{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 1.0, "eval_steps": 500, "global_step": 661, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "eob_count_fixed": 23199, "eob_loss_fixed": 37.5702, "epoch": 0.0015137180700094607, "grad_norm": 22.99761199951172, "learning_rate": 0.0, "loss": 12.2763, "loss_dynamic": 6.3524, "loss_fixed": 18.2002, "step": 1 }, { "eob_count_fixed": 23299, "eob_loss_fixed": 37.6787, "epoch": 0.0030274361400189215, "grad_norm": 23.10260581970215, "learning_rate": 5.000000000000001e-07, "loss": 12.2863, "loss_dynamic": 6.4082, "loss_fixed": 18.1645, "step": 2 }, { "eob_count_fixed": 23139, "eob_loss_fixed": 37.251, "epoch": 0.004541154210028382, "grad_norm": 24.02956771850586, "learning_rate": 1.0000000000000002e-06, "loss": 12.0554, "loss_dynamic": 6.2094, "loss_fixed": 17.9015, "step": 3 }, { "eob_count_fixed": 23099, "eob_loss_fixed": 35.4702, "epoch": 0.006054872280037843, "grad_norm": 25.653656005859375, "learning_rate": 1.5e-06, "loss": 11.4066, "loss_dynamic": 5.6879, "loss_fixed": 17.1253, "step": 4 }, { "eob_count_fixed": 23136, "eob_loss_fixed": 29.686, "epoch": 0.007568590350047304, "grad_norm": 29.759170532226562, "learning_rate": 2.0000000000000003e-06, "loss": 9.0617, "loss_dynamic": 3.676, "loss_fixed": 14.4473, "step": 5 }, { "eob_count_fixed": 23232, "eob_loss_fixed": 23.9456, "epoch": 0.009082308420056764, "grad_norm": 13.6844482421875, "learning_rate": 2.5e-06, "loss": 6.9169, "loss_dynamic": 1.7703, "loss_fixed": 12.0634, "step": 6 }, { "eob_count_fixed": 23018, "eob_loss_fixed": 15.4875, "epoch": 0.010596026490066225, "grad_norm": 12.741179466247559, "learning_rate": 3e-06, "loss": 5.3576, "loss_dynamic": 1.9288, "loss_fixed": 8.7863, "step": 7 }, { "eob_count_fixed": 23309, "eob_loss_fixed": 7.6675, "epoch": 0.012109744560075686, "grad_norm": 7.749817371368408, "learning_rate": 3.5e-06, "loss": 3.6295, "loss_dynamic": 1.8276, "loss_fixed": 5.4313, "step": 8 }, { "eob_count_fixed": 23035, "eob_loss_fixed": 4.717, "epoch": 0.013623462630085147, "grad_norm": 2.5551257133483887, "learning_rate": 4.000000000000001e-06, "loss": 2.6034, "loss_dynamic": 1.5386, "loss_fixed": 3.6682, "step": 9 }, { "eob_count_fixed": 23228, "eob_loss_fixed": 1.3679, "epoch": 0.015137180700094607, "grad_norm": 2.7040646076202393, "learning_rate": 4.5e-06, "loss": 2.0081, "loss_dynamic": 1.6989, "loss_fixed": 2.3172, "step": 10 }, { "eob_count_fixed": 23066, "eob_loss_fixed": 1.1504, "epoch": 0.016650898770104068, "grad_norm": 1.7488154172897339, "learning_rate": 5e-06, "loss": 1.8968, "loss_dynamic": 1.817, "loss_fixed": 1.9767, "step": 11 }, { "eob_count_fixed": 23178, "eob_loss_fixed": 1.2767, "epoch": 0.01816461684011353, "grad_norm": 1.0264207124710083, "learning_rate": 5.500000000000001e-06, "loss": 1.7239, "loss_dynamic": 1.6713, "loss_fixed": 1.7766, "step": 12 }, { "eob_count_fixed": 23009, "eob_loss_fixed": 1.2079, "epoch": 0.01967833491012299, "grad_norm": 0.8516889810562134, "learning_rate": 6e-06, "loss": 1.6123, "loss_dynamic": 1.5604, "loss_fixed": 1.6642, "step": 13 }, { "eob_count_fixed": 22993, "eob_loss_fixed": 0.7291, "epoch": 0.02119205298013245, "grad_norm": 0.6665084958076477, "learning_rate": 6.5000000000000004e-06, "loss": 1.6006, "loss_dynamic": 1.674, "loss_fixed": 1.5273, "step": 14 }, { "eob_count_fixed": 23123, "eob_loss_fixed": 0.4957, "epoch": 0.02270577105014191, "grad_norm": 0.7430596947669983, "learning_rate": 7e-06, "loss": 1.4665, "loss_dynamic": 1.4835, "loss_fixed": 1.4495, "step": 15 }, { "eob_count_fixed": 23221, "eob_loss_fixed": 0.4917, "epoch": 0.024219489120151372, "grad_norm": 0.4606741964817047, "learning_rate": 7.500000000000001e-06, "loss": 1.4298, "loss_dynamic": 1.5104, "loss_fixed": 1.3492, "step": 16 }, { "eob_count_fixed": 23082, "eob_loss_fixed": 0.5079, "epoch": 0.025733207190160833, "grad_norm": 0.3731869161128998, "learning_rate": 8.000000000000001e-06, "loss": 1.4026, "loss_dynamic": 1.4812, "loss_fixed": 1.324, "step": 17 }, { "eob_count_fixed": 23165, "eob_loss_fixed": 0.4555, "epoch": 0.027246925260170293, "grad_norm": 0.4283307194709778, "learning_rate": 8.5e-06, "loss": 1.37, "loss_dynamic": 1.465, "loss_fixed": 1.2751, "step": 18 }, { "eob_count_fixed": 23241, "eob_loss_fixed": 0.3812, "epoch": 0.028760643330179754, "grad_norm": 0.32799842953681946, "learning_rate": 9e-06, "loss": 1.3092, "loss_dynamic": 1.3731, "loss_fixed": 1.2452, "step": 19 }, { "eob_count_fixed": 23265, "eob_loss_fixed": 0.3706, "epoch": 0.030274361400189215, "grad_norm": 0.3386831283569336, "learning_rate": 9.5e-06, "loss": 1.2995, "loss_dynamic": 1.4111, "loss_fixed": 1.1879, "step": 20 }, { "eob_count_fixed": 23060, "eob_loss_fixed": 0.386, "epoch": 0.031788079470198675, "grad_norm": 0.3010108768939972, "learning_rate": 1e-05, "loss": 1.2642, "loss_dynamic": 1.3164, "loss_fixed": 1.2121, "step": 21 }, { "eob_count_fixed": 23187, "eob_loss_fixed": 0.3847, "epoch": 0.033301797540208136, "grad_norm": 0.2816918194293976, "learning_rate": 9.999939948643008e-06, "loss": 1.2607, "loss_dynamic": 1.364, "loss_fixed": 1.1575, "step": 22 }, { "eob_count_fixed": 23085, "eob_loss_fixed": 0.3647, "epoch": 0.0348155156102176, "grad_norm": 0.29582875967025757, "learning_rate": 9.999759796014496e-06, "loss": 1.244, "loss_dynamic": 1.335, "loss_fixed": 1.1531, "step": 23 }, { "eob_count_fixed": 23113, "eob_loss_fixed": 0.3212, "epoch": 0.03632923368022706, "grad_norm": 0.2852838635444641, "learning_rate": 9.999459546441833e-06, "loss": 1.3175, "loss_dynamic": 1.4818, "loss_fixed": 1.1532, "step": 24 }, { "eob_count_fixed": 23113, "eob_loss_fixed": 0.2942, "epoch": 0.03784295175023652, "grad_norm": 0.24114593863487244, "learning_rate": 9.99903920713717e-06, "loss": 1.2316, "loss_dynamic": 1.3213, "loss_fixed": 1.1419, "step": 25 }, { "eob_count_fixed": 23093, "eob_loss_fixed": 0.2906, "epoch": 0.03935666982024598, "grad_norm": 0.2545201778411865, "learning_rate": 9.998498788197287e-06, "loss": 1.2935, "loss_dynamic": 1.4509, "loss_fixed": 1.1361, "step": 26 }, { "eob_count_fixed": 23220, "eob_loss_fixed": 0.2979, "epoch": 0.04087038789025544, "grad_norm": 0.2507128417491913, "learning_rate": 9.997838302603344e-06, "loss": 1.2633, "loss_dynamic": 1.4117, "loss_fixed": 1.1149, "step": 27 }, { "eob_count_fixed": 23189, "eob_loss_fixed": 0.3019, "epoch": 0.0423841059602649, "grad_norm": 0.23537935316562653, "learning_rate": 9.99705776622056e-06, "loss": 1.2344, "loss_dynamic": 1.3644, "loss_fixed": 1.1045, "step": 28 }, { "eob_count_fixed": 23146, "eob_loss_fixed": 0.3207, "epoch": 0.04389782403027436, "grad_norm": 0.24033816158771515, "learning_rate": 9.996157197797843e-06, "loss": 1.2313, "loss_dynamic": 1.353, "loss_fixed": 1.1096, "step": 29 }, { "eob_count_fixed": 23204, "eob_loss_fixed": 0.2988, "epoch": 0.04541154210028382, "grad_norm": 0.22857218980789185, "learning_rate": 9.995136618967336e-06, "loss": 1.2024, "loss_dynamic": 1.317, "loss_fixed": 1.0877, "step": 30 }, { "eob_count_fixed": 23213, "eob_loss_fixed": 0.2751, "epoch": 0.04692526017029328, "grad_norm": 0.2354959398508072, "learning_rate": 9.993996054243897e-06, "loss": 1.2432, "loss_dynamic": 1.3962, "loss_fixed": 1.0903, "step": 31 }, { "eob_count_fixed": 23243, "eob_loss_fixed": 0.2554, "epoch": 0.048438978240302744, "grad_norm": 0.2223290354013443, "learning_rate": 9.992735531024508e-06, "loss": 1.216, "loss_dynamic": 1.3615, "loss_fixed": 1.0705, "step": 32 }, { "eob_count_fixed": 22997, "eob_loss_fixed": 0.2504, "epoch": 0.049952696310312204, "grad_norm": 0.22776173055171967, "learning_rate": 9.991355079587623e-06, "loss": 1.1966, "loss_dynamic": 1.3181, "loss_fixed": 1.075, "step": 33 }, { "eob_count_fixed": 23203, "eob_loss_fixed": 0.2664, "epoch": 0.051466414380321665, "grad_norm": 0.24485988914966583, "learning_rate": 9.989854733092435e-06, "loss": 1.2428, "loss_dynamic": 1.414, "loss_fixed": 1.0717, "step": 34 }, { "eob_count_fixed": 23162, "eob_loss_fixed": 0.2879, "epoch": 0.052980132450331126, "grad_norm": 0.23885564506053925, "learning_rate": 9.98823452757808e-06, "loss": 1.2612, "loss_dynamic": 1.4372, "loss_fixed": 1.0851, "step": 35 }, { "eob_count_fixed": 23090, "eob_loss_fixed": 0.2708, "epoch": 0.054493850520340587, "grad_norm": 0.2277117222547531, "learning_rate": 9.986494501962772e-06, "loss": 1.2014, "loss_dynamic": 1.3393, "loss_fixed": 1.0635, "step": 36 }, { "eob_count_fixed": 23157, "eob_loss_fixed": 0.2529, "epoch": 0.05600756859035005, "grad_norm": 0.27555766701698303, "learning_rate": 9.984634698042875e-06, "loss": 1.2246, "loss_dynamic": 1.3668, "loss_fixed": 1.0824, "step": 37 }, { "eob_count_fixed": 22965, "eob_loss_fixed": 0.2458, "epoch": 0.05752128666035951, "grad_norm": 0.22988812625408173, "learning_rate": 9.982655160491883e-06, "loss": 1.1768, "loss_dynamic": 1.3116, "loss_fixed": 1.0421, "step": 38 }, { "eob_count_fixed": 23219, "eob_loss_fixed": 0.2482, "epoch": 0.05903500473036897, "grad_norm": 0.25177380442619324, "learning_rate": 9.98055593685937e-06, "loss": 1.2074, "loss_dynamic": 1.374, "loss_fixed": 1.0408, "step": 39 }, { "eob_count_fixed": 23114, "eob_loss_fixed": 0.2616, "epoch": 0.06054872280037843, "grad_norm": 0.22855891287326813, "learning_rate": 9.978337077569819e-06, "loss": 1.2153, "loss_dynamic": 1.4029, "loss_fixed": 1.0278, "step": 40 }, { "eob_count_fixed": 23011, "eob_loss_fixed": 0.2605, "epoch": 0.06206244087038789, "grad_norm": 0.22180484235286713, "learning_rate": 9.975998635921442e-06, "loss": 1.1768, "loss_dynamic": 1.3253, "loss_fixed": 1.0283, "step": 41 }, { "eob_count_fixed": 23144, "eob_loss_fixed": 0.2516, "epoch": 0.06357615894039735, "grad_norm": 0.21961922943592072, "learning_rate": 9.973540668084872e-06, "loss": 1.1925, "loss_dynamic": 1.3265, "loss_fixed": 1.0586, "step": 42 }, { "eob_count_fixed": 22962, "eob_loss_fixed": 0.2297, "epoch": 0.06508987701040682, "grad_norm": 0.22070494294166565, "learning_rate": 9.970963233101831e-06, "loss": 1.2036, "loss_dynamic": 1.3599, "loss_fixed": 1.0474, "step": 43 }, { "eob_count_fixed": 23109, "eob_loss_fixed": 0.2431, "epoch": 0.06660359508041627, "grad_norm": 0.22726963460445404, "learning_rate": 9.968266392883708e-06, "loss": 1.2339, "loss_dynamic": 1.4106, "loss_fixed": 1.0572, "step": 44 }, { "eob_count_fixed": 23096, "eob_loss_fixed": 0.2631, "epoch": 0.06811731315042574, "grad_norm": 0.21237440407276154, "learning_rate": 9.965450212210068e-06, "loss": 1.2325, "loss_dynamic": 1.3856, "loss_fixed": 1.0794, "step": 45 }, { "eob_count_fixed": 23267, "eob_loss_fixed": 0.2505, "epoch": 0.0696310312204352, "grad_norm": 0.2193518728017807, "learning_rate": 9.962514758727098e-06, "loss": 1.1381, "loss_dynamic": 1.2606, "loss_fixed": 1.0157, "step": 46 }, { "eob_count_fixed": 23105, "eob_loss_fixed": 0.2397, "epoch": 0.07114474929044466, "grad_norm": 0.24308504164218903, "learning_rate": 9.959460102945986e-06, "loss": 1.2098, "loss_dynamic": 1.3718, "loss_fixed": 1.0477, "step": 47 }, { "eob_count_fixed": 22978, "eob_loss_fixed": 0.24, "epoch": 0.07265846736045412, "grad_norm": 0.22213910520076752, "learning_rate": 9.956286318241222e-06, "loss": 1.1963, "loss_dynamic": 1.3691, "loss_fixed": 1.0235, "step": 48 }, { "eob_count_fixed": 23128, "eob_loss_fixed": 0.2446, "epoch": 0.07417218543046358, "grad_norm": 0.23669806122779846, "learning_rate": 9.952993480848837e-06, "loss": 1.2313, "loss_dynamic": 1.4134, "loss_fixed": 1.0493, "step": 49 }, { "eob_count_fixed": 23254, "eob_loss_fixed": 0.2514, "epoch": 0.07568590350047304, "grad_norm": 0.21434348821640015, "learning_rate": 9.949581669864571e-06, "loss": 1.2048, "loss_dynamic": 1.3701, "loss_fixed": 1.0396, "step": 50 }, { "eob_count_fixed": 23147, "eob_loss_fixed": 0.2258, "epoch": 0.0771996215704825, "grad_norm": 0.21070963144302368, "learning_rate": 9.946050967241979e-06, "loss": 1.0676, "loss_dynamic": 1.1575, "loss_fixed": 0.9778, "step": 51 }, { "eob_count_fixed": 23011, "eob_loss_fixed": 0.2234, "epoch": 0.07871333964049196, "grad_norm": 0.22653864324092865, "learning_rate": 9.94240145779045e-06, "loss": 1.2218, "loss_dynamic": 1.4064, "loss_fixed": 1.0372, "step": 52 }, { "eob_count_fixed": 23107, "eob_loss_fixed": 0.2429, "epoch": 0.08022705771050143, "grad_norm": 0.22146368026733398, "learning_rate": 9.938633229173185e-06, "loss": 1.1956, "loss_dynamic": 1.3452, "loss_fixed": 1.046, "step": 53 }, { "eob_count_fixed": 23237, "eob_loss_fixed": 0.2517, "epoch": 0.08174077578051088, "grad_norm": 0.2323981374502182, "learning_rate": 9.934746371905082e-06, "loss": 1.1979, "loss_dynamic": 1.3471, "loss_fixed": 1.0488, "step": 54 }, { "eob_count_fixed": 23084, "eob_loss_fixed": 0.2374, "epoch": 0.08325449385052035, "grad_norm": 0.2093762755393982, "learning_rate": 9.930740979350557e-06, "loss": 1.1546, "loss_dynamic": 1.2663, "loss_fixed": 1.0429, "step": 55 }, { "eob_count_fixed": 23182, "eob_loss_fixed": 0.2256, "epoch": 0.0847682119205298, "grad_norm": 0.23099260032176971, "learning_rate": 9.926617147721318e-06, "loss": 1.1641, "loss_dynamic": 1.3164, "loss_fixed": 1.0117, "step": 56 }, { "eob_count_fixed": 23218, "eob_loss_fixed": 0.2231, "epoch": 0.08628192999053927, "grad_norm": 0.24069924652576447, "learning_rate": 9.92237497607404e-06, "loss": 1.183, "loss_dynamic": 1.351, "loss_fixed": 1.0149, "step": 57 }, { "eob_count_fixed": 23055, "eob_loss_fixed": 0.2251, "epoch": 0.08779564806054872, "grad_norm": 0.22786876559257507, "learning_rate": 9.918014566307986e-06, "loss": 1.1754, "loss_dynamic": 1.3353, "loss_fixed": 1.0155, "step": 58 }, { "eob_count_fixed": 23078, "eob_loss_fixed": 0.2335, "epoch": 0.08930936613055819, "grad_norm": 0.23150981962680817, "learning_rate": 9.913536023162564e-06, "loss": 1.1526, "loss_dynamic": 1.2999, "loss_fixed": 1.0053, "step": 59 }, { "eob_count_fixed": 23070, "eob_loss_fixed": 0.2281, "epoch": 0.09082308420056764, "grad_norm": 0.23753342032432556, "learning_rate": 9.908939454214814e-06, "loss": 1.1438, "loss_dynamic": 1.2801, "loss_fixed": 1.0076, "step": 60 }, { "eob_count_fixed": 23128, "eob_loss_fixed": 0.2415, "epoch": 0.09233680227057711, "grad_norm": 0.2215557098388672, "learning_rate": 9.904224969876817e-06, "loss": 1.1569, "loss_dynamic": 1.2919, "loss_fixed": 1.0218, "step": 61 }, { "eob_count_fixed": 23085, "eob_loss_fixed": 0.2209, "epoch": 0.09385052034058657, "grad_norm": 0.23713566362857819, "learning_rate": 9.899392683393043e-06, "loss": 1.1457, "loss_dynamic": 1.3116, "loss_fixed": 0.9798, "step": 62 }, { "eob_count_fixed": 23004, "eob_loss_fixed": 0.2288, "epoch": 0.09536423841059603, "grad_norm": 0.22792600095272064, "learning_rate": 9.89444271083764e-06, "loss": 1.188, "loss_dynamic": 1.3678, "loss_fixed": 1.0083, "step": 63 }, { "eob_count_fixed": 23122, "eob_loss_fixed": 0.2209, "epoch": 0.09687795648060549, "grad_norm": 0.2318040281534195, "learning_rate": 9.889375171111635e-06, "loss": 1.1187, "loss_dynamic": 1.2283, "loss_fixed": 1.0092, "step": 64 }, { "eob_count_fixed": 23107, "eob_loss_fixed": 0.2301, "epoch": 0.09839167455061495, "grad_norm": 0.2221527397632599, "learning_rate": 9.88419018594008e-06, "loss": 1.1641, "loss_dynamic": 1.3004, "loss_fixed": 1.0278, "step": 65 }, { "eob_count_fixed": 23083, "eob_loss_fixed": 0.2424, "epoch": 0.09990539262062441, "grad_norm": 0.2430257797241211, "learning_rate": 9.878887879869137e-06, "loss": 1.1961, "loss_dynamic": 1.353, "loss_fixed": 1.0392, "step": 66 }, { "eob_count_fixed": 23120, "eob_loss_fixed": 0.2311, "epoch": 0.10141911069063388, "grad_norm": 0.2224264144897461, "learning_rate": 9.873468380263072e-06, "loss": 1.1113, "loss_dynamic": 1.2305, "loss_fixed": 0.9921, "step": 67 }, { "eob_count_fixed": 23073, "eob_loss_fixed": 0.2248, "epoch": 0.10293282876064333, "grad_norm": 0.22271840274333954, "learning_rate": 9.86793181730121e-06, "loss": 1.1648, "loss_dynamic": 1.3045, "loss_fixed": 1.0251, "step": 68 }, { "eob_count_fixed": 23106, "eob_loss_fixed": 0.2254, "epoch": 0.1044465468306528, "grad_norm": 0.2555963099002838, "learning_rate": 9.862278323974798e-06, "loss": 1.2321, "loss_dynamic": 1.456, "loss_fixed": 1.0082, "step": 69 }, { "eob_count_fixed": 23027, "eob_loss_fixed": 0.2295, "epoch": 0.10596026490066225, "grad_norm": 0.22944504022598267, "learning_rate": 9.856508036083813e-06, "loss": 1.1923, "loss_dynamic": 1.3816, "loss_fixed": 1.0031, "step": 70 }, { "eob_count_fixed": 22965, "eob_loss_fixed": 0.2318, "epoch": 0.10747398297067172, "grad_norm": 0.23594002425670624, "learning_rate": 9.850621092233705e-06, "loss": 1.1865, "loss_dynamic": 1.3575, "loss_fixed": 1.0156, "step": 71 }, { "eob_count_fixed": 23154, "eob_loss_fixed": 0.2302, "epoch": 0.10898770104068117, "grad_norm": 0.22777444124221802, "learning_rate": 9.844617633832058e-06, "loss": 1.1989, "loss_dynamic": 1.3646, "loss_fixed": 1.0332, "step": 72 }, { "eob_count_fixed": 23089, "eob_loss_fixed": 0.2101, "epoch": 0.11050141911069064, "grad_norm": 0.20970605313777924, "learning_rate": 9.838497805085205e-06, "loss": 1.1583, "loss_dynamic": 1.308, "loss_fixed": 1.0087, "step": 73 }, { "eob_count_fixed": 23268, "eob_loss_fixed": 0.2258, "epoch": 0.1120151371807001, "grad_norm": 0.2294132113456726, "learning_rate": 9.832261752994749e-06, "loss": 1.1676, "loss_dynamic": 1.3289, "loss_fixed": 1.0063, "step": 74 }, { "eob_count_fixed": 23221, "eob_loss_fixed": 0.224, "epoch": 0.11352885525070956, "grad_norm": 0.21748290956020355, "learning_rate": 9.825909627354051e-06, "loss": 1.1323, "loss_dynamic": 1.2408, "loss_fixed": 1.0238, "step": 75 }, { "eob_count_fixed": 23250, "eob_loss_fixed": 0.2191, "epoch": 0.11504257332071902, "grad_norm": 0.20814786851406097, "learning_rate": 9.819441580744613e-06, "loss": 1.1534, "loss_dynamic": 1.3047, "loss_fixed": 1.0021, "step": 76 }, { "eob_count_fixed": 23180, "eob_loss_fixed": 0.2272, "epoch": 0.11655629139072848, "grad_norm": 0.22371232509613037, "learning_rate": 9.812857768532429e-06, "loss": 1.1915, "loss_dynamic": 1.3642, "loss_fixed": 1.0189, "step": 77 }, { "eob_count_fixed": 23113, "eob_loss_fixed": 0.2174, "epoch": 0.11807000946073794, "grad_norm": 0.22483351826667786, "learning_rate": 9.806158348864239e-06, "loss": 1.1941, "loss_dynamic": 1.3725, "loss_fixed": 1.0157, "step": 78 }, { "eob_count_fixed": 23000, "eob_loss_fixed": 0.2263, "epoch": 0.1195837275307474, "grad_norm": 0.23057368397712708, "learning_rate": 9.799343482663741e-06, "loss": 1.1913, "loss_dynamic": 1.3606, "loss_fixed": 1.022, "step": 79 }, { "eob_count_fixed": 23094, "eob_loss_fixed": 0.2272, "epoch": 0.12109744560075686, "grad_norm": 0.233402818441391, "learning_rate": 9.792413333627721e-06, "loss": 1.1733, "loss_dynamic": 1.3243, "loss_fixed": 1.0222, "step": 80 }, { "eob_count_fixed": 23004, "eob_loss_fixed": 0.221, "epoch": 0.12261116367076633, "grad_norm": 0.22432219982147217, "learning_rate": 9.785368068222121e-06, "loss": 1.1741, "loss_dynamic": 1.3548, "loss_fixed": 0.9935, "step": 81 }, { "eob_count_fixed": 23032, "eob_loss_fixed": 0.2039, "epoch": 0.12412488174077578, "grad_norm": 0.22042952477931976, "learning_rate": 9.778207855678036e-06, "loss": 1.1762, "loss_dynamic": 1.3688, "loss_fixed": 0.9836, "step": 82 }, { "eob_count_fixed": 23119, "eob_loss_fixed": 0.208, "epoch": 0.12563859981078523, "grad_norm": 0.2125842571258545, "learning_rate": 9.770932867987664e-06, "loss": 1.1275, "loss_dynamic": 1.2751, "loss_fixed": 0.9798, "step": 83 }, { "eob_count_fixed": 