diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,17533 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 2.50070332280954, + "eval_steps": 500, + "global_step": 2500, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.001000281329123816, + "grad_norm": 0.18305286765098572, + "learning_rate": 4e-05, + "loss": 0.6074, + "step": 1 + }, + { + "epoch": 0.002000562658247632, + "grad_norm": 0.1878618746995926, + "learning_rate": 8e-05, + "loss": 0.5961, + "step": 2 + }, + { + "epoch": 0.003000843987371448, + "grad_norm": 0.13772089779376984, + "learning_rate": 0.00012, + "loss": 0.571, + "step": 3 + }, + { + "epoch": 0.004001125316495264, + "grad_norm": 0.14825063943862915, + "learning_rate": 0.00016, + "loss": 0.601, + "step": 4 + }, + { + "epoch": 0.005001406645619081, + "grad_norm": 0.15167813003063202, + "learning_rate": 0.0002, + "loss": 0.5963, + "step": 5 + }, + { + "epoch": 0.006001687974742896, + "grad_norm": 0.12160810828208923, + "learning_rate": 0.0001999331550802139, + "loss": 0.5542, + "step": 6 + }, + { + "epoch": 0.007001969303866712, + "grad_norm": 0.11746848374605179, + "learning_rate": 0.00019986631016042782, + "loss": 0.5468, + "step": 7 + }, + { + "epoch": 0.008002250632990529, + "grad_norm": 0.10123622417449951, + "learning_rate": 0.00019979946524064174, + "loss": 0.5404, + "step": 8 + }, + { + "epoch": 0.009002531962114344, + "grad_norm": 0.10068394988775253, + "learning_rate": 0.00019973262032085563, + "loss": 0.5253, + "step": 9 + }, + { + "epoch": 0.010002813291238161, + "grad_norm": 0.09869424253702164, + "learning_rate": 0.00019966577540106952, + "loss": 0.5204, + "step": 10 + }, + { + "epoch": 0.011003094620361977, + "grad_norm": 0.09765811264514923, + "learning_rate": 0.00019959893048128344, + "loss": 0.518, + "step": 11 + }, + { + "epoch": 0.012003375949485792, + "grad_norm": 0.09476452320814133, + "learning_rate": 0.00019953208556149733, + "loss": 0.5422, + "step": 12 + }, + { + "epoch": 0.01300365727860961, + "grad_norm": 0.09298824518918991, + "learning_rate": 0.00019946524064171124, + "loss": 0.4958, + "step": 13 + }, + { + "epoch": 0.014003938607733425, + "grad_norm": 0.08859862387180328, + "learning_rate": 0.00019939839572192514, + "loss": 0.4935, + "step": 14 + }, + { + "epoch": 0.01500421993685724, + "grad_norm": 0.09502395987510681, + "learning_rate": 0.00019933155080213905, + "loss": 0.5341, + "step": 15 + }, + { + "epoch": 0.016004501265981057, + "grad_norm": 0.0984087735414505, + "learning_rate": 0.00019926470588235294, + "loss": 0.4949, + "step": 16 + }, + { + "epoch": 0.017004782595104875, + "grad_norm": 0.10976890474557877, + "learning_rate": 0.00019919786096256686, + "loss": 0.5041, + "step": 17 + }, + { + "epoch": 0.01800506392422869, + "grad_norm": 0.10657032579183578, + "learning_rate": 0.00019913101604278075, + "loss": 0.5089, + "step": 18 + }, + { + "epoch": 0.019005345253352506, + "grad_norm": 0.25480905175209045, + "learning_rate": 0.00019906417112299467, + "loss": 0.4954, + "step": 19 + }, + { + "epoch": 0.020005626582476323, + "grad_norm": 0.10139419138431549, + "learning_rate": 0.00019899732620320856, + "loss": 0.5021, + "step": 20 + }, + { + "epoch": 0.021005907911600136, + "grad_norm": 0.10845401883125305, + "learning_rate": 0.00019893048128342245, + "loss": 0.4779, + "step": 21 + }, + { + "epoch": 0.022006189240723954, + "grad_norm": 0.10552472621202469, + "learning_rate": 0.00019886363636363637, + "loss": 0.4812, + "step": 22 + }, + { + "epoch": 0.02300647056984777, + "grad_norm": 0.09560544043779373, + "learning_rate": 0.0001987967914438503, + "loss": 0.4391, + "step": 23 + }, + { + "epoch": 0.024006751898971584, + "grad_norm": 0.09816516190767288, + "learning_rate": 0.00019872994652406418, + "loss": 0.4522, + "step": 24 + }, + { + "epoch": 0.0250070332280954, + "grad_norm": 0.09173469245433807, + "learning_rate": 0.00019866310160427807, + "loss": 0.4577, + "step": 25 + }, + { + "epoch": 0.02600731455721922, + "grad_norm": 0.09287692606449127, + "learning_rate": 0.000198596256684492, + "loss": 0.4712, + "step": 26 + }, + { + "epoch": 0.027007595886343033, + "grad_norm": 0.09577341377735138, + "learning_rate": 0.0001985294117647059, + "loss": 0.4471, + "step": 27 + }, + { + "epoch": 0.02800787721546685, + "grad_norm": 0.09503757208585739, + "learning_rate": 0.0001984625668449198, + "loss": 0.4473, + "step": 28 + }, + { + "epoch": 0.029008158544590667, + "grad_norm": 0.09698604792356491, + "learning_rate": 0.0001983957219251337, + "loss": 0.4626, + "step": 29 + }, + { + "epoch": 0.03000843987371448, + "grad_norm": 0.08879084140062332, + "learning_rate": 0.0001983288770053476, + "loss": 0.4556, + "step": 30 + }, + { + "epoch": 0.031008721202838298, + "grad_norm": 0.09425485134124756, + "learning_rate": 0.00019826203208556152, + "loss": 0.4294, + "step": 31 + }, + { + "epoch": 0.032009002531962115, + "grad_norm": 0.1031407043337822, + "learning_rate": 0.00019819518716577541, + "loss": 0.4357, + "step": 32 + }, + { + "epoch": 0.03300928386108593, + "grad_norm": 0.10730168223381042, + "learning_rate": 0.0001981283422459893, + "loss": 0.4448, + "step": 33 + }, + { + "epoch": 0.03400956519020975, + "grad_norm": 0.1108059212565422, + "learning_rate": 0.00019806149732620322, + "loss": 0.4282, + "step": 34 + }, + { + "epoch": 0.03500984651933356, + "grad_norm": 0.09637418389320374, + "learning_rate": 0.0001979946524064171, + "loss": 0.4117, + "step": 35 + }, + { + "epoch": 0.03601012784845738, + "grad_norm": 0.10290028899908066, + "learning_rate": 0.00019792780748663103, + "loss": 0.4284, + "step": 36 + }, + { + "epoch": 0.037010409177581194, + "grad_norm": 0.11193324625492096, + "learning_rate": 0.00019786096256684492, + "loss": 0.4279, + "step": 37 + }, + { + "epoch": 0.03801069050670501, + "grad_norm": 0.10121899098157883, + "learning_rate": 0.00019779411764705884, + "loss": 0.4143, + "step": 38 + }, + { + "epoch": 0.03901097183582883, + "grad_norm": 0.09821845591068268, + "learning_rate": 0.00019772727272727273, + "loss": 0.4307, + "step": 39 + }, + { + "epoch": 0.040011253164952645, + "grad_norm": 0.11662347614765167, + "learning_rate": 0.00019766042780748665, + "loss": 0.4272, + "step": 40 + }, + { + "epoch": 0.041011534494076456, + "grad_norm": 0.11152911931276321, + "learning_rate": 0.00019759358288770054, + "loss": 0.406, + "step": 41 + }, + { + "epoch": 0.04201181582320027, + "grad_norm": 0.10545951128005981, + "learning_rate": 0.00019752673796791446, + "loss": 0.4063, + "step": 42 + }, + { + "epoch": 0.04301209715232409, + "grad_norm": 0.10418938845396042, + "learning_rate": 0.00019745989304812835, + "loss": 0.3904, + "step": 43 + }, + { + "epoch": 0.04401237848144791, + "grad_norm": 0.1134418472647667, + "learning_rate": 0.00019739304812834224, + "loss": 0.3832, + "step": 44 + }, + { + "epoch": 0.045012659810571724, + "grad_norm": 0.10668052732944489, + "learning_rate": 0.00019732620320855616, + "loss": 0.3806, + "step": 45 + }, + { + "epoch": 0.04601294113969554, + "grad_norm": 0.10941950976848602, + "learning_rate": 0.00019725935828877008, + "loss": 0.401, + "step": 46 + }, + { + "epoch": 0.04701322246881936, + "grad_norm": 0.12284600734710693, + "learning_rate": 0.00019719251336898397, + "loss": 0.3859, + "step": 47 + }, + { + "epoch": 0.04801350379794317, + "grad_norm": 0.11308012902736664, + "learning_rate": 0.00019712566844919786, + "loss": 0.371, + "step": 48 + }, + { + "epoch": 0.049013785127066986, + "grad_norm": 0.11386734992265701, + "learning_rate": 0.00019705882352941177, + "loss": 0.3789, + "step": 49 + }, + { + "epoch": 0.0500140664561908, + "grad_norm": 0.12261448800563812, + "learning_rate": 0.0001969919786096257, + "loss": 0.3913, + "step": 50 + }, + { + "epoch": 0.05101434778531462, + "grad_norm": 0.12915784120559692, + "learning_rate": 0.00019692513368983958, + "loss": 0.4119, + "step": 51 + }, + { + "epoch": 0.05201462911443844, + "grad_norm": 0.11582653969526291, + "learning_rate": 0.00019685828877005347, + "loss": 0.393, + "step": 52 + }, + { + "epoch": 0.053014910443562255, + "grad_norm": 0.1418416053056717, + "learning_rate": 0.0001967914438502674, + "loss": 0.3753, + "step": 53 + }, + { + "epoch": 0.054015191772686065, + "grad_norm": 0.15086938440799713, + "learning_rate": 0.0001967245989304813, + "loss": 0.3723, + "step": 54 + }, + { + "epoch": 0.05501547310180988, + "grad_norm": 0.1251247376203537, + "learning_rate": 0.0001966577540106952, + "loss": 0.3735, + "step": 55 + }, + { + "epoch": 0.0560157544309337, + "grad_norm": 0.12578067183494568, + "learning_rate": 0.0001965909090909091, + "loss": 0.3658, + "step": 56 + }, + { + "epoch": 0.05701603576005752, + "grad_norm": 0.1202901154756546, + "learning_rate": 0.000196524064171123, + "loss": 0.3976, + "step": 57 + }, + { + "epoch": 0.058016317089181334, + "grad_norm": 0.13312962651252747, + "learning_rate": 0.0001964572192513369, + "loss": 0.3663, + "step": 58 + }, + { + "epoch": 0.05901659841830515, + "grad_norm": 0.12706811726093292, + "learning_rate": 0.00019639037433155082, + "loss": 0.372, + "step": 59 + }, + { + "epoch": 0.06001687974742896, + "grad_norm": 0.13053391873836517, + "learning_rate": 0.0001963235294117647, + "loss": 0.3831, + "step": 60 + }, + { + "epoch": 0.06101716107655278, + "grad_norm": 0.13940644264221191, + "learning_rate": 0.00019625668449197863, + "loss": 0.3653, + "step": 61 + }, + { + "epoch": 0.062017442405676595, + "grad_norm": 0.13523486256599426, + "learning_rate": 0.00019618983957219252, + "loss": 0.3894, + "step": 62 + }, + { + "epoch": 0.06301772373480041, + "grad_norm": 0.126784548163414, + "learning_rate": 0.00019612299465240644, + "loss": 0.3799, + "step": 63 + }, + { + "epoch": 0.06401800506392423, + "grad_norm": 0.13675376772880554, + "learning_rate": 0.00019605614973262033, + "loss": 0.3684, + "step": 64 + }, + { + "epoch": 0.06501828639304805, + "grad_norm": 0.14057081937789917, + "learning_rate": 0.00019598930481283424, + "loss": 0.3719, + "step": 65 + }, + { + "epoch": 0.06601856772217186, + "grad_norm": 0.1402556449174881, + "learning_rate": 0.00019592245989304814, + "loss": 0.3649, + "step": 66 + }, + { + "epoch": 0.06701884905129568, + "grad_norm": 0.14691877365112305, + "learning_rate": 0.00019585561497326203, + "loss": 0.36, + "step": 67 + }, + { + "epoch": 0.0680191303804195, + "grad_norm": 0.1424274891614914, + "learning_rate": 0.00019578877005347594, + "loss": 0.3593, + "step": 68 + }, + { + "epoch": 0.06901941170954332, + "grad_norm": 0.13602085411548615, + "learning_rate": 0.00019572192513368986, + "loss": 0.3638, + "step": 69 + }, + { + "epoch": 0.07001969303866712, + "grad_norm": 0.14070270955562592, + "learning_rate": 0.00019565508021390375, + "loss": 0.3486, + "step": 70 + }, + { + "epoch": 0.07101997436779094, + "grad_norm": 0.14548255503177643, + "learning_rate": 0.00019558823529411764, + "loss": 0.366, + "step": 71 + }, + { + "epoch": 0.07202025569691475, + "grad_norm": 0.1355600506067276, + "learning_rate": 0.00019552139037433156, + "loss": 0.3452, + "step": 72 + }, + { + "epoch": 0.07302053702603857, + "grad_norm": 0.1325199007987976, + "learning_rate": 0.00019545454545454548, + "loss": 0.3499, + "step": 73 + }, + { + "epoch": 0.07402081835516239, + "grad_norm": 0.14651721715927124, + "learning_rate": 0.00019538770053475937, + "loss": 0.3528, + "step": 74 + }, + { + "epoch": 0.0750210996842862, + "grad_norm": 0.1445704996585846, + "learning_rate": 0.00019532085561497326, + "loss": 0.3476, + "step": 75 + }, + { + "epoch": 0.07602138101341002, + "grad_norm": 0.13428623974323273, + "learning_rate": 0.00019525401069518718, + "loss": 0.3534, + "step": 76 + }, + { + "epoch": 0.07702166234253384, + "grad_norm": 0.1384708285331726, + "learning_rate": 0.00019518716577540107, + "loss": 0.3297, + "step": 77 + }, + { + "epoch": 0.07802194367165766, + "grad_norm": 0.14777661859989166, + "learning_rate": 0.000195120320855615, + "loss": 0.3489, + "step": 78 + }, + { + "epoch": 0.07902222500078147, + "grad_norm": 0.1292216181755066, + "learning_rate": 0.00019505347593582888, + "loss": 0.3365, + "step": 79 + }, + { + "epoch": 0.08002250632990529, + "grad_norm": 0.14391309022903442, + "learning_rate": 0.0001949866310160428, + "loss": 0.359, + "step": 80 + }, + { + "epoch": 0.08102278765902911, + "grad_norm": 0.14055773615837097, + "learning_rate": 0.0001949197860962567, + "loss": 0.3336, + "step": 81 + }, + { + "epoch": 0.08202306898815291, + "grad_norm": 0.13753433525562286, + "learning_rate": 0.0001948529411764706, + "loss": 0.3386, + "step": 82 + }, + { + "epoch": 0.08302335031727673, + "grad_norm": 0.1397666484117508, + "learning_rate": 0.0001947860962566845, + "loss": 0.3461, + "step": 83 + }, + { + "epoch": 0.08402363164640055, + "grad_norm": 0.16351380944252014, + "learning_rate": 0.00019471925133689841, + "loss": 0.3536, + "step": 84 + }, + { + "epoch": 0.08502391297552436, + "grad_norm": 0.1343546211719513, + "learning_rate": 0.0001946524064171123, + "loss": 0.33, + "step": 85 + }, + { + "epoch": 0.08602419430464818, + "grad_norm": 0.1452942192554474, + "learning_rate": 0.0001945855614973262, + "loss": 0.3221, + "step": 86 + }, + { + "epoch": 0.087024475633772, + "grad_norm": 0.1350940316915512, + "learning_rate": 0.0001945187165775401, + "loss": 0.343, + "step": 87 + }, + { + "epoch": 0.08802475696289581, + "grad_norm": 0.13612036406993866, + "learning_rate": 0.00019445187165775403, + "loss": 0.3234, + "step": 88 + }, + { + "epoch": 0.08902503829201963, + "grad_norm": 0.15519319474697113, + "learning_rate": 0.00019438502673796792, + "loss": 0.3483, + "step": 89 + }, + { + "epoch": 0.09002531962114345, + "grad_norm": 0.14684025943279266, + "learning_rate": 0.0001943181818181818, + "loss": 0.3319, + "step": 90 + }, + { + "epoch": 0.09102560095026727, + "grad_norm": 0.14248189330101013, + "learning_rate": 0.00019425133689839573, + "loss": 0.3365, + "step": 91 + }, + { + "epoch": 0.09202588227939108, + "grad_norm": 0.13078805804252625, + "learning_rate": 0.00019418449197860965, + "loss": 0.3129, + "step": 92 + }, + { + "epoch": 0.0930261636085149, + "grad_norm": 0.13958805799484253, + "learning_rate": 0.00019411764705882354, + "loss": 0.3303, + "step": 93 + }, + { + "epoch": 0.09402644493763872, + "grad_norm": 0.15192373096942902, + "learning_rate": 0.00019405080213903743, + "loss": 0.3458, + "step": 94 + }, + { + "epoch": 0.09502672626676252, + "grad_norm": 0.15896309912204742, + "learning_rate": 0.00019398395721925135, + "loss": 0.3505, + "step": 95 + }, + { + "epoch": 0.09602700759588634, + "grad_norm": 0.1371334195137024, + "learning_rate": 0.00019391711229946527, + "loss": 0.3355, + "step": 96 + }, + { + "epoch": 0.09702728892501016, + "grad_norm": 0.15836550295352936, + "learning_rate": 0.00019385026737967916, + "loss": 0.3569, + "step": 97 + }, + { + "epoch": 0.09802757025413397, + "grad_norm": 0.14184708893299103, + "learning_rate": 0.00019378342245989305, + "loss": 0.3379, + "step": 98 + }, + { + "epoch": 0.09902785158325779, + "grad_norm": 0.13973352313041687, + "learning_rate": 0.00019371657754010697, + "loss": 0.3096, + "step": 99 + }, + { + "epoch": 0.1000281329123816, + "grad_norm": 0.16972346603870392, + "learning_rate": 0.00019364973262032086, + "loss": 0.3296, + "step": 100 + }, + { + "epoch": 0.10102841424150542, + "grad_norm": 0.13497419655323029, + "learning_rate": 0.00019358288770053477, + "loss": 0.3386, + "step": 101 + }, + { + "epoch": 0.10202869557062924, + "grad_norm": 0.14974476397037506, + "learning_rate": 0.00019351604278074867, + "loss": 0.3323, + "step": 102 + }, + { + "epoch": 0.10302897689975306, + "grad_norm": 0.17438584566116333, + "learning_rate": 0.00019344919786096258, + "loss": 0.3286, + "step": 103 + }, + { + "epoch": 0.10402925822887688, + "grad_norm": 0.1679316908121109, + "learning_rate": 0.00019338235294117647, + "loss": 0.3328, + "step": 104 + }, + { + "epoch": 0.10502953955800069, + "grad_norm": 0.15288309752941132, + "learning_rate": 0.0001933155080213904, + "loss": 0.3313, + "step": 105 + }, + { + "epoch": 0.10602982088712451, + "grad_norm": 0.15492694079875946, + "learning_rate": 0.00019324866310160428, + "loss": 0.3164, + "step": 106 + }, + { + "epoch": 0.10703010221624833, + "grad_norm": 0.16060768067836761, + "learning_rate": 0.0001931818181818182, + "loss": 0.3391, + "step": 107 + }, + { + "epoch": 0.10803038354537213, + "grad_norm": 0.16690890491008759, + "learning_rate": 0.0001931149732620321, + "loss": 0.3432, + "step": 108 + }, + { + "epoch": 0.10903066487449595, + "grad_norm": 0.13718362152576447, + "learning_rate": 0.00019304812834224598, + "loss": 0.3259, + "step": 109 + }, + { + "epoch": 0.11003094620361976, + "grad_norm": 0.1728982925415039, + "learning_rate": 0.0001929812834224599, + "loss": 0.3287, + "step": 110 + }, + { + "epoch": 0.11103122753274358, + "grad_norm": 0.14428073167800903, + "learning_rate": 0.00019291443850267382, + "loss": 0.3157, + "step": 111 + }, + { + "epoch": 0.1120315088618674, + "grad_norm": 0.14764927327632904, + "learning_rate": 0.0001928475935828877, + "loss": 0.3096, + "step": 112 + }, + { + "epoch": 0.11303179019099122, + "grad_norm": 0.14915820956230164, + "learning_rate": 0.0001927807486631016, + "loss": 0.3242, + "step": 113 + }, + { + "epoch": 0.11403207152011503, + "grad_norm": 0.14097453653812408, + "learning_rate": 0.00019271390374331552, + "loss": 0.3087, + "step": 114 + }, + { + "epoch": 0.11503235284923885, + "grad_norm": 0.14339981973171234, + "learning_rate": 0.00019264705882352944, + "loss": 0.3061, + "step": 115 + }, + { + "epoch": 0.11603263417836267, + "grad_norm": 0.1636725515127182, + "learning_rate": 0.00019258021390374333, + "loss": 0.3208, + "step": 116 + }, + { + "epoch": 0.11703291550748648, + "grad_norm": 0.15016674995422363, + "learning_rate": 0.00019251336898395722, + "loss": 0.3253, + "step": 117 + }, + { + "epoch": 0.1180331968366103, + "grad_norm": 0.16004154086112976, + "learning_rate": 0.00019244652406417114, + "loss": 0.3236, + "step": 118 + }, + { + "epoch": 0.11903347816573412, + "grad_norm": 0.1520974040031433, + "learning_rate": 0.00019237967914438503, + "loss": 0.331, + "step": 119 + }, + { + "epoch": 0.12003375949485792, + "grad_norm": 0.1554533690214157, + "learning_rate": 0.00019231283422459894, + "loss": 0.3317, + "step": 120 + }, + { + "epoch": 0.12103404082398174, + "grad_norm": 0.14271193742752075, + "learning_rate": 0.00019224598930481283, + "loss": 0.3057, + "step": 121 + }, + { + "epoch": 0.12203432215310556, + "grad_norm": 0.14464707672595978, + "learning_rate": 0.00019217914438502675, + "loss": 0.3164, + "step": 122 + }, + { + "epoch": 0.12303460348222937, + "grad_norm": 0.14747333526611328, + "learning_rate": 0.00019211229946524064, + "loss": 0.324, + "step": 123 + }, + { + "epoch": 0.12403488481135319, + "grad_norm": 0.15460710227489471, + "learning_rate": 0.00019204545454545456, + "loss": 0.3338, + "step": 124 + }, + { + "epoch": 0.12503516614047702, + "grad_norm": 0.14597386121749878, + "learning_rate": 0.00019197860962566845, + "loss": 0.3181, + "step": 125 + }, + { + "epoch": 0.12603544746960083, + "grad_norm": 0.16421006619930267, + "learning_rate": 0.00019191176470588237, + "loss": 0.3013, + "step": 126 + }, + { + "epoch": 0.12703572879872463, + "grad_norm": 0.16206569969654083, + "learning_rate": 0.00019184491978609626, + "loss": 0.3082, + "step": 127 + }, + { + "epoch": 0.12803601012784846, + "grad_norm": 0.1381295770406723, + "learning_rate": 0.00019177807486631015, + "loss": 0.2945, + "step": 128 + }, + { + "epoch": 0.12903629145697226, + "grad_norm": 0.1753690391778946, + "learning_rate": 0.00019171122994652407, + "loss": 0.3025, + "step": 129 + }, + { + "epoch": 0.1300365727860961, + "grad_norm": 0.17089487612247467, + "learning_rate": 