diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -11,3645 +11,3645 @@ "log_history": [ { "epoch": 0.005780346820809248, - "grad_norm": 37.11543498357617, + "grad_norm": 21.87793783020131, "learning_rate": 0.0, - "loss": 6.3788, + "loss": 3.0384, "step": 1 }, { "epoch": 0.011560693641618497, - "grad_norm": 40.33716745892234, + "grad_norm": 25.944153812411226, "learning_rate": 1.9230769230769234e-07, - "loss": 6.3346, + "loss": 2.7475, "step": 2 }, { "epoch": 0.017341040462427744, - "grad_norm": 33.72797985924269, + "grad_norm": 27.600907958149808, "learning_rate": 3.846153846153847e-07, - "loss": 5.2429, + "loss": 4.4341, "step": 3 }, { "epoch": 0.023121387283236993, - "grad_norm": 31.313283709227267, + "grad_norm": 32.91981305085188, "learning_rate": 5.76923076923077e-07, - "loss": 5.2123, + "loss": 6.5326, "step": 4 }, { "epoch": 0.028901734104046242, - "grad_norm": 33.753428042253816, + "grad_norm": 28.298528074059195, "learning_rate": 7.692307692307694e-07, - "loss": 5.2401, + "loss": 4.6837, "step": 5 }, { "epoch": 0.03468208092485549, - "grad_norm": 31.856571765727526, + "grad_norm": 28.4809265754676, "learning_rate": 9.615384615384617e-07, - "loss": 5.0233, + "loss": 4.2992, "step": 6 }, { "epoch": 0.04046242774566474, - "grad_norm": 30.554267749429098, + "grad_norm": 34.294499257509685, "learning_rate": 1.153846153846154e-06, - "loss": 4.8847, + "loss": 6.3554, "step": 7 }, { "epoch": 0.046242774566473986, - "grad_norm": 33.271535701662366, + "grad_norm": 32.40670348934399, "learning_rate": 1.3461538461538462e-06, - "loss": 4.7594, + "loss": 5.5468, "step": 8 }, { "epoch": 0.05202312138728324, - "grad_norm": 31.346198862760094, + "grad_norm": 31.088077207788306, "learning_rate": 1.5384615384615387e-06, - "loss": 4.6163, + "loss": 4.7881, "step": 9 }, { "epoch": 0.057803468208092484, - "grad_norm": 31.57620712287447, + "grad_norm": 27.103526225977465, "learning_rate": 1.7307692307692308e-06, - "loss": 4.5756, + "loss": 4.2773, "step": 10 }, { "epoch": 0.06358381502890173, - "grad_norm": 29.890651646322883, + "grad_norm": 23.372545017917634, "learning_rate": 1.9230769230769234e-06, - "loss": 4.5248, + "loss": 3.7714, "step": 11 }, { "epoch": 0.06936416184971098, - "grad_norm": 27.07938902623954, + "grad_norm": 29.536841037137425, "learning_rate": 2.1153846153846155e-06, - "loss": 3.6664, + "loss": 5.4037, "step": 12 }, { "epoch": 0.07514450867052024, - "grad_norm": 25.348009175778653, + "grad_norm": 20.51058072309702, "learning_rate": 2.307692307692308e-06, - "loss": 4.0609, + "loss": 3.0428, "step": 13 }, { "epoch": 0.08092485549132948, - "grad_norm": 26.375202591021132, + "grad_norm": 25.851758056606016, "learning_rate": 2.5e-06, - "loss": 3.8629, + "loss": 3.5816, "step": 14 }, { "epoch": 0.08670520231213873, - "grad_norm": 18.784615156036363, + "grad_norm": 15.01324797937803, "learning_rate": 2.6923076923076923e-06, - "loss": 3.333, + "loss": 2.3105, "step": 15 }, { "epoch": 0.09248554913294797, - "grad_norm": 17.383555382248034, + "grad_norm": 15.199500088532252, "learning_rate": 2.8846153846153845e-06, - "loss": 2.524, + "loss": 2.5381, "step": 16 }, { "epoch": 0.09826589595375723, - "grad_norm": 11.986743526930361, + "grad_norm": 17.505903657285067, "learning_rate": 3.0769230769230774e-06, - "loss": 1.9181, + "loss": 2.9574, "step": 17 }, { "epoch": 0.10404624277456648, - "grad_norm": 14.832587316240966, + "grad_norm": 11.528249965912817, "learning_rate": 3.2692307692307696e-06, - "loss": 3.1124, + "loss": 2.3289, "step": 18 }, { "epoch": 0.10982658959537572, - "grad_norm": 12.566654914969215, + "grad_norm": 12.880494070804799, "learning_rate": 3.4615384615384617e-06, - "loss": 2.4797, + "loss": 2.6934, "step": 19 }, { "epoch": 0.11560693641618497, - "grad_norm": 10.2540748952002, + "grad_norm": 10.865402934129182, "learning_rate": 3.653846153846154e-06, - "loss": 2.4203, + "loss": 2.6421, "step": 20 }, { "epoch": 0.12138728323699421, - "grad_norm": 8.930966424025984, + "grad_norm": 7.125972199900619, "learning_rate": 3.846153846153847e-06, - "loss": 2.3697, + "loss": 2.5033, "step": 21 }, { "epoch": 0.12716763005780346, - "grad_norm": 5.924769584674026, + "grad_norm": 11.21541490251159, "learning_rate": 4.0384615384615385e-06, - "loss": 2.1558, + "loss": 2.6345, "step": 22 }, { "epoch": 0.1329479768786127, - "grad_norm": 5.088251567950973, + "grad_norm": 4.100350430185937, "learning_rate": 4.230769230769231e-06, - "loss": 2.1828, + "loss": 1.9498, "step": 23 }, { "epoch": 0.13872832369942195, - "grad_norm": 3.697328121977238, + "grad_norm": 2.2057204886805226, "learning_rate": 4.423076923076924e-06, - "loss": 1.9942, + "loss": 1.0249, "step": 24 }, { "epoch": 0.14450867052023122, - "grad_norm": 4.315687543784776, + "grad_norm": 3.0429202636869386, "learning_rate": 4.615384615384616e-06, - "loss": 2.3611, + "loss": 1.68, "step": 25 }, { "epoch": 0.15028901734104047, - "grad_norm": 3.7062496976958856, + "grad_norm": 5.131215855608062, "learning_rate": 4.807692307692308e-06, - "loss": 1.9552, + "loss": 2.6619, "step": 26 }, { "epoch": 0.15606936416184972, - "grad_norm": 4.357400309380474, + "grad_norm": 3.4088527219076052, "learning_rate": 5e-06, - "loss": 2.4886, + "loss": 1.8795, "step": 27 }, { "epoch": 0.16184971098265896, - "grad_norm": 3.6115066565591207, + "grad_norm": 3.7165759547218675, "learning_rate": 5.192307692307693e-06, - "loss": 2.4048, + "loss": 2.1434, "step": 28 }, { "epoch": 0.1676300578034682, - "grad_norm": 3.510175992563641, + "grad_norm": 3.4769214622128977, "learning_rate": 5.384615384615385e-06, - "loss": 2.1366, + "loss": 1.9582, "step": 29 }, { "epoch": 0.17341040462427745, - "grad_norm": 4.187147912815141, + "grad_norm": 3.4755064358560963, "learning_rate": 5.576923076923077e-06, - "loss": 2.4846, + "loss": 1.7078, "step": 30 }, { "epoch": 0.1791907514450867, - "grad_norm": 4.567484119316995, + "grad_norm": 2.346759845137902, "learning_rate": 5.769230769230769e-06, - "loss": 2.4767, + "loss": 1.3841, "step": 31 }, { "epoch": 0.18497109826589594, - "grad_norm": 3.0928625436910706, + "grad_norm": 3.3589744625467217, "learning_rate": 5.961538461538462e-06, - "loss": 2.5788, + "loss": 1.9876, "step": 32 }, { "epoch": 0.1907514450867052, - "grad_norm": 2.2973285643194883, + "grad_norm": 2.23530410646522, "learning_rate": 6.153846153846155e-06, - "loss": 1.3782, + "loss": 1.3731, "step": 33 }, { "epoch": 0.19653179190751446, - "grad_norm": 3.035621795378031, + "grad_norm": 3.0334549128243062, "learning_rate": 6.3461538461538466e-06, - "loss": 1.8167, + "loss": 1.954, "step": 34 }, { "epoch": 0.2023121387283237, - "grad_norm": 2.248843299613026, + "grad_norm": 3.019944847537477, "learning_rate": 6.538461538461539e-06, - "loss": 1.6573, + "loss": 1.7887, "step": 35 }, { "epoch": 0.20809248554913296, - "grad_norm": 2.9450508484455624, + "grad_norm": 2.574674373885689, "learning_rate": 6.730769230769232e-06, - "loss": 1.6621, + "loss": 1.3396, "step": 36 }, { "epoch": 0.2138728323699422, - "grad_norm": 2.0245669872834724, + "grad_norm": 1.6788041080361362, "learning_rate": 6.923076923076923e-06, - "loss": 1.3089, + "loss": 1.1171, "step": 37 }, { "epoch": 0.21965317919075145, - "grad_norm": 3.0051251298315313, + "grad_norm": 2.223469814132116, "learning_rate": 7.115384615384616e-06, - "loss": 1.8963, + "loss": 1.1395, "step": 38 }, { "epoch": 0.2254335260115607, - "grad_norm": 2.391162290671958, + "grad_norm": 2.1461305027339996, "learning_rate": 7.307692307692308e-06, - "loss": 1.4552, + "loss": 1.2949, "step": 39 }, { "epoch": 0.23121387283236994, - "grad_norm": 1.882996239084883, + "grad_norm": 1.8459121298686243, "learning_rate": 7.500000000000001e-06, - "loss": 1.3399, + "loss": 1.2421, "step": 40 }, { "epoch": 0.23699421965317918, - "grad_norm": 2.908873697262596, + "grad_norm": 1.6284680007294197, "learning_rate": 7.692307692307694e-06, - "loss": 1.8609, + "loss": 1.2065, "step": 41 }, { "epoch": 0.24277456647398843, - "grad_norm": 1.9301319572306712, + "grad_norm": 2.600840218113501, "learning_rate": 7.884615384615384e-06, - "loss": 1.4359, + "loss": 1.5263, "step": 42 }, { "epoch": 0.24855491329479767, - "grad_norm": 2.2696687500870096, + "grad_norm": 2.4077378791650097, "learning_rate": 8.076923076923077e-06, - "loss": 1.6196, + "loss": 1.6776, "step": 43 }, { "epoch": 0.2543352601156069, - "grad_norm": 2.104357570746364, + "grad_norm": 1.7255161535686105, "learning_rate": 8.26923076923077e-06, - "loss": 1.5003, + "loss": 1.0938, "step": 44 }, { "epoch": 0.26011560693641617, - "grad_norm": 1.5008039120328356, + "grad_norm": 1.4160107000643258, "learning_rate": 8.461538461538462e-06, - "loss": 1.2138, + "loss": 1.0143, "step": 45 }, { "epoch": 0.2658959537572254, - "grad_norm": 1.9450812082745943, + "grad_norm": 1.5785981094401695, "learning_rate": 8.653846153846155e-06, - "loss": 1.3296, + "loss": 1.1311, "step": 46 }, { "epoch": 0.27167630057803466, - "grad_norm": 1.6939530266083083, + "grad_norm": 1.6456550398187915, "learning_rate": 8.846153846153847e-06, - "loss": 1.2155, + "loss": 1.1797, "step": 47 }, { "epoch": 0.2774566473988439, - "grad_norm": 1.4805253940369114, + "grad_norm": 1.043459304370755, "learning_rate": 9.03846153846154e-06, - "loss": 1.0609, + "loss": 0.7448, "step": 48 }, { "epoch": 0.2832369942196532, - "grad_norm": 2.113112852413699, + "grad_norm": 1.219858313334948, "learning_rate": 9.230769230769232e-06, - "loss": 1.5942, + "loss": 0.8615, "step": 49 }, { "epoch": 0.28901734104046245, - "grad_norm": 1.8550520733433014, + "grad_norm": 1.8504734290476341, "learning_rate": 9.423076923076923e-06, - "loss": 1.3397, + "loss": 1.0442, "step": 50 }, { "epoch": 0.2947976878612717, - "grad_norm": 2.318256791349785, + "grad_norm": 1.6671704729926984, "learning_rate": 9.615384615384616e-06, - "loss": 1.2471, + "loss": 1.1062, "step": 51 }, { "epoch": 0.30057803468208094, - "grad_norm": 1.3417626130789702, + "grad_norm": 1.7440575528412874, "learning_rate": 9.807692307692308e-06, - "loss": 1.0969, + "loss": 0.9787, "step": 52 }, { "epoch": 0.3063583815028902, - "grad_norm": 1.4190588634495163, + "grad_norm": 1.6865691956435074, "learning_rate": 1e-05, - "loss": 1.2034, + "loss": 1.039, "step": 53 }, { "epoch": 0.31213872832369943, - "grad_norm": 1.1257105909570848, + "grad_norm": 1.8681738071024672, "learning_rate": 9.99988686307906e-06, - "loss": 0.9821, + "loss": 1.1171, "step": 54 }, { "epoch": 0.3179190751445087, - "grad_norm": 1.746150912676174, + "grad_norm": 1.4603859563880623, "learning_rate": 9.999547457436221e-06, - "loss": 1.0973, + "loss": 1.1437, "step": 55 }, { "epoch": 0.3236994219653179, - "grad_norm": 1.1167709287946892, + "grad_norm": 1.8133100631538102, "learning_rate": 9.99898179843121e-06, - "loss": 0.8932, + "loss": 1.2624, "step": 56 }, { "epoch": 0.32947976878612717, - "grad_norm": 1.256140915327827, + "grad_norm": 0.9630648626847151, "learning_rate": 9.998189911662793e-06, - "loss": 1.019, + "loss": 0.7498, "step": 57 }, { "epoch": 0.3352601156069364, - "grad_norm": 1.2753655875351548, + "grad_norm": 1.9231249269543056, "learning_rate": 9.99717183296762e-06, - "loss": 1.0695, + "loss": 1.4657, "step": 58 }, { "epoch": 0.34104046242774566, - "grad_norm": 1.6244836687299578, + "grad_norm": 1.362454107646963, "learning_rate": 9.995927608418612e-06, - "loss": 1.2366, + "loss": 0.9301, "step": 59 }, { "epoch": 0.3468208092485549, - "grad_norm": 0.7775106736009707, + "grad_norm": 1.5299236968171044, "learning_rate": 9.994457294322858e-06, - "loss": 0.6811, + "loss": 1.0736, "step": 60 }, { "epoch": 0.35260115606936415, - "grad_norm": 0.9440090347669787, + "grad_norm": 1.256654893455161, "learning_rate": 9.992760957219083e-06, - "loss": 0.8703, + "loss": 0.8821, "step": 61 }, { "epoch": 0.3583815028901734, - "grad_norm": 0.9958318951903639, + "grad_norm": 1.1404370436709175, "learning_rate": 9.99083867387463e-06, - "loss": 0.9174, + "loss": 0.8719, "step": 62 }, { "epoch": 0.36416184971098264, - "grad_norm": 0.7634557852779634, + "grad_norm": 1.3053446137350901, "learning_rate": 9.988690531281988e-06, - "loss": 0.6985, + "loss": 0.9312, "step": 63 }, { "epoch": 0.3699421965317919, - "grad_norm": 1.3069604662689145, + "grad_norm": 1.227248883019738, "learning_rate": 9.986316626654851e-06, - "loss": 1.0722, + "loss": 0.976, "step": 64 }, { "epoch": 0.37572254335260113, - "grad_norm": 1.0746802589110356, + "grad_norm": 1.129527697823112, "learning_rate": 9.983717067423721e-06, - "loss": 1.0268, + "loss": 1.0485, "step": 65 }, { "epoch": 0.3815028901734104, - "grad_norm": 