23335, "eob_loss_fixed": 0.2133, "epoch": 0.1271523178807947, "grad_norm": 0.20397071540355682, "learning_rate": 9.763543279900154e-06, "loss": 1.0968, "loss_dynamic": 1.2072, "loss_fixed": 0.9863, "step": 84 }, { "eob_count_fixed": 23092, "eob_loss_fixed": 0.2096, "epoch": 0.12866603595080417, "grad_norm": 0.21145540475845337, "learning_rate": 9.756039268917424e-06, "loss": 1.1249, "loss_dynamic": 1.2857, "loss_fixed": 0.964, "step": 85 }, { "eob_count_fixed": 23085, "eob_loss_fixed": 0.211, "epoch": 0.13017975402081364, "grad_norm": 0.2182677537202835, "learning_rate": 9.74842101528989e-06, "loss": 1.1688, "loss_dynamic": 1.3497, "loss_fixed": 0.988, "step": 86 }, { "eob_count_fixed": 23019, "eob_loss_fixed": 0.2162, "epoch": 0.13169347209082308, "grad_norm": 0.23218557238578796, "learning_rate": 9.74068870201214e-06, "loss": 1.1465, "loss_dynamic": 1.2929, "loss_fixed": 1.0001, "step": 87 }, { "eob_count_fixed": 23138, "eob_loss_fixed": 0.2249, "epoch": 0.13320719016083254, "grad_norm": 0.2216981053352356, "learning_rate": 9.732842514818537e-06, "loss": 1.1794, "loss_dynamic": 1.3547, "loss_fixed": 1.0041, "step": 88 }, { "eob_count_fixed": 23154, "eob_loss_fixed": 0.2218, "epoch": 0.134720908230842, "grad_norm": 0.21972118318080902, "learning_rate": 9.724882642178757e-06, "loss": 1.1451, "loss_dynamic": 1.2969, "loss_fixed": 0.9934, "step": 89 }, { "eob_count_fixed": 23158, "eob_loss_fixed": 0.2067, "epoch": 0.13623462630085148, "grad_norm": 0.250344455242157, "learning_rate": 9.716809275293258e-06, "loss": 1.1561, "loss_dynamic": 1.3091, "loss_fixed": 1.0032, "step": 90 }, { "eob_count_fixed": 23218, "eob_loss_fixed": 0.2006, "epoch": 0.13774834437086092, "grad_norm": 0.225043386220932, "learning_rate": 9.708622608088697e-06, "loss": 1.1628, "loss_dynamic": 1.3573, "loss_fixed": 0.9682, "step": 91 }, { "eob_count_fixed": 22882, "eob_loss_fixed": 0.2271, "epoch": 0.1392620624408704, "grad_norm": 0.21307989954948425, "learning_rate": 9.700322837213263e-06, "loss": 1.1555, "loss_dynamic": 1.3148, "loss_fixed": 0.9963, "step": 92 }, { "eob_count_fixed": 23192, "eob_loss_fixed": 0.23, "epoch": 0.14077578051087986, "grad_norm": 0.20988883078098297, "learning_rate": 9.691910162031958e-06, "loss": 1.1889, "loss_dynamic": 1.3569, "loss_fixed": 1.0209, "step": 93 }, { "eob_count_fixed": 23067, "eob_loss_fixed": 0.2096, "epoch": 0.14228949858088932, "grad_norm": 0.2016877979040146, "learning_rate": 9.683384784621807e-06, "loss": 1.1415, "loss_dynamic": 1.3225, "loss_fixed": 0.9605, "step": 94 }, { "eob_count_fixed": 23129, "eob_loss_fixed": 0.2046, "epoch": 0.14380321665089876, "grad_norm": 0.215549036860466, "learning_rate": 9.674746909767002e-06, "loss": 1.1484, "loss_dynamic": 1.3044, "loss_fixed": 0.9924, "step": 95 }, { "eob_count_fixed": 23034, "eob_loss_fixed": 0.1956, "epoch": 0.14531693472090823, "grad_norm": 0.21526315808296204, "learning_rate": 9.665996744953989e-06, "loss": 1.1554, "loss_dynamic": 1.3594, "loss_fixed": 0.9513, "step": 96 }, { "eob_count_fixed": 23100, "eob_loss_fixed": 0.2029, "epoch": 0.1468306527909177, "grad_norm": 0.23761266469955444, "learning_rate": 9.65713450036647e-06, "loss": 1.1173, "loss_dynamic": 1.2916, "loss_fixed": 0.943, "step": 97 }, { "eob_count_fixed": 23231, "eob_loss_fixed": 0.212, "epoch": 0.14834437086092717, "grad_norm": 0.21916434168815613, "learning_rate": 9.648160388880374e-06, "loss": 1.1499, "loss_dynamic": 1.317, "loss_fixed": 0.9827, "step": 98 }, { "eob_count_fixed": 23184, "eob_loss_fixed": 0.2142, "epoch": 0.1498580889309366, "grad_norm": 0.2286253124475479, "learning_rate": 9.63907462605873e-06, "loss": 1.1442, "loss_dynamic": 1.315, "loss_fixed": 0.9733, "step": 99 }, { "eob_count_fixed": 23249, "eob_loss_fixed": 0.2039, "epoch": 0.15137180700094607, "grad_norm": 0.21630607545375824, "learning_rate": 9.629877430146493e-06, "loss": 1.1521, "loss_dynamic": 1.3272, "loss_fixed": 0.9769, "step": 100 }, { "eob_count_fixed": 23244, "eob_loss_fixed": 0.2132, "epoch": 0.15288552507095554, "grad_norm": 0.23888744413852692, "learning_rate": 9.6205690220653e-06, "loss": 1.1894, "loss_dynamic": 1.4, "loss_fixed": 0.9788, "step": 101 }, { "eob_count_fixed": 23113, "eob_loss_fixed": 0.2095, "epoch": 0.154399243140965, "grad_norm": 0.2389604151248932, "learning_rate": 9.611149625408167e-06, "loss": 1.1499, "loss_dynamic": 1.2982, "loss_fixed": 1.0017, "step": 102 }, { "eob_count_fixed": 22927, "eob_loss_fixed": 0.2087, "epoch": 0.15591296121097445, "grad_norm": 0.2166757434606552, "learning_rate": 9.601619466434113e-06, "loss": 1.1572, "loss_dynamic": 1.3352, "loss_fixed": 0.9791, "step": 103 }, { "eob_count_fixed": 23049, "eob_loss_fixed": 0.1991, "epoch": 0.15742667928098392, "grad_norm": 0.24211056530475616, "learning_rate": 9.59197877406273e-06, "loss": 1.1532, "loss_dynamic": 1.3344, "loss_fixed": 0.9721, "step": 104 }, { "eob_count_fixed": 23185, "eob_loss_fixed": 0.2161, "epoch": 0.15894039735099338, "grad_norm": 0.23450855910778046, "learning_rate": 9.582227779868683e-06, "loss": 1.1501, "loss_dynamic": 1.3249, "loss_fixed": 0.9752, "step": 105 }, { "eob_count_fixed": 23092, "eob_loss_fixed": 0.2124, "epoch": 0.16045411542100285, "grad_norm": 0.2219955325126648, "learning_rate": 9.572366718076142e-06, "loss": 1.1006, "loss_dynamic": 1.255, "loss_fixed": 0.9463, "step": 106 }, { "eob_count_fixed": 23000, "eob_loss_fixed": 0.202, "epoch": 0.1619678334910123, "grad_norm": 0.2090904712677002, "learning_rate": 9.562395825553167e-06, "loss": 1.1412, "loss_dynamic": 1.3048, "loss_fixed": 0.9776, "step": 107 }, { "eob_count_fixed": 23092, "eob_loss_fixed": 0.223, "epoch": 0.16348155156102176, "grad_norm": 0.2288423478603363, "learning_rate": 9.552315341806008e-06, "loss": 1.2093, "loss_dynamic": 1.4011, "loss_fixed": 1.0175, "step": 108 }, { "eob_count_fixed": 22985, "eob_loss_fixed": 0.2025, "epoch": 0.16499526963103123, "grad_norm": 0.2138763964176178, "learning_rate": 9.542125508973356e-06, "loss": 1.163, "loss_dynamic": 1.3438, "loss_fixed": 0.9822, "step": 109 }, { "eob_count_fixed": 23179, "eob_loss_fixed": 0.2001, "epoch": 0.1665089877010407, "grad_norm": 0.22379779815673828, "learning_rate": 9.531826571820527e-06, "loss": 1.1817, "loss_dynamic": 1.3755, "loss_fixed": 0.9878, "step": 110 }, { "eob_count_fixed": 23150, "eob_loss_fixed": 0.2056, "epoch": 0.16802270577105013, "grad_norm": 0.2135486751794815, "learning_rate": 9.52141877773358e-06, "loss": 1.1452, "loss_dynamic": 1.3113, "loss_fixed": 0.9791, "step": 111 }, { "eob_count_fixed": 23174, "eob_loss_fixed": 0.2057, "epoch": 0.1695364238410596, "grad_norm": 0.2012404203414917, "learning_rate": 9.510902376713378e-06, "loss": 1.0864, "loss_dynamic": 1.2158, "loss_fixed": 0.9571, "step": 112 }, { "eob_count_fixed": 22990, "eob_loss_fixed": 0.2092, "epoch": 0.17105014191106907, "grad_norm": 0.20197035372257233, "learning_rate": 9.500277621369585e-06, "loss": 1.1407, "loss_dynamic": 1.2947, "loss_fixed": 0.9867, "step": 113 }, { "eob_count_fixed": 23192, "eob_loss_fixed": 0.199, "epoch": 0.17256385998107854, "grad_norm": 0.22289662063121796, "learning_rate": 9.489544766914591e-06, "loss": 1.1847, "loss_dynamic": 1.4055, "loss_fixed": 0.964, "step": 114 }, { "eob_count_fixed": 23202, "eob_loss_fixed": 0.2076, "epoch": 0.17407757805108798, "grad_norm": 0.2101375162601471, "learning_rate": 9.478704071157382e-06, "loss": 1.1615, "loss_dynamic": 1.335, "loss_fixed": 0.9881, "step": 115 }, { "eob_count_fixed": 23143, "eob_loss_fixed": 0.2169, "epoch": 0.17559129612109745, "grad_norm": 0.21617832779884338, "learning_rate": 9.467755794497358e-06, "loss": 1.172, "loss_dynamic": 1.3568, "loss_fixed": 0.9871, "step": 116 }, { "eob_count_fixed": 23045, "eob_loss_fixed": 0.2167, "epoch": 0.1771050141911069, "grad_norm": 0.21501454710960388, "learning_rate": 9.456700199918065e-06, "loss": 1.1217, "loss_dynamic": 1.2435, "loss_fixed": 0.9999, "step": 117 }, { "eob_count_fixed": 23004, "eob_loss_fixed": 0.1987, "epoch": 0.17861873226111638, "grad_norm": 0.23856137692928314, "learning_rate": 9.445537552980888e-06, "loss": 1.185, "loss_dynamic": 1.4087, "loss_fixed": 0.9613, "step": 118 }, { "eob_count_fixed": 23086, "eob_loss_fixed": 0.1936, "epoch": 0.18013245033112582, "grad_norm": 0.20731163024902344, "learning_rate": 9.434268121818664e-06, "loss": 1.167, "loss_dynamic": 1.3655, "loss_fixed": 0.9685, "step": 119 }, { "eob_count_fixed": 23062, "eob_loss_fixed": 0.2041, "epoch": 0.1816461684011353, "grad_norm": 0.204367995262146, "learning_rate": 9.422892177129245e-06, "loss": 1.1538, "loss_dynamic": 1.3456, "loss_fixed": 0.962, "step": 120 }, { "eob_count_fixed": 22904, "eob_loss_fixed": 0.2167, "epoch": 0.18315988647114476, "grad_norm": 0.22279883921146393, "learning_rate": 9.411409992169001e-06, "loss": 1.1867, "loss_dynamic": 1.3809, "loss_fixed": 0.9925, "step": 121 }, { "eob_count_fixed": 23147, "eob_loss_fixed": 0.2168, "epoch": 0.18467360454115422, "grad_norm": 0.20301565527915955, "learning_rate": 9.399821842746242e-06, "loss": 1.0958, "loss_dynamic": 1.2278, "loss_fixed": 0.9637, "step": 122 }, { "eob_count_fixed": 23141, "eob_loss_fixed": 0.2056, "epoch": 0.18618732261116366, "grad_norm": 0.21497398614883423, "learning_rate": 9.388128007214614e-06, "loss": 1.1672, "loss_dynamic": 1.3601, "loss_fixed": 0.9744, "step": 123 }, { "eob_count_fixed": 23089, "eob_loss_fixed": 0.2039, "epoch": 0.18770104068117313, "grad_norm": 0.20583945512771606, "learning_rate": 9.376328766466387e-06, "loss": 1.137, "loss_dynamic": 1.2816, "loss_fixed": 0.9925, "step": 124 }, { "eob_count_fixed": 23167, "eob_loss_fixed": 0.2029, "epoch": 0.1892147587511826, "grad_norm": 0.20904339849948883, "learning_rate": 9.364424403925734e-06, "loss": 1.1542, "loss_dynamic": 1.321, "loss_fixed": 0.9874, "step": 125 }, { "eob_count_fixed": 23189, "eob_loss_fixed": 0.2022, "epoch": 0.19072847682119207, "grad_norm": 0.216285839676857, "learning_rate": 9.3524152055419e-06, "loss": 1.1175, "loss_dynamic": 1.2817, "loss_fixed": 0.9532, "step": 126 }, { "eob_count_fixed": 23108, "eob_loss_fixed": 0.2, "epoch": 0.1922421948912015, "grad_norm": 0.21317483484745026, "learning_rate": 9.340301459782352e-06, "loss": 1.1287, "loss_dynamic": 1.3073, "loss_fixed": 0.9501, "step": 127 }, { "eob_count_fixed": 23088, "eob_loss_fixed": 0.1913, "epoch": 0.19375591296121097, "grad_norm": 0.23464760184288025, "learning_rate": 9.328083457625838e-06, "loss": 1.1109, "loss_dynamic": 1.2766, "loss_fixed": 0.9451, "step": 128 }, { "eob_count_fixed": 23216, "eob_loss_fixed": 0.2075, "epoch": 0.19526963103122044, "grad_norm": 0.2228466123342514, "learning_rate": 9.315761492555402e-06, "loss": 1.1163, "loss_dynamic": 1.2621, "loss_fixed": 0.9705, "step": 129 }, { "eob_count_fixed": 23066, "eob_loss_fixed": 0.1988, "epoch": 0.1967833491012299, "grad_norm": 0.2207752764225006, "learning_rate": 9.303335860551332e-06, "loss": 1.1362, "loss_dynamic": 1.2743, "loss_fixed": 0.998, "step": 130 }, { "eob_count_fixed": 23152, "eob_loss_fixed": 0.2022, "epoch": 0.19829706717123935, "grad_norm": 0.23500651121139526, "learning_rate": 9.290806860084054e-06, "loss": 1.1288, "loss_dynamic": 1.2828, "loss_fixed": 0.9748, "step": 131 }, { "eob_count_fixed": 23254, "eob_loss_fixed": 0.2042, "epoch": 0.19981078524124882, "grad_norm": 0.20553387701511383, "learning_rate": 9.27817479210696e-06, "loss": 1.1509, "loss_dynamic": 1.33, "loss_fixed": 0.9718, "step": 132 }, { "eob_count_fixed": 23198, "eob_loss_fixed": 0.2006, "epoch": 0.20132450331125828, "grad_norm": 0.2401861995458603, "learning_rate": 9.265439960049178e-06, "loss": 1.1475, "loss_dynamic": 1.321, "loss_fixed": 0.974, "step": 133 }, { "eob_count_fixed": 23218, "eob_loss_fixed": 0.1889, "epoch": 0.20283822138126775, "grad_norm": 0.22657200694084167, "learning_rate": 9.252602669808288e-06, "loss": 1.1665, "loss_dynamic": 1.3727, "loss_fixed": 0.9603, "step": 134 }, { "eob_count_fixed": 23157, "eob_loss_fixed": 0.2109, "epoch": 0.2043519394512772, "grad_norm": 0.2224370241165161, "learning_rate": 9.23966322974297e-06, "loss": 1.1675, "loss_dynamic": 1.3398, "loss_fixed": 0.9952, "step": 135 }, { "eob_count_fixed": 23117, "eob_loss_fixed": 0.2023, "epoch": 0.20586565752128666, "grad_norm": 0.2282455712556839, "learning_rate": 9.226621950665597e-06, "loss": 1.1806, "loss_dynamic": 1.3849, "loss_fixed": 0.9762, "step": 136 }, { "eob_count_fixed": 23252, "eob_loss_fixed": 0.1994, "epoch": 0.20737937559129613, "grad_norm": 0.2160065621137619, "learning_rate": 9.213479145834772e-06, "loss": 1.0947, "loss_dynamic": 1.2454, "loss_fixed": 0.9439, "step": 137 }, { "eob_count_fixed": 23098, "eob_loss_fixed": 0.1989, "epoch": 0.2088930936613056, "grad_norm": 0.24084223806858063, "learning_rate": 9.2002351309478e-06, "loss": 1.128, "loss_dynamic": 1.3116, "loss_fixed": 0.9445, "step": 138 }, { "eob_count_fixed": 23201, "eob_loss_fixed": 0.1959, "epoch": 0.21040681173131504, "grad_norm": 0.3957988917827606, "learning_rate": 9.186890224133106e-06, "loss": 1.1876, "loss_dynamic": 1.3982, "loss_fixed": 0.977, "step": 139 }, { "eob_count_fixed": 23074, "eob_loss_fixed": 0.193, "epoch": 0.2119205298013245, "grad_norm": 0.21406769752502441, "learning_rate": 9.1734447459426e-06, "loss": 1.0897, "loss_dynamic": 1.2278, "loss_fixed": 0.9517, "step": 140 }, { "eob_count_fixed": 23043, "eob_loss_fixed": 0.2083, "epoch": 0.21343424787133397, "grad_norm": 0.2276356965303421, "learning_rate": 9.159899019343962e-06, "loss": 1.1705, "loss_dynamic": 1.38, "loss_fixed": 0.9611, "step": 141 }, { "eob_count_fixed": 23111, "eob_loss_fixed": 0.2009, "epoch": 0.21494796594134344, "grad_norm": 0.2496977150440216, "learning_rate": 9.146253369712899e-06, "loss": 1.1897, "loss_dynamic": 1.4253, "loss_fixed": 0.954, "step": 142 }, { "eob_count_fixed": 22953, "eob_loss_fixed": 0.2129, "epoch": 0.21646168401135288, "grad_norm": 0.23140805959701538, "learning_rate": 9.13250812482532e-06, "loss": 1.1079, "loss_dynamic": 1.2342, "loss_fixed": 0.9817, "step": 143 }, { "eob_count_fixed": 23276, "eob_loss_fixed": 0.2019, "epoch": 0.21797540208136235, "grad_norm": 0.2172962725162506, "learning_rate": 9.118663614849472e-06, "loss": 1.0987, "loss_dynamic": 1.2437, "loss_fixed": 0.9537, "step": 144 }, { "eob_count_fixed": 23115, "eob_loss_fixed": 0.1997, "epoch": 0.2194891201513718, "grad_norm": 0.2137148082256317, "learning_rate": 9.104720172337997e-06, "loss": 1.1344, "loss_dynamic": 1.2979, "loss_fixed": 0.971, "step": 145 }, { "eob_count_fixed": 23181, "eob_loss_fixed": 0.1957, "epoch": 0.22100283822138128, "grad_norm": 0.21047954261302948, "learning_rate": 9.090678132219952e-06, "loss": 1.1341, "loss_dynamic": 1.3006, "loss_fixed": 0.9676, "step": 146 }, { "eob_count_fixed": 23257, "eob_loss_fixed": 0.1982, "epoch": 0.22251655629139072, "grad_norm": 0.21609677374362946, "learning_rate": 9.076537831792764e-06, "loss": 1.188, "loss_dynamic": 1.414, "loss_fixed": 0.962, "step": 147 }, { "eob_count_fixed": 23127, "eob_loss_fixed": 0.2025, "epoch": 0.2240302743614002, "grad_norm": 0.22972749173641205, "learning_rate": 9.062299610714123e-06, "loss": 1.1634, "loss_dynamic": 1.3564, "loss_fixed": 0.9705, "step": 148 }, { "eob_count_fixed": 23124, "eob_loss_fixed": 0.2155, "epoch": 0.22554399243140966, "grad_norm": 0.22412319481372833, "learning_rate": 9.047963810993829e-06, "loss": 1.1827, "loss_dynamic": 1.3917, "loss_fixed": 0.9738, "step": 149 }, { "eob_count_fixed": 23011, "eob_loss_fixed": 0.1937, "epoch": 0.22705771050141912, "grad_norm": 0.22135809063911438, "learning_rate": 9.033530776985573e-06, "loss": 1.0926, "loss_dynamic": 1.2547, "loss_fixed": 0.9306, "step": 150 }, { "eob_count_fixed": 23256, "eob_loss_fixed": 0.183, "epoch": 0.22857142857142856, "grad_norm": 0.23625916242599487, "learning_rate": 9.019000855378666e-06, "loss": 1.1322, "loss_dynamic": 1.3078, "loss_fixed": 0.9567, "step": 151 }, { "eob_count_fixed": 23060, "eob_loss_fixed": 0.2063, "epoch": 0.23008514664143803, "grad_norm": 0.2262950837612152, "learning_rate": 9.00437439518971e-06, "loss": 1.1824, "loss_dynamic": 1.3795, "loss_fixed": 0.9854, "step": 152 }, { "eob_count_fixed": 23189, "eob_loss_fixed": 0.2052, "epoch": 0.2315988647114475, "grad_norm": 0.21782366931438446, "learning_rate": 8.98965174775422e-06, "loss": 1.1437, "loss_dynamic": 1.3066, "loss_fixed": 0.9808, "step": 153 }, { "eob_count_fixed": 23094, "eob_loss_fixed": 0.2025, "epoch": 0.23311258278145697, "grad_norm": 0.2170797884464264, "learning_rate": 8.974833266718179e-06, "loss": 1.1491, "loss_dynamic": 1.3267, "loss_fixed": 0.9716, "step": 154 }, { "eob_count_fixed": 23205, "eob_loss_fixed": 0.196, "epoch": 0.2346263008514664, "grad_norm": 0.21666193008422852, "learning_rate": 8.959919308029543e-06, "loss": 1.1131, "loss_dynamic": 1.2571, "loss_fixed": 0.9691, "step": 155 }, { "eob_count_fixed": 22993, "eob_loss_fixed": 0.1998, "epoch": 0.23614001892147587, "grad_norm": 0.24705326557159424, "learning_rate": 8.944910229929696e-06, "loss": 1.1621, "loss_dynamic": 1.3353, "loss_fixed": 0.989, "step": 156 }, { "eob_count_fixed": 22949, "eob_loss_fixed": 0.1821, "epoch": 0.23765373699148534, "grad_norm": 0.504496157169342, "learning_rate": 8.929806392944839e-06, "loss": 1.1496, "loss_dynamic": 1.353, "loss_fixed": 0.9463, "step": 157 }, { "eob_count_fixed": 23047, "eob_loss_fixed": 0.1969, "epoch": 0.2391674550614948, "grad_norm": 0.22727130353450775, "learning_rate": 8.914608159877338e-06, "loss": 1.1397, "loss_dynamic": 1.3385, "loss_fixed": 0.941, "step": 158 }, { "eob_count_fixed": 22978, "eob_loss_fixed": 0.2034, "epoch": 0.24068117313150425, "grad_norm": 0.20292510092258453, "learning_rate": 8.899315895797e-06, "loss": 1.1547, "loss_dynamic": 1.342, "loss_fixed": 0.9673, "step": 159 }, { "eob_count_fixed": 23240, "eob_loss_fixed": 0.1852, "epoch": 0.24219489120151372, "grad_norm": 0.2147001326084137, "learning_rate": 8.883929968032308e-06, "loss": 1.1387, "loss_dynamic": 1.3257, "loss_fixed": 0.9516, "step": 160 }, { "eob_count_fixed": 22994, "eob_loss_fixed": 0.1896, "epoch": 0.24370860927152319, "grad_norm": 0.2164391428232193, "learning_rate": 8.868450746161598e-06, "loss": 1.1435, "loss_dynamic": 1.3466, "loss_fixed": 0.9404, "step": 161 }, { "eob_count_fixed": 23084, "eob_loss_fixed": 0.2144, "epoch": 0.24522232734153265, "grad_norm": 0.24824491143226624, "learning_rate": 8.85287860200418e-06, "loss": 1.1798, "loss_dynamic": 1.3863, "loss_fixed": 0.9732, "step": 162 }, { "eob_count_fixed": 23182, "eob_loss_fixed": 0.2003, "epoch": 0.2467360454115421, "grad_norm": 0.25680825114250183, "learning_rate": 8.837213909611416e-06, "loss": 1.1786, "loss_dynamic": 1.3934, "loss_fixed": 0.9637, "step": 163 }, { "eob_count_fixed": 23172, "eob_loss_fixed": 0.1963, "epoch": 0.24824976348155156, "grad_norm": 0.2670901119709015, "learning_rate": 8.821457045257712e-06, "loss": 1.1655, "loss_dynamic": 1.3781, "loss_fixed": 0.953, "step": 164 }, { "eob_count_fixed": 23217, "eob_loss_fixed": 0.1942, "epoch": 0.24976348155156103, "grad_norm": 0.3082368075847626, "learning_rate": 8.805608387431504e-06, "loss": 1.1559, "loss_dynamic": 1.3549, "loss_fixed": 0.9569, "step": 165 }, { "eob_count_fixed": 22925, "eob_loss_fixed": 0.1941, "epoch": 0.25127719962157047, "grad_norm": 0.23024457693099976, "learning_rate": 8.78966831682616e-06, "loss": 1.1482, "loss_dynamic": 1.3345, "loss_fixed": 0.9619, "step": 166 }, { "eob_count_fixed": 