0.000191644385026738, + "loss": 0.339, + "step": 130 + }, + { + "epoch": 0.1310368541152199, + "grad_norm": 0.14900460839271545, + "learning_rate": 0.00019157754010695188, + "loss": 0.3006, + "step": 131 + }, + { + "epoch": 0.13203713544434373, + "grad_norm": 0.17873816192150116, + "learning_rate": 0.00019151069518716577, + "loss": 0.3111, + "step": 132 + }, + { + "epoch": 0.13303741677346753, + "grad_norm": 0.15606606006622314, + "learning_rate": 0.0001914438502673797, + "loss": 0.3285, + "step": 133 + }, + { + "epoch": 0.13403769810259136, + "grad_norm": 0.1459713727235794, + "learning_rate": 0.0001913770053475936, + "loss": 0.3277, + "step": 134 + }, + { + "epoch": 0.13503797943171517, + "grad_norm": 0.1609024554491043, + "learning_rate": 0.0001913101604278075, + "loss": 0.2916, + "step": 135 + }, + { + "epoch": 0.136038260760839, + "grad_norm": 0.14843402802944183, + "learning_rate": 0.0001912433155080214, + "loss": 0.3112, + "step": 136 + }, + { + "epoch": 0.1370385420899628, + "grad_norm": 0.14451859891414642, + "learning_rate": 0.0001911764705882353, + "loss": 0.2942, + "step": 137 + }, + { + "epoch": 0.13803882341908663, + "grad_norm": 0.1509537547826767, + "learning_rate": 0.00019110962566844922, + "loss": 0.3123, + "step": 138 + }, + { + "epoch": 0.13903910474821043, + "grad_norm": 0.13653701543807983, + "learning_rate": 0.0001910427807486631, + "loss": 0.2937, + "step": 139 + }, + { + "epoch": 0.14003938607733424, + "grad_norm": 0.15668630599975586, + "learning_rate": 0.000190975935828877, + "loss": 0.3194, + "step": 140 + }, + { + "epoch": 0.14103966740645807, + "grad_norm": 0.1442468911409378, + "learning_rate": 0.00019090909090909092, + "loss": 0.2974, + "step": 141 + }, + { + "epoch": 0.14203994873558187, + "grad_norm": 0.15264883637428284, + "learning_rate": 0.0001908422459893048, + "loss": 0.2944, + "step": 142 + }, + { + "epoch": 0.1430402300647057, + "grad_norm": 0.14924411475658417, + "learning_rate": 0.00019077540106951873, + "loss": 0.3246, + "step": 143 + }, + { + "epoch": 0.1440405113938295, + "grad_norm": 0.14341893792152405, + "learning_rate": 0.00019070855614973262, + "loss": 0.2825, + "step": 144 + }, + { + "epoch": 0.14504079272295334, + "grad_norm": 0.16582003235816956, + "learning_rate": 0.00019064171122994654, + "loss": 0.3415, + "step": 145 + }, + { + "epoch": 0.14604107405207714, + "grad_norm": 0.19595587253570557, + "learning_rate": 0.00019057486631016043, + "loss": 0.3254, + "step": 146 + }, + { + "epoch": 0.14704135538120097, + "grad_norm": 0.15274642407894135, + "learning_rate": 0.00019050802139037435, + "loss": 0.2896, + "step": 147 + }, + { + "epoch": 0.14804163671032478, + "grad_norm": 0.15903377532958984, + "learning_rate": 0.00019044117647058824, + "loss": 0.308, + "step": 148 + }, + { + "epoch": 0.1490419180394486, + "grad_norm": 0.17833171784877777, + "learning_rate": 0.00019037433155080216, + "loss": 0.3319, + "step": 149 + }, + { + "epoch": 0.1500421993685724, + "grad_norm": 0.15367349982261658, + "learning_rate": 0.00019030748663101605, + "loss": 0.309, + "step": 150 + }, + { + "epoch": 0.15104248069769624, + "grad_norm": 0.1506006270647049, + "learning_rate": 0.00019024064171122994, + "loss": 0.2945, + "step": 151 + }, + { + "epoch": 0.15204276202682004, + "grad_norm": 0.14558513462543488, + "learning_rate": 0.00019017379679144388, + "loss": 0.3354, + "step": 152 + }, + { + "epoch": 0.15304304335594385, + "grad_norm": 0.1557934284210205, + "learning_rate": 0.00019010695187165777, + "loss": 0.2996, + "step": 153 + }, + { + "epoch": 0.15404332468506768, + "grad_norm": 0.17990130186080933, + "learning_rate": 0.00019004010695187167, + "loss": 0.325, + "step": 154 + }, + { + "epoch": 0.15504360601419148, + "grad_norm": 0.1437973827123642, + "learning_rate": 0.00018997326203208556, + "loss": 0.3307, + "step": 155 + }, + { + "epoch": 0.1560438873433153, + "grad_norm": 0.15119291841983795, + "learning_rate": 0.00018990641711229947, + "loss": 0.3113, + "step": 156 + }, + { + "epoch": 0.15704416867243912, + "grad_norm": 0.1675308793783188, + "learning_rate": 0.0001898395721925134, + "loss": 0.321, + "step": 157 + }, + { + "epoch": 0.15804445000156295, + "grad_norm": 0.1566278487443924, + "learning_rate": 0.00018977272727272728, + "loss": 0.3148, + "step": 158 + }, + { + "epoch": 0.15904473133068675, + "grad_norm": 0.16762055456638336, + "learning_rate": 0.00018970588235294117, + "loss": 0.3103, + "step": 159 + }, + { + "epoch": 0.16004501265981058, + "grad_norm": 0.1488402634859085, + "learning_rate": 0.0001896390374331551, + "loss": 0.3021, + "step": 160 + }, + { + "epoch": 0.16104529398893438, + "grad_norm": 0.14263427257537842, + "learning_rate": 0.000189572192513369, + "loss": 0.3148, + "step": 161 + }, + { + "epoch": 0.16204557531805822, + "grad_norm": 0.14631019532680511, + "learning_rate": 0.0001895053475935829, + "loss": 0.3095, + "step": 162 + }, + { + "epoch": 0.16304585664718202, + "grad_norm": 0.15839937329292297, + "learning_rate": 0.0001894385026737968, + "loss": 0.3122, + "step": 163 + }, + { + "epoch": 0.16404613797630582, + "grad_norm": 0.15408416092395782, + "learning_rate": 0.0001893716577540107, + "loss": 0.3042, + "step": 164 + }, + { + "epoch": 0.16504641930542965, + "grad_norm": 0.15629133582115173, + "learning_rate": 0.0001893048128342246, + "loss": 0.2911, + "step": 165 + }, + { + "epoch": 0.16604670063455346, + "grad_norm": 0.14410415291786194, + "learning_rate": 0.00018923796791443852, + "loss": 0.2955, + "step": 166 + }, + { + "epoch": 0.1670469819636773, + "grad_norm": 0.17271237075328827, + "learning_rate": 0.0001891711229946524, + "loss": 0.3117, + "step": 167 + }, + { + "epoch": 0.1680472632928011, + "grad_norm": 0.15094706416130066, + "learning_rate": 0.00018910427807486633, + "loss": 0.3023, + "step": 168 + }, + { + "epoch": 0.16904754462192492, + "grad_norm": 0.14633840322494507, + "learning_rate": 0.00018903743315508022, + "loss": 0.3251, + "step": 169 + }, + { + "epoch": 0.17004782595104873, + "grad_norm": 0.14702406525611877, + "learning_rate": 0.00018897058823529413, + "loss": 0.3002, + "step": 170 + }, + { + "epoch": 0.17104810728017256, + "grad_norm": 0.15511703491210938, + "learning_rate": 0.00018890374331550803, + "loss": 0.3027, + "step": 171 + }, + { + "epoch": 0.17204838860929636, + "grad_norm": 0.1390470713376999, + "learning_rate": 0.00018883689839572194, + "loss": 0.3024, + "step": 172 + }, + { + "epoch": 0.1730486699384202, + "grad_norm": 0.15443062782287598, + "learning_rate": 0.00018877005347593583, + "loss": 0.3173, + "step": 173 + }, + { + "epoch": 0.174048951267544, + "grad_norm": 0.14939570426940918, + "learning_rate": 0.00018870320855614973, + "loss": 0.3182, + "step": 174 + }, + { + "epoch": 0.17504923259666783, + "grad_norm": 0.1404954195022583, + "learning_rate": 0.00018863636363636364, + "loss": 0.2977, + "step": 175 + }, + { + "epoch": 0.17604951392579163, + "grad_norm": 0.14608599245548248, + "learning_rate": 0.00018856951871657756, + "loss": 0.2922, + "step": 176 + }, + { + "epoch": 0.17704979525491543, + "grad_norm": 0.13753263652324677, + "learning_rate": 0.00018850267379679145, + "loss": 0.2876, + "step": 177 + }, + { + "epoch": 0.17805007658403926, + "grad_norm": 0.1753247231245041, + "learning_rate": 0.00018843582887700534, + "loss": 0.2972, + "step": 178 + }, + { + "epoch": 0.17905035791316307, + "grad_norm": 0.15599356591701508, + "learning_rate": 0.00018836898395721926, + "loss": 0.2871, + "step": 179 + }, + { + "epoch": 0.1800506392422869, + "grad_norm": 0.13570567965507507, + "learning_rate": 0.00018830213903743318, + "loss": 0.2791, + "step": 180 + }, + { + "epoch": 0.1810509205714107, + "grad_norm": 0.1570330709218979, + "learning_rate": 0.00018823529411764707, + "loss": 0.3046, + "step": 181 + }, + { + "epoch": 0.18205120190053453, + "grad_norm": 0.17737139761447906, + "learning_rate": 0.00018816844919786096, + "loss": 0.3021, + "step": 182 + }, + { + "epoch": 0.18305148322965833, + "grad_norm": 0.1483505219221115, + "learning_rate": 0.00018810160427807488, + "loss": 0.2926, + "step": 183 + }, + { + "epoch": 0.18405176455878217, + "grad_norm": 0.16992217302322388, + "learning_rate": 0.00018803475935828877, + "loss": 0.3192, + "step": 184 + }, + { + "epoch": 0.18505204588790597, + "grad_norm": 0.1548883318901062, + "learning_rate": 0.0001879679144385027, + "loss": 0.2945, + "step": 185 + }, + { + "epoch": 0.1860523272170298, + "grad_norm": 0.14812099933624268, + "learning_rate": 0.00018790106951871658, + "loss": 0.293, + "step": 186 + }, + { + "epoch": 0.1870526085461536, + "grad_norm": 0.14878343045711517, + "learning_rate": 0.0001878342245989305, + "loss": 0.3008, + "step": 187 + }, + { + "epoch": 0.18805288987527743, + "grad_norm": 0.15423160791397095, + "learning_rate": 0.00018776737967914439, + "loss": 0.2947, + "step": 188 + }, + { + "epoch": 0.18905317120440124, + "grad_norm": 0.1366434246301651, + "learning_rate": 0.0001877005347593583, + "loss": 0.3011, + "step": 189 + }, + { + "epoch": 0.19005345253352504, + "grad_norm": 0.15592864155769348, + "learning_rate": 0.0001876336898395722, + "loss": 0.3024, + "step": 190 + }, + { + "epoch": 0.19105373386264887, + "grad_norm": 0.14957195520401, + "learning_rate": 0.0001875668449197861, + "loss": 0.2926, + "step": 191 + }, + { + "epoch": 0.19205401519177268, + "grad_norm": 0.13259784877300262, + "learning_rate": 0.0001875, + "loss": 0.2723, + "step": 192 + }, + { + "epoch": 0.1930542965208965, + "grad_norm": 0.157723069190979, + "learning_rate": 0.0001874331550802139, + "loss": 0.3106, + "step": 193 + }, + { + "epoch": 0.1940545778500203, + "grad_norm": 0.15471331775188446, + "learning_rate": 0.00018736631016042784, + "loss": 0.2881, + "step": 194 + }, + { + "epoch": 0.19505485917914414, + "grad_norm": 0.17348626255989075, + "learning_rate": 0.00018729946524064173, + "loss": 0.2972, + "step": 195 + }, + { + "epoch": 0.19605514050826794, + "grad_norm": 0.17215636372566223, + "learning_rate": 0.00018723262032085562, + "loss": 0.3061, + "step": 196 + }, + { + "epoch": 0.19705542183739178, + "grad_norm": 0.15460576117038727, + "learning_rate": 0.0001871657754010695, + "loss": 0.31, + "step": 197 + }, + { + "epoch": 0.19805570316651558, + "grad_norm": 0.1641826033592224, + "learning_rate": 0.00018709893048128343, + "loss": 0.3039, + "step": 198 + }, + { + "epoch": 0.1990559844956394, + "grad_norm": 0.14432573318481445, + "learning_rate": 0.00018703208556149735, + "loss": 0.2811, + "step": 199 + }, + { + "epoch": 0.2000562658247632, + "grad_norm": 0.12521930038928986, + "learning_rate": 0.00018696524064171124, + "loss": 0.282, + "step": 200 + }, + { + "epoch": 0.20105654715388704, + "grad_norm": 0.16774822771549225, + "learning_rate": 0.00018689839572192513, + "loss": 0.2842, + "step": 201 + }, + { + "epoch": 0.20205682848301085, + "grad_norm": 0.14739921689033508, + "learning_rate": 0.00018683155080213905, + "loss": 0.2909, + "step": 202 + }, + { + "epoch": 0.20305710981213465, + "grad_norm": 0.15254798531532288, + "learning_rate": 0.00018676470588235297, + "loss": 0.2791, + "step": 203 + }, + { + "epoch": 0.20405739114125848, + "grad_norm": 0.15687166154384613, + "learning_rate": 0.00018669786096256686, + "loss": 0.2785, + "step": 204 + }, + { + "epoch": 0.20505767247038229, + "grad_norm": 0.1820078045129776, + "learning_rate": 0.00018663101604278075, + "loss": 0.2901, + "step": 205 + }, + { + "epoch": 0.20605795379950612, + "grad_norm": 0.15682387351989746, + "learning_rate": 0.00018656417112299467, + "loss": 0.295, + "step": 206 + }, + { + "epoch": 0.20705823512862992, + "grad_norm": 0.15712325274944305, + "learning_rate": 0.00018649732620320856, + "loss": 0.2933, + "step": 207 + }, + { + "epoch": 0.20805851645775375, + "grad_norm": 0.17071470618247986, + "learning_rate": 0.00018643048128342247, + "loss": 0.2963, + "step": 208 + }, + { + "epoch": 0.20905879778687755, + "grad_norm": 0.1309688538312912, + "learning_rate": 0.00018636363636363636, + "loss": 0.269, + "step": 209 + }, + { + "epoch": 0.21005907911600138, + "grad_norm": 0.13551504909992218, + "learning_rate": 0.00018629679144385028, + "loss": 0.3089, + "step": 210 + }, + { + "epoch": 0.2110593604451252, + "grad_norm": 0.13836127519607544, + "learning_rate": 0.00018622994652406417, + "loss": 0.2911, + "step": 211 + }, + { + "epoch": 0.21205964177424902, + "grad_norm": 0.14555582404136658, + "learning_rate": 0.0001861631016042781, + "loss": 0.2893, + "step": 212 + }, + { + "epoch": 0.21305992310337282, + "grad_norm": 0.1329416185617447, + "learning_rate": 0.000186096256684492, + "loss": 0.2857, + "step": 213 + }, + { + "epoch": 0.21406020443249665, + "grad_norm": 0.1323581039905548, + "learning_rate": 0.0001860294117647059, + "loss": 0.2765, + "step": 214 + }, + { + "epoch": 0.21506048576162046, + "grad_norm": 0.14781104028224945, + "learning_rate": 0.0001859625668449198, + "loss": 0.2869, + "step": 215 + }, + { + "epoch": 0.21606076709074426, + "grad_norm": 0.16794413328170776, + "learning_rate": 0.00018589572192513368, + "loss": 0.3001, + "step": 216 + }, + { + "epoch": 0.2170610484198681, + "grad_norm": 0.14039048552513123, + "learning_rate": 0.0001858288770053476, + "loss": 0.29, + "step": 217 + }, + { + "epoch": 0.2180613297489919, + "grad_norm": 0.14844009280204773, + "learning_rate": 0.00018576203208556152, + "loss": 0.3084, + "step": 218 + }, + { + "epoch": 0.21906161107811573, + "grad_norm": 0.15255549550056458, + "learning_rate": 0.0001856951871657754, + "loss": 0.2906, + "step": 219 + }, + { + "epoch": 0.22006189240723953, + "grad_norm": 0.14477452635765076, + "learning_rate": 0.0001856283422459893, + "loss": 0.2742, + "step": 220 + }, + { + "epoch": 0.22106217373636336, + "grad_norm": 0.13587240874767303, + "learning_rate": 0.00018556149732620322, + "loss": 0.2812, + "step": 221 + }, + { + "epoch": 0.22206245506548716, + "grad_norm": 0.14347586035728455, + "learning_rate": 0.00018549465240641713, + "loss": 0.2909, + "step": 222 + }, + { + "epoch": 0.223062736394611, + "grad_norm": 0.1650112122297287, + "learning_rate": 0.00018542780748663103, + "loss": 0.2912, + "step": 223 + }, + { + "epoch": 0.2240630177237348, + "grad_norm": 0.1673017293214798, + "learning_rate": 0.00018536096256684492, + "loss": 0.3035, + "step": 224 + }, + { + "epoch": 0.22506329905285863, + "grad_norm": 0.14499802887439728, + "learning_rate": 0.00018529411764705883, + "loss": 0.2925, + "step": 225 + }, + { + "epoch": 0.22606358038198243, + "grad_norm": 0.14276009798049927, + "learning_rate": 0.00018522727272727273, + "loss": 0.2869, + "step": 226 + }, + { + "epoch": 0.22706386171110624, + "grad_norm": 0.15777187049388885, + "learning_rate": 0.00018516042780748664, + "loss": 0.3042, + "step": 227 + }, + { + "epoch": 0.22806414304023007, + "grad_norm": 0.1441316306591034, + "learning_rate": 0.00018509358288770053, + "loss": 0.2939, + "step": 228 + }, + { + "epoch": 0.22906442436935387, + "grad_norm": 0.14155633747577667, + "learning_rate": 0.00018502673796791445, + "loss": 0.2724, + "step": 229 + }, + { + "epoch": 0.2300647056984777, + "grad_norm": 0.1651201844215393, + "learning_rate": 0.00018495989304812834, + "loss": 0.287, + "step": 230 + }, + { + "epoch": 0.2310649870276015, + "grad_norm": 0.1399175077676773, + "learning_rate": 0.00018489304812834226, + "loss": 0.2627, + "step": 231 + }, + { + "epoch": 0.23206526835672533, + "grad_norm": 0.146319180727005, + "learning_rate": 0.00018482620320855615, + "loss": 0.3039, + "step": 232 + }, + { + "epoch": 0.23306554968584914, + "grad_norm": 0.14373745024204254, + "learning_rate": 0.00018475935828877007, + "loss": 0.284, + "step": 233 + }, + { + "epoch": 0.23406583101497297, + "grad_norm": 0.1537570059299469, + "learning_rate": 0.00018469251336898396, + "loss": 0.2862, + "step": 234 + }, + { + "epoch": 0.23506611234409677, + "grad_norm": 0.15432754158973694, + "learning_rate": 0.00018462566844919785, + "loss": 0.2831, + "step": 235 + }, + { + "epoch": 0.2360663936732206, + "grad_norm": 0.1558825522661209, + "learning_rate": 0.0001845588235294118, + "loss": 0.2908, + "step": 236 + }, + { + "epoch": 0.2370666750023444, + "grad_norm": 0.15844637155532837, + "learning_rate": 0.0001844919786096257, + "loss": 0.2869, + "step": 237 + }, + { + "epoch": 0.23806695633146824, + "grad_norm": 0.14787733554840088, + "learning_rate": 0.00018442513368983958, + "loss": 0.303, + "step": 238 + }, + { + "epoch": 0.23906723766059204, + "grad_norm": 0.15325358510017395, + "learning_rate": 0.00018435828877005347, + "loss": 0.2839, + "step": 239 + }, + { + "epoch": 0.24006751898971584, + "grad_norm": 0.1502685397863388, + "learning_rate": 0.00018429144385026739, + "loss": 0.2982, + "step": 240 + }, + { + "epoch": 0.24106780031883968, + "grad_norm": 0.15496723353862762, + "learning_rate": 0.0001842245989304813, + "loss": 0.2788, + "step": 241 + }, + { + "epoch": 0.24206808164796348, + "grad_norm": 0.14616385102272034, + "learning_rate": 0.0001841577540106952, + "loss": 0.2895, + "step": 242 + }, + { + "epoch": 0.2430683629770873, + "grad_norm": 0.16031970083713531, + "learning_rate": 0.00018409090909090909, + "loss": 0.2815, + "step": 243 + }, + { + "epoch": 0.2440686443062111, + "grad_norm": 0.17965003848075867, + "learning_rate": 0.000184024064171123, + "loss": 0.2946, + "step": 244 + }, + { + "epoch": 0.24506892563533494, + "grad_norm": 0.1588526964187622, + "learning_rate": 0.00018395721925133692, + "loss": 0.2885, + "step": 245 + }, + { + "epoch": 0.24606920696445875, + "grad_norm": 0.1620316356420517, + "learning_rate": 0.0001838903743315508, + "loss": 0.2894, + "step": 246 + }, + { + "epoch": 0.24706948829358258, + "grad_norm": 0.14299319684505463, + "learning_rate": 0.0001838235294117647, + "loss": 0.2642, + "step": 247 + }, + { + "epoch": 0.24806976962270638, + "grad_norm": 0.14844638109207153, + "learning_rate": 0.00018375668449197862, + "loss": 0.3015, + "step": 248 + }, + { + "epoch": 0.2490700509518302, + "grad_norm": 0.16899922490119934, + "learning_rate": 0.0001836898395721925, + "loss": 0.2912, + "step": 249 + }, + { + "epoch": 0.25007033228095404, + "grad_norm": 0.15139029920101166, + "learning_rate": 0.00018362299465240643, + "loss": 0.2818, + "step": 250 + }, + { + "epoch": 0.2510706136100778, + "grad_norm": 0.14223715662956238, + "learning_rate": 0.00018355614973262032, + "loss": 0.2741, + "step": 251 + }, + { + "epoch": 0.25207089493920165, + "grad_norm": 0.1626238226890564, + "learning_rate": 0.00018348930481283424, + "loss": 0.2754, + "step": 252 + }, + { + "epoch": 0.2530711762683255, + "grad_norm": 0.15521816909313202, + "learning_rate": 0.00018342245989304813, + "loss": 0.2773, + "step": 253 + }, + { + "epoch": 0.25407145759744926, + "grad_norm": 0.1444212794303894, + "learning_rate": 0.00018335561497326205, + "loss": 0.2898, + "step": 254 + }, + { + "epoch": 0.2550717389265731, + "grad_norm": 0.16217008233070374, + "learning_rate": 0.00018328877005347597, + "loss": 0.2876, + "step": 255 + }, + { + "epoch": 0.2560720202556969, + "grad_norm": 0.1517690122127533, + "learning_rate": 0.00018322192513368986, + "loss": 0.2908, + "step": 256 + }, + { + "epoch": 0.25707230158482075, + "grad_norm": 0.1489913910627365, + "learning_rate": 0.00018315508021390375, + "loss": 0.294, + "step": 257 + }, + { + "epoch": 0.2580725829139445, + "grad_norm": 0.14388781785964966, + "learning_rate": 0.00018308823529411764, + "loss": 0.2838, + "step": 258 + }, + { + "epoch": 0.25907286424306836, + "grad_norm": 0.14525553584098816, + "learning_rate": 0.00018302139037433158, + "loss": 0.286, + "step": 259 + }, + { + "epoch": 0.2600731455721922, + "grad_norm": 0.1541617214679718, + "learning_rate": 0.00018295454545454547, + "loss": 