.197022109168059e-08, "loss": 0.2257, "step": 27642 }, { "epoch": 1.9110266159695817, "grad_norm": 0.6743567404581631, "learning_rate": 5.188975558524711e-08, "loss": 0.1939, "step": 27643 }, { "epoch": 1.9110957483581057, "grad_norm": 0.7293187874848596, "learning_rate": 5.180935209454563e-08, "loss": 0.2898, "step": 27644 }, { "epoch": 1.9111648807466297, "grad_norm": 0.6901486337119067, "learning_rate": 5.172901062058311e-08, "loss": 0.2854, "step": 27645 }, { "epoch": 1.9112340131351537, "grad_norm": 0.6155779091320036, "learning_rate": 5.1648731164367086e-08, "loss": 0.1985, "step": 27646 }, { "epoch": 1.9113031455236777, "grad_norm": 0.6917956610620432, "learning_rate": 5.1568513726903416e-08, "loss": 0.2208, "step": 27647 }, { "epoch": 1.9113722779122018, "grad_norm": 0.7197092803599785, "learning_rate": 5.1488358309197406e-08, "loss": 0.3222, "step": 27648 }, { "epoch": 1.9114414103007258, "grad_norm": 0.6748742584598154, "learning_rate": 5.1408264912253813e-08, "loss": 0.2409, "step": 27649 }, { "epoch": 1.9115105426892498, "grad_norm": 0.6827000298798622, "learning_rate": 5.132823353707683e-08, "loss": 0.2482, "step": 27650 }, { "epoch": 1.9115796750777738, "grad_norm": 0.7421825779755926, "learning_rate": 5.124826418466844e-08, "loss": 0.2855, "step": 27651 }, { "epoch": 1.9116488074662978, "grad_norm": 0.6403142214066291, "learning_rate": 5.1168356856032275e-08, "loss": 0.315, "step": 27652 }, { "epoch": 1.9117179398548219, "grad_norm": 0.6892189021343845, "learning_rate": 5.1088511552169206e-08, "loss": 0.2067, "step": 27653 }, { "epoch": 1.9117870722433459, "grad_norm": 0.6644576729431589, "learning_rate": 5.100872827407954e-08, "loss": 0.2292, "step": 27654 }, { "epoch": 1.91185620463187, "grad_norm": 0.7305160454578605, "learning_rate": 5.09290070227636e-08, "loss": 0.3305, "step": 27655 }, { "epoch": 1.911925337020394, "grad_norm": 0.7160436901537953, "learning_rate": 5.084934779922113e-08, "loss": 0.2342, "step": 27656 }, { "epoch": 1.911994469408918, "grad_norm": 0.6608584688056458, "learning_rate": 5.076975060444911e-08, "loss": 0.2305, "step": 27657 }, { "epoch": 1.912063601797442, "grad_norm": 1.912308231765292, "learning_rate": 5.069021543944619e-08, "loss": 0.3029, "step": 27658 }, { "epoch": 1.912132734185966, "grad_norm": 0.7546066339446175, "learning_rate": 5.061074230520935e-08, "loss": 0.2851, "step": 27659 }, { "epoch": 1.91220186657449, "grad_norm": 0.7095577408271052, "learning_rate": 5.05313312027339e-08, "loss": 0.2104, "step": 27660 }, { "epoch": 1.9122709989630142, "grad_norm": 0.634525567519, "learning_rate": 5.045198213301572e-08, "loss": 0.2518, "step": 27661 }, { "epoch": 1.9123401313515382, "grad_norm": 0.7176023093200516, "learning_rate": 5.037269509704901e-08, "loss": 0.3684, "step": 27662 }, { "epoch": 1.9124092637400623, "grad_norm": 0.9785429598289647, "learning_rate": 5.0293470095827415e-08, "loss": 0.3112, "step": 27663 }, { "epoch": 1.9124783961285863, "grad_norm": 0.7234191933254063, "learning_rate": 5.0214307130344584e-08, "loss": 0.2137, "step": 27664 }, { "epoch": 1.9125475285171103, "grad_norm": 0.7363813946182436, "learning_rate": 5.0135206201591404e-08, "loss": 0.2846, "step": 27665 }, { "epoch": 1.9126166609056343, "grad_norm": 0.6923344910274342, "learning_rate": 5.00561673105604e-08, "loss": 0.3091, "step": 27666 }, { "epoch": 1.9126857932941583, "grad_norm": 0.6896063142805389, "learning_rate": 4.997719045824134e-08, "loss": 0.2299, "step": 27667 }, { "epoch": 1.9127549256826823, "grad_norm": 0.6198260566020345, "learning_rate": 4.98982756456251e-08, "loss": 0.2227, "step": 27668 }, { "epoch": 1.9128240580712064, "grad_norm": 0.7515804617518439, "learning_rate": 4.981942287369979e-08, "loss": 0.3469, "step": 27669 }, { "epoch": 1.9128931904597304, "grad_norm": 0.7572740888716186, "learning_rate": 4.974063214345404e-08, "loss": 0.2337, "step": 27670 }, { "epoch": 1.9129623228482544, "grad_norm": 0.6680160831655305, "learning_rate": 4.9661903455875406e-08, "loss": 0.2146, "step": 27671 }, { "epoch": 1.9130314552367784, "grad_norm": 0.7482828147501436, "learning_rate": 4.958323681195032e-08, "loss": 0.232, "step": 27672 }, { "epoch": 1.9131005876253024, "grad_norm": 0.7412678429425126, "learning_rate": 4.950463221266522e-08, "loss": 0.2988, "step": 27673 }, { "epoch": 1.9131697200138265, "grad_norm": 0.7910537804265966, "learning_rate": 4.942608965900486e-08, "loss": 0.2364, "step": 27674 }, { "epoch": 1.9132388524023505, "grad_norm": 0.6598601194470783, "learning_rate": 4.9347609151953466e-08, "loss": 0.1925, "step": 27675 }, { "epoch": 1.9133079847908745, "grad_norm": 0.7152244833413213, "learning_rate": 4.926919069249525e-08, "loss": 0.2555, "step": 27676 }, { "epoch": 1.9133771171793985, "grad_norm": 0.7267187046833532, "learning_rate": 4.919083428161331e-08, "loss": 0.3655, "step": 27677 }, { "epoch": 1.9134462495679225, "grad_norm": 0.7007570135725895, "learning_rate": 4.9112539920288525e-08, "loss": 0.2769, "step": 27678 }, { "epoch": 1.9135153819564465, "grad_norm": 0.8682390620488668, "learning_rate": 4.903430760950234e-08, "loss": 0.2832, "step": 27679 }, { "epoch": 1.9135845143449706, "grad_norm": 0.7111664997430662, "learning_rate": 4.8956137350236185e-08, "loss": 0.3459, "step": 27680 }, { "epoch": 1.9136536467334948, "grad_norm": 0.6853294444951517, "learning_rate": 4.887802914346984e-08, "loss": 0.2351, "step": 27681 }, { "epoch": 1.9137227791220188, "grad_norm": 0.7136559201040805, "learning_rate": 4.879998299018085e-08, "loss": 0.1659, "step": 27682 }, { "epoch": 1.9137919115105428, "grad_norm": 0.7417038166763229, "learning_rate": 4.872199889134899e-08, "loss": 0.2336, "step": 27683 }, { "epoch": 1.9138610438990669, "grad_norm": 0.7116864506293218, "learning_rate": 4.8644076847950694e-08, "loss": 0.2987, "step": 27684 }, { "epoch": 1.9139301762875909, "grad_norm": 0.7358338911077547, "learning_rate": 4.856621686096241e-08, "loss": 0.2353, "step": 27685 }, { "epoch": 1.913999308676115, "grad_norm": 0.7918384424263862, "learning_rate": 4.848841893136058e-08, "loss": 0.2666, "step": 27686 }, { "epoch": 1.914068441064639, "grad_norm": 0.7761848435279811, "learning_rate": 4.8410683060120536e-08, "loss": 0.3815, "step": 27687 }, { "epoch": 1.914137573453163, "grad_norm": 0.7256443992173827, "learning_rate": 4.833300924821538e-08, "loss": 0.2772, "step": 27688 }, { "epoch": 1.914206705841687, "grad_norm": 0.6863383078412916, "learning_rate": 4.825539749661989e-08, "loss": 0.2533, "step": 27689 }, { "epoch": 1.914275838230211, "grad_norm": 0.8011069652565431, "learning_rate": 4.817784780630552e-08, "loss": 0.1928, "step": 27690 }, { "epoch": 1.914344970618735, "grad_norm": 0.7488410556006544, "learning_rate": 4.8100360178245374e-08, "loss": 0.3163, "step": 27691 }, { "epoch": 1.914414103007259, "grad_norm": 0.711740844370114, "learning_rate": 4.8022934613410346e-08, "loss": 0.2265, "step": 27692 }, { "epoch": 1.914483235395783, "grad_norm": 0.6582095399212659, "learning_rate": 4.7945571112770205e-08, "loss": 0.1853, "step": 27693 }, { "epoch": 1.914552367784307, "grad_norm": 0.6733590901285352, "learning_rate": 4.786826967729474e-08, "loss": 0.337, "step": 27694 }, { "epoch": 1.914621500172831, "grad_norm": 0.6828732405207093, "learning_rate": 4.7791030307953734e-08, "loss": 0.3146, "step": 27695 }, { "epoch": 1.914690632561355, "grad_norm": 0.7638245974340846, "learning_rate": 4.7713853005714186e-08, "loss": 0.2922, "step": 27696 }, { "epoch": 1.914759764949879, "grad_norm": 0.6748871194231844, "learning_rate": 4.7636737771544204e-08, "loss": 0.229, "step": 27697 }, { "epoch": 1.9148288973384031, "grad_norm": 0.7255718284160118, "learning_rate": 4.755968460640914e-08, "loss": 0.3322, "step": 27698 }, { "epoch": 1.9148980297269271, "grad_norm": 0.6850488701446507, "learning_rate": 4.748269351127599e-08, "loss": 0.2839, "step": 27699 }, { "epoch": 1.9149671621154511, "grad_norm": 0.665564903994977, "learning_rate": 4.740576448710843e-08, "loss": 0.193, "step": 27700 }, { "epoch": 1.9150362945039752, "grad_norm": 0.6901835838865454, "learning_rate": 4.7328897534871796e-08, "loss": 0.2619, "step": 27701 }, { "epoch": 1.9151054268924992, "grad_norm": 1.4675020491774353, "learning_rate": 4.725209265552866e-08, "loss": 0.2952, "step": 27702 }, { "epoch": 1.9151745592810232, "grad_norm": 0.6489654264697208, "learning_rate": 4.7175349850042686e-08, "loss": 0.2271, "step": 27703 }, { "epoch": 1.9152436916695472, "grad_norm": 0.7551973977913983, "learning_rate": 4.709866911937422e-08, "loss": 0.2054, "step": 27704 }, { "epoch": 1.9153128240580712, "grad_norm": 0.7219492665971574, "learning_rate": 4.702205046448582e-08, "loss": 0.302, "step": 27705 }, { "epoch": 1.9153819564465953, "grad_norm": 0.793769198303925, "learning_rate": 4.694549388633618e-08, "loss": 0.2662, "step": 27706 }, { "epoch": 1.9154510888351193, "grad_norm": 0.6444067453357669, "learning_rate": 4.6868999385886185e-08, "loss": 0.2065, "step": 27707 }, { "epoch": 1.9155202212236433, "grad_norm": 0.6942853615663831, "learning_rate": 4.67925669640934e-08, "loss": 0.317, "step": 27708 }, { "epoch": 1.9155893536121673, "grad_norm": 0.7321960725761698, "learning_rate": 4.671619662191706e-08, "loss": 0.3207, "step": 27709 }, { "epoch": 1.9156584860006913, "grad_norm": 0.6813548684616985, "learning_rate": 4.663988836031363e-08, "loss": 0.1843, "step": 27710 }, { "epoch": 1.9157276183892153, "grad_norm": 0.7004768832609786, "learning_rate": 4.656364218023956e-08, "loss": 0.2023, "step": 27711 }, { "epoch": 1.9157967507777394, "grad_norm": 0.7640034604970234, "learning_rate": 4.6487458082650204e-08, "loss": 0.3386, "step": 27712 }, { "epoch": 1.9158658831662634, "grad_norm": 0.7529732140987874, "learning_rate": 4.6411336068500903e-08, "loss": 0.2381, "step": 27713 }, { "epoch": 1.9159350155547874, "grad_norm": 0.7507204268173526, "learning_rate": 4.633527613874533e-08, "loss": 0.2481, "step": 27714 }, { "epoch": 1.9160041479433114, "grad_norm": 0.7245357492866075, "learning_rate": 4.625927829433663e-08, "loss": 0.2221, "step": 27715 }, { "epoch": 1.9160732803318354, "grad_norm": 0.7177853208815571, "learning_rate": 4.618334253622736e-08, "loss": 0.3555, "step": 27716 }, { "epoch": 1.9161424127203595, "grad_norm": 0.7378112550216174, "learning_rate": 4.6107468865369544e-08, "loss": 0.1989, "step": 27717 }, { "epoch": 1.9162115451088835, "grad_norm": 0.6460067198144683, "learning_rate": 4.603165728271408e-08, "loss": 0.228, "step": 27718 }, { "epoch": 1.9162806774974075, "grad_norm": 0.7779852893601568, "learning_rate": 4.5955907789211327e-08, "loss": 0.3694, "step": 27719 }, { "epoch": 1.9163498098859315, "grad_norm": 0.6781641308755177, "learning_rate": 4.588022038580997e-08, "loss": 0.2732, "step": 27720 }, { "epoch": 1.9164189422744555, "grad_norm": 0.6694636276830399, "learning_rate": 4.580459507345925e-08, "loss": 0.2312, "step": 27721 }, { "epoch": 1.9164880746629795, "grad_norm": 0.7142757679668774, "learning_rate": 4.572903185310673e-08, "loss": 0.2615, "step": 27722 }, { "epoch": 1.9165572070515036, "grad_norm": 0.7983351878614987, "learning_rate": 4.5653530725699444e-08, "loss": 0.3619, "step": 27723 }, { "epoch": 1.9166263394400276, "grad_norm": 0.7609226081175029, "learning_rate": 4.5578091692183855e-08, "loss": 0.218, "step": 27724 }, { "epoch": 1.9166954718285516, "grad_norm": 0.6043517554518335, "learning_rate": 4.550271475350532e-08, "loss": 0.2257, "step": 27725 }, { "epoch": 1.9167646042170756, "grad_norm": 0.6943905609119745, "learning_rate": 4.5427399910608625e-08, "loss": 0.2852, "step": 27726 }, { "epoch": 1.9168337366055996, "grad_norm": 0.7447642835038107, "learning_rate": 4.5352147164437476e-08, "loss": 0.3479, "step": 27727 }, { "epoch": 1.9169028689941237, "grad_norm": 0.6801930501938818, "learning_rate": 4.5276956515935e-08, "loss": 0.2144, "step": 27728 }, { "epoch": 1.9169720013826477, "grad_norm": 2.787325633693919, "learning_rate": 4.5201827966043774e-08, "loss": 0.232, "step": 27729 }, { "epoch": 1.9170411337711717, "grad_norm": 0.7153861844272595, "learning_rate": 4.512676151570583e-08, "loss": 0.3195, "step": 27730 }, { "epoch": 1.9171102661596957, "grad_norm": 0.7822915549658611, "learning_rate": 4.505175716586152e-08, "loss": 0.2584, "step": 27731 }, { "epoch": 1.9171793985482197, "grad_norm": 0.6346811569936953, "learning_rate": 4.497681491745065e-08, "loss": 0.2664, "step": 27732 }, { "epoch": 1.9172485309367437, "grad_norm": 0.7222656455415917, "learning_rate": 4.490193477141303e-08, "loss": 0.274, "step": 27733 }, { "epoch": 1.9173176633252678, "grad_norm": 0.7638228550446261, "learning_rate": 4.482711672868678e-08, "loss": 0.406, "step": 27734 }, { "epoch": 1.9173867957137918, "grad_norm": 0.6545755682944844, "learning_rate": 4.47523607902095e-08, "loss": 0.2699, "step": 27735 }, { "epoch": 1.9174559281023158, "grad_norm": 0.729272955526884, "learning_rate": 4.4677666956918774e-08, "loss": 0.2009, "step": 27736 }, { "epoch": 1.9175250604908398, "grad_norm": 0.7752696450194361, "learning_rate": 4.460303522974996e-08, "loss": 0.3296, "step": 27737 }, { "epoch": 1.9175941928793638, "grad_norm": 0.7188166019800156, "learning_rate": 4.452846560963897e-08, "loss": 0.2463, "step": 27738 }, { "epoch": 1.9176633252678879, "grad_norm": 0.6858424624248973, "learning_rate": 4.445395809752007e-08, "loss": 0.2393, "step": 27739 }, { "epoch": 1.9177324576564119, "grad_norm": 0.723524055712359, "learning_rate": 4.43795126943275e-08, "loss": 0.2683, "step": 27740 }, { "epoch": 1.9178015900449359, "grad_norm": 0.7011761236657185, "learning_rate": 4.430512940099385e-08, "loss": 0.3673, "step": 27741 }, { "epoch": 1.9178707224334601, "grad_norm": 0.894527323463844, "learning_rate": 4.423080821845116e-08, "loss": 0.2172, "step": 27742 }, { "epoch": 1.9179398548219841, "grad_norm": 0.6909285592488169, "learning_rate": 4.415654914763146e-08, "loss": 0.2594, "step": 27743 }, { "epoch": 1.9180089872105082, "grad_norm": 0.7129932284929849, "learning_rate": 4.408235218946621e-08, "loss": 0.2543, "step": 27744 }, { "epoch": 1.9180781195990322, "grad_norm": 0.7287482046964292, "learning_rate": 4.4008217344883585e-08, "loss": 0.3052, "step": 27745 }, { "epoch": 1.9181472519875562, "grad_norm": 0.7091306976149935, "learning_rate": 4.3934144614813936e-08, "loss": 0.2535, "step": 27746 }, { "epoch": 1.9182163843760802, "grad_norm": 0.7499225388974399, "learning_rate": 4.386013400018485e-08, "loss": 0.2579, "step": 27747 }, { "epoch": 1.9182855167646042, "grad_norm": 0.6939250370093402, "learning_rate": 4.378618550192504e-08, "loss": 0.3178, "step": 27748 }, { "epoch": 1.9183546491531283, "grad_norm": 0.8864024235885246, "learning_rate": 4.371229912096042e-08, "loss": 0.193, "step": 27749 }, { "epoch": 1.9184237815416523, "grad_norm": 0.7824923884303263, "learning_rate": 4.363847485821693e-08, "loss": 0.2521, "step": 27750 }, { "epoch": 1.9184929139301763, "grad_norm": 0.7404844912265024, "learning_rate": 4.356471271461993e-08, "loss": 0.2902, "step": 27751 }, { "epoch": 1.9185620463187003, "grad_norm": 0.7108650045054389, "learning_rate": 4.349101269109479e-08, "loss": 0.2863, "step": 27752 }, { "epoch": 1.9186311787072243, "grad_norm": 0.6750274797570615, "learning_rate": 4.3417374788564114e-08, "loss": 0.2252, "step": 27753 }, { "epoch": 1.9187003110957483, "grad_norm": 0.7146101484619671, "learning_rate": 4.33437990079516e-08, "loss": 0.2092, "step": 27754 }, { "epoch": 1.9187694434842724, "grad_norm": 0.6969415256667288, "learning_rate": 4.3270285350178744e-08, "loss": 0.3029, "step": 27755 }, { "epoch": 1.9188385758727964, "grad_norm": 0.6819888931834688, "learning_rate": 4.3196833816167016e-08, "loss": 0.2474, "step": 27756 }, { "epoch": 1.9189077082613204, "grad_norm": 0.6449968368504077, "learning_rate": 4.31234444068368e-08, "loss": 0.2455, "step": 27757 }, { "epoch": 1.9189768406498444, "grad_norm": 0.7727848159751378, "learning_rate": 4.3050117123109025e-08, "loss": 0.3234, "step": 27758 }, { "epoch": 1.9190459730383684, "grad_norm": 0.7445404416759979, "learning_rate": 4.297685196590184e-08, "loss": 0.3259, "step": 27759 }, { "epoch": 1.9191151054268925, "grad_norm": 0.8968099147436381, "learning_rate": 4.2903648936133414e-08, "loss": 0.2099, "step": 27760 }, { "epoch": 1.9191842378154165, "grad_norm": 0.7338456288429347, "learning_rate": 4.2830508034721327e-08, "loss": 0.2181, "step": 27761 }, { "epoch": 1.9192533702039407, "grad_norm": 0.7566280991938832, "learning_rate": 4.2757429262583196e-08, "loss": 0.3471, "step": 27762 }, { "epoch": 1.9193225025924647, "grad_norm": 0.6827095029983089, "learning_rate": 4.2684412620633275e-08, "loss": 0.294, "step": 27763 }, { "epoch": 1.9193916349809887, "grad_norm": 0.7498248527630826, "learning_rate": 4.2611458109787504e-08, "loss": 0.2158, "step": 27764 }, { "epoch": 1.9194607673695128, "grad_norm": 0.6575661702171836, "learning_rate": 4.2538565730960156e-08, "loss": 0.254, "step": 27765 }, { "epoch": 1.9195298997580368, "grad_norm": 0.699253377834323, "learning_rate": 4.24657354850655e-08, "loss": 0.2962, "step": 27766 }, { "epoch": 1.9195990321465608, "grad_norm": 0.6705788958862914, "learning_rate": 4.239296737301501e-08, "loss": 0.202, "step": 27767 }, { "epoch": 1.9196681645350848, "grad_norm": 0.7143546080804255, "learning_rate": 4.232026139572187e-08, "loss": 0.2446, "step": 27768 }, { "epoch": 1.9197372969236088, "grad_norm": 0.6695328968029597, "learning_rate": 4.224761755409701e-08, "loss": 0.2908, "step": 27769 }, { "epoch": 1.9198064293121329, "grad_norm": 0.7319877741389023, "learning_rate": 4.217503584905025e-08, "loss": 0.3406, "step": 27770 }, { "epoch": 1.9198755617006569, "grad_norm": 0.6695198988679243, "learning_rate": 4.210251628149198e-08, "loss": 0.1991, "step": 27771 }, { "epoch": 1.919944694089181, "grad_norm": 0.6706362248180244, "learning_rate": 4.203005885233091e-08, "loss": 0.2534, "step": 27772 }, { "epoch": 1.920013826477705, "grad_norm": 0.665332410856056, "learning_rate": 4.195766356247521e-08, "loss": 0.2718, "step": 27773 }, { "epoch": 1.920082958866229, "grad_norm": 0.7245855098342633, "learning_rate": 4.188533041283249e-08, "loss": 0.2858, "step": 27774 }, { "epoch": 1.920152091254753, "grad_norm": 0.6919505343843526, "learning_rate": 4.1813059404308686e-08, "loss": 0.1728, "step": 27775 }, { "epoch": 1.920221223643277, "grad_norm": 0.7200065632824966, "learning_rate": 4.17408505378103e-08, "loss": 0.2242, "step": 27776 }, { "epoch": 1.920290356031801, "grad_norm": 0.7211485628269536, "learning_rate": 4.166870381424104e-08, "loss": 0.3185, "step": 27777 }, { "epoch": 1.920359488420325, "grad_norm": 0.705711357933727, "learning_rate": 4.159661923450686e-08, "loss": 0.2271, "step": 27778 }, { "epoch": 1.920428620808849, "grad_norm": 0.7211681396719711, "learning_rate": 4.152459679950982e-08, "loss": 0.2554, "step": 27779 }, { "epoch": 1.920497753197373, "grad_norm": 0.6734919305161938, "learning_rate": 4.1452636510153634e-08, "loss": 0.3522, "step": 27780 }, { "epoch": 1.920566885585897, "grad_norm": 0.7601921737776064, "learning_rate": 4.1380738367339244e-08, "loss": 0.2458, "step": 27781 }, { "epoch": 1.920636017974421, "grad_norm": 0.636733127361534, "learning_rate": 4.130890237196872e-08, "loss": 0.2132, "step": 27782 }, { "epoch": 1.920705150362945, "grad_norm": 0.6996828337699671, "learning_rate": 4.123712852494188e-08, "loss": 0.2282, "step": 27783 }, { "epoch": 1.920774282751469, "grad_norm": 0.8102076259753408, "learning_rate": 4.116541682715802e-08, "loss": 0.3734, "step": 27784 }, { "epoch": 1.9208434151399931, "grad_norm": 0.6462617922496343, "learning_rate": 4.109376727951642e-08, "loss": 0.1572, "step": 27785 }, { "epoch": 1.9209125475285171, "grad_norm": 0.6843580582299399, "learning_rate": 4.102217988291468e-08, "loss": 0.226, "step": 27786 }, { "epoch": 1.9209816799170412, "grad_norm": 0.7039630130682601, "learning_rate": 4.095065463825043e-08, "loss": 0.3368, "step": 27787 }, { "epoch": 1.9210508123055652, "grad_norm": 0.6214653333326654, "learning_rate": 4.087919154642017e-08, "loss": 0.272, "step": 27788 }, { "epoch": 1.9211199446940892, "grad_norm": 0.6275658941309955, "learning_rate": 4.0807790608318744e-08, "loss": 0.2047, "step": 27789 }, { "epoch": 1.9211890770826132, "grad_norm": 0.6975867708523648, "learning_rate": 4.0736451824842097e-08, "loss": 0.2688, "step": 27790 }, { "epoch": 1.9212582094711372, "grad_norm": 0.6810806378681684, "learning_rate": 4.06651751968834e-08, "loss": 0.309, "step": 27791 }, { "epoch": 1.9213273418596613, "grad_norm": 0.6462593318900868, "learning_rate": 4.059396072533639e-08, "loss": 0.2222, "step": 27792 }, { "epoch": 1.9213964742481853, "grad_norm": 0.7999761435884231, "learning_rate": 4.0522808411093687e-08, "loss": 0.2438, "step": 27793 }, { "epoch": 1.9214656066367093, "grad_norm": 0.8755605638752234, "learning_rate": 4.0451718255047344e-08, "loss": 0.2839, "step": 27794 }, { "epoch": 1.9215347390252333, "grad_norm": 0.7185702128483548, "learning_rate": 4.0380690258087217e-08, "loss": 0.2272, "step": 27795 }, { "epoch": 1.9216038714137573, "grad_norm": 0.7968308413737257, "learning_rate": 4.030972442110537e-08, "loss": 0.2465, "step": 27796 }, { "epoch": 1.9216730038022813, "grad_norm": 0.7081633517511691, "learning_rate": 4.0238820744989415e-08, "loss": 0.2772, "step": 27797 }, { "epoch": 1.9217421361908054, "grad_norm": 0.6682779697101036, "learning_rate": 4.016797923062921e-08, "loss": 0.3437, "step": 27798 }, { "epoch": 1.9218112685793294, "grad_norm": 0.7007606222378986, "learning_rate": 4.0097199878911805e-08, "loss": 0.2772, "step": 27799 }, { "epoch": 1.9218804009678534, "grad_norm": 0.8579397089717488, "learning_rate": 4.002648269072484e-08, "loss": 0.1901, "step": 27800 }, { "epoch": 1.9219495333563774, "grad_norm": 0.6378697530005256, "learning_rate": 3.995582766695427e-08, "loss": 0.3012, "step": 27801 }, { "epoch": 1.9220186657449014, "grad_norm": 0.7088240706514509, "learning_rate": 3.988523480848605e-08, "loss": 0.3843, "step": 27802 }, { "epoch": 1.9220877981334255, "grad_norm": 0.6900124100117847, "learning_rate": 3.981470411620447e-08, "loss": 0.1974, "step": 27803 }, { "epoch": 