23244, "eob_loss_fixed": 0.1925, "epoch": 0.25279091769157996, "grad_norm": 0.2338479608297348, "learning_rate": 8.773637216330824e-06, "loss": 1.1457, "loss_dynamic": 1.3368, "loss_fixed": 0.9545, "step": 167 }, { "eob_count_fixed": 23166, "eob_loss_fixed": 0.1888, "epoch": 0.2543046357615894, "grad_norm": 0.21633832156658173, "learning_rate": 8.757515471021232e-06, "loss": 1.0932, "loss_dynamic": 1.2425, "loss_fixed": 0.944, "step": 168 }, { "eob_count_fixed": 23297, "eob_loss_fixed": 0.1958, "epoch": 0.25581835383159884, "grad_norm": 0.23612940311431885, "learning_rate": 8.74130346815046e-06, "loss": 1.1348, "loss_dynamic": 1.3139, "loss_fixed": 0.9556, "step": 169 }, { "eob_count_fixed": 23053, "eob_loss_fixed": 0.2054, "epoch": 0.25733207190160834, "grad_norm": 0.21078245341777802, "learning_rate": 8.725001597139613e-06, "loss": 1.1117, "loss_dynamic": 1.2598, "loss_fixed": 0.9636, "step": 170 }, { "eob_count_fixed": 23095, "eob_loss_fixed": 0.1939, "epoch": 0.2588457899716178, "grad_norm": 0.2276008278131485, "learning_rate": 8.708610249568484e-06, "loss": 1.1498, "loss_dynamic": 1.3314, "loss_fixed": 0.9682, "step": 171 }, { "eob_count_fixed": 23048, "eob_loss_fixed": 0.1927, "epoch": 0.2603595080416273, "grad_norm": 0.21225674450397491, "learning_rate": 8.692129819166138e-06, "loss": 1.1394, "loss_dynamic": 1.333, "loss_fixed": 0.9458, "step": 172 }, { "eob_count_fixed": 23207, "eob_loss_fixed": 0.1956, "epoch": 0.2618732261116367, "grad_norm": 0.22824546694755554, "learning_rate": 8.675560701801458e-06, "loss": 1.1637, "loss_dynamic": 1.3518, "loss_fixed": 0.9755, "step": 173 }, { "eob_count_fixed": 23196, "eob_loss_fixed": 0.1939, "epoch": 0.26338694418164615, "grad_norm": 0.2226727157831192, "learning_rate": 8.65890329547364e-06, "loss": 1.1859, "loss_dynamic": 1.4157, "loss_fixed": 0.956, "step": 174 }, { "eob_count_fixed": 23032, "eob_loss_fixed": 0.1932, "epoch": 0.26490066225165565, "grad_norm": 0.2501350939273834, "learning_rate": 8.642158000302621e-06, "loss": 1.1589, "loss_dynamic": 1.3391, "loss_fixed": 0.9788, "step": 175 }, { "eob_count_fixed": 23258, "eob_loss_fixed": 0.2047, "epoch": 0.2664143803216651, "grad_norm": 0.2134886085987091, "learning_rate": 8.625325218519483e-06, "loss": 1.1292, "loss_dynamic": 1.3193, "loss_fixed": 0.9391, "step": 176 }, { "eob_count_fixed": 23125, "eob_loss_fixed": 0.1949, "epoch": 0.26792809839167453, "grad_norm": 0.2220613956451416, "learning_rate": 8.608405354456779e-06, "loss": 1.092, "loss_dynamic": 1.2445, "loss_fixed": 0.9394, "step": 177 }, { "eob_count_fixed": 23085, "eob_loss_fixed": 0.1936, "epoch": 0.269441816461684, "grad_norm": 0.20297449827194214, "learning_rate": 8.591398814538832e-06, "loss": 1.1084, "loss_dynamic": 1.2708, "loss_fixed": 0.946, "step": 178 }, { "eob_count_fixed": 23124, "eob_loss_fixed": 0.1952, "epoch": 0.27095553453169346, "grad_norm": 0.21689414978027344, "learning_rate": 8.574306007271957e-06, "loss": 1.1964, "loss_dynamic": 1.4258, "loss_fixed": 0.9669, "step": 179 }, { "eob_count_fixed": 23110, "eob_loss_fixed": 0.2005, "epoch": 0.27246925260170296, "grad_norm": 0.22509916126728058, "learning_rate": 8.557127343234667e-06, "loss": 1.1584, "loss_dynamic": 1.3503, "loss_fixed": 0.9664, "step": 180 }, { "eob_count_fixed": 23168, "eob_loss_fixed": 0.2044, "epoch": 0.2739829706717124, "grad_norm": 0.2288724035024643, "learning_rate": 8.539863235067794e-06, "loss": 1.158, "loss_dynamic": 1.3539, "loss_fixed": 0.9621, "step": 181 }, { "eob_count_fixed": 23123, "eob_loss_fixed": 0.1986, "epoch": 0.27549668874172184, "grad_norm": 0.22012163698673248, "learning_rate": 8.522514097464586e-06, "loss": 1.1302, "loss_dynamic": 1.3156, "loss_fixed": 0.9449, "step": 182 }, { "eob_count_fixed": 23102, "eob_loss_fixed": 0.1953, "epoch": 0.27701040681173134, "grad_norm": 0.22424371540546417, "learning_rate": 8.50508034716075e-06, "loss": 1.1863, "loss_dynamic": 1.4061, "loss_fixed": 0.9666, "step": 183 }, { "eob_count_fixed": 23239, "eob_loss_fixed": 0.1924, "epoch": 0.2785241248817408, "grad_norm": 0.2283855527639389, "learning_rate": 8.487562402924426e-06, "loss": 1.1838, "loss_dynamic": 1.3834, "loss_fixed": 0.9842, "step": 184 }, { "eob_count_fixed": 23038, "eob_loss_fixed": 0.1832, "epoch": 0.2800378429517502, "grad_norm": 0.2160593569278717, "learning_rate": 8.469960685546145e-06, "loss": 1.1007, "loss_dynamic": 1.2561, "loss_fixed": 0.9454, "step": 185 }, { "eob_count_fixed": 23203, "eob_loss_fixed": 0.2061, "epoch": 0.2815515610217597, "grad_norm": 0.20844058692455292, "learning_rate": 8.452275617828715e-06, "loss": 1.1531, "loss_dynamic": 1.3574, "loss_fixed": 0.9488, "step": 186 }, { "eob_count_fixed": 23117, "eob_loss_fixed": 0.2033, "epoch": 0.28306527909176915, "grad_norm": 0.2453841269016266, "learning_rate": 8.434507624577058e-06, "loss": 1.2097, "loss_dynamic": 1.4647, "loss_fixed": 0.9547, "step": 187 }, { "eob_count_fixed": 23125, "eob_loss_fixed": 0.1985, "epoch": 0.28457899716177865, "grad_norm": 0.22973443567752838, "learning_rate": 8.41665713258802e-06, "loss": 1.1309, "loss_dynamic": 1.3159, "loss_fixed": 0.9459, "step": 188 }, { "eob_count_fixed": 23194, "eob_loss_fixed": 0.1912, "epoch": 0.2860927152317881, "grad_norm": 0.21650952100753784, "learning_rate": 8.398724570640107e-06, "loss": 1.1367, "loss_dynamic": 1.3173, "loss_fixed": 0.9561, "step": 189 }, { "eob_count_fixed": 23145, "eob_loss_fixed": 0.1896, "epoch": 0.2876064333017975, "grad_norm": 0.20151762664318085, "learning_rate": 8.380710369483186e-06, "loss": 1.1792, "loss_dynamic": 1.3765, "loss_fixed": 0.9819, "step": 190 }, { "eob_count_fixed": 23155, "eob_loss_fixed": 0.1998, "epoch": 0.289120151371807, "grad_norm": 0.2122228443622589, "learning_rate": 8.362614961828153e-06, "loss": 1.1505, "loss_dynamic": 1.3303, "loss_fixed": 0.9708, "step": 191 }, { "eob_count_fixed": 23042, "eob_loss_fixed": 0.1981, "epoch": 0.29063386944181646, "grad_norm": 0.22585619986057281, "learning_rate": 8.34443878233652e-06, "loss": 1.1054, "loss_dynamic": 1.2746, "loss_fixed": 0.9362, "step": 192 }, { "eob_count_fixed": 23042, "eob_loss_fixed": 0.1932, "epoch": 0.2921475875118259, "grad_norm": 0.20969459414482117, "learning_rate": 8.326182267609984e-06, "loss": 1.1388, "loss_dynamic": 1.3217, "loss_fixed": 0.9558, "step": 193 }, { "eob_count_fixed": 22972, "eob_loss_fixed": 0.1941, "epoch": 0.2936613055818354, "grad_norm": 0.2062511295080185, "learning_rate": 8.30784585617994e-06, "loss": 1.1361, "loss_dynamic": 1.3189, "loss_fixed": 0.9533, "step": 194 }, { "eob_count_fixed": 23223, "eob_loss_fixed": 0.1874, "epoch": 0.29517502365184484, "grad_norm": 0.22198927402496338, "learning_rate": 8.289429988496938e-06, "loss": 1.1122, "loss_dynamic": 1.2865, "loss_fixed": 0.938, "step": 195 }, { "eob_count_fixed": 23224, "eob_loss_fixed": 0.195, "epoch": 0.29668874172185433, "grad_norm": 0.20665717124938965, "learning_rate": 8.270935106920122e-06, "loss": 1.1678, "loss_dynamic": 1.3679, "loss_fixed": 0.9676, "step": 196 }, { "eob_count_fixed": 23095, "eob_loss_fixed": 0.1838, "epoch": 0.29820245979186377, "grad_norm": 0.22092752158641815, "learning_rate": 8.252361655706584e-06, "loss": 1.1529, "loss_dynamic": 1.3607, "loss_fixed": 0.945, "step": 197 }, { "eob_count_fixed": 23090, "eob_loss_fixed": 0.2045, "epoch": 0.2997161778618732, "grad_norm": 0.2128957062959671, "learning_rate": 8.233710081000703e-06, "loss": 1.0913, "loss_dynamic": 1.2322, "loss_fixed": 0.9504, "step": 198 }, { "eob_count_fixed": 23177, "eob_loss_fixed": 0.2115, "epoch": 0.3012298959318827, "grad_norm": 0.21134507656097412, "learning_rate": 8.214980830823429e-06, "loss": 1.1267, "loss_dynamic": 1.2823, "loss_fixed": 0.9711, "step": 199 }, { "eob_count_fixed": 23119, "eob_loss_fixed": 0.1899, "epoch": 0.30274361400189215, "grad_norm": 0.20937271416187286, "learning_rate": 8.196174355061517e-06, "loss": 1.138, "loss_dynamic": 1.3312, "loss_fixed": 0.9448, "step": 200 }, { "eob_count_fixed": 23077, "eob_loss_fixed": 0.1841, "epoch": 0.3042573320719016, "grad_norm": 0.22359099984169006, "learning_rate": 8.177291105456723e-06, "loss": 1.1546, "loss_dynamic": 1.3721, "loss_fixed": 0.9371, "step": 201 }, { "eob_count_fixed": 23019, "eob_loss_fixed": 0.1787, "epoch": 0.3057710501419111, "grad_norm": 0.22193999588489532, "learning_rate": 8.15833153559495e-06, "loss": 1.1548, "loss_dynamic": 1.3536, "loss_fixed": 0.956, "step": 202 }, { "eob_count_fixed": 23085, "eob_loss_fixed": 0.1836, "epoch": 0.3072847682119205, "grad_norm": 0.22661146521568298, "learning_rate": 8.13929610089536e-06, "loss": 1.1367, "loss_dynamic": 1.3306, "loss_fixed": 0.9429, "step": 203 }, { "eob_count_fixed": 23044, "eob_loss_fixed": 0.2049, "epoch": 0.30879848628193, "grad_norm": 0.1984066367149353, "learning_rate": 8.120185258599428e-06, "loss": 1.1141, "loss_dynamic": 1.2815, "loss_fixed": 0.9467, "step": 204 }, { "eob_count_fixed": 23264, "eob_loss_fixed": 0.197, "epoch": 0.31031220435193946, "grad_norm": 0.21948891878128052, "learning_rate": 8.100999467759955e-06, "loss": 1.1758, "loss_dynamic": 1.3733, "loss_fixed": 0.9784, "step": 205 }, { "eob_count_fixed": 23185, "eob_loss_fixed": 0.1938, "epoch": 0.3118259224219489, "grad_norm": 0.23247477412223816, "learning_rate": 8.081739189230055e-06, "loss": 1.1347, "loss_dynamic": 1.3183, "loss_fixed": 0.951, "step": 206 }, { "eob_count_fixed": 23258, "eob_loss_fixed": 0.184, "epoch": 0.3133396404919584, "grad_norm": 0.19910964369773865, "learning_rate": 8.062404885652071e-06, "loss": 1.0971, "loss_dynamic": 1.2782, "loss_fixed": 0.916, "step": 207 }, { "eob_count_fixed": 23075, "eob_loss_fixed": 0.1971, "epoch": 0.31485335856196783, "grad_norm": 0.23165415227413177, "learning_rate": 8.042997021446469e-06, "loss": 1.1553, "loss_dynamic": 1.3378, "loss_fixed": 0.9728, "step": 208 }, { "eob_count_fixed": 23027, "eob_loss_fixed": 0.2102, "epoch": 0.3163670766319773, "grad_norm": 0.22059288620948792, "learning_rate": 8.023516062800682e-06, "loss": 1.1802, "loss_dynamic": 1.3799, "loss_fixed": 0.9804, "step": 209 }, { "eob_count_fixed": 23368, "eob_loss_fixed": 0.1935, "epoch": 0.31788079470198677, "grad_norm": 0.21328063309192657, "learning_rate": 8.003962477657911e-06, "loss": 1.1173, "loss_dynamic": 1.3033, "loss_fixed": 0.9314, "step": 210 }, { "eob_count_fixed": 23126, "eob_loss_fixed": 0.2072, "epoch": 0.3193945127719962, "grad_norm": 0.23099765181541443, "learning_rate": 7.984336735705885e-06, "loss": 1.1416, "loss_dynamic": 1.3172, "loss_fixed": 0.966, "step": 211 }, { "eob_count_fixed": 23084, "eob_loss_fixed": 0.1865, "epoch": 0.3209082308420057, "grad_norm": 0.20524200797080994, "learning_rate": 7.964639308365578e-06, "loss": 1.1092, "loss_dynamic": 1.2755, "loss_fixed": 0.9429, "step": 212 }, { "eob_count_fixed": 23149, "eob_loss_fixed": 0.1839, "epoch": 0.32242194891201514, "grad_norm": 0.2050389051437378, "learning_rate": 7.944870668779887e-06, "loss": 1.1481, "loss_dynamic": 1.3361, "loss_fixed": 0.9602, "step": 213 }, { "eob_count_fixed": 23148, "eob_loss_fixed": 0.1863, "epoch": 0.3239356669820246, "grad_norm": 0.2407693862915039, "learning_rate": 7.925031291802267e-06, "loss": 1.1289, "loss_dynamic": 1.3364, "loss_fixed": 0.9214, "step": 214 }, { "eob_count_fixed": 23096, "eob_loss_fixed": 0.202, "epoch": 0.3254493850520341, "grad_norm": 0.22053195536136627, "learning_rate": 7.905121653985317e-06, "loss": 1.1531, "loss_dynamic": 1.337, "loss_fixed": 0.9692, "step": 215 }, { "eob_count_fixed": 23134, "eob_loss_fixed": 0.1984, "epoch": 0.3269631031220435, "grad_norm": 0.23012273013591766, "learning_rate": 7.885142233569349e-06, "loss": 1.1019, "loss_dynamic": 1.245, "loss_fixed": 0.9587, "step": 216 }, { "eob_count_fixed": 23196, "eob_loss_fixed": 0.1943, "epoch": 0.32847682119205296, "grad_norm": 0.2115291804075241, "learning_rate": 7.865093510470885e-06, "loss": 1.1452, "loss_dynamic": 1.3571, "loss_fixed": 0.9332, "step": 217 }, { "eob_count_fixed": 23033, "eob_loss_fixed": 0.1889, "epoch": 0.32999053926206245, "grad_norm": 0.212088942527771, "learning_rate": 7.844975966271134e-06, "loss": 1.1518, "loss_dynamic": 1.3654, "loss_fixed": 0.9381, "step": 218 }, { "eob_count_fixed": 23185, "eob_loss_fixed": 0.1805, "epoch": 0.3315042573320719, "grad_norm": 0.23262086510658264, "learning_rate": 7.824790084204427e-06, "loss": 1.1415, "loss_dynamic": 1.3554, "loss_fixed": 0.9276, "step": 219 }, { "eob_count_fixed": 23196, "eob_loss_fixed": 0.1894, "epoch": 0.3330179754020814, "grad_norm": 0.2145562469959259, "learning_rate": 7.804536349146611e-06, "loss": 1.1558, "loss_dynamic": 1.3667, "loss_fixed": 0.9448, "step": 220 }, { "eob_count_fixed": 23101, "eob_loss_fixed": 0.1874, "epoch": 0.33453169347209083, "grad_norm": 0.21004539728164673, "learning_rate": 7.784215247603397e-06, "loss": 1.1183, "loss_dynamic": 1.3134, "loss_fixed": 0.9233, "step": 221 }, { "eob_count_fixed": 23254, "eob_loss_fixed": 0.1906, "epoch": 0.33604541154210027, "grad_norm": 0.2244272083044052, "learning_rate": 7.76382726769867e-06, "loss": 1.1755, "loss_dynamic": 1.3948, "loss_fixed": 0.9563, "step": 222 }, { "eob_count_fixed": 23147, "eob_loss_fixed": 0.1853, "epoch": 0.33755912961210977, "grad_norm": 0.21895021200180054, "learning_rate": 7.743372899162776e-06, "loss": 1.1704, "loss_dynamic": 1.3897, "loss_fixed": 0.9511, "step": 223 }, { "eob_count_fixed": 23223, "eob_loss_fixed": 0.1874, "epoch": 0.3390728476821192, "grad_norm": 0.21501126885414124, "learning_rate": 7.72285263332075e-06, "loss": 1.1463, "loss_dynamic": 1.351, "loss_fixed": 0.9417, "step": 224 }, { "eob_count_fixed": 22969, "eob_loss_fixed": 0.1865, "epoch": 0.34058656575212864, "grad_norm": 0.250153124332428, "learning_rate": 7.702266963080516e-06, "loss": 1.1655, "loss_dynamic": 1.3778, "loss_fixed": 0.9531, "step": 225 }, { "eob_count_fixed": 23211, "eob_loss_fixed": 0.1953, "epoch": 0.34210028382213814, "grad_norm": 0.20466573536396027, "learning_rate": 7.681616382921044e-06, "loss": 1.1194, "loss_dynamic": 1.3089, "loss_fixed": 0.9298, "step": 226 }, { "eob_count_fixed": 23266, "eob_loss_fixed": 0.2013, "epoch": 0.3436140018921476, "grad_norm": 0.22672031819820404, "learning_rate": 7.660901388880484e-06, "loss": 1.1585, "loss_dynamic": 1.3719, "loss_fixed": 0.9452, "step": 227 }, { "eob_count_fixed": 23137, "eob_loss_fixed": 0.1832, "epoch": 0.3451277199621571, "grad_norm": 0.2075030356645584, "learning_rate": 7.640122478544232e-06, "loss": 1.171, "loss_dynamic": 1.3843, "loss_fixed": 0.9576, "step": 228 }, { "eob_count_fixed": 23007, "eob_loss_fixed": 0.1846, "epoch": 0.3466414380321665, "grad_norm": 0.2191740721464157, "learning_rate": 7.619280151032996e-06, "loss": 1.1751, "loss_dynamic": 1.4068, "loss_fixed": 0.9434, "step": 229 }, { "eob_count_fixed": 23227, "eob_loss_fixed": 0.1871, "epoch": 0.34815515610217596, "grad_norm": 0.21374446153640747, "learning_rate": 7.598374906990795e-06, "loss": 1.1577, "loss_dynamic": 1.3805, "loss_fixed": 0.9349, "step": 230 }, { "eob_count_fixed": 23242, "eob_loss_fixed": 0.1921, "epoch": 0.34966887417218545, "grad_norm": 0.2193359136581421, "learning_rate": 7.577407248572937e-06, "loss": 1.1354, "loss_dynamic": 1.3265, "loss_fixed": 0.9442, "step": 231 }, { "eob_count_fixed": 23165, "eob_loss_fixed": 0.1892, "epoch": 0.3511825922421949, "grad_norm": 0.2195143848657608, "learning_rate": 7.55637767943396e-06, "loss": 1.1292, "loss_dynamic": 1.331, "loss_fixed": 0.9273, "step": 232 }, { "eob_count_fixed": 23164, "eob_loss_fixed": 0.1804, "epoch": 0.35269631031220433, "grad_norm": 0.21375612914562225, "learning_rate": 7.535286704715529e-06, "loss": 1.1496, "loss_dynamic": 1.3755, "loss_fixed": 0.9238, "step": 233 }, { "eob_count_fixed": 23080, "eob_loss_fixed": 0.179, "epoch": 0.3542100283822138, "grad_norm": 0.2254623919725418, "learning_rate": 7.514134831034305e-06, "loss": 1.1506, "loss_dynamic": 1.3514, "loss_fixed": 0.9498, "step": 234 }, { "eob_count_fixed": 23066, "eob_loss_fixed": 0.1902, "epoch": 0.35572374645222327, "grad_norm": 0.21597300469875336, "learning_rate": 7.4929225664697736e-06, "loss": 1.1353, "loss_dynamic": 1.3384, "loss_fixed": 0.9322, "step": 235 }, { "eob_count_fixed": 23135, "eob_loss_fixed": 0.2003, "epoch": 0.35723746452223276, "grad_norm": 0.23177792131900787, "learning_rate": 7.471650420552047e-06, "loss": 1.1832, "loss_dynamic": 1.4142, "loss_fixed": 0.9522, "step": 236 }, { "eob_count_fixed": 23081, "eob_loss_fixed": 0.1847, "epoch": 0.3587511825922422, "grad_norm": 0.2122935801744461, "learning_rate": 7.450318904249612e-06, "loss": 1.1389, "loss_dynamic": 1.3347, "loss_fixed": 0.9432, "step": 237 }, { "eob_count_fixed": 23205, "eob_loss_fixed": 0.187, "epoch": 0.36026490066225164, "grad_norm": 0.24003006517887115, "learning_rate": 7.4289285299570725e-06, "loss": 1.1359, "loss_dynamic": 1.3252, "loss_fixed": 0.9467, "step": 238 }, { "eob_count_fixed": 23165, "eob_loss_fixed": 0.1744, "epoch": 0.36177861873226114, "grad_norm": 0.2316312938928604, "learning_rate": 7.407479811482829e-06, "loss": 1.1247, "loss_dynamic": 1.308, "loss_fixed": 0.9415, "step": 239 }, { "eob_count_fixed": 23163, "eob_loss_fixed": 0.1863, "epoch": 0.3632923368022706, "grad_norm": 0.21367256343364716, "learning_rate": 7.385973264036739e-06, "loss": 1.1236, "loss_dynamic": 1.3062, "loss_fixed": 0.941, "step": 240 }, { "eob_count_fixed": 23085, "eob_loss_fixed": 0.2087, "epoch": 0.36480605487228, "grad_norm": 0.2245267927646637, "learning_rate": 7.364409404217749e-06, "loss": 1.1473, "loss_dynamic": 1.3507, "loss_fixed": 0.9438, "step": 241 }, { "eob_count_fixed": 23196, "eob_loss_fixed": 0.1942, "epoch": 0.3663197729422895, "grad_norm": 0.2000124305486679, "learning_rate": 7.342788750001475e-06, "loss": 1.1229, "loss_dynamic": 1.3114, "loss_fixed": 0.9344, "step": 242 }, { "eob_count_fixed": 23145, "eob_loss_fixed": 0.1857, "epoch": 0.36783349101229895, "grad_norm": 0.20588742196559906, "learning_rate": 7.3211118207277675e-06, "loss": 1.1025, "loss_dynamic": 1.2791, "loss_fixed": 0.9258, "step": 243 }, { "eob_count_fixed": 23129, "eob_loss_fixed": 0.1859, "epoch": 0.36934720908230845, "grad_norm": 0.20222258567810059, "learning_rate": 7.299379137088234e-06, "loss": 1.1476, "loss_dynamic": 1.3399, "loss_fixed": 0.9553, "step": 244 }, { "eob_count_fixed": 23118, "eob_loss_fixed": 0.1823, "epoch": 0.3708609271523179, "grad_norm": 0.22052869200706482, "learning_rate": 7.277591221113731e-06, "loss": 1.1716, "loss_dynamic": 1.3901, "loss_fixed": 0.9531, "step": 245 }, { "eob_count_fixed": 22946, "eob_loss_fixed": 0.1739, "epoch": 0.3723746452223273, "grad_norm": 0.2086203545331955, "learning_rate": 7.255748596161829e-06, "loss": 1.1425, "loss_dynamic": 1.3562, "loss_fixed": 0.9287, "step": 246 }, { "eob_count_fixed": 23157, "eob_loss_fixed": 0.1918, "epoch": 0.3738883632923368, "grad_norm": 0.21078285574913025, "learning_rate": 7.233851786904231e-06, "loss": 1.1016, "loss_dynamic": 1.2569, "loss_fixed": 0.9464, "step": 247 }, { "eob_count_fixed": 23217, "eob_loss_fixed": 0.1968, "epoch": 0.37540208136234626, "grad_norm": 0.2174227088689804, "learning_rate": 7.2119013193141854e-06, "loss": 1.1353, "loss_dynamic": 1.3377, "loss_fixed": 0.933, "step": 248 }, { "eob_count_fixed": 23057, "eob_loss_fixed": 0.1894, "epoch": 0.3769157994323557, "grad_norm": 0.21234743297100067, "learning_rate": 7.189897720653836e-06, "loss": 1.1475, "loss_dynamic": 