0.3068, + "step": 260 + }, + { + "epoch": 0.261073426901316, + "grad_norm": 0.1467447131872177, + "learning_rate": 0.00018288770053475936, + "loss": 0.2786, + "step": 261 + }, + { + "epoch": 0.2620737082304398, + "grad_norm": 0.13658003509044647, + "learning_rate": 0.00018282085561497326, + "loss": 0.2634, + "step": 262 + }, + { + "epoch": 0.2630739895595636, + "grad_norm": 0.133203387260437, + "learning_rate": 0.00018275401069518717, + "loss": 0.2579, + "step": 263 + }, + { + "epoch": 0.26407427088868746, + "grad_norm": 0.14174403250217438, + "learning_rate": 0.0001826871657754011, + "loss": 0.3022, + "step": 264 + }, + { + "epoch": 0.26507455221781123, + "grad_norm": 0.12970340251922607, + "learning_rate": 0.00018262032085561498, + "loss": 0.2833, + "step": 265 + }, + { + "epoch": 0.26607483354693506, + "grad_norm": 0.13447734713554382, + "learning_rate": 0.00018255347593582887, + "loss": 0.2813, + "step": 266 + }, + { + "epoch": 0.2670751148760589, + "grad_norm": 0.14448338747024536, + "learning_rate": 0.0001824866310160428, + "loss": 0.2774, + "step": 267 + }, + { + "epoch": 0.2680753962051827, + "grad_norm": 0.13557809591293335, + "learning_rate": 0.0001824197860962567, + "loss": 0.2583, + "step": 268 + }, + { + "epoch": 0.2690756775343065, + "grad_norm": 0.1400664746761322, + "learning_rate": 0.0001823529411764706, + "loss": 0.2859, + "step": 269 + }, + { + "epoch": 0.27007595886343033, + "grad_norm": 0.14144816994667053, + "learning_rate": 0.0001822860962566845, + "loss": 0.2762, + "step": 270 + }, + { + "epoch": 0.27107624019255416, + "grad_norm": 0.1487583965063095, + "learning_rate": 0.0001822192513368984, + "loss": 0.2708, + "step": 271 + }, + { + "epoch": 0.272076521521678, + "grad_norm": 0.14664174616336823, + "learning_rate": 0.0001821524064171123, + "loss": 0.2842, + "step": 272 + }, + { + "epoch": 0.27307680285080177, + "grad_norm": 0.14909137785434723, + "learning_rate": 0.00018208556149732622, + "loss": 0.2867, + "step": 273 + }, + { + "epoch": 0.2740770841799256, + "grad_norm": 0.13501740992069244, + "learning_rate": 0.0001820187165775401, + "loss": 0.2796, + "step": 274 + }, + { + "epoch": 0.27507736550904943, + "grad_norm": 0.1478702425956726, + "learning_rate": 0.00018195187165775403, + "loss": 0.2782, + "step": 275 + }, + { + "epoch": 0.27607764683817326, + "grad_norm": 0.13471487164497375, + "learning_rate": 0.00018188502673796792, + "loss": 0.2748, + "step": 276 + }, + { + "epoch": 0.27707792816729704, + "grad_norm": 0.15584290027618408, + "learning_rate": 0.00018181818181818183, + "loss": 0.2933, + "step": 277 + }, + { + "epoch": 0.27807820949642087, + "grad_norm": 0.16546475887298584, + "learning_rate": 0.00018175133689839575, + "loss": 0.2712, + "step": 278 + }, + { + "epoch": 0.2790784908255447, + "grad_norm": 0.15253497660160065, + "learning_rate": 0.00018168449197860964, + "loss": 0.2784, + "step": 279 + }, + { + "epoch": 0.2800787721546685, + "grad_norm": 0.15355491638183594, + "learning_rate": 0.00018161764705882353, + "loss": 0.2652, + "step": 280 + }, + { + "epoch": 0.2810790534837923, + "grad_norm": 0.1511094570159912, + "learning_rate": 0.00018155080213903742, + "loss": 0.294, + "step": 281 + }, + { + "epoch": 0.28207933481291614, + "grad_norm": 0.14573180675506592, + "learning_rate": 0.00018148395721925134, + "loss": 0.2981, + "step": 282 + }, + { + "epoch": 0.28307961614203997, + "grad_norm": 0.14542432129383087, + "learning_rate": 0.00018141711229946526, + "loss": 0.2996, + "step": 283 + }, + { + "epoch": 0.28407989747116374, + "grad_norm": 0.1633489578962326, + "learning_rate": 0.00018135026737967915, + "loss": 0.295, + "step": 284 + }, + { + "epoch": 0.2850801788002876, + "grad_norm": 0.13544511795043945, + "learning_rate": 0.00018128342245989304, + "loss": 0.2746, + "step": 285 + }, + { + "epoch": 0.2860804601294114, + "grad_norm": 0.14513398706912994, + "learning_rate": 0.00018121657754010696, + "loss": 0.2875, + "step": 286 + }, + { + "epoch": 0.28708074145853524, + "grad_norm": 0.16061873733997345, + "learning_rate": 0.00018114973262032088, + "loss": 0.2813, + "step": 287 + }, + { + "epoch": 0.288081022787659, + "grad_norm": 0.16886045038700104, + "learning_rate": 0.00018108288770053477, + "loss": 0.292, + "step": 288 + }, + { + "epoch": 0.28908130411678284, + "grad_norm": 0.1356947422027588, + "learning_rate": 0.00018101604278074866, + "loss": 0.283, + "step": 289 + }, + { + "epoch": 0.2900815854459067, + "grad_norm": 0.15411818027496338, + "learning_rate": 0.00018094919786096258, + "loss": 0.2827, + "step": 290 + }, + { + "epoch": 0.29108186677503045, + "grad_norm": 0.15824365615844727, + "learning_rate": 0.00018088235294117647, + "loss": 0.2703, + "step": 291 + }, + { + "epoch": 0.2920821481041543, + "grad_norm": 0.149041086435318, + "learning_rate": 0.00018081550802139039, + "loss": 0.281, + "step": 292 + }, + { + "epoch": 0.2930824294332781, + "grad_norm": 0.13340702652931213, + "learning_rate": 0.00018074866310160428, + "loss": 0.2733, + "step": 293 + }, + { + "epoch": 0.29408271076240194, + "grad_norm": 0.17881901562213898, + "learning_rate": 0.0001806818181818182, + "loss": 0.2914, + "step": 294 + }, + { + "epoch": 0.2950829920915257, + "grad_norm": 0.1549568772315979, + "learning_rate": 0.00018061497326203209, + "loss": 0.2795, + "step": 295 + }, + { + "epoch": 0.29608327342064955, + "grad_norm": 0.16672222316265106, + "learning_rate": 0.000180548128342246, + "loss": 0.2867, + "step": 296 + }, + { + "epoch": 0.2970835547497734, + "grad_norm": 0.14212043583393097, + "learning_rate": 0.00018048128342245992, + "loss": 0.2664, + "step": 297 + }, + { + "epoch": 0.2980838360788972, + "grad_norm": 0.14593954384326935, + "learning_rate": 0.0001804144385026738, + "loss": 0.283, + "step": 298 + }, + { + "epoch": 0.299084117408021, + "grad_norm": 0.14363057911396027, + "learning_rate": 0.0001803475935828877, + "loss": 0.2594, + "step": 299 + }, + { + "epoch": 0.3000843987371448, + "grad_norm": 0.15335021913051605, + "learning_rate": 0.0001802807486631016, + "loss": 0.2989, + "step": 300 + }, + { + "epoch": 0.30108468006626865, + "grad_norm": 0.14953511953353882, + "learning_rate": 0.00018021390374331554, + "loss": 0.2897, + "step": 301 + }, + { + "epoch": 0.3020849613953925, + "grad_norm": 0.1523541659116745, + "learning_rate": 0.00018014705882352943, + "loss": 0.2878, + "step": 302 + }, + { + "epoch": 0.30308524272451626, + "grad_norm": 0.14485718309879303, + "learning_rate": 0.00018008021390374332, + "loss": 0.3004, + "step": 303 + }, + { + "epoch": 0.3040855240536401, + "grad_norm": 0.13954873383045197, + "learning_rate": 0.0001800133689839572, + "loss": 0.2666, + "step": 304 + }, + { + "epoch": 0.3050858053827639, + "grad_norm": 0.1484750509262085, + "learning_rate": 0.00017994652406417113, + "loss": 0.279, + "step": 305 + }, + { + "epoch": 0.3060860867118877, + "grad_norm": 0.14913174510002136, + "learning_rate": 0.00017987967914438505, + "loss": 0.2976, + "step": 306 + }, + { + "epoch": 0.3070863680410115, + "grad_norm": 0.1236981526017189, + "learning_rate": 0.00017981283422459894, + "loss": 0.2683, + "step": 307 + }, + { + "epoch": 0.30808664937013536, + "grad_norm": 0.12732578814029694, + "learning_rate": 0.00017974598930481283, + "loss": 0.2757, + "step": 308 + }, + { + "epoch": 0.3090869306992592, + "grad_norm": 0.13497596979141235, + "learning_rate": 0.00017967914438502675, + "loss": 0.2581, + "step": 309 + }, + { + "epoch": 0.31008721202838296, + "grad_norm": 0.142715722322464, + "learning_rate": 0.00017961229946524066, + "loss": 0.2597, + "step": 310 + }, + { + "epoch": 0.3110874933575068, + "grad_norm": 0.142337828874588, + "learning_rate": 0.00017954545454545456, + "loss": 0.2697, + "step": 311 + }, + { + "epoch": 0.3120877746866306, + "grad_norm": 0.1301991492509842, + "learning_rate": 0.00017947860962566845, + "loss": 0.2688, + "step": 312 + }, + { + "epoch": 0.31308805601575446, + "grad_norm": 0.14765168726444244, + "learning_rate": 0.00017941176470588236, + "loss": 0.2811, + "step": 313 + }, + { + "epoch": 0.31408833734487823, + "grad_norm": 0.12940143048763275, + "learning_rate": 0.00017934491978609626, + "loss": 0.2507, + "step": 314 + }, + { + "epoch": 0.31508861867400206, + "grad_norm": 0.14036771655082703, + "learning_rate": 0.00017927807486631017, + "loss": 0.2627, + "step": 315 + }, + { + "epoch": 0.3160889000031259, + "grad_norm": 0.14210949838161469, + "learning_rate": 0.00017921122994652406, + "loss": 0.2763, + "step": 316 + }, + { + "epoch": 0.31708918133224967, + "grad_norm": 0.1410045474767685, + "learning_rate": 0.00017914438502673798, + "loss": 0.2741, + "step": 317 + }, + { + "epoch": 0.3180894626613735, + "grad_norm": 0.1302804946899414, + "learning_rate": 0.00017907754010695187, + "loss": 0.297, + "step": 318 + }, + { + "epoch": 0.31908974399049733, + "grad_norm": 0.1306772232055664, + "learning_rate": 0.0001790106951871658, + "loss": 0.2648, + "step": 319 + }, + { + "epoch": 0.32009002531962116, + "grad_norm": 0.14245423674583435, + "learning_rate": 0.0001789438502673797, + "loss": 0.2692, + "step": 320 + }, + { + "epoch": 0.32109030664874494, + "grad_norm": 0.13119956851005554, + "learning_rate": 0.0001788770053475936, + "loss": 0.2625, + "step": 321 + }, + { + "epoch": 0.32209058797786877, + "grad_norm": 0.1373457908630371, + "learning_rate": 0.0001788101604278075, + "loss": 0.2744, + "step": 322 + }, + { + "epoch": 0.3230908693069926, + "grad_norm": 0.16108274459838867, + "learning_rate": 0.00017874331550802138, + "loss": 0.2759, + "step": 323 + }, + { + "epoch": 0.32409115063611643, + "grad_norm": 0.13662441074848175, + "learning_rate": 0.0001786764705882353, + "loss": 0.2596, + "step": 324 + }, + { + "epoch": 0.3250914319652402, + "grad_norm": 0.16478440165519714, + "learning_rate": 0.00017860962566844922, + "loss": 0.2627, + "step": 325 + }, + { + "epoch": 0.32609171329436404, + "grad_norm": 0.1390889436006546, + "learning_rate": 0.0001785427807486631, + "loss": 0.2493, + "step": 326 + }, + { + "epoch": 0.32709199462348787, + "grad_norm": 0.15374642610549927, + "learning_rate": 0.000178475935828877, + "loss": 0.2885, + "step": 327 + }, + { + "epoch": 0.32809227595261165, + "grad_norm": 0.1456235647201538, + "learning_rate": 0.00017840909090909092, + "loss": 0.2755, + "step": 328 + }, + { + "epoch": 0.3290925572817355, + "grad_norm": 0.1325979381799698, + "learning_rate": 0.00017834224598930483, + "loss": 0.2596, + "step": 329 + }, + { + "epoch": 0.3300928386108593, + "grad_norm": 0.134046733379364, + "learning_rate": 0.00017827540106951872, + "loss": 0.2615, + "step": 330 + }, + { + "epoch": 0.33109311993998314, + "grad_norm": 0.137125164270401, + "learning_rate": 0.00017820855614973262, + "loss": 0.2916, + "step": 331 + }, + { + "epoch": 0.3320934012691069, + "grad_norm": 0.15468603372573853, + "learning_rate": 0.00017814171122994653, + "loss": 0.2872, + "step": 332 + }, + { + "epoch": 0.33309368259823074, + "grad_norm": 0.12741553783416748, + "learning_rate": 0.00017807486631016042, + "loss": 0.2668, + "step": 333 + }, + { + "epoch": 0.3340939639273546, + "grad_norm": 0.16241924464702606, + "learning_rate": 0.00017800802139037434, + "loss": 0.2839, + "step": 334 + }, + { + "epoch": 0.3350942452564784, + "grad_norm": 0.14307934045791626, + "learning_rate": 0.00017794117647058823, + "loss": 0.2815, + "step": 335 + }, + { + "epoch": 0.3360945265856022, + "grad_norm": 0.11836118996143341, + "learning_rate": 0.00017787433155080215, + "loss": 0.2508, + "step": 336 + }, + { + "epoch": 0.337094807914726, + "grad_norm": 0.1503506898880005, + "learning_rate": 0.00017780748663101604, + "loss": 0.2783, + "step": 337 + }, + { + "epoch": 0.33809508924384984, + "grad_norm": 0.1553022712469101, + "learning_rate": 0.00017774064171122996, + "loss": 0.2727, + "step": 338 + }, + { + "epoch": 0.3390953705729737, + "grad_norm": 0.139094740152359, + "learning_rate": 0.00017767379679144388, + "loss": 0.2885, + "step": 339 + }, + { + "epoch": 0.34009565190209745, + "grad_norm": 0.14895129203796387, + "learning_rate": 0.00017760695187165777, + "loss": 0.2773, + "step": 340 + }, + { + "epoch": 0.3410959332312213, + "grad_norm": 0.14662131667137146, + "learning_rate": 0.00017754010695187166, + "loss": 0.2764, + "step": 341 + }, + { + "epoch": 0.3420962145603451, + "grad_norm": 0.14086416363716125, + "learning_rate": 0.00017747326203208555, + "loss": 0.2704, + "step": 342 + }, + { + "epoch": 0.3430964958894689, + "grad_norm": 0.14517523348331451, + "learning_rate": 0.0001774064171122995, + "loss": 0.2885, + "step": 343 + }, + { + "epoch": 0.3440967772185927, + "grad_norm": 0.14759743213653564, + "learning_rate": 0.00017733957219251339, + "loss": 0.2871, + "step": 344 + }, + { + "epoch": 0.34509705854771655, + "grad_norm": 0.14926595985889435, + "learning_rate": 0.00017727272727272728, + "loss": 0.2915, + "step": 345 + }, + { + "epoch": 0.3460973398768404, + "grad_norm": 0.1312466412782669, + "learning_rate": 0.00017720588235294117, + "loss": 0.2624, + "step": 346 + }, + { + "epoch": 0.34709762120596416, + "grad_norm": 0.13725310564041138, + "learning_rate": 0.00017713903743315509, + "loss": 0.2777, + "step": 347 + }, + { + "epoch": 0.348097902535088, + "grad_norm": 0.13410036265850067, + "learning_rate": 0.000177072192513369, + "loss": 0.2755, + "step": 348 + }, + { + "epoch": 0.3490981838642118, + "grad_norm": 0.12318189442157745, + "learning_rate": 0.0001770053475935829, + "loss": 0.2741, + "step": 349 + }, + { + "epoch": 0.35009846519333565, + "grad_norm": 0.13404425978660583, + "learning_rate": 0.00017693850267379679, + "loss": 0.2657, + "step": 350 + }, + { + "epoch": 0.3510987465224594, + "grad_norm": 0.12469787150621414, + "learning_rate": 0.0001768716577540107, + "loss": 0.2642, + "step": 351 + }, + { + "epoch": 0.35209902785158326, + "grad_norm": 0.1452178955078125, + "learning_rate": 0.00017680481283422462, + "loss": 0.2688, + "step": 352 + }, + { + "epoch": 0.3530993091807071, + "grad_norm": 0.1566033959388733, + "learning_rate": 0.0001767379679144385, + "loss": 0.2786, + "step": 353 + }, + { + "epoch": 0.35409959050983086, + "grad_norm": 0.13707688450813293, + "learning_rate": 0.0001766711229946524, + "loss": 0.2674, + "step": 354 + }, + { + "epoch": 0.3550998718389547, + "grad_norm": 0.145652174949646, + "learning_rate": 0.00017660427807486632, + "loss": 0.2806, + "step": 355 + }, + { + "epoch": 0.3561001531680785, + "grad_norm": 0.13377350568771362, + "learning_rate": 0.0001765374331550802, + "loss": 0.269, + "step": 356 + }, + { + "epoch": 0.35710043449720236, + "grad_norm": 0.12570109963417053, + "learning_rate": 0.00017647058823529413, + "loss": 0.2721, + "step": 357 + }, + { + "epoch": 0.35810071582632613, + "grad_norm": 0.1427479088306427, + "learning_rate": 0.00017640374331550802, + "loss": 0.2657, + "step": 358 + }, + { + "epoch": 0.35910099715544996, + "grad_norm": 0.14124266803264618, + "learning_rate": 0.00017633689839572194, + "loss": 0.2645, + "step": 359 + }, + { + "epoch": 0.3601012784845738, + "grad_norm": 0.13596995174884796, + "learning_rate": 0.00017627005347593583, + "loss": 0.2586, + "step": 360 + }, + { + "epoch": 0.3611015598136976, + "grad_norm": 0.13423608243465424, + "learning_rate": 0.00017620320855614975, + "loss": 0.2681, + "step": 361 + }, + { + "epoch": 0.3621018411428214, + "grad_norm": 0.15114182233810425, + "learning_rate": 0.00017613636363636366, + "loss": 0.2757, + "step": 362 + }, + { + "epoch": 0.36310212247194523, + "grad_norm": 0.13942475616931915, + "learning_rate": 0.00017606951871657756, + "loss": 0.2773, + "step": 363 + }, + { + "epoch": 0.36410240380106906, + "grad_norm": 0.14439767599105835, + "learning_rate": 0.00017600267379679145, + "loss": 0.2741, + "step": 364 + }, + { + "epoch": 0.3651026851301929, + "grad_norm": 0.1474488079547882, + "learning_rate": 0.00017593582887700534, + "loss": 0.282, + "step": 365 + }, + { + "epoch": 0.36610296645931667, + "grad_norm": 0.13497796654701233, + "learning_rate": 0.00017586898395721928, + "loss": 0.2764, + "step": 366 + }, + { + "epoch": 0.3671032477884405, + "grad_norm": 0.13684698939323425, + "learning_rate": 0.00017580213903743317, + "loss": 0.2663, + "step": 367 + }, + { + "epoch": 0.36810352911756433, + "grad_norm": 0.13904264569282532, + "learning_rate": 0.00017573529411764706, + "loss": 0.2817, + "step": 368 + }, + { + "epoch": 0.3691038104466881, + "grad_norm": 0.13045138120651245, + "learning_rate": 0.00017566844919786095, + "loss": 0.2671, + "step": 369 + }, + { + "epoch": 0.37010409177581194, + "grad_norm": 0.13704635202884674, + "learning_rate": 0.00017560160427807487, + "loss": 0.2772, + "step": 370 + }, + { + "epoch": 0.37110437310493577, + "grad_norm": 0.13385838270187378, + "learning_rate": 0.0001755347593582888, + "loss": 0.2548, + "step": 371 + }, + { + "epoch": 0.3721046544340596, + "grad_norm": 0.14585889875888824, + "learning_rate": 0.00017546791443850268, + "loss": 0.272, + "step": 372 + }, + { + "epoch": 0.3731049357631834, + "grad_norm": 0.13362255692481995, + "learning_rate": 0.00017540106951871657, + "loss": 0.266, + "step": 373 + }, + { + "epoch": 0.3741052170923072, + "grad_norm": 0.12938883900642395, + "learning_rate": 0.0001753342245989305, + "loss": 0.2712, + "step": 374 + }, + { + "epoch": 0.37510549842143104, + "grad_norm": 0.13322357833385468, + "learning_rate": 0.0001752673796791444, + "loss": 0.2681, + "step": 375 + }, + { + "epoch": 0.37610577975055487, + "grad_norm": 0.13961443305015564, + "learning_rate": 0.0001752005347593583, + "loss": 0.2692, + "step": 376 + }, + { + "epoch": 0.37710606107967864, + "grad_norm": 0.18336084485054016, + "learning_rate": 0.0001751336898395722, + "loss": 0.2832, + "step": 377 + }, + { + "epoch": 0.3781063424088025, + "grad_norm": 0.16892533004283905, + "learning_rate": 0.0001750668449197861, + "loss": 0.2807, + "step": 378 + }, + { + "epoch": 0.3791066237379263, + "grad_norm": 0.14116264879703522, + "learning_rate": 0.000175, + "loss": 0.2889, + "step": 379 + }, + { + "epoch": 0.3801069050670501, + "grad_norm": 0.14313367009162903, + "learning_rate": 0.00017493315508021392, + "loss": 0.2729, + "step": 380 + }, + { + "epoch": 0.3811071863961739, + "grad_norm": 0.13969191908836365, + "learning_rate": 0.00017486631016042783, + "loss": 0.2605, + "step": 381 + }, + { + "epoch": 0.38210746772529774, + "grad_norm": 0.5227783918380737, + "learning_rate": 0.00017479946524064172, + "loss": 0.2817, + "step": 382 + }, + { + "epoch": 0.3831077490544216, + "grad_norm": 0.133822500705719, + "learning_rate": 0.00017473262032085562, + "loss": 0.2549, + "step": 383 + }, + { + "epoch": 0.38410803038354535, + "grad_norm": 0.12532807886600494, + "learning_rate": 0.00017466577540106953, + "loss": 0.2555, + "step": 384 + }, + { + "epoch": 0.3851083117126692, + "grad_norm": 0.1421874612569809, + "learning_rate": 0.00017459893048128345, + "loss": 0.3041, + "step": 385 + }, + { + "epoch": 0.386108593041793, + "grad_norm": 0.14090843498706818, + "learning_rate": 0.00017453208556149734, + "loss": 0.275, + "step": 386 + }, + { + "epoch": 0.38710887437091684, + "grad_norm": 0.1319083571434021, + "learning_rate": 0.00017446524064171123, + "loss": 0.2653, + "step": 387 + }, + { + "epoch": 0.3881091557000406, + "grad_norm": 0.14791524410247803, + "learning_rate": 0.00017439839572192512, + "loss": 0.2811, + "step": 388 + }, + { + "epoch": 0.38910943702916445, + "grad_norm": 0.12477608770132065, + "learning_rate": 0.00017433155080213904, + "loss": 0.2466, + "step": 389 + }, + { + "epoch": 0.3901097183582883, + "grad_norm": 0.13591758906841278, + "learning_rate": 0.00017426470588235296, + "loss": 0.2642, + "step": 390 + }, + { + "epoch": 0.39110999968741206, + "grad_norm": 