1.9221569305219495, "grad_norm": 0.7151877362661954, "learning_rate": 3.974423559099383e-08, "loss": 0.2609, "step": 27804 }, { "epoch": 1.9222260629104735, "grad_norm": 0.6913577917473623, "learning_rate": 3.96738292337373e-08, "loss": 0.3265, "step": 27805 }, { "epoch": 1.9222951952989975, "grad_norm": 0.7415262099176816, "learning_rate": 3.960348504531697e-08, "loss": 0.2554, "step": 27806 }, { "epoch": 1.9223643276875215, "grad_norm": 0.7543781666167816, "learning_rate": 3.9533203026614896e-08, "loss": 0.218, "step": 27807 }, { "epoch": 1.9224334600760455, "grad_norm": 0.7669751353925793, "learning_rate": 3.946298317851149e-08, "loss": 0.2739, "step": 27808 }, { "epoch": 1.9225025924645696, "grad_norm": 0.7147703897246676, "learning_rate": 3.939282550188772e-08, "loss": 0.3643, "step": 27809 }, { "epoch": 1.9225717248530936, "grad_norm": 0.6764081235760028, "learning_rate": 3.9322729997621764e-08, "loss": 0.2949, "step": 27810 }, { "epoch": 1.9226408572416176, "grad_norm": 0.7298528747291099, "learning_rate": 3.925269666659293e-08, "loss": 0.2461, "step": 27811 }, { "epoch": 1.9227099896301416, "grad_norm": 0.7121284851735045, "learning_rate": 3.918272550967883e-08, "loss": 0.3293, "step": 27812 }, { "epoch": 1.9227791220186656, "grad_norm": 0.8340587039941137, "learning_rate": 3.911281652775545e-08, "loss": 0.3346, "step": 27813 }, { "epoch": 1.9228482544071896, "grad_norm": 0.6428512632930967, "learning_rate": 3.904296972170041e-08, "loss": 0.2273, "step": 27814 }, { "epoch": 1.9229173867957137, "grad_norm": 0.6878596593379271, "learning_rate": 3.897318509238912e-08, "loss": 0.2542, "step": 27815 }, { "epoch": 1.9229865191842377, "grad_norm": 0.8016502233510618, "learning_rate": 3.890346264069478e-08, "loss": 0.3544, "step": 27816 }, { "epoch": 1.9230556515727617, "grad_norm": 0.715324567435577, "learning_rate": 3.883380236749224e-08, "loss": 0.2329, "step": 27817 }, { "epoch": 1.9231247839612857, "grad_norm": 0.6314864293316403, "learning_rate": 3.876420427365468e-08, "loss": 0.2018, "step": 27818 }, { "epoch": 1.9231939163498097, "grad_norm": 0.7272891520497905, "learning_rate": 3.8694668360053645e-08, "loss": 0.3269, "step": 27819 }, { "epoch": 1.9232630487383338, "grad_norm": 1.1031893885370847, "learning_rate": 3.8625194627561204e-08, "loss": 0.2917, "step": 27820 }, { "epoch": 1.9233321811268578, "grad_norm": 0.6842356132735055, "learning_rate": 3.8555783077048325e-08, "loss": 0.2841, "step": 27821 }, { "epoch": 1.9234013135153818, "grad_norm": 0.7817805887509599, "learning_rate": 3.8486433709384317e-08, "loss": 0.2799, "step": 27822 }, { "epoch": 1.923470445903906, "grad_norm": 0.7173673223013858, "learning_rate": 3.8417146525438485e-08, "loss": 0.3381, "step": 27823 }, { "epoch": 1.92353957829243, "grad_norm": 0.7507710699362311, "learning_rate": 3.834792152607958e-08, "loss": 0.2129, "step": 27824 }, { "epoch": 1.923608710680954, "grad_norm": 0.6088310281238501, "learning_rate": 3.8278758712175234e-08, "loss": 0.2083, "step": 27825 }, { "epoch": 1.923677843069478, "grad_norm": 0.7631751608587446, "learning_rate": 3.820965808459198e-08, "loss": 0.3319, "step": 27826 }, { "epoch": 1.923746975458002, "grad_norm": 0.6843560577874617, "learning_rate": 3.814061964419524e-08, "loss": 0.3101, "step": 27827 }, { "epoch": 1.9238161078465261, "grad_norm": 0.6695172944893543, "learning_rate": 3.8071643391851544e-08, "loss": 0.2076, "step": 27828 }, { "epoch": 1.9238852402350501, "grad_norm": 0.6903311244734085, "learning_rate": 3.800272932842519e-08, "loss": 0.2491, "step": 27829 }, { "epoch": 1.9239543726235742, "grad_norm": 0.7281376454075406, "learning_rate": 3.7933877454779386e-08, "loss": 0.3758, "step": 27830 }, { "epoch": 1.9240235050120982, "grad_norm": 0.7109735228065124, "learning_rate": 3.7865087771776776e-08, "loss": 0.2649, "step": 27831 }, { "epoch": 1.9240926374006222, "grad_norm": 0.6561114612944238, "learning_rate": 3.7796360280280554e-08, "loss": 0.2301, "step": 27832 }, { "epoch": 1.9241617697891462, "grad_norm": 0.6705661116850034, "learning_rate": 3.772769498115114e-08, "loss": 0.227, "step": 27833 }, { "epoch": 1.9242309021776702, "grad_norm": 0.7125386187777559, "learning_rate": 3.765909187524952e-08, "loss": 0.3287, "step": 27834 }, { "epoch": 1.9243000345661943, "grad_norm": 0.6234406702709576, "learning_rate": 3.7590550963435554e-08, "loss": 0.2258, "step": 27835 }, { "epoch": 1.9243691669547183, "grad_norm": 0.8362342411805567, "learning_rate": 3.752207224656801e-08, "loss": 0.2159, "step": 27836 }, { "epoch": 1.9244382993432423, "grad_norm": 1.5931681968898734, "learning_rate": 3.7453655725505635e-08, "loss": 0.3655, "step": 27837 }, { "epoch": 1.9245074317317663, "grad_norm": 0.6540789896065123, "learning_rate": 3.7385301401105524e-08, "loss": 0.2477, "step": 27838 }, { "epoch": 1.9245765641202903, "grad_norm": 0.8651375361176077, "learning_rate": 3.731700927422477e-08, "loss": 0.241, "step": 27839 }, { "epoch": 1.9246456965088143, "grad_norm": 0.724943785262848, "learning_rate": 3.724877934571824e-08, "loss": 0.224, "step": 27840 }, { "epoch": 1.9247148288973384, "grad_norm": 0.7581573104897493, "learning_rate": 3.718061161644249e-08, "loss": 0.316, "step": 27841 }, { "epoch": 1.9247839612858624, "grad_norm": 0.8638482956820689, "learning_rate": 3.7112506087251253e-08, "loss": 0.2237, "step": 27842 }, { "epoch": 1.9248530936743864, "grad_norm": 0.6208958207863937, "learning_rate": 3.7044462758997755e-08, "loss": 0.2406, "step": 27843 }, { "epoch": 1.9249222260629106, "grad_norm": 0.6848401471530846, "learning_rate": 3.697648163253464e-08, "loss": 0.3048, "step": 27844 }, { "epoch": 1.9249913584514347, "grad_norm": 0.7619812827624379, "learning_rate": 3.690856270871512e-08, "loss": 0.2799, "step": 27845 }, { "epoch": 1.9250604908399587, "grad_norm": 0.82838277688505, "learning_rate": 3.6840705988389624e-08, "loss": 0.2478, "step": 27846 }, { "epoch": 1.9251296232284827, "grad_norm": 0.7089385554949277, "learning_rate": 3.677291147240914e-08, "loss": 0.2351, "step": 27847 }, { "epoch": 1.9251987556170067, "grad_norm": 0.6760351127425162, "learning_rate": 3.670517916162242e-08, "loss": 0.3551, "step": 27848 }, { "epoch": 1.9252678880055307, "grad_norm": 0.7299065536454279, "learning_rate": 3.66375090568788e-08, "loss": 0.2941, "step": 27849 }, { "epoch": 1.9253370203940547, "grad_norm": 0.6941628536115683, "learning_rate": 3.656990115902648e-08, "loss": 0.1736, "step": 27850 }, { "epoch": 1.9254061527825788, "grad_norm": 0.6450064675434113, "learning_rate": 3.650235546891312e-08, "loss": 0.2678, "step": 27851 }, { "epoch": 1.9254752851711028, "grad_norm": 0.7314576424540306, "learning_rate": 3.6434871987384715e-08, "loss": 0.2876, "step": 27852 }, { "epoch": 1.9255444175596268, "grad_norm": 0.8075903204671229, "learning_rate": 3.63674507152878e-08, "loss": 0.2369, "step": 27853 }, { "epoch": 1.9256135499481508, "grad_norm": 0.646700732626812, "learning_rate": 3.630009165346615e-08, "loss": 0.2476, "step": 27854 }, { "epoch": 1.9256826823366748, "grad_norm": 0.7272721991869875, "learning_rate": 3.6232794802765204e-08, "loss": 0.3739, "step": 27855 }, { "epoch": 1.9257518147251989, "grad_norm": 0.700097296312416, "learning_rate": 3.616556016402817e-08, "loss": 0.2305, "step": 27856 }, { "epoch": 1.9258209471137229, "grad_norm": 0.6658554428056193, "learning_rate": 3.6098387738096593e-08, "loss": 0.2222, "step": 27857 }, { "epoch": 1.9258900795022469, "grad_norm": 0.7256625249945293, "learning_rate": 3.60312775258137e-08, "loss": 0.2595, "step": 27858 }, { "epoch": 1.925959211890771, "grad_norm": 0.6553501817033587, "learning_rate": 3.596422952802048e-08, "loss": 0.2602, "step": 27859 }, { "epoch": 1.926028344279295, "grad_norm": 0.6966631730902456, "learning_rate": 3.589724374555681e-08, "loss": 0.2322, "step": 27860 }, { "epoch": 1.926097476667819, "grad_norm": 0.7918019618501981, "learning_rate": 3.583032017926202e-08, "loss": 0.2415, "step": 27861 }, { "epoch": 1.926166609056343, "grad_norm": 0.7070610542134409, "learning_rate": 3.576345882997545e-08, "loss": 0.3049, "step": 27862 }, { "epoch": 1.926235741444867, "grad_norm": 0.6829802196107421, "learning_rate": 3.569665969853475e-08, "loss": 0.2346, "step": 27863 }, { "epoch": 1.926304873833391, "grad_norm": 0.6950747642884892, "learning_rate": 3.562992278577704e-08, "loss": 0.2356, "step": 27864 }, { "epoch": 1.926374006221915, "grad_norm": 0.7610990959152886, "learning_rate": 3.5563248092539413e-08, "loss": 0.2399, "step": 27865 }, { "epoch": 1.926443138610439, "grad_norm": 0.7885238639466884, "learning_rate": 3.549663561965677e-08, "loss": 0.3527, "step": 27866 }, { "epoch": 1.926512270998963, "grad_norm": 0.7822158559254017, "learning_rate": 3.543008536796455e-08, "loss": 0.2309, "step": 27867 }, { "epoch": 1.926581403387487, "grad_norm": 0.6834414817806703, "learning_rate": 3.536359733829653e-08, "loss": 0.2499, "step": 27868 }, { "epoch": 1.926650535776011, "grad_norm": 0.7916097747653912, "learning_rate": 3.5297171531485374e-08, "loss": 0.3037, "step": 27869 }, { "epoch": 1.926719668164535, "grad_norm": 0.7904897507962872, "learning_rate": 3.523080794836431e-08, "loss": 0.3177, "step": 27870 }, { "epoch": 1.9267888005530591, "grad_norm": 0.6399607625293766, "learning_rate": 3.516450658976545e-08, "loss": 0.2444, "step": 27871 }, { "epoch": 1.9268579329415831, "grad_norm": 0.8977352404878909, "learning_rate": 3.5098267456519786e-08, "loss": 0.2528, "step": 27872 }, { "epoch": 1.9269270653301072, "grad_norm": 0.7221419016642258, "learning_rate": 3.503209054945611e-08, "loss": 0.3435, "step": 27873 }, { "epoch": 1.9269961977186312, "grad_norm": 0.7186231410208168, "learning_rate": 3.496597586940542e-08, "loss": 0.2111, "step": 27874 }, { "epoch": 1.9270653301071552, "grad_norm": 0.6946580916691877, "learning_rate": 3.489992341719539e-08, "loss": 0.1837, "step": 27875 }, { "epoch": 1.9271344624956792, "grad_norm": 0.6633462261949062, "learning_rate": 3.48339331936548e-08, "loss": 0.276, "step": 27876 }, { "epoch": 1.9272035948842032, "grad_norm": 0.7282210884593429, "learning_rate": 3.4768005199609656e-08, "loss": 0.272, "step": 27877 }, { "epoch": 1.9272727272727272, "grad_norm": 0.7434103008414963, "learning_rate": 3.470213943588652e-08, "loss": 0.2435, "step": 27878 }, { "epoch": 1.9273418596612513, "grad_norm": 0.6621674777236234, "learning_rate": 3.463633590331083e-08, "loss": 0.2155, "step": 27879 }, { "epoch": 1.9274109920497753, "grad_norm": 0.7469296950689455, "learning_rate": 3.457059460270862e-08, "loss": 0.335, "step": 27880 }, { "epoch": 1.9274801244382993, "grad_norm": 0.827366253737065, "learning_rate": 3.450491553490198e-08, "loss": 0.2291, "step": 27881 }, { "epoch": 1.9275492568268233, "grad_norm": 0.6115149142887766, "learning_rate": 3.4439298700715275e-08, "loss": 0.1884, "step": 27882 }, { "epoch": 1.9276183892153473, "grad_norm": 0.7307792806792927, "learning_rate": 3.437374410097061e-08, "loss": 0.2549, "step": 27883 }, { "epoch": 1.9276875216038714, "grad_norm": 0.7025496410740955, "learning_rate": 3.4308251736488996e-08, "loss": 0.2645, "step": 27884 }, { "epoch": 1.9277566539923954, "grad_norm": 0.772183447229429, "learning_rate": 3.4242821608092004e-08, "loss": 0.2173, "step": 27885 }, { "epoch": 1.9278257863809194, "grad_norm": 0.7245248038816368, "learning_rate": 3.4177453716599526e-08, "loss": 0.1883, "step": 27886 }, { "epoch": 1.9278949187694434, "grad_norm": 0.7871661951486731, "learning_rate": 3.411214806283092e-08, "loss": 0.362, "step": 27887 }, { "epoch": 1.9279640511579674, "grad_norm": 0.800232294162044, "learning_rate": 3.404690464760441e-08, "loss": 0.278, "step": 27888 }, { "epoch": 1.9280331835464914, "grad_norm": 0.7561104004770532, "learning_rate": 3.398172347173767e-08, "loss": 0.2356, "step": 27889 }, { "epoch": 1.9281023159350155, "grad_norm": 0.6595620882335403, "learning_rate": 3.3916604536047836e-08, "loss": 0.2458, "step": 27890 }, { "epoch": 1.9281714483235395, "grad_norm": 0.6706469628888589, "learning_rate": 3.385154784135092e-08, "loss": 0.3287, "step": 27891 }, { "epoch": 1.9282405807120635, "grad_norm": 0.795819674697061, "learning_rate": 3.3786553388462376e-08, "loss": 0.2279, "step": 27892 }, { "epoch": 1.9283097131005875, "grad_norm": 0.6514171449666091, "learning_rate": 3.372162117819655e-08, "loss": 0.2844, "step": 27893 }, { "epoch": 1.9283788454891115, "grad_norm": 0.7376509684938124, "learning_rate": 3.3656751211368355e-08, "loss": 0.3035, "step": 27894 }, { "epoch": 1.9284479778776356, "grad_norm": 0.7197678201666735, "learning_rate": 3.3591943488788805e-08, "loss": 0.2456, "step": 27895 }, { "epoch": 1.9285171102661596, "grad_norm": 0.912491268078412, "learning_rate": 3.3527198011272245e-08, "loss": 0.2426, "step": 27896 }, { "epoch": 1.9285862426546836, "grad_norm": 0.6743869211546379, "learning_rate": 3.3462514779628584e-08, "loss": 0.2682, "step": 27897 }, { "epoch": 1.9286553750432076, "grad_norm": 0.7154390704223244, "learning_rate": 3.33978937946694e-08, "loss": 0.3454, "step": 27898 }, { "epoch": 1.9287245074317316, "grad_norm": 0.679890278802065, "learning_rate": 3.333333505720404e-08, "loss": 0.2012, "step": 27899 }, { "epoch": 1.9287936398202556, "grad_norm": 0.6470188806155831, "learning_rate": 3.3268838568042416e-08, "loss": 0.2195, "step": 27900 }, { "epoch": 1.9288627722087797, "grad_norm": 0.7265239383404275, "learning_rate": 3.320440432799166e-08, "loss": 0.3173, "step": 27901 }, { "epoch": 1.9289319045973037, "grad_norm": 1.7070759420424504, "learning_rate": 3.3140032337861115e-08, "loss": 0.32, "step": 27902 }, { "epoch": 1.9290010369858277, "grad_norm": 0.6558859457556745, "learning_rate": 3.3075722598455706e-08, "loss": 0.2288, "step": 27903 }, { "epoch": 1.9290701693743517, "grad_norm": 0.6750943743540325, "learning_rate": 3.301147511058256e-08, "loss": 0.2199, "step": 27904 }, { "epoch": 1.929139301762876, "grad_norm": 0.7152060235098696, "learning_rate": 3.294728987504603e-08, "loss": 0.2714, "step": 27905 }, { "epoch": 1.9292084341514, "grad_norm": 0.659400105194789, "learning_rate": 3.2883166892651585e-08, "loss": 0.2092, "step": 27906 }, { "epoch": 1.929277566539924, "grad_norm": 0.708361775572246, "learning_rate": 3.281910616420192e-08, "loss": 0.1868, "step": 27907 }, { "epoch": 1.929346698928448, "grad_norm": 0.6667486267720643, "learning_rate": 3.2755107690501386e-08, "loss": 0.2119, "step": 27908 }, { "epoch": 1.929415831316972, "grad_norm": 0.7110154387856858, "learning_rate": 3.269117147235046e-08, "loss": 0.3488, "step": 27909 }, { "epoch": 1.929484963705496, "grad_norm": 0.6573119270670326, "learning_rate": 3.262729751055127e-08, "loss": 0.2461, "step": 27910 }, { "epoch": 1.92955409609402, "grad_norm": 0.6678171495398688, "learning_rate": 3.2563485805904294e-08, "loss": 0.2417, "step": 27911 }, { "epoch": 1.929623228482544, "grad_norm": 0.7193603402734563, "learning_rate": 3.249973635920889e-08, "loss": 0.3364, "step": 27912 }, { "epoch": 1.929692360871068, "grad_norm": 0.7805022365026096, "learning_rate": 3.243604917126497e-08, "loss": 0.3067, "step": 27913 }, { "epoch": 1.9297614932595921, "grad_norm": 0.8468481324339527, "learning_rate": 3.237242424286968e-08, "loss": 0.2364, "step": 27914 }, { "epoch": 1.9298306256481161, "grad_norm": 0.6771978994671065, "learning_rate": 3.230886157482072e-08, "loss": 0.2512, "step": 27915 }, { "epoch": 1.9298997580366402, "grad_norm": 0.7279276974071626, "learning_rate": 3.2245361167915766e-08, "loss": 0.3785, "step": 27916 }, { "epoch": 1.9299688904251642, "grad_norm": 0.6728439115231221, "learning_rate": 3.21819230229492e-08, "loss": 0.2307, "step": 27917 }, { "epoch": 1.9300380228136882, "grad_norm": 0.8183521172570023, "learning_rate": 3.211854714071705e-08, "loss": 0.1671, "step": 27918 }, { "epoch": 1.9301071552022122, "grad_norm": 1.1547310178487198, "learning_rate": 3.205523352201312e-08, "loss": 0.3096, "step": 27919 }, { "epoch": 1.9301762875907362, "grad_norm": 0.7342291222836841, "learning_rate": 3.199198216763122e-08, "loss": 0.3191, "step": 27920 }, { "epoch": 1.9302454199792602, "grad_norm": 0.716429580689905, "learning_rate": 3.192879307836405e-08, "loss": 0.2582, "step": 27921 }, { "epoch": 1.9303145523677843, "grad_norm": 0.7171629593939052, "learning_rate": 3.186566625500376e-08, "loss": 0.3054, "step": 27922 }, { "epoch": 1.9303836847563083, "grad_norm": 0.6450659652634502, "learning_rate": 3.180260169834082e-08, "loss": 0.2921, "step": 27923 }, { "epoch": 1.9304528171448323, "grad_norm": 0.7407945727086226, "learning_rate": 3.173959940916682e-08, "loss": 0.2437, "step": 27924 }, { "epoch": 1.9305219495333565, "grad_norm": 0.7078270761798972, "learning_rate": 3.1676659388270023e-08, "loss": 0.2081, "step": 27925 }, { "epoch": 1.9305910819218806, "grad_norm": 0.685137268241222, "learning_rate": 3.1613781636440354e-08, "loss": 0.2817, "step": 27926 }, { "epoch": 1.9306602143104046, "grad_norm": 0.6868492403370668, "learning_rate": 3.155096615446496e-08, "loss": 0.3003, "step": 27927 }, { "epoch": 1.9307293466989286, "grad_norm": 0.6766054985175481, "learning_rate": 3.148821294313209e-08, "loss": 0.2698, "step": 27928 }, { "epoch": 1.9307984790874526, "grad_norm": 0.7020194250657447, "learning_rate": 3.1425522003227235e-08, "loss": 0.2594, "step": 27929 }, { "epoch": 1.9308676114759766, "grad_norm": 0.7672645413548846, "learning_rate": 3.136289333553699e-08, "loss": 0.3575, "step": 27930 }, { "epoch": 1.9309367438645006, "grad_norm": 0.6955600942515422, "learning_rate": 3.130032694084628e-08, "loss": 0.2582, "step": 27931 }, { "epoch": 1.9310058762530247, "grad_norm": 0.6992071638353328, "learning_rate": 3.123782281993837e-08, "loss": 0.2028, "step": 27932 }, { "epoch": 1.9310750086415487, "grad_norm": 0.8121737735495373, "learning_rate": 3.1175380973597624e-08, "loss": 0.2432, "step": 27933 }, { "epoch": 1.9311441410300727, "grad_norm": 0.975019146868967, "learning_rate": 3.1113001402605645e-08, "loss": 0.3575, "step": 27934 }, { "epoch": 1.9312132734185967, "grad_norm": 0.8232791099628222, "learning_rate": 3.10506841077457e-08, "loss": 0.2492, "step": 27935 }, { "epoch": 1.9312824058071207, "grad_norm": 0.6749995629157023, "learning_rate": 3.098842908979716e-08, "loss": 0.2135, "step": 27936 }, { "epoch": 1.9313515381956448, "grad_norm": 0.7981516026263236, "learning_rate": 3.0926236349541616e-08, "loss": 0.376, "step": 27937 }, { "epoch": 1.9314206705841688, "grad_norm": 0.7874053710968414, "learning_rate": 3.0864105887757346e-08, "loss": 0.3222, "step": 27938 }, { "epoch": 1.9314898029726928, "grad_norm": 0.6769253842605515, "learning_rate": 3.0802037705224275e-08, "loss": 0.2262, "step": 27939 }, { "epoch": 1.9315589353612168, "grad_norm": 0.805495938318475, "learning_rate": 3.074003180271956e-08, "loss": 0.2514, "step": 27940 }, { "epoch": 1.9316280677497408, "grad_norm": 0.7038044711337776, "learning_rate": 3.067808818102036e-08, "loss": 0.3183, "step": 27941 }, { "epoch": 1.9316972001382648, "grad_norm": 0.7120544226754243, "learning_rate": 3.061620684090272e-08, "loss": 0.234, "step": 27942 }, { "epoch": 1.9317663325267889, "grad_norm": 1.6715536306140912, "learning_rate": 3.055438778314379e-08, "loss": 0.2232, "step": 27943 }, { "epoch": 1.9318354649153129, "grad_norm": 0.7426435368100269, "learning_rate": 3.049263100851629e-08, "loss": 0.3738, "step": 27944 }, { "epoch": 1.931904597303837, "grad_norm": 0.708889887206565, "learning_rate": 3.0430936517795716e-08, "loss": 0.289, "step": 27945 }, { "epoch": 1.931973729692361, "grad_norm": 0.6876886235669336, "learning_rate": 3.0369304311754225e-08, "loss": 0.2697, "step": 27946 }, { "epoch": 1.932042862080885, "grad_norm": 0.7310486345562744, "learning_rate": 3.0307734391165635e-08, "loss": 0.2478, "step": 27947 }, { "epoch": 1.932111994469409, "grad_norm": 0.7286540581659577, "learning_rate": 3.024622675679989e-08, "loss": 0.3793, "step": 27948 }, { "epoch": 1.932181126857933, "grad_norm": 0.729141090097371, "learning_rate": 3.018478140942915e-08, "loss": 0.2345, "step": 27949 }, { "epoch": 1.932250259246457, "grad_norm": 0.7351145234486428, "learning_rate": 3.012339834982336e-08, "loss": 0.2921, "step": 27950 }, { "epoch": 1.932319391634981, "grad_norm": 0.7829027936262717, "learning_rate": 3.0062077578751344e-08, "loss": 0.2607, "step": 27951 }, { "epoch": 1.932388524023505, "grad_norm": 0.7764062609731356, "learning_rate": 3.0000819096981934e-08, "loss": 0.3207, "step": 27952 }, { "epoch": 1.932457656412029, "grad_norm": 0.7577193944358479, "learning_rate": 2.993962290528285e-08, "loss": 0.1886, "step": 27953 }, { "epoch": 1.932526788800553, "grad_norm": 0.7291437728445008, "learning_rate": 2.987848900442125e-08, "loss": 0.2281, "step": 27954 }, { "epoch": 1.932595921189077, "grad_norm": 0.6576914328389251, "learning_rate": 2.9817417395163196e-08, "loss": 0.2864, "step": 27955 }, { "epoch": 1.932665053577601, "grad_norm": 0.8210247164160797, "learning_rate": 2.975640807827418e-08, "loss": 0.3402, "step": 27956 }, { "epoch": 1.9327341859661251, "grad_norm": 0.6751434138058185, "learning_rate": 2.9695461054518594e-08, "loss": 0.2176, "step": 27957 }, { "epoch": 1.9328033183546491, "grad_norm": 0.6716514851152317, "learning_rate": 2.9634576324660824e-08, "loss": 0.2196, "step": 27958 }, { "epoch": 1.9328724507431732, "grad_norm": 0.6907354458693998, "learning_rate": 2.9573753889463598e-08, "loss": 0.2977, "step": 27959 }, { "epoch": 1.9329415831316972, "grad_norm": 0.720184708060551, "learning_rate": 2.9512993749689077e-08, "loss": 0.2028, "step": 27960 }, { "epoch": 1.9330107155202212, "grad_norm": 0.7532251441597325, "learning_rate": 2.9452295906098883e-08, "loss": 0.2468, "step": 27961 }, { "epoch": 1.9330798479087452, "grad_norm": 0.7101144659688314, "learning_rate": 2.9391660359454065e-08, "loss": 0.2892, "step": 27962 }, { "epoch": 1.9331489802972692, "grad_norm": 0.7250203543107866, "learning_rate": 2.9331087110514024e-08, "loss": 0.2824, "step": 27963 }, { "epoch": 1.9332181126857932, "grad_norm": 1.259887448389533, "learning_rate": 2.9270576160038145e-08, "loss": 0.2273, "step": 27964 }, { "epoch": 1.9332872450743173, "grad_norm": 0.6622234090959797, "learning_rate": 