0.8438968159980698, + "grad_norm": 2.2554059120706835, "learning_rate": 9.980891971231052e-06, - "loss": 0.7803, + "loss": 1.2171, "step": 66 }, { "epoch": 0.3872832369942196, - "grad_norm": 1.1334177185581455, + "grad_norm": 0.904872887481673, "learning_rate": 9.977841465925918e-06, - "loss": 0.9242, + "loss": 0.7695, "step": 67 }, { "epoch": 0.3930635838150289, - "grad_norm": 0.8394453857662859, + "grad_norm": 1.0481280098300976, "learning_rate": 9.974565689558228e-06, - "loss": 0.769, + "loss": 0.9209, "step": 68 }, { "epoch": 0.3988439306358382, - "grad_norm": 1.530644455515152, + "grad_norm": 1.2781650561787112, "learning_rate": 9.971064790372484e-06, - "loss": 1.0502, + "loss": 1.1006, "step": 69 }, { "epoch": 0.4046242774566474, - "grad_norm": 0.949416066019725, + "grad_norm": 0.8969433268648974, "learning_rate": 9.967338926801066e-06, - "loss": 0.8783, + "loss": 0.7209, "step": 70 }, { "epoch": 0.41040462427745666, - "grad_norm": 0.7463616862166296, + "grad_norm": 1.1393852014047097, "learning_rate": 9.963388267457071e-06, - "loss": 0.7611, + "loss": 0.7685, "step": 71 }, { "epoch": 0.4161849710982659, - "grad_norm": 1.894168527637676, + "grad_norm": 0.8409711904675974, "learning_rate": 9.959212991126668e-06, - "loss": 1.2004, + "loss": 0.5886, "step": 72 }, { "epoch": 0.42196531791907516, - "grad_norm": 0.9196092202638283, + "grad_norm": 1.0097041809739957, "learning_rate": 9.954813286761021e-06, - "loss": 0.7431, + "loss": 0.8177, "step": 73 }, { "epoch": 0.4277456647398844, - "grad_norm": 0.8018230986583454, + "grad_norm": 0.8674783887677202, "learning_rate": 9.950189353467735e-06, - "loss": 0.7637, + "loss": 0.6911, "step": 74 }, { "epoch": 0.43352601156069365, - "grad_norm": 2.0393273882588985, + "grad_norm": 1.0032789649476923, "learning_rate": 9.945341400501838e-06, - "loss": 0.9983, + "loss": 0.9463, "step": 75 }, { "epoch": 0.4393063583815029, - "grad_norm": 0.867447783004979, + "grad_norm": 1.0495612381066386, "learning_rate": 9.940269647256319e-06, - "loss": 0.9035, + "loss": 0.9201, "step": 76 }, { "epoch": 0.44508670520231214, - "grad_norm": 0.9193425231200821, + "grad_norm": 0.7000102566512686, "learning_rate": 9.934974323252195e-06, - "loss": 0.9127, + "loss": 0.633, "step": 77 }, { "epoch": 0.4508670520231214, - "grad_norm": 1.1024698344826929, + "grad_norm": 0.8404727978210973, "learning_rate": 9.929455668128129e-06, - "loss": 0.9904, + "loss": 0.7939, "step": 78 }, { "epoch": 0.45664739884393063, - "grad_norm": 0.7383825161647908, + "grad_norm": 0.8979481667152188, "learning_rate": 9.923713931629582e-06, - "loss": 0.6404, + "loss": 0.7995, "step": 79 }, { "epoch": 0.4624277456647399, - "grad_norm": 0.7098921654251555, + "grad_norm": 1.006209086886219, "learning_rate": 9.917749373597506e-06, - "loss": 0.7, + "loss": 0.9377, "step": 80 }, { "epoch": 0.4682080924855491, - "grad_norm": 1.0319611220794853, + "grad_norm": 1.3079350852166611, "learning_rate": 9.911562263956593e-06, - "loss": 0.9071, + "loss": 0.8293, "step": 81 }, { "epoch": 0.47398843930635837, - "grad_norm": 0.8526142560020595, + "grad_norm": 1.0302107201825823, "learning_rate": 9.90515288270306e-06, - "loss": 0.7953, + "loss": 0.8907, "step": 82 }, { "epoch": 0.4797687861271676, - "grad_norm": 0.8414547107608813, + "grad_norm": 0.8512910211686245, "learning_rate": 9.898521519891968e-06, - "loss": 0.7819, + "loss": 0.6677, "step": 83 }, { "epoch": 0.48554913294797686, - "grad_norm": 1.3215045319910295, + "grad_norm": 1.4034538610622003, "learning_rate": 9.891668475624106e-06, - "loss": 0.9444, + "loss": 1.011, "step": 84 }, { "epoch": 0.4913294797687861, - "grad_norm": 0.7952308185586521, + "grad_norm": 1.1133640764754114, "learning_rate": 9.884594060032407e-06, - "loss": 0.7343, + "loss": 0.9868, "step": 85 }, { "epoch": 0.49710982658959535, - "grad_norm": 1.1562300771664449, + "grad_norm": 0.8832987768292198, "learning_rate": 9.877298593267906e-06, - "loss": 0.7701, + "loss": 0.7435, "step": 86 }, { "epoch": 0.5028901734104047, - "grad_norm": 0.8099789150838541, + "grad_norm": 0.9793843183088854, "learning_rate": 9.869782405485267e-06, - "loss": 0.8307, + "loss": 0.9289, "step": 87 }, { "epoch": 0.5086705202312138, - "grad_norm": 0.964215484115652, + "grad_norm": 0.9159241257550397, "learning_rate": 9.862045836827821e-06, - "loss": 0.7516, + "loss": 0.704, "step": 88 }, { "epoch": 0.5144508670520231, - "grad_norm": 1.203561439516184, + "grad_norm": 0.8669077131194463, "learning_rate": 9.854089237412194e-06, - "loss": 0.8082, + "loss": 0.7518, "step": 89 }, { "epoch": 0.5202312138728323, - "grad_norm": 0.9332097842144084, + "grad_norm": 0.9177313381025624, "learning_rate": 9.84591296731245e-06, - "loss": 0.9396, + "loss": 0.6925, "step": 90 }, { "epoch": 0.5260115606936416, - "grad_norm": 0.7012438892577367, + "grad_norm": 0.7724908538126755, "learning_rate": 9.837517396543799e-06, - "loss": 0.7377, + "loss": 0.6358, "step": 91 }, { "epoch": 0.5317919075144508, - "grad_norm": 0.6854419049253191, + "grad_norm": 1.6095008914897926, "learning_rate": 9.82890290504585e-06, - "loss": 0.7204, + "loss": 0.9213, "step": 92 }, { "epoch": 0.5375722543352601, - "grad_norm": 0.9781279283861866, + "grad_norm": 0.9894464003328666, "learning_rate": 9.82006988266542e-06, - "loss": 0.924, + "loss": 0.8624, "step": 93 }, { "epoch": 0.5433526011560693, - "grad_norm": 0.9782316300254307, + "grad_norm": 0.7889218642430385, "learning_rate": 9.811018729138892e-06, - "loss": 1.0374, + "loss": 0.7508, "step": 94 }, { "epoch": 0.5491329479768786, - "grad_norm": 0.9982601558000294, + "grad_norm": 1.0722731395513763, "learning_rate": 9.801749854074122e-06, - "loss": 0.8366, + "loss": 0.7923, "step": 95 }, { "epoch": 0.5549132947976878, - "grad_norm": 0.8305948258770038, + "grad_norm": 1.0436817532057514, "learning_rate": 9.792263676931906e-06, - "loss": 0.7175, + "loss": 0.7715, "step": 96 }, { "epoch": 0.5606936416184971, - "grad_norm": 0.7030335335365292, + "grad_norm": 1.7989526448931117, "learning_rate": 9.78256062700699e-06, - "loss": 0.7123, + "loss": 1.2198, "step": 97 }, { "epoch": 0.5664739884393064, - "grad_norm": 0.706003197595692, + "grad_norm": 1.0674303681403066, "learning_rate": 9.772641143408652e-06, - "loss": 0.7211, + "loss": 0.7883, "step": 98 }, { "epoch": 0.5722543352601156, - "grad_norm": 0.7849334413450759, + "grad_norm": 0.9353488034991134, "learning_rate": 9.762505675040826e-06, - "loss": 0.7539, + "loss": 0.7686, "step": 99 }, { "epoch": 0.5780346820809249, - "grad_norm": 3.6761886807353137, + "grad_norm": 1.1674201445045589, "learning_rate": 9.752154680581783e-06, - "loss": 0.6494, + "loss": 0.995, "step": 100 }, { "epoch": 0.5838150289017341, - "grad_norm": 0.7807197922428346, + "grad_norm": 0.9466404615356364, "learning_rate": 9.741588628463384e-06, - "loss": 0.7134, + "loss": 0.8552, "step": 101 }, { "epoch": 0.5895953757225434, - "grad_norm": 0.7632334516305908, + "grad_norm": 0.7681659885617845, "learning_rate": 9.730807996849864e-06, - "loss": 0.6662, + "loss": 0.6232, "step": 102 }, { "epoch": 0.5953757225433526, - "grad_norm": 0.8451679526800531, + "grad_norm": 1.0787984383542042, "learning_rate": 9.719813273616216e-06, - "loss": 0.7518, + "loss": 0.9384, "step": 103 }, { "epoch": 0.6011560693641619, - "grad_norm": 0.8989668499217826, + "grad_norm": 1.163335897640102, "learning_rate": 9.70860495632609e-06, - "loss": 0.8596, + "loss": 1.0792, "step": 104 }, { "epoch": 0.6069364161849711, - "grad_norm": 0.844314990838292, + "grad_norm": 1.131274450256789, "learning_rate": 9.697183552209289e-06, - "loss": 0.7271, + "loss": 0.8781, "step": 105 }, { "epoch": 0.6127167630057804, - "grad_norm": 0.9028200521575998, + "grad_norm": 1.0095856194543398, "learning_rate": 9.68554957813881e-06, - "loss": 0.7853, + "loss": 0.8006, "step": 106 }, { "epoch": 0.6184971098265896, - "grad_norm": 0.8371176283258334, + "grad_norm": 0.835619336255323, "learning_rate": 9.673703560607459e-06, - "loss": 0.7162, + "loss": 0.6371, "step": 107 }, { "epoch": 0.6242774566473989, - "grad_norm": 0.8275153796290823, + "grad_norm": 0.7655234428141927, "learning_rate": 9.661646035704009e-06, - "loss": 0.7313, + "loss": 0.5344, "step": 108 }, { "epoch": 0.630057803468208, - "grad_norm": 0.9030218537849709, + "grad_norm": 0.9520070072269193, "learning_rate": 9.649377549088962e-06, - "loss": 0.8741, + "loss": 0.8094, "step": 109 }, { "epoch": 0.6358381502890174, - "grad_norm": 0.8873056298834013, + "grad_norm": 0.9429087170034833, "learning_rate": 9.636898655969837e-06, - "loss": 0.8216, + "loss": 0.7412, "step": 110 }, { "epoch": 0.6416184971098265, - "grad_norm": 1.0574331855159385, + "grad_norm": 0.9882443151038491, "learning_rate": 9.62420992107605e-06, - "loss": 0.8499, + "loss": 0.864, "step": 111 }, { "epoch": 0.6473988439306358, - "grad_norm": 0.6709941792795421, + "grad_norm": 0.9980586995279209, "learning_rate": 9.61131191863336e-06, - "loss": 0.6977, + "loss": 0.883, "step": 112 }, { "epoch": 0.653179190751445, - "grad_norm": 1.021656041099865, + "grad_norm": 0.8143392097546589, "learning_rate": 9.598205232337882e-06, - "loss": 0.9163, + "loss": 0.593, "step": 113 }, { "epoch": 0.6589595375722543, - "grad_norm": 0.8250260244592214, + "grad_norm": 0.8759873715277585, "learning_rate": 9.584890455329667e-06, - "loss": 0.7709, + "loss": 0.7435, "step": 114 }, { "epoch": 0.6647398843930635, - "grad_norm": 0.793931542867849, + "grad_norm": 0.787760238489214, "learning_rate": 9.571368190165865e-06, - "loss": 0.7113, + "loss": 0.6313, "step": 115 }, { "epoch": 0.6705202312138728, - "grad_norm": 0.9754802271233276, + "grad_norm": 0.7813584309524949, "learning_rate": 9.557639048793453e-06, - "loss": 0.9266, + "loss": 0.5329, "step": 116 }, { "epoch": 0.6763005780346821, - "grad_norm": 0.7769622044727718, + "grad_norm": 0.810792296710866, "learning_rate": 9.543703652521543e-06, - "loss": 0.799, + "loss": 0.6099, "step": 117 }, { "epoch": 0.6820809248554913, - "grad_norm": 0.7244784605356199, + "grad_norm": 0.8539955958073703, "learning_rate": 9.52956263199327e-06, - "loss": 0.7489, + "loss": 0.7579, "step": 118 }, { "epoch": 0.6878612716763006, - "grad_norm": 0.7104597019333324, + "grad_norm": 0.973039298272916, "learning_rate": 9.515216627157238e-06, - "loss": 0.7662, + "loss": 0.7422, "step": 119 }, { "epoch": 0.6936416184971098, - "grad_norm": 0.8517106378930308, + "grad_norm": 0.8789927069109355, "learning_rate": 9.500666287238573e-06, - "loss": 0.7974, + "loss": 0.7703, "step": 120 }, { "epoch": 0.6994219653179191, - "grad_norm": 0.6943395529233615, + "grad_norm": 0.8280305404585776, "learning_rate": 9.485912270709542e-06, - "loss": 0.7254, + "loss": 0.5447, "step": 121 }, { "epoch": 0.7052023121387283, - "grad_norm": 0.6914087429469479, + "grad_norm": 0.8306195346423975, "learning_rate": 9.470955245259742e-06, - "loss": 0.7508, + "loss": 0.55, "step": 122 }, { "epoch": 0.7109826589595376, - "grad_norm": 0.7826622618008323, + "grad_norm": 0.8569125077265108, "learning_rate": 9.455795887765896e-06, - "loss": 0.825, + "loss": 0.7363, "step": 123 }, { "epoch": 0.7167630057803468, - "grad_norm": 1.1355838214209095, + "grad_norm": 0.8571238902612902, "learning_rate": 9.440434884261216e-06, - "loss": 0.7426, + "loss": 0.6441, "step": 124 }, { "epoch": 0.7225433526011561, - "grad_norm": 0.6322644005827122, + "grad_norm": 0.8036597031305626, "learning_rate": 9.424872929904359e-06, - "loss": 0.6198, + "loss": 0.6873, "step": 125 }, { "epoch": 0.7283236994219653, - "grad_norm": 0.7362155540142553, + "grad_norm": 0.9237720491861983, "learning_rate": 9.409110728947964e-06, - "loss": 0.7477, + "loss": 0.7905, "step": 126 }, { "epoch": 0.7341040462427746, - "grad_norm": 0.83852942563382, + "grad_norm": 0.9973443566645532, "learning_rate": 9.393148994706785e-06, - "loss": 0.7004, + "loss": 0.4917, "step": 127 }, { "epoch": 0.7398843930635838, - "grad_norm": 0.6748981862594124, + "grad_norm": 1.0006509192435291, "learning_rate": 9.376988449525405e-06, - "loss": 0.5962, + "loss": 0.8303, "step": 128 }, { "epoch": 0.7456647398843931, - "grad_norm": 1.0188783299334387, + "grad_norm": 0.9847352318496524, "learning_rate": 9.360629824745558e-06, - "loss": 0.8826, + "loss": 0.8072, "step": 129 }, { "epoch": 0.7514450867052023, - "grad_norm": 0.7432721538845581, + "grad_norm": 0.9057800124682823, "learning_rate": 9.344073860673016e-06, - "loss": 0.6724, + "loss": 0.7909, "step": 130 }, { "epoch": 0.7572254335260116, - "grad_norm": 0.6103487166007125, + "grad_norm": 0.751051206899431, "learning_rate": 