1.3509, "loss_fixed": 0.9442, "step": 249 }, { "eob_count_fixed": 23104, "eob_loss_fixed": 0.1785, "epoch": 0.3784295175023652, "grad_norm": 0.21280284225940704, "learning_rate": 7.167841519461567e-06, "loss": 1.1572, "loss_dynamic": 1.3745, "loss_fixed": 0.9399, "step": 250 }, { "eob_count_fixed": 22969, "eob_loss_fixed": 0.1706, "epoch": 0.37994323557237464, "grad_norm": 0.2146778702735901, "learning_rate": 7.145733245539303e-06, "loss": 1.1169, "loss_dynamic": 1.3181, "loss_fixed": 0.9157, "step": 251 }, { "eob_count_fixed": 23238, "eob_loss_fixed": 0.1855, "epoch": 0.38145695364238413, "grad_norm": 0.21017496287822723, "learning_rate": 7.123573429939783e-06, "loss": 1.1331, "loss_dynamic": 1.3358, "loss_fixed": 0.9304, "step": 252 }, { "eob_count_fixed": 23035, "eob_loss_fixed": 0.1942, "epoch": 0.3829706717123936, "grad_norm": 0.20081499218940735, "learning_rate": 7.101362604953806e-06, "loss": 1.1034, "loss_dynamic": 1.2758, "loss_fixed": 0.931, "step": 253 }, { "eob_count_fixed": 23130, "eob_loss_fixed": 0.1878, "epoch": 0.384484389782403, "grad_norm": 0.2026153802871704, "learning_rate": 7.079101304097446e-06, "loss": 1.146, "loss_dynamic": 1.3636, "loss_fixed": 0.9284, "step": 254 }, { "eob_count_fixed": 23163, "eob_loss_fixed": 0.1827, "epoch": 0.3859981078524125, "grad_norm": 0.20794685184955597, "learning_rate": 7.056790062099231e-06, "loss": 1.1559, "loss_dynamic": 1.3946, "loss_fixed": 0.9172, "step": 255 }, { "eob_count_fixed": 23046, "eob_loss_fixed": 0.1834, "epoch": 0.38751182592242195, "grad_norm": 0.1961076259613037, "learning_rate": 7.034429414887305e-06, "loss": 1.0975, "loss_dynamic": 1.2479, "loss_fixed": 0.9471, "step": 256 }, { "eob_count_fixed": 23181, "eob_loss_fixed": 0.1837, "epoch": 0.3890255439924314, "grad_norm": 0.2136988341808319, "learning_rate": 7.012019899576552e-06, "loss": 1.1614, "loss_dynamic": 1.3766, "loss_fixed": 0.9462, "step": 257 }, { "eob_count_fixed": 23145, "eob_loss_fixed": 0.1801, "epoch": 0.3905392620624409, "grad_norm": 0.20374900102615356, "learning_rate": 6.989562054455692e-06, "loss": 1.108, "loss_dynamic": 1.2936, "loss_fixed": 0.9224, "step": 258 }, { "eob_count_fixed": 23150, "eob_loss_fixed": 0.181, "epoch": 0.3920529801324503, "grad_norm": 0.2196567803621292, "learning_rate": 6.967056418974356e-06, "loss": 1.1425, "loss_dynamic": 1.3671, "loss_fixed": 0.9179, "step": 259 }, { "eob_count_fixed": 23065, "eob_loss_fixed": 0.1853, "epoch": 0.3935666982024598, "grad_norm": 0.21198992431163788, "learning_rate": 6.944503533730125e-06, "loss": 1.1071, "loss_dynamic": 1.2935, "loss_fixed": 0.9208, "step": 260 }, { "eob_count_fixed": 23078, "eob_loss_fixed": 0.1882, "epoch": 0.39508041627246926, "grad_norm": 0.22205325961112976, "learning_rate": 6.921903940455541e-06, "loss": 1.1482, "loss_dynamic": 1.3589, "loss_fixed": 0.9374, "step": 261 }, { "eob_count_fixed": 22989, "eob_loss_fixed": 0.1877, "epoch": 0.3965941343424787, "grad_norm": 0.22633060812950134, "learning_rate": 6.899258182005106e-06, "loss": 1.1122, "loss_dynamic": 1.2861, "loss_fixed": 0.9382, "step": 262 }, { "eob_count_fixed": 23114, "eob_loss_fixed": 0.1963, "epoch": 0.3981078524124882, "grad_norm": 0.21773554384708405, "learning_rate": 6.876566802342226e-06, "loss": 1.1709, "loss_dynamic": 1.3951, "loss_fixed": 0.9468, "step": 263 }, { "eob_count_fixed": 23064, "eob_loss_fixed": 0.1855, "epoch": 0.39962157048249763, "grad_norm": 0.2154804766178131, "learning_rate": 6.853830346526159e-06, "loss": 1.1261, "loss_dynamic": 1.3143, "loss_fixed": 0.9378, "step": 264 }, { "eob_count_fixed": 23213, "eob_loss_fixed": 0.1745, "epoch": 0.4011352885525071, "grad_norm": 0.23099182546138763, "learning_rate": 6.8310493606989155e-06, "loss": 1.1344, "loss_dynamic": 1.3463, "loss_fixed": 0.9224, "step": 265 }, { "eob_count_fixed": 23149, "eob_loss_fixed": 0.1788, "epoch": 0.40264900662251657, "grad_norm": 0.22170846164226532, "learning_rate": 6.80822439207214e-06, "loss": 1.1342, "loss_dynamic": 1.3339, "loss_fixed": 0.9345, "step": 266 }, { "eob_count_fixed": 23039, "eob_loss_fixed": 0.1811, "epoch": 0.404162724692526, "grad_norm": 0.20232586562633514, "learning_rate": 6.7853559889139675e-06, "loss": 1.1102, "loss_dynamic": 1.2947, "loss_fixed": 0.9258, "step": 267 }, { "eob_count_fixed": 22947, "eob_loss_fixed": 0.188, "epoch": 0.4056764427625355, "grad_norm": 0.21055904030799866, "learning_rate": 6.7624447005358565e-06, "loss": 1.1206, "loss_dynamic": 1.3215, "loss_fixed": 0.9196, "step": 268 }, { "eob_count_fixed": 23042, "eob_loss_fixed": 0.1922, "epoch": 0.40719016083254495, "grad_norm": 0.20497803390026093, "learning_rate": 6.739491077279389e-06, "loss": 1.1322, "loss_dynamic": 1.3325, "loss_fixed": 0.932, "step": 269 }, { "eob_count_fixed": 23095, "eob_loss_fixed": 0.1977, "epoch": 0.4087038789025544, "grad_norm": 0.23152382671833038, "learning_rate": 6.7164956705030534e-06, "loss": 1.13, "loss_dynamic": 1.3142, "loss_fixed": 0.9458, "step": 270 }, { "eob_count_fixed": 23001, "eob_loss_fixed": 0.1694, "epoch": 0.4102175969725639, "grad_norm": 0.21911920607089996, "learning_rate": 6.693459032569005e-06, "loss": 1.1267, "loss_dynamic": 1.3365, "loss_fixed": 0.9168, "step": 271 }, { "eob_count_fixed": 23114, "eob_loss_fixed": 0.1788, "epoch": 0.4117313150425733, "grad_norm": 0.22078031301498413, "learning_rate": 6.670381716829789e-06, "loss": 1.1369, "loss_dynamic": 1.3339, "loss_fixed": 0.94, "step": 272 }, { "eob_count_fixed": 23255, "eob_loss_fixed": 0.1819, "epoch": 0.41324503311258276, "grad_norm": 0.21395455300807953, "learning_rate": 6.6472642776150575e-06, "loss": 1.1005, "loss_dynamic": 1.2552, "loss_fixed": 0.9458, "step": 273 }, { "eob_count_fixed": 23131, "eob_loss_fixed": 0.1952, "epoch": 0.41475875118259226, "grad_norm": 0.21822787821292877, "learning_rate": 6.6241072702182475e-06, "loss": 1.1723, "loss_dynamic": 1.4001, "loss_fixed": 0.9445, "step": 274 }, { "eob_count_fixed": 23190, "eob_loss_fixed": 0.1818, "epoch": 0.4162724692526017, "grad_norm": 0.20708195865154266, "learning_rate": 6.600911250883245e-06, "loss": 1.0875, "loss_dynamic": 1.2598, "loss_fixed": 0.9152, "step": 275 }, { "eob_count_fixed": 23217, "eob_loss_fixed": 0.1857, "epoch": 0.4177861873226112, "grad_norm": 0.2083449810743332, "learning_rate": 6.577676776791029e-06, "loss": 1.1704, "loss_dynamic": 1.3784, "loss_fixed": 0.9624, "step": 276 }, { "eob_count_fixed": 23048, "eob_loss_fixed": 0.1783, "epoch": 0.41929990539262063, "grad_norm": 0.20417432487010956, "learning_rate": 6.554404406046273e-06, "loss": 1.107, "loss_dynamic": 1.299, "loss_fixed": 0.915, "step": 277 }, { "eob_count_fixed": 23087, "eob_loss_fixed": 0.1689, "epoch": 0.42081362346263007, "grad_norm": 0.23975564539432526, "learning_rate": 6.531094697663963e-06, "loss": 1.1617, "loss_dynamic": 1.4031, "loss_fixed": 0.9203, "step": 278 }, { "eob_count_fixed": 23014, "eob_loss_fixed": 0.1844, "epoch": 0.42232734153263957, "grad_norm": 0.21082700788974762, "learning_rate": 6.507748211555936e-06, "loss": 1.1858, "loss_dynamic": 1.4256, "loss_fixed": 0.9459, "step": 279 }, { "eob_count_fixed": 23112, "eob_loss_fixed": 0.1836, "epoch": 0.423841059602649, "grad_norm": 0.22500140964984894, "learning_rate": 6.4843655085174685e-06, "loss": 1.1299, "loss_dynamic": 1.3245, "loss_fixed": 0.9352, "step": 280 }, { "eob_count_fixed": 23115, "eob_loss_fixed": 0.1884, "epoch": 0.42535477767265845, "grad_norm": 0.22141921520233154, "learning_rate": 6.460947150213778e-06, "loss": 1.1189, "loss_dynamic": 1.3315, "loss_fixed": 0.9062, "step": 281 }, { "eob_count_fixed": 23253, "eob_loss_fixed": 0.1849, "epoch": 0.42686849574266794, "grad_norm": 0.20540092885494232, "learning_rate": 6.437493699166539e-06, "loss": 1.0664, "loss_dynamic": 1.2081, "loss_fixed": 0.9248, "step": 282 }, { "eob_count_fixed": 23346, "eob_loss_fixed": 0.1816, "epoch": 0.4283822138126774, "grad_norm": 0.2138291895389557, "learning_rate": 6.414005718740383e-06, "loss": 1.0956, "loss_dynamic": 1.2619, "loss_fixed": 0.9293, "step": 283 }, { "eob_count_fixed": 23066, "eob_loss_fixed": 0.1766, "epoch": 0.4298959318826869, "grad_norm": 0.206741601228714, "learning_rate": 6.3904837731293415e-06, "loss": 1.1323, "loss_dynamic": 1.3311, "loss_fixed": 0.9335, "step": 284 }, { "eob_count_fixed": 23102, "eob_loss_fixed": 0.1847, "epoch": 0.4314096499526963, "grad_norm": 0.20858481526374817, "learning_rate": 6.366928427343319e-06, "loss": 1.0652, "loss_dynamic": 1.2236, "loss_fixed": 0.9068, "step": 285 }, { "eob_count_fixed": 23107, "eob_loss_fixed": 0.193, "epoch": 0.43292336802270576, "grad_norm": 0.22202512621879578, "learning_rate": 6.343340247194509e-06, "loss": 1.1437, "loss_dynamic": 1.3365, "loss_fixed": 0.9509, "step": 286 }, { "eob_count_fixed": 23106, "eob_loss_fixed": 0.1859, "epoch": 0.43443708609271525, "grad_norm": 0.19832028448581696, "learning_rate": 6.319719799283801e-06, "loss": 1.0924, "loss_dynamic": 1.2723, "loss_fixed": 0.9126, "step": 287 }, { "eob_count_fixed": 22981, "eob_loss_fixed": 0.1749, "epoch": 0.4359508041627247, "grad_norm": 0.2197405993938446, "learning_rate": 6.296067650987172e-06, "loss": 1.0708, "loss_dynamic": 1.2219, "loss_fixed": 0.9198, "step": 288 }, { "eob_count_fixed": 22924, "eob_loss_fixed": 0.1802, "epoch": 0.43746452223273413, "grad_norm": 0.241365447640419, "learning_rate": 6.272384370442065e-06, "loss": 1.1002, "loss_dynamic": 1.2758, "loss_fixed": 0.9247, "step": 289 }, { "eob_count_fixed": 22981, "eob_loss_fixed": 0.1769, "epoch": 0.4389782403027436, "grad_norm": 0.1928035318851471, "learning_rate": 6.248670526533734e-06, "loss": 1.1067, "loss_dynamic": 1.3029, "loss_fixed": 0.9105, "step": 290 }, { "eob_count_fixed": 23082, "eob_loss_fixed": 0.1799, "epoch": 0.44049195837275307, "grad_norm": 0.20250217616558075, "learning_rate": 6.224926688881583e-06, "loss": 1.1129, "loss_dynamic": 1.2961, "loss_fixed": 0.9298, "step": 291 }, { "eob_count_fixed": 23034, "eob_loss_fixed": 0.1928, "epoch": 0.44200567644276256, "grad_norm": 0.21595007181167603, "learning_rate": 6.201153427825477e-06, "loss": 1.1251, "loss_dynamic": 1.3273, "loss_fixed": 0.9228, "step": 292 }, { "eob_count_fixed": 23081, "eob_loss_fixed": 0.1756, "epoch": 0.443519394512772, "grad_norm": 0.21127356588840485, "learning_rate": 6.17735131441205e-06, "loss": 1.1337, "loss_dynamic": 1.3306, "loss_fixed": 0.9367, "step": 293 }, { "eob_count_fixed": 23308, "eob_loss_fixed": 0.1804, "epoch": 0.44503311258278144, "grad_norm": 0.2083386331796646, "learning_rate": 6.153520920380991e-06, "loss": 1.1144, "loss_dynamic": 1.3026, "loss_fixed": 0.9262, "step": 294 }, { "eob_count_fixed": 22992, "eob_loss_fixed": 0.1816, "epoch": 0.44654683065279094, "grad_norm": 0.20555631816387177, "learning_rate": 6.1296628181512966e-06, "loss": 1.1151, "loss_dynamic": 1.3088, "loss_fixed": 0.9215, "step": 295 }, { "eob_count_fixed": 22854, "eob_loss_fixed": 0.1969, "epoch": 0.4480605487228004, "grad_norm": 0.20284795761108398, "learning_rate": 6.10577758080753e-06, "loss": 1.1198, "loss_dynamic": 1.2996, "loss_fixed": 0.9399, "step": 296 }, { "eob_count_fixed": 23064, "eob_loss_fixed": 0.1856, "epoch": 0.4495742667928098, "grad_norm": 0.21810874342918396, "learning_rate": 6.0818657820860635e-06, "loss": 1.1284, "loss_dynamic": 1.3221, "loss_fixed": 0.9347, "step": 297 }, { "eob_count_fixed": 23069, "eob_loss_fixed": 0.1856, "epoch": 0.4510879848628193, "grad_norm": 0.20820103585720062, "learning_rate": 6.057927996361276e-06, "loss": 1.1274, "loss_dynamic": 1.3403, "loss_fixed": 0.9145, "step": 298 }, { "eob_count_fixed": 23154, "eob_loss_fixed": 0.169, "epoch": 0.45260170293282875, "grad_norm": 0.2861765921115875, "learning_rate": 6.033964798631776e-06, "loss": 1.0971, "loss_dynamic": 1.2873, "loss_fixed": 0.907, "step": 299 }, { "eob_count_fixed": 23271, "eob_loss_fixed": 0.1706, "epoch": 0.45411542100283825, "grad_norm": 0.2304537147283554, "learning_rate": 6.00997676450658e-06, "loss": 1.0998, "loss_dynamic": 1.3018, "loss_fixed": 0.8978, "step": 300 }, { "eob_count_fixed": 22886, "eob_loss_fixed": 0.1804, "epoch": 0.4556291390728477, "grad_norm": 0.20172514021396637, "learning_rate": 5.985964470191288e-06, "loss": 1.0691, "loss_dynamic": 1.2163, "loss_fixed": 0.922, "step": 301 }, { "eob_count_fixed": 23022, "eob_loss_fixed": 0.1831, "epoch": 0.45714285714285713, "grad_norm": 0.2111663818359375, "learning_rate": 5.961928492474243e-06, "loss": 1.0857, "loss_dynamic": 1.2701, "loss_fixed": 0.9013, "step": 302 }, { "eob_count_fixed": 23126, "eob_loss_fixed": 0.1857, "epoch": 0.4586565752128666, "grad_norm": 0.20045487582683563, "learning_rate": 5.937869408712676e-06, "loss": 1.1335, "loss_dynamic": 1.3372, "loss_fixed": 0.9299, "step": 303 }, { "eob_count_fixed": 23260, "eob_loss_fixed": 0.1826, "epoch": 0.46017029328287606, "grad_norm": 0.21822811663150787, "learning_rate": 5.913787796818841e-06, "loss": 1.1028, "loss_dynamic": 1.2768, "loss_fixed": 0.9287, "step": 304 }, { "eob_count_fixed": 23107, "eob_loss_fixed": 0.1718, "epoch": 0.4616840113528855, "grad_norm": 0.21210405230522156, "learning_rate": 5.889684235246125e-06, "loss": 1.1291, "loss_dynamic": 1.3333, "loss_fixed": 0.925, "step": 305 }, { "eob_count_fixed": 23092, "eob_loss_fixed": 0.1732, "epoch": 0.463197729422895, "grad_norm": 0.2097574770450592, "learning_rate": 5.86555930297516e-06, "loss": 1.0918, "loss_dynamic": 1.2631, "loss_fixed": 0.9205, "step": 306 }, { "eob_count_fixed": 23208, "eob_loss_fixed": 0.1874, "epoch": 0.46471144749290444, "grad_norm": 0.22585654258728027, "learning_rate": 5.841413579499915e-06, "loss": 1.1268, "loss_dynamic": 1.324, "loss_fixed": 0.9296, "step": 307 }, { "eob_count_fixed": 23304, "eob_loss_fixed": 0.1804, "epoch": 0.46622516556291393, "grad_norm": 0.2288316935300827, "learning_rate": 5.817247644813774e-06, "loss": 1.1074, "loss_dynamic": 1.2706, "loss_fixed": 0.9443, "step": 308 }, { "eob_count_fixed": 23294, "eob_loss_fixed": 0.182, "epoch": 0.4677388836329234, "grad_norm": 0.21061351895332336, "learning_rate": 5.793062079395602e-06, "loss": 1.0823, "loss_dynamic": 1.2419, "loss_fixed": 0.9228, "step": 309 }, { "eob_count_fixed": 23021, "eob_loss_fixed": 0.1745, "epoch": 0.4692526017029328, "grad_norm": 0.21294938027858734, "learning_rate": 5.768857464195815e-06, "loss": 1.1222, "loss_dynamic": 1.3267, "loss_fixed": 0.9178, "step": 310 }, { "eob_count_fixed": 23051, "eob_loss_fixed": 0.1795, "epoch": 0.4707663197729423, "grad_norm": 0.21609732508659363, "learning_rate": 5.744634380622402e-06, "loss": 1.1231, "loss_dynamic": 1.3223, "loss_fixed": 0.9238, "step": 311 }, { "eob_count_fixed": 23137, "eob_loss_fixed": 0.1847, "epoch": 0.47228003784295175, "grad_norm": 0.2041160762310028, "learning_rate": 5.720393410526982e-06, "loss": 1.1214, "loss_dynamic": 1.312, "loss_fixed": 0.9308, "step": 312 }, { "eob_count_fixed": 23128, "eob_loss_fixed": 0.1823, "epoch": 0.4737937559129612, "grad_norm": 0.21784985065460205, "learning_rate": 5.696135136190815e-06, "loss": 1.1041, "loss_dynamic": 1.3073, "loss_fixed": 0.9008, "step": 313 }, { "eob_count_fixed": 23074, "eob_loss_fixed": 0.1758, "epoch": 0.4753074739829707, "grad_norm": 0.1973234862089157, "learning_rate": 5.671860140310815e-06, "loss": 1.0901, "loss_dynamic": 1.2606, "loss_fixed": 0.9195, "step": 314 }, { "eob_count_fixed": 23288, "eob_loss_fixed": 0.173, "epoch": 0.4768211920529801, "grad_norm": 0.2092382162809372, "learning_rate": 5.647569005985563e-06, "loss": 1.1355, "loss_dynamic": 1.3479, "loss_fixed": 0.9231, "step": 315 }, { "eob_count_fixed": 22983, "eob_loss_fixed": 0.1678, "epoch": 0.4783349101229896, "grad_norm": 0.20711465179920197, "learning_rate": 5.623262316701287e-06, "loss": 1.1448, "loss_dynamic": 1.3637, "loss_fixed": 0.9259, "step": 316 }, { "eob_count_fixed": 23116, "eob_loss_fixed": 0.1841, "epoch": 0.47984862819299906, "grad_norm": 0.22020606696605682, "learning_rate": 5.598940656317859e-06, "loss": 1.1354, "loss_dynamic": 1.3442, "loss_fixed": 0.9267, "step": 317 }, { "eob_count_fixed": 23110, "eob_loss_fixed": 0.1872, "epoch": 0.4813623462630085, "grad_norm": 0.21547763049602509, "learning_rate": 5.574604609054764e-06, "loss": 1.1378, "loss_dynamic": 1.3259, "loss_fixed": 0.9496, "step": 318 }, { "eob_count_fixed": 23286, "eob_loss_fixed": 0.1857, "epoch": 0.482876064333018, "grad_norm": 0.22140221297740936, "learning_rate": 5.550254759477065e-06, "loss": 1.1044, "loss_dynamic": 1.3004, "loss_fixed": 0.9084, "step": 319 }, { "eob_count_fixed": 23153, "eob_loss_fixed": 0.1812, "epoch": 0.48438978240302744, "grad_norm": 0.2056232988834381, "learning_rate": 5.525891692481367e-06, "loss": 1.0937, "loss_dynamic": 1.2577, "loss_fixed": 0.9298, "step": 320 }, { "eob_count_fixed": 23044, "eob_loss_fixed": 0.1769, "epoch": 0.4859035004730369, "grad_norm": 0.21026545763015747, "learning_rate": 5.501515993281763e-06, "loss": 1.1792, "loss_dynamic": 1.4201, "loss_fixed": 0.9383, "step": 321 }, { "eob_count_fixed": 23068, "eob_loss_fixed": 0.1798, "epoch": 0.48741721854304637, "grad_norm": 0.21225573122501373, "learning_rate": 5.47712824739578e-06, "loss": 1.1612, "loss_dynamic": 1.3978, "loss_fixed": 0.9246, "step": 322 }, { "eob_count_fixed": 23174, "eob_loss_fixed": 0.1714, "epoch": 0.4889309366130558, "grad_norm": 0.21689331531524658, "learning_rate": 5.45272904063031e-06, "loss": 1.1103, "loss_dynamic": 1.3007, "loss_fixed": 0.9199, "step": 323 }, { "eob_count_fixed": 23067, "eob_loss_fixed": 0.1795, "epoch": 0.4904446546830653, "grad_norm": 0.2203279286623001, "learning_rate": 5.428318959067543e-06, "loss": 1.0856, "loss_dynamic": 1.2407, "loss_fixed": 0.9305, "step": 324 }, { "eob_count_fixed": 23262, "eob_loss_fixed": 0.1887, "epoch": 0.49195837275307475, "grad_norm": 0.20611776411533356, "learning_rate": 5.403898589050891e-06, "loss": 1.1473, "loss_dynamic": 1.3558, "loss_fixed": 0.9387, "step": 325 }, { "eob_count_fixed": 22965, "eob_loss_fixed": 0.1831, "epoch": 0.4934720908230842, "grad_norm": 0.21164338290691376, "learning_rate": 5.379468517170894e-06, "loss": 1.089, "loss_dynamic": 1.2699, "loss_fixed": 0.9081, "step": 326 }, { "eob_count_fixed": 22925, "eob_loss_fixed": 0.1776, "epoch": 0.4949858088930937, "grad_norm": 0.2125881314277649, "learning_rate": 5.355029330251142e-06, "loss": 1.1042, "loss_dynamic": 1.2994, "loss_fixed": 0.909, "step": 327 }, { "eob_count_fixed": 23206, "eob_loss_fixed": 0.1857, "epoch": 0.4964995269631031, "grad_norm": 0.20022618770599365, "learning_rate": 5.330581615334165e-06, "loss": 1.1328, "loss_dynamic": 1.3175, "loss_fixed": 0.9481, "step": 328 }, { "eob_count_fixed": 23102, "eob_loss_fixed": 0.1701, "epoch": 0.49801324503311256, "grad_norm": 0.21667128801345825, "learning_rate": 5.306125959667352e-06, "loss": 1.131, "loss_dynamic": 1.3626, "loss_fixed": 0.8995, "step": 329 }, { "eob_count_fixed": 23146, "eob_loss_fixed": 0.1736, "epoch": 0.49952696310312206, "grad_norm": 0.2071097493171692, "learning_rate": 5.281662950688822e-06, "loss": 1.1193, "loss_dynamic": 1.3127, "loss_fixed": 0.9259, "step": 330 }, { "eob_count_fixed": 23168, "eob_loss_fixed": 0.1752, "epoch": 0.5010406811731315, "grad_norm": 0.2101971060037613, "learning_rate": 5.257193176013331e-06, "loss": 1.0793, "loss_dynamic": 1.2531, "loss_fixed": 0.9055, "step": 331 }, { "eob_count_fixed": 23074, "eob_loss_fixed": 0.177, "epoch": 0.5025543992431409, "grad_norm": 0.20413200557231903, "learning_rate": 