2.9210127508785268e-08, "loss": 0.1855, "step": 27965 }, { "epoch": 1.9333563774628413, "grad_norm": 0.6983442578008595, "learning_rate": 2.914974115751257e-08, "loss": 0.3672, "step": 27966 }, { "epoch": 1.9334255098513653, "grad_norm": 0.6986089003717662, "learning_rate": 2.9089417106977214e-08, "loss": 0.2712, "step": 27967 }, { "epoch": 1.9334946422398893, "grad_norm": 0.6399202929455925, "learning_rate": 2.9029155357934714e-08, "loss": 0.2399, "step": 27968 }, { "epoch": 1.9335637746284133, "grad_norm": 0.6931449516696317, "learning_rate": 2.896895591114057e-08, "loss": 0.2773, "step": 27969 }, { "epoch": 1.9336329070169374, "grad_norm": 0.7275918398714144, "learning_rate": 2.8908818767349745e-08, "loss": 0.2945, "step": 27970 }, { "epoch": 1.9337020394054614, "grad_norm": 1.3658362909115787, "learning_rate": 2.8848743927315514e-08, "loss": 0.1973, "step": 27971 }, { "epoch": 1.9337711717939854, "grad_norm": 0.7747971548758376, "learning_rate": 2.8788731391790614e-08, "loss": 0.2904, "step": 27972 }, { "epoch": 1.9338403041825094, "grad_norm": 0.7539817294968008, "learning_rate": 2.872878116152722e-08, "loss": 0.3239, "step": 27973 }, { "epoch": 1.9339094365710334, "grad_norm": 0.6873357366066156, "learning_rate": 2.8668893237277506e-08, "loss": 0.258, "step": 27974 }, { "epoch": 1.9339785689595574, "grad_norm": 0.6388947343555748, "learning_rate": 2.860906761979143e-08, "loss": 0.2014, "step": 27975 }, { "epoch": 1.9340477013480815, "grad_norm": 0.6725726011705474, "learning_rate": 2.854930430981839e-08, "loss": 0.2666, "step": 27976 }, { "epoch": 1.9341168337366055, "grad_norm": 0.709753706598882, "learning_rate": 2.8489603308107793e-08, "loss": 0.2712, "step": 27977 }, { "epoch": 1.9341859661251295, "grad_norm": 0.699126042998302, "learning_rate": 2.8429964615408478e-08, "loss": 0.2952, "step": 27978 }, { "epoch": 1.9342550985136535, "grad_norm": 0.6622787247725044, "learning_rate": 2.837038823246707e-08, "loss": 0.2229, "step": 27979 }, { "epoch": 1.9343242309021775, "grad_norm": 0.70819935290361, "learning_rate": 2.8310874160030756e-08, "loss": 0.3187, "step": 27980 }, { "epoch": 1.9343933632907016, "grad_norm": 0.7486940404539665, "learning_rate": 2.8251422398845042e-08, "loss": 0.2584, "step": 27981 }, { "epoch": 1.9344624956792256, "grad_norm": 0.7483420781758455, "learning_rate": 2.8192032949655445e-08, "loss": 0.2061, "step": 27982 }, { "epoch": 1.9345316280677496, "grad_norm": 0.7213576832059957, "learning_rate": 2.8132705813206372e-08, "loss": 0.2368, "step": 27983 }, { "epoch": 1.9346007604562736, "grad_norm": 0.7170218212245805, "learning_rate": 2.8073440990240563e-08, "loss": 0.3178, "step": 27984 }, { "epoch": 1.9346698928447976, "grad_norm": 0.6597432157653764, "learning_rate": 2.8014238481501866e-08, "loss": 0.1922, "step": 27985 }, { "epoch": 1.9347390252333219, "grad_norm": 0.6442229876058705, "learning_rate": 2.7955098287731354e-08, "loss": 0.2301, "step": 27986 }, { "epoch": 1.9348081576218459, "grad_norm": 0.7082369946661582, "learning_rate": 2.789602040967121e-08, "loss": 0.3112, "step": 27987 }, { "epoch": 1.93487729001037, "grad_norm": 0.7834181254600016, "learning_rate": 2.7837004848060846e-08, "loss": 0.2672, "step": 27988 }, { "epoch": 1.934946422398894, "grad_norm": 0.6633809615384132, "learning_rate": 2.7778051603640778e-08, "loss": 0.2372, "step": 27989 }, { "epoch": 1.935015554787418, "grad_norm": 0.7445697441410015, "learning_rate": 2.7719160677149303e-08, "loss": 0.2528, "step": 27990 }, { "epoch": 1.935084687175942, "grad_norm": 0.6587260366982587, "learning_rate": 2.7660332069324168e-08, "loss": 0.3449, "step": 27991 }, { "epoch": 1.935153819564466, "grad_norm": 0.7467928864538791, "learning_rate": 2.760156578090367e-08, "loss": 0.2524, "step": 27992 }, { "epoch": 1.93522295195299, "grad_norm": 0.6959769594600053, "learning_rate": 2.7542861812624443e-08, "loss": 0.2591, "step": 27993 }, { "epoch": 1.935292084341514, "grad_norm": 0.7004697155981929, "learning_rate": 2.7484220165220897e-08, "loss": 0.3098, "step": 27994 }, { "epoch": 1.935361216730038, "grad_norm": 0.7257975051158299, "learning_rate": 2.7425640839429666e-08, "loss": 0.3052, "step": 27995 }, { "epoch": 1.935430349118562, "grad_norm": 0.6495761513938179, "learning_rate": 2.73671238359835e-08, "loss": 0.2348, "step": 27996 }, { "epoch": 1.935499481507086, "grad_norm": 0.69521554443565, "learning_rate": 2.730866915561625e-08, "loss": 0.274, "step": 27997 }, { "epoch": 1.93556861389561, "grad_norm": 0.6656534568683355, "learning_rate": 2.725027679906067e-08, "loss": 0.2966, "step": 27998 }, { "epoch": 1.935637746284134, "grad_norm": 0.7938350524690154, "learning_rate": 2.719194676704895e-08, "loss": 0.2218, "step": 27999 }, { "epoch": 1.9357068786726581, "grad_norm": 0.7876919801891838, "learning_rate": 2.713367906031161e-08, "loss": 0.2345, "step": 28000 }, { "epoch": 1.9357760110611821, "grad_norm": 0.731530249584551, "learning_rate": 2.707547367957919e-08, "loss": 0.3151, "step": 28001 }, { "epoch": 1.9358451434497062, "grad_norm": 0.7677100336463646, "learning_rate": 2.7017330625581096e-08, "loss": 0.279, "step": 28002 }, { "epoch": 1.9359142758382302, "grad_norm": 0.7229020149349389, "learning_rate": 2.695924989904619e-08, "loss": 0.2417, "step": 28003 }, { "epoch": 1.9359834082267542, "grad_norm": 0.6317619026117867, "learning_rate": 2.690123150070223e-08, "loss": 0.1935, "step": 28004 }, { "epoch": 1.9360525406152782, "grad_norm": 0.9163202235801656, "learning_rate": 2.684327543127585e-08, "loss": 0.3639, "step": 28005 }, { "epoch": 1.9361216730038024, "grad_norm": 0.6816067954827049, "learning_rate": 2.6785381691494805e-08, "loss": 0.215, "step": 28006 }, { "epoch": 1.9361908053923265, "grad_norm": 0.7312204970021436, "learning_rate": 2.6727550282083513e-08, "loss": 0.2399, "step": 28007 }, { "epoch": 1.9362599377808505, "grad_norm": 0.6632273450273738, "learning_rate": 2.666978120376751e-08, "loss": 0.265, "step": 28008 }, { "epoch": 1.9363290701693745, "grad_norm": 0.7753120801138421, "learning_rate": 2.66120744572701e-08, "loss": 0.3692, "step": 28009 }, { "epoch": 1.9363982025578985, "grad_norm": 0.6649867217023101, "learning_rate": 2.6554430043315705e-08, "loss": 0.2102, "step": 28010 }, { "epoch": 1.9364673349464225, "grad_norm": 0.7201922985439354, "learning_rate": 2.649684796262486e-08, "loss": 0.2646, "step": 28011 }, { "epoch": 1.9365364673349466, "grad_norm": 0.7845526808413178, "learning_rate": 2.643932821592088e-08, "loss": 0.3379, "step": 28012 }, { "epoch": 1.9366055997234706, "grad_norm": 0.7984801561601531, "learning_rate": 2.6381870803924293e-08, "loss": 0.2722, "step": 28013 }, { "epoch": 1.9366747321119946, "grad_norm": 0.797698800552077, "learning_rate": 2.6324475727355636e-08, "loss": 0.2198, "step": 28014 }, { "epoch": 1.9367438645005186, "grad_norm": 0.7027014463371866, "learning_rate": 2.6267142986933225e-08, "loss": 0.2246, "step": 28015 }, { "epoch": 1.9368129968890426, "grad_norm": 0.7647218058828984, "learning_rate": 2.620987258337593e-08, "loss": 0.3795, "step": 28016 }, { "epoch": 1.9368821292775666, "grad_norm": 1.4149721513263787, "learning_rate": 2.615266451740206e-08, "loss": 0.2054, "step": 28017 }, { "epoch": 1.9369512616660907, "grad_norm": 0.6913815184762695, "learning_rate": 2.609551878972827e-08, "loss": 0.2203, "step": 28018 }, { "epoch": 1.9370203940546147, "grad_norm": 0.7888681122124511, "learning_rate": 2.603843540107065e-08, "loss": 0.3304, "step": 28019 }, { "epoch": 1.9370895264431387, "grad_norm": 1.2678177142088134, "learning_rate": 2.5981414352145295e-08, "loss": 0.3242, "step": 28020 }, { "epoch": 1.9371586588316627, "grad_norm": 0.8003595256614545, "learning_rate": 2.592445564366608e-08, "loss": 0.2497, "step": 28021 }, { "epoch": 1.9372277912201867, "grad_norm": 0.7211532378836176, "learning_rate": 2.5867559276346877e-08, "loss": 0.2669, "step": 28022 }, { "epoch": 1.9372969236087108, "grad_norm": 0.747308235703646, "learning_rate": 2.581072525090156e-08, "loss": 0.3592, "step": 28023 }, { "epoch": 1.9373660559972348, "grad_norm": 0.7860922430729942, "learning_rate": 2.575395356804178e-08, "loss": 0.2301, "step": 28024 }, { "epoch": 1.9374351883857588, "grad_norm": 0.7149601172137766, "learning_rate": 2.569724422847919e-08, "loss": 0.2254, "step": 28025 }, { "epoch": 1.9375043207742828, "grad_norm": 0.6493719413029164, "learning_rate": 2.5640597232924335e-08, "loss": 0.2694, "step": 28026 }, { "epoch": 1.9375734531628068, "grad_norm": 0.7927376238687586, "learning_rate": 2.5584012582087758e-08, "loss": 0.3401, "step": 28027 }, { "epoch": 1.9376425855513308, "grad_norm": 0.8591449818532617, "learning_rate": 2.5527490276678334e-08, "loss": 0.2233, "step": 28028 }, { "epoch": 1.9377117179398549, "grad_norm": 0.8285598966678956, "learning_rate": 2.5471030317404388e-08, "loss": 0.2292, "step": 28029 }, { "epoch": 1.9377808503283789, "grad_norm": 0.7605017304338711, "learning_rate": 2.541463270497424e-08, "loss": 0.3684, "step": 28030 }, { "epoch": 1.937849982716903, "grad_norm": 0.6805227517324353, "learning_rate": 2.5358297440093438e-08, "loss": 0.2291, "step": 28031 }, { "epoch": 1.937919115105427, "grad_norm": 0.6154763576943513, "learning_rate": 2.5302024523469192e-08, "loss": 0.2062, "step": 28032 }, { "epoch": 1.937988247493951, "grad_norm": 0.6913688494146787, "learning_rate": 2.5245813955805944e-08, "loss": 0.2484, "step": 28033 }, { "epoch": 1.938057379882475, "grad_norm": 0.7743982761838732, "learning_rate": 2.5189665737808123e-08, "loss": 0.3298, "step": 28034 }, { "epoch": 1.938126512270999, "grad_norm": 0.7527302204358899, "learning_rate": 2.5133579870180723e-08, "loss": 0.2815, "step": 28035 }, { "epoch": 1.938195644659523, "grad_norm": 0.8979938663122129, "learning_rate": 2.5077556353625408e-08, "loss": 0.3688, "step": 28036 }, { "epoch": 1.938264777048047, "grad_norm": 0.6635816275293892, "learning_rate": 2.502159518884495e-08, "loss": 0.293, "step": 28037 }, { "epoch": 1.938333909436571, "grad_norm": 0.6563047973065338, "learning_rate": 2.496569637654045e-08, "loss": 0.1949, "step": 28038 }, { "epoch": 1.938403041825095, "grad_norm": 0.6488012384626115, "learning_rate": 2.4909859917413015e-08, "loss": 0.21, "step": 28039 }, { "epoch": 1.938472174213619, "grad_norm": 0.7609302040519268, "learning_rate": 2.4854085812161533e-08, "loss": 0.2719, "step": 28040 }, { "epoch": 1.938541306602143, "grad_norm": 0.7280168700624932, "learning_rate": 2.479837406148544e-08, "loss": 0.3156, "step": 28041 }, { "epoch": 1.938610438990667, "grad_norm": 0.7752276971284094, "learning_rate": 2.4742724666083628e-08, "loss": 0.2613, "step": 28042 }, { "epoch": 1.9386795713791911, "grad_norm": 0.7260529585109985, "learning_rate": 2.4687137626652757e-08, "loss": 0.2213, "step": 28043 }, { "epoch": 1.9387487037677151, "grad_norm": 0.796977610225095, "learning_rate": 2.4631612943889494e-08, "loss": 0.3752, "step": 28044 }, { "epoch": 1.9388178361562391, "grad_norm": 0.6936888482588612, "learning_rate": 2.45761506184905e-08, "loss": 0.3527, "step": 28045 }, { "epoch": 1.9388869685447632, "grad_norm": 0.6808937793181743, "learning_rate": 2.4520750651150227e-08, "loss": 0.2432, "step": 28046 }, { "epoch": 1.9389561009332872, "grad_norm": 0.7426533208540596, "learning_rate": 2.4465413042563115e-08, "loss": 0.2524, "step": 28047 }, { "epoch": 1.9390252333218112, "grad_norm": 0.7483257590142377, "learning_rate": 2.4410137793422494e-08, "loss": 0.3449, "step": 28048 }, { "epoch": 1.9390943657103352, "grad_norm": 0.7327522865116368, "learning_rate": 2.4354924904421707e-08, "loss": 0.2204, "step": 28049 }, { "epoch": 1.9391634980988592, "grad_norm": 0.7672033320282365, "learning_rate": 2.429977437625297e-08, "loss": 0.187, "step": 28050 }, { "epoch": 1.9392326304873833, "grad_norm": 0.7526398767579547, "learning_rate": 2.4244686209606848e-08, "loss": 0.2751, "step": 28051 }, { "epoch": 1.9393017628759073, "grad_norm": 0.7191540982940205, "learning_rate": 2.4189660405173897e-08, "loss": 0.3203, "step": 28052 }, { "epoch": 1.9393708952644313, "grad_norm": 0.7180308721835819, "learning_rate": 2.4134696963643566e-08, "loss": 0.228, "step": 28053 }, { "epoch": 1.9394400276529553, "grad_norm": 0.7444736314993596, "learning_rate": 2.4079795885705304e-08, "loss": 0.2816, "step": 28054 }, { "epoch": 1.9395091600414793, "grad_norm": 0.7380440337716921, "learning_rate": 2.4024957172046336e-08, "loss": 0.3271, "step": 28055 }, { "epoch": 1.9395782924300033, "grad_norm": 0.7146823316293813, "learning_rate": 2.3970180823355005e-08, "loss": 0.2071, "step": 28056 }, { "epoch": 1.9396474248185274, "grad_norm": 0.6374909854604914, "learning_rate": 2.391546684031687e-08, "loss": 0.2046, "step": 28057 }, { "epoch": 1.9397165572070514, "grad_norm": 0.7092138249409988, "learning_rate": 2.3860815223618606e-08, "loss": 0.2803, "step": 28058 }, { "epoch": 1.9397856895955754, "grad_norm": 0.7170527006042556, "learning_rate": 2.3806225973944662e-08, "loss": 0.3569, "step": 28059 }, { "epoch": 1.9398548219840994, "grad_norm": 0.6688069198672848, "learning_rate": 2.375169909197894e-08, "loss": 0.2416, "step": 28060 }, { "epoch": 1.9399239543726234, "grad_norm": 1.289118151146169, "learning_rate": 2.3697234578405337e-08, "loss": 0.2239, "step": 28061 }, { "epoch": 1.9399930867611475, "grad_norm": 0.7040781679420224, "learning_rate": 2.3642832433906082e-08, "loss": 0.3676, "step": 28062 }, { "epoch": 1.9400622191496715, "grad_norm": 0.7723903420344245, "learning_rate": 2.358849265916341e-08, "loss": 0.2867, "step": 28063 }, { "epoch": 1.9401313515381955, "grad_norm": 0.7252175358442935, "learning_rate": 2.353421525485844e-08, "loss": 0.2513, "step": 28064 }, { "epoch": 1.9402004839267195, "grad_norm": 0.7062146898743502, "learning_rate": 2.348000022167063e-08, "loss": 0.3019, "step": 28065 }, { "epoch": 1.9402696163152435, "grad_norm": 0.6906723207816903, "learning_rate": 2.3425847560280546e-08, "loss": 0.2898, "step": 28066 }, { "epoch": 1.9403387487037678, "grad_norm": 0.7983574115440484, "learning_rate": 2.3371757271366534e-08, "loss": 0.1962, "step": 28067 }, { "epoch": 1.9404078810922918, "grad_norm": 0.8691769108324416, "learning_rate": 2.3317729355605832e-08, "loss": 0.2, "step": 28068 }, { "epoch": 1.9404770134808158, "grad_norm": 0.7389345723783287, "learning_rate": 2.3263763813676234e-08, "loss": 0.2794, "step": 28069 }, { "epoch": 1.9405461458693398, "grad_norm": 0.7288487693080449, "learning_rate": 2.3209860646253858e-08, "loss": 0.2737, "step": 28070 }, { "epoch": 1.9406152782578638, "grad_norm": 0.6355649306800537, "learning_rate": 2.3156019854014277e-08, "loss": 0.1986, "step": 28071 }, { "epoch": 1.9406844106463879, "grad_norm": 0.6871164149527325, "learning_rate": 2.310224143763251e-08, "loss": 0.2189, "step": 28072 }, { "epoch": 1.9407535430349119, "grad_norm": 0.6404408982253037, "learning_rate": 2.3048525397783018e-08, "loss": 0.3015, "step": 28073 }, { "epoch": 1.940822675423436, "grad_norm": 0.718626379390398, "learning_rate": 2.2994871735138036e-08, "loss": 0.2362, "step": 28074 }, { "epoch": 1.94089180781196, "grad_norm": 0.7298993714887608, "learning_rate": 2.2941280450370363e-08, "loss": 0.2391, "step": 28075 }, { "epoch": 1.940960940200484, "grad_norm": 0.6955355566110183, "learning_rate": 2.2887751544151683e-08, "loss": 0.3074, "step": 28076 }, { "epoch": 1.941030072589008, "grad_norm": 0.7506950671924417, "learning_rate": 2.283428501715368e-08, "loss": 0.2876, "step": 28077 }, { "epoch": 1.941099204977532, "grad_norm": 0.6640735131992975, "learning_rate": 2.2780880870045263e-08, "loss": 0.2616, "step": 28078 }, { "epoch": 1.941168337366056, "grad_norm": 0.6461741762336131, "learning_rate": 2.2727539103496453e-08, "loss": 0.2992, "step": 28079 }, { "epoch": 1.94123746975458, "grad_norm": 0.7106290721099017, "learning_rate": 2.267425971817505e-08, "loss": 0.2927, "step": 28080 }, { "epoch": 1.941306602143104, "grad_norm": 0.8125038749468612, "learning_rate": 2.2621042714749964e-08, "loss": 0.2723, "step": 28081 }, { "epoch": 1.941375734531628, "grad_norm": 0.7161342879066603, "learning_rate": 2.2567888093887324e-08, "loss": 0.2209, "step": 28082 }, { "epoch": 1.941444866920152, "grad_norm": 0.7357602948676852, "learning_rate": 2.2514795856253825e-08, "loss": 0.2896, "step": 28083 }, { "epoch": 1.941513999308676, "grad_norm": 0.6880950427861462, "learning_rate": 2.2461766002514485e-08, "loss": 0.3337, "step": 28084 }, { "epoch": 1.9415831316972, "grad_norm": 0.6699660729980864, "learning_rate": 2.2408798533333777e-08, "loss": 0.2549, "step": 28085 }, { "epoch": 1.9416522640857241, "grad_norm": 0.699955285270624, "learning_rate": 2.2355893449376165e-08, "loss": 0.2334, "step": 28086 }, { "epoch": 1.9417213964742484, "grad_norm": 0.6336760539404503, "learning_rate": 2.2303050751304457e-08, "loss": 0.3043, "step": 28087 }, { "epoch": 1.9417905288627724, "grad_norm": 0.6961630587555606, "learning_rate": 2.2250270439780897e-08, "loss": 0.235, "step": 28088 }, { "epoch": 1.9418596612512964, "grad_norm": 0.6089162835061157, "learning_rate": 2.219755251546718e-08, "loss": 0.2034, "step": 28089 }, { "epoch": 1.9419287936398204, "grad_norm": 0.7526881599603324, "learning_rate": 2.2144896979023333e-08, "loss": 0.2263, "step": 28090 }, { "epoch": 1.9419979260283444, "grad_norm": 0.7422342101178361, "learning_rate": 2.2092303831110496e-08, "loss": 0.3384, "step": 28091 }, { "epoch": 1.9420670584168684, "grad_norm": 0.8733429580753362, "learning_rate": 2.203977307238647e-08, "loss": 0.2171, "step": 28092 }, { "epoch": 1.9421361908053925, "grad_norm": 0.761371065651325, "learning_rate": 2.1987304703510736e-08, "loss": 0.237, "step": 28093 }, { "epoch": 1.9422053231939165, "grad_norm": 0.7174274710238949, "learning_rate": 2.193489872514054e-08, "loss": 0.3079, "step": 28094 }, { "epoch": 1.9422744555824405, "grad_norm": 0.7095222116535713, "learning_rate": 2.188255513793258e-08, "loss": 0.281, "step": 28095 }, { "epoch": 1.9423435879709645, "grad_norm": 0.7197752780888762, "learning_rate": 2.1830273942542445e-08, "loss": 0.2042, "step": 28096 }, { "epoch": 1.9424127203594885, "grad_norm": 0.6506082613726792, "learning_rate": 2.1778055139626277e-08, "loss": 0.2352, "step": 28097 }, { "epoch": 1.9424818527480125, "grad_norm": 0.7509547273393746, "learning_rate": 2.172589872983799e-08, "loss": 0.3211, "step": 28098 }, { "epoch": 1.9425509851365366, "grad_norm": 0.6593584478411174, "learning_rate": 2.167380471383207e-08, "loss": 0.2434, "step": 28099 }, { "epoch": 1.9426201175250606, "grad_norm": 0.5982481887076273, "learning_rate": 2.1621773092260212e-08, "loss": 0.1958, "step": 28100 }, { "epoch": 1.9426892499135846, "grad_norm": 0.6463078696549865, "learning_rate": 2.156980386577523e-08, "loss": 0.2528, "step": 28101 }, { "epoch": 1.9427583823021086, "grad_norm": 1.3031586578079846, "learning_rate": 2.1517897035028268e-08, "loss": 0.3387, "step": 28102 }, { "epoch": 1.9428275146906326, "grad_norm": 0.637937198398267, "learning_rate": 2.146605260066992e-08, "loss": 0.2112, "step": 28103 }, { "epoch": 1.9428966470791567, "grad_norm": 0.737190461584772, "learning_rate": 2.1414270563350214e-08, "loss": 0.2462, "step": 28104 }, { "epoch": 1.9429657794676807, "grad_norm": 0.6845158570631642, "learning_rate": 2.136255092371753e-08, "loss": 0.336, "step": 28105 }, { "epoch": 1.9430349118562047, "grad_norm": 0.7204511661618644, "learning_rate": 2.1310893682420786e-08, "loss": 0.2765, "step": 28106 }, { "epoch": 1.9431040442447287, "grad_norm": 0.8666190552487085, "learning_rate": 2.1259298840107246e-08, "loss": 0.2346, "step": 28107 }, { "epoch": 1.9431731766332527, "grad_norm": 0.8405736374357689, "learning_rate": 2.1207766397423612e-08, "loss": 0.2488, "step": 28108 }, { "epoch": 1.9432423090217767, "grad_norm": 0.6891153333445998, "learning_rate": 2.1156296355015483e-08, "loss": 0.3312, "step": 28109 }, { "epoch": 1.9433114414103008, "grad_norm": 0.6739883578969307, "learning_rate": 2.110488871352734e-08, "loss": 0.2222, "step": 28110 }, { "epoch": 1.9433805737988248, "grad_norm": 0.7033559218994767, "learning_rate": 2.1053543473604775e-08, "loss": 0.2621, "step": 28111 }, { "epoch": 1.9434497061873488, "grad_norm": 0.6969094154566382, "learning_rate": 2.1002260635890614e-08, "loss": 0.274, "step": 28112 }, { "epoch": 1.9435188385758728, "grad_norm": 0.7865652089814061, "learning_rate": 2.0951040201027672e-08, "loss": 0.2616, "step": 28113 }, { "epoch": 1.9435879709643968, "grad_norm": 0.6827247265208306, "learning_rate": 2.089988216965766e-08, "loss": 0.216, "step": 28114 }, { "epoch": 1.9436571033529209, "grad_norm": 0.7880949660600444, "learning_rate": 2.0848786542422838e-08, "loss": 0.2888, "step": 28115 }, { "epoch": 1.9437262357414449, "grad_norm": 0.7800306590736256, "learning_rate": 2.0797753319962142e-08, "loss": 0.3872, "step": 28116 }, { "epoch": 1.943795368129969, "grad_norm": 0.7302103520233593, "learning_rate": 2.0746782502916175e-08, "loss": 0.2673, "step": 28117 }, { "epoch": 1.943864500518493, "grad_norm": 0.8355074354565762, "learning_rate": 2.0695874091922753e-08, "loss": 0.2522, "step": 28118 }, { "epoch": 1.943933632907017, "grad_norm": 0.7144210271776529, "learning_rate": 2.0645028087621366e-08, "loss": 0.3331, "step": 28119 }, { "epoch": 1.944002765295541, "grad_norm": 0.7748953031688207, "learning_rate": 2.059424449064762e-08, "loss": 0.2155, "step": 28120 }, { "epoch": 1.944071897684065, "grad_norm": 0.6726614087559083, "learning_rate": 2.054352330163989e-08, "loss": 0.2321, "step": 28121 }, { "epoch": 1.944141030072589, "grad_norm": 0.8469436678053572, "learning_rate": 2.0492864521232113e-08, "loss": 0.2507, "step": 28122 }, { "epoch": 1.944210162461113, "grad_norm": 0.8013583871612465, "learning_rate": 2.044226815006045e-08, "loss": 0.3768, "step": 28123 }, { "epoch": 1.944279294849637, "grad_norm": 0.7325490045698797, "learning_rate": 2.0391734188758284e-08, "loss": 0.2493, "step": 28124 }, { "epoch": 1.944348427238161, "grad_norm": 0.6541283199997612, "learning_rate": 2.0341262637958992e-08, "loss": 0.1811, "step": 28125 }, { "epoch": 1.944417559626685, "grad_norm": 0.6856867478047334, "learning_rate": 2.0290853498295403e-08, "loss": 0.2919, "step": 28126 }, { "epoch": 1.944486692015209, "grad_norm": 0.7516414617071758, "learning_rate": 2.0240506770399792e-08, "loss": 0.2516, "step": 28127 }, { "epoch": 1.944555824403733, "grad_norm": 0.7043571974568782, "learning_rate": 2.019022245490221e-08, "loss": 0.2163, "step": 28128 }, { "epoch": 1.944624956792257, "grad_norm": 0.7378071171876929, "learning_rate": 2.0140000552433257e-08, "loss": 0.2501, "step": 28129 }, { "epoch": 1.9446940891807811, "grad_norm": 0.7006244875586811, "learning_rate": 2.008984106362244e-08, "loss": 0.3286, "step": 28130 }, { "epoch": 1.9447632215693051, "grad_norm": 0.7207777718399337, "learning_rate": 2.0039743989098693e-08, "loss": 0.2546, "step": 28131 }, { "epoch": 1.9448323539578292, "grad_norm": 0.6832569340330862, "learning_rate": 1.998970932948985e-08, "loss": 0.1774, "step": 28132 }, { "epoch": 1.9449014863463532, "grad_norm": 0.7064798968236539, "learning_rate": 1.9939737085422072e-08, "loss": 0.2643, "step": 28133 }, { "epoch": 1.9449706187348772, "grad_norm": 0.7189891079875154, "learning_rate": 1.9889827257523198e-08, "loss": 0.3319, "step": 28134 }, { "epoch": 1.9450397511234012, "grad_norm": 0.7310609013240226, "learning_rate": 1.983997984641717e-08, "loss": 0.2115, "step": 28135 }, { "epoch": 1.9451088835119252, "grad_norm": 0.7069641934939469, "learning_rate": 1.979019485273015e-08, "loss": 0.2054, "step": 28136 }, { "epoch": 1.9451780159004493, "grad_norm": 0.7061062443299208, "learning_rate": 1.9740472277084976e-08, "loss": 0.3494, "step": 28137 }, { "epoch": 1.9452471482889733, "grad_norm": 0.7248706496416839, "learning_rate": 1.9690812120105596e-08, "loss": 0.2834, "step": 28138 }, { "epoch": 1.9453162806774973, "grad_norm": 0.7115208061235753, "learning_rate": 1.9641214382413732e-08, "loss": 0.2278, "step": 28139 }, { "epoch": 1.9453854130660213, "grad_norm": 0.664203790921547, "learning_rate": 1.9591679064631662e-08, "loss": 0.256, "step": 28140 }, { "epoch": 1.9454545454545453, "grad_norm": 0.7700312005406386, "learning_rate": 1.954220616737945e-08, "loss": 0.4088, "step": 28141 }, { "epoch": 1.9455236778430693, "grad_norm": 0.709851798850788, "learning_rate": 1.9492795691278265e-08, "loss": 0.2129, "step": 28142 }, { "epoch": 1.9455928102315934, "grad_norm": 0.6007294685570473, "learning_rate": 1.9443447636947056e-08, "loss": 0.2221, "step": 28143 }, { "epoch": 1.9456619426201174, "grad_norm": 0.698667672292681, "learning_rate": 1.939416200500366e-08, "loss": 0.339, "step": 28144 }, { "epoch": 1.9457310750086414, "grad_norm": 0.7133952116796336, "learning_rate": 1.9344938796065915e-08, "loss": 0.3388, "step": 28145 }, { "epoch": 1.9458002073971654, "grad_norm": 0.7122058151392036, "learning_rate": 1.929577801075111e-08, "loss": 0.2327, "step": 28146 }, { "epoch": 1.9458693397856894, "grad_norm": 0.7431532355262264, "learning_rate": 1.9246679649675416e-08, "loss": 0.2604, "step": 28147 }, { "epoch": 1.9459384721742137, "grad_norm": 0.6682140106391766, "learning_rate": 1.9197643713453896e-08, "loss": 0.2931, "step": 28148 }, { "epoch": 1.9460076045627377, "grad_norm": 0.7169677192791837, "learning_rate": 1.9148670202701058e-08, "loss": 0.2031, "step": 28149 }, { "epoch": 1.9460767369512617, "grad_norm": 0.7058306997533158, "learning_rate": 1.9099759118031412e-08, "loss": 0.236, "step": 28150 }, { "epoch": 1.9461458693397857, "grad_norm": 0.6688038708932017, "learning_rate": 1.9050910460057247e-08, "loss": 0.3116, "step": 28151 }, { "epoch": 1.9462150017283097, "grad_norm": 0.7415523360053775, "learning_rate": 1.900212422939085e-08, "loss": 0.3562, "step": 28152 }, { "epoch": 1.9462841341168338, "grad_norm": 0.6554488492572967, "learning_rate": 1.8953400426643954e-08, "loss": 0.2391, "step": 28153 }, { "epoch": 1.9463532665053578, "grad_norm": 0.6448213643495125, "learning_rate": 1.890473905242718e-08, "loss": 0.2327, "step": 28154 }, { "epoch": 1.9464223988938818, "grad_norm": 0.7373866355138394, "learning_rate": 1.8856140107350595e-08, "loss": 0.2739, "step": 28155 }, { "epoch": 1.9464915312824058, "grad_norm": 0.7099446761119272, "learning_rate": 1.8807603592022605e-08, "loss": 0.3074, "step": 28156 }, { "epoch": 1.9465606636709298, "grad_norm": 0.6449350013578844, "learning_rate": 1.8759129507052164e-08, "loss": 0.1783, "step": 28157 }, { "epoch": 1.9466297960594539, "grad_norm": 0.716207207165831, "learning_rate": 1.8710717853046567e-08, "loss": 0.2505, "step": 28158 }, { "epoch": 1.9466989284479779, "grad_norm": 0.7056081547893576, "learning_rate": 1.866236863061255e-08, "loss": 0.2747, "step": 28159 }, { "epoch": 1.946768060836502, "grad_norm": 0.6963257030439565, "learning_rate": 1.861408184035629e-08, "loss": 0.2451, "step": 28160 }, { "epoch": 1.946837193225026, "grad_norm": 0.7822098398722807, "learning_rate": 1.8565857482882865e-08, "loss": 0.219, "step": 28161 }, { "epoch": 1.94690632561355, "grad_norm": 0.6518927967557782, "learning_rate": 1.8517695558796234e-08, "loss": 0.2867, "step": 28162 }, { "epoch": 1.946975458002074, "grad_norm": 0.7987886142279664, "learning_rate": 1.8469596068700913e-08, "loss": 0.3114, "step": 28163 }, { "epoch": 1.947044590390598, "grad_norm": 0.7800755625332848, "learning_rate": 1.8421559013199198e-08, "loss": 0.2471, "step": 28164 }, { "epoch": 1.947113722779122, "grad_norm": 0.684536809476325, "learning_rate": 1.837358439289283e-08, "loss": 0.2618, "step": 28165 }, { "epoch": 1.947182855167646, "grad_norm": 0.7301498055578456, "learning_rate": 1.8325672208384104e-08, "loss": 0.291, "step": 28166 }, { "epoch": 1.94725198755617, "grad_norm": 0.7646215445093469, "learning_rate": 1.827782246027199e-08, "loss": 0.2511, "step": 28167 }, { "epoch": 1.947321119944694, "grad_norm": 0.7792911863823043, "learning_rate": 1.8230035149157664e-08, "loss": 0.2344, "step": 28168 }, { "epoch": 1.9473902523332183, "grad_norm": 0.6847000670269044, "learning_rate": 1.818231027563955e-08, "loss": 0.2982, "step": 28169 }, { "epoch": 1.9474593847217423, "grad_norm": 0.7834429319962142, "learning_rate": 1.813464784031549e-08, "loss": 0.3939, "step": 28170 }, { "epoch": 1.9475285171102663, "grad_norm": 0.6786730398843822, "learning_rate": 1.8087047843783345e-08, "loss": 0.2031, "step": 28171 }, { "epoch": 1.9475976494987903, "grad_norm": 0.7390202595885016, "learning_rate": 1.8039510286639305e-08, "loss": 0.2197, "step": 28172 }, { "epoch": 1.9476667818873143, "grad_norm": 0.6759071555936497, "learning_rate": 1.7992035169479005e-08, "loss": 0.2942, "step": 28173 }, { "epoch": 1.9477359142758384, "grad_norm": 0.6714145851667974, "learning_rate": 1.794462249289808e-08, "loss": 0.252, "step": 28174 }, { "epoch": 1.9478050466643624, "grad_norm": 0.7157431458491941, "learning_rate": 1.7897272257490494e-08, "loss": 0.2189, "step": 28175 }, { "epoch": 1.9478741790528864, "grad_norm": 0.7516569448654321, "learning_rate": 1.7849984463849667e-08, "loss": 0.2442, "step": 28176 }, { "epoch": 1.9479433114414104, "grad_norm": 0.7655467031677926, "learning_rate": 1.7802759112567903e-08, "loss": 0.3241, "step": 28177 }, { "epoch": 1.9480124438299344, "grad_norm": 0.6670452064471373, "learning_rate": 1.77555962042375e-08, "loss": 0.2273, "step": 28178 }, { "epoch": 1.9480815762184585, "grad_norm": 0.7086530046614784, "learning_rate": 1.770849573944966e-08, "loss": 0.2521, "step": 28179 }, { "epoch": 1.9481507086069825, "grad_norm": 0.6603374323256941, "learning_rate": 1.766145771879446e-08, "loss": 0.3109, "step": 28180 }, { "epoch": 1.9482198409955065, "grad_norm": 0.7276353381106367, "learning_rate": 1.7614482142861433e-08, "loss": 0.3355, "step": 28181 }, { "epoch": 1.9482889733840305, "grad_norm": 0.7899807882821945, "learning_rate": 1.756756901223955e-08, "loss": 0.2365, "step": 28182 }, { "epoch": 1.9483581057725545, "grad_norm": 0.6952073169634382, "learning_rate": 1.7520718327516672e-08, "loss": 0.2467, "step": 28183 }, { "epoch": 1.9484272381610785, "grad_norm": 0.6928609348954248, "learning_rate": 1.7473930089279555e-08, "loss": 0.2891, "step": 28184 }, { "epoch": 1.9484963705496026, "grad_norm": 0.7177469967885297, "learning_rate": 1.742720429811551e-08, "loss": 0.2234, "step": 28185 }, { "epoch": 1.9485655029381266, "grad_norm": 0.6769619295485616, "learning_rate": 1.7380540954609617e-08, "loss": 0.2141, "step": 28186 }, { "epoch": 1.9486346353266506, "grad_norm": 0.8489211497790172, "learning_rate": 1.733394005934641e-08, "loss": 0.3095, "step": 28187 }, { "epoch": 1.9487037677151746, "grad_norm": 0.7533244024660094, "learning_rate": 1.7287401612910427e-08, "loss": 0.2811, "step": 28188 }, { "epoch": 1.9487729001036986, "grad_norm": 0.6582769611612724, "learning_rate": 1.7240925615885085e-08, "loss": 0.2057, "step": 28189 }, { "epoch": 1.9488420324922227, "grad_norm": 0.6628078593009116, "learning_rate": 1.719451206885214e-08, "loss": 0.2375, "step": 28190 }, { "epoch": 1.9489111648807467, "grad_norm": 0.6549437247659345, "learning_rate": 1.7148160972393913e-08, "loss": 0.3316, "step": 28191 }, { "epoch": 1.9489802972692707, "grad_norm": 0.7441574324514467, "learning_rate": 1.7101872327091595e-08, "loss": 0.2463, "step": 28192 }, { "epoch": 1.9490494296577947, "grad_norm": 0.6612627110629075, "learning_rate": 1.705564613352473e-08, "loss": 0.2262, "step": 28193 }, { "epoch": 1.9491185620463187, "grad_norm": 0.7155976567226978, "learning_rate": 1.7009482392272848e-08, "loss": 0.306, "step": 28194 }, { "epoch": 1.9491876944348427, "grad_norm": 0.8523386825014553, "learning_rate": 1.696338110391438e-08, "loss": 0.2922, "step": 28195 }, { "epoch": 1.9492568268233668, "grad_norm": 0.9858040067745274, "learning_rate": 1.691734226902719e-08, "loss": 0.2678, "step": 28196 }, { "epoch": 1.9493259592118908, "grad_norm": 0.6945109100744379, "learning_rate": 1.6871365888188607e-08, "loss": 0.2247, "step": 28197 }, { "epoch": 1.9493950916004148, "grad_norm": 0.7203940525684918, "learning_rate": 1.6825451961974827e-08, "loss": 0.3075, "step": 28198 }, { "epoch": 1.9494642239889388, "grad_norm": 1.9820923617592796, "learning_rate": 1.6779600490960946e-08, "loss": 0.2768, "step": 28199 }, { "epoch": 1.9495333563774628, "grad_norm": 0.7559621013500111, "learning_rate": 1.673381147572206e-08, "loss": 0.2384, "step": 28200 }, { "epoch": 1.9496024887659869, "grad_norm": 0.7019847288106774, "learning_rate": 1.6688084916831603e-08, "loss": 0.31, "step": 28201 }, { "epoch": 1.9496716211545109, "grad_norm": 0.7476762392736152, "learning_rate": 1.6642420814862447e-08, "loss": 0.3239, "step": 28202 }, { "epoch": 1.9497407535430349, "grad_norm": 0.7972561831018221, "learning_rate": 1.6596819170388024e-08, "loss": 0.2067, "step": 28203 }, { "epoch": 1.949809885931559, "grad_norm": 0.713229686207514, "learning_rate": 1.6551279983978984e-08, "loss": 0.3046, "step": 28204 }, { "epoch": 1.949879018320083, "grad_norm": 0.7923449936740238, "learning_rate": 1.650580325620599e-08, "loss": 0.3689, "step": 28205 }, { "epoch": 1.949948150708607, "grad_norm": 0.710260576736511, "learning_rate": 1.6460388987639686e-08, "loss": 0.2298, "step": 28206 }, { "epoch": 1.950017283097131, "grad_norm": 0.7061163595965415, "learning_rate": 1.6415037178849068e-08, "loss": 0.235, "step": 28207 }, { "epoch": 1.950086415485655, "grad_norm": 0.772640383924655, "learning_rate": 1.6369747830402017e-08, "loss": 0.2961, "step": 28208 }, { "epoch": 1.950155547874179, "grad_norm": 0.7440002993987256, "learning_rate": 1.6324520942866405e-08, "loss": 0.3749, "step": 28209 }, { "epoch": 1.950224680262703, "grad_norm": 0.6377512489108143, "learning_rate": 1.6279356516809007e-08, "loss": 0.2359, "step": 28210 }, { "epoch": 1.950293812651227, "grad_norm": 0.6827497966388922, "learning_rate": 1.623425455279659e-08, "loss": 0.2237, "step": 28211 }, { "epoch": 1.950362945039751, "grad_norm": 0.7867390928060221, "learning_rate": 1.6189215051393148e-08, "loss": 0.3542, "step": 28212 }, { "epoch": 1.950432077428275, "grad_norm": 0.6924794228696387, "learning_rate": 1.6144238013164337e-08, "loss": 0.2622, "step": 28213 }, { "epoch": 1.950501209816799, "grad_norm": 0.7048432356514741, "learning_rate": 1.6099323438673597e-08, "loss": 0.3148, "step": 28214 }, { "epoch": 1.950570342205323, "grad_norm": 0.695008620787656, "learning_rate": 1.6054471328483255e-08, "loss": 0.2837, "step": 28215 }, { "epoch": 1.9506394745938471, "grad_norm": 0.7675951149855744, "learning_rate": 1.6009681683156196e-08, "loss": 0.3458, "step": 28216 }, { "epoch": 1.9507086069823711, "grad_norm": 1.181575225800398, "learning_rate": 1.5964954503253637e-08, "loss": 0.23, "step": 28217 }, { "epoch": 1.9507777393708952, "grad_norm": 0.8903218445710965, "learning_rate": 1.592028978933513e-08, "loss": 0.2692, "step": 28218 }, { "epoch": 1.9508468717594192, "grad_norm": 0.7561630313240056, "learning_rate": 1.5875687541961894e-08, "loss": 0.3184, "step": 28219 }, { "epoch": 1.9509160041479432, "grad_norm": 0.7831517908341221, "learning_rate": 1.5831147761692368e-08, "loss": 0.2882, "step": 28220 }, { "epoch": 1.9509851365364672, "grad_norm": 0.6869576204244754, "learning_rate": 1.578667044908444e-08, "loss": 0.2064, "step": 28221 }, { "epoch": 1.9510542689249912, "grad_norm": 0.6522647741315883, "learning_rate": 1.5742255604696e-08, "loss": 0.2403, "step": 28222 }, { "epoch": 1.9511234013135152, "grad_norm": 0.7905115954925996, "learning_rate": 1.569790322908382e-08, "loss": 0.3739, "step": 28223 }, { "epoch": 1.9511925337020393, "grad_norm": 0.8296630724689422, "learning_rate": 1.5653613322803573e-08, "loss": 0.2514, "step": 28224 }, { "epoch": 1.9512616660905633, "grad_norm": 0.6862111943536703, "learning_rate": 1.5609385886409813e-08, "loss": 0.2571, "step": 28225 }, { "epoch": 1.9513307984790873, "grad_norm": 0.6724060855262869, "learning_rate": 1.5565220920457646e-08, "loss": 0.2949, "step": 28226 }, { "epoch": 1.9513999308676113, "grad_norm": 0.7675174849782784, "learning_rate": 1.5521118425500524e-08, "loss": 0.3088, "step": 28227 }, { "epoch": 1.9514690632561353, "grad_norm": 0.7125326239922045, "learning_rate": 1.5477078402090784e-08, "loss": 0.2207, "step": 28228 }, { "epoch": 1.9515381956446594, "grad_norm": 0.8944452724634587, "learning_rate": 1.5433100850780203e-08, "loss": 0.1921, "step": 28229 }, { "epoch": 1.9516073280331836, "grad_norm": 0.7848439311301091, "learning_rate": 1.5389185772121118e-08, "loss": 0.3477, "step": 28230 }, { "epoch": 1.9516764604217076, "grad_norm": 0.7280465517874415, "learning_rate": 1.5345333166662533e-08, "loss": 0.2144, "step": 28231 }, { "epoch": 1.9517455928102316, "grad_norm": 0.6695780108421328, "learning_rate": 1.5301543034955124e-08, "loss": 0.2271, "step": 28232 }, { "epoch": 1.9518147251987557, "grad_norm": 0.7141971328692797, "learning_rate": 1.5257815377546782e-08, "loss": 0.2912, "step": 28233 }, { "epoch": 1.9518838575872797, "grad_norm": 0.6578698598993714, "learning_rate": 1.521415019498651e-08, "loss": 0.3094, "step": 28234 }, { "epoch": 1.9519529899758037, "grad_norm": 0.6673351299842389, "learning_rate": 1.517054748782054e-08, "loss": 0.2309, "step": 28235 }, { "epoch": 1.9520221223643277, "grad_norm": 0.7126151931271915, "learning_rate": 1.5127007256596772e-08, "loss": 0.2342, "step": 28236 }, { "epoch": 1.9520912547528517, "grad_norm": 0.6870161463415436, "learning_rate": 1.5083529501859206e-08, "loss": 0.3441, "step": 28237 }, { "epoch": 1.9521603871413757, "grad_norm": 0.7462502892669566, "learning_rate": 1.504011422415408e-08, "loss": 0.2929, "step": 28238 }, { "epoch": 1.9522295195298998, "grad_norm": 0.7741308941640394, "learning_rate": 1.499676142402484e-08, "loss": 0.1774, "step": 28239 }, { "epoch": 1.9522986519184238, "grad_norm": 0.8478311229287019, "learning_rate": 1.4953471102015503e-08, "loss": 0.2656, "step": 28240 }, { "epoch": 1.9523677843069478, "grad_norm": 0.6819416441178949, "learning_rate": 1.4910243258667855e-08, "loss": 0.3973, "step": 28241 }, { "epoch": 1.9524369166954718, "grad_norm": 0.7424705545981864, "learning_rate": 1.486707789452424e-08, "loss": 0.291, "step": 28242 }, { "epoch": 1.9525060490839958, "grad_norm": 0.7475258703452313, "learning_rate": 1.4823975010125335e-08, "loss": 0.2617, "step": 28243 }, { "epoch": 1.9525751814725198, "grad_norm": 0.7032880998364477, "learning_rate": 1.4780934606011266e-08, "loss": 0.3215, "step": 28244 }, { "epoch": 1.9526443138610439, "grad_norm": 0.729271621066715, "learning_rate": 1.4737956682721599e-08, "loss": 0.2452, "step": 28245 }, { "epoch": 1.9527134462495679, "grad_norm": 0.6908004516054828, "learning_rate": 1.4695041240795349e-08, "loss": 0.2404, "step": 28246 }, { "epoch": 1.952782578638092, "grad_norm": 0.6853445154157428, "learning_rate": 1.4652188280769864e-08, "loss": 0.2279, "step": 28247 }, { "epoch": 1.952851711026616, "grad_norm": 0.6921991266841456, "learning_rate": 1.460939780318249e-08, "loss": 0.3455, "step": 28248 }, { "epoch": 1.95292084341514, "grad_norm": 0.7713765070479689, "learning_rate": 1.4566669808569467e-08, "loss": 0.2623, "step": 28249 }, { "epoch": 1.9529899758036642, "grad_norm": 0.7178048584120932, "learning_rate": 1.4524004297466476e-08, "loss": 0.2705, "step": 28250 }, { "epoch": 1.9530591081921882, "grad_norm": 0.6599336092419122, "learning_rate": 1.4481401270408091e-08, "loss": 0.2685, "step": 28251 }, { "epoch": 1.9531282405807122, "grad_norm": 0.7095371654926466, "learning_rate": 1.4438860727928327e-08, "loss": 0.304, "step": 28252 }, { "epoch": 1.9531973729692362, "grad_norm": 0.7687584887956076, "learning_rate": 1.4396382670560094e-08, "loss": 0.2885, "step": 28253 }, { "epoch": 1.9532665053577603, "grad_norm": 0.6867885401646913, "learning_rate": 1.4353967098835742e-08, "loss": 0.2161, "step": 28254 }, { "epoch": 1.9533356377462843, "grad_norm": 0.7068741969964338, "learning_rate": 1.4311614013287623e-08, "loss": 0.3207, "step": 28255 }, { "epoch": 1.9534047701348083, "grad_norm": 0.9112797213183175, "learning_rate": 1.4269323414445868e-08, "loss": 0.228, "step": 28256 }, { "epoch": 1.9534739025233323, "grad_norm": 0.6493964106723744, "learning_rate": 1.4227095302841165e-08, "loss": 0.2555, "step": 28257 }, { "epoch": 1.9535430349118563, "grad_norm": 0.7076256861751099, "learning_rate": 1.4184929679001425e-08, "loss": 0.2203, "step": 28258 }, { "epoch": 1.9536121673003803, "grad_norm": 0.6867240515521987, "learning_rate": 1.4142826543456778e-08, "loss": 0.3036, "step": 28259 }, { "epoch": 1.9536812996889044, "grad_norm": 0.7583272530934391, "learning_rate": 1.4100785896734026e-08, "loss": 0.2036, "step": 28260 }, { "epoch": 1.9537504320774284, "grad_norm": 1.3331508862880852, "learning_rate": 1.405880773935997e-08, "loss": 0.2713, "step": 28261 }, { "epoch": 1.9538195644659524, "grad_norm": 0.9637532730731048, "learning_rate": 1.4016892071860855e-08, "loss": 0.3122, "step": 28262 }, { "epoch": 1.9538886968544764, "grad_norm": 0.690834608081229, "learning_rate": 1.3975038894762372e-08, "loss": 0.2367, "step": 28263 }, { "epoch": 1.9539578292430004, "grad_norm": 0.7449957050321906, "learning_rate": 1.3933248208588546e-08, "loss": 0.2575, "step": 28264 }, { "epoch": 1.9540269616315245, "grad_norm": 0.7002713420851754, "learning_rate": 1.3891520013863402e-08, "loss": 0.2458, "step": 28265 }, { "epoch": 1.9540960940200485, "grad_norm": 0.7438002048880982, "learning_rate": 1.3849854311110412e-08, "loss": 0.3293, "step": 28266 }, { "epoch": 1.9541652264085725, "grad_norm": 0.7292062072897738, "learning_rate": 1.3808251100850822e-08, "loss": 0.2261, "step": 28267 }, { "epoch": 1.9542343587970965, "grad_norm": 0.7218951012210314, "learning_rate": 1.3766710383606996e-08, "loss": 0.2205, "step": 28268 }, { "epoch": 1.9543034911856205, "grad_norm": 0.7138146268120596, "learning_rate": 1.3725232159898516e-08, "loss": 0.3546, "step": 28269 }, { "epoch": 1.9543726235741445, "grad_norm": 0.7417436783332104, "learning_rate": 1.3683816430246633e-08, "loss": 0.2837, "step": 28270 }, { "epoch": 1.9544417559626686, "grad_norm": 0.6752902765947608, "learning_rate": 1.3642463195168709e-08, "loss": 0.2425, "step": 28271 }, { "epoch": 1.9545108883511926, "grad_norm": 0.6659315287884208, "learning_rate": 1.3601172455184886e-08, "loss": 0.2295, "step": 28272 }, { "epoch": 1.9545800207397166, "grad_norm": 0.7353798208508054, "learning_rate": 1.3559944210810861e-08, "loss": 0.3222, "step": 28273 }, { "epoch": 1.9546491531282406, "grad_norm": 0.7757339368664616, "learning_rate": 1.3518778462564552e-08, "loss": 0.2399, "step": 28274 }, { "epoch": 1.9547182855167646, "grad_norm": 0.635454403084104, "learning_rate": 1.3477675210961661e-08, "loss": 0.2135, "step": 28275 }, { "epoch": 1.9547874179052886, "grad_norm": 0.7767873460450061, "learning_rate": 1.3436634456517328e-08, "loss": 0.2796, "step": 28276 }, { "epoch": 1.9548565502938127, "grad_norm": 0.7124198988477138, "learning_rate": 1.3395656199746142e-08, "loss": 0.2945, "step": 28277 }, { "epoch": 1.9549256826823367, "grad_norm": 0.8640929551808783, "learning_rate": 1.335474044116103e-08, "loss": 0.203, "step": 28278 }, { "epoch": 1.9549948150708607, "grad_norm": 0.7176203185027807, "learning_rate": 1.3313887181274909e-08, "loss": 0.2297, "step": 28279 }, { "epoch": 1.9550639474593847, "grad_norm": 0.6831491476365327, "learning_rate": 1.3273096420600707e-08, "loss": 0.3131, "step": 28280 }, { "epoch": 1.9551330798479087, "grad_norm": 0.7693318733889796, "learning_rate": 1.3232368159648567e-08, "loss": 0.2914, "step": 28281 }, { "epoch": 1.9552022122364328, "grad_norm": 0.7197240859526436, "learning_rate": 1.3191702398929752e-08, "loss": 0.2061, "step": 28282 }, { "epoch": 1.9552713446249568, "grad_norm": 1.0692265315275287, "learning_rate": 1.3151099138953293e-08, "loss": 0.239, "step": 28283 }, { "epoch": 1.9553404770134808, "grad_norm": 0.7290834021765628, "learning_rate": 1.3110558380228233e-08, "loss": 0.3486, "step": 28284 }, { "epoch": 1.9554096094020048, "grad_norm": 0.7460851857677294, "learning_rate": 1.3070080123263051e-08, "loss": 0.2755, "step": 28285 }, { "epoch": 1.9554787417905288, "grad_norm": 0.658413302708512, "learning_rate": 1.302966436856512e-08, "loss": 0.2117, "step": 28286 }, { "epoch": 1.9555478741790528, "grad_norm": 0.684158271607603, "learning_rate": 1.2989311116640147e-08, "loss": 0.3606, "step": 