9.327321306544097e-06, - "loss": 0.639, + "loss": 0.6188, "step": 131 }, { "epoch": 0.7630057803468208, - "grad_norm": 0.8945254680236359, + "grad_norm": 0.7918961715149377, "learning_rate": 9.310372920491761e-06, - "loss": 0.8035, + "loss": 0.5697, "step": 132 }, { "epoch": 0.7687861271676301, - "grad_norm": 0.8488865300229482, + "grad_norm": 0.8912684513395774, "learning_rate": 9.293229469511293e-06, - "loss": 0.9083, + "loss": 0.7477, "step": 133 }, { "epoch": 0.7745664739884393, - "grad_norm": 0.6756399548937257, + "grad_norm": 0.6633704205093338, "learning_rate": 9.275891729425595e-06, - "loss": 0.7156, + "loss": 0.5831, "step": 134 }, { "epoch": 0.7803468208092486, - "grad_norm": 0.7184301845822525, + "grad_norm": 0.8899258986178338, "learning_rate": 9.25836048485008e-06, - "loss": 0.7173, + "loss": 0.7249, "step": 135 }, { "epoch": 0.7861271676300579, - "grad_norm": 0.9317951335803887, + "grad_norm": 0.7490887846503816, "learning_rate": 9.240636529157158e-06, - "loss": 0.8955, + "loss": 0.6534, "step": 136 }, { "epoch": 0.791907514450867, - "grad_norm": 0.6935745149751389, + "grad_norm": 1.3911616822761665, "learning_rate": 9.22272066444034e-06, - "loss": 0.7382, + "loss": 1.1257, "step": 137 }, { "epoch": 0.7976878612716763, - "grad_norm": 0.6161623212880647, + "grad_norm": 0.8807323209346695, "learning_rate": 9.204613701477935e-06, - "loss": 0.6026, + "loss": 0.5506, "step": 138 }, { "epoch": 0.8034682080924855, - "grad_norm": 0.8437926156424123, + "grad_norm": 1.2063249188448404, "learning_rate": 9.186316459696359e-06, - "loss": 0.7998, + "loss": 1.0353, "step": 139 }, { "epoch": 0.8092485549132948, - "grad_norm": 1.1707771174001835, + "grad_norm": 0.780960970851516, "learning_rate": 9.167829767133047e-06, - "loss": 0.8504, + "loss": 0.6501, "step": 140 }, { "epoch": 0.815028901734104, - "grad_norm": 0.853766834777908, + "grad_norm": 1.0465279583342073, "learning_rate": 9.149154460398993e-06, - "loss": 0.7923, + "loss": 1.0094, "step": 141 }, { "epoch": 0.8208092485549133, - "grad_norm": 0.8468174728471449, + "grad_norm": 0.8493522401893265, "learning_rate": 9.130291384640873e-06, - "loss": 0.6997, + "loss": 0.6026, "step": 142 }, { "epoch": 0.8265895953757225, - "grad_norm": 0.7921502381441966, + "grad_norm": 1.0520248613531298, "learning_rate": 9.111241393502814e-06, - "loss": 0.8331, + "loss": 0.9218, "step": 143 }, { "epoch": 0.8323699421965318, - "grad_norm": 0.9161796068427475, + "grad_norm": 0.8062150660900307, "learning_rate": 9.092005349087754e-06, - "loss": 0.9119, + "loss": 0.5588, "step": 144 }, { "epoch": 0.838150289017341, - "grad_norm": 0.8632742492898611, + "grad_norm": 1.1800864030349798, "learning_rate": 9.072584121918426e-06, - "loss": 0.7559, + "loss": 0.8445, "step": 145 }, { "epoch": 0.8439306358381503, - "grad_norm": 0.692533183756357, + "grad_norm": 1.204141856153108, "learning_rate": 9.052978590897964e-06, - "loss": 0.6906, + "loss": 0.8043, "step": 146 }, { "epoch": 0.8497109826589595, - "grad_norm": 0.6977551180379914, + "grad_norm": 0.8162766248184291, "learning_rate": 9.033189643270139e-06, - "loss": 0.7512, + "loss": 0.6661, "step": 147 }, { "epoch": 0.8554913294797688, - "grad_norm": 0.9171792960669404, + "grad_norm": 1.0427374615350444, "learning_rate": 9.013218174579189e-06, - "loss": 0.7806, + "loss": 0.8038, "step": 148 }, { "epoch": 0.861271676300578, - "grad_norm": 1.0581091004774794, + "grad_norm": 0.9354749929394746, "learning_rate": 8.993065088629304e-06, - "loss": 0.8988, + "loss": 0.7768, "step": 149 }, { "epoch": 0.8670520231213873, - "grad_norm": 0.6691099185467246, + "grad_norm": 0.9287721633918639, "learning_rate": 8.972731297443722e-06, - "loss": 0.6422, + "loss": 0.731, "step": 150 }, { "epoch": 0.8728323699421965, - "grad_norm": 0.7521912970247356, + "grad_norm": 0.9230583505002304, "learning_rate": 8.95221772122345e-06, - "loss": 0.6641, + "loss": 0.6313, "step": 151 }, { "epoch": 0.8786127167630058, - "grad_norm": 0.9303337252800314, + "grad_norm": 0.9349454478962814, "learning_rate": 8.931525288305633e-06, - "loss": 0.8101, + "loss": 0.8083, "step": 152 }, { "epoch": 0.884393063583815, - "grad_norm": 0.7519577210478734, + "grad_norm": 0.913057550393148, "learning_rate": 8.910654935121528e-06, - "loss": 0.7043, + "loss": 0.7647, "step": 153 }, { "epoch": 0.8901734104046243, - "grad_norm": 0.6428049747490524, + "grad_norm": 0.7348197855556597, "learning_rate": 8.889607606154132e-06, - "loss": 0.6431, + "loss": 0.5773, "step": 154 }, { "epoch": 0.8959537572254336, - "grad_norm": 0.693906731694735, + "grad_norm": 0.7867588221031618, "learning_rate": 8.868384253895445e-06, - "loss": 0.6837, + "loss": 0.7065, "step": 155 }, { "epoch": 0.9017341040462428, - "grad_norm": 0.6183649988892536, + "grad_norm": 0.8228491320872959, "learning_rate": 8.846985838803357e-06, - "loss": 0.6175, + "loss": 0.7271, "step": 156 }, { "epoch": 0.9075144508670521, - "grad_norm": 0.8331580784240926, + "grad_norm": 1.3053568158332713, "learning_rate": 8.825413329258187e-06, - "loss": 0.7155, + "loss": 0.9825, "step": 157 }, { "epoch": 0.9132947976878613, - "grad_norm": 0.7253442766088543, + "grad_norm": 1.0945947320675422, "learning_rate": 8.803667701518857e-06, - "loss": 0.6564, + "loss": 0.9142, "step": 158 }, { "epoch": 0.9190751445086706, - "grad_norm": 0.8610636535683205, + "grad_norm": 0.9764593529348718, "learning_rate": 8.781749939678712e-06, - "loss": 0.823, + "loss": 0.7848, "step": 159 }, { "epoch": 0.9248554913294798, - "grad_norm": 0.6591293339802592, + "grad_norm": 0.6941029727730416, "learning_rate": 8.759661035620992e-06, - "loss": 0.6539, + "loss": 0.5887, "step": 160 }, { "epoch": 0.930635838150289, - "grad_norm": 0.7548758291288769, + "grad_norm": 0.7925590817706383, "learning_rate": 8.73740198897393e-06, - "loss": 0.6444, + "loss": 0.6934, "step": 161 }, { "epoch": 0.9364161849710982, - "grad_norm": 1.131439269762958, + "grad_norm": 0.7656873002992562, "learning_rate": 8.714973807065525e-06, - "loss": 0.7512, + "loss": 0.6853, "step": 162 }, { "epoch": 0.9421965317919075, - "grad_norm": 0.9844357900489483, + "grad_norm": 1.1161472512069044, "learning_rate": 8.69237750487796e-06, - "loss": 0.6915, + "loss": 0.853, "step": 163 }, { "epoch": 0.9479768786127167, - "grad_norm": 0.712529866098214, + "grad_norm": 0.9749363576847155, "learning_rate": 8.669614105001652e-06, - "loss": 0.7822, + "loss": 0.8424, "step": 164 }, { "epoch": 0.953757225433526, - "grad_norm": 0.8200345751340958, + "grad_norm": 0.8325202621241643, "learning_rate": 8.646684637588992e-06, - "loss": 0.7891, + "loss": 0.8077, "step": 165 }, { "epoch": 0.9595375722543352, - "grad_norm": 0.861370522678107, + "grad_norm": 0.9783393336805141, "learning_rate": 8.623590140307715e-06, - "loss": 0.656, + "loss": 0.7791, "step": 166 }, { "epoch": 0.9653179190751445, - "grad_norm": 0.8672087735680497, + "grad_norm": 0.8080744052126594, "learning_rate": 8.600331658293948e-06, - "loss": 0.7101, + "loss": 0.6518, "step": 167 }, { "epoch": 0.9710982658959537, - "grad_norm": 0.7600294500783626, + "grad_norm": 1.1474927961761585, "learning_rate": 8.576910244104905e-06, - "loss": 0.796, + "loss": 0.8716, "step": 168 }, { "epoch": 0.976878612716763, - "grad_norm": 0.7729597442643854, + "grad_norm": 0.9719680493249003, "learning_rate": 8.553326957671264e-06, - "loss": 0.7136, + "loss": 0.8288, "step": 169 }, { "epoch": 0.9826589595375722, - "grad_norm": 0.701072160894904, + "grad_norm": 0.8706832814278217, "learning_rate": 8.529582866249187e-06, - "loss": 0.8065, + "loss": 0.592, "step": 170 }, { "epoch": 0.9884393063583815, - "grad_norm": 0.5925613316023062, + "grad_norm": 0.929595095010918, "learning_rate": 8.50567904437203e-06, - "loss": 0.5756, + "loss": 0.7816, "step": 171 }, { "epoch": 0.9942196531791907, - "grad_norm": 0.7820436692457375, + "grad_norm": 0.7657958882007067, "learning_rate": 8.48161657380172e-06, - "loss": 0.6814, + "loss": 0.4877, "step": 172 }, { "epoch": 1.0, - "grad_norm": 0.8236411546051227, + "grad_norm": 1.1436020534304703, "learning_rate": 8.457396543479787e-06, - "loss": 0.8232, + "loss": 0.8026, "step": 173 }, { "epoch": 1.0057803468208093, - "grad_norm": 0.7330585578803988, + "grad_norm": 0.7855851232058355, "learning_rate": 8.433020049478093e-06, - "loss": 0.6667, + "loss": 0.5871, "step": 174 }, { "epoch": 1.0115606936416186, - "grad_norm": 0.6946768946250981, + "grad_norm": 0.7646534714069891, "learning_rate": 8.408488194949229e-06, - "loss": 0.666, + "loss": 0.6586, "step": 175 }, { "epoch": 1.0173410404624277, - "grad_norm": 0.6440746694570499, + "grad_norm": 0.8230539636052089, "learning_rate": 8.383802090076589e-06, - "loss": 0.6788, + "loss": 0.6081, "step": 176 }, { "epoch": 1.023121387283237, - "grad_norm": 0.9466375715702998, + "grad_norm": 0.9538635888680288, "learning_rate": 8.358962852024128e-06, - "loss": 0.5753, + "loss": 0.7103, "step": 177 }, { "epoch": 1.0289017341040463, - "grad_norm": 0.729911905767581, + "grad_norm": 0.9031383623963593, "learning_rate": 8.333971604885817e-06, - "loss": 0.6307, + "loss": 0.5453, "step": 178 }, { "epoch": 1.0346820809248556, - "grad_norm": 0.6904725332273605, + "grad_norm": 0.8947067399258345, "learning_rate": 8.308829479634753e-06, - "loss": 0.5312, + "loss": 0.6581, "step": 179 }, { "epoch": 1.0404624277456647, - "grad_norm": 0.6534648081872814, + "grad_norm": 0.7906077257603986, "learning_rate": 8.283537614071987e-06, - "loss": 0.5647, + "loss": 0.6724, "step": 180 }, { "epoch": 1.046242774566474, - "grad_norm": 0.6390187064815486, + "grad_norm": 0.9275191131931334, "learning_rate": 8.258097152775045e-06, - "loss": 0.5973, + "loss": 0.6303, "step": 181 }, { "epoch": 1.0520231213872833, - "grad_norm": 0.6459178187162624, + "grad_norm": 1.0819371958975426, "learning_rate": 8.232509247046106e-06, - "loss": 0.6151, + "loss": 0.7358, "step": 182 }, { "epoch": 1.0578034682080926, - "grad_norm": 0.8089745359730721, + "grad_norm": 0.9172402539596392, "learning_rate": 8.206775054859914e-06, - "loss": 0.6653, + "loss": 0.6278, "step": 183 }, { "epoch": 1.0635838150289016, - "grad_norm": 0.7704942142349642, + "grad_norm": 0.7024408514337848, "learning_rate": 8.180895740811381e-06, - "loss": 0.5279, + "loss": 0.5198, "step": 184 }, { "epoch": 1.069364161849711, - "grad_norm": 0.8526091896188928, + "grad_norm": 0.6236935477897357, "learning_rate": 8.154872476062868e-06, - "loss": 0.7164, + "loss": 0.4613, "step": 185 }, { "epoch": 1.0751445086705202, - "grad_norm": 0.9129705077010957, + "grad_norm": 0.8702316152168408, "learning_rate": 8.128706438291193e-06, - "loss": 0.5609, + "loss": 0.5644, "step": 186 }, { "epoch": 1.0809248554913296, - "grad_norm": 0.6913786990222495, + "grad_norm": 0.8493352120847084, "learning_rate": 8.102398811634338e-06, - "loss": 0.6435, + "loss": 0.6492, "step": 187 }, { "epoch": 1.0867052023121386, - "grad_norm": 0.6157362513970346, + "grad_norm": 0.77177764951958, "learning_rate": 8.075950786637847e-06, - "loss": 0.6101, + "loss": 0.6817, "step": 188 }, { "epoch": 1.092485549132948, - "grad_norm": 0.7791529870282913, + "grad_norm": 0.7604368990026783, "learning_rate": 8.049363560200972e-06, - "loss": 0.6, + "loss": 0.4197, "step": 189 }, { "epoch": 1.0982658959537572, - "grad_norm": 0.7272600842835277, + "grad_norm": 0.7547200781480733, "learning_rate": 8.022638335522484e-06, - "loss": 0.5431, + "loss": 0.4693, "step": 190 }, { "epoch": 1.1040462427745665, - "grad_norm": 0.7897989566376171, + "grad_norm": 0.8363425113790128, "learning_rate": 7.995776322046236e-06, - "loss": 0.8725, + "loss": 0.5314, "step": 191 }, { "epoch": 1.1098265895953756, - "grad_norm": 0.8057843444247929, + "grad_norm": 0.7732661043588892, "learning_rate": 7.968778735406426e-06, - "loss": 0.8086, + "loss": 0.5379, "step": 192 }, { "epoch": 1.115606936416185, - "grad_norm": 0.9463678364280204, + "grad_norm": 0.7255116220582496, "learning_rate": 7.941646797372584e-06, - "loss": 0.7486, + "loss": 0.5787, "step": 193 }, { "epoch": 1.1213872832369942, - "grad_norm": 0.6848932352162628, + "grad_norm": 0.7847798977214011, "learning_rate": 7.914381735794282e-06, - "loss": 0.602, + "loss": 0.6169, "step": 194 }, { "epoch": 1.1271676300578035, - "grad_norm": 0.6790628184875152, + "grad_norm": 0.6595825059371818, "learning_rate": 7.886984784545565e-06, - "loss": 0.5483, + "loss": 0.4591, "step": 195 }, { "epoch": 1.1329479768786128, - "grad_norm": 0.6778767427169146, + "grad_norm": 0.7278796876236617, "learning_rate": 7.859457183469119e-06, - "loss": 0.6552, + "loss": 