0.15453137457370758, + "learning_rate": 0.00017419786096256685, + "loss": 0.2688, + "step": 391 + }, + { + "epoch": 0.3921102810165359, + "grad_norm": 0.1306883841753006, + "learning_rate": 0.00017413101604278074, + "loss": 0.2571, + "step": 392 + }, + { + "epoch": 0.3931105623456597, + "grad_norm": 0.13438357412815094, + "learning_rate": 0.00017406417112299466, + "loss": 0.2747, + "step": 393 + }, + { + "epoch": 0.39411084367478355, + "grad_norm": 0.14441339671611786, + "learning_rate": 0.00017399732620320858, + "loss": 0.2745, + "step": 394 + }, + { + "epoch": 0.3951111250039073, + "grad_norm": 0.1330602467060089, + "learning_rate": 0.00017393048128342247, + "loss": 0.265, + "step": 395 + }, + { + "epoch": 0.39611140633303116, + "grad_norm": 0.13888601958751678, + "learning_rate": 0.00017386363636363636, + "loss": 0.3017, + "step": 396 + }, + { + "epoch": 0.397111687662155, + "grad_norm": 0.440023273229599, + "learning_rate": 0.00017379679144385028, + "loss": 0.27, + "step": 397 + }, + { + "epoch": 0.3981119689912788, + "grad_norm": 0.14055335521697998, + "learning_rate": 0.00017372994652406417, + "loss": 0.2704, + "step": 398 + }, + { + "epoch": 0.3991122503204026, + "grad_norm": 0.12351582944393158, + "learning_rate": 0.00017366310160427809, + "loss": 0.2541, + "step": 399 + }, + { + "epoch": 0.4001125316495264, + "grad_norm": 0.13435687124729156, + "learning_rate": 0.000173596256684492, + "loss": 0.2806, + "step": 400 + }, + { + "epoch": 0.40111281297865026, + "grad_norm": 0.14805035293102264, + "learning_rate": 0.0001735294117647059, + "loss": 0.2623, + "step": 401 + }, + { + "epoch": 0.4021130943077741, + "grad_norm": 0.13022367656230927, + "learning_rate": 0.00017346256684491979, + "loss": 0.2579, + "step": 402 + }, + { + "epoch": 0.40311337563689786, + "grad_norm": 0.1321382373571396, + "learning_rate": 0.0001733957219251337, + "loss": 0.288, + "step": 403 + }, + { + "epoch": 0.4041136569660217, + "grad_norm": 0.14684803783893585, + "learning_rate": 0.00017332887700534762, + "loss": 0.2911, + "step": 404 + }, + { + "epoch": 0.4051139382951455, + "grad_norm": 0.13008224964141846, + "learning_rate": 0.0001732620320855615, + "loss": 0.2866, + "step": 405 + }, + { + "epoch": 0.4061142196242693, + "grad_norm": 0.1350080817937851, + "learning_rate": 0.0001731951871657754, + "loss": 0.2797, + "step": 406 + }, + { + "epoch": 0.40711450095339313, + "grad_norm": 0.14416401088237762, + "learning_rate": 0.0001731283422459893, + "loss": 0.2738, + "step": 407 + }, + { + "epoch": 0.40811478228251696, + "grad_norm": 0.12459777295589447, + "learning_rate": 0.00017306149732620324, + "loss": 0.2525, + "step": 408 + }, + { + "epoch": 0.4091150636116408, + "grad_norm": 0.14084003865718842, + "learning_rate": 0.00017299465240641713, + "loss": 0.2509, + "step": 409 + }, + { + "epoch": 0.41011534494076457, + "grad_norm": 0.13740015029907227, + "learning_rate": 0.00017292780748663102, + "loss": 0.2717, + "step": 410 + }, + { + "epoch": 0.4111156262698884, + "grad_norm": 0.13260775804519653, + "learning_rate": 0.0001728609625668449, + "loss": 0.2716, + "step": 411 + }, + { + "epoch": 0.41211590759901223, + "grad_norm": 0.13791711628437042, + "learning_rate": 0.00017279411764705883, + "loss": 0.2786, + "step": 412 + }, + { + "epoch": 0.41311618892813606, + "grad_norm": 0.1407199651002884, + "learning_rate": 0.00017272727272727275, + "loss": 0.2573, + "step": 413 + }, + { + "epoch": 0.41411647025725984, + "grad_norm": 0.1278151422739029, + "learning_rate": 0.00017266042780748664, + "loss": 0.2552, + "step": 414 + }, + { + "epoch": 0.41511675158638367, + "grad_norm": 0.1330394744873047, + "learning_rate": 0.00017259358288770053, + "loss": 0.2688, + "step": 415 + }, + { + "epoch": 0.4161170329155075, + "grad_norm": 0.17083314061164856, + "learning_rate": 0.00017252673796791445, + "loss": 0.2799, + "step": 416 + }, + { + "epoch": 0.4171173142446313, + "grad_norm": 0.15132571756839752, + "learning_rate": 0.00017245989304812836, + "loss": 0.2719, + "step": 417 + }, + { + "epoch": 0.4181175955737551, + "grad_norm": 0.13255450129508972, + "learning_rate": 0.00017239304812834225, + "loss": 0.2567, + "step": 418 + }, + { + "epoch": 0.41911787690287894, + "grad_norm": 0.1479891538619995, + "learning_rate": 0.00017232620320855615, + "loss": 0.2841, + "step": 419 + }, + { + "epoch": 0.42011815823200277, + "grad_norm": 0.13702897727489471, + "learning_rate": 0.00017225935828877006, + "loss": 0.2664, + "step": 420 + }, + { + "epoch": 0.42111843956112655, + "grad_norm": 0.12373264878988266, + "learning_rate": 0.00017219251336898395, + "loss": 0.2721, + "step": 421 + }, + { + "epoch": 0.4221187208902504, + "grad_norm": 0.14097996056079865, + "learning_rate": 0.00017212566844919787, + "loss": 0.2791, + "step": 422 + }, + { + "epoch": 0.4231190022193742, + "grad_norm": 0.13768741488456726, + "learning_rate": 0.0001720588235294118, + "loss": 0.2523, + "step": 423 + }, + { + "epoch": 0.42411928354849804, + "grad_norm": 0.12540988624095917, + "learning_rate": 0.00017199197860962568, + "loss": 0.2655, + "step": 424 + }, + { + "epoch": 0.4251195648776218, + "grad_norm": 0.13412897288799286, + "learning_rate": 0.00017192513368983957, + "loss": 0.2551, + "step": 425 + }, + { + "epoch": 0.42611984620674564, + "grad_norm": 0.1302645355463028, + "learning_rate": 0.0001718582887700535, + "loss": 0.2824, + "step": 426 + }, + { + "epoch": 0.4271201275358695, + "grad_norm": 0.13050320744514465, + "learning_rate": 0.0001717914438502674, + "loss": 0.2767, + "step": 427 + }, + { + "epoch": 0.4281204088649933, + "grad_norm": 0.14933830499649048, + "learning_rate": 0.0001717245989304813, + "loss": 0.272, + "step": 428 + }, + { + "epoch": 0.4291206901941171, + "grad_norm": 0.13217081129550934, + "learning_rate": 0.0001716577540106952, + "loss": 0.2868, + "step": 429 + }, + { + "epoch": 0.4301209715232409, + "grad_norm": 0.1315024197101593, + "learning_rate": 0.00017159090909090908, + "loss": 0.2795, + "step": 430 + }, + { + "epoch": 0.43112125285236474, + "grad_norm": 0.13341479003429413, + "learning_rate": 0.000171524064171123, + "loss": 0.2869, + "step": 431 + }, + { + "epoch": 0.4321215341814885, + "grad_norm": 0.12039531767368317, + "learning_rate": 0.00017145721925133692, + "loss": 0.2526, + "step": 432 + }, + { + "epoch": 0.43312181551061235, + "grad_norm": 0.13261151313781738, + "learning_rate": 0.0001713903743315508, + "loss": 0.2545, + "step": 433 + }, + { + "epoch": 0.4341220968397362, + "grad_norm": 0.12078419327735901, + "learning_rate": 0.0001713235294117647, + "loss": 0.2499, + "step": 434 + }, + { + "epoch": 0.43512237816886, + "grad_norm": 0.12651419639587402, + "learning_rate": 0.00017125668449197862, + "loss": 0.2553, + "step": 435 + }, + { + "epoch": 0.4361226594979838, + "grad_norm": 0.14438410103321075, + "learning_rate": 0.00017118983957219253, + "loss": 0.2667, + "step": 436 + }, + { + "epoch": 0.4371229408271076, + "grad_norm": 0.12641195952892303, + "learning_rate": 0.00017112299465240642, + "loss": 0.2698, + "step": 437 + }, + { + "epoch": 0.43812322215623145, + "grad_norm": 0.12843714654445648, + "learning_rate": 0.00017105614973262032, + "loss": 0.2712, + "step": 438 + }, + { + "epoch": 0.4391235034853553, + "grad_norm": 0.1262800544500351, + "learning_rate": 0.00017098930481283423, + "loss": 0.259, + "step": 439 + }, + { + "epoch": 0.44012378481447906, + "grad_norm": 0.1376781463623047, + "learning_rate": 0.00017092245989304812, + "loss": 0.2553, + "step": 440 + }, + { + "epoch": 0.4411240661436029, + "grad_norm": 0.13698402047157288, + "learning_rate": 0.00017085561497326204, + "loss": 0.279, + "step": 441 + }, + { + "epoch": 0.4421243474727267, + "grad_norm": 0.13023951649665833, + "learning_rate": 0.00017078877005347596, + "loss": 0.2571, + "step": 442 + }, + { + "epoch": 0.4431246288018505, + "grad_norm": 0.12800049781799316, + "learning_rate": 0.00017072192513368985, + "loss": 0.2779, + "step": 443 + }, + { + "epoch": 0.4441249101309743, + "grad_norm": 0.12756869196891785, + "learning_rate": 0.00017065508021390374, + "loss": 0.2613, + "step": 444 + }, + { + "epoch": 0.44512519146009816, + "grad_norm": 0.13758492469787598, + "learning_rate": 0.00017058823529411766, + "loss": 0.2735, + "step": 445 + }, + { + "epoch": 0.446125472789222, + "grad_norm": 0.13032783567905426, + "learning_rate": 0.00017052139037433158, + "loss": 0.2583, + "step": 446 + }, + { + "epoch": 0.44712575411834576, + "grad_norm": 0.12345101684331894, + "learning_rate": 0.00017045454545454547, + "loss": 0.2529, + "step": 447 + }, + { + "epoch": 0.4481260354474696, + "grad_norm": 0.12137975543737411, + "learning_rate": 0.00017038770053475936, + "loss": 0.2486, + "step": 448 + }, + { + "epoch": 0.4491263167765934, + "grad_norm": 0.15893349051475525, + "learning_rate": 0.00017032085561497325, + "loss": 0.2614, + "step": 449 + }, + { + "epoch": 0.45012659810571726, + "grad_norm": 0.12508884072303772, + "learning_rate": 0.0001702540106951872, + "loss": 0.265, + "step": 450 + }, + { + "epoch": 0.45112687943484103, + "grad_norm": 0.11829554289579391, + "learning_rate": 0.00017018716577540109, + "loss": 0.2461, + "step": 451 + }, + { + "epoch": 0.45212716076396486, + "grad_norm": 0.14072422683238983, + "learning_rate": 0.00017012032085561498, + "loss": 0.2461, + "step": 452 + }, + { + "epoch": 0.4531274420930887, + "grad_norm": 0.137125164270401, + "learning_rate": 0.00017005347593582887, + "loss": 0.2662, + "step": 453 + }, + { + "epoch": 0.45412772342221247, + "grad_norm": 0.15341967344284058, + "learning_rate": 0.00016998663101604278, + "loss": 0.2825, + "step": 454 + }, + { + "epoch": 0.4551280047513363, + "grad_norm": 0.12128650397062302, + "learning_rate": 0.0001699197860962567, + "loss": 0.2567, + "step": 455 + }, + { + "epoch": 0.45612828608046013, + "grad_norm": 0.13926555216312408, + "learning_rate": 0.0001698529411764706, + "loss": 0.2602, + "step": 456 + }, + { + "epoch": 0.45712856740958396, + "grad_norm": 0.12147518992424011, + "learning_rate": 0.00016978609625668448, + "loss": 0.2584, + "step": 457 + }, + { + "epoch": 0.45812884873870774, + "grad_norm": 0.12368662655353546, + "learning_rate": 0.0001697192513368984, + "loss": 0.2446, + "step": 458 + }, + { + "epoch": 0.45912913006783157, + "grad_norm": 0.12873952090740204, + "learning_rate": 0.00016965240641711232, + "loss": 0.2637, + "step": 459 + }, + { + "epoch": 0.4601294113969554, + "grad_norm": 0.1259600669145584, + "learning_rate": 0.0001695855614973262, + "loss": 0.2552, + "step": 460 + }, + { + "epoch": 0.46112969272607923, + "grad_norm": 0.12401297688484192, + "learning_rate": 0.0001695187165775401, + "loss": 0.261, + "step": 461 + }, + { + "epoch": 0.462129974055203, + "grad_norm": 0.13291938602924347, + "learning_rate": 0.00016945187165775402, + "loss": 0.2609, + "step": 462 + }, + { + "epoch": 0.46313025538432684, + "grad_norm": 0.14955592155456543, + "learning_rate": 0.0001693850267379679, + "loss": 0.2543, + "step": 463 + }, + { + "epoch": 0.46413053671345067, + "grad_norm": 0.15458466112613678, + "learning_rate": 0.00016931818181818183, + "loss": 0.2978, + "step": 464 + }, + { + "epoch": 0.4651308180425745, + "grad_norm": 0.14671652019023895, + "learning_rate": 0.00016925133689839575, + "loss": 0.2717, + "step": 465 + }, + { + "epoch": 0.4661310993716983, + "grad_norm": 0.1314944177865982, + "learning_rate": 0.00016918449197860964, + "loss": 0.267, + "step": 466 + }, + { + "epoch": 0.4671313807008221, + "grad_norm": 0.12745250761508942, + "learning_rate": 0.00016911764705882353, + "loss": 0.2415, + "step": 467 + }, + { + "epoch": 0.46813166202994594, + "grad_norm": 0.12175773829221725, + "learning_rate": 0.00016905080213903745, + "loss": 0.2625, + "step": 468 + }, + { + "epoch": 0.4691319433590697, + "grad_norm": 0.14158126711845398, + "learning_rate": 0.00016898395721925136, + "loss": 0.2585, + "step": 469 + }, + { + "epoch": 0.47013222468819355, + "grad_norm": 0.13295957446098328, + "learning_rate": 0.00016891711229946525, + "loss": 0.2598, + "step": 470 + }, + { + "epoch": 0.4711325060173174, + "grad_norm": 0.13304173946380615, + "learning_rate": 0.00016885026737967915, + "loss": 0.2653, + "step": 471 + }, + { + "epoch": 0.4721327873464412, + "grad_norm": 0.12392549216747284, + "learning_rate": 0.00016878342245989304, + "loss": 0.2482, + "step": 472 + }, + { + "epoch": 0.473133068675565, + "grad_norm": 0.1471811830997467, + "learning_rate": 0.00016871657754010698, + "loss": 0.2562, + "step": 473 + }, + { + "epoch": 0.4741333500046888, + "grad_norm": 0.12605445086956024, + "learning_rate": 0.00016864973262032087, + "loss": 0.2547, + "step": 474 + }, + { + "epoch": 0.47513363133381264, + "grad_norm": 0.1423640251159668, + "learning_rate": 0.00016858288770053476, + "loss": 0.2565, + "step": 475 + }, + { + "epoch": 0.4761339126629365, + "grad_norm": 0.13975542783737183, + "learning_rate": 0.00016851604278074865, + "loss": 0.272, + "step": 476 + }, + { + "epoch": 0.47713419399206025, + "grad_norm": 0.13784927129745483, + "learning_rate": 0.00016844919786096257, + "loss": 0.2651, + "step": 477 + }, + { + "epoch": 0.4781344753211841, + "grad_norm": 0.14744417369365692, + "learning_rate": 0.0001683823529411765, + "loss": 0.2638, + "step": 478 + }, + { + "epoch": 0.4791347566503079, + "grad_norm": 0.1385982781648636, + "learning_rate": 0.00016831550802139038, + "loss": 0.2692, + "step": 479 + }, + { + "epoch": 0.4801350379794317, + "grad_norm": 0.1384260654449463, + "learning_rate": 0.00016824866310160427, + "loss": 0.2586, + "step": 480 + }, + { + "epoch": 0.4811353193085555, + "grad_norm": 0.13358135521411896, + "learning_rate": 0.0001681818181818182, + "loss": 0.2866, + "step": 481 + }, + { + "epoch": 0.48213560063767935, + "grad_norm": 0.14295433461666107, + "learning_rate": 0.0001681149732620321, + "loss": 0.2777, + "step": 482 + }, + { + "epoch": 0.4831358819668032, + "grad_norm": 0.14078019559383392, + "learning_rate": 0.000168048128342246, + "loss": 0.2546, + "step": 483 + }, + { + "epoch": 0.48413616329592696, + "grad_norm": 0.13567811250686646, + "learning_rate": 0.00016798128342245992, + "loss": 0.2686, + "step": 484 + }, + { + "epoch": 0.4851364446250508, + "grad_norm": 0.12754715979099274, + "learning_rate": 0.0001679144385026738, + "loss": 0.2519, + "step": 485 + }, + { + "epoch": 0.4861367259541746, + "grad_norm": 0.13427051901817322, + "learning_rate": 0.0001678475935828877, + "loss": 0.2619, + "step": 486 + }, + { + "epoch": 0.48713700728329845, + "grad_norm": 0.12028037011623383, + "learning_rate": 0.00016778074866310162, + "loss": 0.2538, + "step": 487 + }, + { + "epoch": 0.4881372886124222, + "grad_norm": 0.1394716054201126, + "learning_rate": 0.00016771390374331553, + "loss": 0.2674, + "step": 488 + }, + { + "epoch": 0.48913756994154606, + "grad_norm": 0.12311938405036926, + "learning_rate": 0.00016764705882352942, + "loss": 0.2508, + "step": 489 + }, + { + "epoch": 0.4901378512706699, + "grad_norm": 0.1261584460735321, + "learning_rate": 0.00016758021390374331, + "loss": 0.2573, + "step": 490 + }, + { + "epoch": 0.4911381325997937, + "grad_norm": 0.12742400169372559, + "learning_rate": 0.00016751336898395723, + "loss": 0.2585, + "step": 491 + }, + { + "epoch": 0.4921384139289175, + "grad_norm": 0.1366826891899109, + "learning_rate": 0.00016744652406417115, + "loss": 0.2762, + "step": 492 + }, + { + "epoch": 0.4931386952580413, + "grad_norm": 0.13107511401176453, + "learning_rate": 0.00016737967914438504, + "loss": 0.2466, + "step": 493 + }, + { + "epoch": 0.49413897658716516, + "grad_norm": 0.11613395810127258, + "learning_rate": 0.00016731283422459893, + "loss": 0.2393, + "step": 494 + }, + { + "epoch": 0.49513925791628893, + "grad_norm": 0.1225895881652832, + "learning_rate": 0.00016724598930481282, + "loss": 0.2616, + "step": 495 + }, + { + "epoch": 0.49613953924541276, + "grad_norm": 0.12995213270187378, + "learning_rate": 0.00016717914438502674, + "loss": 0.2739, + "step": 496 + }, + { + "epoch": 0.4971398205745366, + "grad_norm": 0.12202667444944382, + "learning_rate": 0.00016711229946524066, + "loss": 0.2475, + "step": 497 + }, + { + "epoch": 0.4981401019036604, + "grad_norm": 0.12723197042942047, + "learning_rate": 0.00016704545454545455, + "loss": 0.2716, + "step": 498 + }, + { + "epoch": 0.4991403832327842, + "grad_norm": 0.11711172759532928, + "learning_rate": 0.00016697860962566844, + "loss": 0.2597, + "step": 499 + }, + { + "epoch": 0.5001406645619081, + "grad_norm": 0.13298696279525757, + "learning_rate": 0.00016691176470588236, + "loss": 0.254, + "step": 500 + }, + { + "epoch": 0.5011409458910319, + "grad_norm": 0.12314271181821823, + "learning_rate": 0.00016684491978609628, + "loss": 0.2573, + "step": 501 + }, + { + "epoch": 0.5021412272201556, + "grad_norm": 0.1264898180961609, + "learning_rate": 0.00016677807486631017, + "loss": 0.2592, + "step": 502 + }, + { + "epoch": 0.5031415085492795, + "grad_norm": 0.13533622026443481, + "learning_rate": 0.00016671122994652406, + "loss": 0.2568, + "step": 503 + }, + { + "epoch": 0.5041417898784033, + "grad_norm": 0.12481293082237244, + "learning_rate": 0.00016664438502673798, + "loss": 0.2591, + "step": 504 + }, + { + "epoch": 0.5051420712075271, + "grad_norm": 0.12333500385284424, + "learning_rate": 0.00016657754010695187, + "loss": 0.2539, + "step": 505 + }, + { + "epoch": 0.506142352536651, + "grad_norm": 0.12423723191022873, + "learning_rate": 0.00016651069518716578, + "loss": 0.2745, + "step": 506 + }, + { + "epoch": 0.5071426338657747, + "grad_norm": 0.11766920238733292, + "learning_rate": 0.0001664438502673797, + "loss": 0.2379, + "step": 507 + }, + { + "epoch": 0.5081429151948985, + "grad_norm": 0.12503701448440552, + "learning_rate": 0.0001663770053475936, + "loss": 0.2495, + "step": 508 + }, + { + "epoch": 0.5091431965240224, + "grad_norm": 0.12612241506576538, + "learning_rate": 0.00016631016042780748, + "loss": 0.2475, + "step": 509 + }, + { + "epoch": 0.5101434778531462, + "grad_norm": 0.13151812553405762, + "learning_rate": 0.0001662433155080214, + "loss": 0.263, + "step": 510 + }, + { + "epoch": 0.5111437591822701, + "grad_norm": 0.12832529842853546, + "learning_rate": 0.00016617647058823532, + "loss": 0.2528, + "step": 511 + }, + { + "epoch": 0.5121440405113938, + "grad_norm": 0.13635586202144623, + "learning_rate": 0.0001661096256684492, + "loss": 0.2651, + "step": 512 + }, + { + "epoch": 0.5131443218405176, + "grad_norm": 0.13641208410263062, + "learning_rate": 0.0001660427807486631, + "loss": 0.2623, + "step": 513 + }, + { + "epoch": 0.5141446031696415, + "grad_norm": 0.1351822316646576, + "learning_rate": 0.000165975935828877, + "loss": 0.2572, + "step": 514 + }, + { + "epoch": 0.5151448844987653, + "grad_norm": 0.11805664747953415, + "learning_rate": 0.00016590909090909094, + "loss": 0.2554, + "step": 515 + }, + { + "epoch": 0.516145165827889, + "grad_norm": 0.1333199441432953, + "learning_rate": 0.00016584224598930483, + "loss": 0.2595, + "step": 516 + }, + { + "epoch": 0.5171454471570129, + "grad_norm": 0.12360018491744995, + "learning_rate": 0.00016577540106951872, + "loss": 0.2498, + "step": 517 + }, + { + "epoch": 0.5181457284861367, + "grad_norm": 0.13380879163742065, + "learning_rate": 0.0001657085561497326, + "loss": 0.2446, + "step": 518 + }, + { + "epoch": 0.5191460098152605, + "grad_norm": 0.1117963120341301, + "learning_rate": 0.00016564171122994653, + "loss": 0.2414, + "step": 519 + }, + { + "epoch": 0.5201462911443844, + "grad_norm": 0.1254943162202835, + "learning_rate": 0.00016557486631016045, + "loss": 0.2727, + "step": 520 + }, + { + "epoch": 0.5211465724735082, + "grad_norm": 0.13776473701000214, + "learning_rate": 0.00016550802139037434, + "loss": 