28287 }, { "epoch": 1.9556170065675769, "grad_norm": 0.7772929898101867, "learning_rate": 1.2949020367994947e-08, "loss": 0.2853, "step": 28288 }, { "epoch": 1.9556861389561009, "grad_norm": 0.7022085123097564, "learning_rate": 1.2908792123133562e-08, "loss": 0.2, "step": 28289 }, { "epoch": 1.955755271344625, "grad_norm": 0.826558782178007, "learning_rate": 1.2868626382561144e-08, "loss": 0.2123, "step": 28290 }, { "epoch": 1.955824403733149, "grad_norm": 0.7320064263367044, "learning_rate": 1.282852314678007e-08, "loss": 0.4075, "step": 28291 }, { "epoch": 1.955893536121673, "grad_norm": 0.7472312618054627, "learning_rate": 1.2788482416293824e-08, "loss": 0.287, "step": 28292 }, { "epoch": 1.955962668510197, "grad_norm": 0.7613429991822964, "learning_rate": 1.2748504191603117e-08, "loss": 0.2197, "step": 28293 }, { "epoch": 1.956031800898721, "grad_norm": 0.7045725716223763, "learning_rate": 1.2708588473210326e-08, "loss": 0.3006, "step": 28294 }, { "epoch": 1.956100933287245, "grad_norm": 0.7583401310729908, "learning_rate": 1.266873526161505e-08, "loss": 0.2638, "step": 28295 }, { "epoch": 1.956170065675769, "grad_norm": 0.7166976390112064, "learning_rate": 1.262894455731689e-08, "loss": 0.2234, "step": 28296 }, { "epoch": 1.956239198064293, "grad_norm": 0.7232745524405112, "learning_rate": 1.2589216360814893e-08, "loss": 0.1997, "step": 28297 }, { "epoch": 1.956308330452817, "grad_norm": 0.7407619540070375, "learning_rate": 1.2549550672606437e-08, "loss": 0.3251, "step": 28298 }, { "epoch": 1.956377462841341, "grad_norm": 1.026034754215155, "learning_rate": 1.2509947493188345e-08, "loss": 0.2547, "step": 28299 }, { "epoch": 1.956446595229865, "grad_norm": 0.7232748647643009, "learning_rate": 1.2470406823058001e-08, "loss": 0.2482, "step": 28300 }, { "epoch": 1.956515727618389, "grad_norm": 2.3617299281505137, "learning_rate": 1.2430928662710562e-08, "loss": 0.328, "step": 28301 }, { "epoch": 1.9565848600069131, "grad_norm": 0.7390263630811524, "learning_rate": 1.2391513012640633e-08, "loss": 0.2987, "step": 28302 }, { "epoch": 1.9566539923954371, "grad_norm": 0.6604043314922534, "learning_rate": 1.2352159873342262e-08, "loss": 0.1999, "step": 28303 }, { "epoch": 1.9567231247839612, "grad_norm": 0.7342792891410215, "learning_rate": 1.231286924530839e-08, "loss": 0.2759, "step": 28304 }, { "epoch": 1.9567922571724852, "grad_norm": 0.7082943024854009, "learning_rate": 1.2273641129031954e-08, "loss": 0.3146, "step": 28305 }, { "epoch": 1.9568613895610092, "grad_norm": 0.6984801754394255, "learning_rate": 1.2234475525004785e-08, "loss": 0.206, "step": 28306 }, { "epoch": 1.9569305219495332, "grad_norm": 0.6767351519241637, "learning_rate": 1.2195372433717045e-08, "loss": 0.2292, "step": 28307 }, { "epoch": 1.9569996543380572, "grad_norm": 0.6284551250373475, "learning_rate": 1.2156331855659453e-08, "loss": 0.2389, "step": 28308 }, { "epoch": 1.9570687867265812, "grad_norm": 0.8196792448528684, "learning_rate": 1.2117353791320507e-08, "loss": 0.4226, "step": 28309 }, { "epoch": 1.9571379191151053, "grad_norm": 0.707955213160823, "learning_rate": 1.2078438241189261e-08, "loss": 0.2402, "step": 28310 }, { "epoch": 1.9572070515036295, "grad_norm": 0.6656616835129475, "learning_rate": 1.2039585205753657e-08, "loss": 0.2943, "step": 28311 }, { "epoch": 1.9572761838921535, "grad_norm": 0.6357582942285642, "learning_rate": 1.2000794685501082e-08, "loss": 0.3013, "step": 28312 }, { "epoch": 1.9573453162806775, "grad_norm": 0.749722024937695, "learning_rate": 1.196206668091615e-08, "loss": 0.2843, "step": 28313 }, { "epoch": 1.9574144486692016, "grad_norm": 0.6819187941534359, "learning_rate": 1.1923401192485695e-08, "loss": 0.2229, "step": 28314 }, { "epoch": 1.9574835810577256, "grad_norm": 0.7386644841802734, "learning_rate": 1.1884798220693771e-08, "loss": 0.262, "step": 28315 }, { "epoch": 1.9575527134462496, "grad_norm": 0.7768718005291144, "learning_rate": 1.1846257766023882e-08, "loss": 0.3332, "step": 28316 }, { "epoch": 1.9576218458347736, "grad_norm": 0.9550392860736122, "learning_rate": 1.1807779828959532e-08, "loss": 0.2769, "step": 28317 }, { "epoch": 1.9576909782232976, "grad_norm": 0.7557444779991211, "learning_rate": 1.1769364409983109e-08, "loss": 0.242, "step": 28318 }, { "epoch": 1.9577601106118216, "grad_norm": 0.7704372061635691, "learning_rate": 1.1731011509575342e-08, "loss": 0.285, "step": 28319 }, { "epoch": 1.9578292430003457, "grad_norm": 0.7046888141552121, "learning_rate": 1.1692721128217511e-08, "loss": 0.2178, "step": 28320 }, { "epoch": 1.9578983753888697, "grad_norm": 0.6707783224714983, "learning_rate": 1.1654493266389233e-08, "loss": 0.2366, "step": 28321 }, { "epoch": 1.9579675077773937, "grad_norm": 0.6668913063090671, "learning_rate": 1.1616327924570126e-08, "loss": 0.261, "step": 28322 }, { "epoch": 1.9580366401659177, "grad_norm": 0.7282156729359995, "learning_rate": 1.157822510323814e-08, "loss": 0.3267, "step": 28323 }, { "epoch": 1.9581057725544417, "grad_norm": 0.6906855609923752, "learning_rate": 1.1540184802871224e-08, "loss": 0.2556, "step": 28324 }, { "epoch": 1.9581749049429658, "grad_norm": 0.6362331891104289, "learning_rate": 1.1502207023945111e-08, "loss": 0.2172, "step": 28325 }, { "epoch": 1.9582440373314898, "grad_norm": 0.7432573763387984, "learning_rate": 1.1464291766936641e-08, "loss": 0.2586, "step": 28326 }, { "epoch": 1.9583131697200138, "grad_norm": 0.7217071773567235, "learning_rate": 1.142643903232099e-08, "loss": 0.3042, "step": 28327 }, { "epoch": 1.9583823021085378, "grad_norm": 0.7394751160273977, "learning_rate": 1.1388648820572224e-08, "loss": 0.2423, "step": 28328 }, { "epoch": 1.9584514344970618, "grad_norm": 0.6250130673836418, "learning_rate": 1.1350921132164405e-08, "loss": 0.2523, "step": 28329 }, { "epoch": 1.9585205668855858, "grad_norm": 0.7450886669794412, "learning_rate": 1.1313255967569936e-08, "loss": 0.297, "step": 28330 }, { "epoch": 1.95858969927411, "grad_norm": 0.7618230277477411, "learning_rate": 1.1275653327261216e-08, "loss": 0.3027, "step": 28331 }, { "epoch": 1.958658831662634, "grad_norm": 0.6947646830977392, "learning_rate": 1.1238113211708979e-08, "loss": 0.1748, "step": 28332 }, { "epoch": 1.9587279640511581, "grad_norm": 0.7062224425739602, "learning_rate": 1.1200635621384515e-08, "loss": 0.2737, "step": 28333 }, { "epoch": 1.9587970964396821, "grad_norm": 0.6978178349112424, "learning_rate": 1.1163220556757447e-08, "loss": 0.323, "step": 28334 }, { "epoch": 1.9588662288282062, "grad_norm": 0.6570094784415795, "learning_rate": 1.1125868018296293e-08, "loss": 0.1992, "step": 28335 }, { "epoch": 1.9589353612167302, "grad_norm": 0.6767733550604299, "learning_rate": 1.1088578006469008e-08, "loss": 0.2863, "step": 28336 }, { "epoch": 1.9590044936052542, "grad_norm": 0.6817621513708437, "learning_rate": 1.1051350521743554e-08, "loss": 0.3244, "step": 28337 }, { "epoch": 1.9590736259937782, "grad_norm": 0.8485741835235948, "learning_rate": 1.1014185564586222e-08, "loss": 0.3065, "step": 28338 }, { "epoch": 1.9591427583823022, "grad_norm": 0.6943802715558728, "learning_rate": 1.0977083135462751e-08, "loss": 0.2377, "step": 28339 }, { "epoch": 1.9592118907708262, "grad_norm": 0.7443632734068434, "learning_rate": 1.0940043234837772e-08, "loss": 0.2758, "step": 28340 }, { "epoch": 1.9592810231593503, "grad_norm": 0.7666684853439627, "learning_rate": 1.090306586317702e-08, "loss": 0.2988, "step": 28341 }, { "epoch": 1.9593501555478743, "grad_norm": 0.7343563412901072, "learning_rate": 1.0866151020941796e-08, "loss": 0.2437, "step": 28342 }, { "epoch": 1.9594192879363983, "grad_norm": 0.6782577299349773, "learning_rate": 1.0829298708596724e-08, "loss": 0.202, "step": 28343 }, { "epoch": 1.9594884203249223, "grad_norm": 0.7463336981612966, "learning_rate": 1.0792508926602552e-08, "loss": 0.2983, "step": 28344 }, { "epoch": 1.9595575527134463, "grad_norm": 0.7284953220005423, "learning_rate": 1.0755781675420573e-08, "loss": 0.2659, "step": 28345 }, { "epoch": 1.9596266851019704, "grad_norm": 0.6504475605262293, "learning_rate": 1.0719116955510978e-08, "loss": 0.1875, "step": 28346 }, { "epoch": 1.9596958174904944, "grad_norm": 0.6900794073153729, "learning_rate": 1.068251476733395e-08, "loss": 0.2119, "step": 28347 }, { "epoch": 1.9597649498790184, "grad_norm": 0.6923816937258561, "learning_rate": 1.0645975111347462e-08, "loss": 0.3094, "step": 28348 }, { "epoch": 1.9598340822675424, "grad_norm": 0.7253572257690866, "learning_rate": 1.0609497988009477e-08, "loss": 0.243, "step": 28349 }, { "epoch": 1.9599032146560664, "grad_norm": 0.6402640544681575, "learning_rate": 1.0573083397777961e-08, "loss": 0.2118, "step": 28350 }, { "epoch": 1.9599723470445904, "grad_norm": 0.6763399711505855, "learning_rate": 1.0536731341108664e-08, "loss": 0.2595, "step": 28351 }, { "epoch": 1.9600414794331145, "grad_norm": 0.748214837785749, "learning_rate": 1.050044181845733e-08, "loss": 0.321, "step": 28352 }, { "epoch": 1.9601106118216385, "grad_norm": 0.6620500622966314, "learning_rate": 1.0464214830278597e-08, "loss": 0.2038, "step": 28353 }, { "epoch": 1.9601797442101625, "grad_norm": 0.6977303677792936, "learning_rate": 1.0428050377027099e-08, "loss": 0.2416, "step": 28354 }, { "epoch": 1.9602488765986865, "grad_norm": 0.7477963995216549, "learning_rate": 1.0391948459155254e-08, "loss": 0.3767, "step": 28355 }, { "epoch": 1.9603180089872105, "grad_norm": 0.7049531092881571, "learning_rate": 1.0355909077116033e-08, "loss": 0.3436, "step": 28356 }, { "epoch": 1.9603871413757346, "grad_norm": 0.6812215760571101, "learning_rate": 1.0319932231361296e-08, "loss": 0.2243, "step": 28357 }, { "epoch": 1.9604562737642586, "grad_norm": 0.7586237518493992, "learning_rate": 1.0284017922341793e-08, "loss": 0.2489, "step": 28358 }, { "epoch": 1.9605254061527826, "grad_norm": 0.6682964836291527, "learning_rate": 1.0248166150507721e-08, "loss": 0.3213, "step": 28359 }, { "epoch": 1.9605945385413066, "grad_norm": 0.7791866783930557, "learning_rate": 1.021237691630761e-08, "loss": 0.2365, "step": 28360 }, { "epoch": 1.9606636709298306, "grad_norm": 0.6926000110975794, "learning_rate": 1.0176650220191098e-08, "loss": 0.2449, "step": 28361 }, { "epoch": 1.9607328033183546, "grad_norm": 0.7057333069438744, "learning_rate": 1.0140986062605606e-08, "loss": 0.2997, "step": 28362 }, { "epoch": 1.9608019357068787, "grad_norm": 0.674033285907999, "learning_rate": 1.0105384443998e-08, "loss": 0.3144, "step": 28363 }, { "epoch": 1.9608710680954027, "grad_norm": 0.8013765936157966, "learning_rate": 1.006984536481459e-08, "loss": 0.1871, "step": 28364 }, { "epoch": 1.9609402004839267, "grad_norm": 0.6471759389528244, "learning_rate": 1.0034368825500573e-08, "loss": 0.2449, "step": 28365 }, { "epoch": 1.9610093328724507, "grad_norm": 0.7004263747259277, "learning_rate": 9.998954826500595e-09, "loss": 0.3776, "step": 28366 }, { "epoch": 1.9610784652609747, "grad_norm": 0.6788160523159805, "learning_rate": 9.9636033682593e-09, "loss": 0.2651, "step": 28367 }, { "epoch": 1.9611475976494988, "grad_norm": 0.6489561308264935, "learning_rate": 9.928314451218558e-09, "loss": 0.2495, "step": 28368 }, { "epoch": 1.9612167300380228, "grad_norm": 0.7293174746322403, "learning_rate": 9.893088075821345e-09, "loss": 0.3418, "step": 28369 }, { "epoch": 1.9612858624265468, "grad_norm": 0.7337515682880816, "learning_rate": 9.857924242508976e-09, "loss": 0.3741, "step": 28370 }, { "epoch": 1.9613549948150708, "grad_norm": 0.7640408918633411, "learning_rate": 9.82282295172221e-09, "loss": 0.2575, "step": 28371 }, { "epoch": 1.9614241272035948, "grad_norm": 0.7401726885521014, "learning_rate": 9.787784203901252e-09, "loss": 0.2729, "step": 28372 }, { "epoch": 1.9614932595921188, "grad_norm": 0.6892433735685567, "learning_rate": 9.752807999484636e-09, "loss": 0.4096, "step": 28373 }, { "epoch": 1.9615623919806429, "grad_norm": 0.7013436635537665, "learning_rate": 9.71789433891146e-09, "loss": 0.2532, "step": 28374 }, { "epoch": 1.9616315243691669, "grad_norm": 0.6763610209704288, "learning_rate": 9.683043222618594e-09, "loss": 0.2628, "step": 28375 }, { "epoch": 1.961700656757691, "grad_norm": 0.6973452463973924, "learning_rate": 9.64825465104291e-09, "loss": 0.2946, "step": 28376 }, { "epoch": 1.961769789146215, "grad_norm": 0.6785241228311241, "learning_rate": 9.613528624621282e-09, "loss": 0.2727, "step": 28377 }, { "epoch": 1.961838921534739, "grad_norm": 0.8148999420549679, "learning_rate": 9.578865143787808e-09, "loss": 0.2594, "step": 28378 }, { "epoch": 1.961908053923263, "grad_norm": 0.7771569808893704, "learning_rate": 9.544264208978249e-09, "loss": 0.2151, "step": 28379 }, { "epoch": 1.961977186311787, "grad_norm": 0.7083457254978608, "learning_rate": 9.509725820624482e-09, "loss": 0.3162, "step": 28380 }, { "epoch": 1.962046318700311, "grad_norm": 0.774055557134466, "learning_rate": 9.47524997916116e-09, "loss": 0.2554, "step": 28381 }, { "epoch": 1.962115451088835, "grad_norm": 0.7377557175448239, "learning_rate": 9.44083668501905e-09, "loss": 0.191, "step": 28382 }, { "epoch": 1.962184583477359, "grad_norm": 0.6887982984117043, "learning_rate": 9.406485938630027e-09, "loss": 0.2544, "step": 28383 }, { "epoch": 1.962253715865883, "grad_norm": 0.7712496096753614, "learning_rate": 9.37219774042486e-09, "loss": 0.3195, "step": 28384 }, { "epoch": 1.962322848254407, "grad_norm": 0.8211656574098309, "learning_rate": 9.337972090833203e-09, "loss": 0.2232, "step": 28385 }, { "epoch": 1.962391980642931, "grad_norm": 0.7345544725178947, "learning_rate": 9.303808990283603e-09, "loss": 0.2669, "step": 28386 }, { "epoch": 1.962461113031455, "grad_norm": 0.8647725925510172, "learning_rate": 9.26970843920405e-09, "loss": 0.4195, "step": 28387 }, { "epoch": 1.9625302454199791, "grad_norm": 0.7534092120012824, "learning_rate": 9.235670438023092e-09, "loss": 0.2522, "step": 28388 }, { "epoch": 1.9625993778085031, "grad_norm": 0.7633742437772689, "learning_rate": 9.201694987165943e-09, "loss": 0.2157, "step": 28389 }, { "epoch": 1.9626685101970271, "grad_norm": 0.7113218251027421, "learning_rate": 9.167782087059485e-09, "loss": 0.2408, "step": 28390 }, { "epoch": 1.9627376425855512, "grad_norm": 0.9898451559124184, "learning_rate": 9.13393173812782e-09, "loss": 0.3706, "step": 28391 }, { "epoch": 1.9628067749740754, "grad_norm": 0.7186913148122437, "learning_rate": 9.100143940796169e-09, "loss": 0.2409, "step": 28392 }, { "epoch": 1.9628759073625994, "grad_norm": 0.6630162505699729, "learning_rate": 9.066418695487522e-09, "loss": 0.1864, "step": 28393 }, { "epoch": 1.9629450397511234, "grad_norm": 0.7082142147643622, "learning_rate": 9.03275600262432e-09, "loss": 0.2719, "step": 28394 }, { "epoch": 1.9630141721396475, "grad_norm": 0.659893938273702, "learning_rate": 8.999155862629006e-09, "loss": 0.2623, "step": 28395 }, { "epoch": 1.9630833045281715, "grad_norm": 0.6815893253037817, "learning_rate": 8.96561827592235e-09, "loss": 0.2198, "step": 28396 }, { "epoch": 1.9631524369166955, "grad_norm": 1.0731989474658985, "learning_rate": 8.932143242925128e-09, "loss": 0.2787, "step": 28397 }, { "epoch": 1.9632215693052195, "grad_norm": 0.7235177263259328, "learning_rate": 8.898730764055897e-09, "loss": 0.318, "step": 28398 }, { "epoch": 1.9632907016937435, "grad_norm": 0.7678949962389415, "learning_rate": 8.865380839734316e-09, "loss": 0.2583, "step": 28399 }, { "epoch": 1.9633598340822676, "grad_norm": 0.75913880752834, "learning_rate": 8.832093470378389e-09, "loss": 0.2328, "step": 28400 }, { "epoch": 1.9634289664707916, "grad_norm": 0.7386899813862127, "learning_rate": 8.798868656405002e-09, "loss": 0.2841, "step": 28401 }, { "epoch": 1.9634980988593156, "grad_norm": 0.7645455799197922, "learning_rate": 8.765706398229934e-09, "loss": 0.3368, "step": 28402 }, { "epoch": 1.9635672312478396, "grad_norm": 0.7511171060558535, "learning_rate": 8.732606696270074e-09, "loss": 0.2231, "step": 28403 }, { "epoch": 1.9636363636363636, "grad_norm": 0.7248353754434039, "learning_rate": 8.699569550939534e-09, "loss": 0.2792, "step": 28404 }, { "epoch": 1.9637054960248876, "grad_norm": 0.6796588477442402, "learning_rate": 8.66659496265243e-09, "loss": 0.2972, "step": 28405 }, { "epoch": 1.9637746284134117, "grad_norm": 0.7430029193566071, "learning_rate": 8.633682931822318e-09, "loss": 0.2304, "step": 28406 }, { "epoch": 1.9638437608019357, "grad_norm": 0.6549212039172773, "learning_rate": 8.60083345886109e-09, "loss": 0.256, "step": 28407 }, { "epoch": 1.9639128931904597, "grad_norm": 0.7107712307797356, "learning_rate": 8.568046544181196e-09, "loss": 0.3027, "step": 28408 }, { "epoch": 1.9639820255789837, "grad_norm": 0.7561512298495228, "learning_rate": 8.535322188193418e-09, "loss": 0.3156, "step": 28409 }, { "epoch": 1.9640511579675077, "grad_norm": 0.7294780885926986, "learning_rate": 8.502660391307426e-09, "loss": 0.2421, "step": 28410 }, { "epoch": 1.9641202903560318, "grad_norm": 0.6847618070586532, "learning_rate": 8.470061153932896e-09, "loss": 0.2255, "step": 28411 }, { "epoch": 1.964189422744556, "grad_norm": 0.7710415412689341, "learning_rate": 8.437524476478386e-09, "loss": 0.3529, "step": 28412 }, { "epoch": 1.96425855513308, "grad_norm": 0.7955847531135255, "learning_rate": 8.405050359351351e-09, "loss": 0.2285, "step": 28413 }, { "epoch": 1.964327687521604, "grad_norm": 0.7323959770576773, "learning_rate": 8.372638802959798e-09, "loss": 0.2345, "step": 28414 }, { "epoch": 1.964396819910128, "grad_norm": 0.6909590866509048, "learning_rate": 8.340289807708956e-09, "loss": 0.2143, "step": 28415 }, { "epoch": 1.964465952298652, "grad_norm": 0.724382936211319, "learning_rate": 8.30800337400406e-09, "loss": 0.4048, "step": 28416 }, { "epoch": 1.964535084687176, "grad_norm": 0.7460614540111632, "learning_rate": 8.27577950225089e-09, "loss": 0.2782, "step": 28417 }, { "epoch": 1.9646042170757, "grad_norm": 0.716172291897701, "learning_rate": 8.243618192852465e-09, "loss": 0.2353, "step": 28418 }, { "epoch": 1.9646733494642241, "grad_norm": 0.8072305039632389, "learning_rate": 8.211519446212346e-09, "loss": 0.3032, "step": 28419 }, { "epoch": 1.9647424818527481, "grad_norm": 0.6522081394515092, "learning_rate": 8.179483262732435e-09, "loss": 0.2584, "step": 28420 }, { "epoch": 1.9648116142412722, "grad_norm": 0.6614710274279286, "learning_rate": 8.147509642814078e-09, "loss": 0.2447, "step": 28421 }, { "epoch": 1.9648807466297962, "grad_norm": 0.7236762640703165, "learning_rate": 8.115598586858619e-09, "loss": 0.2205, "step": 28422 }, { "epoch": 1.9649498790183202, "grad_norm": 0.6947815759267969, "learning_rate": 8.08375009526574e-09, "loss": 0.3467, "step": 28423 }, { "epoch": 1.9650190114068442, "grad_norm": 0.9905364112628083, "learning_rate": 8.051964168434568e-09, "loss": 0.2839, "step": 28424 }, { "epoch": 1.9650881437953682, "grad_norm": 0.7953332514025304, "learning_rate": 8.020240806763114e-09, "loss": 0.2176, "step": 28425 }, { "epoch": 1.9651572761838922, "grad_norm": 0.705239828698516, "learning_rate": 7.988580010649948e-09, "loss": 0.2553, "step": 28426 }, { "epoch": 1.9652264085724163, "grad_norm": 0.9971547276249452, "learning_rate": 7.95698178049087e-09, "loss": 0.3175, "step": 28427 }, { "epoch": 1.9652955409609403, "grad_norm": 0.6621764495899009, "learning_rate": 7.925446116682223e-09, "loss": 0.2372, "step": 28428 }, { "epoch": 1.9653646733494643, "grad_norm": 0.6706977402273095, "learning_rate": 7.893973019619805e-09, "loss": 0.2342, "step": 28429 }, { "epoch": 1.9654338057379883, "grad_norm": 0.6997375026775267, "learning_rate": 7.86256248969719e-09, "loss": 0.301, "step": 28430 }, { "epoch": 1.9655029381265123, "grad_norm": 0.7040290244420572, "learning_rate": 7.831214527309062e-09, "loss": 0.2443, "step": 28431 }, { "epoch": 1.9655720705150364, "grad_norm": 0.6640927205995023, "learning_rate": 7.799929132847328e-09, "loss": 0.2217, "step": 28432 }, { "epoch": 1.9656412029035604, "grad_norm": 0.685683453689678, "learning_rate": 7.768706306704455e-09, "loss": 0.2418, "step": 28433 }, { "epoch": 1.9657103352920844, "grad_norm": 0.6995289910571842, "learning_rate": 7.737546049271793e-09, "loss": 0.3447, "step": 28434 }, { "epoch": 1.9657794676806084, "grad_norm": 0.8292958273945966, "learning_rate": 7.706448360940145e-09, "loss": 0.2637, "step": 28435 }, { "epoch": 1.9658486000691324, "grad_norm": 0.6618257505537056, "learning_rate": 7.67541324209864e-09, "loss": 0.2225, "step": 28436 }, { "epoch": 1.9659177324576564, "grad_norm": 0.6669660930952417, "learning_rate": 7.644440693136967e-09, "loss": 0.3111, "step": 28437 }, { "epoch": 1.9659868648461805, "grad_norm": 0.7160578366112519, "learning_rate": 7.613530714442596e-09, "loss": 0.2818, "step": 28438 }, { "epoch": 1.9660559972347045, "grad_norm": 1.2782833836412562, "learning_rate": 7.582683306403548e-09, "loss": 0.2748, "step": 28439 }, { "epoch": 1.9661251296232285, "grad_norm": 0.7107162777776983, "learning_rate": 7.551898469406738e-09, "loss": 0.2469, "step": 28440 }, { "epoch": 1.9661942620117525, "grad_norm": 0.6877669479566826, "learning_rate": 7.521176203836856e-09, "loss": 0.3262, "step": 28441 }, { "epoch": 1.9662633944002765, "grad_norm": 0.693203879944901, "learning_rate": 7.490516510079704e-09, "loss": 0.242, "step": 28442 }, { "epoch": 1.9663325267888005, "grad_norm": 0.7298028496459475, "learning_rate": 7.45991938851942e-09, "loss": 0.2641, "step": 28443 }, { "epoch": 1.9664016591773246, "grad_norm": 0.7146839721481343, "learning_rate": 7.429384839539033e-09, "loss": 0.3559, "step": 28444 }, { "epoch": 1.9664707915658486, "grad_norm": 0.7460435861016487, "learning_rate": 7.398912863522123e-09, "loss": 0.242, "step": 28445 }, { "epoch": 1.9665399239543726, "grad_norm": 0.6700623131896554, "learning_rate": 7.368503460850051e-09, "loss": 0.2455, "step": 28446 }, { "epoch": 1.9666090563428966, "grad_norm": 0.706899761011641, "learning_rate": 7.338156631903625e-09, "loss": 0.216, "step": 28447 }, { "epoch": 1.9666781887314206, "grad_norm": 0.6588493130741704, "learning_rate": 7.307872377064207e-09, "loss": 0.3137, "step": 28448 }, { "epoch": 1.9667473211199447, "grad_norm": 0.737477490729147, "learning_rate": 