0.5845, "step": 196 }, { "epoch": 1.138728323699422, - "grad_norm": 0.8137551601742041, + "grad_norm": 0.7528110685404278, "learning_rate": 7.831800178320153e-06, - "loss": 0.9145, + "loss": 0.5812, "step": 197 }, { "epoch": 1.1445086705202312, - "grad_norm": 0.5973889529894271, + "grad_norm": 0.890267436582, "learning_rate": 7.804015020710028e-06, - "loss": 0.5632, + "loss": 0.6572, "step": 198 }, { "epoch": 1.1502890173410405, - "grad_norm": 0.6294985184604123, + "grad_norm": 0.8254440289935021, "learning_rate": 7.776102968049616e-06, - "loss": 0.5479, + "loss": 0.6755, "step": 199 }, { "epoch": 1.1560693641618498, - "grad_norm": 0.746529782704009, + "grad_norm": 0.6507477500486915, "learning_rate": 7.748065283492397e-06, - "loss": 0.5878, + "loss": 0.554, "step": 200 }, { "epoch": 1.1618497109826589, - "grad_norm": 0.7779139219101336, + "grad_norm": 0.6283177198784258, "learning_rate": 7.719903235877289e-06, - "loss": 0.7232, + "loss": 0.5776, "step": 201 }, { "epoch": 1.1676300578034682, - "grad_norm": 0.6482680670587473, + "grad_norm": 0.8918223872817956, "learning_rate": 7.691618099671235e-06, - "loss": 0.6254, + "loss": 0.8012, "step": 202 }, { "epoch": 1.1734104046242775, - "grad_norm": 0.7938649268425443, + "grad_norm": 0.9344902364876053, "learning_rate": 7.663211154911523e-06, - "loss": 0.6094, + "loss": 0.8115, "step": 203 }, { "epoch": 1.1791907514450868, - "grad_norm": 1.463550262254941, + "grad_norm": 0.7794391355469076, "learning_rate": 7.634683687147857e-06, - "loss": 0.5983, + "loss": 0.6142, "step": 204 }, { "epoch": 1.1849710982658959, - "grad_norm": 0.7149523848762201, + "grad_norm": 0.8659277724734541, "learning_rate": 7.606036987384185e-06, - "loss": 0.622, + "loss": 0.7304, "step": 205 }, { "epoch": 1.1907514450867052, - "grad_norm": 0.7116091272777324, + "grad_norm": 0.9502452996067565, "learning_rate": 7.577272352020269e-06, - "loss": 0.5982, + "loss": 0.8352, "step": 206 }, { "epoch": 1.1965317919075145, - "grad_norm": 0.7533870640980641, + "grad_norm": 0.7759351158211815, "learning_rate": 7.5483910827930186e-06, - "loss": 0.5959, + "loss": 0.645, "step": 207 }, { "epoch": 1.2023121387283238, - "grad_norm": 0.6586197594120814, + "grad_norm": 0.6696108650382663, "learning_rate": 7.519394486717583e-06, - "loss": 0.512, + "loss": 0.5238, "step": 208 }, { "epoch": 1.208092485549133, - "grad_norm": 0.8449855704082665, + "grad_norm": 0.8786759679795678, "learning_rate": 7.4902838760282024e-06, - "loss": 0.7867, + "loss": 0.74, "step": 209 }, { "epoch": 1.2138728323699421, - "grad_norm": 0.7864108058193743, + "grad_norm": 0.7066794429138108, "learning_rate": 7.461060568118822e-06, - "loss": 0.573, + "loss": 0.5637, "step": 210 }, { "epoch": 1.2196531791907514, - "grad_norm": 0.713488945394281, + "grad_norm": 0.7820317919250585, "learning_rate": 7.43172588548347e-06, - "loss": 0.6831, + "loss": 0.6129, "step": 211 }, { "epoch": 1.2254335260115607, - "grad_norm": 0.6686237008986486, + "grad_norm": 0.8213464787047494, "learning_rate": 7.402281155656415e-06, - "loss": 0.7005, + "loss": 0.7279, "step": 212 }, { "epoch": 1.2312138728323698, - "grad_norm": 0.7778732829123279, + "grad_norm": 1.2811020609856818, "learning_rate": 7.3727277111520864e-06, - "loss": 0.7207, + "loss": 1.027, "step": 213 }, { "epoch": 1.2369942196531791, - "grad_norm": 0.7707906968898907, + "grad_norm": 0.7203326251914638, "learning_rate": 7.343066889404769e-06, - "loss": 0.6786, + "loss": 0.5618, "step": 214 }, { "epoch": 1.2427745664739884, - "grad_norm": 0.6532276866723493, + "grad_norm": 0.7490042739512657, "learning_rate": 7.313300032708081e-06, - "loss": 0.617, + "loss": 0.4959, "step": 215 }, { "epoch": 1.2485549132947977, - "grad_norm": 0.9291907058395319, + "grad_norm": 0.8330199251865329, "learning_rate": 7.283428488154227e-06, - "loss": 0.7249, + "loss": 0.6419, "step": 216 }, { "epoch": 1.254335260115607, - "grad_norm": 0.7875932252872507, + "grad_norm": 0.8109431295258052, "learning_rate": 7.253453607573037e-06, - "loss": 0.7395, + "loss": 0.5864, "step": 217 }, { "epoch": 1.260115606936416, - "grad_norm": 0.8014022002969718, + "grad_norm": 0.6789438861293353, "learning_rate": 7.223376747470792e-06, - "loss": 0.6499, + "loss": 0.5748, "step": 218 }, { "epoch": 1.2658959537572254, - "grad_norm": 0.7045574520911717, + "grad_norm": 0.7905554276439895, "learning_rate": 7.193199268968825e-06, - "loss": 0.6274, + "loss": 0.4877, "step": 219 }, { "epoch": 1.2716763005780347, - "grad_norm": 0.7474810639105346, + "grad_norm": 0.7672450268723763, "learning_rate": 7.162922537741937e-06, - "loss": 0.758, + "loss": 0.488, "step": 220 }, { "epoch": 1.2774566473988438, - "grad_norm": 0.6298329208923502, + "grad_norm": 0.7113314734848389, "learning_rate": 7.1325479239565875e-06, - "loss": 0.5848, + "loss": 0.496, "step": 221 }, { "epoch": 1.2832369942196533, - "grad_norm": 0.6382265876528984, + "grad_norm": 0.775371647892662, "learning_rate": 7.102076802208887e-06, - "loss": 0.6311, + "loss": 0.5555, "step": 222 }, { "epoch": 1.2890173410404624, - "grad_norm": 0.8581527995655507, + "grad_norm": 0.8155548293800693, "learning_rate": 7.071510551462395e-06, - "loss": 0.5809, + "loss": 0.6871, "step": 223 }, { "epoch": 1.2947976878612717, - "grad_norm": 0.6435054997232182, + "grad_norm": 0.869775047071274, "learning_rate": 7.040850554985706e-06, - "loss": 0.6554, + "loss": 0.6442, "step": 224 }, { "epoch": 1.300578034682081, - "grad_norm": 0.7069082328493708, + "grad_norm": 0.8896294697712317, "learning_rate": 7.01009820028986e-06, - "loss": 0.825, + "loss": 0.6255, "step": 225 }, { "epoch": 1.30635838150289, - "grad_norm": 0.6518137576025003, + "grad_norm": 0.8155789387389749, "learning_rate": 6.9792548790655465e-06, - "loss": 0.6556, + "loss": 0.7973, "step": 226 }, { "epoch": 1.3121387283236994, - "grad_norm": 0.7783627296839134, + "grad_norm": 0.7230343705020649, "learning_rate": 6.948321987120122e-06, - "loss": 0.7556, + "loss": 0.5421, "step": 227 }, { "epoch": 1.3179190751445087, - "grad_norm": 0.724833483004496, + "grad_norm": 0.7291631821760047, "learning_rate": 6.9173009243144485e-06, - "loss": 0.6996, + "loss": 0.431, "step": 228 }, { "epoch": 1.323699421965318, - "grad_norm": 0.6581459996747386, + "grad_norm": 0.7287156885787672, "learning_rate": 6.886193094499537e-06, - "loss": 0.5828, + "loss": 0.558, "step": 229 }, { "epoch": 1.3294797687861273, - "grad_norm": 0.8459594697685823, + "grad_norm": 1.0724029075135588, "learning_rate": 6.854999905453022e-06, - "loss": 0.6812, + "loss": 0.5622, "step": 230 }, { "epoch": 1.3352601156069364, - "grad_norm": 0.6004468175928412, + "grad_norm": 0.8822778651565606, "learning_rate": 6.823722768815446e-06, - "loss": 0.4808, + "loss": 0.6397, "step": 231 }, { "epoch": 1.3410404624277457, - "grad_norm": 0.8282645817161267, + "grad_norm": 0.7985828585892485, "learning_rate": 6.792363100026383e-06, - "loss": 0.6486, + "loss": 0.6297, "step": 232 }, { "epoch": 1.346820809248555, - "grad_norm": 0.66517260435302, + "grad_norm": 0.670328316646438, "learning_rate": 6.760922318260384e-06, - "loss": 0.6522, + "loss": 0.6461, "step": 233 }, { "epoch": 1.352601156069364, - "grad_norm": 0.5807981788583886, + "grad_norm": 0.7789795421340042, "learning_rate": 6.729401846362743e-06, - "loss": 0.5449, + "loss": 0.6102, "step": 234 }, { "epoch": 1.3583815028901733, - "grad_norm": 0.5902371061520595, + "grad_norm": 1.05298645942576, "learning_rate": 6.697803110785115e-06, - "loss": 0.5506, + "loss": 0.7174, "step": 235 }, { "epoch": 1.3641618497109826, - "grad_norm": 0.6150334998706332, + "grad_norm": 0.6391669736154241, "learning_rate": 6.666127541520958e-06, - "loss": 0.5158, + "loss": 0.4992, "step": 236 }, { "epoch": 1.369942196531792, - "grad_norm": 0.6897841056630322, + "grad_norm": 0.8465130878040216, "learning_rate": 6.634376572040826e-06, - "loss": 0.7367, + "loss": 0.5653, "step": 237 }, { "epoch": 1.3757225433526012, - "grad_norm": 0.668191514030944, + "grad_norm": 0.8552412125002662, "learning_rate": 6.602551639227486e-06, - "loss": 0.6003, + "loss": 0.4559, "step": 238 }, { "epoch": 1.3815028901734103, - "grad_norm": 0.6482070367483068, + "grad_norm": 1.0924656658631755, "learning_rate": 6.570654183310901e-06, - "loss": 0.5534, + "loss": 0.8661, "step": 239 }, { "epoch": 1.3872832369942196, - "grad_norm": 1.478028386990335, + "grad_norm": 0.7794676129362818, "learning_rate": 6.538685647803049e-06, - "loss": 0.8822, + "loss": 0.6934, "step": 240 }, { "epoch": 1.393063583815029, - "grad_norm": 0.5653608270449567, + "grad_norm": 0.7171760865328232, "learning_rate": 6.506647479432604e-06, - "loss": 0.5803, + "loss": 0.5452, "step": 241 }, { "epoch": 1.3988439306358382, - "grad_norm": 0.6559185317520129, + "grad_norm": 0.8512263697543654, "learning_rate": 6.474541128079452e-06, - "loss": 0.5799, + "loss": 0.539, "step": 242 }, { "epoch": 1.4046242774566475, - "grad_norm": 0.7110078220778856, + "grad_norm": 0.7418655135071943, "learning_rate": 6.442368046709087e-06, - "loss": 0.5281, + "loss": 0.5844, "step": 243 }, { "epoch": 1.4104046242774566, - "grad_norm": 0.6182909344208828, + "grad_norm": 0.7995078110330098, "learning_rate": 6.410129691306855e-06, - "loss": 0.5398, + "loss": 0.4744, "step": 244 }, { "epoch": 1.416184971098266, - "grad_norm": 0.6814885840461983, + "grad_norm": 0.9331606469598925, "learning_rate": 6.377827520812061e-06, - "loss": 0.7044, + "loss": 0.7489, "step": 245 }, { "epoch": 1.4219653179190752, - "grad_norm": 0.5832538378556845, + "grad_norm": 0.8268022509923469, "learning_rate": 6.34546299705195e-06, - "loss": 0.4371, + "loss": 0.7182, "step": 246 }, { "epoch": 1.4277456647398843, - "grad_norm": 0.6803189711086476, + "grad_norm": 0.6686099679963994, "learning_rate": 6.31303758467555e-06, - "loss": 0.6393, + "loss": 0.4366, "step": 247 }, { "epoch": 1.4335260115606936, - "grad_norm": 0.6053394812193783, + "grad_norm": 0.8037659222821476, "learning_rate": 6.280552751087384e-06, - "loss": 0.5314, + "loss": 0.6675, "step": 248 }, { "epoch": 1.439306358381503, - "grad_norm": 0.6681704114033317, + "grad_norm": 0.9009423195262899, "learning_rate": 6.248009966381074e-06, - "loss": 0.6185, + "loss": 0.6042, "step": 249 }, { "epoch": 1.4450867052023122, - "grad_norm": 0.6757039362728535, + "grad_norm": 0.9903928033155953, "learning_rate": 6.215410703272805e-06, - "loss": 0.627, + "loss": 0.7219, "step": 250 }, { "epoch": 1.4508670520231215, - "grad_norm": 0.6114539462595463, + "grad_norm": 0.6591691197785381, "learning_rate": 6.182756437034677e-06, - "loss": 0.5209, + "loss": 0.4372, "step": 251 }, { "epoch": 1.4566473988439306, - "grad_norm": 0.8330405171146432, + "grad_norm": 0.695635459295935, "learning_rate": 6.150048645427944e-06, - "loss": 0.8047, + "loss": 0.496, "step": 252 }, { "epoch": 1.4624277456647399, - "grad_norm": 0.656139553434916, + "grad_norm": 0.5945863816394287, "learning_rate": 6.117288808636141e-06, - "loss": 0.5713, + "loss": 0.5174, "step": 253 }, { "epoch": 1.4682080924855492, - "grad_norm": 0.8569781392487273, + "grad_norm": 0.7602116565399197, "learning_rate": 6.0844784091980955e-06, - "loss": 0.6085, + "loss": 0.645, "step": 254 }, { "epoch": 1.4739884393063583, - "grad_norm": 0.5538339672083201, + "grad_norm": 0.7090698522334424, "learning_rate": 6.05161893194083e-06, - "loss": 0.58, + "loss": 0.5536, "step": 255 }, { "epoch": 1.4797687861271676, - "grad_norm": 1.472559963634999, + "grad_norm": 0.725729229031602, "learning_rate": 6.018711863912381e-06, - "loss": 0.6159, + "loss": 0.6477, "step": 256 }, { "epoch": 1.4855491329479769, - "grad_norm": 0.7185599108174412, + "grad_norm": 0.6357974105573619, "learning_rate": 5.985758694314486e-06, - "loss": 0.7419, + "loss": 0.5217, "step": 257 }, { "epoch": 1.4913294797687862, - "grad_norm": 0.6311434379513406, + "grad_norm": 1.0443530505670549, "learning_rate": 5.9527609144352075e-06, - "loss": 0.6612, + "loss": 0.6152, "step": 258 }, { "epoch": 1.4971098265895955, - "grad_norm": 0.7097391328621409, + "grad_norm": 0.8202669154391847, "learning_rate": 5.9197200175814275e-06, - "loss": 0.5787, + "loss": 0.5029, "step": 259 }, { "epoch": 1.5028901734104045, - "grad_norm": 0.7927118174990451, + "grad_norm": 0.6862183739194273, "learning_rate": 5.8866374990112785e-06, - "loss": 0.6979, + "loss": 0.5836, "step": 260 }, { "epoch": 1.5086705202312138, - "grad_norm": 0.5574774651086752, + "grad_norm": 0.898525799970463, "learning_rate": 5.853514855866481e-06, - "loss": 0.5459, + "loss": 0.8393, "step": 261 }, { "epoch": 1.5144508670520231, - "grad_norm": 