0.2601, + "step": 521 + }, + { + "epoch": 0.522146853802632, + "grad_norm": 0.12182191014289856, + "learning_rate": 0.00016544117647058823, + "loss": 0.2566, + "step": 522 + }, + { + "epoch": 0.5231471351317558, + "grad_norm": 0.10954313725233078, + "learning_rate": 0.00016537433155080215, + "loss": 0.2477, + "step": 523 + }, + { + "epoch": 0.5241474164608796, + "grad_norm": 0.11662272363901138, + "learning_rate": 0.00016530748663101606, + "loss": 0.2619, + "step": 524 + }, + { + "epoch": 0.5251476977900035, + "grad_norm": 0.12759320437908173, + "learning_rate": 0.00016524064171122995, + "loss": 0.2661, + "step": 525 + }, + { + "epoch": 0.5261479791191273, + "grad_norm": 0.12759540975093842, + "learning_rate": 0.00016517379679144387, + "loss": 0.2464, + "step": 526 + }, + { + "epoch": 0.527148260448251, + "grad_norm": 0.12509910762310028, + "learning_rate": 0.00016510695187165776, + "loss": 0.2491, + "step": 527 + }, + { + "epoch": 0.5281485417773749, + "grad_norm": 0.11390827596187592, + "learning_rate": 0.00016504010695187165, + "loss": 0.2435, + "step": 528 + }, + { + "epoch": 0.5291488231064987, + "grad_norm": 0.13615134358406067, + "learning_rate": 0.00016497326203208557, + "loss": 0.2917, + "step": 529 + }, + { + "epoch": 0.5301491044356225, + "grad_norm": 0.12437987327575684, + "learning_rate": 0.0001649064171122995, + "loss": 0.2504, + "step": 530 + }, + { + "epoch": 0.5311493857647464, + "grad_norm": 0.12392181158065796, + "learning_rate": 0.00016483957219251338, + "loss": 0.258, + "step": 531 + }, + { + "epoch": 0.5321496670938701, + "grad_norm": 0.1262609362602234, + "learning_rate": 0.00016477272727272727, + "loss": 0.2456, + "step": 532 + }, + { + "epoch": 0.533149948422994, + "grad_norm": 0.12495619058609009, + "learning_rate": 0.0001647058823529412, + "loss": 0.2436, + "step": 533 + }, + { + "epoch": 0.5341502297521178, + "grad_norm": 0.12220554053783417, + "learning_rate": 0.0001646390374331551, + "loss": 0.2478, + "step": 534 + }, + { + "epoch": 0.5351505110812416, + "grad_norm": 0.11760423332452774, + "learning_rate": 0.000164572192513369, + "loss": 0.2496, + "step": 535 + }, + { + "epoch": 0.5361507924103655, + "grad_norm": 0.11106106638908386, + "learning_rate": 0.0001645053475935829, + "loss": 0.245, + "step": 536 + }, + { + "epoch": 0.5371510737394892, + "grad_norm": 0.12540924549102783, + "learning_rate": 0.00016443850267379678, + "loss": 0.2467, + "step": 537 + }, + { + "epoch": 0.538151355068613, + "grad_norm": 0.13700392842292786, + "learning_rate": 0.0001643716577540107, + "loss": 0.2678, + "step": 538 + }, + { + "epoch": 0.5391516363977369, + "grad_norm": 0.12768560647964478, + "learning_rate": 0.00016430481283422462, + "loss": 0.2493, + "step": 539 + }, + { + "epoch": 0.5401519177268607, + "grad_norm": 0.12279049307107925, + "learning_rate": 0.0001642379679144385, + "loss": 0.2406, + "step": 540 + }, + { + "epoch": 0.5411521990559846, + "grad_norm": 0.1347498744726181, + "learning_rate": 0.0001641711229946524, + "loss": 0.2595, + "step": 541 + }, + { + "epoch": 0.5421524803851083, + "grad_norm": 0.12546177208423615, + "learning_rate": 0.00016410427807486631, + "loss": 0.2429, + "step": 542 + }, + { + "epoch": 0.5431527617142321, + "grad_norm": 0.13467158377170563, + "learning_rate": 0.00016403743315508023, + "loss": 0.246, + "step": 543 + }, + { + "epoch": 0.544153043043356, + "grad_norm": 0.146375834941864, + "learning_rate": 0.00016397058823529412, + "loss": 0.2594, + "step": 544 + }, + { + "epoch": 0.5451533243724798, + "grad_norm": 0.11646528542041779, + "learning_rate": 0.00016390374331550801, + "loss": 0.2473, + "step": 545 + }, + { + "epoch": 0.5461536057016035, + "grad_norm": 0.13053809106349945, + "learning_rate": 0.00016383689839572193, + "loss": 0.2621, + "step": 546 + }, + { + "epoch": 0.5471538870307274, + "grad_norm": 0.124763622879982, + "learning_rate": 0.00016377005347593582, + "loss": 0.2546, + "step": 547 + }, + { + "epoch": 0.5481541683598512, + "grad_norm": 0.12095394730567932, + "learning_rate": 0.00016370320855614974, + "loss": 0.2725, + "step": 548 + }, + { + "epoch": 0.549154449688975, + "grad_norm": 0.1425929069519043, + "learning_rate": 0.00016363636363636366, + "loss": 0.2634, + "step": 549 + }, + { + "epoch": 0.5501547310180989, + "grad_norm": 0.12262514978647232, + "learning_rate": 0.00016356951871657755, + "loss": 0.2632, + "step": 550 + }, + { + "epoch": 0.5511550123472226, + "grad_norm": 0.1277114897966385, + "learning_rate": 0.00016350267379679144, + "loss": 0.2596, + "step": 551 + }, + { + "epoch": 0.5521552936763465, + "grad_norm": 0.12494557350873947, + "learning_rate": 0.00016343582887700536, + "loss": 0.2636, + "step": 552 + }, + { + "epoch": 0.5531555750054703, + "grad_norm": 0.1309121549129486, + "learning_rate": 0.00016336898395721928, + "loss": 0.2549, + "step": 553 + }, + { + "epoch": 0.5541558563345941, + "grad_norm": 0.13584892451763153, + "learning_rate": 0.00016330213903743317, + "loss": 0.2556, + "step": 554 + }, + { + "epoch": 0.555156137663718, + "grad_norm": 0.13664132356643677, + "learning_rate": 0.00016323529411764706, + "loss": 0.2721, + "step": 555 + }, + { + "epoch": 0.5561564189928417, + "grad_norm": 0.12577325105667114, + "learning_rate": 0.00016316844919786095, + "loss": 0.2442, + "step": 556 + }, + { + "epoch": 0.5571567003219655, + "grad_norm": 0.13021226227283478, + "learning_rate": 0.0001631016042780749, + "loss": 0.2472, + "step": 557 + }, + { + "epoch": 0.5581569816510894, + "grad_norm": 0.11993830651044846, + "learning_rate": 0.00016303475935828878, + "loss": 0.255, + "step": 558 + }, + { + "epoch": 0.5591572629802132, + "grad_norm": 0.12341359257698059, + "learning_rate": 0.00016296791443850268, + "loss": 0.2527, + "step": 559 + }, + { + "epoch": 0.560157544309337, + "grad_norm": 0.1230374202132225, + "learning_rate": 0.00016290106951871657, + "loss": 0.2481, + "step": 560 + }, + { + "epoch": 0.5611578256384608, + "grad_norm": 0.13662074506282806, + "learning_rate": 0.00016283422459893048, + "loss": 0.2761, + "step": 561 + }, + { + "epoch": 0.5621581069675846, + "grad_norm": 0.11973608285188675, + "learning_rate": 0.0001627673796791444, + "loss": 0.2474, + "step": 562 + }, + { + "epoch": 0.5631583882967085, + "grad_norm": 0.13744860887527466, + "learning_rate": 0.0001627005347593583, + "loss": 0.2792, + "step": 563 + }, + { + "epoch": 0.5641586696258323, + "grad_norm": 0.1280643492937088, + "learning_rate": 0.00016263368983957218, + "loss": 0.2622, + "step": 564 + }, + { + "epoch": 0.565158950954956, + "grad_norm": 0.11882605403661728, + "learning_rate": 0.0001625668449197861, + "loss": 0.2546, + "step": 565 + }, + { + "epoch": 0.5661592322840799, + "grad_norm": 0.13407133519649506, + "learning_rate": 0.00016250000000000002, + "loss": 0.2581, + "step": 566 + }, + { + "epoch": 0.5671595136132037, + "grad_norm": 0.12686043977737427, + "learning_rate": 0.0001624331550802139, + "loss": 0.2414, + "step": 567 + }, + { + "epoch": 0.5681597949423275, + "grad_norm": 0.13043813407421112, + "learning_rate": 0.00016236631016042783, + "loss": 0.2337, + "step": 568 + }, + { + "epoch": 0.5691600762714514, + "grad_norm": 0.13808302581310272, + "learning_rate": 0.00016229946524064172, + "loss": 0.2548, + "step": 569 + }, + { + "epoch": 0.5701603576005752, + "grad_norm": 0.14317278563976288, + "learning_rate": 0.0001622326203208556, + "loss": 0.2701, + "step": 570 + }, + { + "epoch": 0.5711606389296989, + "grad_norm": 0.1243097335100174, + "learning_rate": 0.00016216577540106953, + "loss": 0.2575, + "step": 571 + }, + { + "epoch": 0.5721609202588228, + "grad_norm": 0.11183468997478485, + "learning_rate": 0.00016209893048128345, + "loss": 0.2555, + "step": 572 + }, + { + "epoch": 0.5731612015879466, + "grad_norm": 0.1294286698102951, + "learning_rate": 0.00016203208556149734, + "loss": 0.2647, + "step": 573 + }, + { + "epoch": 0.5741614829170705, + "grad_norm": 0.13197191059589386, + "learning_rate": 0.00016196524064171123, + "loss": 0.2637, + "step": 574 + }, + { + "epoch": 0.5751617642461943, + "grad_norm": 0.12841752171516418, + "learning_rate": 0.00016189839572192515, + "loss": 0.2647, + "step": 575 + }, + { + "epoch": 0.576162045575318, + "grad_norm": 0.11670025438070297, + "learning_rate": 0.00016183155080213906, + "loss": 0.2443, + "step": 576 + }, + { + "epoch": 0.5771623269044419, + "grad_norm": 0.12932729721069336, + "learning_rate": 0.00016176470588235295, + "loss": 0.2413, + "step": 577 + }, + { + "epoch": 0.5781626082335657, + "grad_norm": 0.12478774785995483, + "learning_rate": 0.00016169786096256684, + "loss": 0.2456, + "step": 578 + }, + { + "epoch": 0.5791628895626895, + "grad_norm": 0.13175909221172333, + "learning_rate": 0.00016163101604278074, + "loss": 0.2607, + "step": 579 + }, + { + "epoch": 0.5801631708918134, + "grad_norm": 0.13016562163829803, + "learning_rate": 0.00016156417112299468, + "loss": 0.2488, + "step": 580 + }, + { + "epoch": 0.5811634522209371, + "grad_norm": 0.13259732723236084, + "learning_rate": 0.00016149732620320857, + "loss": 0.2586, + "step": 581 + }, + { + "epoch": 0.5821637335500609, + "grad_norm": 0.12380441278219223, + "learning_rate": 0.00016143048128342246, + "loss": 0.26, + "step": 582 + }, + { + "epoch": 0.5831640148791848, + "grad_norm": 0.12515774369239807, + "learning_rate": 0.00016136363636363635, + "loss": 0.2459, + "step": 583 + }, + { + "epoch": 0.5841642962083086, + "grad_norm": 0.13320566713809967, + "learning_rate": 0.00016129679144385027, + "loss": 0.2594, + "step": 584 + }, + { + "epoch": 0.5851645775374325, + "grad_norm": 0.11543724685907364, + "learning_rate": 0.0001612299465240642, + "loss": 0.2499, + "step": 585 + }, + { + "epoch": 0.5861648588665562, + "grad_norm": 0.12284442782402039, + "learning_rate": 0.00016116310160427808, + "loss": 0.2472, + "step": 586 + }, + { + "epoch": 0.58716514019568, + "grad_norm": 0.12199356406927109, + "learning_rate": 0.000161096256684492, + "loss": 0.2454, + "step": 587 + }, + { + "epoch": 0.5881654215248039, + "grad_norm": 0.11472079902887344, + "learning_rate": 0.0001610294117647059, + "loss": 0.2633, + "step": 588 + }, + { + "epoch": 0.5891657028539277, + "grad_norm": 0.1262284517288208, + "learning_rate": 0.0001609625668449198, + "loss": 0.2683, + "step": 589 + }, + { + "epoch": 0.5901659841830514, + "grad_norm": 0.11997257173061371, + "learning_rate": 0.0001608957219251337, + "loss": 0.2478, + "step": 590 + }, + { + "epoch": 0.5911662655121753, + "grad_norm": 0.13336658477783203, + "learning_rate": 0.00016082887700534762, + "loss": 0.2688, + "step": 591 + }, + { + "epoch": 0.5921665468412991, + "grad_norm": 0.12463680654764175, + "learning_rate": 0.0001607620320855615, + "loss": 0.2485, + "step": 592 + }, + { + "epoch": 0.5931668281704229, + "grad_norm": 0.11950656771659851, + "learning_rate": 0.0001606951871657754, + "loss": 0.2439, + "step": 593 + }, + { + "epoch": 0.5941671094995468, + "grad_norm": 0.10655203461647034, + "learning_rate": 0.00016062834224598931, + "loss": 0.2286, + "step": 594 + }, + { + "epoch": 0.5951673908286705, + "grad_norm": 0.12858037650585175, + "learning_rate": 0.00016056149732620323, + "loss": 0.2491, + "step": 595 + }, + { + "epoch": 0.5961676721577944, + "grad_norm": 0.1166888102889061, + "learning_rate": 0.00016049465240641712, + "loss": 0.2524, + "step": 596 + }, + { + "epoch": 0.5971679534869182, + "grad_norm": 0.1221320778131485, + "learning_rate": 0.00016042780748663101, + "loss": 0.2475, + "step": 597 + }, + { + "epoch": 0.598168234816042, + "grad_norm": 0.1133677139878273, + "learning_rate": 0.00016036096256684493, + "loss": 0.2374, + "step": 598 + }, + { + "epoch": 0.5991685161451659, + "grad_norm": 0.12312517315149307, + "learning_rate": 0.00016029411764705885, + "loss": 0.245, + "step": 599 + }, + { + "epoch": 0.6001687974742896, + "grad_norm": 0.15150807797908783, + "learning_rate": 0.00016022727272727274, + "loss": 0.2657, + "step": 600 + }, + { + "epoch": 0.6011690788034134, + "grad_norm": 0.12168843299150467, + "learning_rate": 0.00016016042780748663, + "loss": 0.2444, + "step": 601 + }, + { + "epoch": 0.6021693601325373, + "grad_norm": 0.12176227569580078, + "learning_rate": 0.00016009358288770052, + "loss": 0.2593, + "step": 602 + }, + { + "epoch": 0.6031696414616611, + "grad_norm": 0.11734271049499512, + "learning_rate": 0.00016002673796791444, + "loss": 0.2451, + "step": 603 + }, + { + "epoch": 0.604169922790785, + "grad_norm": 0.1430376023054123, + "learning_rate": 0.00015995989304812836, + "loss": 0.2444, + "step": 604 + }, + { + "epoch": 0.6051702041199087, + "grad_norm": 0.11792927980422974, + "learning_rate": 0.00015989304812834225, + "loss": 0.2564, + "step": 605 + }, + { + "epoch": 0.6061704854490325, + "grad_norm": 0.14576008915901184, + "learning_rate": 0.00015982620320855614, + "loss": 0.2557, + "step": 606 + }, + { + "epoch": 0.6071707667781564, + "grad_norm": 0.13490746915340424, + "learning_rate": 0.00015975935828877006, + "loss": 0.2432, + "step": 607 + }, + { + "epoch": 0.6081710481072802, + "grad_norm": 0.1228419840335846, + "learning_rate": 0.00015969251336898398, + "loss": 0.273, + "step": 608 + }, + { + "epoch": 0.609171329436404, + "grad_norm": 0.11748429387807846, + "learning_rate": 0.00015962566844919787, + "loss": 0.2583, + "step": 609 + }, + { + "epoch": 0.6101716107655278, + "grad_norm": 0.11770988255739212, + "learning_rate": 0.00015955882352941178, + "loss": 0.2353, + "step": 610 + }, + { + "epoch": 0.6111718920946516, + "grad_norm": 0.1256726235151291, + "learning_rate": 0.00015949197860962568, + "loss": 0.2634, + "step": 611 + }, + { + "epoch": 0.6121721734237754, + "grad_norm": 0.12202870100736618, + "learning_rate": 0.00015942513368983957, + "loss": 0.261, + "step": 612 + }, + { + "epoch": 0.6131724547528993, + "grad_norm": 0.10914970934391022, + "learning_rate": 0.00015935828877005348, + "loss": 0.2458, + "step": 613 + }, + { + "epoch": 0.614172736082023, + "grad_norm": 0.11851291358470917, + "learning_rate": 0.0001592914438502674, + "loss": 0.2408, + "step": 614 + }, + { + "epoch": 0.6151730174111469, + "grad_norm": 0.12364912778139114, + "learning_rate": 0.0001592245989304813, + "loss": 0.2569, + "step": 615 + }, + { + "epoch": 0.6161732987402707, + "grad_norm": 0.1220770999789238, + "learning_rate": 0.00015915775401069518, + "loss": 0.2578, + "step": 616 + }, + { + "epoch": 0.6171735800693945, + "grad_norm": 0.12284126877784729, + "learning_rate": 0.0001590909090909091, + "loss": 0.2443, + "step": 617 + }, + { + "epoch": 0.6181738613985184, + "grad_norm": 0.12593187391757965, + "learning_rate": 0.00015902406417112302, + "loss": 0.2516, + "step": 618 + }, + { + "epoch": 0.6191741427276422, + "grad_norm": 0.10996383428573608, + "learning_rate": 0.0001589572192513369, + "loss": 0.2412, + "step": 619 + }, + { + "epoch": 0.6201744240567659, + "grad_norm": 0.11267533898353577, + "learning_rate": 0.0001588903743315508, + "loss": 0.2461, + "step": 620 + }, + { + "epoch": 0.6211747053858898, + "grad_norm": 0.12729158997535706, + "learning_rate": 0.0001588235294117647, + "loss": 0.2587, + "step": 621 + }, + { + "epoch": 0.6221749867150136, + "grad_norm": 0.12738372385501862, + "learning_rate": 0.00015875668449197864, + "loss": 0.2469, + "step": 622 + }, + { + "epoch": 0.6231752680441374, + "grad_norm": 0.13399261236190796, + "learning_rate": 0.00015868983957219253, + "loss": 0.2504, + "step": 623 + }, + { + "epoch": 0.6241755493732613, + "grad_norm": 0.127264142036438, + "learning_rate": 0.00015862299465240642, + "loss": 0.2632, + "step": 624 + }, + { + "epoch": 0.625175830702385, + "grad_norm": 0.12200193852186203, + "learning_rate": 0.0001585561497326203, + "loss": 0.2478, + "step": 625 + }, + { + "epoch": 0.6261761120315089, + "grad_norm": 0.11718714982271194, + "learning_rate": 0.00015848930481283423, + "loss": 0.2346, + "step": 626 + }, + { + "epoch": 0.6271763933606327, + "grad_norm": 0.12062099575996399, + "learning_rate": 0.00015842245989304815, + "loss": 0.2418, + "step": 627 + }, + { + "epoch": 0.6281766746897565, + "grad_norm": 0.11644895374774933, + "learning_rate": 0.00015835561497326204, + "loss": 0.2409, + "step": 628 + }, + { + "epoch": 0.6291769560188804, + "grad_norm": 0.1144137904047966, + "learning_rate": 0.00015828877005347595, + "loss": 0.2523, + "step": 629 + }, + { + "epoch": 0.6301772373480041, + "grad_norm": 0.13702888786792755, + "learning_rate": 0.00015822192513368984, + "loss": 0.2461, + "step": 630 + }, + { + "epoch": 0.6311775186771279, + "grad_norm": 0.13842301070690155, + "learning_rate": 0.00015815508021390376, + "loss": 0.2505, + "step": 631 + }, + { + "epoch": 0.6321778000062518, + "grad_norm": 0.10883615911006927, + "learning_rate": 0.00015808823529411765, + "loss": 0.2369, + "step": 632 + }, + { + "epoch": 0.6331780813353756, + "grad_norm": 0.11406701058149338, + "learning_rate": 0.00015802139037433157, + "loss": 0.248, + "step": 633 + }, + { + "epoch": 0.6341783626644993, + "grad_norm": 0.11928481608629227, + "learning_rate": 0.00015795454545454546, + "loss": 0.2581, + "step": 634 + }, + { + "epoch": 0.6351786439936232, + "grad_norm": 0.11724626272916794, + "learning_rate": 0.00015788770053475935, + "loss": 0.2507, + "step": 635 + }, + { + "epoch": 0.636178925322747, + "grad_norm": 0.11574576050043106, + "learning_rate": 0.00015782085561497327, + "loss": 0.2356, + "step": 636 + }, + { + "epoch": 0.6371792066518709, + "grad_norm": 0.11877516657114029, + "learning_rate": 0.0001577540106951872, + "loss": 0.2447, + "step": 637 + }, + { + "epoch": 0.6381794879809947, + "grad_norm": 0.10799930989742279, + "learning_rate": 0.00015768716577540108, + "loss": 0.2453, + "step": 638 + }, + { + "epoch": 0.6391797693101184, + "grad_norm": 0.11875832825899124, + "learning_rate": 0.00015762032085561497, + "loss": 0.2367, + "step": 639 + }, + { + "epoch": 0.6401800506392423, + "grad_norm": 0.11106745898723602, + "learning_rate": 0.0001575534759358289, + "loss": 0.2388, + "step": 640 + }, + { + "epoch": 0.6411803319683661, + "grad_norm": 0.11930815130472183, + "learning_rate": 0.0001574866310160428, + "loss": 0.2446, + "step": 641 + }, + { + "epoch": 0.6421806132974899, + "grad_norm": 0.12062037736177444, + "learning_rate": 0.0001574197860962567, + "loss": 0.2526, + "step": 642 + }, + { + "epoch": 0.6431808946266138, + "grad_norm": 0.13425442576408386, + "learning_rate": 0.0001573529411764706, + "loss": 0.2553, + "step": 643 + }, + { + "epoch": 0.6441811759557375, + "grad_norm": 0.12765665352344513, + "learning_rate": 0.00015728609625668448, + "loss": 0.2706, + "step": 644 + }, + { + "epoch": 0.6451814572848613, + "grad_norm": 0.11713891476392746, + "learning_rate": 0.0001572192513368984, + "loss": 0.2546, + "step": 645 + }, + { + "epoch": 0.6461817386139852, + "grad_norm": 0.1253063678741455, + "learning_rate": 0.00015715240641711231, + "loss": 0.2696, + "step": 646 + }, + { + "epoch": 0.647182019943109, + "grad_norm": 0.11812195181846619, + "learning_rate": 0.0001570855614973262, + "loss": 0.2537, + "step": 647 + }, + { + "epoch": 0.6481823012722329, + "grad_norm": 0.11805175244808197, + "learning_rate": 0.0001570187165775401, + "loss": 0.2453, + "step": 648 + }, + { + "epoch": 0.6491825826013566, + "grad_norm": 0.12036430090665817, + "learning_rate": 0.00015695187165775401, + "loss": 0.2357, + "step": 649 + }, + { + "epoch": 0.6501828639304804, + "grad_norm": 0.11339133977890015, + "learning_rate": 0.00015688502673796793, + "loss": 0.2471, + "step": 650 + }, + { + "epoch": 0.6511831452596043, + "grad_norm": 0.13621383905410767, + "learning_rate": 0.00015681818181818182, + "loss": 0.2789, + "step": 651 + }, + { + "epoch": 0.6521834265887281, + "grad_norm": 0.13126809895038605, + "learning_rate": 