7.277650696710381e-09, "loss": 0.2559, "step": 28449 }, { "epoch": 1.9668164535084687, "grad_norm": 0.5904072091447736, "learning_rate": 7.247491591221845e-09, "loss": 0.1693, "step": 28450 }, { "epoch": 1.9668855858969927, "grad_norm": 0.8638732936522253, "learning_rate": 7.217395060975518e-09, "loss": 0.3082, "step": 28451 }, { "epoch": 1.9669547182855167, "grad_norm": 0.9483298507137395, "learning_rate": 7.187361106348878e-09, "loss": 0.3474, "step": 28452 }, { "epoch": 1.9670238506740407, "grad_norm": 0.9810526642869013, "learning_rate": 7.157389727719399e-09, "loss": 0.2261, "step": 28453 }, { "epoch": 1.9670929830625647, "grad_norm": 0.7780064096476174, "learning_rate": 7.127480925461227e-09, "loss": 0.2235, "step": 28454 }, { "epoch": 1.9671621154510888, "grad_norm": 0.7319214811681215, "learning_rate": 7.097634699950173e-09, "loss": 0.3154, "step": 28455 }, { "epoch": 1.9672312478396128, "grad_norm": 1.869224909256369, "learning_rate": 7.0678510515598266e-09, "loss": 0.2984, "step": 28456 }, { "epoch": 1.9673003802281368, "grad_norm": 0.7438872587627771, "learning_rate": 7.038129980663777e-09, "loss": 0.2189, "step": 28457 }, { "epoch": 1.9673695126166608, "grad_norm": 0.7477223550172677, "learning_rate": 7.00847148763506e-09, "loss": 0.2723, "step": 28458 }, { "epoch": 1.9674386450051848, "grad_norm": 0.7422374471290432, "learning_rate": 6.978875572843935e-09, "loss": 0.3691, "step": 28459 }, { "epoch": 1.9675077773937089, "grad_norm": 0.7887198260647624, "learning_rate": 6.949342236662326e-09, "loss": 0.2659, "step": 28460 }, { "epoch": 1.9675769097822329, "grad_norm": 0.6589227256908203, "learning_rate": 6.919871479460494e-09, "loss": 0.1952, "step": 28461 }, { "epoch": 1.967646042170757, "grad_norm": 0.6837678842698564, "learning_rate": 6.890463301607031e-09, "loss": 0.3455, "step": 28462 }, { "epoch": 1.967715174559281, "grad_norm": 0.7187652563961238, "learning_rate": 6.861117703471643e-09, "loss": 0.2503, "step": 28463 }, { "epoch": 1.967784306947805, "grad_norm": 0.9926442747116653, "learning_rate": 6.831834685421257e-09, "loss": 0.2207, "step": 28464 }, { "epoch": 1.967853439336329, "grad_norm": 0.7980647312676078, "learning_rate": 6.802614247822803e-09, "loss": 0.2959, "step": 28465 }, { "epoch": 1.967922571724853, "grad_norm": 0.7328896068199647, "learning_rate": 6.773456391043209e-09, "loss": 0.3748, "step": 28466 }, { "epoch": 1.967991704113377, "grad_norm": 0.7142492342144974, "learning_rate": 6.744361115447184e-09, "loss": 0.2302, "step": 28467 }, { "epoch": 1.968060836501901, "grad_norm": 0.6954040951011281, "learning_rate": 6.715328421399992e-09, "loss": 0.2296, "step": 28468 }, { "epoch": 1.968129968890425, "grad_norm": 0.7570431332594192, "learning_rate": 6.686358309265228e-09, "loss": 0.243, "step": 28469 }, { "epoch": 1.968199101278949, "grad_norm": 0.7141252764194195, "learning_rate": 6.6574507794053835e-09, "loss": 0.3333, "step": 28470 }, { "epoch": 1.968268233667473, "grad_norm": 0.6829579790852797, "learning_rate": 6.628605832184054e-09, "loss": 0.2405, "step": 28471 }, { "epoch": 1.968337366055997, "grad_norm": 0.7348056814861645, "learning_rate": 6.5998234679620636e-09, "loss": 0.2567, "step": 28472 }, { "epoch": 1.9684064984445213, "grad_norm": 0.7791618123907529, "learning_rate": 6.571103687099678e-09, "loss": 0.3593, "step": 28473 }, { "epoch": 1.9684756308330453, "grad_norm": 0.8043686796023365, "learning_rate": 6.542446489957721e-09, "loss": 0.2674, "step": 28474 }, { "epoch": 1.9685447632215693, "grad_norm": 0.6873275997559435, "learning_rate": 6.5138518768953495e-09, "loss": 0.2063, "step": 28475 }, { "epoch": 1.9686138956100934, "grad_norm": 0.6654604908616529, "learning_rate": 6.485319848270055e-09, "loss": 0.2979, "step": 28476 }, { "epoch": 1.9686830279986174, "grad_norm": 0.8297960475521311, "learning_rate": 6.456850404439885e-09, "loss": 0.2825, "step": 28477 }, { "epoch": 1.9687521603871414, "grad_norm": 0.9942742994130606, "learning_rate": 6.428443545761775e-09, "loss": 0.2634, "step": 28478 }, { "epoch": 1.9688212927756654, "grad_norm": 0.647603805665136, "learning_rate": 6.400099272592109e-09, "loss": 0.2365, "step": 28479 }, { "epoch": 1.9688904251641894, "grad_norm": 0.674801880968789, "learning_rate": 6.371817585285045e-09, "loss": 0.3214, "step": 28480 }, { "epoch": 1.9689595575527135, "grad_norm": 0.7255434310010872, "learning_rate": 6.3435984841964115e-09, "loss": 0.3216, "step": 28481 }, { "epoch": 1.9690286899412375, "grad_norm": 0.6889146351429132, "learning_rate": 6.315441969678704e-09, "loss": 0.226, "step": 28482 }, { "epoch": 1.9690978223297615, "grad_norm": 0.8009094485916954, "learning_rate": 6.287348042085528e-09, "loss": 0.2414, "step": 28483 }, { "epoch": 1.9691669547182855, "grad_norm": 0.7378792944959796, "learning_rate": 6.259316701768825e-09, "loss": 0.2642, "step": 28484 }, { "epoch": 1.9692360871068095, "grad_norm": 0.697939784727754, "learning_rate": 6.231347949079425e-09, "loss": 0.261, "step": 28485 }, { "epoch": 1.9693052194953335, "grad_norm": 1.2255917936413174, "learning_rate": 6.203441784368714e-09, "loss": 0.2237, "step": 28486 }, { "epoch": 1.9693743518838576, "grad_norm": 0.8218385919928756, "learning_rate": 6.175598207985856e-09, "loss": 0.3813, "step": 28487 }, { "epoch": 1.9694434842723816, "grad_norm": 0.7459080029164026, "learning_rate": 6.147817220280017e-09, "loss": 0.3036, "step": 28488 }, { "epoch": 1.9695126166609056, "grad_norm": 0.5722382147963032, "learning_rate": 6.120098821599252e-09, "loss": 0.2184, "step": 28489 }, { "epoch": 1.9695817490494296, "grad_norm": 0.7403545267748101, "learning_rate": 6.092443012291061e-09, "loss": 0.2349, "step": 28490 }, { "epoch": 1.9696508814379536, "grad_norm": 0.7320989746885973, "learning_rate": 6.064849792702388e-09, "loss": 0.3744, "step": 28491 }, { "epoch": 1.9697200138264777, "grad_norm": 0.6934310299398982, "learning_rate": 6.037319163178512e-09, "loss": 0.23, "step": 28492 }, { "epoch": 1.9697891462150017, "grad_norm": 0.9253114875630037, "learning_rate": 6.009851124065269e-09, "loss": 0.2699, "step": 28493 }, { "epoch": 1.969858278603526, "grad_norm": 0.8278036826935747, "learning_rate": 5.982445675705717e-09, "loss": 0.3068, "step": 28494 }, { "epoch": 1.96992741099205, "grad_norm": 0.8125942379491695, "learning_rate": 5.955102818444025e-09, "loss": 0.2792, "step": 28495 }, { "epoch": 1.969996543380574, "grad_norm": 0.7278096430370418, "learning_rate": 5.927822552622698e-09, "loss": 0.2114, "step": 28496 }, { "epoch": 1.970065675769098, "grad_norm": 0.7308767813820094, "learning_rate": 5.900604878584238e-09, "loss": 0.2692, "step": 28497 }, { "epoch": 1.970134808157622, "grad_norm": 0.7439182986333532, "learning_rate": 5.873449796668929e-09, "loss": 0.3308, "step": 28498 }, { "epoch": 1.970203940546146, "grad_norm": 0.7514594245012068, "learning_rate": 5.8463573072181646e-09, "loss": 0.2551, "step": 28499 }, { "epoch": 1.97027307293467, "grad_norm": 0.6405675451563817, "learning_rate": 5.819327410570008e-09, "loss": 0.1827, "step": 28500 }, { "epoch": 1.970342205323194, "grad_norm": 0.7382286993975503, "learning_rate": 5.7923601070647425e-09, "loss": 0.3285, "step": 28501 }, { "epoch": 1.970411337711718, "grad_norm": 0.716273655600603, "learning_rate": 5.7654553970393215e-09, "loss": 0.3495, "step": 28502 }, { "epoch": 1.970480470100242, "grad_norm": 0.6663932471680588, "learning_rate": 5.738613280831251e-09, "loss": 0.2575, "step": 28503 }, { "epoch": 1.970549602488766, "grad_norm": 0.6420733914457672, "learning_rate": 5.71183375877693e-09, "loss": 0.2327, "step": 28504 }, { "epoch": 1.9706187348772901, "grad_norm": 0.7259600322740515, "learning_rate": 5.685116831212756e-09, "loss": 0.3348, "step": 28505 }, { "epoch": 1.9706878672658141, "grad_norm": 0.7395135786606598, "learning_rate": 5.65846249847235e-09, "loss": 0.2855, "step": 28506 }, { "epoch": 1.9707569996543381, "grad_norm": 0.6755039227995403, "learning_rate": 5.6318707608904454e-09, "loss": 0.208, "step": 28507 }, { "epoch": 1.9708261320428622, "grad_norm": 0.672127382312317, "learning_rate": 5.605341618800109e-09, "loss": 0.2561, "step": 28508 }, { "epoch": 1.9708952644313862, "grad_norm": 0.7328927283808646, "learning_rate": 5.578875072533851e-09, "loss": 0.4001, "step": 28509 }, { "epoch": 1.9709643968199102, "grad_norm": 0.7127718426545604, "learning_rate": 5.552471122423075e-09, "loss": 0.2363, "step": 28510 }, { "epoch": 1.9710335292084342, "grad_norm": 1.0029892036662988, "learning_rate": 5.526129768799738e-09, "loss": 0.2046, "step": 28511 }, { "epoch": 1.9711026615969582, "grad_norm": 0.6511713944675452, "learning_rate": 5.499851011993018e-09, "loss": 0.3076, "step": 28512 }, { "epoch": 1.9711717939854823, "grad_norm": 0.7085379613109339, "learning_rate": 5.4736348523326544e-09, "loss": 0.2852, "step": 28513 }, { "epoch": 1.9712409263740063, "grad_norm": 0.6625835914504797, "learning_rate": 5.447481290147272e-09, "loss": 0.2338, "step": 28514 }, { "epoch": 1.9713100587625303, "grad_norm": 0.6814583497901002, "learning_rate": 5.421390325764386e-09, "loss": 0.2057, "step": 28515 }, { "epoch": 1.9713791911510543, "grad_norm": 0.7057777519681021, "learning_rate": 5.395361959510958e-09, "loss": 0.3307, "step": 28516 }, { "epoch": 1.9714483235395783, "grad_norm": 0.672585672078848, "learning_rate": 5.3693961917139495e-09, "loss": 0.2319, "step": 28517 }, { "epoch": 1.9715174559281023, "grad_norm": 0.8998613068934357, "learning_rate": 5.3434930226975434e-09, "loss": 0.2204, "step": 28518 }, { "epoch": 1.9715865883166264, "grad_norm": 0.7518186294212665, "learning_rate": 5.317652452787591e-09, "loss": 0.2936, "step": 28519 }, { "epoch": 1.9716557207051504, "grad_norm": 0.7176570107405104, "learning_rate": 5.291874482307169e-09, "loss": 0.2892, "step": 28520 }, { "epoch": 1.9717248530936744, "grad_norm": 0.6660540089933575, "learning_rate": 5.266159111579905e-09, "loss": 0.2098, "step": 28521 }, { "epoch": 1.9717939854821984, "grad_norm": 0.6774676998204825, "learning_rate": 5.2405063409277645e-09, "loss": 0.2545, "step": 28522 }, { "epoch": 1.9718631178707224, "grad_norm": 0.7088345811460951, "learning_rate": 5.214916170672157e-09, "loss": 0.3005, "step": 28523 }, { "epoch": 1.9719322502592465, "grad_norm": 0.7587502790937382, "learning_rate": 5.189388601134493e-09, "loss": 0.2404, "step": 28524 }, { "epoch": 1.9720013826477705, "grad_norm": 1.0131756422708564, "learning_rate": 5.1639236326334055e-09, "loss": 0.2703, "step": 28525 }, { "epoch": 1.9720705150362945, "grad_norm": 0.9011103707147516, "learning_rate": 5.138521265489749e-09, "loss": 0.2392, "step": 28526 }, { "epoch": 1.9721396474248185, "grad_norm": 0.8544634848720547, "learning_rate": 5.113181500020492e-09, "loss": 0.3271, "step": 28527 }, { "epoch": 1.9722087798133425, "grad_norm": 0.7218540444079926, "learning_rate": 5.087904336543714e-09, "loss": 0.2431, "step": 28528 }, { "epoch": 1.9722779122018665, "grad_norm": 0.7010574610743753, "learning_rate": 5.062689775376384e-09, "loss": 0.2676, "step": 28529 }, { "epoch": 1.9723470445903906, "grad_norm": 0.6954133108147097, "learning_rate": 5.037537816833804e-09, "loss": 0.3213, "step": 28530 }, { "epoch": 1.9724161769789146, "grad_norm": 0.865640516393572, "learning_rate": 5.0124484612323884e-09, "loss": 0.2329, "step": 28531 }, { "epoch": 1.9724853093674386, "grad_norm": 0.719479450788192, "learning_rate": 4.987421708885776e-09, "loss": 0.2172, "step": 28532 }, { "epoch": 1.9725544417559626, "grad_norm": 0.6617079339380689, "learning_rate": 4.962457560108158e-09, "loss": 0.2939, "step": 28533 }, { "epoch": 1.9726235741444866, "grad_norm": 0.7025814147708451, "learning_rate": 4.937556015211509e-09, "loss": 0.3101, "step": 28534 }, { "epoch": 1.9726927065330107, "grad_norm": 0.6872369728264817, "learning_rate": 4.912717074508911e-09, "loss": 0.2401, "step": 28535 }, { "epoch": 1.9727618389215347, "grad_norm": 0.6983175366666929, "learning_rate": 4.8879407383112255e-09, "loss": 0.1674, "step": 28536 }, { "epoch": 1.9728309713100587, "grad_norm": 0.7091479586879897, "learning_rate": 4.863227006928762e-09, "loss": 0.2781, "step": 28537 }, { "epoch": 1.9729001036985827, "grad_norm": 0.6774660826754217, "learning_rate": 4.838575880671826e-09, "loss": 0.2487, "step": 28538 }, { "epoch": 1.9729692360871067, "grad_norm": 0.667383591730554, "learning_rate": 4.8139873598490595e-09, "loss": 0.231, "step": 28539 }, { "epoch": 1.9730383684756307, "grad_norm": 0.6723054953283483, "learning_rate": 4.7894614447685506e-09, "loss": 0.2688, "step": 28540 }, { "epoch": 1.9731075008641548, "grad_norm": 0.6787973903693082, "learning_rate": 4.764998135737831e-09, "loss": 0.3929, "step": 28541 }, { "epoch": 1.9731766332526788, "grad_norm": 0.7044703725702638, "learning_rate": 4.7405974330633206e-09, "loss": 0.2565, "step": 28542 }, { "epoch": 1.9732457656412028, "grad_norm": 0.6784325472849795, "learning_rate": 4.716259337050888e-09, "loss": 0.3116, "step": 28543 }, { "epoch": 1.9733148980297268, "grad_norm": 0.7216946243717258, "learning_rate": 4.6919838480058435e-09, "loss": 0.3729, "step": 28544 }, { "epoch": 1.9733840304182508, "grad_norm": 1.243722799941045, "learning_rate": 4.667770966232388e-09, "loss": 0.2969, "step": 28545 }, { "epoch": 1.9734531628067749, "grad_norm": 0.6516560344986325, "learning_rate": 4.643620692033612e-09, "loss": 0.2244, "step": 28546 }, { "epoch": 1.9735222951952989, "grad_norm": 0.71549120894034, "learning_rate": 4.6195330257126084e-09, "loss": 0.2536, "step": 28547 }, { "epoch": 1.9735914275838229, "grad_norm": 0.7184094335478262, "learning_rate": 4.595507967571356e-09, "loss": 0.3762, "step": 28548 }, { "epoch": 1.973660559972347, "grad_norm": 0.7205373593407481, "learning_rate": 4.5715455179101694e-09, "loss": 0.2257, "step": 28549 }, { "epoch": 1.973729692360871, "grad_norm": 0.6333034992205082, "learning_rate": 4.5476456770304765e-09, "loss": 0.2034, "step": 28550 }, { "epoch": 1.973798824749395, "grad_norm": 0.6449890189994645, "learning_rate": 4.523808445230926e-09, "loss": 0.2999, "step": 28551 }, { "epoch": 1.973867957137919, "grad_norm": 0.7470666539179311, "learning_rate": 4.500033822810724e-09, "loss": 0.3279, "step": 28552 }, { "epoch": 1.973937089526443, "grad_norm": 0.73394182842114, "learning_rate": 4.4763218100674075e-09, "loss": 0.2402, "step": 28553 }, { "epoch": 1.974006221914967, "grad_norm": 0.6530037571218609, "learning_rate": 4.45267240729852e-09, "loss": 0.2204, "step": 28554 }, { "epoch": 1.9740753543034912, "grad_norm": 1.0233332503502308, "learning_rate": 4.429085614800488e-09, "loss": 0.3172, "step": 28555 }, { "epoch": 1.9741444866920153, "grad_norm": 0.7219200836402447, "learning_rate": 4.405561432869188e-09, "loss": 0.2756, "step": 28556 }, { "epoch": 1.9742136190805393, "grad_norm": 0.6413430835673837, "learning_rate": 4.382099861798828e-09, "loss": 0.2141, "step": 28557 }, { "epoch": 1.9742827514690633, "grad_norm": 0.7103710817167711, "learning_rate": 4.358700901884172e-09, "loss": 0.2366, "step": 28558 }, { "epoch": 1.9743518838575873, "grad_norm": 0.7224848222711066, "learning_rate": 4.335364553417765e-09, "loss": 0.3192, "step": 28559 }, { "epoch": 1.9744210162461113, "grad_norm": 0.7366355362322912, "learning_rate": 4.312090816692149e-09, "loss": 0.2207, "step": 28560 }, { "epoch": 1.9744901486346353, "grad_norm": 0.69952920186785, "learning_rate": 4.28887969199987e-09, "loss": 0.2462, "step": 28561 }, { "epoch": 1.9745592810231594, "grad_norm": 0.8591993310726779, "learning_rate": 4.265731179630694e-09, "loss": 0.3381, "step": 28562 }, { "epoch": 1.9746284134116834, "grad_norm": 0.7286579164805239, "learning_rate": 4.242645279875501e-09, "loss": 0.3127, "step": 28563 }, { "epoch": 1.9746975458002074, "grad_norm": 0.6126795798602218, "learning_rate": 4.2196219930235035e-09, "loss": 0.1714, "step": 28564 }, { "epoch": 1.9747666781887314, "grad_norm": 0.7839048868787324, "learning_rate": 4.19666131936336e-09, "loss": 0.2322, "step": 28565 }, { "epoch": 1.9748358105772554, "grad_norm": 0.6783095190799769, "learning_rate": 4.173763259182062e-09, "loss": 0.319, "step": 28566 }, { "epoch": 1.9749049429657795, "grad_norm": 0.7236449332601687, "learning_rate": 4.1509278127677135e-09, "loss": 0.2637, "step": 28567 }, { "epoch": 1.9749740753543035, "grad_norm": 0.7219787633573199, "learning_rate": 4.128154980405641e-09, "loss": 0.2737, "step": 28568 }, { "epoch": 1.9750432077428275, "grad_norm": 0.884769060649421, "learning_rate": 4.105444762381727e-09, "loss": 0.3164, "step": 28569 }, { "epoch": 1.9751123401313515, "grad_norm": 0.7707035897540636, "learning_rate": 4.0827971589807445e-09, "loss": 0.3202, "step": 28570 }, { "epoch": 1.9751814725198755, "grad_norm": 0.7085316559414677, "learning_rate": 4.0602121704858e-09, "loss": 0.222, "step": 28571 }, { "epoch": 1.9752506049083995, "grad_norm": 0.7357562476186933, "learning_rate": 4.03768979718111e-09, "loss": 0.2293, "step": 28572 }, { "epoch": 1.9753197372969236, "grad_norm": 0.7191949597950437, "learning_rate": 4.015230039347562e-09, "loss": 0.2959, "step": 28573 }, { "epoch": 1.9753888696854476, "grad_norm": 0.7490441966663962, "learning_rate": 3.992832897267707e-09, "loss": 0.2222, "step": 28574 }, { "epoch": 1.9754580020739718, "grad_norm": 0.6676345605217732, "learning_rate": 3.9704983712218755e-09, "loss": 0.218, "step": 28575 }, { "epoch": 1.9755271344624958, "grad_norm": 0.7894336140426288, "learning_rate": 3.948226461490401e-09, "loss": 0.3218, "step": 28576 }, { "epoch": 1.9755962668510199, "grad_norm": 0.7461951064741091, "learning_rate": 3.926017168351393e-09, "loss": 0.3231, "step": 28577 }, { "epoch": 1.9756653992395439, "grad_norm": 0.7217111226442424, "learning_rate": 3.9038704920846275e-09, "loss": 0.2196, "step": 28578 }, { "epoch": 1.975734531628068, "grad_norm": 0.7230288349995371, "learning_rate": 3.881786432965995e-09, "loss": 0.2846, "step": 28579 }, { "epoch": 1.975803664016592, "grad_norm": 0.7556564053039301, "learning_rate": 3.859764991273607e-09, "loss": 0.3091, "step": 28580 }, { "epoch": 1.975872796405116, "grad_norm": 0.9188214253212963, "learning_rate": 3.837806167283354e-09, "loss": 0.2299, "step": 28581 }, { "epoch": 1.97594192879364, "grad_norm": 0.5759040095906337, "learning_rate": 3.815909961270014e-09, "loss": 0.209, "step": 28582 }, { "epoch": 1.976011061182164, "grad_norm": 0.6536081026394803, "learning_rate": 3.794076373507816e-09, "loss": 0.244, "step": 28583 }, { "epoch": 1.976080193570688, "grad_norm": 0.7110528206095756, "learning_rate": 3.772305404270982e-09, "loss": 0.3604, "step": 28584 }, { "epoch": 1.976149325959212, "grad_norm": 0.962950373990036, "learning_rate": 3.750597053832073e-09, "loss": 0.2567, "step": 28585 }, { "epoch": 1.976218458347736, "grad_norm": 0.6582939225387349, "learning_rate": 3.728951322463647e-09, "loss": 0.2753, "step": 28586 }, { "epoch": 1.97628759073626, "grad_norm": 0.7848936914127158, "learning_rate": 3.7073682104360465e-09, "loss": 0.3429, "step": 28587 }, { "epoch": 1.976356723124784, "grad_norm": 0.6677506213237446, "learning_rate": 3.6858477180207187e-09, "loss": 0.2749, "step": 28588 }, { "epoch": 1.976425855513308, "grad_norm": 0.7026307584766115, "learning_rate": 3.664389845486338e-09, "loss": 0.2207, "step": 28589 }, { "epoch": 1.976494987901832, "grad_norm": 0.6829276607163214, "learning_rate": 3.6429945931026887e-09, "loss": 0.2527, "step": 28590 }, { "epoch": 1.976564120290356, "grad_norm": 0.7155699026546144, "learning_rate": 3.62166196113789e-09, "loss": 0.3478, "step": 28591 }, { "epoch": 1.9766332526788801, "grad_norm": 0.6617580663298768, "learning_rate": 3.600391949859505e-09, "loss": 0.2521, "step": 28592 }, { "epoch": 1.9767023850674041, "grad_norm": 0.900641318932126, "learning_rate": 3.579184559532878e-09, "loss": 0.2188, "step": 28593 }, { "epoch": 1.9767715174559282, "grad_norm": 0.6729347352249435, "learning_rate": 3.5580397904250165e-09, "loss": 0.3004, "step": 28594 }, { "epoch": 1.9768406498444522, "grad_norm": 0.7205935702588251, "learning_rate": 3.536957642800709e-09, "loss": 0.3268, "step": 28595 }, { "epoch": 1.9769097822329762, "grad_norm": 0.6684407501559354, "learning_rate": 3.515938116924189e-09, "loss": 0.2215, "step": 28596 }, { "epoch": 1.9769789146215002, "grad_norm": 0.7039100211614825, "learning_rate": 3.494981213058579e-09, "loss": 0.2729, "step": 28597 }, { "epoch": 1.9770480470100242, "grad_norm": 0.7675987487831859, "learning_rate": 3.474086931466447e-09, "loss": 0.3409, "step": 28598 }, { "epoch": 1.9771171793985483, "grad_norm": 0.7641109663975065, "learning_rate": 3.453255272410361e-09, "loss": 0.2884, "step": 28599 }, { "epoch": 1.9771863117870723, "grad_norm": 0.6857288381136392, "learning_rate": 3.4324862361506673e-09, "loss": 0.2445, "step": 28600 }, { "epoch": 1.9772554441755963, "grad_norm": 0.7694456521423484, "learning_rate": 3.411779822948269e-09, "loss": 0.3091, "step": 28601 }, { "epoch": 1.9773245765641203, "grad_norm": 0.6202629909749446, "learning_rate": 3.391136033062403e-09, "loss": 0.2993, "step": 28602 }, { "epoch": 1.9773937089526443, "grad_norm": 0.7068944304493476, "learning_rate": 3.370554866751752e-09, "loss": 0.2002, "step": 28603 }, { "epoch": 1.9774628413411683, "grad_norm": 0.7380633018607137, "learning_rate": 3.3500363242738866e-09, "loss": 0.2905, "step": 28604 }, { "epoch": 1.9775319737296924, "grad_norm": 0.7181123725099413, "learning_rate": 3.329580405886934e-09, "loss": 0.3874, "step": 28605 }, { "epoch": 1.9776011061182164, "grad_norm": 0.6959581137109245, "learning_rate": 3.309187111846246e-09, "loss": 0.2093, "step": 28606 }, { "epoch": 1.9776702385067404, "grad_norm": 0.6376373091341121, "learning_rate": 3.2888564424082834e-09, "loss": 0.201, "step": 28607 }, { "epoch": 1.9777393708952644, "grad_norm": 0.7295576113776289, "learning_rate": 3.268588397827288e-09, "loss": 0.2484, "step": 28608 }, { "epoch": 1.9778085032837884, "grad_norm": 0.7816257657033453, "learning_rate": 3.2483829783575005e-09, "loss": 0.3482, "step": 28609 }, { "epoch": 1.9778776356723125, "grad_norm": 0.7585861402670663, "learning_rate": 3.2282401842520516e-09, "loss": 0.2587, "step": 28610 }, { "epoch": 1.9779467680608365, "grad_norm": 