0.6920042466264643, + "grad_norm": 0.9820808227078155, "learning_rate": 5.8203535871045735e-06, - "loss": 0.624, + "loss": 0.6995, "step": 262 }, { "epoch": 1.5202312138728322, - "grad_norm": 0.6212505344697492, + "grad_norm": 0.7519418179376166, "learning_rate": 5.787155193431095e-06, - "loss": 0.4608, + "loss": 0.5861, "step": 263 }, { "epoch": 1.5260115606936417, - "grad_norm": 0.5149097528709641, + "grad_norm": 0.9393739826771281, "learning_rate": 5.75392117723166e-06, - "loss": 0.5498, + "loss": 0.5841, "step": 264 }, { "epoch": 1.5317919075144508, - "grad_norm": 0.6477265686222932, + "grad_norm": 0.7596859297052929, "learning_rate": 5.7206530425039785e-06, - "loss": 0.5609, + "loss": 0.5483, "step": 265 }, { "epoch": 1.5375722543352601, - "grad_norm": 0.5255664769204426, + "grad_norm": 0.7596810301974631, "learning_rate": 5.687352294789776e-06, - "loss": 0.603, + "loss": 0.601, "step": 266 }, { "epoch": 1.5433526011560694, - "grad_norm": 0.5915239937509899, + "grad_norm": 0.8936156970474353, "learning_rate": 5.654020441106682e-06, - "loss": 0.5864, + "loss": 0.8358, "step": 267 }, { "epoch": 1.5491329479768785, - "grad_norm": 0.6676526153340506, + "grad_norm": 0.8101430405700003, "learning_rate": 5.620658989880011e-06, - "loss": 0.6021, + "loss": 0.5428, "step": 268 }, { "epoch": 1.5549132947976878, - "grad_norm": 0.600671127368277, + "grad_norm": 0.7262001629259932, "learning_rate": 5.587269450874513e-06, - "loss": 0.6013, + "loss": 0.4783, "step": 269 }, { "epoch": 1.560693641618497, - "grad_norm": 0.5844697566376813, + "grad_norm": 0.8075662054671324, "learning_rate": 5.5538533351260395e-06, - "loss": 0.529, + "loss": 0.6543, "step": 270 }, { "epoch": 1.5664739884393064, - "grad_norm": 0.6882910968205374, + "grad_norm": 1.0619586513294879, "learning_rate": 5.52041215487317e-06, - "loss": 0.7213, + "loss": 0.6379, "step": 271 }, { "epoch": 1.5722543352601157, - "grad_norm": 0.6287255808288411, + "grad_norm": 0.6915488345799145, "learning_rate": 5.486947423488774e-06, - "loss": 0.5056, + "loss": 0.5354, "step": 272 }, { "epoch": 1.5780346820809248, - "grad_norm": 0.7365490633929451, + "grad_norm": 0.9595574551452887, "learning_rate": 5.453460655411515e-06, - "loss": 0.6733, + "loss": 0.6359, "step": 273 }, { "epoch": 1.583815028901734, - "grad_norm": 0.6158229226620566, + "grad_norm": 0.6746217408473637, "learning_rate": 5.4199533660773276e-06, - "loss": 0.4215, + "loss": 0.442, "step": 274 }, { "epoch": 1.5895953757225434, - "grad_norm": 0.6098363365047987, + "grad_norm": 0.6875447362464598, "learning_rate": 5.3864270718508305e-06, - "loss": 0.558, + "loss": 0.6612, "step": 275 }, { "epoch": 1.5953757225433525, - "grad_norm": 0.6749997930429508, + "grad_norm": 1.0006017308905897, "learning_rate": 5.352883289956701e-06, - "loss": 0.549, + "loss": 0.7303, "step": 276 }, { "epoch": 1.601156069364162, - "grad_norm": 0.6219987255688616, + "grad_norm": 0.8962934512743116, "learning_rate": 5.319323538411021e-06, - "loss": 0.5607, + "loss": 0.6552, "step": 277 }, { "epoch": 1.606936416184971, - "grad_norm": 0.6534834061456652, + "grad_norm": 1.3267780546969365, "learning_rate": 5.285749335952573e-06, - "loss": 0.6234, + "loss": 0.7785, "step": 278 }, { "epoch": 1.6127167630057804, - "grad_norm": 0.6477067038037368, + "grad_norm": 0.9202137432547965, "learning_rate": 5.252162201974112e-06, - "loss": 0.6205, + "loss": 0.6865, "step": 279 }, { "epoch": 1.6184971098265897, - "grad_norm": 0.9044848085787234, + "grad_norm": 1.0125080800080466, "learning_rate": 5.218563656453609e-06, - "loss": 0.6134, + "loss": 0.645, "step": 280 }, { "epoch": 1.6242774566473988, - "grad_norm": 0.6395032966455607, + "grad_norm": 0.7442509258922682, "learning_rate": 5.184955219885457e-06, - "loss": 0.5684, + "loss": 0.512, "step": 281 }, { "epoch": 1.630057803468208, - "grad_norm": 0.7583127138048714, + "grad_norm": 0.9033544155384335, "learning_rate": 5.15133841321167e-06, - "loss": 0.7791, + "loss": 0.694, "step": 282 }, { "epoch": 1.6358381502890174, - "grad_norm": 0.854312985647277, + "grad_norm": 0.6900225057997373, "learning_rate": 5.117714757753045e-06, - "loss": 0.6784, + "loss": 0.4881, "step": 283 }, { "epoch": 1.6416184971098264, - "grad_norm": 0.7280985217866222, + "grad_norm": 0.6742757865306379, "learning_rate": 5.084085775140324e-06, - "loss": 0.6255, + "loss": 0.5103, "step": 284 }, { "epoch": 1.647398843930636, - "grad_norm": 0.6793910694159482, + "grad_norm": 0.7059532473644158, "learning_rate": 5.050452987245325e-06, - "loss": 0.5657, + "loss": 0.4349, "step": 285 }, { "epoch": 1.653179190751445, - "grad_norm": 0.6515150296173606, + "grad_norm": 0.6564844507129209, "learning_rate": 5.016817916112075e-06, - "loss": 0.5571, + "loss": 0.5836, "step": 286 }, { "epoch": 1.6589595375722543, - "grad_norm": 0.7363630099803015, + "grad_norm": 0.6951495372130232, "learning_rate": 4.9831820838879255e-06, - "loss": 0.662, + "loss": 0.624, "step": 287 }, { "epoch": 1.6647398843930636, - "grad_norm": 0.6850360945395945, + "grad_norm": 0.8002765691543589, "learning_rate": 4.949547012754676e-06, - "loss": 0.6451, + "loss": 0.6247, "step": 288 }, { "epoch": 1.6705202312138727, - "grad_norm": 0.7394402316965297, + "grad_norm": 0.6945205161978173, "learning_rate": 4.915914224859677e-06, - "loss": 0.6139, + "loss": 0.5317, "step": 289 }, { "epoch": 1.6763005780346822, - "grad_norm": 0.6190408305988775, + "grad_norm": 1.0063599327693165, "learning_rate": 4.882285242246958e-06, - "loss": 0.4859, + "loss": 0.8422, "step": 290 }, { "epoch": 1.6820809248554913, - "grad_norm": 1.315159775153852, + "grad_norm": 0.687707175040322, "learning_rate": 4.848661586788334e-06, - "loss": 0.729, + "loss": 0.6278, "step": 291 }, { "epoch": 1.6878612716763006, - "grad_norm": 0.7487282006485257, + "grad_norm": 0.894386154392975, "learning_rate": 4.815044780114544e-06, - "loss": 0.68, + "loss": 0.7546, "step": 292 }, { "epoch": 1.69364161849711, - "grad_norm": 0.7498129968033505, + "grad_norm": 0.6603346113796635, "learning_rate": 4.781436343546392e-06, - "loss": 0.687, + "loss": 0.6156, "step": 293 }, { "epoch": 1.699421965317919, - "grad_norm": 0.6758544868018486, + "grad_norm": 0.886454781643024, "learning_rate": 4.7478377980258885e-06, - "loss": 0.7497, + "loss": 0.5466, "step": 294 }, { "epoch": 1.7052023121387283, - "grad_norm": 0.7633724359669948, + "grad_norm": 0.5996464506334955, "learning_rate": 4.714250664047428e-06, - "loss": 0.7237, + "loss": 0.5173, "step": 295 }, { "epoch": 1.7109826589595376, - "grad_norm": 0.7072277738512419, + "grad_norm": 0.74394974483986, "learning_rate": 4.68067646158898e-06, - "loss": 0.6202, + "loss": 0.5619, "step": 296 }, { "epoch": 1.7167630057803467, - "grad_norm": 0.6287286977794924, + "grad_norm": 0.8641214674891129, "learning_rate": 4.647116710043302e-06, - "loss": 0.6002, + "loss": 0.5851, "step": 297 }, { "epoch": 1.7225433526011562, - "grad_norm": 0.7733929992401346, + "grad_norm": 0.6473738175450051, "learning_rate": 4.613572928149172e-06, - "loss": 0.6308, + "loss": 0.5937, "step": 298 }, { "epoch": 1.7283236994219653, - "grad_norm": 0.5628771279192218, + "grad_norm": 0.7911023989217947, "learning_rate": 4.580046633922675e-06, - "loss": 0.5094, + "loss": 0.6134, "step": 299 }, { "epoch": 1.7341040462427746, - "grad_norm": 0.5481137194588697, + "grad_norm": 0.7199248015046684, "learning_rate": 4.546539344588486e-06, - "loss": 0.5015, + "loss": 0.4924, "step": 300 }, { "epoch": 1.739884393063584, - "grad_norm": 0.6933912131629874, + "grad_norm": 0.8611464036629595, "learning_rate": 4.513052576511227e-06, - "loss": 0.7205, + "loss": 0.5567, "step": 301 }, { "epoch": 1.745664739884393, - "grad_norm": 0.5763974173505807, + "grad_norm": 0.9153747256071443, "learning_rate": 4.47958784512683e-06, - "loss": 0.6069, + "loss": 0.5906, "step": 302 }, { "epoch": 1.7514450867052023, - "grad_norm": 0.49587843076014865, + "grad_norm": 0.5570439958075847, "learning_rate": 4.446146664873961e-06, - "loss": 0.5101, + "loss": 0.5195, "step": 303 }, { "epoch": 1.7572254335260116, - "grad_norm": 1.3731966079165951, + "grad_norm": 0.9281845417649991, "learning_rate": 4.41273054912549e-06, - "loss": 0.5849, + "loss": 0.6441, "step": 304 }, { "epoch": 1.7630057803468207, - "grad_norm": 0.7806243602412305, + "grad_norm": 0.788201691868334, "learning_rate": 4.379341010119992e-06, - "loss": 0.6318, + "loss": 0.6661, "step": 305 }, { "epoch": 1.7687861271676302, - "grad_norm": 0.9760423705832353, + "grad_norm": 0.9529288524795544, "learning_rate": 4.34597955889332e-06, - "loss": 0.7286, + "loss": 0.7531, "step": 306 }, { "epoch": 1.7745664739884393, - "grad_norm": 0.6486163940935261, + "grad_norm": 0.7779671173146703, "learning_rate": 4.312647705210226e-06, - "loss": 0.6663, + "loss": 0.6333, "step": 307 }, { "epoch": 1.7803468208092486, - "grad_norm": 0.693902741919887, + "grad_norm": 0.932290786792518, "learning_rate": 4.279346957496023e-06, - "loss": 0.6049, + "loss": 0.6065, "step": 308 }, { "epoch": 1.7861271676300579, - "grad_norm": 0.8543384276360204, + "grad_norm": 0.8791757542275365, "learning_rate": 4.246078822768339e-06, - "loss": 0.8289, + "loss": 0.593, "step": 309 }, { "epoch": 1.791907514450867, - "grad_norm": 0.7161042253323527, + "grad_norm": 0.8229021097129374, "learning_rate": 4.212844806568906e-06, - "loss": 0.6728, + "loss": 0.7998, "step": 310 }, { "epoch": 1.7976878612716765, - "grad_norm": 0.7384136827970472, + "grad_norm": 0.7027823693838428, "learning_rate": 4.17964641289543e-06, - "loss": 0.6617, + "loss": 0.4884, "step": 311 }, { "epoch": 1.8034682080924855, - "grad_norm": 0.5333323971401331, + "grad_norm": 0.5638484231065025, "learning_rate": 4.1464851441335215e-06, - "loss": 0.4945, + "loss": 0.5413, "step": 312 }, { "epoch": 1.8092485549132948, - "grad_norm": 0.7296175768296973, + "grad_norm": 0.7504131374172438, "learning_rate": 4.113362500988722e-06, - "loss": 0.6816, + "loss": 0.5272, "step": 313 }, { "epoch": 1.8150289017341041, - "grad_norm": 0.7042220930352442, + "grad_norm": 0.6740181052595745, "learning_rate": 4.080279982418574e-06, - "loss": 0.6159, + "loss": 0.5272, "step": 314 }, { "epoch": 1.8208092485549132, - "grad_norm": 0.6967338173420907, + "grad_norm": 0.6590261922608894, "learning_rate": 4.047239085564794e-06, - "loss": 0.6197, + "loss": 0.5185, "step": 315 }, { "epoch": 1.8265895953757225, - "grad_norm": 0.665707168138219, + "grad_norm": 0.6354968708763732, "learning_rate": 4.014241305685514e-06, - "loss": 0.6668, + "loss": 0.5437, "step": 316 }, { "epoch": 1.8323699421965318, - "grad_norm": 0.6159113597910466, + "grad_norm": 0.6088112809354961, "learning_rate": 3.98128813608762e-06, - "loss": 0.5254, + "loss": 0.4712, "step": 317 }, { "epoch": 1.838150289017341, - "grad_norm": 0.6932400491954773, + "grad_norm": 0.6374602552721457, "learning_rate": 3.948381068059171e-06, - "loss": 0.4996, + "loss": 0.5713, "step": 318 }, { "epoch": 1.8439306358381504, - "grad_norm": 0.8447991818763431, + "grad_norm": 0.7213250401566385, "learning_rate": 3.915521590801907e-06, - "loss": 0.6843, + "loss": 0.4117, "step": 319 }, { "epoch": 1.8497109826589595, - "grad_norm": 0.7133507213085982, + "grad_norm": 0.720008546371465, "learning_rate": 3.88271119136386e-06, - "loss": 0.6865, + "loss": 0.5622, "step": 320 }, { "epoch": 1.8554913294797688, - "grad_norm": 0.7144531197197936, + "grad_norm": 0.8620941640593333, "learning_rate": 3.849951354572057e-06, - "loss": 0.7027, + "loss": 0.7968, "step": 321 }, { "epoch": 1.861271676300578, - "grad_norm": 0.6206053123328196, + "grad_norm": 1.056732583118651, "learning_rate": 3.817243562965324e-06, - "loss": 0.5549, + "loss": 0.6582, "step": 322 }, { "epoch": 1.8670520231213872, - "grad_norm": 0.8373060466314365, + "grad_norm": 0.6578509759019269, "learning_rate": 3.7845892967271963e-06, - "loss": 0.7472, + "loss": 0.541, "step": 323 }, { "epoch": 1.8728323699421965, - "grad_norm": 0.650701693000165, + "grad_norm": 0.6057121810732851, "learning_rate": 3.7519900336189265e-06, - "loss": 0.5064, + "loss": 0.4909, "step": 324 }, { "epoch": 1.8786127167630058, - "grad_norm": 0.6541281290116006, + "grad_norm": 0.7726511998943779, "learning_rate": 3.7194472489126176e-06, - "loss": 0.5565, + "loss": 0.6899, "step": 325 }, { "epoch": 1.8843930635838149, - "grad_norm": 0.6965779346499603, + "grad_norm": 0.8109214205884281, "learning_rate": 3.686962415324452e-06, - "loss": 0.5364, + "loss": 0.5595, "step": 326 }, { "epoch": 1.8901734104046244, - "grad_norm": 0.7072305070269004, + "grad_norm": 