5.232717223418149e-06, "loss": 1.0755, "loss_dynamic": 1.2489, "loss_fixed": 0.9021, "step": 332 }, { "eob_count_fixed": 23077, "eob_loss_fixed": 0.1846, "epoch": 0.5040681173131504, "grad_norm": 0.21084538102149963, "learning_rate": 5.208235680828941e-06, "loss": 1.1293, "loss_dynamic": 1.332, "loss_fixed": 0.9266, "step": 333 }, { "eob_count_fixed": 23156, "eob_loss_fixed": 0.1796, "epoch": 0.5055818353831599, "grad_norm": 0.2145121544599533, "learning_rate": 5.183749136305651e-06, "loss": 1.13, "loss_dynamic": 1.3181, "loss_fixed": 0.942, "step": 334 }, { "eob_count_fixed": 23000, "eob_loss_fixed": 0.1736, "epoch": 0.5070955534531694, "grad_norm": 0.20461194217205048, "learning_rate": 5.1592581780283676e-06, "loss": 1.0852, "loss_dynamic": 1.267, "loss_fixed": 0.9034, "step": 335 }, { "eob_count_fixed": 23023, "eob_loss_fixed": 0.1801, "epoch": 0.5086092715231788, "grad_norm": 0.21498383581638336, "learning_rate": 5.134763394283204e-06, "loss": 1.1583, "loss_dynamic": 1.3831, "loss_fixed": 0.9334, "step": 336 }, { "eob_count_fixed": 23273, "eob_loss_fixed": 0.1812, "epoch": 0.5101229895931882, "grad_norm": 0.21106356382369995, "learning_rate": 5.110265373448161e-06, "loss": 1.1004, "loss_dynamic": 1.2902, "loss_fixed": 0.9106, "step": 337 }, { "eob_count_fixed": 23281, "eob_loss_fixed": 0.1912, "epoch": 0.5116367076631977, "grad_norm": 0.21634458005428314, "learning_rate": 5.085764703978995e-06, "loss": 1.125, "loss_dynamic": 1.3562, "loss_fixed": 0.8938, "step": 338 }, { "eob_count_fixed": 23050, "eob_loss_fixed": 0.1725, "epoch": 0.5131504257332072, "grad_norm": 0.21512791514396667, "learning_rate": 5.061261974395087e-06, "loss": 1.0419, "loss_dynamic": 1.1881, "loss_fixed": 0.8956, "step": 339 }, { "eob_count_fixed": 22963, "eob_loss_fixed": 0.177, "epoch": 0.5146641438032167, "grad_norm": 0.22542926669120789, "learning_rate": 5.036757773265301e-06, "loss": 1.1129, "loss_dynamic": 1.3227, "loss_fixed": 0.903, "step": 340 }, { "eob_count_fixed": 23116, "eob_loss_fixed": 0.17, "epoch": 0.5161778618732261, "grad_norm": 0.19417139887809753, "learning_rate": 5.0122526891938495e-06, "loss": 1.0868, "loss_dynamic": 1.2853, "loss_fixed": 0.8882, "step": 341 }, { "eob_count_fixed": 23089, "eob_loss_fixed": 0.1774, "epoch": 0.5176915799432356, "grad_norm": 0.21303316950798035, "learning_rate": 4.987747310806152e-06, "loss": 1.0684, "loss_dynamic": 1.2349, "loss_fixed": 0.9019, "step": 342 }, { "eob_count_fixed": 22988, "eob_loss_fixed": 0.1776, "epoch": 0.519205298013245, "grad_norm": 0.2118719518184662, "learning_rate": 4.963242226734701e-06, "loss": 1.1569, "loss_dynamic": 1.3913, "loss_fixed": 0.9226, "step": 343 }, { "eob_count_fixed": 23067, "eob_loss_fixed": 0.1834, "epoch": 0.5207190160832545, "grad_norm": 0.21722571551799774, "learning_rate": 4.938738025604914e-06, "loss": 1.1275, "loss_dynamic": 1.3204, "loss_fixed": 0.9347, "step": 344 }, { "eob_count_fixed": 23124, "eob_loss_fixed": 0.1893, "epoch": 0.522232734153264, "grad_norm": 0.21412838995456696, "learning_rate": 4.914235296021006e-06, "loss": 1.1355, "loss_dynamic": 1.3073, "loss_fixed": 0.9638, "step": 345 }, { "eob_count_fixed": 23185, "eob_loss_fixed": 0.1755, "epoch": 0.5237464522232734, "grad_norm": 0.23030295968055725, "learning_rate": 4.889734626551842e-06, "loss": 1.1233, "loss_dynamic": 1.3283, "loss_fixed": 0.9184, "step": 346 }, { "eob_count_fixed": 23227, "eob_loss_fixed": 0.1725, "epoch": 0.5252601702932829, "grad_norm": 0.2090342938899994, "learning_rate": 4.865236605716796e-06, "loss": 1.1114, "loss_dynamic": 1.3223, "loss_fixed": 0.9006, "step": 347 }, { "eob_count_fixed": 23168, "eob_loss_fixed": 0.182, "epoch": 0.5267738883632923, "grad_norm": 0.21028518676757812, "learning_rate": 4.840741821971633e-06, "loss": 1.1108, "loss_dynamic": 1.2895, "loss_fixed": 0.9321, "step": 348 }, { "eob_count_fixed": 23201, "eob_loss_fixed": 0.1804, "epoch": 0.5282876064333017, "grad_norm": 0.21016070246696472, "learning_rate": 4.81625086369435e-06, "loss": 1.1269, "loss_dynamic": 1.325, "loss_fixed": 0.9287, "step": 349 }, { "eob_count_fixed": 22850, "eob_loss_fixed": 0.1784, "epoch": 0.5298013245033113, "grad_norm": 0.20363572239875793, "learning_rate": 4.7917643191710594e-06, "loss": 1.1004, "loss_dynamic": 1.2896, "loss_fixed": 0.9112, "step": 350 }, { "eob_count_fixed": 23349, "eob_loss_fixed": 0.1829, "epoch": 0.5313150425733207, "grad_norm": 0.20569369196891785, "learning_rate": 4.767282776581853e-06, "loss": 1.1302, "loss_dynamic": 1.3338, "loss_fixed": 0.9265, "step": 351 }, { "eob_count_fixed": 23227, "eob_loss_fixed": 0.1736, "epoch": 0.5328287606433302, "grad_norm": 0.20726697146892548, "learning_rate": 4.74280682398667e-06, "loss": 1.1039, "loss_dynamic": 1.3043, "loss_fixed": 0.9035, "step": 352 }, { "eob_count_fixed": 23151, "eob_loss_fixed": 0.1775, "epoch": 0.5343424787133396, "grad_norm": 0.23888881504535675, "learning_rate": 4.718337049311178e-06, "loss": 1.1206, "loss_dynamic": 1.3253, "loss_fixed": 0.9158, "step": 353 }, { "eob_count_fixed": 23193, "eob_loss_fixed": 0.1783, "epoch": 0.5358561967833491, "grad_norm": 0.2170393019914627, "learning_rate": 4.693874040332649e-06, "loss": 1.1988, "loss_dynamic": 1.452, "loss_fixed": 0.9456, "step": 354 }, { "eob_count_fixed": 23274, "eob_loss_fixed": 0.171, "epoch": 0.5373699148533586, "grad_norm": 0.21413402259349823, "learning_rate": 4.669418384665836e-06, "loss": 1.1238, "loss_dynamic": 1.3308, "loss_fixed": 0.9169, "step": 355 }, { "eob_count_fixed": 23144, "eob_loss_fixed": 0.1717, "epoch": 0.538883632923368, "grad_norm": 0.2286302149295807, "learning_rate": 4.644970669748859e-06, "loss": 1.1871, "loss_dynamic": 1.4385, "loss_fixed": 0.9357, "step": 356 }, { "eob_count_fixed": 23103, "eob_loss_fixed": 0.1858, "epoch": 0.5403973509933775, "grad_norm": 0.2103649228811264, "learning_rate": 4.620531482829107e-06, "loss": 1.1351, "loss_dynamic": 1.34, "loss_fixed": 0.9301, "step": 357 }, { "eob_count_fixed": 23145, "eob_loss_fixed": 0.1785, "epoch": 0.5419110690633869, "grad_norm": 0.21782265603542328, "learning_rate": 4.59610141094911e-06, "loss": 1.091, "loss_dynamic": 1.2746, "loss_fixed": 0.9074, "step": 358 }, { "eob_count_fixed": 23129, "eob_loss_fixed": 0.1844, "epoch": 0.5434247871333964, "grad_norm": 0.21757525205612183, "learning_rate": 4.571681040932458e-06, "loss": 1.1134, "loss_dynamic": 1.3115, "loss_fixed": 0.9153, "step": 359 }, { "eob_count_fixed": 23060, "eob_loss_fixed": 0.1814, "epoch": 0.5449385052034059, "grad_norm": 0.22426600754261017, "learning_rate": 4.547270959369692e-06, "loss": 1.1026, "loss_dynamic": 1.2945, "loss_fixed": 0.9106, "step": 360 }, { "eob_count_fixed": 23156, "eob_loss_fixed": 0.1791, "epoch": 0.5464522232734154, "grad_norm": 0.21176950633525848, "learning_rate": 4.522871752604221e-06, "loss": 1.1398, "loss_dynamic": 1.3473, "loss_fixed": 0.9323, "step": 361 }, { "eob_count_fixed": 23032, "eob_loss_fixed": 0.172, "epoch": 0.5479659413434248, "grad_norm": 0.20242908596992493, "learning_rate": 4.498484006718239e-06, "loss": 1.0743, "loss_dynamic": 1.2464, "loss_fixed": 0.9022, "step": 362 }, { "eob_count_fixed": 23098, "eob_loss_fixed": 0.1745, "epoch": 0.5494796594134342, "grad_norm": 0.21102313697338104, "learning_rate": 4.474108307518633e-06, "loss": 1.1018, "loss_dynamic": 1.2882, "loss_fixed": 0.9153, "step": 363 }, { "eob_count_fixed": 23155, "eob_loss_fixed": 0.1814, "epoch": 0.5509933774834437, "grad_norm": 0.21991442143917084, "learning_rate": 4.449745240522937e-06, "loss": 1.1088, "loss_dynamic": 1.3063, "loss_fixed": 0.9112, "step": 364 }, { "eob_count_fixed": 23285, "eob_loss_fixed": 0.177, "epoch": 0.5525070955534531, "grad_norm": 0.21315021812915802, "learning_rate": 4.425395390945238e-06, "loss": 1.1322, "loss_dynamic": 1.3504, "loss_fixed": 0.914, "step": 365 }, { "eob_count_fixed": 23249, "eob_loss_fixed": 0.1794, "epoch": 0.5540208136234627, "grad_norm": 0.2163325846195221, "learning_rate": 4.4010593436821416e-06, "loss": 1.1392, "loss_dynamic": 1.3623, "loss_fixed": 0.9161, "step": 366 }, { "eob_count_fixed": 23228, "eob_loss_fixed": 0.1781, "epoch": 0.5555345316934721, "grad_norm": 0.21668337285518646, "learning_rate": 4.376737683298715e-06, "loss": 1.1041, "loss_dynamic": 1.2651, "loss_fixed": 0.943, "step": 367 }, { "eob_count_fixed": 22968, "eob_loss_fixed": 0.1827, "epoch": 0.5570482497634816, "grad_norm": 0.22099612653255463, "learning_rate": 4.352430994014439e-06, "loss": 1.0873, "loss_dynamic": 1.2702, "loss_fixed": 0.9044, "step": 368 }, { "eob_count_fixed": 23164, "eob_loss_fixed": 0.1845, "epoch": 0.558561967833491, "grad_norm": 0.21094657480716705, "learning_rate": 4.328139859689185e-06, "loss": 1.0885, "loss_dynamic": 1.257, "loss_fixed": 0.92, "step": 369 }, { "eob_count_fixed": 22941, "eob_loss_fixed": 0.1856, "epoch": 0.5600756859035004, "grad_norm": 0.21676358580589294, "learning_rate": 4.303864863809186e-06, "loss": 1.1115, "loss_dynamic": 1.2837, "loss_fixed": 0.9393, "step": 370 }, { "eob_count_fixed": 23157, "eob_loss_fixed": 0.1766, "epoch": 0.56158940397351, "grad_norm": 0.20060774683952332, "learning_rate": 4.279606589473019e-06, "loss": 1.0846, "loss_dynamic": 1.2673, "loss_fixed": 0.902, "step": 371 }, { "eob_count_fixed": 23152, "eob_loss_fixed": 0.1852, "epoch": 0.5631031220435194, "grad_norm": 0.21493381261825562, "learning_rate": 4.255365619377597e-06, "loss": 1.1415, "loss_dynamic": 1.3507, "loss_fixed": 0.9323, "step": 372 }, { "eob_count_fixed": 23125, "eob_loss_fixed": 0.1855, "epoch": 0.5646168401135289, "grad_norm": 0.20571355521678925, "learning_rate": 4.2311425358041865e-06, "loss": 1.132, "loss_dynamic": 1.3418, "loss_fixed": 0.9222, "step": 373 }, { "eob_count_fixed": 23158, "eob_loss_fixed": 0.174, "epoch": 0.5661305581835383, "grad_norm": 0.22777028381824493, "learning_rate": 4.2069379206043985e-06, "loss": 1.1212, "loss_dynamic": 1.3453, "loss_fixed": 0.8972, "step": 374 }, { "eob_count_fixed": 23176, "eob_loss_fixed": 0.1808, "epoch": 0.5676442762535477, "grad_norm": 0.21594975888729095, "learning_rate": 4.18275235518623e-06, "loss": 1.1254, "loss_dynamic": 1.3271, "loss_fixed": 0.9237, "step": 375 }, { "eob_count_fixed": 23087, "eob_loss_fixed": 0.1749, "epoch": 0.5691579943235573, "grad_norm": 0.21433459222316742, "learning_rate": 4.158586420500086e-06, "loss": 1.1447, "loss_dynamic": 1.3571, "loss_fixed": 0.9323, "step": 376 }, { "eob_count_fixed": 23076, "eob_loss_fixed": 0.1832, "epoch": 0.5706717123935667, "grad_norm": 0.2405467927455902, "learning_rate": 4.134440697024841e-06, "loss": 1.0946, "loss_dynamic": 1.2752, "loss_fixed": 0.9139, "step": 377 }, { "eob_count_fixed": 23110, "eob_loss_fixed": 0.1848, "epoch": 0.5721854304635762, "grad_norm": 0.2120642215013504, "learning_rate": 4.110315764753877e-06, "loss": 1.1196, "loss_dynamic": 1.2997, "loss_fixed": 0.9395, "step": 378 }, { "eob_count_fixed": 23214, "eob_loss_fixed": 0.1773, "epoch": 0.5736991485335856, "grad_norm": 0.1975390613079071, "learning_rate": 4.0862122031811594e-06, "loss": 1.1083, "loss_dynamic": 1.2739, "loss_fixed": 0.9426, "step": 379 }, { "eob_count_fixed": 22930, "eob_loss_fixed": 0.1841, "epoch": 0.575212866603595, "grad_norm": 0.2325475513935089, "learning_rate": 4.062130591287325e-06, "loss": 1.1706, "loss_dynamic": 1.3869, "loss_fixed": 0.9543, "step": 380 }, { "eob_count_fixed": 23100, "eob_loss_fixed": 0.172, "epoch": 0.5767265846736045, "grad_norm": 0.2254154533147812, "learning_rate": 4.038071507525759e-06, "loss": 1.148, "loss_dynamic": 1.3627, "loss_fixed": 0.9334, "step": 381 }, { "eob_count_fixed": 23012, "eob_loss_fixed": 0.1795, "epoch": 0.578240302743614, "grad_norm": 0.21779169142246246, "learning_rate": 4.0140355298087145e-06, "loss": 1.1162, "loss_dynamic": 1.2946, "loss_fixed": 0.9378, "step": 382 }, { "eob_count_fixed": 23099, "eob_loss_fixed": 0.1681, "epoch": 0.5797540208136235, "grad_norm": 0.1937403678894043, "learning_rate": 3.990023235493422e-06, "loss": 1.0953, "loss_dynamic": 1.2903, "loss_fixed": 0.9002, "step": 383 }, { "eob_count_fixed": 23182, "eob_loss_fixed": 0.1676, "epoch": 0.5812677388836329, "grad_norm": 0.2050800919532776, "learning_rate": 3.9660352013682255e-06, "loss": 1.039, "loss_dynamic": 1.1877, "loss_fixed": 0.8903, "step": 384 }, { "eob_count_fixed": 23135, "eob_loss_fixed": 0.1729, "epoch": 0.5827814569536424, "grad_norm": 0.2182866632938385, "learning_rate": 3.942072003638725e-06, "loss": 1.1148, "loss_dynamic": 1.3112, "loss_fixed": 0.9183, "step": 385 }, { "eob_count_fixed": 23024, "eob_loss_fixed": 0.1803, "epoch": 0.5842951750236518, "grad_norm": 0.22225452959537506, "learning_rate": 3.918134217913937e-06, "loss": 1.1321, "loss_dynamic": 1.3513, "loss_fixed": 0.9129, "step": 386 }, { "eob_count_fixed": 23060, "eob_loss_fixed": 0.1746, "epoch": 0.5858088930936614, "grad_norm": 0.19819849729537964, "learning_rate": 3.8942224191924705e-06, "loss": 1.0934, "loss_dynamic": 1.2776, "loss_fixed": 0.9093, "step": 387 }, { "eob_count_fixed": 23152, "eob_loss_fixed": 0.1623, "epoch": 0.5873226111636708, "grad_norm": 0.20713703334331512, "learning_rate": 3.870337181848704e-06, "loss": 1.1177, "loss_dynamic": 1.3293, "loss_fixed": 0.9062, "step": 388 }, { "eob_count_fixed": 23216, "eob_loss_fixed": 0.1665, "epoch": 0.5888363292336802, "grad_norm": 0.20775070786476135, "learning_rate": 3.84647907961901e-06, "loss": 1.0658, "loss_dynamic": 1.2235, "loss_fixed": 0.9081, "step": 389 }, { "eob_count_fixed": 22957, "eob_loss_fixed": 0.1783, "epoch": 0.5903500473036897, "grad_norm": 0.2085498720407486, "learning_rate": 3.822648685587951e-06, "loss": 1.1032, "loss_dynamic": 1.2987, "loss_fixed": 0.9077, "step": 390 }, { "eob_count_fixed": 23215, "eob_loss_fixed": 0.1723, "epoch": 0.5918637653736991, "grad_norm": 0.20074789226055145, "learning_rate": 3.798846572174526e-06, "loss": 1.0994, "loss_dynamic": 1.294, "loss_fixed": 0.9047, "step": 391 }, { "eob_count_fixed": 23372, "eob_loss_fixed": 0.1793, "epoch": 0.5933774834437087, "grad_norm": 0.19605715572834015, "learning_rate": 3.775073311118419e-06, "loss": 1.0811, "loss_dynamic": 1.2612, "loss_fixed": 0.901, "step": 392 }, { "eob_count_fixed": 23304, "eob_loss_fixed": 0.1794, "epoch": 0.5948912015137181, "grad_norm": 0.2011772096157074, "learning_rate": 3.7513294734662663e-06, "loss": 1.0786, "loss_dynamic": 1.2504, "loss_fixed": 0.9069, "step": 393 }, { "eob_count_fixed": 23163, "eob_loss_fixed": 0.1681, "epoch": 0.5964049195837275, "grad_norm": 0.20544405281543732, "learning_rate": 3.727615629557936e-06, "loss": 1.0775, "loss_dynamic": 1.2711, "loss_fixed": 0.8839, "step": 394 }, { "eob_count_fixed": 23047, "eob_loss_fixed": 0.1784, "epoch": 0.597918637653737, "grad_norm": 0.23554764688014984, "learning_rate": 3.703932349012829e-06, "loss": 1.0703, "loss_dynamic": 1.2262, "loss_fixed": 0.9145, "step": 395 }, { "eob_count_fixed": 23087, "eob_loss_fixed": 0.1739, "epoch": 0.5994323557237464, "grad_norm": 0.20225323736667633, "learning_rate": 3.680280200716201e-06, "loss": 1.1144, "loss_dynamic": 1.313, "loss_fixed": 0.9158, "step": 396 }, { "eob_count_fixed": 23045, "eob_loss_fixed": 0.1768, "epoch": 0.6009460737937559, "grad_norm": 0.20049576461315155, "learning_rate": 3.6566597528054925e-06, "loss": 1.0879, "loss_dynamic": 1.2726, "loss_fixed": 0.9032, "step": 397 }, { "eob_count_fixed": 23180, "eob_loss_fixed": 0.1742, "epoch": 0.6024597918637654, "grad_norm": 0.21610061824321747, "learning_rate": 3.6330715726566803e-06, "loss": 1.103, "loss_dynamic": 1.3142, "loss_fixed": 0.8918, "step": 398 }, { "eob_count_fixed": 23092, "eob_loss_fixed": 0.1847, "epoch": 0.6039735099337749, "grad_norm": 0.1942455768585205, "learning_rate": 3.6095162268706598e-06, "loss": 1.0782, "loss_dynamic": 1.242, "loss_fixed": 0.9143, "step": 399 }, { "eob_count_fixed": 22986, "eob_loss_fixed": 0.1833, "epoch": 0.6054872280037843, "grad_norm": 0.20393240451812744, "learning_rate": 3.58599428125962e-06, "loss": 1.1089, "loss_dynamic": 1.3091, "loss_fixed": 0.9086, "step": 400 }, { "eob_count_fixed": 23114, "eob_loss_fixed": 0.1803, "epoch": 0.6070009460737937, "grad_norm": 0.2003854215145111, "learning_rate": 3.5625063008334603e-06, "loss": 1.0628, "loss_dynamic": 1.2243, "loss_fixed": 0.9013, "step": 401 }, { "eob_count_fixed": 23107, "eob_loss_fixed": 0.1822, "epoch": 0.6085146641438032, "grad_norm": 0.1951499730348587, "learning_rate": 3.539052849786224e-06, "loss": 1.1024, "loss_dynamic": 1.2645, "loss_fixed": 0.9403, "step": 402 }, { "eob_count_fixed": 23096, "eob_loss_fixed": 0.1711, "epoch": 0.6100283822138127, "grad_norm": 0.21109794080257416, "learning_rate": 3.5156344914825324e-06, "loss": 1.0797, "loss_dynamic": 1.2493, "loss_fixed": 0.9101, "step": 403 }, { "eob_count_fixed": 23214, "eob_loss_fixed": 0.183, "epoch": 0.6115421002838222, "grad_norm": 0.21089673042297363, "learning_rate": 3.4922517884440643e-06, "loss": 1.1215, "loss_dynamic": 1.3215, "loss_fixed": 0.9215, "step": 404 }, { "eob_count_fixed": 22944, "eob_loss_fixed": 0.1834, "epoch": 0.6130558183538316, "grad_norm": 0.20545899868011475, "learning_rate": 3.46890530233604e-06, "loss": 1.1198, "loss_dynamic": 1.3243, "loss_fixed": 0.9153, "step": 405 }, { "eob_count_fixed": 23014, "eob_loss_fixed": 0.1721, "epoch": 0.614569536423841, "grad_norm": 0.21258802711963654, "learning_rate": 3.445595593953728e-06, "loss": 1.111, "loss_dynamic": 1.2887, "loss_fixed": 0.9334, "step": 406 }, { "eob_count_fixed": 23159, "eob_loss_fixed": 0.1746, "epoch": 0.6160832544938505, "grad_norm": 0.2126374989748001, "learning_rate": 3.4223232232089735e-06, "loss": 1.0636, "loss_dynamic": 1.2337, "loss_fixed": 0.8935, "step": 407 }, { "eob_count_fixed": 23193, "eob_loss_fixed": 0.1802, "epoch": 0.61759697256386, "grad_norm": 0.20966534316539764, "learning_rate": 3.399088749116756e-06, "loss": 1.1083, "loss_dynamic": 1.2906, "loss_fixed": 0.9259, "step": 408 }, { "eob_count_fixed": 23069, "eob_loss_fixed": 0.1831, "epoch": 0.6191106906338695, "grad_norm": 0.21623684465885162, "learning_rate": 3.3758927297817546e-06, "loss": 1.1426, "loss_dynamic": 1.3356, "loss_fixed": 0.9495, "step": 409 }, { "eob_count_fixed": 23184, "eob_loss_fixed": 0.1764, "epoch": 0.6206244087038789, "grad_norm": 0.20789849758148193, "learning_rate": 3.3527357223849446e-06, "loss": 1.0688, "loss_dynamic": 1.2343, "loss_fixed": 0.9034, "step": 410 }, { "eob_count_fixed": 23104, "eob_loss_fixed": 0.1833, "epoch": 0.6221381267738884, "grad_norm": 0.19965368509292603, "learning_rate": 3.329618283170211e-06, "loss": 1.1198, "loss_dynamic": 1.3363, "loss_fixed": 0.9034, "step": 411 }, { "eob_count_fixed": 23052, "eob_loss_fixed": 0.1802, "epoch": 0.6236518448438978, "grad_norm": 0.21918784081935883, "learning_rate": 3.3065409674309954e-06, "loss": 1.1157, "loss_dynamic": 1.3217, "loss_fixed": 0.9098, "step": 412 }, { "eob_count_fixed": 23118, "eob_loss_fixed": 0.1837, "epoch": 0.6251655629139072, "grad_norm": 0.21683639287948608, "learning_rate": 3.2835043294969482e-06, "loss": 1.1217, "loss_dynamic": 1.304, "loss_fixed": 0.9394, "step": 413 }, { "eob_count_fixed": 23193, "eob_loss_fixed": 0.1707, "epoch": 0.6266792809839168, "grad_norm": 0.20917080342769623, "learning_rate": 3.2605089227206124e-06, "loss": 1.1121, "loss_dynamic": 1.3269, "loss_fixed": 0.8973, "step": 414 }, { "eob_count_fixed": 23219, "eob_loss_fixed": 0.1722, "epoch": 0.6281929990539262, "grad_norm": 