0.6471976994018773, "learning_rate": 3.2081600157635175e-09, "loss": 0.2226, "step": 28611 }, { "epoch": 1.9780159004493605, "grad_norm": 0.7683523390146456, "learning_rate": 3.188142473143918e-09, "loss": 0.3459, "step": 28612 }, { "epoch": 1.9780850328378845, "grad_norm": 0.7866211638541573, "learning_rate": 3.168187556643054e-09, "loss": 0.3113, "step": 28613 }, { "epoch": 1.9781541652264085, "grad_norm": 0.7626283611166156, "learning_rate": 3.1482952665123913e-09, "loss": 0.2271, "step": 28614 }, { "epoch": 1.9782232976149325, "grad_norm": 0.700254611516815, "learning_rate": 3.1284656030000636e-09, "loss": 0.2796, "step": 28615 }, { "epoch": 1.9782924300034566, "grad_norm": 0.7160567448063861, "learning_rate": 3.108698566355317e-09, "loss": 0.2867, "step": 28616 }, { "epoch": 1.9783615623919806, "grad_norm": 0.6629589423283385, "learning_rate": 3.088994156825731e-09, "loss": 0.2436, "step": 28617 }, { "epoch": 1.9784306947805046, "grad_norm": 0.7388741385405677, "learning_rate": 3.0693523746577746e-09, "loss": 0.2711, "step": 28618 }, { "epoch": 1.9784998271690286, "grad_norm": 0.6610966282859886, "learning_rate": 3.049773220098473e-09, "loss": 0.2708, "step": 28619 }, { "epoch": 1.9785689595575526, "grad_norm": 0.7892248101219846, "learning_rate": 3.0302566933931852e-09, "loss": 0.2715, "step": 28620 }, { "epoch": 1.9786380919460766, "grad_norm": 0.7186044139066111, "learning_rate": 3.0108027947856057e-09, "loss": 0.1975, "step": 28621 }, { "epoch": 1.9787072243346007, "grad_norm": 0.6996691954979175, "learning_rate": 2.9914115245199827e-09, "loss": 0.24, "step": 28622 }, { "epoch": 1.9787763567231247, "grad_norm": 0.9739680055316098, "learning_rate": 2.9720828828394554e-09, "loss": 0.3259, "step": 28623 }, { "epoch": 1.9788454891116487, "grad_norm": 0.6963716146288331, "learning_rate": 2.9528168699866076e-09, "loss": 0.2474, "step": 28624 }, { "epoch": 1.9789146215001727, "grad_norm": 0.6258001505561331, "learning_rate": 2.9336134862023578e-09, "loss": 0.1943, "step": 28625 }, { "epoch": 1.9789837538886967, "grad_norm": 0.6951086298941588, "learning_rate": 2.9144727317276243e-09, "loss": 0.2579, "step": 28626 }, { "epoch": 1.9790528862772208, "grad_norm": 0.7693209349143965, "learning_rate": 2.8953946068027707e-09, "loss": 0.3709, "step": 28627 }, { "epoch": 1.9791220186657448, "grad_norm": 0.7514591946710333, "learning_rate": 2.8763791116659388e-09, "loss": 0.218, "step": 28628 }, { "epoch": 1.9791911510542688, "grad_norm": 0.6799838568642284, "learning_rate": 2.8574262465558277e-09, "loss": 0.2041, "step": 28629 }, { "epoch": 1.9792602834427928, "grad_norm": 0.6782996208624185, "learning_rate": 2.8385360117100245e-09, "loss": 0.3313, "step": 28630 }, { "epoch": 1.9793294158313168, "grad_norm": 0.7585874555705858, "learning_rate": 2.8197084073655622e-09, "loss": 0.2838, "step": 28631 }, { "epoch": 1.9793985482198408, "grad_norm": 0.8213098359372965, "learning_rate": 2.8009434337578077e-09, "loss": 0.2296, "step": 28632 }, { "epoch": 1.9794676806083649, "grad_norm": 0.7100049099649369, "learning_rate": 2.7822410911226836e-09, "loss": 0.2699, "step": 28633 }, { "epoch": 1.9795368129968889, "grad_norm": 0.7321683998982765, "learning_rate": 2.763601379694447e-09, "loss": 0.3087, "step": 28634 }, { "epoch": 1.979605945385413, "grad_norm": 0.734788456380787, "learning_rate": 2.745024299706245e-09, "loss": 0.1971, "step": 28635 }, { "epoch": 1.9796750777739371, "grad_norm": 0.7550316483629625, "learning_rate": 2.7265098513912238e-09, "loss": 0.231, "step": 28636 }, { "epoch": 1.9797442101624612, "grad_norm": 0.838964413768752, "learning_rate": 2.70805803498142e-09, "loss": 0.3101, "step": 28637 }, { "epoch": 1.9798133425509852, "grad_norm": 0.7075876218290622, "learning_rate": 2.6896688507077605e-09, "loss": 0.2898, "step": 28638 }, { "epoch": 1.9798824749395092, "grad_norm": 0.6591198441928487, "learning_rate": 2.6713422988011716e-09, "loss": 0.2105, "step": 28639 }, { "epoch": 1.9799516073280332, "grad_norm": 0.7507360716524167, "learning_rate": 2.653078379490914e-09, "loss": 0.2954, "step": 28640 }, { "epoch": 1.9800207397165572, "grad_norm": 0.7296920374840131, "learning_rate": 2.634877093006805e-09, "loss": 0.3755, "step": 28641 }, { "epoch": 1.9800898721050813, "grad_norm": 0.7474442701573912, "learning_rate": 2.6167384395758833e-09, "loss": 0.2134, "step": 28642 }, { "epoch": 1.9801590044936053, "grad_norm": 0.6983473807632412, "learning_rate": 2.598662419425746e-09, "loss": 0.2243, "step": 28643 }, { "epoch": 1.9802281368821293, "grad_norm": 0.6712990711122679, "learning_rate": 2.580649032783433e-09, "loss": 0.2823, "step": 28644 }, { "epoch": 1.9802972692706533, "grad_norm": 1.8869242248610893, "learning_rate": 2.5626982798743207e-09, "loss": 0.2697, "step": 28645 }, { "epoch": 1.9803664016591773, "grad_norm": 0.6990207172554509, "learning_rate": 2.5448101609232278e-09, "loss": 0.2663, "step": 28646 }, { "epoch": 1.9804355340477013, "grad_norm": 0.7224313124412043, "learning_rate": 2.5269846761549755e-09, "loss": 0.2333, "step": 28647 }, { "epoch": 1.9805046664362254, "grad_norm": 0.7413792234014184, "learning_rate": 2.509221825792163e-09, "loss": 0.3802, "step": 28648 }, { "epoch": 1.9805737988247494, "grad_norm": 0.8437900892880476, "learning_rate": 2.4915216100579453e-09, "loss": 0.2161, "step": 28649 }, { "epoch": 1.9806429312132734, "grad_norm": 0.9120107724862542, "learning_rate": 2.473884029174367e-09, "loss": 0.2292, "step": 28650 }, { "epoch": 1.9807120636017974, "grad_norm": 0.7261068348859484, "learning_rate": 2.4563090833618074e-09, "loss": 0.2995, "step": 28651 }, { "epoch": 1.9807811959903214, "grad_norm": 0.7457153953571104, "learning_rate": 2.438796772841201e-09, "loss": 0.307, "step": 28652 }, { "epoch": 1.9808503283788454, "grad_norm": 0.7628170642214385, "learning_rate": 2.4213470978312615e-09, "loss": 0.2194, "step": 28653 }, { "epoch": 1.9809194607673695, "grad_norm": 0.7440187980791131, "learning_rate": 2.403960058551813e-09, "loss": 0.2443, "step": 28654 }, { "epoch": 1.9809885931558935, "grad_norm": 0.7022229854623205, "learning_rate": 2.3866356552193494e-09, "loss": 0.352, "step": 28655 }, { "epoch": 1.9810577255444177, "grad_norm": 0.6901182523921956, "learning_rate": 2.3693738880525842e-09, "loss": 0.2975, "step": 28656 }, { "epoch": 1.9811268579329417, "grad_norm": 0.7056065792181709, "learning_rate": 2.3521747572663458e-09, "loss": 0.2869, "step": 28657 }, { "epoch": 1.9811959903214658, "grad_norm": 0.68035210205489, "learning_rate": 2.3350382630771275e-09, "loss": 0.2393, "step": 28658 }, { "epoch": 1.9812651227099898, "grad_norm": 0.7367410650641543, "learning_rate": 2.3179644056997574e-09, "loss": 0.4029, "step": 28659 }, { "epoch": 1.9813342550985138, "grad_norm": 0.6883170378765643, "learning_rate": 2.300953185347399e-09, "loss": 0.237, "step": 28660 }, { "epoch": 1.9814033874870378, "grad_norm": 0.7559896983417881, "learning_rate": 2.2840046022343242e-09, "loss": 0.2348, "step": 28661 }, { "epoch": 1.9814725198755618, "grad_norm": 0.7156160106065398, "learning_rate": 2.2671186565720318e-09, "loss": 0.3002, "step": 28662 }, { "epoch": 1.9815416522640859, "grad_norm": 0.7392844014903762, "learning_rate": 2.250295348573128e-09, "loss": 0.2915, "step": 28663 }, { "epoch": 1.9816107846526099, "grad_norm": 0.6694955763130477, "learning_rate": 2.2335346784474465e-09, "loss": 0.2496, "step": 28664 }, { "epoch": 1.9816799170411339, "grad_norm": 0.6534112563019031, "learning_rate": 2.2168366464053735e-09, "loss": 0.2258, "step": 28665 }, { "epoch": 1.981749049429658, "grad_norm": 0.6821812070720445, "learning_rate": 2.200201252656187e-09, "loss": 0.319, "step": 28666 }, { "epoch": 1.981818181818182, "grad_norm": 0.6959554503323353, "learning_rate": 2.1836284974091626e-09, "loss": 0.2302, "step": 28667 }, { "epoch": 1.981887314206706, "grad_norm": 0.6555121363211774, "learning_rate": 2.167118380870803e-09, "loss": 0.2171, "step": 28668 }, { "epoch": 1.98195644659523, "grad_norm": 0.6786422459127677, "learning_rate": 2.1506709032487194e-09, "loss": 0.3319, "step": 28669 }, { "epoch": 1.982025578983754, "grad_norm": 0.716188486292941, "learning_rate": 2.1342860647488583e-09, "loss": 0.2511, "step": 28670 }, { "epoch": 1.982094711372278, "grad_norm": 0.6787670200499101, "learning_rate": 2.1179638655766112e-09, "loss": 0.2123, "step": 28671 }, { "epoch": 1.982163843760802, "grad_norm": 0.663433859289538, "learning_rate": 2.101704305936814e-09, "loss": 0.2413, "step": 28672 }, { "epoch": 1.982232976149326, "grad_norm": 0.6737732621481667, "learning_rate": 2.085507386032637e-09, "loss": 0.3046, "step": 28673 }, { "epoch": 1.98230210853785, "grad_norm": 0.8184244583438581, "learning_rate": 2.069373106067807e-09, "loss": 0.2339, "step": 28674 }, { "epoch": 1.982371240926374, "grad_norm": 0.6578663800274588, "learning_rate": 2.0533014662443838e-09, "loss": 0.1854, "step": 28675 }, { "epoch": 1.982440373314898, "grad_norm": 0.6658405668138307, "learning_rate": 2.0372924667633185e-09, "loss": 0.2658, "step": 28676 }, { "epoch": 1.982509505703422, "grad_norm": 0.6795453719713184, "learning_rate": 2.0213461078250064e-09, "loss": 0.3057, "step": 28677 }, { "epoch": 1.9825786380919461, "grad_norm": 3.33388720537835, "learning_rate": 2.005462389630397e-09, "loss": 0.2301, "step": 28678 }, { "epoch": 1.9826477704804701, "grad_norm": 0.7133768365891896, "learning_rate": 1.9896413123776657e-09, "loss": 0.3098, "step": 28679 }, { "epoch": 1.9827169028689942, "grad_norm": 0.7278523658081864, "learning_rate": 1.973882876266098e-09, "loss": 0.2578, "step": 28680 }, { "epoch": 1.9827860352575182, "grad_norm": 0.6625566377103825, "learning_rate": 1.958187081491647e-09, "loss": 0.2245, "step": 28681 }, { "epoch": 1.9828551676460422, "grad_norm": 1.1639532533031307, "learning_rate": 1.9425539282524887e-09, "loss": 0.2178, "step": 28682 }, { "epoch": 1.9829243000345662, "grad_norm": 0.935969881239624, "learning_rate": 1.9269834167440217e-09, "loss": 0.2103, "step": 28683 }, { "epoch": 1.9829934324230902, "grad_norm": 0.7541264880931814, "learning_rate": 1.91147554716109e-09, "loss": 0.3596, "step": 28684 }, { "epoch": 1.9830625648116142, "grad_norm": 0.7371845855316432, "learning_rate": 1.8960303196985385e-09, "loss": 0.234, "step": 28685 }, { "epoch": 1.9831316972001383, "grad_norm": 0.6867487474926569, "learning_rate": 1.8806477345495456e-09, "loss": 0.247, "step": 28686 }, { "epoch": 1.9832008295886623, "grad_norm": 0.7015482133805983, "learning_rate": 1.865327791906735e-09, "loss": 0.3459, "step": 28687 }, { "epoch": 1.9832699619771863, "grad_norm": 0.8280388623740951, "learning_rate": 1.850070491963285e-09, "loss": 0.3191, "step": 28688 }, { "epoch": 1.9833390943657103, "grad_norm": 0.6291147419037575, "learning_rate": 1.8348758349090446e-09, "loss": 0.2012, "step": 28689 }, { "epoch": 1.9834082267542343, "grad_norm": 0.7344396487777143, "learning_rate": 1.8197438209349716e-09, "loss": 0.2591, "step": 28690 }, { "epoch": 1.9834773591427584, "grad_norm": 0.6679841908918258, "learning_rate": 1.8046744502309144e-09, "loss": 0.338, "step": 28691 }, { "epoch": 1.9835464915312824, "grad_norm": 0.6833133645417814, "learning_rate": 1.7896677229856107e-09, "loss": 0.2549, "step": 28692 }, { "epoch": 1.9836156239198064, "grad_norm": 0.718728403728518, "learning_rate": 1.7747236393872436e-09, "loss": 0.2592, "step": 28693 }, { "epoch": 1.9836847563083304, "grad_norm": 0.863499870361344, "learning_rate": 1.7598421996228853e-09, "loss": 0.3261, "step": 28694 }, { "epoch": 1.9837538886968544, "grad_norm": 0.6650023767588483, "learning_rate": 1.7450234038790536e-09, "loss": 0.2879, "step": 28695 }, { "epoch": 1.9838230210853784, "grad_norm": 0.7635175337734176, "learning_rate": 1.7302672523417107e-09, "loss": 0.1914, "step": 28696 }, { "epoch": 1.9838921534739025, "grad_norm": 0.7032703176265824, "learning_rate": 1.7155737451951538e-09, "loss": 0.2166, "step": 28697 }, { "epoch": 1.9839612858624265, "grad_norm": 0.6773192295438505, "learning_rate": 1.7009428826247898e-09, "loss": 0.3537, "step": 28698 }, { "epoch": 1.9840304182509505, "grad_norm": 0.6910443033977371, "learning_rate": 1.6863746648126955e-09, "loss": 0.1731, "step": 28699 }, { "epoch": 1.9840995506394745, "grad_norm": 0.7695089028096683, "learning_rate": 1.6718690919420578e-09, "loss": 0.2564, "step": 28700 }, { "epoch": 1.9841686830279985, "grad_norm": 0.7293483608932086, "learning_rate": 1.657426164194953e-09, "loss": 0.3098, "step": 28701 }, { "epoch": 1.9842378154165226, "grad_norm": 0.7585845981656473, "learning_rate": 1.6430458817517924e-09, "loss": 0.3106, "step": 28702 }, { "epoch": 1.9843069478050466, "grad_norm": 0.9485887793134252, "learning_rate": 1.6287282447935427e-09, "loss": 0.1857, "step": 28703 }, { "epoch": 1.9843760801935706, "grad_norm": 0.6662577888459471, "learning_rate": 1.6144732534983943e-09, "loss": 0.2565, "step": 28704 }, { "epoch": 1.9844452125820946, "grad_norm": 0.7692469203663318, "learning_rate": 1.6002809080462034e-09, "loss": 0.3309, "step": 28705 }, { "epoch": 1.9845143449706186, "grad_norm": 0.6812245283815457, "learning_rate": 1.5861512086146058e-09, "loss": 0.2345, "step": 28706 }, { "epoch": 1.9845834773591426, "grad_norm": 0.7443566385326194, "learning_rate": 1.572084155380682e-09, "loss": 0.2682, "step": 28707 }, { "epoch": 1.9846526097476667, "grad_norm": 0.9812577599926292, "learning_rate": 1.5580797485204024e-09, "loss": 0.2567, "step": 28708 }, { "epoch": 1.9847217421361907, "grad_norm": 0.690377561323311, "learning_rate": 1.5441379882091822e-09, "loss": 0.3689, "step": 28709 }, { "epoch": 1.9847908745247147, "grad_norm": 0.7205273833530389, "learning_rate": 1.5302588746224368e-09, "loss": 0.262, "step": 28710 }, { "epoch": 1.9848600069132387, "grad_norm": 0.7144953449189296, "learning_rate": 1.5164424079333606e-09, "loss": 0.2328, "step": 28711 }, { "epoch": 1.9849291393017627, "grad_norm": 0.7333302850744458, "learning_rate": 1.5026885883157038e-09, "loss": 0.2949, "step": 28712 }, { "epoch": 1.9849982716902868, "grad_norm": 0.7183775428699329, "learning_rate": 1.488997415941551e-09, "loss": 0.2475, "step": 28713 }, { "epoch": 1.9850674040788108, "grad_norm": 0.6924161568927255, "learning_rate": 1.4753688909824315e-09, "loss": 0.2225, "step": 28714 }, { "epoch": 1.9851365364673348, "grad_norm": 0.6552723819800714, "learning_rate": 1.4618030136093197e-09, "loss": 0.2367, "step": 28715 }, { "epoch": 1.9852056688558588, "grad_norm": 0.7424705824463372, "learning_rate": 1.4482997839926349e-09, "loss": 0.3395, "step": 28716 }, { "epoch": 1.985274801244383, "grad_norm": 0.6855982977380406, "learning_rate": 1.434859202300576e-09, "loss": 0.2717, "step": 28717 }, { "epoch": 1.985343933632907, "grad_norm": 0.6681976019079726, "learning_rate": 1.4214812687030067e-09, "loss": 0.1799, "step": 28718 }, { "epoch": 1.985413066021431, "grad_norm": 0.811448133677801, "learning_rate": 1.4081659833664607e-09, "loss": 0.3159, "step": 28719 }, { "epoch": 1.985482198409955, "grad_norm": 0.7324703839564126, "learning_rate": 1.3949133464580266e-09, "loss": 0.334, "step": 28720 }, { "epoch": 1.9855513307984791, "grad_norm": 0.6549170832646276, "learning_rate": 1.3817233581436828e-09, "loss": 0.259, "step": 28721 }, { "epoch": 1.9856204631870031, "grad_norm": 0.7524850112698458, "learning_rate": 1.3685960185894075e-09, "loss": 0.2329, "step": 28722 }, { "epoch": 1.9856895955755272, "grad_norm": 0.751080838727549, "learning_rate": 1.3555313279589588e-09, "loss": 0.3277, "step": 28723 }, { "epoch": 1.9857587279640512, "grad_norm": 0.7669888710848989, "learning_rate": 1.3425292864172045e-09, "loss": 0.2959, "step": 28724 }, { "epoch": 1.9858278603525752, "grad_norm": 0.7097138698197832, "learning_rate": 1.3295898941256823e-09, "loss": 0.2141, "step": 28725 }, { "epoch": 1.9858969927410992, "grad_norm": 0.702836929568197, "learning_rate": 1.31671315124704e-09, "loss": 0.2295, "step": 28726 }, { "epoch": 1.9859661251296232, "grad_norm": 0.7346992013997771, "learning_rate": 1.3038990579433696e-09, "loss": 0.302, "step": 28727 }, { "epoch": 1.9860352575181472, "grad_norm": 0.7729872288248282, "learning_rate": 1.2911476143745438e-09, "loss": 0.275, "step": 28728 }, { "epoch": 1.9861043899066713, "grad_norm": 0.7010431208515908, "learning_rate": 1.2784588207004345e-09, "loss": 0.2633, "step": 28729 }, { "epoch": 1.9861735222951953, "grad_norm": 0.6887794379238127, "learning_rate": 1.2658326770809136e-09, "loss": 0.3028, "step": 28730 }, { "epoch": 1.9862426546837193, "grad_norm": 0.7329990087971745, "learning_rate": 1.2532691836725231e-09, "loss": 0.2493, "step": 28731 }, { "epoch": 1.9863117870722433, "grad_norm": 0.6826946083870572, "learning_rate": 1.2407683406345794e-09, "loss": 0.17, "step": 28732 }, { "epoch": 1.9863809194607673, "grad_norm": 0.7303261405822962, "learning_rate": 1.2283301481225141e-09, "loss": 0.3064, "step": 28733 }, { "epoch": 1.9864500518492914, "grad_norm": 0.7024540515767697, "learning_rate": 1.2159546062928685e-09, "loss": 0.3288, "step": 28734 }, { "epoch": 1.9865191842378154, "grad_norm": 0.7139683497745904, "learning_rate": 1.2036417152999635e-09, "loss": 0.265, "step": 28735 }, { "epoch": 1.9865883166263394, "grad_norm": 0.7393075026217658, "learning_rate": 1.1913914752992305e-09, "loss": 0.2666, "step": 28736 }, { "epoch": 1.9866574490148636, "grad_norm": 0.727748369033508, "learning_rate": 1.1792038864433254e-09, "loss": 0.3216, "step": 28737 }, { "epoch": 1.9867265814033876, "grad_norm": 0.780268130331028, "learning_rate": 1.1670789488854584e-09, "loss": 0.2936, "step": 28738 }, { "epoch": 1.9867957137919117, "grad_norm": 0.694028268039186, "learning_rate": 1.1550166627771754e-09, "loss": 0.2629, "step": 28739 }, { "epoch": 1.9868648461804357, "grad_norm": 0.6697106024403173, "learning_rate": 1.1430170282700215e-09, "loss": 0.2151, "step": 28740 }, { "epoch": 1.9869339785689597, "grad_norm": 0.7320525663056934, "learning_rate": 1.1310800455144322e-09, "loss": 0.3467, "step": 28741 }, { "epoch": 1.9870031109574837, "grad_norm": 0.7456657252115112, "learning_rate": 1.1192057146602874e-09, "loss": 0.2182, "step": 28742 }, { "epoch": 1.9870722433460077, "grad_norm": 0.8660384425511095, "learning_rate": 1.107394035855247e-09, "loss": 0.2118, "step": 28743 }, { "epoch": 1.9871413757345318, "grad_norm": 0.7323126793902561, "learning_rate": 1.095645009248636e-09, "loss": 0.2671, "step": 28744 }, { "epoch": 1.9872105081230558, "grad_norm": 0.7869195007004308, "learning_rate": 1.0839586349864484e-09, "loss": 0.3198, "step": 28745 }, { "epoch": 1.9872796405115798, "grad_norm": 0.7279135399113764, "learning_rate": 1.072334913216344e-09, "loss": 0.2703, "step": 28746 }, { "epoch": 1.9873487729001038, "grad_norm": 0.6859435462383148, "learning_rate": 1.0607738440837623e-09, "loss": 0.3184, "step": 28747 }, { "epoch": 1.9874179052886278, "grad_norm": 0.7369779517921538, "learning_rate": 1.049275427733032e-09, "loss": 0.4243, "step": 28748 }, { "epoch": 1.9874870376771518, "grad_norm": 0.8558712000860101, "learning_rate": 1.037839664309037e-09, "loss": 0.2714, "step": 28749 }, { "epoch": 1.9875561700656759, "grad_norm": 0.6450049609850872, "learning_rate": 1.0264665539538865e-09, "loss": 0.2259, "step": 28750 }, { "epoch": 1.9876253024541999, "grad_norm": 0.7117529585544176, "learning_rate": 1.0151560968113538e-09, "loss": 0.2993, "step": 28751 }, { "epoch": 1.987694434842724, "grad_norm": 0.6998413258173697, "learning_rate": 1.0039082930229927e-09, "loss": 0.3267, "step": 28752 }, { "epoch": 1.987763567231248, "grad_norm": 0.7025497639011207, "learning_rate": 9.92723142729246e-10, "loss": 0.2429, "step": 28753 }, { "epoch": 1.987832699619772, "grad_norm": 0.7187741012583224, "learning_rate": 9.816006460705573e-10, "loss": 0.2556, "step": 28754 }, { "epoch": 1.987901832008296, "grad_norm": 0.7419832630122306, "learning_rate": 9.705408031868147e-10, "loss": 0.2943, "step": 28755 }, { "epoch": 1.98797096439682, "grad_norm": 0.8854669896487007, "learning_rate": 9.595436142156856e-10, "loss": 0.2661, "step": 28756 }, { "epoch": 1.988040096785344, "grad_norm": 0.6225741049179656, "learning_rate": 9.486090792953928e-10, "loss": 0.1951, "step": 28757 }, { "epoch": 1.988109229173868, "grad_norm": 0.6777091666690156, "learning_rate": 9.377371985636042e-10, "loss": 0.2692, "step": 28758 }, { "epoch": 1.988178361562392, "grad_norm": 0.7101736756853486, "learning_rate": 9.269279721557667e-10, "loss": 0.3094, "step": 28759 }, { "epoch": 1.988247493950916, "grad_norm": 0.7228236943084825, "learning_rate": 9.161814002073277e-10, "loss": 0.2585, "step": 28760 }, { "epoch": 1.98831662633944, "grad_norm": 0.6531711498555774, "learning_rate": 9.054974828537344e-10, "loss": 0.1777, "step": 28761 }, { "epoch": 1.988385758727964, "grad_norm": 0.7185533886639855, "learning_rate": 8.948762202282135e-10, "loss": 0.3682, "step": 28762 }, { "epoch": 1.988454891116488, "grad_norm": 0.7410455990994264, "learning_rate": 8.843176124639919e-10, "loss": 0.2873, "step": 28763 }, { "epoch": 1.9885240235050121, "grad_norm": 0.678701929647736, "learning_rate": 8.73821659693741e-10, "loss": 0.232, "step": 28764 }, { "epoch": 1.9885931558935361, "grad_norm": 0.8352803423242907, "learning_rate": 8.633883620490224e-10, "loss": 0.2709, "step": 28765 }, { "epoch": 1.9886622882820602, "grad_norm": 0.6771563937971536, "learning_rate": 8.530177196602874e-10, "loss": 0.3166, "step": 28766 }, { "epoch": 1.9887314206705842, "grad_norm": 0.6319134664519545, "learning_rate": 8.427097326574319e-10, "loss": 0.2358, "step": 28767 }, { "epoch": 1.9888005530591082, "grad_norm": 0.746763527297131, "learning_rate": 8.324644011697969e-10, "loss": 0.2133, "step": 28768 }, { "epoch": 1.9888696854476322, "grad_norm": 0.8210555500667525, "learning_rate": 8.222817253256132e-10, "loss": 0.3205, "step": 28769 }, { "epoch": 1.9889388178361562, "grad_norm": 0.7807108255170525, "learning_rate": 8.121617052531116e-10, "loss": 0.2975, "step": 28770 }, { "epoch": 1.9890079502246802, "grad_norm": 0.6737102932356895, "learning_rate": 8.021043410783025e-10, "loss": 0.2232, "step": 28771 }, { "epoch": 1.9890770826132043, "grad_norm": 0.6204285177242967, "learning_rate": 