1.0646480130164524, "learning_rate": 3.6545370029480515e-06, - "loss": 0.6338, + "loss": 0.606, "step": 327 }, { "epoch": 1.8959537572254335, - "grad_norm": 0.5913567231700574, + "grad_norm": 0.705969500214819, "learning_rate": 3.6221724791879406e-06, - "loss": 0.599, + "loss": 0.7158, "step": 328 }, { "epoch": 1.9017341040462428, - "grad_norm": 0.6031535031072002, + "grad_norm": 0.8864721580498878, "learning_rate": 3.5898703086931474e-06, - "loss": 0.6207, + "loss": 0.7594, "step": 329 }, { "epoch": 1.907514450867052, - "grad_norm": 0.6878489234806915, + "grad_norm": 0.7557686350987431, "learning_rate": 3.557631953290914e-06, - "loss": 0.613, + "loss": 0.4859, "step": 330 }, { "epoch": 1.9132947976878611, - "grad_norm": 0.6319103343525578, + "grad_norm": 1.0351089698366276, "learning_rate": 3.5254588719205494e-06, - "loss": 0.5556, + "loss": 0.7508, "step": 331 }, { "epoch": 1.9190751445086707, - "grad_norm": 0.6372301808239768, + "grad_norm": 0.7085730332833243, "learning_rate": 3.4933525205673977e-06, - "loss": 0.5828, + "loss": 0.6455, "step": 332 }, { "epoch": 1.9248554913294798, - "grad_norm": 0.5433466791208938, + "grad_norm": 0.9200031613758001, "learning_rate": 3.461314352196952e-06, - "loss": 0.5706, + "loss": 0.5611, "step": 333 }, { "epoch": 1.930635838150289, - "grad_norm": 0.8079989116456336, + "grad_norm": 0.6429597743080722, "learning_rate": 3.429345816689101e-06, - "loss": 0.6919, + "loss": 0.4856, "step": 334 }, { "epoch": 1.9364161849710984, - "grad_norm": 0.6259865195671478, + "grad_norm": 0.9298164493352193, "learning_rate": 3.397448360772516e-06, - "loss": 0.6851, + "loss": 0.5274, "step": 335 }, { "epoch": 1.9421965317919074, - "grad_norm": 0.6299895359395308, + "grad_norm": 0.614550273953037, "learning_rate": 3.365623427959175e-06, - "loss": 0.5846, + "loss": 0.6099, "step": 336 }, { "epoch": 1.9479768786127167, - "grad_norm": 0.8579386338530067, + "grad_norm": 0.8692548756868597, "learning_rate": 3.3338724584790427e-06, - "loss": 0.7579, + "loss": 0.6163, "step": 337 }, { "epoch": 1.953757225433526, - "grad_norm": 0.5543425764470207, + "grad_norm": 0.7544063125807522, "learning_rate": 3.302196889214886e-06, - "loss": 0.5719, + "loss": 0.6332, "step": 338 }, { "epoch": 1.9595375722543351, - "grad_norm": 0.7219911920632239, + "grad_norm": 0.794867920805677, "learning_rate": 3.270598153637259e-06, - "loss": 0.6064, + "loss": 0.6922, "step": 339 }, { "epoch": 1.9653179190751446, - "grad_norm": 0.6815192695527847, + "grad_norm": 0.7609575009320511, "learning_rate": 3.239077681739618e-06, - "loss": 0.5839, + "loss": 0.5322, "step": 340 }, { "epoch": 1.9710982658959537, - "grad_norm": 0.6657484441757247, + "grad_norm": 0.6900161454249018, "learning_rate": 3.2076368999736175e-06, - "loss": 0.6297, + "loss": 0.5402, "step": 341 }, { "epoch": 1.976878612716763, - "grad_norm": 1.089888903882402, + "grad_norm": 0.8506561948618055, "learning_rate": 3.176277231184556e-06, - "loss": 0.6139, + "loss": 0.6172, "step": 342 }, { "epoch": 1.9826589595375723, - "grad_norm": 0.5748793577735618, + "grad_norm": 0.6157710849420072, "learning_rate": 3.14500009454698e-06, - "loss": 0.5663, + "loss": 0.5905, "step": 343 }, { "epoch": 1.9884393063583814, - "grad_norm": 0.6825700603419808, + "grad_norm": 0.6934477411408183, "learning_rate": 3.1138069055004628e-06, - "loss": 0.7335, + "loss": 0.5682, "step": 344 }, { "epoch": 1.9942196531791907, - "grad_norm": 0.6683638368010179, + "grad_norm": 0.7277401390122208, "learning_rate": 3.0826990756855528e-06, - "loss": 0.7046, + "loss": 0.5831, "step": 345 }, { "epoch": 2.0, - "grad_norm": 0.5315456540935085, + "grad_norm": 0.7255445232592516, "learning_rate": 3.0516780128798794e-06, - "loss": 0.5607, + "loss": 0.4161, "step": 346 }, { "epoch": 2.005780346820809, - "grad_norm": 0.5684154918398037, + "grad_norm": 0.7986659641333982, "learning_rate": 3.020745120934455e-06, - "loss": 0.5098, + "loss": 0.6286, "step": 347 }, { "epoch": 2.0115606936416186, - "grad_norm": 0.641750759514859, + "grad_norm": 0.6679979342656628, "learning_rate": 2.989901799710142e-06, - "loss": 0.607, + "loss": 0.4853, "step": 348 }, { "epoch": 2.0173410404624277, - "grad_norm": 0.575050648435363, + "grad_norm": 0.9986811632523198, "learning_rate": 2.9591494450142957e-06, - "loss": 0.538, + "loss": 0.6472, "step": 349 }, { "epoch": 2.023121387283237, - "grad_norm": 0.6058803545728428, + "grad_norm": 0.8311828194343122, "learning_rate": 2.9284894485376057e-06, - "loss": 0.4878, + "loss": 0.4739, "step": 350 }, { "epoch": 2.0289017341040463, - "grad_norm": 0.6307587625882025, + "grad_norm": 0.6678215245683566, "learning_rate": 2.8979231977911127e-06, - "loss": 0.5576, + "loss": 0.3969, "step": 351 }, { "epoch": 2.0346820809248554, - "grad_norm": 0.508580655308353, + "grad_norm": 0.6708618173706526, "learning_rate": 2.8674520760434145e-06, - "loss": 0.4493, + "loss": 0.6153, "step": 352 }, { "epoch": 2.040462427745665, - "grad_norm": 0.5954153002168113, + "grad_norm": 0.8515187777361045, "learning_rate": 2.8370774622580644e-06, - "loss": 0.4852, + "loss": 0.5254, "step": 353 }, { "epoch": 2.046242774566474, - "grad_norm": 0.5357226362074531, + "grad_norm": 0.6016593646695987, "learning_rate": 2.806800731031176e-06, - "loss": 0.4343, + "loss": 0.4817, "step": 354 }, { "epoch": 2.052023121387283, - "grad_norm": 0.5383881393206567, + "grad_norm": 0.7719152929747317, "learning_rate": 2.7766232525292104e-06, - "loss": 0.4945, + "loss": 0.5555, "step": 355 }, { "epoch": 2.0578034682080926, - "grad_norm": 0.746028884996199, + "grad_norm": 0.6612185750070453, "learning_rate": 2.746546392426963e-06, - "loss": 0.6516, + "loss": 0.5369, "step": 356 }, { "epoch": 2.0635838150289016, - "grad_norm": 0.6092364042997683, + "grad_norm": 0.7359611638945163, "learning_rate": 2.7165715118457735e-06, - "loss": 0.4545, + "loss": 0.5443, "step": 357 }, { "epoch": 2.069364161849711, - "grad_norm": 0.6927146873134861, + "grad_norm": 0.7388558303950356, "learning_rate": 2.6866999672919198e-06, - "loss": 0.5629, + "loss": 0.5849, "step": 358 }, { "epoch": 2.0751445086705202, - "grad_norm": 0.6275293386174793, + "grad_norm": 0.8550602709603156, "learning_rate": 2.656933110595233e-06, - "loss": 0.5077, + "loss": 0.6346, "step": 359 }, { "epoch": 2.0809248554913293, - "grad_norm": 0.9186098220090658, + "grad_norm": 0.6813466801259701, "learning_rate": 2.6272722888479152e-06, - "loss": 0.5118, + "loss": 0.4533, "step": 360 }, { "epoch": 2.086705202312139, - "grad_norm": 0.5621280732308401, + "grad_norm": 0.9184018851746331, "learning_rate": 2.5977188443435874e-06, - "loss": 0.5013, + "loss": 0.5008, "step": 361 }, { "epoch": 2.092485549132948, - "grad_norm": 1.7564780123883166, + "grad_norm": 0.8469746620559047, "learning_rate": 2.5682741145165325e-06, - "loss": 0.6418, + "loss": 0.5568, "step": 362 }, { "epoch": 2.098265895953757, - "grad_norm": 0.6277196375137782, + "grad_norm": 0.8248199191803897, "learning_rate": 2.53893943188118e-06, - "loss": 0.4274, + "loss": 0.5211, "step": 363 }, { "epoch": 2.1040462427745665, - "grad_norm": 0.6095177489897702, + "grad_norm": 0.6499874475711263, "learning_rate": 2.5097161239717975e-06, - "loss": 0.4569, + "loss": 0.5555, "step": 364 }, { "epoch": 2.1098265895953756, - "grad_norm": 0.5754914160245718, + "grad_norm": 0.7813430444756865, "learning_rate": 2.4806055132824186e-06, - "loss": 0.5343, + "loss": 0.6461, "step": 365 }, { "epoch": 2.115606936416185, - "grad_norm": 0.6279968548558946, + "grad_norm": 0.84725387994917, "learning_rate": 2.4516089172069852e-06, - "loss": 0.5135, + "loss": 0.496, "step": 366 }, { "epoch": 2.121387283236994, - "grad_norm": 0.6225824537375481, + "grad_norm": 0.873004350480058, "learning_rate": 2.422727647979734e-06, - "loss": 0.5482, + "loss": 0.6613, "step": 367 }, { "epoch": 2.1271676300578033, - "grad_norm": 0.6798967734438467, + "grad_norm": 0.675120156410227, "learning_rate": 2.393963012615817e-06, - "loss": 0.5724, + "loss": 0.4495, "step": 368 }, { "epoch": 2.132947976878613, - "grad_norm": 0.6169394117268941, + "grad_norm": 0.7490115944986748, "learning_rate": 2.365316312852144e-06, - "loss": 0.5377, + "loss": 0.5333, "step": 369 }, { "epoch": 2.138728323699422, - "grad_norm": 0.5413158818542305, + "grad_norm": 0.6570964764736189, "learning_rate": 2.336788845088478e-06, - "loss": 0.4396, + "loss": 0.523, "step": 370 }, { "epoch": 2.1445086705202314, - "grad_norm": 0.5525982766283964, + "grad_norm": 1.5378979727083335, "learning_rate": 2.308381900328767e-06, - "loss": 0.5596, + "loss": 0.4837, "step": 371 }, { "epoch": 2.1502890173410405, - "grad_norm": 0.514986628205561, + "grad_norm": 0.8452714858786144, "learning_rate": 2.2800967641227127e-06, - "loss": 0.6204, + "loss": 0.4845, "step": 372 }, { "epoch": 2.1560693641618496, - "grad_norm": 0.650457231558704, + "grad_norm": 0.7851766322063914, "learning_rate": 2.2519347165076067e-06, - "loss": 0.5517, + "loss": 0.5861, "step": 373 }, { "epoch": 2.161849710982659, - "grad_norm": 0.5584034500753343, + "grad_norm": 0.7653788839856677, "learning_rate": 2.223897031950386e-06, - "loss": 0.5557, + "loss": 0.4178, "step": 374 }, { "epoch": 2.167630057803468, - "grad_norm": 0.585394823431568, + "grad_norm": 0.7712566207030392, "learning_rate": 2.195984979289974e-06, - "loss": 0.6333, + "loss": 0.4706, "step": 375 }, { "epoch": 2.1734104046242773, - "grad_norm": 0.6044138538566343, + "grad_norm": 0.6331367645198177, "learning_rate": 2.1681998216798476e-06, - "loss": 0.5585, + "loss": 0.4277, "step": 376 }, { "epoch": 2.179190751445087, - "grad_norm": 0.6543883235567041, + "grad_norm": 0.6644887559230318, "learning_rate": 2.140542816530882e-06, - "loss": 0.5951, + "loss": 0.4447, "step": 377 }, { "epoch": 2.184971098265896, - "grad_norm": 0.5982897707090032, + "grad_norm": 0.6124699875496277, "learning_rate": 2.1130152154544346e-06, - "loss": 0.5302, + "loss": 0.5207, "step": 378 }, { "epoch": 2.1907514450867054, - "grad_norm": 0.7191124285251307, + "grad_norm": 0.6887031182813522, "learning_rate": 2.0856182642057182e-06, - "loss": 0.641, + "loss": 0.5315, "step": 379 }, { "epoch": 2.1965317919075145, - "grad_norm": 0.5389834615259756, + "grad_norm": 0.67468259231777, "learning_rate": 2.058353202627417e-06, - "loss": 0.4903, + "loss": 0.4879, "step": 380 }, { "epoch": 2.2023121387283235, - "grad_norm": 0.6851138108926675, + "grad_norm": 0.7788323939968849, "learning_rate": 2.0312212645935755e-06, - "loss": 0.5627, + "loss": 0.497, "step": 381 }, { "epoch": 2.208092485549133, - "grad_norm": 0.5197231185111274, + "grad_norm": 0.6709526713707947, "learning_rate": 2.0042236779537668e-06, - "loss": 0.5472, + "loss": 0.6146, "step": 382 }, { "epoch": 2.213872832369942, - "grad_norm": 0.5164429827496514, + "grad_norm": 0.6359997471362626, "learning_rate": 1.977361664477518e-06, - "loss": 0.4583, + "loss": 0.5428, "step": 383 }, { "epoch": 2.2196531791907512, - "grad_norm": 0.6340175873272497, + "grad_norm": 0.8088756710901137, "learning_rate": 1.950636439799029e-06, - "loss": 0.6284, + "loss": 0.5739, "step": 384 }, { "epoch": 2.2254335260115607, - "grad_norm": 0.6491652436113424, + "grad_norm": 0.670342241899264, "learning_rate": 1.924049213362153e-06, - "loss": 0.5686, + "loss": 0.4031, "step": 385 }, { "epoch": 2.23121387283237, - "grad_norm": 0.6294267158763216, + "grad_norm": 0.8131170961600367, "learning_rate": 1.8976011883656632e-06, - "loss": 0.5455, + "loss": 0.5596, "step": 386 }, { "epoch": 2.2369942196531793, - "grad_norm": 0.5899575922299914, + "grad_norm": 0.6561035218947915, "learning_rate": 1.8712935617088067e-06, - "loss": 0.4749, + "loss": 0.5486, "step": 387 }, { "epoch": 2.2427745664739884, - "grad_norm": 0.6303599078993858, + "grad_norm": 0.6820151386870905, "learning_rate": 1.8451275239371337e-06, - "loss": 0.5286, + "loss": 0.6458, "step": 388 }, { "epoch": 2.2485549132947975, - "grad_norm": 0.8006095838627105, + "grad_norm": 0.6098685514723682, "learning_rate": 1.8191042591886198e-06, - "loss": 0.7174, + "loss": 0.4581, "step": 389 }, { "epoch": 2.254335260115607, - "grad_norm": 0.5080290834095371, + "grad_norm": 0.8452109569036743, "learning_rate": 1.7932249451400863e-06, - "loss": 0.4262, + "loss": 0.6904, "step": 390 }, { "epoch": 2.260115606936416, - "grad_norm": 0.6661809112956046, + "grad_norm": 0.6659828594692649, "learning_rate": 1.767490752953896e-06, - "loss": 0.5024, + "loss": 0.3941, "step": 391 }, { "epoch": 2.2658959537572256, - "grad_norm": 0.6638153139955407, + "grad_norm": 0.5756879764229126, "learning_rate": 