0.1888306587934494, "learning_rate": 3.237555299464145e-06, "loss": 1.0989, "loss_dynamic": 1.2641, "loss_fixed": 0.9337, "step": 415 }, { "eob_count_fixed": 23207, "eob_loss_fixed": 0.1689, "epoch": 0.6297067171239357, "grad_norm": 0.20328998565673828, "learning_rate": 3.2146440110860338e-06, "loss": 1.1403, "loss_dynamic": 1.3718, "loss_fixed": 0.9089, "step": 416 }, { "eob_count_fixed": 23213, "eob_loss_fixed": 0.1689, "epoch": 0.6312204351939451, "grad_norm": 0.20456993579864502, "learning_rate": 3.1917756079278613e-06, "loss": 1.0566, "loss_dynamic": 1.2341, "loss_fixed": 0.8791, "step": 417 }, { "eob_count_fixed": 23177, "eob_loss_fixed": 0.1673, "epoch": 0.6327341532639545, "grad_norm": 0.2164163589477539, "learning_rate": 3.168950639301086e-06, "loss": 1.063, "loss_dynamic": 1.2427, "loss_fixed": 0.8833, "step": 418 }, { "eob_count_fixed": 23236, "eob_loss_fixed": 0.1803, "epoch": 0.6342478713339641, "grad_norm": 0.19798113405704498, "learning_rate": 3.1461696534738425e-06, "loss": 1.1126, "loss_dynamic": 1.3012, "loss_fixed": 0.924, "step": 419 }, { "eob_count_fixed": 22858, "eob_loss_fixed": 0.18, "epoch": 0.6357615894039735, "grad_norm": 0.20521534979343414, "learning_rate": 3.123433197657777e-06, "loss": 1.1377, "loss_dynamic": 1.3483, "loss_fixed": 0.927, "step": 420 }, { "eob_count_fixed": 23120, "eob_loss_fixed": 0.1833, "epoch": 0.637275307473983, "grad_norm": 0.19997207820415497, "learning_rate": 3.1007418179948954e-06, "loss": 1.1259, "loss_dynamic": 1.3265, "loss_fixed": 0.9253, "step": 421 }, { "eob_count_fixed": 23257, "eob_loss_fixed": 0.1768, "epoch": 0.6387890255439924, "grad_norm": 0.19501835107803345, "learning_rate": 3.078096059544461e-06, "loss": 1.0654, "loss_dynamic": 1.2145, "loss_fixed": 0.9162, "step": 422 }, { "eob_count_fixed": 23080, "eob_loss_fixed": 0.1684, "epoch": 0.6403027436140019, "grad_norm": 0.2246396243572235, "learning_rate": 3.055496466269878e-06, "loss": 1.1132, "loss_dynamic": 1.3076, "loss_fixed": 0.9189, "step": 423 }, { "eob_count_fixed": 23109, "eob_loss_fixed": 0.1693, "epoch": 0.6418164616840114, "grad_norm": 0.20476385951042175, "learning_rate": 3.0329435810256454e-06, "loss": 1.1054, "loss_dynamic": 1.3061, "loss_fixed": 0.9047, "step": 424 }, { "eob_count_fixed": 23207, "eob_loss_fixed": 0.1725, "epoch": 0.6433301797540208, "grad_norm": 0.20611892640590668, "learning_rate": 3.0104379455443096e-06, "loss": 1.0888, "loss_dynamic": 1.2637, "loss_fixed": 0.9139, "step": 425 }, { "eob_count_fixed": 23121, "eob_loss_fixed": 0.1756, "epoch": 0.6448438978240303, "grad_norm": 0.18951097130775452, "learning_rate": 2.98798010042345e-06, "loss": 1.0755, "loss_dynamic": 1.2427, "loss_fixed": 0.9083, "step": 426 }, { "eob_count_fixed": 23180, "eob_loss_fixed": 0.1845, "epoch": 0.6463576158940397, "grad_norm": 0.19419370591640472, "learning_rate": 2.965570585112696e-06, "loss": 1.1052, "loss_dynamic": 1.2935, "loss_fixed": 0.9169, "step": 427 }, { "eob_count_fixed": 23198, "eob_loss_fixed": 0.1828, "epoch": 0.6478713339640492, "grad_norm": 0.18645638227462769, "learning_rate": 2.94320993790077e-06, "loss": 1.0646, "loss_dynamic": 1.2104, "loss_fixed": 0.9188, "step": 428 }, { "eob_count_fixed": 23184, "eob_loss_fixed": 0.1664, "epoch": 0.6493850520340586, "grad_norm": 0.20874756574630737, "learning_rate": 2.9208986959025567e-06, "loss": 1.0369, "loss_dynamic": 1.1969, "loss_fixed": 0.877, "step": 429 }, { "eob_count_fixed": 23102, "eob_loss_fixed": 0.1712, "epoch": 0.6508987701040682, "grad_norm": 0.19865557551383972, "learning_rate": 2.8986373950461945e-06, "loss": 1.0486, "loss_dynamic": 1.2158, "loss_fixed": 0.8814, "step": 430 }, { "eob_count_fixed": 23056, "eob_loss_fixed": 0.1687, "epoch": 0.6524124881740776, "grad_norm": 0.20677290856838226, "learning_rate": 2.876426570060218e-06, "loss": 1.0658, "loss_dynamic": 1.2174, "loss_fixed": 0.9142, "step": 431 }, { "eob_count_fixed": 23076, "eob_loss_fixed": 0.1694, "epoch": 0.653926206244087, "grad_norm": 0.21170108020305634, "learning_rate": 2.8542667544606985e-06, "loss": 1.0858, "loss_dynamic": 1.2873, "loss_fixed": 0.8844, "step": 432 }, { "eob_count_fixed": 23189, "eob_loss_fixed": 0.1799, "epoch": 0.6554399243140965, "grad_norm": 0.2079583704471588, "learning_rate": 2.8321584805384338e-06, "loss": 1.0759, "loss_dynamic": 1.2426, "loss_fixed": 0.9092, "step": 433 }, { "eob_count_fixed": 23121, "eob_loss_fixed": 0.1814, "epoch": 0.6569536423841059, "grad_norm": 0.19858893752098083, "learning_rate": 2.8101022793461653e-06, "loss": 1.1037, "loss_dynamic": 1.2875, "loss_fixed": 0.9199, "step": 434 }, { "eob_count_fixed": 23046, "eob_loss_fixed": 0.1689, "epoch": 0.6584673604541155, "grad_norm": 0.19992844760417938, "learning_rate": 2.788098680685816e-06, "loss": 1.0673, "loss_dynamic": 1.2512, "loss_fixed": 0.8833, "step": 435 }, { "eob_count_fixed": 23056, "eob_loss_fixed": 0.1728, "epoch": 0.6599810785241249, "grad_norm": 0.21476487815380096, "learning_rate": 2.7661482130957697e-06, "loss": 1.0977, "loss_dynamic": 1.3044, "loss_fixed": 0.8909, "step": 436 }, { "eob_count_fixed": 23150, "eob_loss_fixed": 0.1759, "epoch": 0.6614947965941343, "grad_norm": 0.1949796825647354, "learning_rate": 2.7442514038381735e-06, "loss": 1.0483, "loss_dynamic": 1.2207, "loss_fixed": 0.8759, "step": 437 }, { "eob_count_fixed": 23172, "eob_loss_fixed": 0.1727, "epoch": 0.6630085146641438, "grad_norm": 0.19797374308109283, "learning_rate": 2.7224087788862707e-06, "loss": 1.1228, "loss_dynamic": 1.3244, "loss_fixed": 0.9212, "step": 438 }, { "eob_count_fixed": 23085, "eob_loss_fixed": 0.1741, "epoch": 0.6645222327341532, "grad_norm": 0.20930786430835724, "learning_rate": 2.700620862911768e-06, "loss": 1.0956, "loss_dynamic": 1.3057, "loss_fixed": 0.8854, "step": 439 }, { "eob_count_fixed": 23050, "eob_loss_fixed": 0.1754, "epoch": 0.6660359508041628, "grad_norm": 0.19536089897155762, "learning_rate": 2.678888179272232e-06, "loss": 1.103, "loss_dynamic": 1.3086, "loss_fixed": 0.8973, "step": 440 }, { "eob_count_fixed": 23230, "eob_loss_fixed": 0.174, "epoch": 0.6675496688741722, "grad_norm": 0.19721335172653198, "learning_rate": 2.6572112499985254e-06, "loss": 1.0944, "loss_dynamic": 1.2753, "loss_fixed": 0.9135, "step": 441 }, { "eob_count_fixed": 23122, "eob_loss_fixed": 0.1717, "epoch": 0.6690633869441817, "grad_norm": 0.2105029672384262, "learning_rate": 2.6355905957822537e-06, "loss": 1.0837, "loss_dynamic": 1.2701, "loss_fixed": 0.8973, "step": 442 }, { "eob_count_fixed": 23107, "eob_loss_fixed": 0.1793, "epoch": 0.6705771050141911, "grad_norm": 0.20162716507911682, "learning_rate": 2.614026735963262e-06, "loss": 1.1226, "loss_dynamic": 1.332, "loss_fixed": 0.9131, "step": 443 }, { "eob_count_fixed": 23030, "eob_loss_fixed": 0.1697, "epoch": 0.6720908230842005, "grad_norm": 0.2108498215675354, "learning_rate": 2.5925201885171735e-06, "loss": 1.0969, "loss_dynamic": 1.2941, "loss_fixed": 0.8997, "step": 444 }, { "eob_count_fixed": 23079, "eob_loss_fixed": 0.1779, "epoch": 0.67360454115421, "grad_norm": 0.20946940779685974, "learning_rate": 2.571071470042929e-06, "loss": 1.1251, "loss_dynamic": 1.3044, "loss_fixed": 0.9458, "step": 445 }, { "eob_count_fixed": 23238, "eob_loss_fixed": 0.1698, "epoch": 0.6751182592242195, "grad_norm": 0.2010679692029953, "learning_rate": 2.549681095750389e-06, "loss": 1.0352, "loss_dynamic": 1.175, "loss_fixed": 0.8955, "step": 446 }, { "eob_count_fixed": 23206, "eob_loss_fixed": 0.1805, "epoch": 0.676631977294229, "grad_norm": 0.21623045206069946, "learning_rate": 2.528349579447956e-06, "loss": 1.091, "loss_dynamic": 1.2611, "loss_fixed": 0.9209, "step": 447 }, { "eob_count_fixed": 23097, "eob_loss_fixed": 0.1779, "epoch": 0.6781456953642384, "grad_norm": 0.1949029564857483, "learning_rate": 2.5070774335302285e-06, "loss": 1.0571, "loss_dynamic": 1.201, "loss_fixed": 0.9131, "step": 448 }, { "eob_count_fixed": 22983, "eob_loss_fixed": 0.1718, "epoch": 0.6796594134342478, "grad_norm": 0.2259640395641327, "learning_rate": 2.4858651689656955e-06, "loss": 1.0904, "loss_dynamic": 1.2895, "loss_fixed": 0.8914, "step": 449 }, { "eob_count_fixed": 23180, "eob_loss_fixed": 0.1738, "epoch": 0.6811731315042573, "grad_norm": 0.20354484021663666, "learning_rate": 2.4647132952844717e-06, "loss": 1.1196, "loss_dynamic": 1.326, "loss_fixed": 0.9132, "step": 450 }, { "eob_count_fixed": 23046, "eob_loss_fixed": 0.1799, "epoch": 0.6826868495742668, "grad_norm": 0.22074927389621735, "learning_rate": 2.4436223205660427e-06, "loss": 1.1568, "loss_dynamic": 1.38, "loss_fixed": 0.9337, "step": 451 }, { "eob_count_fixed": 23140, "eob_loss_fixed": 0.1808, "epoch": 0.6842005676442763, "grad_norm": 0.19707825779914856, "learning_rate": 2.422592751427066e-06, "loss": 1.1271, "loss_dynamic": 1.3442, "loss_fixed": 0.9099, "step": 452 }, { "eob_count_fixed": 23180, "eob_loss_fixed": 0.1718, "epoch": 0.6857142857142857, "grad_norm": 0.20337405800819397, "learning_rate": 2.4016250930092063e-06, "loss": 1.1112, "loss_dynamic": 1.3076, "loss_fixed": 0.9147, "step": 453 }, { "eob_count_fixed": 23129, "eob_loss_fixed": 0.171, "epoch": 0.6872280037842952, "grad_norm": 0.21362006664276123, "learning_rate": 2.3807198489670045e-06, "loss": 1.098, "loss_dynamic": 1.2708, "loss_fixed": 0.9252, "step": 454 }, { "eob_count_fixed": 23271, "eob_loss_fixed": 0.1649, "epoch": 0.6887417218543046, "grad_norm": 0.2075643390417099, "learning_rate": 2.3598775214557682e-06, "loss": 1.0728, "loss_dynamic": 1.2837, "loss_fixed": 0.8619, "step": 455 }, { "eob_count_fixed": 23013, "eob_loss_fixed": 0.1725, "epoch": 0.6902554399243142, "grad_norm": 0.216401144862175, "learning_rate": 2.3390986111195175e-06, "loss": 1.1337, "loss_dynamic": 1.3549, "loss_fixed": 0.9126, "step": 456 }, { "eob_count_fixed": 23058, "eob_loss_fixed": 0.1719, "epoch": 0.6917691579943236, "grad_norm": 0.22068504989147186, "learning_rate": 2.3183836170789565e-06, "loss": 1.1324, "loss_dynamic": 1.3644, "loss_fixed": 0.9003, "step": 457 }, { "eob_count_fixed": 23278, "eob_loss_fixed": 0.1815, "epoch": 0.693282876064333, "grad_norm": 0.1986350268125534, "learning_rate": 2.2977330369194865e-06, "loss": 1.0739, "loss_dynamic": 1.2351, "loss_fixed": 0.9128, "step": 458 }, { "eob_count_fixed": 23124, "eob_loss_fixed": 0.1679, "epoch": 0.6947965941343425, "grad_norm": 0.19748318195343018, "learning_rate": 2.27714736667925e-06, "loss": 1.105, "loss_dynamic": 1.2926, "loss_fixed": 0.9174, "step": 459 }, { "eob_count_fixed": 23075, "eob_loss_fixed": 0.1718, "epoch": 0.6963103122043519, "grad_norm": 0.19933710992336273, "learning_rate": 2.2566271008372258e-06, "loss": 1.086, "loss_dynamic": 1.2755, "loss_fixed": 0.8966, "step": 460 }, { "eob_count_fixed": 23043, "eob_loss_fixed": 0.1799, "epoch": 0.6978240302743614, "grad_norm": 0.20548279583454132, "learning_rate": 2.236172732301332e-06, "loss": 1.1544, "loss_dynamic": 1.3935, "loss_fixed": 0.9153, "step": 461 }, { "eob_count_fixed": 23092, "eob_loss_fixed": 0.1742, "epoch": 0.6993377483443709, "grad_norm": 0.1929888278245926, "learning_rate": 2.2157847523966043e-06, "loss": 1.1063, "loss_dynamic": 1.3148, "loss_fixed": 0.8978, "step": 462 }, { "eob_count_fixed": 23069, "eob_loss_fixed": 0.1628, "epoch": 0.7008514664143803, "grad_norm": 0.21323232352733612, "learning_rate": 2.1954636508533887e-06, "loss": 1.1145, "loss_dynamic": 1.3281, "loss_fixed": 0.9009, "step": 463 }, { "eob_count_fixed": 23147, "eob_loss_fixed": 0.1699, "epoch": 0.7023651844843898, "grad_norm": 0.19127249717712402, "learning_rate": 2.1752099157955736e-06, "loss": 1.0152, "loss_dynamic": 1.1582, "loss_fixed": 0.8723, "step": 464 }, { "eob_count_fixed": 23199, "eob_loss_fixed": 0.1799, "epoch": 0.7038789025543992, "grad_norm": 0.2166828215122223, "learning_rate": 2.1550240337288684e-06, "loss": 1.1259, "loss_dynamic": 1.3426, "loss_fixed": 0.9091, "step": 465 }, { "eob_count_fixed": 23339, "eob_loss_fixed": 0.1669, "epoch": 0.7053926206244087, "grad_norm": 0.2086198478937149, "learning_rate": 2.1349064895291176e-06, "loss": 1.0958, "loss_dynamic": 1.2837, "loss_fixed": 0.908, "step": 466 }, { "eob_count_fixed": 23163, "eob_loss_fixed": 0.1652, "epoch": 0.7069063386944182, "grad_norm": 0.19981300830841064, "learning_rate": 2.1148577664306525e-06, "loss": 1.0938, "loss_dynamic": 1.2937, "loss_fixed": 0.894, "step": 467 }, { "eob_count_fixed": 23133, "eob_loss_fixed": 0.1881, "epoch": 0.7084200567644277, "grad_norm": 0.21530763804912567, "learning_rate": 2.094878346014684e-06, "loss": 1.1541, "loss_dynamic": 1.3874, "loss_fixed": 0.9209, "step": 468 }, { "eob_count_fixed": 23004, "eob_loss_fixed": 0.1807, "epoch": 0.7099337748344371, "grad_norm": 0.2075398713350296, "learning_rate": 2.074968708197734e-06, "loss": 1.1622, "loss_dynamic": 1.3849, "loss_fixed": 0.9395, "step": 469 }, { "eob_count_fixed": 23120, "eob_loss_fixed": 0.1645, "epoch": 0.7114474929044465, "grad_norm": 0.21184225380420685, "learning_rate": 2.0551293312201144e-06, "loss": 1.061, "loss_dynamic": 1.2168, "loss_fixed": 0.9052, "step": 470 }, { "eob_count_fixed": 23095, "eob_loss_fixed": 0.1676, "epoch": 0.712961210974456, "grad_norm": 0.20390024781227112, "learning_rate": 2.0353606916344237e-06, "loss": 1.1171, "loss_dynamic": 1.3326, "loss_fixed": 0.9016, "step": 471 }, { "eob_count_fixed": 23155, "eob_loss_fixed": 0.1664, "epoch": 0.7144749290444655, "grad_norm": 0.20452170073986053, "learning_rate": 2.015663264294116e-06, "loss": 1.0949, "loss_dynamic": 1.2867, "loss_fixed": 0.9031, "step": 472 }, { "eob_count_fixed": 23082, "eob_loss_fixed": 0.1795, "epoch": 0.715988647114475, "grad_norm": 0.2253059297800064, "learning_rate": 1.9960375223420903e-06, "loss": 1.112, "loss_dynamic": 1.3079, "loss_fixed": 0.916, "step": 473 }, { "eob_count_fixed": 23183, "eob_loss_fixed": 0.1833, "epoch": 0.7175023651844844, "grad_norm": 0.2100057601928711, "learning_rate": 1.97648393719932e-06, "loss": 1.0874, "loss_dynamic": 1.2591, "loss_fixed": 0.9158, "step": 474 }, { "eob_count_fixed": 23111, "eob_loss_fixed": 0.1811, "epoch": 0.7190160832544938, "grad_norm": 0.21353377401828766, "learning_rate": 1.957002978553533e-06, "loss": 1.1184, "loss_dynamic": 1.3361, "loss_fixed": 0.9007, "step": 475 }, { "eob_count_fixed": 23251, "eob_loss_fixed": 0.1764, "epoch": 0.7205298013245033, "grad_norm": 0.1881416290998459, "learning_rate": 1.9375951143479312e-06, "loss": 1.0332, "loss_dynamic": 1.1937, "loss_fixed": 0.8726, "step": 476 }, { "eob_count_fixed": 23172, "eob_loss_fixed": 0.1817, "epoch": 0.7220435193945127, "grad_norm": 0.20165178179740906, "learning_rate": 1.9182608107699464e-06, "loss": 1.0739, "loss_dynamic": 1.2503, "loss_fixed": 0.8974, "step": 477 }, { "eob_count_fixed": 23145, "eob_loss_fixed": 0.1745, "epoch": 0.7235572374645223, "grad_norm": 0.20751385390758514, "learning_rate": 1.899000532240044e-06, "loss": 1.1006, "loss_dynamic": 1.2942, "loss_fixed": 0.9069, "step": 478 }, { "eob_count_fixed": 23179, "eob_loss_fixed": 0.1708, "epoch": 0.7250709555345317, "grad_norm": 0.20481151342391968, "learning_rate": 1.8798147414005742e-06, "loss": 1.0926, "loss_dynamic": 1.2702, "loss_fixed": 0.915, "step": 479 }, { "eob_count_fixed": 23141, "eob_loss_fixed": 0.1764, "epoch": 0.7265846736045412, "grad_norm": 0.21131667494773865, "learning_rate": 1.860703899104641e-06, "loss": 1.1197, "loss_dynamic": 1.3365, "loss_fixed": 0.903, "step": 480 }, { "eob_count_fixed": 23323, "eob_loss_fixed": 0.1769, "epoch": 0.7280983916745506, "grad_norm": 0.184462308883667, "learning_rate": 1.8416684644050497e-06, "loss": 1.0833, "loss_dynamic": 1.2697, "loss_fixed": 0.8968, "step": 481 }, { "eob_count_fixed": 23143, "eob_loss_fixed": 0.1689, "epoch": 0.72961210974456, "grad_norm": 0.19851557910442352, "learning_rate": 1.8227088945432787e-06, "loss": 1.0952, "loss_dynamic": 1.3056, "loss_fixed": 0.8849, "step": 482 }, { "eob_count_fixed": 23166, "eob_loss_fixed": 0.168, "epoch": 0.7311258278145696, "grad_norm": 0.20582623779773712, "learning_rate": 1.803825644938484e-06, "loss": 1.051, "loss_dynamic": 1.2217, "loss_fixed": 0.8803, "step": 483 }, { "eob_count_fixed": 22897, "eob_loss_fixed": 0.1699, "epoch": 0.732639545884579, "grad_norm": 0.19415777921676636, "learning_rate": 1.7850191691765734e-06, "loss": 1.0725, "loss_dynamic": 1.2503, "loss_fixed": 0.8947, "step": 484 }, { "eob_count_fixed": 23059, "eob_loss_fixed": 0.1601, "epoch": 0.7341532639545885, "grad_norm": 0.21426227688789368, "learning_rate": 1.7662899189992977e-06, "loss": 1.0752, "loss_dynamic": 1.2765, "loss_fixed": 0.874, "step": 485 }, { "eob_count_fixed": 23026, "eob_loss_fixed": 0.1683, "epoch": 0.7356669820245979, "grad_norm": 0.21013692021369934, "learning_rate": 1.7476383442934175e-06, "loss": 1.1002, "loss_dynamic": 1.2977, "loss_fixed": 0.9028, "step": 486 }, { "eob_count_fixed": 22987, "eob_loss_fixed": 0.1715, "epoch": 0.7371807000946073, "grad_norm": 0.21278217434883118, "learning_rate": 1.7290648930798791e-06, "loss": 1.1185, "loss_dynamic": 1.3294, "loss_fixed": 0.9077, "step": 487 }, { "eob_count_fixed": 23066, "eob_loss_fixed": 0.1773, "epoch": 0.7386944181646169, "grad_norm": 0.20638079941272736, "learning_rate": 1.7105700115030632e-06, "loss": 1.139, "loss_dynamic": 1.3642, "loss_fixed": 0.9138, "step": 488 }, { "eob_count_fixed": 23117, "eob_loss_fixed": 0.163, "epoch": 0.7402081362346263, "grad_norm": 0.19844624400138855, "learning_rate": 1.6921541438200633e-06, "loss": 1.0743, "loss_dynamic": 1.2655, "loss_fixed": 0.8831, "step": 489 }, { "eob_count_fixed": 23233, "eob_loss_fixed": 0.1741, "epoch": 0.7417218543046358, "grad_norm": 0.21111704409122467, "learning_rate": 1.6738177323900173e-06, "loss": 1.1125, "loss_dynamic": 1.3279, "loss_fixed": 0.8971, "step": 490 }, { "eob_count_fixed": 23009, "eob_loss_fixed": 0.1792, "epoch": 0.7432355723746452, "grad_norm": 0.20069438219070435, "learning_rate": 1.6555612176634795e-06, "loss": 1.1319, "loss_dynamic": 1.3384, "loss_fixed": 0.9254, "step": 491 }, { "eob_count_fixed": 23080, "eob_loss_fixed": 0.1775, "epoch": 0.7447492904446547, "grad_norm": 0.19551318883895874, "learning_rate": 1.6373850381718464e-06, "loss": 1.1041, "loss_dynamic": 1.2839, "loss_fixed": 0.9243, "step": 492 }, { "eob_count_fixed": 23208, "eob_loss_fixed": 0.1731, "epoch": 0.7462630085146641, "grad_norm": 0.21083058416843414, "learning_rate": 1.6192896305168154e-06, "loss": 1.1268, "loss_dynamic": 1.3486, "loss_fixed": 0.9051, "step": 493 }, { "eob_count_fixed": 23089, "eob_loss_fixed": 0.1706, "epoch": 0.7477767265846736, "grad_norm": 0.1927291750907898, "learning_rate": 1.6012754293598953e-06, "loss": 1.0595, "loss_dynamic": 1.2319, "loss_fixed": 0.8872, "step": 494 }, { "eob_count_fixed": 23199, "eob_loss_fixed": 0.1845, "epoch": 0.7492904446546831, "grad_norm": 0.1904156655073166, "learning_rate": 1.5833428674119804e-06, "loss": 1.039, "loss_dynamic": 1.1526, "loss_fixed": 0.9254, "step": 495 }, { "eob_count_fixed": 23033, "eob_loss_fixed": 0.1681, "epoch": 0.7508041627246925, "grad_norm": 0.20740769803524017, "learning_rate": 1.5654923754229422e-06, "loss": 1.0826, "loss_dynamic": 1.274, "loss_fixed": 0.8912, "step": 496 }, { "eob_count_fixed": 23243, "eob_loss_fixed": 0.1631, "epoch": 0.752317880794702, "grad_norm": 0.20365136861801147, "learning_rate": 1.5477243821712861e-06, "loss": 1.074, "loss_dynamic": 1.2656, "loss_fixed": 0.8825, "step": 497 }, { "eob_count_fixed": 23024, "eob_loss_fixed": 