0.00015675133689839574, + "loss": 0.264, + "step": 652 + }, + { + "epoch": 0.6531837079178519, + "grad_norm": 0.1214464008808136, + "learning_rate": 0.00015668449197860963, + "loss": 0.2404, + "step": 653 + }, + { + "epoch": 0.6541839892469757, + "grad_norm": 0.11669190973043442, + "learning_rate": 0.00015661764705882352, + "loss": 0.2404, + "step": 654 + }, + { + "epoch": 0.6551842705760995, + "grad_norm": 0.12724418938159943, + "learning_rate": 0.00015655080213903744, + "loss": 0.2688, + "step": 655 + }, + { + "epoch": 0.6561845519052233, + "grad_norm": 0.12160211056470871, + "learning_rate": 0.00015648395721925136, + "loss": 0.2739, + "step": 656 + }, + { + "epoch": 0.6571848332343472, + "grad_norm": 0.11892884224653244, + "learning_rate": 0.00015641711229946525, + "loss": 0.2438, + "step": 657 + }, + { + "epoch": 0.658185114563471, + "grad_norm": 0.12483203411102295, + "learning_rate": 0.00015635026737967914, + "loss": 0.2738, + "step": 658 + }, + { + "epoch": 0.6591853958925948, + "grad_norm": 0.1167154312133789, + "learning_rate": 0.00015628342245989306, + "loss": 0.2461, + "step": 659 + }, + { + "epoch": 0.6601856772217186, + "grad_norm": 0.12697070837020874, + "learning_rate": 0.00015621657754010698, + "loss": 0.249, + "step": 660 + }, + { + "epoch": 0.6611859585508424, + "grad_norm": 0.1177363395690918, + "learning_rate": 0.00015614973262032087, + "loss": 0.2439, + "step": 661 + }, + { + "epoch": 0.6621862398799663, + "grad_norm": 0.11203393340110779, + "learning_rate": 0.00015608288770053476, + "loss": 0.2261, + "step": 662 + }, + { + "epoch": 0.66318652120909, + "grad_norm": 0.13510490953922272, + "learning_rate": 0.00015601604278074865, + "loss": 0.248, + "step": 663 + }, + { + "epoch": 0.6641868025382138, + "grad_norm": 0.12733061611652374, + "learning_rate": 0.0001559491978609626, + "loss": 0.2341, + "step": 664 + }, + { + "epoch": 0.6651870838673377, + "grad_norm": 0.130961075425148, + "learning_rate": 0.00015588235294117648, + "loss": 0.2564, + "step": 665 + }, + { + "epoch": 0.6661873651964615, + "grad_norm": 0.12575258314609528, + "learning_rate": 0.00015581550802139037, + "loss": 0.2519, + "step": 666 + }, + { + "epoch": 0.6671876465255854, + "grad_norm": 0.11712280660867691, + "learning_rate": 0.00015574866310160427, + "loss": 0.2471, + "step": 667 + }, + { + "epoch": 0.6681879278547092, + "grad_norm": 0.11082516610622406, + "learning_rate": 0.00015568181818181818, + "loss": 0.2325, + "step": 668 + }, + { + "epoch": 0.6691882091838329, + "grad_norm": 0.1253071129322052, + "learning_rate": 0.0001556149732620321, + "loss": 0.2533, + "step": 669 + }, + { + "epoch": 0.6701884905129568, + "grad_norm": 0.1490536332130432, + "learning_rate": 0.000155548128342246, + "loss": 0.2538, + "step": 670 + }, + { + "epoch": 0.6711887718420806, + "grad_norm": 0.13031023740768433, + "learning_rate": 0.0001554812834224599, + "loss": 0.2575, + "step": 671 + }, + { + "epoch": 0.6721890531712044, + "grad_norm": 0.12916699051856995, + "learning_rate": 0.0001554144385026738, + "loss": 0.2462, + "step": 672 + }, + { + "epoch": 0.6731893345003283, + "grad_norm": 0.12561458349227905, + "learning_rate": 0.00015534759358288772, + "loss": 0.2409, + "step": 673 + }, + { + "epoch": 0.674189615829452, + "grad_norm": 0.12291021645069122, + "learning_rate": 0.0001552807486631016, + "loss": 0.2581, + "step": 674 + }, + { + "epoch": 0.6751898971585758, + "grad_norm": 0.11148926615715027, + "learning_rate": 0.00015521390374331553, + "loss": 0.2375, + "step": 675 + }, + { + "epoch": 0.6761901784876997, + "grad_norm": 0.12354960292577744, + "learning_rate": 0.00015514705882352942, + "loss": 0.2716, + "step": 676 + }, + { + "epoch": 0.6771904598168235, + "grad_norm": 0.12153778225183487, + "learning_rate": 0.0001550802139037433, + "loss": 0.2588, + "step": 677 + }, + { + "epoch": 0.6781907411459474, + "grad_norm": 0.11479944735765457, + "learning_rate": 0.00015501336898395723, + "loss": 0.2521, + "step": 678 + }, + { + "epoch": 0.6791910224750711, + "grad_norm": 0.1302914321422577, + "learning_rate": 0.00015494652406417115, + "loss": 0.2716, + "step": 679 + }, + { + "epoch": 0.6801913038041949, + "grad_norm": 0.11325110495090485, + "learning_rate": 0.00015487967914438504, + "loss": 0.2471, + "step": 680 + }, + { + "epoch": 0.6811915851333188, + "grad_norm": 0.12691296637058258, + "learning_rate": 0.00015481283422459893, + "loss": 0.2439, + "step": 681 + }, + { + "epoch": 0.6821918664624426, + "grad_norm": 0.11070741713047028, + "learning_rate": 0.00015474598930481284, + "loss": 0.2387, + "step": 682 + }, + { + "epoch": 0.6831921477915663, + "grad_norm": 0.11579304933547974, + "learning_rate": 0.00015467914438502676, + "loss": 0.235, + "step": 683 + }, + { + "epoch": 0.6841924291206902, + "grad_norm": 0.12047869712114334, + "learning_rate": 0.00015461229946524065, + "loss": 0.272, + "step": 684 + }, + { + "epoch": 0.685192710449814, + "grad_norm": 0.11976123601198196, + "learning_rate": 0.00015454545454545454, + "loss": 0.2452, + "step": 685 + }, + { + "epoch": 0.6861929917789378, + "grad_norm": 0.13170906901359558, + "learning_rate": 0.00015447860962566844, + "loss": 0.2521, + "step": 686 + }, + { + "epoch": 0.6871932731080617, + "grad_norm": 0.1220417320728302, + "learning_rate": 0.00015441176470588238, + "loss": 0.2362, + "step": 687 + }, + { + "epoch": 0.6881935544371854, + "grad_norm": 0.12599536776542664, + "learning_rate": 0.00015434491978609627, + "loss": 0.2478, + "step": 688 + }, + { + "epoch": 0.6891938357663093, + "grad_norm": 0.12105037271976471, + "learning_rate": 0.00015427807486631016, + "loss": 0.2375, + "step": 689 + }, + { + "epoch": 0.6901941170954331, + "grad_norm": 0.12330012768507004, + "learning_rate": 0.00015421122994652405, + "loss": 0.2489, + "step": 690 + }, + { + "epoch": 0.6911943984245569, + "grad_norm": 0.1253179907798767, + "learning_rate": 0.00015414438502673797, + "loss": 0.2367, + "step": 691 + }, + { + "epoch": 0.6921946797536808, + "grad_norm": 0.11838237941265106, + "learning_rate": 0.0001540775401069519, + "loss": 0.2623, + "step": 692 + }, + { + "epoch": 0.6931949610828045, + "grad_norm": 0.13540798425674438, + "learning_rate": 0.00015401069518716578, + "loss": 0.2539, + "step": 693 + }, + { + "epoch": 0.6941952424119283, + "grad_norm": 0.14386463165283203, + "learning_rate": 0.0001539438502673797, + "loss": 0.2632, + "step": 694 + }, + { + "epoch": 0.6951955237410522, + "grad_norm": 0.12164293229579926, + "learning_rate": 0.0001538770053475936, + "loss": 0.2577, + "step": 695 + }, + { + "epoch": 0.696195805070176, + "grad_norm": 0.117221400141716, + "learning_rate": 0.0001538101604278075, + "loss": 0.2414, + "step": 696 + }, + { + "epoch": 0.6971960863992998, + "grad_norm": 0.12161596119403839, + "learning_rate": 0.0001537433155080214, + "loss": 0.2399, + "step": 697 + }, + { + "epoch": 0.6981963677284236, + "grad_norm": 0.1220688745379448, + "learning_rate": 0.00015367647058823531, + "loss": 0.2353, + "step": 698 + }, + { + "epoch": 0.6991966490575474, + "grad_norm": 0.1035013198852539, + "learning_rate": 0.0001536096256684492, + "loss": 0.2374, + "step": 699 + }, + { + "epoch": 0.7001969303866713, + "grad_norm": 0.11900071799755096, + "learning_rate": 0.0001535427807486631, + "loss": 0.2552, + "step": 700 + }, + { + "epoch": 0.7011972117157951, + "grad_norm": 0.11466988176107407, + "learning_rate": 0.00015347593582887701, + "loss": 0.2307, + "step": 701 + }, + { + "epoch": 0.7021974930449189, + "grad_norm": 0.12527823448181152, + "learning_rate": 0.00015340909090909093, + "loss": 0.2422, + "step": 702 + }, + { + "epoch": 0.7031977743740427, + "grad_norm": 0.11435528844594955, + "learning_rate": 0.00015334224598930482, + "loss": 0.2307, + "step": 703 + }, + { + "epoch": 0.7041980557031665, + "grad_norm": 0.11573023349046707, + "learning_rate": 0.00015327540106951871, + "loss": 0.2518, + "step": 704 + }, + { + "epoch": 0.7051983370322903, + "grad_norm": 0.12702280282974243, + "learning_rate": 0.00015320855614973263, + "loss": 0.2566, + "step": 705 + }, + { + "epoch": 0.7061986183614142, + "grad_norm": 0.11725278198719025, + "learning_rate": 0.00015314171122994655, + "loss": 0.2547, + "step": 706 + }, + { + "epoch": 0.707198899690538, + "grad_norm": 0.12754283845424652, + "learning_rate": 0.00015307486631016044, + "loss": 0.2405, + "step": 707 + }, + { + "epoch": 0.7081991810196617, + "grad_norm": 0.12315244227647781, + "learning_rate": 0.00015300802139037433, + "loss": 0.2512, + "step": 708 + }, + { + "epoch": 0.7091994623487856, + "grad_norm": 0.11458302289247513, + "learning_rate": 0.00015294117647058822, + "loss": 0.2319, + "step": 709 + }, + { + "epoch": 0.7101997436779094, + "grad_norm": 0.11183693259954453, + "learning_rate": 0.00015287433155080214, + "loss": 0.2488, + "step": 710 + }, + { + "epoch": 0.7112000250070333, + "grad_norm": 0.11887528002262115, + "learning_rate": 0.00015280748663101606, + "loss": 0.2379, + "step": 711 + }, + { + "epoch": 0.712200306336157, + "grad_norm": 0.1158289909362793, + "learning_rate": 0.00015274064171122995, + "loss": 0.235, + "step": 712 + }, + { + "epoch": 0.7132005876652808, + "grad_norm": 0.1151076927781105, + "learning_rate": 0.00015267379679144387, + "loss": 0.2584, + "step": 713 + }, + { + "epoch": 0.7142008689944047, + "grad_norm": 0.120317742228508, + "learning_rate": 0.00015260695187165776, + "loss": 0.2756, + "step": 714 + }, + { + "epoch": 0.7152011503235285, + "grad_norm": 0.11019955575466156, + "learning_rate": 0.00015254010695187168, + "loss": 0.2395, + "step": 715 + }, + { + "epoch": 0.7162014316526523, + "grad_norm": 0.12281118333339691, + "learning_rate": 0.00015247326203208557, + "loss": 0.2478, + "step": 716 + }, + { + "epoch": 0.7172017129817762, + "grad_norm": 0.11180153489112854, + "learning_rate": 0.00015240641711229948, + "loss": 0.235, + "step": 717 + }, + { + "epoch": 0.7182019943108999, + "grad_norm": 0.11637281626462936, + "learning_rate": 0.00015233957219251337, + "loss": 0.2423, + "step": 718 + }, + { + "epoch": 0.7192022756400237, + "grad_norm": 0.11919128149747849, + "learning_rate": 0.00015227272727272727, + "loss": 0.2499, + "step": 719 + }, + { + "epoch": 0.7202025569691476, + "grad_norm": 0.1156754121184349, + "learning_rate": 0.00015220588235294118, + "loss": 0.2602, + "step": 720 + }, + { + "epoch": 0.7212028382982714, + "grad_norm": 0.10709173232316971, + "learning_rate": 0.0001521390374331551, + "loss": 0.2384, + "step": 721 + }, + { + "epoch": 0.7222031196273953, + "grad_norm": 0.11579756438732147, + "learning_rate": 0.000152072192513369, + "loss": 0.2392, + "step": 722 + }, + { + "epoch": 0.723203400956519, + "grad_norm": 0.13141584396362305, + "learning_rate": 0.00015200534759358288, + "loss": 0.2516, + "step": 723 + }, + { + "epoch": 0.7242036822856428, + "grad_norm": 0.11077112704515457, + "learning_rate": 0.0001519385026737968, + "loss": 0.2315, + "step": 724 + }, + { + "epoch": 0.7252039636147667, + "grad_norm": 0.11116013675928116, + "learning_rate": 0.00015187165775401072, + "loss": 0.2485, + "step": 725 + }, + { + "epoch": 0.7262042449438905, + "grad_norm": 0.1086612194776535, + "learning_rate": 0.0001518048128342246, + "loss": 0.2437, + "step": 726 + }, + { + "epoch": 0.7272045262730142, + "grad_norm": 0.11183172464370728, + "learning_rate": 0.0001517379679144385, + "loss": 0.2463, + "step": 727 + }, + { + "epoch": 0.7282048076021381, + "grad_norm": 0.12008163332939148, + "learning_rate": 0.0001516711229946524, + "loss": 0.2515, + "step": 728 + }, + { + "epoch": 0.7292050889312619, + "grad_norm": 0.1177927553653717, + "learning_rate": 0.00015160427807486634, + "loss": 0.2587, + "step": 729 + }, + { + "epoch": 0.7302053702603858, + "grad_norm": 0.11384297907352448, + "learning_rate": 0.00015153743315508023, + "loss": 0.2549, + "step": 730 + }, + { + "epoch": 0.7312056515895096, + "grad_norm": 0.11532732099294662, + "learning_rate": 0.00015147058823529412, + "loss": 0.2448, + "step": 731 + }, + { + "epoch": 0.7322059329186333, + "grad_norm": 0.11613297462463379, + "learning_rate": 0.000151403743315508, + "loss": 0.2548, + "step": 732 + }, + { + "epoch": 0.7332062142477572, + "grad_norm": 0.11316312849521637, + "learning_rate": 0.00015133689839572193, + "loss": 0.2467, + "step": 733 + }, + { + "epoch": 0.734206495576881, + "grad_norm": 0.11662183701992035, + "learning_rate": 0.00015127005347593584, + "loss": 0.2352, + "step": 734 + }, + { + "epoch": 0.7352067769060048, + "grad_norm": 0.12867549061775208, + "learning_rate": 0.00015120320855614974, + "loss": 0.2484, + "step": 735 + }, + { + "epoch": 0.7362070582351287, + "grad_norm": 0.10538063943386078, + "learning_rate": 0.00015113636363636365, + "loss": 0.246, + "step": 736 + }, + { + "epoch": 0.7372073395642524, + "grad_norm": 0.1263480931520462, + "learning_rate": 0.00015106951871657754, + "loss": 0.2392, + "step": 737 + }, + { + "epoch": 0.7382076208933762, + "grad_norm": 0.14239145815372467, + "learning_rate": 0.00015100267379679146, + "loss": 0.2444, + "step": 738 + }, + { + "epoch": 0.7392079022225001, + "grad_norm": 0.11437618732452393, + "learning_rate": 0.00015093582887700535, + "loss": 0.2507, + "step": 739 + }, + { + "epoch": 0.7402081835516239, + "grad_norm": 0.12424953281879425, + "learning_rate": 0.00015086898395721927, + "loss": 0.2422, + "step": 740 + }, + { + "epoch": 0.7412084648807478, + "grad_norm": 0.14059457182884216, + "learning_rate": 0.00015080213903743316, + "loss": 0.2659, + "step": 741 + }, + { + "epoch": 0.7422087462098715, + "grad_norm": 0.114485003054142, + "learning_rate": 0.00015073529411764705, + "loss": 0.2325, + "step": 742 + }, + { + "epoch": 0.7432090275389953, + "grad_norm": 0.12970639765262604, + "learning_rate": 0.00015066844919786097, + "loss": 0.2444, + "step": 743 + }, + { + "epoch": 0.7442093088681192, + "grad_norm": 0.10770043730735779, + "learning_rate": 0.0001506016042780749, + "loss": 0.2466, + "step": 744 + }, + { + "epoch": 0.745209590197243, + "grad_norm": 0.10905428975820541, + "learning_rate": 0.00015053475935828878, + "loss": 0.2377, + "step": 745 + }, + { + "epoch": 0.7462098715263668, + "grad_norm": 0.1166926920413971, + "learning_rate": 0.00015046791443850267, + "loss": 0.2582, + "step": 746 + }, + { + "epoch": 0.7472101528554906, + "grad_norm": 0.11056409776210785, + "learning_rate": 0.0001504010695187166, + "loss": 0.2511, + "step": 747 + }, + { + "epoch": 0.7482104341846144, + "grad_norm": 0.11626572161912918, + "learning_rate": 0.0001503342245989305, + "loss": 0.2491, + "step": 748 + }, + { + "epoch": 0.7492107155137382, + "grad_norm": 0.12436028569936752, + "learning_rate": 0.0001502673796791444, + "loss": 0.2561, + "step": 749 + }, + { + "epoch": 0.7502109968428621, + "grad_norm": 0.11258986592292786, + "learning_rate": 0.0001502005347593583, + "loss": 0.2372, + "step": 750 + }, + { + "epoch": 0.7512112781719859, + "grad_norm": 0.11168336123228073, + "learning_rate": 0.00015013368983957218, + "loss": 0.2439, + "step": 751 + }, + { + "epoch": 0.7522115595011097, + "grad_norm": 0.12370362132787704, + "learning_rate": 0.0001500668449197861, + "loss": 0.2554, + "step": 752 + }, + { + "epoch": 0.7532118408302335, + "grad_norm": 0.12653499841690063, + "learning_rate": 0.00015000000000000001, + "loss": 0.2513, + "step": 753 + }, + { + "epoch": 0.7542121221593573, + "grad_norm": 0.11492380499839783, + "learning_rate": 0.0001499331550802139, + "loss": 0.2496, + "step": 754 + }, + { + "epoch": 0.7552124034884812, + "grad_norm": 0.11227967590093613, + "learning_rate": 0.00014986631016042782, + "loss": 0.2314, + "step": 755 + }, + { + "epoch": 0.756212684817605, + "grad_norm": 0.12645453214645386, + "learning_rate": 0.0001497994652406417, + "loss": 0.2613, + "step": 756 + }, + { + "epoch": 0.7572129661467287, + "grad_norm": 0.16427937150001526, + "learning_rate": 0.00014973262032085563, + "loss": 0.2551, + "step": 757 + }, + { + "epoch": 0.7582132474758526, + "grad_norm": 0.12859593331813812, + "learning_rate": 0.00014966577540106952, + "loss": 0.2694, + "step": 758 + }, + { + "epoch": 0.7592135288049764, + "grad_norm": 0.11306439340114594, + "learning_rate": 0.00014959893048128344, + "loss": 0.2542, + "step": 759 + }, + { + "epoch": 0.7602138101341002, + "grad_norm": 0.1307079792022705, + "learning_rate": 0.00014953208556149733, + "loss": 0.2631, + "step": 760 + }, + { + "epoch": 0.761214091463224, + "grad_norm": 0.14013920724391937, + "learning_rate": 0.00014946524064171122, + "loss": 0.2524, + "step": 761 + }, + { + "epoch": 0.7622143727923478, + "grad_norm": 0.1274118572473526, + "learning_rate": 0.00014939839572192514, + "loss": 0.2614, + "step": 762 + }, + { + "epoch": 0.7632146541214717, + "grad_norm": 0.11558174341917038, + "learning_rate": 0.00014933155080213906, + "loss": 0.2437, + "step": 763 + }, + { + "epoch": 0.7642149354505955, + "grad_norm": 0.11618223041296005, + "learning_rate": 0.00014926470588235295, + "loss": 0.259, + "step": 764 + }, + { + "epoch": 0.7652152167797193, + "grad_norm": 0.121167853474617, + "learning_rate": 0.00014919786096256684, + "loss": 0.248, + "step": 765 + }, + { + "epoch": 0.7662154981088432, + "grad_norm": 0.10980004817247391, + "learning_rate": 0.00014913101604278076, + "loss": 0.2315, + "step": 766 + }, + { + "epoch": 0.7672157794379669, + "grad_norm": 0.11669566482305527, + "learning_rate": 0.00014906417112299468, + "loss": 0.2483, + "step": 767 + }, + { + "epoch": 0.7682160607670907, + "grad_norm": 0.12446890026330948, + "learning_rate": 0.00014899732620320857, + "loss": 0.2469, + "step": 768 + }, + { + "epoch": 0.7692163420962146, + "grad_norm": 0.11734813451766968, + "learning_rate": 0.00014893048128342246, + "loss": 0.2406, + "step": 769 + }, + { + "epoch": 0.7702166234253384, + "grad_norm": 0.11169088631868362, + "learning_rate": 0.00014886363636363635, + "loss": 0.2371, + "step": 770 + }, + { + "epoch": 0.7712169047544621, + "grad_norm": 0.12620817124843597, + "learning_rate": 0.0001487967914438503, + "loss": 0.2485, + "step": 771 + }, + { + "epoch": 0.772217186083586, + "grad_norm": 0.13355882465839386, + "learning_rate": 0.00014872994652406418, + "loss": 0.2609, + "step": 772 + }, + { + "epoch": 0.7732174674127098, + "grad_norm": 0.12451037019491196, + "learning_rate": 0.00014866310160427807, + "loss": 0.2422, + "step": 773 + }, + { + "epoch": 0.7742177487418337, + "grad_norm": 0.11673011630773544, + "learning_rate": 0.000148596256684492, + "loss": 0.2395, + "step": 774 + }, + { + "epoch": 0.7752180300709575, + "grad_norm": 0.12303411215543747, + "learning_rate": 0.00014852941176470588, + "loss": 0.2421, + "step": 775 + }, + { + "epoch": 0.7762183114000812, + "grad_norm": 0.12523183226585388, + "learning_rate": 0.0001484625668449198, + "loss": 0.2413, + "step": 776 + }, + { + "epoch": 0.7772185927292051, + "grad_norm": 0.11674949526786804, + "learning_rate": 0.0001483957219251337, + "loss": 0.2209, + "step": 777 + }, + { + "epoch": 0.7782188740583289, + "grad_norm": 0.12070966511964798, + "learning_rate": 0.0001483288770053476, + "loss": 0.2345, + "step": 778 + }, + { + "epoch": 0.7792191553874527, + "grad_norm": 0.10922983288764954, + "learning_rate": 0.0001482620320855615, + "loss": 0.2506, + "step": 779 + }, + { + "epoch": 0.7802194367165766, + "grad_norm": 0.11496447771787643, + "learning_rate": 0.00014819518716577542, + "loss": 0.2464, + "step": 780 + }, + { + "epoch": 0.7812197180457003, + "grad_norm": 0.1120864599943161, + "learning_rate": 0.0001481283422459893, + "loss": 0.2507, + "step": 781 + }, + { + "epoch": 0.7822199993748241, + "grad_norm": 0.1168716624379158, + "learning_rate": 0.00014806149732620323, + "loss": 0.2517, + "step": 782 + }, + { + "epoch": 0.783220280703948, + "grad_norm": 