1.7419028472249566e-06, - "loss": 0.6463, + "loss": 0.5507, "step": 392 }, { "epoch": 2.2716763005780347, - "grad_norm": 1.1183110402363063, + "grad_norm": 0.6604711511382437, "learning_rate": 1.7164623859280144e-06, - "loss": 0.6754, + "loss": 0.5287, "step": 393 }, { "epoch": 2.277456647398844, - "grad_norm": 0.9573431684168563, + "grad_norm": 1.5345932250020973, "learning_rate": 1.6911705203652506e-06, - "loss": 0.5985, + "loss": 0.5368, "step": 394 }, { "epoch": 2.2832369942196533, - "grad_norm": 0.6242174335726994, + "grad_norm": 0.6641692298107834, "learning_rate": 1.6660283951141847e-06, - "loss": 0.6711, + "loss": 0.4684, "step": 395 }, { "epoch": 2.2890173410404624, - "grad_norm": 0.5529433995446528, + "grad_norm": 0.8484405008890961, "learning_rate": 1.641037147975872e-06, - "loss": 0.4153, + "loss": 0.5993, "step": 396 }, { "epoch": 2.294797687861272, - "grad_norm": 0.6060325674323018, + "grad_norm": 0.9366024809353254, "learning_rate": 1.616197909923412e-06, - "loss": 0.3823, + "loss": 0.5794, "step": 397 }, { "epoch": 2.300578034682081, - "grad_norm": 0.5754897781203018, + "grad_norm": 0.5673626418707731, "learning_rate": 1.591511805050772e-06, - "loss": 0.5719, + "loss": 0.4806, "step": 398 }, { "epoch": 2.30635838150289, - "grad_norm": 0.554971267529678, + "grad_norm": 0.7242857939505446, "learning_rate": 1.5669799505219069e-06, - "loss": 0.5268, + "loss": 0.502, "step": 399 }, { "epoch": 2.3121387283236996, - "grad_norm": 0.5646431996196378, + "grad_norm": 0.6060165227720642, "learning_rate": 1.542603456520214e-06, - "loss": 0.4572, + "loss": 0.4041, "step": 400 }, { "epoch": 2.3179190751445087, - "grad_norm": 0.5907367137443995, + "grad_norm": 0.6560686130548388, "learning_rate": 1.5183834261982804e-06, - "loss": 0.5036, + "loss": 0.5262, "step": 401 }, { "epoch": 2.3236994219653178, - "grad_norm": 0.563208594776787, + "grad_norm": 0.6541682337960161, "learning_rate": 1.4943209556279698e-06, - "loss": 0.531, + "loss": 0.5337, "step": 402 }, { "epoch": 2.3294797687861273, - "grad_norm": 0.6168629280334367, + "grad_norm": 0.6410584716703046, "learning_rate": 1.4704171337508144e-06, - "loss": 0.6965, + "loss": 0.5113, "step": 403 }, { "epoch": 2.3352601156069364, - "grad_norm": 0.8573246264679767, + "grad_norm": 0.688298063944939, "learning_rate": 1.4466730423287385e-06, - "loss": 0.5759, + "loss": 0.5875, "step": 404 }, { "epoch": 2.3410404624277454, - "grad_norm": 0.5442707592190321, + "grad_norm": 0.6724254637374641, "learning_rate": 1.423089755895095e-06, - "loss": 0.4777, + "loss": 0.396, "step": 405 }, { "epoch": 2.346820809248555, - "grad_norm": 0.5690201423494262, + "grad_norm": 0.5864952096498138, "learning_rate": 1.399668341706053e-06, - "loss": 0.5007, + "loss": 0.4239, "step": 406 }, { "epoch": 2.352601156069364, - "grad_norm": 0.6902369415191765, + "grad_norm": 0.6178918581100669, "learning_rate": 1.3764098596922865e-06, - "loss": 0.5636, + "loss": 0.5631, "step": 407 }, { "epoch": 2.3583815028901736, - "grad_norm": 0.604233302044609, + "grad_norm": 0.9247148328484425, "learning_rate": 1.3533153624110097e-06, - "loss": 0.6368, + "loss": 0.4963, "step": 408 }, { "epoch": 2.3641618497109826, - "grad_norm": 0.7435259953395013, + "grad_norm": 0.8654766383717402, "learning_rate": 1.3303858949983495e-06, - "loss": 0.6803, + "loss": 0.5263, "step": 409 }, { "epoch": 2.3699421965317917, - "grad_norm": 1.1489500046757974, + "grad_norm": 0.7195607163743138, "learning_rate": 1.3076224951220413e-06, - "loss": 0.5989, + "loss": 0.6329, "step": 410 }, { "epoch": 2.3757225433526012, - "grad_norm": 0.6087279138747027, + "grad_norm": 0.6949276909817108, "learning_rate": 1.2850261929344748e-06, - "loss": 0.5246, + "loss": 0.4078, "step": 411 }, { "epoch": 2.3815028901734103, - "grad_norm": 0.5345619801290881, + "grad_norm": 0.6664530713874662, "learning_rate": 1.2625980110260711e-06, - "loss": 0.4948, + "loss": 0.5412, "step": 412 }, { "epoch": 2.38728323699422, - "grad_norm": 0.5878225913732457, + "grad_norm": 0.7267558282861785, "learning_rate": 1.2403389643790086e-06, - "loss": 0.5019, + "loss": 0.4662, "step": 413 }, { "epoch": 2.393063583815029, - "grad_norm": 0.6389372845791784, + "grad_norm": 0.6623588377036678, "learning_rate": 1.2182500603212882e-06, - "loss": 0.431, + "loss": 0.4992, "step": 414 }, { "epoch": 2.398843930635838, - "grad_norm": 0.5026201883584769, + "grad_norm": 0.6718558580372862, "learning_rate": 1.1963322984811454e-06, - "loss": 0.5098, + "loss": 0.5274, "step": 415 }, { "epoch": 2.4046242774566475, - "grad_norm": 0.7775616323290275, + "grad_norm": 0.9631367487372555, "learning_rate": 1.1745866707418146e-06, - "loss": 0.5866, + "loss": 0.643, "step": 416 }, { "epoch": 2.4104046242774566, - "grad_norm": 0.5676960014718839, + "grad_norm": 0.7063167334871303, "learning_rate": 1.1530141611966438e-06, - "loss": 0.6236, + "loss": 0.4825, "step": 417 }, { "epoch": 2.416184971098266, - "grad_norm": 0.5085080276824953, + "grad_norm": 0.7292394956959912, "learning_rate": 1.1316157461045553e-06, - "loss": 0.4246, + "loss": 0.6665, "step": 418 }, { "epoch": 2.421965317919075, - "grad_norm": 0.6280259456235967, + "grad_norm": 0.926762344420454, "learning_rate": 1.1103923938458677e-06, - "loss": 0.438, + "loss": 0.6534, "step": 419 }, { "epoch": 2.4277456647398843, - "grad_norm": 0.5366162365214106, + "grad_norm": 0.6784779321154496, "learning_rate": 1.0893450648784736e-06, - "loss": 0.5336, + "loss": 0.5104, "step": 420 }, { "epoch": 2.433526011560694, - "grad_norm": 0.7043968367145309, + "grad_norm": 0.7772613387332611, "learning_rate": 1.0684747116943683e-06, - "loss": 0.585, + "loss": 0.3761, "step": 421 }, { "epoch": 2.439306358381503, - "grad_norm": 0.5921575855806483, + "grad_norm": 0.6162486774128405, "learning_rate": 1.04778227877655e-06, - "loss": 0.4999, + "loss": 0.4899, "step": 422 }, { "epoch": 2.445086705202312, - "grad_norm": 0.8283463200935905, + "grad_norm": 0.961701488339332, "learning_rate": 1.0272687025562794e-06, - "loss": 0.8082, + "loss": 0.6117, "step": 423 }, { "epoch": 2.4508670520231215, - "grad_norm": 0.5301743043908443, + "grad_norm": 0.6313815028131994, "learning_rate": 1.006934911370696e-06, - "loss": 0.45, + "loss": 0.4986, "step": 424 }, { "epoch": 2.4566473988439306, - "grad_norm": 0.8808810550471253, + "grad_norm": 0.6603889532264356, "learning_rate": 9.867818254208122e-07, - "loss": 0.6784, + "loss": 0.5369, "step": 425 }, { "epoch": 2.4624277456647397, - "grad_norm": 0.6588800348981321, + "grad_norm": 0.8843950594070151, "learning_rate": 9.668103567298615e-07, - "loss": 0.6731, + "loss": 0.4081, "step": 426 }, { "epoch": 2.468208092485549, - "grad_norm": 0.6725707327488296, + "grad_norm": 0.5618293210160273, "learning_rate": 9.470214091020358e-07, - "loss": 0.566, + "loss": 0.4613, "step": 427 }, { "epoch": 2.4739884393063583, - "grad_norm": 0.5773534593364248, + "grad_norm": 0.7055186578758241, "learning_rate": 9.274158780815767e-07, - "loss": 0.5828, + "loss": 0.6163, "step": 428 }, { "epoch": 2.479768786127168, - "grad_norm": 0.5555280028477948, + "grad_norm": 0.5685837831584806, "learning_rate": 9.079946509122473e-07, - "loss": 0.4959, + "loss": 0.3661, "step": 429 }, { "epoch": 2.485549132947977, - "grad_norm": 0.6190788294082807, + "grad_norm": 0.784214174921459, "learning_rate": 8.887586064971859e-07, - "loss": 0.6247, + "loss": 0.6345, "step": 430 }, { "epoch": 2.491329479768786, - "grad_norm": 0.537715419582133, + "grad_norm": 0.6305599218370685, "learning_rate": 8.697086153591289e-07, - "loss": 0.4283, + "loss": 0.416, "step": 431 }, { "epoch": 2.4971098265895955, - "grad_norm": 0.7493663050353904, + "grad_norm": 0.6688718546399536, "learning_rate": 8.508455396010096e-07, - "loss": 0.6529, + "loss": 0.4818, "step": 432 }, { "epoch": 2.5028901734104045, - "grad_norm": 0.6224875423728089, + "grad_norm": 0.6161066570746109, "learning_rate": 8.321702328669534e-07, - "loss": 0.5323, + "loss": 0.5282, "step": 433 }, { "epoch": 2.508670520231214, - "grad_norm": 0.5788785567581803, + "grad_norm": 0.6566268049892484, "learning_rate": 8.136835403036413e-07, - "loss": 0.4799, + "loss": 0.4428, "step": 434 }, { "epoch": 2.514450867052023, - "grad_norm": 0.7140821524671198, + "grad_norm": 1.006353487708078, "learning_rate": 7.95386298522065e-07, - "loss": 0.4584, + "loss": 0.5345, "step": 435 }, { "epoch": 2.520231213872832, - "grad_norm": 0.5402958107219245, + "grad_norm": 0.6413665105044725, "learning_rate": 7.772793355596597e-07, - "loss": 0.5912, + "loss": 0.4737, "step": 436 }, { "epoch": 2.5260115606936417, - "grad_norm": 0.6108455295418388, + "grad_norm": 0.6293528837134872, "learning_rate": 7.593634708428438e-07, - "loss": 0.5867, + "loss": 0.5381, "step": 437 }, { "epoch": 2.531791907514451, - "grad_norm": 0.5874865003824954, + "grad_norm": 0.6738700311860045, "learning_rate": 7.416395151499223e-07, - "loss": 0.5841, + "loss": 0.5094, "step": 438 }, { "epoch": 2.5375722543352603, - "grad_norm": 0.5856839051093293, + "grad_norm": 0.6898932232980165, "learning_rate": 7.241082705744057e-07, - "loss": 0.4494, + "loss": 0.4759, "step": 439 }, { "epoch": 2.5433526011560694, - "grad_norm": 0.6399543535807632, + "grad_norm": 0.5745027363154075, "learning_rate": 7.067705304887074e-07, - "loss": 0.6127, + "loss": 0.5029, "step": 440 }, { "epoch": 2.5491329479768785, - "grad_norm": 0.6536590065584015, + "grad_norm": 0.6453821301523448, "learning_rate": 6.896270795082394e-07, - "loss": 0.5916, + "loss": 0.5204, "step": 441 }, { "epoch": 2.5549132947976876, - "grad_norm": 0.58199737697071, + "grad_norm": 0.6922347786512014, "learning_rate": 6.726786934559048e-07, - "loss": 0.6094, + "loss": 0.3755, "step": 442 }, { "epoch": 2.560693641618497, - "grad_norm": 0.548212355952789, + "grad_norm": 0.6866975025738358, "learning_rate": 6.559261393269872e-07, - "loss": 0.5028, + "loss": 0.5367, "step": 443 }, { "epoch": 2.5664739884393066, - "grad_norm": 0.5970978636199156, + "grad_norm": 0.6700088622432993, "learning_rate": 6.39370175254444e-07, - "loss": 0.4781, + "loss": 0.3622, "step": 444 }, { "epoch": 2.5722543352601157, - "grad_norm": 0.5705156891583142, + "grad_norm": 0.5402057213183034, "learning_rate": 6.230115504745954e-07, - "loss": 0.5195, + "loss": 0.3243, "step": 445 }, { "epoch": 2.578034682080925, - "grad_norm": 0.5518267393211346, + "grad_norm": 0.7350281504880568, "learning_rate": 6.06851005293217e-07, - "loss": 0.4891, + "loss": 0.5981, "step": 446 }, { "epoch": 2.583815028901734, - "grad_norm": 0.5832356154325689, + "grad_norm": 0.6703540902498925, "learning_rate": 5.908892710520375e-07, - "loss": 0.4374, + "loss": 0.4066, "step": 447 }, { "epoch": 2.5895953757225434, - "grad_norm": 0.5337949568987072, + "grad_norm": 0.7258590657543835, "learning_rate": 5.75127070095643e-07, - "loss": 0.4627, + "loss": 0.4559, "step": 448 }, { "epoch": 2.5953757225433525, - "grad_norm": 0.5739548220958064, + "grad_norm": 0.6704672360240732, "learning_rate": 5.595651157387855e-07, - "loss": 0.504, + "loss": 0.4375, "step": 449 }, { "epoch": 2.601156069364162, - "grad_norm": 0.5538232731901092, + "grad_norm": 0.6579385787797936, "learning_rate": 5.442041122341057e-07, - "loss": 0.5525, + "loss": 0.5388, "step": 450 }, { "epoch": 2.606936416184971, - "grad_norm": 0.5226922049408729, + "grad_norm": 0.8553046424907753, "learning_rate": 5.290447547402594e-07, - "loss": 0.4906, + "loss": 0.7152, "step": 451 }, { "epoch": 2.61271676300578, - "grad_norm": 0.5795099841167338, + "grad_norm": 0.6877484771956439, "learning_rate": 5.14087729290459e-07, - "loss": 0.5554, + "loss": 0.5358, "step": 452 }, { "epoch": 2.6184971098265897, - "grad_norm": 0.576589047489958, + "grad_norm": 0.6498973479008244, "learning_rate": 4.993337127614273e-07, - "loss": 0.6471, + "loss": 0.4932, "step": 453 }, { "epoch": 2.6242774566473988, - "grad_norm": 0.5755403512070262, + "grad_norm": 0.770132063847202, "learning_rate": 4.847833728427636e-07, - "loss": 0.4703, + "loss": 0.6222, "step": 454 }, { "epoch": 2.6300578034682083, - "grad_norm": 0.5882682826679644, + "grad_norm": 0.7044746862566583, "learning_rate": 4.7043736800673254e-07, - "loss": 0.6217, + "loss": 0.3938, "step": 455 }, { "epoch": 2.6358381502890174, - "grad_norm": 0.5565289700153734, + "grad_norm": 0.6702798408184912, "learning_rate": 4.5629634747845764e-07, - "loss": 0.4244, + "loss": 0.4261, "step": 456 }, { "epoch": 2.6416184971098264, - "grad_norm": 0.6031724742315551, + "grad_norm": 1.1172345643942705, "learning_rate": 4.423609512065485e-07, - "loss": 