0.1684, "epoch": 0.7538315988647114, "grad_norm": 0.20416443049907684, "learning_rate": 1.5300393144538556e-06, "loss": 1.1322, "loss_dynamic": 1.3578, "loss_fixed": 0.9066, "step": 498 }, { "eob_count_fixed": 23169, "eob_loss_fixed": 0.1746, "epoch": 0.755345316934721, "grad_norm": 0.22976593673229218, "learning_rate": 1.5124375970755755e-06, "loss": 1.1092, "loss_dynamic": 1.3302, "loss_fixed": 0.8882, "step": 499 }, { "eob_count_fixed": 23140, "eob_loss_fixed": 0.1788, "epoch": 0.7568590350047304, "grad_norm": 0.21354855597019196, "learning_rate": 1.4949196528392523e-06, "loss": 1.1257, "loss_dynamic": 1.3472, "loss_fixed": 0.9042, "step": 500 }, { "eob_count_fixed": 22998, "eob_loss_fixed": 0.1708, "epoch": 0.7583727530747398, "grad_norm": 0.19752137362957, "learning_rate": 1.4774859025354133e-06, "loss": 1.0622, "loss_dynamic": 1.2346, "loss_fixed": 0.8898, "step": 501 }, { "eob_count_fixed": 23226, "eob_loss_fixed": 0.177, "epoch": 0.7598864711447493, "grad_norm": 0.20483042299747467, "learning_rate": 1.4601367649322084e-06, "loss": 1.1203, "loss_dynamic": 1.3326, "loss_fixed": 0.9081, "step": 502 }, { "eob_count_fixed": 23282, "eob_loss_fixed": 0.173, "epoch": 0.7614001892147587, "grad_norm": 0.19694310426712036, "learning_rate": 1.442872656765335e-06, "loss": 1.0851, "loss_dynamic": 1.2705, "loss_fixed": 0.8998, "step": 503 }, { "eob_count_fixed": 23040, "eob_loss_fixed": 0.1682, "epoch": 0.7629139072847683, "grad_norm": 0.20435911417007446, "learning_rate": 1.4256939927280428e-06, "loss": 1.1509, "loss_dynamic": 1.4041, "loss_fixed": 0.8977, "step": 504 }, { "eob_count_fixed": 23126, "eob_loss_fixed": 0.1738, "epoch": 0.7644276253547777, "grad_norm": 0.21666571497917175, "learning_rate": 1.4086011854611691e-06, "loss": 1.1127, "loss_dynamic": 1.3176, "loss_fixed": 0.9078, "step": 505 }, { "eob_count_fixed": 23154, "eob_loss_fixed": 0.1665, "epoch": 0.7659413434247871, "grad_norm": 0.19174206256866455, "learning_rate": 1.3915946455432216e-06, "loss": 1.0916, "loss_dynamic": 1.2707, "loss_fixed": 0.9124, "step": 506 }, { "eob_count_fixed": 23218, "eob_loss_fixed": 0.1685, "epoch": 0.7674550614947966, "grad_norm": 0.20413535833358765, "learning_rate": 1.3746747814805189e-06, "loss": 1.1235, "loss_dynamic": 1.3272, "loss_fixed": 0.9197, "step": 507 }, { "eob_count_fixed": 23135, "eob_loss_fixed": 0.1752, "epoch": 0.768968779564806, "grad_norm": 0.19877751171588898, "learning_rate": 1.3578419996973808e-06, "loss": 1.08, "loss_dynamic": 1.2642, "loss_fixed": 0.8958, "step": 508 }, { "eob_count_fixed": 23091, "eob_loss_fixed": 0.1755, "epoch": 0.7704824976348155, "grad_norm": 0.2044394612312317, "learning_rate": 1.3410967045263622e-06, "loss": 1.1126, "loss_dynamic": 1.3288, "loss_fixed": 0.8964, "step": 509 }, { "eob_count_fixed": 23037, "eob_loss_fixed": 0.175, "epoch": 0.771996215704825, "grad_norm": 0.20264782011508942, "learning_rate": 1.3244392981985416e-06, "loss": 1.1466, "loss_dynamic": 1.3815, "loss_fixed": 0.9117, "step": 510 }, { "eob_count_fixed": 23091, "eob_loss_fixed": 0.1793, "epoch": 0.7735099337748345, "grad_norm": 0.19920742511749268, "learning_rate": 1.307870180833864e-06, "loss": 1.1234, "loss_dynamic": 1.3322, "loss_fixed": 0.9146, "step": 511 }, { "eob_count_fixed": 23222, "eob_loss_fixed": 0.1718, "epoch": 0.7750236518448439, "grad_norm": 0.19918370246887207, "learning_rate": 1.291389750431518e-06, "loss": 1.0627, "loss_dynamic": 1.2166, "loss_fixed": 0.9088, "step": 512 }, { "eob_count_fixed": 23177, "eob_loss_fixed": 0.1676, "epoch": 0.7765373699148533, "grad_norm": 0.1911431849002838, "learning_rate": 1.274998402860388e-06, "loss": 1.0591, "loss_dynamic": 1.2285, "loss_fixed": 0.8897, "step": 513 }, { "eob_count_fixed": 22944, "eob_loss_fixed": 0.1685, "epoch": 0.7780510879848628, "grad_norm": 0.20139557123184204, "learning_rate": 1.258696531849542e-06, "loss": 1.0689, "loss_dynamic": 1.2574, "loss_fixed": 0.8805, "step": 514 }, { "eob_count_fixed": 23194, "eob_loss_fixed": 0.18, "epoch": 0.7795648060548723, "grad_norm": 0.20752547681331635, "learning_rate": 1.242484528978769e-06, "loss": 1.125, "loss_dynamic": 1.356, "loss_fixed": 0.8939, "step": 515 }, { "eob_count_fixed": 23076, "eob_loss_fixed": 0.1747, "epoch": 0.7810785241248818, "grad_norm": 0.20313572883605957, "learning_rate": 1.2263627836691783e-06, "loss": 1.0897, "loss_dynamic": 1.2788, "loss_fixed": 0.9007, "step": 516 }, { "eob_count_fixed": 23001, "eob_loss_fixed": 0.1694, "epoch": 0.7825922421948912, "grad_norm": 0.22779418528079987, "learning_rate": 1.2103316831738426e-06, "loss": 1.116, "loss_dynamic": 1.3219, "loss_fixed": 0.9101, "step": 517 }, { "eob_count_fixed": 23296, "eob_loss_fixed": 0.1728, "epoch": 0.7841059602649006, "grad_norm": 0.20598873496055603, "learning_rate": 1.1943916125684974e-06, "loss": 1.091, "loss_dynamic": 1.2954, "loss_fixed": 0.8866, "step": 518 }, { "eob_count_fixed": 23197, "eob_loss_fixed": 0.1751, "epoch": 0.7856196783349101, "grad_norm": 0.18500293791294098, "learning_rate": 1.1785429547422912e-06, "loss": 1.0322, "loss_dynamic": 1.1604, "loss_fixed": 0.9041, "step": 519 }, { "eob_count_fixed": 23211, "eob_loss_fixed": 0.1774, "epoch": 0.7871333964049196, "grad_norm": 0.19441650807857513, "learning_rate": 1.1627860903885857e-06, "loss": 1.0684, "loss_dynamic": 1.2473, "loss_fixed": 0.8895, "step": 520 }, { "eob_count_fixed": 23057, "eob_loss_fixed": 0.1684, "epoch": 0.7886471144749291, "grad_norm": 0.2138061225414276, "learning_rate": 1.1471213979958202e-06, "loss": 1.1197, "loss_dynamic": 1.3377, "loss_fixed": 0.9016, "step": 521 }, { "eob_count_fixed": 23105, "eob_loss_fixed": 0.1735, "epoch": 0.7901608325449385, "grad_norm": 0.19494998455047607, "learning_rate": 1.1315492538384044e-06, "loss": 1.0697, "loss_dynamic": 1.2401, "loss_fixed": 0.8994, "step": 522 }, { "eob_count_fixed": 23176, "eob_loss_fixed": 0.1707, "epoch": 0.791674550614948, "grad_norm": 0.20065844058990479, "learning_rate": 1.1160700319676926e-06, "loss": 1.0968, "loss_dynamic": 1.2894, "loss_fixed": 0.9042, "step": 523 }, { "eob_count_fixed": 23022, "eob_loss_fixed": 0.1615, "epoch": 0.7931882686849574, "grad_norm": 0.19735577702522278, "learning_rate": 1.1006841042030004e-06, "loss": 1.0903, "loss_dynamic": 1.2955, "loss_fixed": 0.8851, "step": 524 }, { "eob_count_fixed": 23182, "eob_loss_fixed": 0.178, "epoch": 0.7947019867549668, "grad_norm": 0.20147641003131866, "learning_rate": 1.085391840122662e-06, "loss": 1.1301, "loss_dynamic": 1.3465, "loss_fixed": 0.9138, "step": 525 }, { "eob_count_fixed": 23109, "eob_loss_fixed": 0.18, "epoch": 0.7962157048249764, "grad_norm": 0.21134053170681, "learning_rate": 1.070193607055161e-06, "loss": 1.0975, "loss_dynamic": 1.2971, "loss_fixed": 0.8978, "step": 526 }, { "eob_count_fixed": 23074, "eob_loss_fixed": 0.1701, "epoch": 0.7977294228949858, "grad_norm": 0.19800984859466553, "learning_rate": 1.055089770070306e-06, "loss": 1.0635, "loss_dynamic": 1.2423, "loss_fixed": 0.8847, "step": 527 }, { "eob_count_fixed": 23202, "eob_loss_fixed": 0.1692, "epoch": 0.7992431409649953, "grad_norm": 0.18660441040992737, "learning_rate": 1.0400806919704591e-06, "loss": 1.051, "loss_dynamic": 1.2214, "loss_fixed": 0.8807, "step": 528 }, { "eob_count_fixed": 23121, "eob_loss_fixed": 0.1657, "epoch": 0.8007568590350047, "grad_norm": 0.19741471111774445, "learning_rate": 1.0251667332818215e-06, "loss": 1.0608, "loss_dynamic": 1.2321, "loss_fixed": 0.8896, "step": 529 }, { "eob_count_fixed": 23020, "eob_loss_fixed": 0.1705, "epoch": 0.8022705771050141, "grad_norm": 0.20535947382450104, "learning_rate": 1.0103482522457814e-06, "loss": 1.1214, "loss_dynamic": 1.328, "loss_fixed": 0.9148, "step": 530 }, { "eob_count_fixed": 23018, "eob_loss_fixed": 0.1704, "epoch": 0.8037842951750237, "grad_norm": 0.20263829827308655, "learning_rate": 9.956256048102914e-07, "loss": 1.112, "loss_dynamic": 1.3256, "loss_fixed": 0.8983, "step": 531 }, { "eob_count_fixed": 23018, "eob_loss_fixed": 0.1684, "epoch": 0.8052980132450331, "grad_norm": 0.2210431694984436, "learning_rate": 9.809991446213358e-07, "loss": 1.1462, "loss_dynamic": 1.3944, "loss_fixed": 0.898, "step": 532 }, { "eob_count_fixed": 23023, "eob_loss_fixed": 0.1647, "epoch": 0.8068117313150426, "grad_norm": 0.2001654952764511, "learning_rate": 9.664692230144267e-07, "loss": 1.0548, "loss_dynamic": 1.2374, "loss_fixed": 0.8721, "step": 533 }, { "eob_count_fixed": 23105, "eob_loss_fixed": 0.169, "epoch": 0.808325449385052, "grad_norm": 0.2072724997997284, "learning_rate": 9.520361890061708e-07, "loss": 1.1394, "loss_dynamic": 1.3659, "loss_fixed": 0.9129, "step": 534 }, { "eob_count_fixed": 23175, "eob_loss_fixed": 0.1718, "epoch": 0.8098391674550615, "grad_norm": 0.20203897356987, "learning_rate": 9.377003892858788e-07, "loss": 1.0868, "loss_dynamic": 1.2777, "loss_fixed": 0.8958, "step": 535 }, { "eob_count_fixed": 23157, "eob_loss_fixed": 0.1727, "epoch": 0.811352885525071, "grad_norm": 0.2001199573278427, "learning_rate": 9.23462168207237e-07, "loss": 1.0865, "loss_dynamic": 1.2772, "loss_fixed": 0.8957, "step": 536 }, { "eob_count_fixed": 23089, "eob_loss_fixed": 0.1835, "epoch": 0.8128666035950805, "grad_norm": 0.1988053023815155, "learning_rate": 9.093218677800486e-07, "loss": 1.1082, "loss_dynamic": 1.3052, "loss_fixed": 0.9112, "step": 537 }, { "eob_count_fixed": 23071, "eob_loss_fixed": 0.1699, "epoch": 0.8143803216650899, "grad_norm": 0.20419570803642273, "learning_rate": 8.952798276620039e-07, "loss": 1.1229, "loss_dynamic": 1.3475, "loss_fixed": 0.8983, "step": 538 }, { "eob_count_fixed": 23107, "eob_loss_fixed": 0.1769, "epoch": 0.8158940397350993, "grad_norm": 0.20018355548381805, "learning_rate": 8.813363851505285e-07, "loss": 1.076, "loss_dynamic": 1.2524, "loss_fixed": 0.8995, "step": 539 }, { "eob_count_fixed": 23011, "eob_loss_fixed": 0.1733, "epoch": 0.8174077578051088, "grad_norm": 0.20038960874080658, "learning_rate": 8.674918751746802e-07, "loss": 1.07, "loss_dynamic": 1.2646, "loss_fixed": 0.8753, "step": 540 }, { "eob_count_fixed": 23193, "eob_loss_fixed": 0.1744, "epoch": 0.8189214758751182, "grad_norm": 0.19085867702960968, "learning_rate": 8.537466302871028e-07, "loss": 1.0538, "loss_dynamic": 1.2128, "loss_fixed": 0.8947, "step": 541 }, { "eob_count_fixed": 23056, "eob_loss_fixed": 0.1719, "epoch": 0.8204351939451278, "grad_norm": 0.20804941654205322, "learning_rate": 8.401009806560384e-07, "loss": 1.0968, "loss_dynamic": 1.2964, "loss_fixed": 0.8972, "step": 542 }, { "eob_count_fixed": 23102, "eob_loss_fixed": 0.1793, "epoch": 0.8219489120151372, "grad_norm": 0.19972850382328033, "learning_rate": 8.265552540574001e-07, "loss": 1.1172, "loss_dynamic": 1.3316, "loss_fixed": 0.9029, "step": 543 }, { "eob_count_fixed": 23126, "eob_loss_fixed": 0.1749, "epoch": 0.8234626300851466, "grad_norm": 0.19851398468017578, "learning_rate": 8.131097758668944e-07, "loss": 1.0731, "loss_dynamic": 1.2552, "loss_fixed": 0.891, "step": 544 }, { "eob_count_fixed": 23019, "eob_loss_fixed": 0.1708, "epoch": 0.8249763481551561, "grad_norm": 0.1939672976732254, "learning_rate": 7.99764869052202e-07, "loss": 1.0784, "loss_dynamic": 1.2308, "loss_fixed": 0.926, "step": 545 }, { "eob_count_fixed": 23059, "eob_loss_fixed": 0.1756, "epoch": 0.8264900662251655, "grad_norm": 0.20951946079730988, "learning_rate": 7.865208541652287e-07, "loss": 1.1309, "loss_dynamic": 1.3608, "loss_fixed": 0.9011, "step": 546 }, { "eob_count_fixed": 23059, "eob_loss_fixed": 0.1623, "epoch": 0.8280037842951751, "grad_norm": 0.2065434753894806, "learning_rate": 7.733780493344034e-07, "loss": 1.0782, "loss_dynamic": 1.2686, "loss_fixed": 0.8879, "step": 547 }, { "eob_count_fixed": 23069, "eob_loss_fixed": 0.169, "epoch": 0.8295175023651845, "grad_norm": 0.2194949835538864, "learning_rate": 7.603367702570308e-07, "loss": 1.0833, "loss_dynamic": 1.2687, "loss_fixed": 0.898, "step": 548 }, { "eob_count_fixed": 22965, "eob_loss_fixed": 0.1741, "epoch": 0.831031220435194, "grad_norm": 0.2021845430135727, "learning_rate": 7.473973301917126e-07, "loss": 1.0853, "loss_dynamic": 1.2628, "loss_fixed": 0.9077, "step": 549 }, { "eob_count_fixed": 23077, "eob_loss_fixed": 0.1674, "epoch": 0.8325449385052034, "grad_norm": 0.20137476921081543, "learning_rate": 7.345600399508235e-07, "loss": 1.1138, "loss_dynamic": 1.3333, "loss_fixed": 0.8943, "step": 550 }, { "eob_count_fixed": 23055, "eob_loss_fixed": 0.1591, "epoch": 0.8340586565752128, "grad_norm": 0.20036745071411133, "learning_rate": 7.218252078930422e-07, "loss": 1.0779, "loss_dynamic": 1.2602, "loss_fixed": 0.8955, "step": 551 }, { "eob_count_fixed": 23055, "eob_loss_fixed": 0.1707, "epoch": 0.8355723746452224, "grad_norm": 0.2440427988767624, "learning_rate": 7.091931399159463e-07, "loss": 1.0625, "loss_dynamic": 1.233, "loss_fixed": 0.892, "step": 552 }, { "eob_count_fixed": 23093, "eob_loss_fixed": 0.179, "epoch": 0.8370860927152318, "grad_norm": 0.18559592962265015, "learning_rate": 6.966641394486695e-07, "loss": 1.098, "loss_dynamic": 1.2945, "loss_fixed": 0.9015, "step": 553 }, { "eob_count_fixed": 23026, "eob_loss_fixed": 0.1758, "epoch": 0.8385998107852413, "grad_norm": 0.20346888899803162, "learning_rate": 6.842385074445995e-07, "loss": 1.0689, "loss_dynamic": 1.2326, "loss_fixed": 0.9052, "step": 554 }, { "eob_count_fixed": 23041, "eob_loss_fixed": 0.1778, "epoch": 0.8401135288552507, "grad_norm": 0.19728370010852814, "learning_rate": 6.719165423741614e-07, "loss": 1.0911, "loss_dynamic": 1.2853, "loss_fixed": 0.8969, "step": 555 }, { "eob_count_fixed": 23118, "eob_loss_fixed": 0.1657, "epoch": 0.8416272469252601, "grad_norm": 0.20472094416618347, "learning_rate": 6.596985402176476e-07, "loss": 1.1381, "loss_dynamic": 1.3654, "loss_fixed": 0.9107, "step": 556 }, { "eob_count_fixed": 23069, "eob_loss_fixed": 0.1689, "epoch": 0.8431409649952696, "grad_norm": 0.20042504370212555, "learning_rate": 6.475847944580998e-07, "loss": 1.0366, "loss_dynamic": 1.1833, "loss_fixed": 0.8899, "step": 557 }, { "eob_count_fixed": 22992, "eob_loss_fixed": 0.1687, "epoch": 0.8446546830652791, "grad_norm": 0.18335440754890442, "learning_rate": 6.355755960742676e-07, "loss": 1.0659, "loss_dynamic": 1.2537, "loss_fixed": 0.878, "step": 558 }, { "eob_count_fixed": 23185, "eob_loss_fixed": 0.1654, "epoch": 0.8461684011352886, "grad_norm": 0.2045193314552307, "learning_rate": 6.236712335336132e-07, "loss": 1.0336, "loss_dynamic": 1.2016, "loss_fixed": 0.8656, "step": 559 }, { "eob_count_fixed": 23036, "eob_loss_fixed": 0.1719, "epoch": 0.847682119205298, "grad_norm": 0.18460145592689514, "learning_rate": 6.118719927853878e-07, "loss": 1.0922, "loss_dynamic": 1.2939, "loss_fixed": 0.8905, "step": 560 }, { "eob_count_fixed": 23023, "eob_loss_fixed": 0.1721, "epoch": 0.8491958372753075, "grad_norm": 0.19638977944850922, "learning_rate": 6.001781572537585e-07, "loss": 1.0874, "loss_dynamic": 1.2779, "loss_fixed": 0.8968, "step": 561 }, { "eob_count_fixed": 23186, "eob_loss_fixed": 0.1703, "epoch": 0.8507095553453169, "grad_norm": 0.18931467831134796, "learning_rate": 5.885900078310014e-07, "loss": 1.0723, "loss_dynamic": 1.2435, "loss_fixed": 0.9012, "step": 562 }, { "eob_count_fixed": 22994, "eob_loss_fixed": 0.1786, "epoch": 0.8522232734153264, "grad_norm": 0.18982499837875366, "learning_rate": 5.771078228707566e-07, "loss": 1.0592, "loss_dynamic": 1.2099, "loss_fixed": 0.9084, "step": 563 }, { "eob_count_fixed": 23147, "eob_loss_fixed": 0.1732, "epoch": 0.8537369914853359, "grad_norm": 0.19139231741428375, "learning_rate": 5.657318781813387e-07, "loss": 1.1357, "loss_dynamic": 1.3677, "loss_fixed": 0.9037, "step": 564 }, { "eob_count_fixed": 23248, "eob_loss_fixed": 0.1732, "epoch": 0.8552507095553453, "grad_norm": 0.20967084169387817, "learning_rate": 5.544624470191129e-07, "loss": 1.1493, "loss_dynamic": 1.3931, "loss_fixed": 0.9055, "step": 565 }, { "eob_count_fixed": 22968, "eob_loss_fixed": 0.1759, "epoch": 0.8567644276253548, "grad_norm": 0.20143187046051025, "learning_rate": 5.43299800081935e-07, "loss": 1.0854, "loss_dynamic": 1.2706, "loss_fixed": 0.9002, "step": 566 }, { "eob_count_fixed": 23056, "eob_loss_fixed": 0.1721, "epoch": 0.8582781456953642, "grad_norm": 0.19537176191806793, "learning_rate": 5.322442055026428e-07, "loss": 1.079, "loss_dynamic": 1.25, "loss_fixed": 0.908, "step": 567 }, { "eob_count_fixed": 23239, "eob_loss_fixed": 0.1719, "epoch": 0.8597918637653738, "grad_norm": 0.20242100954055786, "learning_rate": 5.212959288426189e-07, "loss": 1.0925, "loss_dynamic": 1.2938, "loss_fixed": 0.8911, "step": 568 }, { "eob_count_fixed": 23218, "eob_loss_fixed": 0.1722, "epoch": 0.8613055818353832, "grad_norm": 0.19811059534549713, "learning_rate": 5.104552330854113e-07, "loss": 1.1154, "loss_dynamic": 1.3409, "loss_fixed": 0.8898, "step": 569 }, { "eob_count_fixed": 23084, "eob_loss_fixed": 0.168, "epoch": 0.8628192999053926, "grad_norm": 0.19515381753444672, "learning_rate": 4.997223786304156e-07, "loss": 1.0681, "loss_dynamic": 1.255, "loss_fixed": 0.8813, "step": 570 }, { "eob_count_fixed": 22997, "eob_loss_fixed": 0.1678, "epoch": 0.8643330179754021, "grad_norm": 0.20603272318840027, "learning_rate": 4.890976232866218e-07, "loss": 1.1496, "loss_dynamic": 1.3986, "loss_fixed": 0.9006, "step": 571 }, { "eob_count_fixed": 23080, "eob_loss_fixed": 0.1804, "epoch": 0.8658467360454115, "grad_norm": 0.20551015436649323, "learning_rate": 4.785812222664227e-07, "loss": 1.1015, "loss_dynamic": 1.2978, "loss_fixed": 0.9051, "step": 572 }, { "eob_count_fixed": 23084, "eob_loss_fixed": 0.1688, "epoch": 0.867360454115421, "grad_norm": 0.21004240214824677, "learning_rate": 4.681734281794759e-07, "loss": 1.0863, "loss_dynamic": 1.289, "loss_fixed": 0.8836, "step": 573 }, { "eob_count_fixed": 23350, "eob_loss_fixed": 0.1721, "epoch": 0.8688741721854305, "grad_norm": 0.19750754535198212, "learning_rate": 4.5787449102664527e-07, "loss": 1.0892, "loss_dynamic": 1.277, "loss_fixed": 0.9013, "step": 574 }, { "eob_count_fixed": 22980, "eob_loss_fixed": 0.17, "epoch": 0.8703878902554399, "grad_norm": 0.19520039856433868, "learning_rate": 4.476846581939925e-07, "loss": 1.077, "loss_dynamic": 1.2674, "loss_fixed": 0.8865, "step": 575 }, { "eob_count_fixed": 23117, "eob_loss_fixed": 0.1631, "epoch": 0.8719016083254494, "grad_norm": 0.20797178149223328, "learning_rate": 4.3760417444683344e-07, "loss": 1.0967, "loss_dynamic": 1.3235, "loss_fixed": 0.8699, "step": 576 }, { "eob_count_fixed": 23162, "eob_loss_fixed": 0.1752, "epoch": 0.8734153263954588, "grad_norm": 0.19275124371051788, "learning_rate": 4.276332819238599e-07, "loss": 1.0962, "loss_dynamic": 1.304, "loss_fixed": 0.8884, "step": 577 }, { "eob_count_fixed": 23198, "eob_loss_fixed": 0.1661, "epoch": 0.8749290444654683, "grad_norm": 0.21545203030109406, "learning_rate": 4.1777222013131893e-07, "loss": 1.0753, "loss_dynamic": 1.2733, "loss_fixed": 0.8772, "step": 578 }, { "eob_count_fixed": 23094, "eob_loss_fixed": 0.1664, "epoch": 0.8764427625354778, "grad_norm": 0.2113175094127655, "learning_rate": 4.0802122593727125e-07, "loss": 1.0961, "loss_dynamic": 1.3268, "loss_fixed": 0.8655, "step": 579 }, { "eob_count_fixed": 23064, "eob_loss_fixed": 0.173, "epoch": 0.8779564806054873, "grad_norm": 0.19722746312618256, "learning_rate": 3.98380533565888e-07, "loss": 1.0301, "loss_dynamic": 1.1866, "loss_fixed": 0.8736, "step": 580 }, { "eob_count_fixed": 