0.11765114217996597, + "learning_rate": 0.00014799465240641712, + "loss": 0.2389, + "step": 783 + }, + { + "epoch": 0.7842205620330718, + "grad_norm": 0.11662907898426056, + "learning_rate": 0.000147927807486631, + "loss": 0.2598, + "step": 784 + }, + { + "epoch": 0.7852208433621957, + "grad_norm": 0.1114870011806488, + "learning_rate": 0.00014786096256684493, + "loss": 0.2373, + "step": 785 + }, + { + "epoch": 0.7862211246913194, + "grad_norm": 0.12278923392295837, + "learning_rate": 0.00014779411764705884, + "loss": 0.2466, + "step": 786 + }, + { + "epoch": 0.7872214060204432, + "grad_norm": 0.10689809173345566, + "learning_rate": 0.00014772727272727274, + "loss": 0.2209, + "step": 787 + }, + { + "epoch": 0.7882216873495671, + "grad_norm": 0.12380310893058777, + "learning_rate": 0.00014766042780748663, + "loss": 0.2501, + "step": 788 + }, + { + "epoch": 0.7892219686786909, + "grad_norm": 0.11559835076332092, + "learning_rate": 0.00014759358288770054, + "loss": 0.2395, + "step": 789 + }, + { + "epoch": 0.7902222500078147, + "grad_norm": 0.11426148563623428, + "learning_rate": 0.00014752673796791446, + "loss": 0.2565, + "step": 790 + }, + { + "epoch": 0.7912225313369385, + "grad_norm": 0.11781702190637589, + "learning_rate": 0.00014745989304812835, + "loss": 0.2316, + "step": 791 + }, + { + "epoch": 0.7922228126660623, + "grad_norm": 0.11257804930210114, + "learning_rate": 0.00014739304812834224, + "loss": 0.2318, + "step": 792 + }, + { + "epoch": 0.7932230939951862, + "grad_norm": 0.12982343137264252, + "learning_rate": 0.00014732620320855613, + "loss": 0.2527, + "step": 793 + }, + { + "epoch": 0.79422337532431, + "grad_norm": 0.11986357718706131, + "learning_rate": 0.00014725935828877008, + "loss": 0.2399, + "step": 794 + }, + { + "epoch": 0.7952236566534338, + "grad_norm": 0.12158485502004623, + "learning_rate": 0.00014719251336898397, + "loss": 0.2542, + "step": 795 + }, + { + "epoch": 0.7962239379825576, + "grad_norm": 0.1091170385479927, + "learning_rate": 0.00014712566844919786, + "loss": 0.2351, + "step": 796 + }, + { + "epoch": 0.7972242193116814, + "grad_norm": 0.1128731220960617, + "learning_rate": 0.00014705882352941178, + "loss": 0.2508, + "step": 797 + }, + { + "epoch": 0.7982245006408052, + "grad_norm": 0.10471203178167343, + "learning_rate": 0.00014699197860962567, + "loss": 0.2378, + "step": 798 + }, + { + "epoch": 0.7992247819699291, + "grad_norm": 0.10250696539878845, + "learning_rate": 0.0001469251336898396, + "loss": 0.2266, + "step": 799 + }, + { + "epoch": 0.8002250632990529, + "grad_norm": 0.10389907658100128, + "learning_rate": 0.00014685828877005348, + "loss": 0.2393, + "step": 800 + }, + { + "epoch": 0.8012253446281766, + "grad_norm": 0.11774886399507523, + "learning_rate": 0.0001467914438502674, + "loss": 0.233, + "step": 801 + }, + { + "epoch": 0.8022256259573005, + "grad_norm": 0.09853003919124603, + "learning_rate": 0.0001467245989304813, + "loss": 0.2174, + "step": 802 + }, + { + "epoch": 0.8032259072864243, + "grad_norm": 0.12377273291349411, + "learning_rate": 0.0001466577540106952, + "loss": 0.2449, + "step": 803 + }, + { + "epoch": 0.8042261886155482, + "grad_norm": 0.12278609722852707, + "learning_rate": 0.0001465909090909091, + "loss": 0.2449, + "step": 804 + }, + { + "epoch": 0.805226469944672, + "grad_norm": 0.12045742571353912, + "learning_rate": 0.00014652406417112301, + "loss": 0.247, + "step": 805 + }, + { + "epoch": 0.8062267512737957, + "grad_norm": 0.12080372869968414, + "learning_rate": 0.0001464572192513369, + "loss": 0.2501, + "step": 806 + }, + { + "epoch": 0.8072270326029196, + "grad_norm": 0.1103830561041832, + "learning_rate": 0.0001463903743315508, + "loss": 0.2416, + "step": 807 + }, + { + "epoch": 0.8082273139320434, + "grad_norm": 0.10650736093521118, + "learning_rate": 0.0001463235294117647, + "loss": 0.2363, + "step": 808 + }, + { + "epoch": 0.8092275952611672, + "grad_norm": 0.10822493582963943, + "learning_rate": 0.00014625668449197863, + "loss": 0.2315, + "step": 809 + }, + { + "epoch": 0.810227876590291, + "grad_norm": 0.13452479243278503, + "learning_rate": 0.00014618983957219252, + "loss": 0.2431, + "step": 810 + }, + { + "epoch": 0.8112281579194148, + "grad_norm": 0.12023711204528809, + "learning_rate": 0.0001461229946524064, + "loss": 0.2462, + "step": 811 + }, + { + "epoch": 0.8122284392485386, + "grad_norm": 0.10861025005578995, + "learning_rate": 0.00014605614973262033, + "loss": 0.2495, + "step": 812 + }, + { + "epoch": 0.8132287205776625, + "grad_norm": 0.11870504170656204, + "learning_rate": 0.00014598930481283425, + "loss": 0.2431, + "step": 813 + }, + { + "epoch": 0.8142290019067863, + "grad_norm": 0.11424995958805084, + "learning_rate": 0.00014592245989304814, + "loss": 0.2387, + "step": 814 + }, + { + "epoch": 0.8152292832359102, + "grad_norm": 0.12039054930210114, + "learning_rate": 0.00014585561497326203, + "loss": 0.2451, + "step": 815 + }, + { + "epoch": 0.8162295645650339, + "grad_norm": 0.13258178532123566, + "learning_rate": 0.00014578877005347595, + "loss": 0.2717, + "step": 816 + }, + { + "epoch": 0.8172298458941577, + "grad_norm": 0.11285381764173508, + "learning_rate": 0.00014572192513368984, + "loss": 0.2589, + "step": 817 + }, + { + "epoch": 0.8182301272232816, + "grad_norm": 0.1185690313577652, + "learning_rate": 0.00014565508021390376, + "loss": 0.2539, + "step": 818 + }, + { + "epoch": 0.8192304085524054, + "grad_norm": 0.11847860366106033, + "learning_rate": 0.00014558823529411765, + "loss": 0.2415, + "step": 819 + }, + { + "epoch": 0.8202306898815291, + "grad_norm": 0.11036616563796997, + "learning_rate": 0.00014552139037433157, + "loss": 0.2567, + "step": 820 + }, + { + "epoch": 0.821230971210653, + "grad_norm": 0.1137470081448555, + "learning_rate": 0.00014545454545454546, + "loss": 0.2441, + "step": 821 + }, + { + "epoch": 0.8222312525397768, + "grad_norm": 0.12307382375001907, + "learning_rate": 0.00014538770053475937, + "loss": 0.2532, + "step": 822 + }, + { + "epoch": 0.8232315338689006, + "grad_norm": 0.10465478897094727, + "learning_rate": 0.00014532085561497327, + "loss": 0.2417, + "step": 823 + }, + { + "epoch": 0.8242318151980245, + "grad_norm": 0.10899168252944946, + "learning_rate": 0.00014525401069518718, + "loss": 0.2449, + "step": 824 + }, + { + "epoch": 0.8252320965271482, + "grad_norm": 0.1114077940583229, + "learning_rate": 0.00014518716577540107, + "loss": 0.2282, + "step": 825 + }, + { + "epoch": 0.8262323778562721, + "grad_norm": 0.11783557385206223, + "learning_rate": 0.00014512032085561496, + "loss": 0.2537, + "step": 826 + }, + { + "epoch": 0.8272326591853959, + "grad_norm": 0.1194356232881546, + "learning_rate": 0.00014505347593582888, + "loss": 0.2476, + "step": 827 + }, + { + "epoch": 0.8282329405145197, + "grad_norm": 0.10246395319700241, + "learning_rate": 0.0001449866310160428, + "loss": 0.2307, + "step": 828 + }, + { + "epoch": 0.8292332218436436, + "grad_norm": 0.12487491220235825, + "learning_rate": 0.0001449197860962567, + "loss": 0.2488, + "step": 829 + }, + { + "epoch": 0.8302335031727673, + "grad_norm": 0.10684099048376083, + "learning_rate": 0.00014485294117647058, + "loss": 0.2567, + "step": 830 + }, + { + "epoch": 0.8312337845018911, + "grad_norm": 0.12656627595424652, + "learning_rate": 0.0001447860962566845, + "loss": 0.2291, + "step": 831 + }, + { + "epoch": 0.832234065831015, + "grad_norm": 0.11824151873588562, + "learning_rate": 0.00014471925133689842, + "loss": 0.2267, + "step": 832 + }, + { + "epoch": 0.8332343471601388, + "grad_norm": 0.1229713037610054, + "learning_rate": 0.0001446524064171123, + "loss": 0.2486, + "step": 833 + }, + { + "epoch": 0.8342346284892626, + "grad_norm": 0.12732572853565216, + "learning_rate": 0.0001445855614973262, + "loss": 0.2456, + "step": 834 + }, + { + "epoch": 0.8352349098183864, + "grad_norm": 0.12420819699764252, + "learning_rate": 0.0001445187165775401, + "loss": 0.2415, + "step": 835 + }, + { + "epoch": 0.8362351911475102, + "grad_norm": 0.11336098611354828, + "learning_rate": 0.00014445187165775404, + "loss": 0.2399, + "step": 836 + }, + { + "epoch": 0.8372354724766341, + "grad_norm": 0.11686563491821289, + "learning_rate": 0.00014438502673796793, + "loss": 0.2563, + "step": 837 + }, + { + "epoch": 0.8382357538057579, + "grad_norm": 0.11366021633148193, + "learning_rate": 0.00014431818181818182, + "loss": 0.2532, + "step": 838 + }, + { + "epoch": 0.8392360351348817, + "grad_norm": 0.11067181080579758, + "learning_rate": 0.00014425133689839574, + "loss": 0.2246, + "step": 839 + }, + { + "epoch": 0.8402363164640055, + "grad_norm": 0.10358266532421112, + "learning_rate": 0.00014418449197860963, + "loss": 0.2244, + "step": 840 + }, + { + "epoch": 0.8412365977931293, + "grad_norm": 0.12209904938936234, + "learning_rate": 0.00014411764705882354, + "loss": 0.264, + "step": 841 + }, + { + "epoch": 0.8422368791222531, + "grad_norm": 0.11398942768573761, + "learning_rate": 0.00014405080213903743, + "loss": 0.2467, + "step": 842 + }, + { + "epoch": 0.843237160451377, + "grad_norm": 0.1180686429142952, + "learning_rate": 0.00014398395721925135, + "loss": 0.2448, + "step": 843 + }, + { + "epoch": 0.8442374417805008, + "grad_norm": 0.11448219418525696, + "learning_rate": 0.00014391711229946524, + "loss": 0.2422, + "step": 844 + }, + { + "epoch": 0.8452377231096245, + "grad_norm": 0.10853015631437302, + "learning_rate": 0.00014385026737967916, + "loss": 0.2302, + "step": 845 + }, + { + "epoch": 0.8462380044387484, + "grad_norm": 0.13390398025512695, + "learning_rate": 0.00014378342245989305, + "loss": 0.2643, + "step": 846 + }, + { + "epoch": 0.8472382857678722, + "grad_norm": 0.11720698326826096, + "learning_rate": 0.00014371657754010697, + "loss": 0.2465, + "step": 847 + }, + { + "epoch": 0.8482385670969961, + "grad_norm": 0.10493404418230057, + "learning_rate": 0.00014364973262032086, + "loss": 0.2267, + "step": 848 + }, + { + "epoch": 0.8492388484261199, + "grad_norm": 0.11664943397045135, + "learning_rate": 0.00014358288770053475, + "loss": 0.2421, + "step": 849 + }, + { + "epoch": 0.8502391297552436, + "grad_norm": 0.1090892031788826, + "learning_rate": 0.00014351604278074867, + "loss": 0.2425, + "step": 850 + }, + { + "epoch": 0.8512394110843675, + "grad_norm": 0.11527938395738602, + "learning_rate": 0.0001434491978609626, + "loss": 0.2432, + "step": 851 + }, + { + "epoch": 0.8522396924134913, + "grad_norm": 0.11178325116634369, + "learning_rate": 0.00014338235294117648, + "loss": 0.2469, + "step": 852 + }, + { + "epoch": 0.8532399737426151, + "grad_norm": 0.11904676258563995, + "learning_rate": 0.00014331550802139037, + "loss": 0.2373, + "step": 853 + }, + { + "epoch": 0.854240255071739, + "grad_norm": 0.12647341191768646, + "learning_rate": 0.0001432486631016043, + "loss": 0.2361, + "step": 854 + }, + { + "epoch": 0.8552405364008627, + "grad_norm": 0.11088274419307709, + "learning_rate": 0.0001431818181818182, + "loss": 0.2414, + "step": 855 + }, + { + "epoch": 0.8562408177299866, + "grad_norm": 0.11001602560281754, + "learning_rate": 0.0001431149732620321, + "loss": 0.2258, + "step": 856 + }, + { + "epoch": 0.8572410990591104, + "grad_norm": 0.12491337209939957, + "learning_rate": 0.000143048128342246, + "loss": 0.2503, + "step": 857 + }, + { + "epoch": 0.8582413803882342, + "grad_norm": 0.11464844644069672, + "learning_rate": 0.0001429812834224599, + "loss": 0.2278, + "step": 858 + }, + { + "epoch": 0.859241661717358, + "grad_norm": 0.11179960519075394, + "learning_rate": 0.0001429144385026738, + "loss": 0.2499, + "step": 859 + }, + { + "epoch": 0.8602419430464818, + "grad_norm": 0.10024150460958481, + "learning_rate": 0.0001428475935828877, + "loss": 0.2246, + "step": 860 + }, + { + "epoch": 0.8612422243756056, + "grad_norm": 0.10340822488069534, + "learning_rate": 0.0001427807486631016, + "loss": 0.228, + "step": 861 + }, + { + "epoch": 0.8622425057047295, + "grad_norm": 0.10974396020174026, + "learning_rate": 0.00014271390374331552, + "loss": 0.2335, + "step": 862 + }, + { + "epoch": 0.8632427870338533, + "grad_norm": 0.11434578895568848, + "learning_rate": 0.0001426470588235294, + "loss": 0.2462, + "step": 863 + }, + { + "epoch": 0.864243068362977, + "grad_norm": 0.10660667717456818, + "learning_rate": 0.00014258021390374333, + "loss": 0.2446, + "step": 864 + }, + { + "epoch": 0.8652433496921009, + "grad_norm": 0.10910869389772415, + "learning_rate": 0.00014251336898395722, + "loss": 0.2304, + "step": 865 + }, + { + "epoch": 0.8662436310212247, + "grad_norm": 0.11090603470802307, + "learning_rate": 0.00014244652406417114, + "loss": 0.2243, + "step": 866 + }, + { + "epoch": 0.8672439123503486, + "grad_norm": 0.1097351461648941, + "learning_rate": 0.00014237967914438503, + "loss": 0.251, + "step": 867 + }, + { + "epoch": 0.8682441936794724, + "grad_norm": 0.11390431225299835, + "learning_rate": 0.00014231283422459892, + "loss": 0.2681, + "step": 868 + }, + { + "epoch": 0.8692444750085961, + "grad_norm": 0.1101333275437355, + "learning_rate": 0.00014224598930481284, + "loss": 0.2368, + "step": 869 + }, + { + "epoch": 0.87024475633772, + "grad_norm": 0.1092897579073906, + "learning_rate": 0.00014217914438502676, + "loss": 0.2407, + "step": 870 + }, + { + "epoch": 0.8712450376668438, + "grad_norm": 0.1118299588561058, + "learning_rate": 0.00014211229946524065, + "loss": 0.2466, + "step": 871 + }, + { + "epoch": 0.8722453189959676, + "grad_norm": 0.10369225591421127, + "learning_rate": 0.00014204545454545454, + "loss": 0.2535, + "step": 872 + }, + { + "epoch": 0.8732456003250915, + "grad_norm": 0.11793062090873718, + "learning_rate": 0.00014197860962566846, + "loss": 0.254, + "step": 873 + }, + { + "epoch": 0.8742458816542152, + "grad_norm": 0.11111178994178772, + "learning_rate": 0.00014191176470588237, + "loss": 0.237, + "step": 874 + }, + { + "epoch": 0.875246162983339, + "grad_norm": 0.10416046530008316, + "learning_rate": 0.00014184491978609627, + "loss": 0.2331, + "step": 875 + }, + { + "epoch": 0.8762464443124629, + "grad_norm": 0.11279145628213882, + "learning_rate": 0.00014177807486631016, + "loss": 0.2349, + "step": 876 + }, + { + "epoch": 0.8772467256415867, + "grad_norm": 0.10356856882572174, + "learning_rate": 0.00014171122994652405, + "loss": 0.2345, + "step": 877 + }, + { + "epoch": 0.8782470069707106, + "grad_norm": 0.10975120961666107, + "learning_rate": 0.000141644385026738, + "loss": 0.2326, + "step": 878 + }, + { + "epoch": 0.8792472882998343, + "grad_norm": 0.11057771742343903, + "learning_rate": 0.00014157754010695188, + "loss": 0.235, + "step": 879 + }, + { + "epoch": 0.8802475696289581, + "grad_norm": 0.09839839488267899, + "learning_rate": 0.00014151069518716577, + "loss": 0.2319, + "step": 880 + }, + { + "epoch": 0.881247850958082, + "grad_norm": 0.11734974384307861, + "learning_rate": 0.0001414438502673797, + "loss": 0.2357, + "step": 881 + }, + { + "epoch": 0.8822481322872058, + "grad_norm": 0.11259441822767258, + "learning_rate": 0.00014137700534759358, + "loss": 0.2417, + "step": 882 + }, + { + "epoch": 0.8832484136163296, + "grad_norm": 0.10897365212440491, + "learning_rate": 0.0001413101604278075, + "loss": 0.2337, + "step": 883 + }, + { + "epoch": 0.8842486949454534, + "grad_norm": 0.12866760790348053, + "learning_rate": 0.0001412433155080214, + "loss": 0.2313, + "step": 884 + }, + { + "epoch": 0.8852489762745772, + "grad_norm": 0.1488298773765564, + "learning_rate": 0.0001411764705882353, + "loss": 0.2531, + "step": 885 + }, + { + "epoch": 0.886249257603701, + "grad_norm": 0.11166534572839737, + "learning_rate": 0.0001411096256684492, + "loss": 0.2411, + "step": 886 + }, + { + "epoch": 0.8872495389328249, + "grad_norm": 0.10767518728971481, + "learning_rate": 0.00014104278074866312, + "loss": 0.2263, + "step": 887 + }, + { + "epoch": 0.8882498202619487, + "grad_norm": 0.12601135671138763, + "learning_rate": 0.000140975935828877, + "loss": 0.2449, + "step": 888 + }, + { + "epoch": 0.8892501015910725, + "grad_norm": 0.11647336184978485, + "learning_rate": 0.00014090909090909093, + "loss": 0.2471, + "step": 889 + }, + { + "epoch": 0.8902503829201963, + "grad_norm": 0.12583930790424347, + "learning_rate": 0.00014084224598930482, + "loss": 0.244, + "step": 890 + }, + { + "epoch": 0.8912506642493201, + "grad_norm": 0.1087995395064354, + "learning_rate": 0.0001407754010695187, + "loss": 0.2382, + "step": 891 + }, + { + "epoch": 0.892250945578444, + "grad_norm": 0.1260484755039215, + "learning_rate": 0.00014070855614973263, + "loss": 0.2408, + "step": 892 + }, + { + "epoch": 0.8932512269075678, + "grad_norm": 0.12253131717443466, + "learning_rate": 0.00014064171122994654, + "loss": 0.2307, + "step": 893 + }, + { + "epoch": 0.8942515082366915, + "grad_norm": 0.11156216263771057, + "learning_rate": 0.00014057486631016043, + "loss": 0.2428, + "step": 894 + }, + { + "epoch": 0.8952517895658154, + "grad_norm": 0.12055166810750961, + "learning_rate": 0.00014050802139037433, + "loss": 0.2481, + "step": 895 + }, + { + "epoch": 0.8962520708949392, + "grad_norm": 0.13422726094722748, + "learning_rate": 0.00014044117647058824, + "loss": 0.2547, + "step": 896 + }, + { + "epoch": 0.897252352224063, + "grad_norm": 0.11690551787614822, + "learning_rate": 0.00014037433155080216, + "loss": 0.2474, + "step": 897 + }, + { + "epoch": 0.8982526335531869, + "grad_norm": 0.10050872713327408, + "learning_rate": 0.00014030748663101605, + "loss": 0.2514, + "step": 898 + }, + { + "epoch": 0.8992529148823106, + "grad_norm": 0.1152438074350357, + "learning_rate": 0.00014024064171122994, + "loss": 0.2794, + "step": 899 + }, + { + "epoch": 0.9002531962114345, + "grad_norm": 0.11295252293348312, + "learning_rate": 0.00014017379679144386, + "loss": 0.2427, + "step": 900 + }, + { + "epoch": 0.9012534775405583, + "grad_norm": 0.11782575398683548, + "learning_rate": 0.00014010695187165778, + "loss": 0.2511, + "step": 901 + }, + { + "epoch": 0.9022537588696821, + "grad_norm": 0.10582475364208221, + "learning_rate": 0.00014004010695187167, + "loss": 0.2225, + "step": 902 + }, + { + "epoch": 0.903254040198806, + "grad_norm": 0.11037655919790268, + "learning_rate": 0.00013997326203208556, + "loss": 0.2319, + "step": 903 + }, + { + "epoch": 0.9042543215279297, + "grad_norm": 0.1153755784034729, + "learning_rate": 0.00013990641711229948, + "loss": 0.2442, + "step": 904 + }, + { + "epoch": 0.9052546028570535, + "grad_norm": 0.11509797722101212, + "learning_rate": 0.00013983957219251337, + "loss": 0.2372, + "step": 905 + }, + { + "epoch": 0.9062548841861774, + "grad_norm": 0.105307936668396, + "learning_rate": 0.0001397727272727273, + "loss": 0.2449, + "step": 906 + }, + { + "epoch": 0.9072551655153012, + "grad_norm": 0.11316991597414017, + "learning_rate": 0.00013970588235294118, + "loss": 0.2563, + "step": 907 + }, + { + "epoch": 0.9082554468444249, + "grad_norm": 0.11107899248600006, + "learning_rate": 0.0001396390374331551, + "loss": 0.2548, + "step": 908 + }, + { + "epoch": 0.9092557281735488, + "grad_norm": 0.11439424753189087, + "learning_rate": 0.000139572192513369, + "loss": 0.2508, + "step": 909 + }, + { + "epoch": 0.9102560095026726, + "grad_norm": 0.10585097223520279, + "learning_rate": 0.0001395053475935829, + "loss": 0.2317, + "step": 910 + }, + { + "epoch": 0.9112562908317965, + "grad_norm": 0.09622346609830856, + "learning_rate": 0.0001394385026737968, + "loss": 0.2215, + "step": 911 + }, + { + "epoch": 0.9122565721609203, + "grad_norm": 0.10366550087928772, + "learning_rate": 0.0001393716577540107, + "loss": 0.2368, + "step": 912 + }, + { + "epoch": 0.913256853490044, + "grad_norm": 0.12692417204380035, + "learning_rate": 0.0001393048128342246, + "loss": 0.2562, + "step": 913 + }, + { + "epoch": 