7.921096329277512e-10, "loss": 0.1963, "step": 28772 }, { "epoch": 1.9891462150017283, "grad_norm": 0.8742440356230978, "learning_rate": 7.821775809269128e-10, "loss": 0.273, "step": 28773 }, { "epoch": 1.9892153473902523, "grad_norm": 0.9930309389124646, "learning_rate": 7.723081851995773e-10, "loss": 0.2719, "step": 28774 }, { "epoch": 1.9892844797787763, "grad_norm": 0.6450577980455214, "learning_rate": 7.625014458700897e-10, "loss": 0.1573, "step": 28775 }, { "epoch": 1.9893536121673003, "grad_norm": 0.6817783862653561, "learning_rate": 7.527573630611295e-10, "loss": 0.3094, "step": 28776 }, { "epoch": 1.9894227445558244, "grad_norm": 0.7318421100106353, "learning_rate": 7.430759368948215e-10, "loss": 0.2982, "step": 28777 }, { "epoch": 1.9894918769443484, "grad_norm": 0.661616363786137, "learning_rate": 7.334571674921797e-10, "loss": 0.2242, "step": 28778 }, { "epoch": 1.9895610093328724, "grad_norm": 0.6781954061341292, "learning_rate": 7.239010549742187e-10, "loss": 0.2655, "step": 28779 }, { "epoch": 1.9896301417213964, "grad_norm": 0.8045269723984787, "learning_rate": 7.144075994608424e-10, "loss": 0.3087, "step": 28780 }, { "epoch": 1.9896992741099204, "grad_norm": 0.8093180724373712, "learning_rate": 7.049768010702895e-10, "loss": 0.2416, "step": 28781 }, { "epoch": 1.9897684064984444, "grad_norm": 0.6303261225436866, "learning_rate": 6.956086599213541e-10, "loss": 0.246, "step": 28782 }, { "epoch": 1.9898375388869685, "grad_norm": 0.7073515484829668, "learning_rate": 6.863031761317196e-10, "loss": 0.2465, "step": 28783 }, { "epoch": 1.9899066712754925, "grad_norm": 0.7557875367661365, "learning_rate": 6.770603498168493e-10, "loss": 0.3367, "step": 28784 }, { "epoch": 1.9899758036640165, "grad_norm": 0.7083491439753481, "learning_rate": 6.678801810938718e-10, "loss": 0.2118, "step": 28785 }, { "epoch": 1.9900449360525405, "grad_norm": 0.6775085829206356, "learning_rate": 6.587626700771399e-10, "loss": 0.2236, "step": 28786 }, { "epoch": 1.9901140684410645, "grad_norm": 0.7739166422605749, "learning_rate": 6.497078168810067e-10, "loss": 0.2728, "step": 28787 }, { "epoch": 1.9901832008295886, "grad_norm": 0.6988819044672345, "learning_rate": 6.4071562161927e-10, "loss": 0.2967, "step": 28788 }, { "epoch": 1.9902523332181126, "grad_norm": 0.6819259319742837, "learning_rate": 6.317860844046175e-10, "loss": 0.2479, "step": 28789 }, { "epoch": 1.9903214656066366, "grad_norm": 0.777038272430047, "learning_rate": 6.229192053480715e-10, "loss": 0.3042, "step": 28790 }, { "epoch": 1.9903905979951606, "grad_norm": 0.7153904326356276, "learning_rate": 6.141149845617644e-10, "loss": 0.376, "step": 28791 }, { "epoch": 1.9904597303836846, "grad_norm": 0.7749875227660543, "learning_rate": 6.053734221561636e-10, "loss": 0.2334, "step": 28792 }, { "epoch": 1.9905288627722086, "grad_norm": 0.7106348829012632, "learning_rate": 5.966945182395156e-10, "loss": 0.2801, "step": 28793 }, { "epoch": 1.9905979951607327, "grad_norm": 0.7036863652776841, "learning_rate": 5.880782729222878e-10, "loss": 0.3035, "step": 28794 }, { "epoch": 1.9906671275492567, "grad_norm": 0.6989706886728028, "learning_rate": 5.795246863110615e-10, "loss": 0.254, "step": 28795 }, { "epoch": 1.9907362599377807, "grad_norm": 0.7355541310105612, "learning_rate": 5.710337585140835e-10, "loss": 0.1921, "step": 28796 }, { "epoch": 1.9908053923263047, "grad_norm": 0.739562516787971, "learning_rate": 5.62605489636825e-10, "loss": 0.2403, "step": 28797 }, { "epoch": 1.990874524714829, "grad_norm": 0.8436299755390483, "learning_rate": 5.542398797858673e-10, "loss": 0.3165, "step": 28798 }, { "epoch": 1.990943657103353, "grad_norm": 5.961298224339283, "learning_rate": 5.459369290655713e-10, "loss": 0.2145, "step": 28799 }, { "epoch": 1.991012789491877, "grad_norm": 0.6798521609093073, "learning_rate": 5.376966375797432e-10, "loss": 0.2659, "step": 28800 }, { "epoch": 1.991081921880401, "grad_norm": 0.6816414510864884, "learning_rate": 5.295190054316335e-10, "loss": 0.2747, "step": 28801 }, { "epoch": 1.991151054268925, "grad_norm": 0.7073917813275237, "learning_rate": 5.21404032725048e-10, "loss": 0.3411, "step": 28802 }, { "epoch": 1.991220186657449, "grad_norm": 0.6611944166470435, "learning_rate": 5.133517195599069e-10, "loss": 0.2528, "step": 28803 }, { "epoch": 1.991289319045973, "grad_norm": 0.6524879364153915, "learning_rate": 5.053620660383507e-10, "loss": 0.1884, "step": 28804 }, { "epoch": 1.991358451434497, "grad_norm": 0.6904663661612556, "learning_rate": 4.974350722597443e-10, "loss": 0.368, "step": 28805 }, { "epoch": 1.991427583823021, "grad_norm": 0.6919510279585692, "learning_rate": 4.895707383240078e-10, "loss": 0.3112, "step": 28806 }, { "epoch": 1.9914967162115451, "grad_norm": 0.7230317875050783, "learning_rate": 4.817690643288408e-10, "loss": 0.2483, "step": 28807 }, { "epoch": 1.9915658486000691, "grad_norm": 0.7757869129213987, "learning_rate": 4.740300503736084e-10, "loss": 0.3084, "step": 28808 }, { "epoch": 1.9916349809885932, "grad_norm": 0.7494696257089379, "learning_rate": 4.663536965537896e-10, "loss": 0.3443, "step": 28809 }, { "epoch": 1.9917041133771172, "grad_norm": 0.631612127352572, "learning_rate": 4.58740002966529e-10, "loss": 0.1806, "step": 28810 }, { "epoch": 1.9917732457656412, "grad_norm": 0.7087137764309718, "learning_rate": 4.511889697061955e-10, "loss": 0.2365, "step": 28811 }, { "epoch": 1.9918423781541652, "grad_norm": 0.8902640429859409, "learning_rate": 4.437005968688235e-10, "loss": 0.3297, "step": 28812 }, { "epoch": 1.9919115105426892, "grad_norm": 0.7609527811982646, "learning_rate": 4.3627488454711653e-10, "loss": 0.3126, "step": 28813 }, { "epoch": 1.9919806429312132, "grad_norm": 0.7398135435602817, "learning_rate": 4.289118328348885e-10, "loss": 0.2429, "step": 28814 }, { "epoch": 1.9920497753197373, "grad_norm": 0.736291608403621, "learning_rate": 4.216114418237327e-10, "loss": 0.2458, "step": 28815 }, { "epoch": 1.9921189077082613, "grad_norm": 0.7379091118626147, "learning_rate": 4.1437371160579773e-10, "loss": 0.3846, "step": 28816 }, { "epoch": 1.9921880400967853, "grad_norm": 0.7731086038419165, "learning_rate": 4.071986422715668e-10, "loss": 0.2123, "step": 28817 }, { "epoch": 1.9922571724853093, "grad_norm": 0.6106487350651334, "learning_rate": 4.000862339109679e-10, "loss": 0.2224, "step": 28818 }, { "epoch": 1.9923263048738336, "grad_norm": 0.7583390373422565, "learning_rate": 3.9303648661337404e-10, "loss": 0.2833, "step": 28819 }, { "epoch": 1.9923954372623576, "grad_norm": 0.7824342947081611, "learning_rate": 3.8604940046649274e-10, "loss": 0.2509, "step": 28820 }, { "epoch": 1.9924645696508816, "grad_norm": 0.6176629462311566, "learning_rate": 3.791249755580317e-10, "loss": 0.2034, "step": 28821 }, { "epoch": 1.9925337020394056, "grad_norm": 0.7360399364505185, "learning_rate": 3.7226321197514346e-10, "loss": 0.222, "step": 28822 }, { "epoch": 1.9926028344279296, "grad_norm": 0.7415816452414927, "learning_rate": 3.6546410980387026e-10, "loss": 0.3393, "step": 28823 }, { "epoch": 1.9926719668164536, "grad_norm": 1.207748671298846, "learning_rate": 3.587276691291441e-10, "loss": 0.2707, "step": 28824 }, { "epoch": 1.9927410992049777, "grad_norm": 0.7956835730525373, "learning_rate": 3.520538900358972e-10, "loss": 0.2658, "step": 28825 }, { "epoch": 1.9928102315935017, "grad_norm": 0.7335434326384479, "learning_rate": 3.45442772606841e-10, "loss": 0.2863, "step": 28826 }, { "epoch": 1.9928793639820257, "grad_norm": 0.7984833887506169, "learning_rate": 3.3889431692579744e-10, "loss": 0.3562, "step": 28827 }, { "epoch": 1.9929484963705497, "grad_norm": 0.6730040261331468, "learning_rate": 3.324085230738128e-10, "loss": 0.2302, "step": 28828 }, { "epoch": 1.9930176287590737, "grad_norm": 0.6765315661192596, "learning_rate": 3.259853911335986e-10, "loss": 0.2647, "step": 28829 }, { "epoch": 1.9930867611475978, "grad_norm": 0.701877983314906, "learning_rate": 3.196249211839808e-10, "loss": 0.2981, "step": 28830 }, { "epoch": 1.9931558935361218, "grad_norm": 0.7349060944570824, "learning_rate": 3.133271133060056e-10, "loss": 0.2407, "step": 28831 }, { "epoch": 1.9932250259246458, "grad_norm": 0.7441630158326209, "learning_rate": 3.070919675784989e-10, "loss": 0.2215, "step": 28832 }, { "epoch": 1.9932941583131698, "grad_norm": 0.7152072774957241, "learning_rate": 3.0091948407862116e-10, "loss": 0.2222, "step": 28833 }, { "epoch": 1.9933632907016938, "grad_norm": 0.6845718342612463, "learning_rate": 2.9480966288464305e-10, "loss": 0.3368, "step": 28834 }, { "epoch": 1.9934324230902178, "grad_norm": 0.8521133278283738, "learning_rate": 2.8876250407261497e-10, "loss": 0.2376, "step": 28835 }, { "epoch": 1.9935015554787419, "grad_norm": 0.6805052263665855, "learning_rate": 2.827780077185871e-10, "loss": 0.2187, "step": 28836 }, { "epoch": 1.9935706878672659, "grad_norm": 0.6956247266472299, "learning_rate": 2.7685617389805463e-10, "loss": 0.3199, "step": 28837 }, { "epoch": 1.99363982025579, "grad_norm": 0.7510689820168994, "learning_rate": 2.709970026842923e-10, "loss": 0.2892, "step": 28838 }, { "epoch": 1.993708952644314, "grad_norm": 0.7335223638124039, "learning_rate": 2.652004941511299e-10, "loss": 0.2681, "step": 28839 }, { "epoch": 1.993778085032838, "grad_norm": 0.7073688790916559, "learning_rate": 2.594666483712871e-10, "loss": 0.2745, "step": 28840 }, { "epoch": 1.993847217421362, "grad_norm": 0.9537992507346141, "learning_rate": 2.5379546541692835e-10, "loss": 0.3566, "step": 28841 }, { "epoch": 1.993916349809886, "grad_norm": 0.6785529709779569, "learning_rate": 2.481869453585528e-10, "loss": 0.2667, "step": 28842 }, { "epoch": 1.99398548219841, "grad_norm": 0.7572011299447893, "learning_rate": 2.4264108826665967e-10, "loss": 0.2356, "step": 28843 }, { "epoch": 1.994054614586934, "grad_norm": 0.7227496651933148, "learning_rate": 2.3715789421063784e-10, "loss": 0.3067, "step": 28844 }, { "epoch": 1.994123746975458, "grad_norm": 0.6830814983738559, "learning_rate": 2.3173736325932118e-10, "loss": 0.2935, "step": 28845 }, { "epoch": 1.994192879363982, "grad_norm": 0.6303041203748393, "learning_rate": 2.2637949548098836e-10, "loss": 0.2537, "step": 28846 }, { "epoch": 1.994262011752506, "grad_norm": 0.7482146132668307, "learning_rate": 2.2108429094225282e-10, "loss": 0.2166, "step": 28847 }, { "epoch": 1.99433114414103, "grad_norm": 0.7368607532745445, "learning_rate": 2.15851749710283e-10, "loss": 0.3591, "step": 28848 }, { "epoch": 1.994400276529554, "grad_norm": 0.9684982514742132, "learning_rate": 2.1068187184947185e-10, "loss": 0.2462, "step": 28849 }, { "epoch": 1.9944694089180781, "grad_norm": 0.6827855956804483, "learning_rate": 2.0557465742532256e-10, "loss": 0.2329, "step": 28850 }, { "epoch": 1.9945385413066021, "grad_norm": 0.6851515867787881, "learning_rate": 2.0053010650167292e-10, "loss": 0.2484, "step": 28851 }, { "epoch": 1.9946076736951261, "grad_norm": 0.7299280490324511, "learning_rate": 1.9554821914236078e-10, "loss": 0.3045, "step": 28852 }, { "epoch": 1.9946768060836502, "grad_norm": 1.0146450171535872, "learning_rate": 1.9062899540900347e-10, "loss": 0.2185, "step": 28853 }, { "epoch": 1.9947459384721742, "grad_norm": 0.7248881787087164, "learning_rate": 1.8577243536321843e-10, "loss": 0.259, "step": 28854 }, { "epoch": 1.9948150708606982, "grad_norm": 0.7768816182229266, "learning_rate": 1.8097853906662298e-10, "loss": 0.3536, "step": 28855 }, { "epoch": 1.9948842032492222, "grad_norm": 0.7283287252361998, "learning_rate": 1.7624730657916922e-10, "loss": 0.2954, "step": 28856 }, { "epoch": 1.9949533356377462, "grad_norm": 0.7216925771352714, "learning_rate": 1.7157873795914382e-10, "loss": 0.2484, "step": 28857 }, { "epoch": 1.9950224680262703, "grad_norm": 0.7625167576317008, "learning_rate": 1.669728332659437e-10, "loss": 0.2767, "step": 28858 }, { "epoch": 1.9950916004147943, "grad_norm": 0.7174882839250708, "learning_rate": 1.624295925573005e-10, "loss": 0.305, "step": 28859 }, { "epoch": 1.9951607328033183, "grad_norm": 0.6769821346603997, "learning_rate": 1.579490158903907e-10, "loss": 0.2689, "step": 28860 }, { "epoch": 1.9952298651918423, "grad_norm": 0.6864949764727546, "learning_rate": 1.535311033201703e-10, "loss": 0.1986, "step": 28861 }, { "epoch": 1.9952989975803663, "grad_norm": 0.6872179275859172, "learning_rate": 1.491758549032607e-10, "loss": 0.3116, "step": 28862 }, { "epoch": 1.9953681299688903, "grad_norm": 0.7177161555090465, "learning_rate": 1.4488327069406283e-10, "loss": 0.2337, "step": 28863 }, { "epoch": 1.9954372623574144, "grad_norm": 0.6455173803785526, "learning_rate": 1.406533507458674e-10, "loss": 0.2452, "step": 28864 }, { "epoch": 1.9955063947459384, "grad_norm": 0.6705375132944257, "learning_rate": 1.3648609511196509e-10, "loss": 0.25, "step": 28865 }, { "epoch": 1.9955755271344624, "grad_norm": 0.7600435477364542, "learning_rate": 1.323815038439813e-10, "loss": 0.3225, "step": 28866 }, { "epoch": 1.9956446595229864, "grad_norm": 0.696990915621174, "learning_rate": 1.283395769946516e-10, "loss": 0.1875, "step": 28867 }, { "epoch": 1.9957137919115104, "grad_norm": 0.6510546099919348, "learning_rate": 1.2436031461393605e-10, "loss": 0.2263, "step": 28868 }, { "epoch": 1.9957829243000345, "grad_norm": 0.703147204680434, "learning_rate": 1.2044371675123956e-10, "loss": 0.3049, "step": 28869 }, { "epoch": 1.9958520566885585, "grad_norm": 0.6532987741913174, "learning_rate": 1.1658978345596705e-10, "loss": 0.2447, "step": 28870 }, { "epoch": 1.9959211890770825, "grad_norm": 0.701172334424065, "learning_rate": 1.1279851477696835e-10, "loss": 0.2196, "step": 28871 }, { "epoch": 1.9959903214656065, "grad_norm": 0.6595011819789385, "learning_rate": 1.0906991076142793e-10, "loss": 0.1999, "step": 28872 }, { "epoch": 1.9960594538541305, "grad_norm": 0.7438063532996918, "learning_rate": 1.0540397145597514e-10, "loss": 0.364, "step": 28873 }, { "epoch": 1.9961285862426545, "grad_norm": 0.7150542168746032, "learning_rate": 1.0180069690612915e-10, "loss": 0.2666, "step": 28874 }, { "epoch": 1.9961977186311786, "grad_norm": 0.62226514571863, "learning_rate": 9.82600871579642e-11, "loss": 0.2311, "step": 28875 }, { "epoch": 1.9962668510197026, "grad_norm": 0.6894663413531769, "learning_rate": 9.4782142254779e-11, "loss": 0.2677, "step": 28876 }, { "epoch": 1.9963359834082266, "grad_norm": 0.7961503731820748, "learning_rate": 9.136686224153757e-11, "loss": 0.3468, "step": 28877 }, { "epoch": 1.9964051157967506, "grad_norm": 0.6949437786180582, "learning_rate": 8.801424715987327e-11, "loss": 0.2572, "step": 28878 }, { "epoch": 1.9964742481852746, "grad_norm": 1.026620446784883, "learning_rate": 8.47242970519746e-11, "loss": 0.2424, "step": 28879 }, { "epoch": 1.9965433805737989, "grad_norm": 0.7834479060416648, "learning_rate": 8.149701195947491e-11, "loss": 0.3846, "step": 28880 }, { "epoch": 1.996612512962323, "grad_norm": 0.6930658091546016, "learning_rate": 7.833239192289732e-11, "loss": 0.2506, "step": 28881 }, { "epoch": 1.996681645350847, "grad_norm": 0.6983242067104605, "learning_rate": 7.523043698165477e-11, "loss": 0.1984, "step": 28882 }, { "epoch": 1.996750777739371, "grad_norm": 0.7800820567983153, "learning_rate": 7.219114717460507e-11, "loss": 0.2254, "step": 28883 }, { "epoch": 1.996819910127895, "grad_norm": 0.7461330163841804, "learning_rate": 6.921452253949579e-11, "loss": 0.3558, "step": 28884 }, { "epoch": 1.996889042516419, "grad_norm": 0.7099246601659975, "learning_rate": 6.630056311407452e-11, "loss": 0.2514, "step": 28885 }, { "epoch": 1.996958174904943, "grad_norm": 0.6491658163733116, "learning_rate": 6.34492689349786e-11, "loss": 0.2442, "step": 28886 }, { "epoch": 1.997027307293467, "grad_norm": 0.6906512319422398, "learning_rate": 6.06606400377352e-11, "loss": 0.3137, "step": 28887 }, { "epoch": 1.997096439681991, "grad_norm": 0.7914147293473117, "learning_rate": 5.793467645731632e-11, "loss": 0.3189, "step": 28888 }, { "epoch": 1.997165572070515, "grad_norm": 0.7067101424506305, "learning_rate": 5.527137822813888e-11, "loss": 0.2345, "step": 28889 }, { "epoch": 1.997234704459039, "grad_norm": 0.6697728397549267, "learning_rate": 5.267074538295447e-11, "loss": 0.1873, "step": 28890 }, { "epoch": 1.997303836847563, "grad_norm": 0.718359775033468, "learning_rate": 5.0132777955069766e-11, "loss": 0.3156, "step": 28891 }, { "epoch": 1.997372969236087, "grad_norm": 0.6554928534058176, "learning_rate": 4.765747597557102e-11, "loss": 0.2117, "step": 28892 }, { "epoch": 1.997442101624611, "grad_norm": 0.6886464977425578, "learning_rate": 4.524483947609959e-11, "loss": 0.2279, "step": 28893 }, { "epoch": 1.9975112340131351, "grad_norm": 0.7288786426222404, "learning_rate": 4.289486848663149e-11, "loss": 0.3187, "step": 28894 }, { "epoch": 1.9975803664016591, "grad_norm": 0.8834860406638064, "learning_rate": 4.060756303658764e-11, "loss": 0.3358, "step": 28895 }, { "epoch": 1.9976494987901832, "grad_norm": 0.7090774769479236, "learning_rate": 3.838292315483383e-11, "loss": 0.2399, "step": 28896 }, { "epoch": 1.9977186311787072, "grad_norm": 0.7331309154326539, "learning_rate": 3.622094886912564e-11, "loss": 0.2739, "step": 28897 }, { "epoch": 1.9977877635672312, "grad_norm": 0.6884478686654878, "learning_rate": 3.412164020666353e-11, "loss": 0.315, "step": 28898 }, { "epoch": 1.9978568959557552, "grad_norm": 0.7355088137974062, "learning_rate": 3.2084997193537745e-11, "loss": 0.2355, "step": 28899 }, { "epoch": 1.9979260283442795, "grad_norm": 0.7131432323174279, "learning_rate": 3.011101985528342e-11, "loss": 0.2226, "step": 28900 }, { "epoch": 1.9979951607328035, "grad_norm": 0.7438168225125615, "learning_rate": 2.8199708216880562e-11, "loss": 0.3163, "step": 28901 }, { "epoch": 1.9980642931213275, "grad_norm": 0.7353502994405315, "learning_rate": 2.635106230164386e-11, "loss": 0.245, "step": 28902 }, { "epoch": 1.9981334255098515, "grad_norm": 0.7969485852697847, "learning_rate": 2.456508213399822e-11, "loss": 0.2205, "step": 28903 }, { "epoch": 1.9982025578983755, "grad_norm": 0.6994911708825158, "learning_rate": 2.284176773503788e-11, "loss": 0.2964, "step": 28904 }, { "epoch": 1.9982716902868995, "grad_norm": 0.683398460713319, "learning_rate": 2.1181119126967298e-11, "loss": 0.3334, "step": 28905 }, { "epoch": 1.9983408226754236, "grad_norm": 0.7064478177361788, "learning_rate": 1.9583136330880713e-11, "loss": 0.2423, "step": 28906 }, { "epoch": 1.9984099550639476, "grad_norm": 0.7167619755283111, "learning_rate": 1.8047819366207032e-11, "loss": 0.2497, "step": 28907 }, { "epoch": 1.9984790874524716, "grad_norm": 0.7180749843462999, "learning_rate": 1.6575168252375152e-11, "loss": 0.3499, "step": 28908 }, { "epoch": 1.9985482198409956, "grad_norm": 0.7760785162082305, "learning_rate": 1.5165183007703755e-11, "loss": 0.3366, "step": 28909 }, { "epoch": 1.9986173522295196, "grad_norm": 0.7068412865191604, "learning_rate": 1.381786365051152e-11, "loss": 0.2909, "step": 28910 }, { "epoch": 1.9986864846180437, "grad_norm": 0.6884793457621707, "learning_rate": 1.2533210196896684e-11, "loss": 0.2057, "step": 28911 }, { "epoch": 1.9987556170065677, "grad_norm": 0.7338796003409158, "learning_rate": 1.1311222663512589e-11, "loss": 0.3997, "step": 28912 }, { "epoch": 1.9988247493950917, "grad_norm": 1.1323326564715626, "learning_rate": 1.0151901065347248e-11, "loss": 0.2393, "step": 28913 }, { "epoch": 1.9988938817836157, "grad_norm": 0.6864426606552839, "learning_rate": 9.055245416833558e-12, "loss": 0.2052, "step": 28914 }, { "epoch": 1.9989630141721397, "grad_norm": 0.7199115916492747, "learning_rate": 8.02125573240442e-12, "loss": 0.2582, "step": 28915 }, { "epoch": 1.9990321465606637, "grad_norm": 10.235087003405711, "learning_rate": 7.049932024272288e-12, "loss": 0.4179, "step": 28916 }, { "epoch": 1.9991012789491878, "grad_norm": 0.6897554185847004, "learning_rate": 6.141274305204725e-12, "loss": 0.2264, "step": 28917 }, { "epoch": 1.9991704113377118, "grad_norm": 0.8200011326318364, "learning_rate": 5.295282585748851e-12, "loss": 0.2448, "step": 28918 }, { "epoch": 1.9992395437262358, "grad_norm": 0.7571179867986748, "learning_rate": 4.5119568770068956e-12, "loss": 0.3149, "step": 28919 }, { "epoch": 1.9993086761147598, "grad_norm": 0.7273211090144452, "learning_rate": 3.791297189525978e-12, "loss": 0.3244, "step": 28920 }, { "epoch": 1.9993778085032838, "grad_norm": 0.6371198098135522, "learning_rate": 3.1333035310776583e-12, "loss": 0.2094, "step": 28921 }, { "epoch": 1.9994469408918079, "grad_norm": 0.6810573107926885, "learning_rate": 2.537975910543722e-12, "loss": 0.2253, "step": 28922 }, { "epoch": 1.9995160732803319, "grad_norm": 0.7573456052776731, "learning_rate": 2.005314335140618e-12, "loss": 0.3781, "step": 28923 }, { "epoch": 1.999585205668856, "grad_norm": 0.7039809061800506, "learning_rate": 1.5353188120847962e-12, "loss": 0.2792, "step": 28924 }, { "epoch": 1.99965433805738, "grad_norm": 0.7492377929172311, "learning_rate": 1.1279893463722602e-12, "loss": 0.2532, "step": 28925 }, { "epoch": 1.999723470445904, "grad_norm": 0.7244275463649561, "learning_rate": 7.833259441092366e-13, "loss": 0.2286, "step": 28926 }, { "epoch": 1.999792602834428, "grad_norm": 0.7550757684559005, "learning_rate": 5.013286091815061e-13, "loss": 0.3237, "step": 28927 }, { "epoch": 1.999861735222952, "grad_norm": 0.7413146847490546, "learning_rate": 2.8199734436462624e-13, "loss": 0.2373, "step": 28928 }, { "epoch": 1.999930867611476, "grad_norm": 0.7172542032923016, "learning_rate": 1.2533215354437745e-13, "loss": 0.2244, "step": 28929 }, { "epoch": 2.0, "grad_norm": 0.6900019794188633, "learning_rate": 3.1333038386094363e-14, "loss": 0.2268, "step": 28930 }, { "epoch": 2.0, "step": 28930, "total_flos": 1.2606539534696448e+17, "train_loss": 0.31978979713511724, "train_runtime": 151004.5625, "train_samples_per_second": 49.043, "train_steps_per_second": 0.192 } ], "logging_steps": 1.0, "max_steps": 28930, "num_input_tokens_seen": 0, "num_train_epochs": 2, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 1.2606539534696448e+17, "train_batch_size": 4, "trial_name": null, "trial_params": null }