23127, "eob_loss_fixed": 0.1717, "epoch": 0.8794701986754967, "grad_norm": 0.19765855371952057, "learning_rate": 3.888503745918343e-07, "loss": 1.1399, "loss_dynamic": 1.3667, "loss_fixed": 0.9132, "step": 581 }, { "eob_count_fixed": 23172, "eob_loss_fixed": 0.1693, "epoch": 0.8809839167455061, "grad_norm": 0.20713695883750916, "learning_rate": 3.7943097793470054e-07, "loss": 1.0859, "loss_dynamic": 1.2718, "loss_fixed": 0.8999, "step": 582 }, { "eob_count_fixed": 22983, "eob_loss_fixed": 0.1641, "epoch": 0.8824976348155156, "grad_norm": 0.20887163281440735, "learning_rate": 3.701225698535077e-07, "loss": 1.1044, "loss_dynamic": 1.3283, "loss_fixed": 0.8805, "step": 583 }, { "eob_count_fixed": 23090, "eob_loss_fixed": 0.1625, "epoch": 0.8840113528855251, "grad_norm": 0.21147388219833374, "learning_rate": 3.6092537394126937e-07, "loss": 1.0746, "loss_dynamic": 1.2618, "loss_fixed": 0.8873, "step": 584 }, { "eob_count_fixed": 22991, "eob_loss_fixed": 0.166, "epoch": 0.8855250709555346, "grad_norm": 0.19041500985622406, "learning_rate": 3.5183961111962596e-07, "loss": 1.0753, "loss_dynamic": 1.2649, "loss_fixed": 0.8857, "step": 585 }, { "eob_count_fixed": 23202, "eob_loss_fixed": 0.1746, "epoch": 0.887038789025544, "grad_norm": 0.20787997543811798, "learning_rate": 3.4286549963353123e-07, "loss": 1.1521, "loss_dynamic": 1.3948, "loss_fixed": 0.9094, "step": 586 }, { "eob_count_fixed": 23250, "eob_loss_fixed": 0.1663, "epoch": 0.8885525070955534, "grad_norm": 0.20761670172214508, "learning_rate": 3.3400325504601263e-07, "loss": 1.0645, "loss_dynamic": 1.2421, "loss_fixed": 0.8869, "step": 587 }, { "eob_count_fixed": 23345, "eob_loss_fixed": 0.17, "epoch": 0.8900662251655629, "grad_norm": 0.20258772373199463, "learning_rate": 3.25253090232997e-07, "loss": 1.0644, "loss_dynamic": 1.2399, "loss_fixed": 0.889, "step": 588 }, { "eob_count_fixed": 23154, "eob_loss_fixed": 0.1725, "epoch": 0.8915799432355723, "grad_norm": 0.19433066248893738, "learning_rate": 3.1661521537819263e-07, "loss": 1.0523, "loss_dynamic": 1.2106, "loss_fixed": 0.894, "step": 589 }, { "eob_count_fixed": 23150, "eob_loss_fixed": 0.1719, "epoch": 0.8930936613055819, "grad_norm": 0.19783279299736023, "learning_rate": 3.080898379680419e-07, "loss": 1.0798, "loss_dynamic": 1.2324, "loss_fixed": 0.9273, "step": 590 }, { "eob_count_fixed": 23056, "eob_loss_fixed": 0.1715, "epoch": 0.8946073793755913, "grad_norm": 0.18526658415794373, "learning_rate": 2.996771627867379e-07, "loss": 1.0907, "loss_dynamic": 1.2842, "loss_fixed": 0.8973, "step": 591 }, { "eob_count_fixed": 23048, "eob_loss_fixed": 0.1749, "epoch": 0.8961210974456008, "grad_norm": 0.19446301460266113, "learning_rate": 2.9137739191130445e-07, "loss": 1.0972, "loss_dynamic": 1.3053, "loss_fixed": 0.8891, "step": 592 }, { "eob_count_fixed": 23234, "eob_loss_fixed": 0.1663, "epoch": 0.8976348155156102, "grad_norm": 0.18977250158786774, "learning_rate": 2.8319072470674334e-07, "loss": 1.0674, "loss_dynamic": 1.2415, "loss_fixed": 0.8933, "step": 593 }, { "eob_count_fixed": 23233, "eob_loss_fixed": 0.1775, "epoch": 0.8991485335856196, "grad_norm": 0.2210838496685028, "learning_rate": 2.7511735782124406e-07, "loss": 1.1604, "loss_dynamic": 1.4062, "loss_fixed": 0.9145, "step": 594 }, { "eob_count_fixed": 23106, "eob_loss_fixed": 0.1825, "epoch": 0.9006622516556292, "grad_norm": 0.2056199610233307, "learning_rate": 2.6715748518146347e-07, "loss": 1.0755, "loss_dynamic": 1.2478, "loss_fixed": 0.9033, "step": 595 }, { "eob_count_fixed": 23031, "eob_loss_fixed": 0.1652, "epoch": 0.9021759697256386, "grad_norm": 0.1901683211326599, "learning_rate": 2.593112979878604e-07, "loss": 1.0821, "loss_dynamic": 1.2817, "loss_fixed": 0.8824, "step": 596 }, { "eob_count_fixed": 23101, "eob_loss_fixed": 0.1733, "epoch": 0.9036896877956481, "grad_norm": 0.19854025542736053, "learning_rate": 2.5157898471011076e-07, "loss": 1.1085, "loss_dynamic": 1.3142, "loss_fixed": 0.9028, "step": 597 }, { "eob_count_fixed": 23191, "eob_loss_fixed": 0.1751, "epoch": 0.9052034058656575, "grad_norm": 0.20916058123111725, "learning_rate": 2.439607310825776e-07, "loss": 1.0914, "loss_dynamic": 1.2883, "loss_fixed": 0.8945, "step": 598 }, { "eob_count_fixed": 23186, "eob_loss_fixed": 0.1756, "epoch": 0.906717123935667, "grad_norm": 0.2138511836528778, "learning_rate": 2.3645672009984687e-07, "loss": 1.0952, "loss_dynamic": 1.2897, "loss_fixed": 0.9007, "step": 599 }, { "eob_count_fixed": 23030, "eob_loss_fixed": 0.1667, "epoch": 0.9082308420056765, "grad_norm": 0.1940704882144928, "learning_rate": 2.2906713201233666e-07, "loss": 1.0561, "loss_dynamic": 1.2246, "loss_fixed": 0.8876, "step": 600 }, { "eob_count_fixed": 23173, "eob_loss_fixed": 0.1698, "epoch": 0.9097445600756859, "grad_norm": 0.2067168653011322, "learning_rate": 2.217921443219634e-07, "loss": 1.1021, "loss_dynamic": 1.2917, "loss_fixed": 0.9126, "step": 601 }, { "eob_count_fixed": 23211, "eob_loss_fixed": 0.1635, "epoch": 0.9112582781456954, "grad_norm": 0.19758015871047974, "learning_rate": 2.1463193177788054e-07, "loss": 1.0257, "loss_dynamic": 1.1706, "loss_fixed": 0.8807, "step": 602 }, { "eob_count_fixed": 23109, "eob_loss_fixed": 0.1649, "epoch": 0.9127719962157048, "grad_norm": 0.20059126615524292, "learning_rate": 2.0758666637227865e-07, "loss": 1.0953, "loss_dynamic": 1.3012, "loss_fixed": 0.8894, "step": 603 }, { "eob_count_fixed": 23083, "eob_loss_fixed": 0.168, "epoch": 0.9142857142857143, "grad_norm": 0.21166343986988068, "learning_rate": 2.0065651733625902e-07, "loss": 1.0762, "loss_dynamic": 1.253, "loss_fixed": 0.8994, "step": 604 }, { "eob_count_fixed": 23164, "eob_loss_fixed": 0.1675, "epoch": 0.9157994323557237, "grad_norm": 0.20116858184337616, "learning_rate": 1.9384165113576215e-07, "loss": 1.0913, "loss_dynamic": 1.2829, "loss_fixed": 0.8996, "step": 605 }, { "eob_count_fixed": 23107, "eob_loss_fixed": 0.1675, "epoch": 0.9173131504257332, "grad_norm": 0.19238729774951935, "learning_rate": 1.8714223146757194e-07, "loss": 1.1029, "loss_dynamic": 1.3171, "loss_fixed": 0.8887, "step": 606 }, { "eob_count_fixed": 23002, "eob_loss_fixed": 0.163, "epoch": 0.9188268684957427, "grad_norm": 0.2087770700454712, "learning_rate": 1.805584192553872e-07, "loss": 1.0818, "loss_dynamic": 1.2869, "loss_fixed": 0.8766, "step": 607 }, { "eob_count_fixed": 23146, "eob_loss_fixed": 0.1759, "epoch": 0.9203405865657521, "grad_norm": 0.18620748817920685, "learning_rate": 1.740903726459503e-07, "loss": 1.0721, "loss_dynamic": 1.2474, "loss_fixed": 0.8968, "step": 608 }, { "eob_count_fixed": 23074, "eob_loss_fixed": 0.1779, "epoch": 0.9218543046357616, "grad_norm": 0.20933733880519867, "learning_rate": 1.6773824700525133e-07, "loss": 1.1555, "loss_dynamic": 1.3958, "loss_fixed": 0.9153, "step": 609 }, { "eob_count_fixed": 23091, "eob_loss_fixed": 0.1721, "epoch": 0.923368022705771, "grad_norm": 0.20692850649356842, "learning_rate": 1.6150219491479656e-07, "loss": 1.1306, "loss_dynamic": 1.3679, "loss_fixed": 0.8933, "step": 610 }, { "eob_count_fixed": 23070, "eob_loss_fixed": 0.1724, "epoch": 0.9248817407757806, "grad_norm": 0.19492830336093903, "learning_rate": 1.553823661679421e-07, "loss": 1.0618, "loss_dynamic": 1.224, "loss_fixed": 0.8996, "step": 611 }, { "eob_count_fixed": 23046, "eob_loss_fixed": 0.1785, "epoch": 0.92639545884579, "grad_norm": 0.19594687223434448, "learning_rate": 1.493789077662955e-07, "loss": 1.1307, "loss_dynamic": 1.3488, "loss_fixed": 0.9127, "step": 612 }, { "eob_count_fixed": 23046, "eob_loss_fixed": 0.1708, "epoch": 0.9279091769157994, "grad_norm": 0.20037534832954407, "learning_rate": 1.434919639161869e-07, "loss": 1.0412, "loss_dynamic": 1.1847, "loss_fixed": 0.8977, "step": 613 }, { "eob_count_fixed": 23088, "eob_loss_fixed": 0.1685, "epoch": 0.9294228949858089, "grad_norm": 0.2017602026462555, "learning_rate": 1.3772167602520313e-07, "loss": 1.1134, "loss_dynamic": 1.3293, "loss_fixed": 0.8975, "step": 614 }, { "eob_count_fixed": 23224, "eob_loss_fixed": 0.173, "epoch": 0.9309366130558183, "grad_norm": 0.18743064999580383, "learning_rate": 1.320681826987913e-07, "loss": 1.0329, "loss_dynamic": 1.1873, "loss_fixed": 0.8785, "step": 615 }, { "eob_count_fixed": 23131, "eob_loss_fixed": 0.1691, "epoch": 0.9324503311258279, "grad_norm": 0.19095174968242645, "learning_rate": 1.2653161973692873e-07, "loss": 1.0588, "loss_dynamic": 1.2278, "loss_fixed": 0.8898, "step": 616 }, { "eob_count_fixed": 23091, "eob_loss_fixed": 0.1764, "epoch": 0.9339640491958373, "grad_norm": 0.1938062161207199, "learning_rate": 1.211121201308646e-07, "loss": 1.087, "loss_dynamic": 1.268, "loss_fixed": 0.906, "step": 617 }, { "eob_count_fixed": 23084, "eob_loss_fixed": 0.1698, "epoch": 0.9354777672658467, "grad_norm": 0.20505136251449585, "learning_rate": 1.1580981405992065e-07, "loss": 1.114, "loss_dynamic": 1.3224, "loss_fixed": 0.9055, "step": 618 }, { "eob_count_fixed": 23039, "eob_loss_fixed": 0.1773, "epoch": 0.9369914853358562, "grad_norm": 0.18493802845478058, "learning_rate": 1.1062482888836656e-07, "loss": 1.0812, "loss_dynamic": 1.2346, "loss_fixed": 0.9277, "step": 619 }, { "eob_count_fixed": 23230, "eob_loss_fixed": 0.1708, "epoch": 0.9385052034058656, "grad_norm": 0.1911139339208603, "learning_rate": 1.0555728916236019e-07, "loss": 1.1158, "loss_dynamic": 1.3122, "loss_fixed": 0.9193, "step": 620 }, { "eob_count_fixed": 23008, "eob_loss_fixed": 0.1764, "epoch": 0.9400189214758751, "grad_norm": 0.19547654688358307, "learning_rate": 1.006073166069571e-07, "loss": 1.0737, "loss_dynamic": 1.2166, "loss_fixed": 0.9307, "step": 621 }, { "eob_count_fixed": 22966, "eob_loss_fixed": 0.1659, "epoch": 0.9415326395458846, "grad_norm": 0.21599315106868744, "learning_rate": 9.577503012318468e-08, "loss": 1.0716, "loss_dynamic": 1.2566, "loss_fixed": 0.8866, "step": 622 }, { "eob_count_fixed": 23053, "eob_loss_fixed": 0.1722, "epoch": 0.9430463576158941, "grad_norm": 0.18753939867019653, "learning_rate": 9.106054578518708e-08, "loss": 1.0933, "loss_dynamic": 1.299, "loss_fixed": 0.8877, "step": 623 }, { "eob_count_fixed": 23165, "eob_loss_fixed": 0.1713, "epoch": 0.9445600756859035, "grad_norm": 0.19258585572242737, "learning_rate": 8.646397683743756e-08, "loss": 1.1024, "loss_dynamic": 1.2904, "loss_fixed": 0.9144, "step": 624 }, { "eob_count_fixed": 23060, "eob_loss_fixed": 0.172, "epoch": 0.9460737937559129, "grad_norm": 0.20476429164409637, "learning_rate": 8.198543369201618e-08, "loss": 1.1128, "loss_dynamic": 1.3089, "loss_fixed": 0.9167, "step": 625 }, { "eob_count_fixed": 23182, "eob_loss_fixed": 0.1721, "epoch": 0.9475875118259224, "grad_norm": 0.2051982581615448, "learning_rate": 7.762502392596072e-08, "loss": 1.105, "loss_dynamic": 1.3047, "loss_fixed": 0.9054, "step": 626 }, { "eob_count_fixed": 23272, "eob_loss_fixed": 0.1741, "epoch": 0.9491012298959319, "grad_norm": 0.19247262179851532, "learning_rate": 7.33828522786817e-08, "loss": 1.0713, "loss_dynamic": 1.2313, "loss_fixed": 0.9113, "step": 627 }, { "eob_count_fixed": 23161, "eob_loss_fixed": 0.1688, "epoch": 0.9506149479659414, "grad_norm": 0.20321109890937805, "learning_rate": 6.925902064944367e-08, "loss": 1.1046, "loss_dynamic": 1.3313, "loss_fixed": 0.8779, "step": 628 }, { "eob_count_fixed": 23100, "eob_loss_fixed": 0.1628, "epoch": 0.9521286660359508, "grad_norm": 0.19299806654453278, "learning_rate": 6.525362809492008e-08, "loss": 1.0533, "loss_dynamic": 1.2211, "loss_fixed": 0.8855, "step": 629 }, { "eob_count_fixed": 22962, "eob_loss_fixed": 0.1775, "epoch": 0.9536423841059603, "grad_norm": 0.2061794102191925, "learning_rate": 6.136677082681563e-08, "loss": 1.2005, "loss_dynamic": 1.4798, "loss_fixed": 0.9212, "step": 630 }, { "eob_count_fixed": 23211, "eob_loss_fixed": 0.1706, "epoch": 0.9551561021759697, "grad_norm": 0.2016502320766449, "learning_rate": 5.7598542209550966e-08, "loss": 1.0864, "loss_dynamic": 1.2845, "loss_fixed": 0.8883, "step": 631 }, { "eob_count_fixed": 23083, "eob_loss_fixed": 0.1703, "epoch": 0.9566698202459792, "grad_norm": 0.20341387391090393, "learning_rate": 5.394903275802277e-08, "loss": 1.0763, "loss_dynamic": 1.249, "loss_fixed": 0.9035, "step": 632 }, { "eob_count_fixed": 23137, "eob_loss_fixed": 0.1735, "epoch": 0.9581835383159887, "grad_norm": 0.1964695155620575, "learning_rate": 5.0418330135429406e-08, "loss": 1.0794, "loss_dynamic": 1.2818, "loss_fixed": 0.877, "step": 633 }, { "eob_count_fixed": 22922, "eob_loss_fixed": 0.17, "epoch": 0.9596972563859981, "grad_norm": 0.19936531782150269, "learning_rate": 4.7006519151164273e-08, "loss": 1.121, "loss_dynamic": 1.3269, "loss_fixed": 0.915, "step": 634 }, { "eob_count_fixed": 23106, "eob_loss_fixed": 0.168, "epoch": 0.9612109744560076, "grad_norm": 0.2070865035057068, "learning_rate": 4.3713681758777995e-08, "loss": 1.1388, "loss_dynamic": 1.3815, "loss_fixed": 0.8962, "step": 635 }, { "eob_count_fixed": 23030, "eob_loss_fixed": 0.1751, "epoch": 0.962724692526017, "grad_norm": 0.2018270194530487, "learning_rate": 4.053989705401384e-08, "loss": 1.0732, "loss_dynamic": 1.2559, "loss_fixed": 0.8904, "step": 636 }, { "eob_count_fixed": 23216, "eob_loss_fixed": 0.1695, "epoch": 0.9642384105960264, "grad_norm": 0.20239993929862976, "learning_rate": 3.748524127290265e-08, "loss": 1.1527, "loss_dynamic": 1.4048, "loss_fixed": 0.9006, "step": 637 }, { "eob_count_fixed": 23234, "eob_loss_fixed": 0.1638, "epoch": 0.965752128666036, "grad_norm": 0.19750122725963593, "learning_rate": 3.454978778993367e-08, "loss": 1.1002, "loss_dynamic": 1.3013, "loss_fixed": 0.899, "step": 638 }, { "eob_count_fixed": 23182, "eob_loss_fixed": 0.1726, "epoch": 0.9672658467360454, "grad_norm": 0.19213822484016418, "learning_rate": 3.173360711629325e-08, "loss": 1.1006, "loss_dynamic": 1.2834, "loss_fixed": 0.9179, "step": 639 }, { "eob_count_fixed": 23207, "eob_loss_fixed": 0.169, "epoch": 0.9687795648060549, "grad_norm": 0.19141767919063568, "learning_rate": 2.9036766898170032e-08, "loss": 1.0797, "loss_dynamic": 1.2556, "loss_fixed": 0.9039, "step": 640 }, { "eob_count_fixed": 23134, "eob_loss_fixed": 0.1675, "epoch": 0.9702932828760643, "grad_norm": 0.20646406710147858, "learning_rate": 2.6459331915129617e-08, "loss": 1.0588, "loss_dynamic": 1.2332, "loss_fixed": 0.8844, "step": 641 }, { "eob_count_fixed": 23196, "eob_loss_fixed": 0.1722, "epoch": 0.9718070009460738, "grad_norm": 0.1970716118812561, "learning_rate": 2.400136407855913e-08, "loss": 1.0703, "loss_dynamic": 1.2198, "loss_fixed": 0.9207, "step": 642 }, { "eob_count_fixed": 22994, "eob_loss_fixed": 0.1698, "epoch": 0.9733207190160833, "grad_norm": 0.19086982309818268, "learning_rate": 2.16629224301812e-08, "loss": 1.0553, "loss_dynamic": 1.2229, "loss_fixed": 0.8877, "step": 643 }, { "eob_count_fixed": 23006, "eob_loss_fixed": 0.1771, "epoch": 0.9748344370860927, "grad_norm": 0.19106905162334442, "learning_rate": 1.9444063140632297e-08, "loss": 1.0905, "loss_dynamic": 1.283, "loss_fixed": 0.898, "step": 644 }, { "eob_count_fixed": 23249, "eob_loss_fixed": 0.1648, "epoch": 0.9763481551561022, "grad_norm": 0.20776167511940002, "learning_rate": 1.734483950811716e-08, "loss": 1.0909, "loss_dynamic": 1.2944, "loss_fixed": 0.8875, "step": 645 }, { "eob_count_fixed": 23138, "eob_loss_fixed": 0.1581, "epoch": 0.9778618732261116, "grad_norm": 0.18972894549369812, "learning_rate": 1.53653019571276e-08, "loss": 1.0637, "loss_dynamic": 1.2576, "loss_fixed": 0.8698, "step": 646 }, { "eob_count_fixed": 23113, "eob_loss_fixed": 0.1728, "epoch": 0.9793755912961211, "grad_norm": 0.19805556535720825, "learning_rate": 1.3505498037229003e-08, "loss": 1.1074, "loss_dynamic": 1.3129, "loss_fixed": 0.9019, "step": 647 }, { "eob_count_fixed": 23222, "eob_loss_fixed": 0.1665, "epoch": 0.9808893093661306, "grad_norm": 0.21592503786087036, "learning_rate": 1.1765472421921831e-08, "loss": 1.0913, "loss_dynamic": 1.2885, "loss_fixed": 0.8941, "step": 648 }, { "eob_count_fixed": 23143, "eob_loss_fixed": 0.1773, "epoch": 0.98240302743614, "grad_norm": 0.19953139126300812, "learning_rate": 1.0145266907566342e-08, "loss": 1.0896, "loss_dynamic": 1.2626, "loss_fixed": 0.9166, "step": 649 }, { "eob_count_fixed": 23039, "eob_loss_fixed": 0.1799, "epoch": 0.9839167455061495, "grad_norm": 0.2071475088596344, "learning_rate": 8.6449204123773e-09, "loss": 1.1365, "loss_dynamic": 1.3498, "loss_fixed": 0.9232, "step": 650 }, { "eob_count_fixed": 23165, "eob_loss_fixed": 0.1656, "epoch": 0.9854304635761589, "grad_norm": 0.19088952243328094, "learning_rate": 7.264468975491934e-09, "loss": 1.0868, "loss_dynamic": 1.2835, "loss_fixed": 0.8902, "step": 651 }, { "eob_count_fixed": 22987, "eob_loss_fixed": 0.169, "epoch": 0.9869441816461684, "grad_norm": 0.20201347768306732, "learning_rate": 6.003945756103968e-09, "loss": 1.1035, "loss_dynamic": 1.3287, "loss_fixed": 0.8783, "step": 652 }, { "eob_count_fixed": 23109, "eob_loss_fixed": 0.1669, "epoch": 0.9884578997161778, "grad_norm": 0.20471379160881042, "learning_rate": 4.863381032664816e-09, "loss": 1.1245, "loss_dynamic": 1.3538, "loss_fixed": 0.8953, "step": 653 }, { "eob_count_fixed": 23249, "eob_loss_fixed": 0.1687, "epoch": 0.9899716177861874, "grad_norm": 0.19555577635765076, "learning_rate": 3.842802202158047e-09, "loss": 1.0414, "loss_dynamic": 1.1802, "loss_fixed": 0.9025, "step": 654 }, { "eob_count_fixed": 22970, "eob_loss_fixed": 0.1733, "epoch": 0.9914853358561968, "grad_norm": 0.20639273524284363, "learning_rate": 2.9422337794415836e-09, "loss": 1.1668, "loss_dynamic": 1.4217, "loss_fixed": 0.912, "step": 655 }, { "eob_count_fixed": 23176, "eob_loss_fixed": 0.1669, "epoch": 0.9929990539262062, "grad_norm": 0.1977607011795044, "learning_rate": 2.161697396657614e-09, "loss": 1.0491, "loss_dynamic": 1.2152, "loss_fixed": 0.8829, "step": 656 }, { "eob_count_fixed": 23159, "eob_loss_fixed": 0.1775, "epoch": 0.9945127719962157, "grad_norm": 0.20130042731761932, "learning_rate": 1.5012118027135647e-09, "loss": 1.0938, "loss_dynamic": 1.2893, "loss_fixed": 0.8984, "step": 657 }, { "eob_count_fixed": 22971, "eob_loss_fixed": 0.1703, "epoch": 0.9960264900662251, "grad_norm": 0.20477260649204254, "learning_rate": 9.607928628319052e-10, "loss": 1.1521, "loss_dynamic": 1.3931, "loss_fixed": 0.9112, "step": 658 }, { "eob_count_fixed": 23131, "eob_loss_fixed": 0.1717, "epoch": 0.9975402081362347, "grad_norm": 0.20038621127605438, "learning_rate": 5.404535581693404e-10, "loss": 1.0954, "loss_dynamic": 1.2884, "loss_fixed": 0.9024, "step": 659 }, { "eob_count_fixed": 22920, "eob_loss_fixed": 0.1664, "epoch": 0.9990539262062441, "grad_norm": 0.19627010822296143, "learning_rate": 2.4020398550372857e-10, "loss": 1.0831, "loss_dynamic": 1.2928, "loss_fixed": 0.8734, "step": 660 }, { "eob_count_fixed": 14402, "eob_loss_fixed": 0.1081, "epoch": 1.0, "grad_norm": 0.2122781127691269, "learning_rate": 6.005135699260755e-11, "loss": 0.6989, "loss_dynamic": 0.8263, "loss_fixed": 0.5715, "step": 661 }, { "epoch": 1.0, "step": 661, "total_flos": 6.687422305608191e+19, "train_loss": 1.228308567210874, "train_runtime": 136377.5959, "train_samples_per_second": 9.92, "train_steps_per_second": 0.005 } ], "logging_steps": 1, "max_steps": 661, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 25, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 6.687422305608191e+19, "train_batch_size": 32, "trial_name": null, "trial_params": null }