0.556, + "loss": 0.6234, "step": 457 }, { "epoch": 2.647398843930636, - "grad_norm": 0.7911392018178747, + "grad_norm": 0.7659921270327145, "learning_rate": 4.2863180983413744e-07, - "loss": 0.6715, + "loss": 0.6352, "step": 458 }, { "epoch": 2.653179190751445, - "grad_norm": 0.6092093161730765, + "grad_norm": 0.7276018590380591, "learning_rate": 4.1510954467033457e-07, - "loss": 0.6141, + "loss": 0.5297, "step": 459 }, { "epoch": 2.6589595375722546, - "grad_norm": 0.5734786572879191, + "grad_norm": 0.6190001997300761, "learning_rate": 4.0179476766211865e-07, - "loss": 0.5655, + "loss": 0.5252, "step": 460 }, { "epoch": 2.6647398843930636, - "grad_norm": 0.5712325877144385, + "grad_norm": 0.6344762030667028, "learning_rate": 3.8868808136663995e-07, - "loss": 0.5119, + "loss": 0.5306, "step": 461 }, { "epoch": 2.6705202312138727, - "grad_norm": 0.7041948468929646, + "grad_norm": 0.5759828123397588, "learning_rate": 3.757900789239516e-07, - "loss": 0.581, + "loss": 0.5379, "step": 462 }, { "epoch": 2.6763005780346822, - "grad_norm": 0.5969385403566825, + "grad_norm": 0.6866121028719112, "learning_rate": 3.631013440301645e-07, - "loss": 0.4646, + "loss": 0.4335, "step": 463 }, { "epoch": 2.6820809248554913, - "grad_norm": 0.5057426242616975, + "grad_norm": 0.6463962239601395, "learning_rate": 3.50622450911039e-07, - "loss": 0.3963, + "loss": 0.5133, "step": 464 }, { "epoch": 2.687861271676301, - "grad_norm": 0.5159079345294197, + "grad_norm": 0.653588946548295, "learning_rate": 3.383539642959915e-07, - "loss": 0.5408, + "loss": 0.5014, "step": 465 }, { "epoch": 2.69364161849711, - "grad_norm": 0.7666687962973947, + "grad_norm": 0.7652801505954091, "learning_rate": 3.262964393925433e-07, - "loss": 0.6175, + "loss": 0.4342, "step": 466 }, { "epoch": 2.699421965317919, - "grad_norm": 0.5002821996507574, + "grad_norm": 0.722324748328315, "learning_rate": 3.144504218611899e-07, - "loss": 0.4331, + "loss": 0.4172, "step": 467 }, { "epoch": 2.705202312138728, - "grad_norm": 0.587477426621882, + "grad_norm": 0.7449073337272253, "learning_rate": 3.028164477907125e-07, - "loss": 0.5323, + "loss": 0.5408, "step": 468 }, { "epoch": 2.7109826589595376, - "grad_norm": 0.6588519684187805, + "grad_norm": 1.2797339305753486, "learning_rate": 2.913950436739116e-07, - "loss": 0.4924, + "loss": 0.5264, "step": 469 }, { "epoch": 2.7167630057803467, - "grad_norm": 0.5815438656010812, + "grad_norm": 0.8483318778291473, "learning_rate": 2.8018672638378486e-07, - "loss": 0.532, + "loss": 0.5537, "step": 470 }, { "epoch": 2.722543352601156, - "grad_norm": 0.7544369562719615, + "grad_norm": 0.6475041734782216, "learning_rate": 2.6919200315013606e-07, - "loss": 0.5593, + "loss": 0.598, "step": 471 }, { "epoch": 2.7283236994219653, - "grad_norm": 0.6301787404273509, + "grad_norm": 0.6088717228752758, "learning_rate": 2.5841137153661765e-07, - "loss": 0.5001, + "loss": 0.3216, "step": 472 }, { "epoch": 2.7341040462427744, - "grad_norm": 0.5993245109288315, + "grad_norm": 0.6831655278694656, "learning_rate": 2.4784531941821675e-07, - "loss": 0.7161, + "loss": 0.5108, "step": 473 }, { "epoch": 2.739884393063584, - "grad_norm": 0.6643382977320614, + "grad_norm": 0.7006518455929439, "learning_rate": 2.3749432495917546e-07, - "loss": 0.4652, + "loss": 0.4978, "step": 474 }, { "epoch": 2.745664739884393, - "grad_norm": 0.5190910480988365, + "grad_norm": 0.6297202502249554, "learning_rate": 2.2735885659134927e-07, - "loss": 0.5041, + "loss": 0.4321, "step": 475 }, { "epoch": 2.7514450867052025, - "grad_norm": 0.6630309587957454, + "grad_norm": 0.6678663798391579, "learning_rate": 2.1743937299301242e-07, - "loss": 0.491, + "loss": 0.5351, "step": 476 }, { "epoch": 2.7572254335260116, - "grad_norm": 0.5674562925635366, + "grad_norm": 0.6561058848775081, "learning_rate": 2.0773632306809622e-07, - "loss": 0.53, + "loss": 0.519, "step": 477 }, { "epoch": 2.7630057803468207, - "grad_norm": 0.7484735180985417, + "grad_norm": 0.7009614607634279, "learning_rate": 1.9825014592587844e-07, - "loss": 0.5673, + "loss": 0.4933, "step": 478 }, { "epoch": 2.76878612716763, - "grad_norm": 0.5416867923408462, + "grad_norm": 0.642529585526624, "learning_rate": 1.889812708611083e-07, - "loss": 0.546, + "loss": 0.4666, "step": 479 }, { "epoch": 2.7745664739884393, - "grad_norm": 0.6390120484336053, + "grad_norm": 0.7789251421262561, "learning_rate": 1.7993011733458077e-07, - "loss": 0.5194, + "loss": 0.5676, "step": 480 }, { "epoch": 2.7803468208092488, - "grad_norm": 0.550004690933299, + "grad_norm": 0.7662588228210795, "learning_rate": 1.7109709495415073e-07, - "loss": 0.5918, + "loss": 0.5911, "step": 481 }, { "epoch": 2.786127167630058, - "grad_norm": 0.5913114986820611, + "grad_norm": 0.6573273387890879, "learning_rate": 1.624826034562016e-07, - "loss": 0.5211, + "loss": 0.4786, "step": 482 }, { "epoch": 2.791907514450867, - "grad_norm": 0.5786641063697584, + "grad_norm": 0.7395909205331832, "learning_rate": 1.5408703268754988e-07, - "loss": 0.5039, + "loss": 0.4978, "step": 483 }, { "epoch": 2.7976878612716765, - "grad_norm": 0.6765832837369992, + "grad_norm": 0.7598293296000616, "learning_rate": 1.459107625878059e-07, - "loss": 0.6065, + "loss": 0.7086, "step": 484 }, { "epoch": 2.8034682080924855, - "grad_norm": 0.5926302057082565, + "grad_norm": 0.8008269666978963, "learning_rate": 1.3795416317218036e-07, - "loss": 0.5904, + "loss": 0.441, "step": 485 }, { "epoch": 2.809248554913295, - "grad_norm": 0.6023460772144975, + "grad_norm": 0.607269773087072, "learning_rate": 1.3021759451473548e-07, - "loss": 0.4301, + "loss": 0.4577, "step": 486 }, { "epoch": 2.815028901734104, - "grad_norm": 0.5866888679602545, + "grad_norm": 0.8080424613928077, "learning_rate": 1.2270140673209473e-07, - "loss": 0.4801, + "loss": 0.4894, "step": 487 }, { "epoch": 2.820809248554913, - "grad_norm": 0.5518850867259907, + "grad_norm": 0.6835264471975695, "learning_rate": 1.1540593996759441e-07, - "loss": 0.5613, + "loss": 0.4831, "step": 488 }, { "epoch": 2.8265895953757223, - "grad_norm": 0.6474731731284514, + "grad_norm": 0.5617365597730085, "learning_rate": 1.0833152437589423e-07, - "loss": 0.593, + "loss": 0.4427, "step": 489 }, { "epoch": 2.832369942196532, - "grad_norm": 0.6141953013443167, + "grad_norm": 0.740772247640691, "learning_rate": 1.0147848010803319e-07, - "loss": 0.4921, + "loss": 0.6623, "step": 490 }, { "epoch": 2.838150289017341, - "grad_norm": 0.546559959952295, + "grad_norm": 0.6572245995068094, "learning_rate": 9.484711729694229e-08, - "loss": 0.5289, + "loss": 0.349, "step": 491 }, { "epoch": 2.8439306358381504, - "grad_norm": 0.5779046309961086, + "grad_norm": 0.5552515983374141, "learning_rate": 8.8437736043408e-08, - "loss": 0.5319, + "loss": 0.3654, "step": 492 }, { "epoch": 2.8497109826589595, - "grad_norm": 0.5485269439683567, + "grad_norm": 0.8034542181899899, "learning_rate": 8.225062640249636e-08, - "loss": 0.5299, + "loss": 0.5365, "step": 493 }, { "epoch": 2.8554913294797686, - "grad_norm": 0.5171310156336446, + "grad_norm": 0.8168265074856756, "learning_rate": 7.628606837041974e-08, - "loss": 0.5587, + "loss": 0.5066, "step": 494 }, { "epoch": 2.861271676300578, - "grad_norm": 0.6218694524421764, + "grad_norm": 0.7056216331593352, "learning_rate": 7.05443318718707e-08, - "loss": 0.6296, + "loss": 0.5115, "step": 495 }, { "epoch": 2.867052023121387, - "grad_norm": 0.6241941304950734, + "grad_norm": 0.5871150000820912, "learning_rate": 6.502567674780524e-08, - "loss": 0.5433, + "loss": 0.4359, "step": 496 }, { "epoch": 2.8728323699421967, - "grad_norm": 0.6893401927215842, + "grad_norm": 0.7322597178201294, "learning_rate": 5.973035274368266e-08, - "loss": 0.5421, + "loss": 0.5456, "step": 497 }, { "epoch": 2.878612716763006, - "grad_norm": 0.587590924135814, + "grad_norm": 0.8233622518580206, "learning_rate": 5.465859949816299e-08, - "loss": 0.6174, + "loss": 0.4307, "step": 498 }, { "epoch": 2.884393063583815, - "grad_norm": 0.7952368790341441, + "grad_norm": 0.5931005955566576, "learning_rate": 4.981064653226564e-08, - "loss": 0.6357, + "loss": 0.4317, "step": 499 }, { "epoch": 2.8901734104046244, - "grad_norm": 0.5934063847455681, + "grad_norm": 0.6617780255291629, "learning_rate": 4.5186713238979385e-08, - "loss": 0.5128, + "loss": 0.4231, "step": 500 }, { "epoch": 2.8959537572254335, - "grad_norm": 0.644744384208876, + "grad_norm": 0.7058701623939151, "learning_rate": 4.078700887333365e-08, - "loss": 0.5189, + "loss": 0.4358, "step": 501 }, { "epoch": 2.901734104046243, - "grad_norm": 0.6647294274280695, + "grad_norm": 0.8650529123196282, "learning_rate": 3.6611732542931044e-08, - "loss": 0.7207, + "loss": 0.4707, "step": 502 }, { "epoch": 2.907514450867052, - "grad_norm": 0.7535481253417466, + "grad_norm": 0.9022908071705149, "learning_rate": 3.266107319893463e-08, - "loss": 0.6227, + "loss": 0.58, "step": 503 }, { "epoch": 2.913294797687861, - "grad_norm": 1.0027257925226223, + "grad_norm": 0.6494927730691019, "learning_rate": 2.89352096275175e-08, - "loss": 0.5107, + "loss": 0.4633, "step": 504 }, { "epoch": 2.9190751445086707, - "grad_norm": 0.5743150125395796, + "grad_norm": 0.7088859627525681, "learning_rate": 2.5434310441773135e-08, - "loss": 0.4914, + "loss": 0.4791, "step": 505 }, { "epoch": 2.9248554913294798, - "grad_norm": 0.6044934735184301, + "grad_norm": 0.5591513082758558, "learning_rate": 2.2158534074083193e-08, - "loss": 0.5253, + "loss": 0.5138, "step": 506 }, { "epoch": 2.9306358381502893, - "grad_norm": 0.48762018383676176, + "grad_norm": 0.5945985036565976, "learning_rate": 1.910802876894824e-08, - "loss": 0.4177, + "loss": 0.4747, "step": 507 }, { "epoch": 2.9364161849710984, - "grad_norm": 0.5420472068506899, + "grad_norm": 0.8039612887428521, "learning_rate": 1.6282932576279775e-08, - "loss": 0.53, + "loss": 0.5382, "step": 508 }, { "epoch": 2.9421965317919074, - "grad_norm": 0.6399738810658941, + "grad_norm": 0.6149926780604529, "learning_rate": 1.3683373345150796e-08, - "loss": 0.5894, + "loss": 0.4795, "step": 509 }, { "epoch": 2.9479768786127165, - "grad_norm": 0.5764220429572922, + "grad_norm": 0.6444388317466117, "learning_rate": 1.1309468718013194e-08, - "loss": 0.4792, + "loss": 0.3789, "step": 510 }, { "epoch": 2.953757225433526, - "grad_norm": 0.5318872547753678, + "grad_norm": 0.5826132360555071, "learning_rate": 9.16132612537035e-09, - "loss": 0.5329, + "loss": 0.5196, "step": 511 }, { "epoch": 2.959537572254335, - "grad_norm": 0.540580058732667, + "grad_norm": 0.8093687170769972, "learning_rate": 7.2390427809176934e-09, - "loss": 0.5065, + "loss": 0.5751, "step": 512 }, { "epoch": 2.9653179190751446, - "grad_norm": 0.6051544824457664, + "grad_norm": 0.8287185571079843, "learning_rate": 5.542705677143434e-09, - "loss": 0.5385, + "loss": 0.7386, "step": 513 }, { "epoch": 2.9710982658959537, - "grad_norm": 0.6211767993411176, + "grad_norm": 0.6711414362177306, "learning_rate": 4.072391581388946e-09, - "loss": 0.5342, + "loss": 0.4838, "step": 514 }, { "epoch": 2.976878612716763, - "grad_norm": 0.6112074242934528, + "grad_norm": 0.6330747713176181, "learning_rate": 2.8281670323798693e-09, - "loss": 0.586, + "loss": 0.442, "step": 515 }, { "epoch": 2.9826589595375723, - "grad_norm": 0.5272549910032829, + "grad_norm": 0.5833032431838548, "learning_rate": 1.8100883372085266e-09, - "loss": 0.5224, + "loss": 0.4679, "step": 516 }, { "epoch": 2.9884393063583814, - "grad_norm": 0.6169947160654549, + "grad_norm": 0.6448004089646303, "learning_rate": 1.0182015687909552e-09, - "loss": 0.5685, + "loss": 0.5389, "step": 517 }, { "epoch": 2.994219653179191, - "grad_norm": 0.5336160799812013, + "grad_norm": 0.6724467232133748, "learning_rate": 4.5254256377913474e-10, - "loss": 0.5236, + "loss": 0.5033, "step": 518 }, { "epoch": 3.0, - "grad_norm": 0.5915025477424289, + "grad_norm": 0.6000692854824138, "learning_rate": 1.1313692094117034e-10, - "loss": 0.5201, + "loss": 0.4207, "step": 519 }, { "epoch": 3.0, "step": 519, - "total_flos": 90606008975360.0, - "train_loss": 0.8465411536151504, - "train_runtime": 4209.1284, - "train_samples_per_second": 1.48, - "train_steps_per_second": 0.123 + "total_flos": 64152198111232.0, + "train_loss": 0.8013873731929213, + "train_runtime": 3362.3689, + "train_samples_per_second": 1.852, + "train_steps_per_second": 0.154 } ], "logging_steps": 1, @@ -3669,7 +3669,7 @@ "attributes": {} } }, - "total_flos": 90606008975360.0, + "total_flos": 64152198111232.0, "train_batch_size": 1, "trial_name": null, "trial_params": null