0.9142571348191679, + "grad_norm": 0.1047922819852829, + "learning_rate": 0.0001392379679144385, + "loss": 0.2463, + "step": 914 + }, + { + "epoch": 0.9152574161482917, + "grad_norm": 0.1251329630613327, + "learning_rate": 0.0001391711229946524, + "loss": 0.2307, + "step": 915 + }, + { + "epoch": 0.9162576974774155, + "grad_norm": 0.10285955667495728, + "learning_rate": 0.00013910427807486633, + "loss": 0.2202, + "step": 916 + }, + { + "epoch": 0.9172579788065394, + "grad_norm": 0.1245562881231308, + "learning_rate": 0.00013903743315508022, + "loss": 0.2568, + "step": 917 + }, + { + "epoch": 0.9182582601356631, + "grad_norm": 0.11022251844406128, + "learning_rate": 0.0001389705882352941, + "loss": 0.2427, + "step": 918 + }, + { + "epoch": 0.919258541464787, + "grad_norm": 0.10444619506597519, + "learning_rate": 0.00013890374331550803, + "loss": 0.227, + "step": 919 + }, + { + "epoch": 0.9202588227939108, + "grad_norm": 0.1037181168794632, + "learning_rate": 0.00013883689839572195, + "loss": 0.2408, + "step": 920 + }, + { + "epoch": 0.9212591041230346, + "grad_norm": 0.11057087779045105, + "learning_rate": 0.00013877005347593584, + "loss": 0.2664, + "step": 921 + }, + { + "epoch": 0.9222593854521585, + "grad_norm": 0.10188641399145126, + "learning_rate": 0.00013870320855614973, + "loss": 0.2464, + "step": 922 + }, + { + "epoch": 0.9232596667812822, + "grad_norm": 0.11430062353610992, + "learning_rate": 0.00013863636363636365, + "loss": 0.2476, + "step": 923 + }, + { + "epoch": 0.924259948110406, + "grad_norm": 0.12008033692836761, + "learning_rate": 0.00013856951871657754, + "loss": 0.2332, + "step": 924 + }, + { + "epoch": 0.9252602294395299, + "grad_norm": 0.10947943478822708, + "learning_rate": 0.00013850267379679146, + "loss": 0.2466, + "step": 925 + }, + { + "epoch": 0.9262605107686537, + "grad_norm": 0.1111648678779602, + "learning_rate": 0.00013843582887700535, + "loss": 0.2408, + "step": 926 + }, + { + "epoch": 0.9272607920977775, + "grad_norm": 0.11220583319664001, + "learning_rate": 0.00013836898395721927, + "loss": 0.2312, + "step": 927 + }, + { + "epoch": 0.9282610734269013, + "grad_norm": 0.1208961084485054, + "learning_rate": 0.00013830213903743316, + "loss": 0.2385, + "step": 928 + }, + { + "epoch": 0.9292613547560251, + "grad_norm": 0.10978847742080688, + "learning_rate": 0.00013823529411764707, + "loss": 0.2351, + "step": 929 + }, + { + "epoch": 0.930261636085149, + "grad_norm": 0.10520518571138382, + "learning_rate": 0.00013816844919786096, + "loss": 0.246, + "step": 930 + }, + { + "epoch": 0.9312619174142728, + "grad_norm": 0.10968729108572006, + "learning_rate": 0.00013810160427807488, + "loss": 0.2344, + "step": 931 + }, + { + "epoch": 0.9322621987433966, + "grad_norm": 0.1055382490158081, + "learning_rate": 0.00013803475935828877, + "loss": 0.2403, + "step": 932 + }, + { + "epoch": 0.9332624800725204, + "grad_norm": 0.11266198754310608, + "learning_rate": 0.00013796791443850266, + "loss": 0.2499, + "step": 933 + }, + { + "epoch": 0.9342627614016442, + "grad_norm": 0.11343249678611755, + "learning_rate": 0.00013790106951871658, + "loss": 0.24, + "step": 934 + }, + { + "epoch": 0.935263042730768, + "grad_norm": 0.1116170585155487, + "learning_rate": 0.0001378342245989305, + "loss": 0.2434, + "step": 935 + }, + { + "epoch": 0.9362633240598919, + "grad_norm": 0.10493092238903046, + "learning_rate": 0.0001377673796791444, + "loss": 0.2417, + "step": 936 + }, + { + "epoch": 0.9372636053890157, + "grad_norm": 0.13521946966648102, + "learning_rate": 0.00013770053475935828, + "loss": 0.2315, + "step": 937 + }, + { + "epoch": 0.9382638867181394, + "grad_norm": 0.11061996221542358, + "learning_rate": 0.0001376336898395722, + "loss": 0.2622, + "step": 938 + }, + { + "epoch": 0.9392641680472633, + "grad_norm": 0.10797733813524246, + "learning_rate": 0.00013756684491978612, + "loss": 0.2395, + "step": 939 + }, + { + "epoch": 0.9402644493763871, + "grad_norm": 0.11262505501508713, + "learning_rate": 0.0001375, + "loss": 0.2468, + "step": 940 + }, + { + "epoch": 0.941264730705511, + "grad_norm": 0.12419302016496658, + "learning_rate": 0.0001374331550802139, + "loss": 0.2424, + "step": 941 + }, + { + "epoch": 0.9422650120346348, + "grad_norm": 0.11063933372497559, + "learning_rate": 0.00013736631016042782, + "loss": 0.231, + "step": 942 + }, + { + "epoch": 0.9432652933637585, + "grad_norm": 0.10237149894237518, + "learning_rate": 0.00013729946524064173, + "loss": 0.2244, + "step": 943 + }, + { + "epoch": 0.9442655746928824, + "grad_norm": 0.1112745925784111, + "learning_rate": 0.00013723262032085563, + "loss": 0.239, + "step": 944 + }, + { + "epoch": 0.9452658560220062, + "grad_norm": 0.11695201694965363, + "learning_rate": 0.00013716577540106952, + "loss": 0.2455, + "step": 945 + }, + { + "epoch": 0.94626613735113, + "grad_norm": 0.11528485268354416, + "learning_rate": 0.00013709893048128343, + "loss": 0.232, + "step": 946 + }, + { + "epoch": 0.9472664186802539, + "grad_norm": 0.10710245370864868, + "learning_rate": 0.00013703208556149733, + "loss": 0.2327, + "step": 947 + }, + { + "epoch": 0.9482667000093776, + "grad_norm": 0.12125382572412491, + "learning_rate": 0.00013696524064171124, + "loss": 0.2445, + "step": 948 + }, + { + "epoch": 0.9492669813385014, + "grad_norm": 0.11560297757387161, + "learning_rate": 0.00013689839572192513, + "loss": 0.2362, + "step": 949 + }, + { + "epoch": 0.9502672626676253, + "grad_norm": 0.11220725625753403, + "learning_rate": 0.00013683155080213905, + "loss": 0.2324, + "step": 950 + }, + { + "epoch": 0.9512675439967491, + "grad_norm": 0.11505668610334396, + "learning_rate": 0.00013676470588235294, + "loss": 0.2325, + "step": 951 + }, + { + "epoch": 0.952267825325873, + "grad_norm": 0.11229875683784485, + "learning_rate": 0.00013669786096256686, + "loss": 0.2383, + "step": 952 + }, + { + "epoch": 0.9532681066549967, + "grad_norm": 0.11714407801628113, + "learning_rate": 0.00013663101604278075, + "loss": 0.2342, + "step": 953 + }, + { + "epoch": 0.9542683879841205, + "grad_norm": 0.1235777735710144, + "learning_rate": 0.00013656417112299467, + "loss": 0.2461, + "step": 954 + }, + { + "epoch": 0.9552686693132444, + "grad_norm": 0.11029482632875443, + "learning_rate": 0.00013649732620320856, + "loss": 0.2402, + "step": 955 + }, + { + "epoch": 0.9562689506423682, + "grad_norm": 0.10886912792921066, + "learning_rate": 0.00013643048128342245, + "loss": 0.2414, + "step": 956 + }, + { + "epoch": 0.9572692319714919, + "grad_norm": 0.10697195678949356, + "learning_rate": 0.00013636363636363637, + "loss": 0.2393, + "step": 957 + }, + { + "epoch": 0.9582695133006158, + "grad_norm": 0.11168259382247925, + "learning_rate": 0.0001362967914438503, + "loss": 0.2498, + "step": 958 + }, + { + "epoch": 0.9592697946297396, + "grad_norm": 0.11057666689157486, + "learning_rate": 0.00013622994652406418, + "loss": 0.2399, + "step": 959 + }, + { + "epoch": 0.9602700759588634, + "grad_norm": 0.10389312356710434, + "learning_rate": 0.00013616310160427807, + "loss": 0.2395, + "step": 960 + }, + { + "epoch": 0.9612703572879873, + "grad_norm": 0.11256705224514008, + "learning_rate": 0.00013609625668449199, + "loss": 0.2419, + "step": 961 + }, + { + "epoch": 0.962270638617111, + "grad_norm": 0.1166057139635086, + "learning_rate": 0.0001360294117647059, + "loss": 0.254, + "step": 962 + }, + { + "epoch": 0.9632709199462349, + "grad_norm": 0.11802296340465546, + "learning_rate": 0.0001359625668449198, + "loss": 0.2451, + "step": 963 + }, + { + "epoch": 0.9642712012753587, + "grad_norm": 0.11779710650444031, + "learning_rate": 0.00013589572192513369, + "loss": 0.239, + "step": 964 + }, + { + "epoch": 0.9652714826044825, + "grad_norm": 0.0998842641711235, + "learning_rate": 0.0001358288770053476, + "loss": 0.2246, + "step": 965 + }, + { + "epoch": 0.9662717639336064, + "grad_norm": 0.14835113286972046, + "learning_rate": 0.0001357620320855615, + "loss": 0.2429, + "step": 966 + }, + { + "epoch": 0.9672720452627301, + "grad_norm": 0.11874979734420776, + "learning_rate": 0.0001356951871657754, + "loss": 0.2569, + "step": 967 + }, + { + "epoch": 0.9682723265918539, + "grad_norm": 0.1061815544962883, + "learning_rate": 0.0001356283422459893, + "loss": 0.2503, + "step": 968 + }, + { + "epoch": 0.9692726079209778, + "grad_norm": 0.10713056474924088, + "learning_rate": 0.00013556149732620322, + "loss": 0.2226, + "step": 969 + }, + { + "epoch": 0.9702728892501016, + "grad_norm": 0.10585816204547882, + "learning_rate": 0.0001354946524064171, + "loss": 0.2429, + "step": 970 + }, + { + "epoch": 0.9712731705792254, + "grad_norm": 0.11206710338592529, + "learning_rate": 0.00013542780748663103, + "loss": 0.2376, + "step": 971 + }, + { + "epoch": 0.9722734519083492, + "grad_norm": 0.10295365005731583, + "learning_rate": 0.00013536096256684492, + "loss": 0.2464, + "step": 972 + }, + { + "epoch": 0.973273733237473, + "grad_norm": 0.11962182819843292, + "learning_rate": 0.00013529411764705884, + "loss": 0.2347, + "step": 973 + }, + { + "epoch": 0.9742740145665969, + "grad_norm": 0.10361276566982269, + "learning_rate": 0.00013522727272727273, + "loss": 0.231, + "step": 974 + }, + { + "epoch": 0.9752742958957207, + "grad_norm": 0.10610037297010422, + "learning_rate": 0.00013516042780748662, + "loss": 0.2344, + "step": 975 + }, + { + "epoch": 0.9762745772248445, + "grad_norm": 0.11070679873228073, + "learning_rate": 0.00013509358288770054, + "loss": 0.2544, + "step": 976 + }, + { + "epoch": 0.9772748585539683, + "grad_norm": 0.10964801162481308, + "learning_rate": 0.00013502673796791446, + "loss": 0.2472, + "step": 977 + }, + { + "epoch": 0.9782751398830921, + "grad_norm": 0.10850351303815842, + "learning_rate": 0.00013495989304812835, + "loss": 0.2399, + "step": 978 + }, + { + "epoch": 0.9792754212122159, + "grad_norm": 0.10610129684209824, + "learning_rate": 0.00013489304812834224, + "loss": 0.234, + "step": 979 + }, + { + "epoch": 0.9802757025413398, + "grad_norm": 0.10784168541431427, + "learning_rate": 0.00013482620320855616, + "loss": 0.2307, + "step": 980 + }, + { + "epoch": 0.9812759838704636, + "grad_norm": 0.10602518916130066, + "learning_rate": 0.00013475935828877007, + "loss": 0.2468, + "step": 981 + }, + { + "epoch": 0.9822762651995874, + "grad_norm": 0.113640196621418, + "learning_rate": 0.00013469251336898396, + "loss": 0.2486, + "step": 982 + }, + { + "epoch": 0.9832765465287112, + "grad_norm": 0.13207989931106567, + "learning_rate": 0.00013462566844919786, + "loss": 0.2517, + "step": 983 + }, + { + "epoch": 0.984276827857835, + "grad_norm": 0.10458721965551376, + "learning_rate": 0.00013455882352941177, + "loss": 0.2271, + "step": 984 + }, + { + "epoch": 0.9852771091869589, + "grad_norm": 0.10744502395391464, + "learning_rate": 0.0001344919786096257, + "loss": 0.2475, + "step": 985 + }, + { + "epoch": 0.9862773905160827, + "grad_norm": 0.12058056145906448, + "learning_rate": 0.00013442513368983958, + "loss": 0.2279, + "step": 986 + }, + { + "epoch": 0.9872776718452064, + "grad_norm": 0.10551847517490387, + "learning_rate": 0.00013435828877005347, + "loss": 0.2327, + "step": 987 + }, + { + "epoch": 0.9882779531743303, + "grad_norm": 0.1183042824268341, + "learning_rate": 0.0001342914438502674, + "loss": 0.2381, + "step": 988 + }, + { + "epoch": 0.9892782345034541, + "grad_norm": 0.1135777160525322, + "learning_rate": 0.00013422459893048128, + "loss": 0.2411, + "step": 989 + }, + { + "epoch": 0.9902785158325779, + "grad_norm": 0.1156490296125412, + "learning_rate": 0.0001341577540106952, + "loss": 0.2366, + "step": 990 + }, + { + "epoch": 0.9912787971617018, + "grad_norm": 0.11141631007194519, + "learning_rate": 0.0001340909090909091, + "loss": 0.2386, + "step": 991 + }, + { + "epoch": 0.9922790784908255, + "grad_norm": 0.10216771066188812, + "learning_rate": 0.000134024064171123, + "loss": 0.2304, + "step": 992 + }, + { + "epoch": 0.9932793598199494, + "grad_norm": 0.10122135281562805, + "learning_rate": 0.0001339572192513369, + "loss": 0.2475, + "step": 993 + }, + { + "epoch": 0.9942796411490732, + "grad_norm": 0.11433159559965134, + "learning_rate": 0.00013389037433155082, + "loss": 0.2445, + "step": 994 + }, + { + "epoch": 0.995279922478197, + "grad_norm": 0.12163514643907547, + "learning_rate": 0.0001338235294117647, + "loss": 0.2329, + "step": 995 + }, + { + "epoch": 0.9962802038073209, + "grad_norm": 0.10961396992206573, + "learning_rate": 0.00013375668449197863, + "loss": 0.2267, + "step": 996 + }, + { + "epoch": 0.9972804851364446, + "grad_norm": 0.10671857744455338, + "learning_rate": 0.00013368983957219252, + "loss": 0.2337, + "step": 997 + }, + { + "epoch": 0.9982807664655684, + "grad_norm": 0.11740228533744812, + "learning_rate": 0.0001336229946524064, + "loss": 0.2404, + "step": 998 + }, + { + "epoch": 0.9992810477946923, + "grad_norm": 0.10227134823799133, + "learning_rate": 0.00013355614973262033, + "loss": 0.2473, + "step": 999 + }, + { + "epoch": 1.0002813291238162, + "grad_norm": 0.10737957805395126, + "learning_rate": 0.00013348930481283424, + "loss": 0.2404, + "step": 1000 + }, + { + "epoch": 1.0012816104529398, + "grad_norm": 0.10383178293704987, + "learning_rate": 0.00013342245989304813, + "loss": 0.2193, + "step": 1001 + }, + { + "epoch": 1.0022818917820637, + "grad_norm": 0.09554237872362137, + "learning_rate": 0.00013335561497326202, + "loss": 0.2278, + "step": 1002 + }, + { + "epoch": 1.0032821731111876, + "grad_norm": 0.10792408883571625, + "learning_rate": 0.00013328877005347594, + "loss": 0.2314, + "step": 1003 + }, + { + "epoch": 1.0042824544403113, + "grad_norm": 0.1073218584060669, + "learning_rate": 0.00013322192513368986, + "loss": 0.23, + "step": 1004 + }, + { + "epoch": 1.0052827357694352, + "grad_norm": 0.09901339560747147, + "learning_rate": 0.00013315508021390375, + "loss": 0.2091, + "step": 1005 + }, + { + "epoch": 1.006283017098559, + "grad_norm": 0.11102905869483948, + "learning_rate": 0.00013308823529411764, + "loss": 0.2341, + "step": 1006 + }, + { + "epoch": 1.0072832984276827, + "grad_norm": 0.09842706471681595, + "learning_rate": 0.00013302139037433156, + "loss": 0.2048, + "step": 1007 + }, + { + "epoch": 1.0082835797568066, + "grad_norm": 0.10446655750274658, + "learning_rate": 0.00013295454545454548, + "loss": 0.2081, + "step": 1008 + }, + { + "epoch": 1.0092838610859305, + "grad_norm": 0.10733439773321152, + "learning_rate": 0.00013288770053475937, + "loss": 0.223, + "step": 1009 + }, + { + "epoch": 1.0102841424150542, + "grad_norm": 0.1143801286816597, + "learning_rate": 0.00013282085561497326, + "loss": 0.2202, + "step": 1010 + }, + { + "epoch": 1.011284423744178, + "grad_norm": 0.1163199171423912, + "learning_rate": 0.00013275401069518718, + "loss": 0.2345, + "step": 1011 + }, + { + "epoch": 1.012284705073302, + "grad_norm": 0.10841790586709976, + "learning_rate": 0.00013268716577540107, + "loss": 0.2143, + "step": 1012 + }, + { + "epoch": 1.0132849864024256, + "grad_norm": 0.10988374799489975, + "learning_rate": 0.00013262032085561499, + "loss": 0.2145, + "step": 1013 + }, + { + "epoch": 1.0142852677315495, + "grad_norm": 0.10899233818054199, + "learning_rate": 0.00013255347593582888, + "loss": 0.2201, + "step": 1014 + }, + { + "epoch": 1.0152855490606734, + "grad_norm": 0.10377402603626251, + "learning_rate": 0.0001324866310160428, + "loss": 0.2144, + "step": 1015 + }, + { + "epoch": 1.016285830389797, + "grad_norm": 0.10859181731939316, + "learning_rate": 0.00013241978609625669, + "loss": 0.2257, + "step": 1016 + }, + { + "epoch": 1.017286111718921, + "grad_norm": 0.1121574267745018, + "learning_rate": 0.0001323529411764706, + "loss": 0.2103, + "step": 1017 + }, + { + "epoch": 1.0182863930480448, + "grad_norm": 0.09887727349996567, + "learning_rate": 0.0001322860962566845, + "loss": 0.2144, + "step": 1018 + }, + { + "epoch": 1.0192866743771685, + "grad_norm": 0.1135513111948967, + "learning_rate": 0.0001322192513368984, + "loss": 0.2312, + "step": 1019 + }, + { + "epoch": 1.0202869557062924, + "grad_norm": 0.10303981602191925, + "learning_rate": 0.0001321524064171123, + "loss": 0.2299, + "step": 1020 + }, + { + "epoch": 1.0212872370354162, + "grad_norm": 0.09777767956256866, + "learning_rate": 0.0001320855614973262, + "loss": 0.2203, + "step": 1021 + }, + { + "epoch": 1.0222875183645401, + "grad_norm": 0.09875063598155975, + "learning_rate": 0.0001320187165775401, + "loss": 0.2078, + "step": 1022 + }, + { + "epoch": 1.0232877996936638, + "grad_norm": 0.10464679449796677, + "learning_rate": 0.00013195187165775403, + "loss": 0.2313, + "step": 1023 + }, + { + "epoch": 1.0242880810227877, + "grad_norm": 0.10746071487665176, + "learning_rate": 0.00013188502673796792, + "loss": 0.2236, + "step": 1024 + }, + { + "epoch": 1.0252883623519116, + "grad_norm": 0.09986191242933273, + "learning_rate": 0.0001318181818181818, + "loss": 0.2186, + "step": 1025 + }, + { + "epoch": 1.0262886436810352, + "grad_norm": 0.10748348385095596, + "learning_rate": 0.00013175133689839573, + "loss": 0.1989, + "step": 1026 + }, + { + "epoch": 1.0272889250101591, + "grad_norm": 0.10878057032823563, + "learning_rate": 0.00013168449197860965, + "loss": 0.2318, + "step": 1027 + }, + { + "epoch": 1.028289206339283, + "grad_norm": 0.11410006880760193, + "learning_rate": 0.00013161764705882354, + "loss": 0.2239, + "step": 1028 + }, + { + "epoch": 1.0292894876684067, + "grad_norm": 0.11048425734043121, + "learning_rate": 0.00013155080213903743, + "loss": 0.236, + "step": 1029 + }, + { + "epoch": 1.0302897689975306, + "grad_norm": 0.11481732875108719, + "learning_rate": 0.00013148395721925135, + "loss": 0.2178, + "step": 1030 + }, + { + "epoch": 1.0312900503266544, + "grad_norm": 0.1073327511548996, + "learning_rate": 0.00013141711229946524, + "loss": 0.2196, + "step": 1031 + }, + { + "epoch": 1.032290331655778, + "grad_norm": 0.10710236430168152, + "learning_rate": 0.00013135026737967916, + "loss": 0.228, + "step": 1032 + }, + { + "epoch": 1.033290612984902, + "grad_norm": 0.10308988392353058, + "learning_rate": 0.00013128342245989305, + "loss": 0.2092, + "step": 1033 + }, + { + "epoch": 1.0342908943140259, + "grad_norm": 0.11342065036296844, + "learning_rate": 0.00013121657754010696, + "loss": 0.2422, + "step": 1034 + }, + { + "epoch": 1.0352911756431495, + "grad_norm": 0.10920470952987671, + "learning_rate": 0.00013114973262032086, + "loss": 0.2068, + "step": 1035 + }, + { + "epoch": 1.0362914569722734, + "grad_norm": 0.11763305217027664, + "learning_rate": 0.00013108288770053477, + "loss": 0.2341, + "step": 1036 + }, + { + "epoch": 1.0372917383013973, + "grad_norm": 0.11884487420320511, + "learning_rate": 0.00013101604278074866, + "loss": 0.2266, + "step": 1037 + }, + { + "epoch": 1.038292019630521, + "grad_norm": 0.1029181182384491, + "learning_rate": 0.00013094919786096258, + "loss": 0.224, + "step": 1038 + }, + { + "epoch": 1.0392923009596449, + "grad_norm": 0.10172492265701294, + "learning_rate": 0.00013088235294117647, + "loss": 0.2182, + "step": 1039 + }, + { + "epoch": 1.0402925822887688, + "grad_norm": 0.11021002382040024, + "learning_rate": 0.00013081550802139036, + "loss": 0.2219, + "step": 1040 + }, + { + "epoch": 1.0412928636178926, + "grad_norm": 0.10181029886007309, + "learning_rate": 0.00013074866310160428, + "loss": 0.2235, + "step": 1041 + }, + { + "epoch": 1.0422931449470163, + "grad_norm": 0.11245550960302353, + "learning_rate": 0.0001306818181818182, + "loss": 0.2064, + "step": 1042 + }, + { + "epoch": 1.0432934262761402, + "grad_norm": 0.10707695037126541, + "learning_rate": 0.0001306149732620321, + "loss": 0.2151, + "step": 1043 + }, + { + "epoch": 1.044293707605264, + "grad_norm": 0.10446685552597046, + "learning_rate": 0.00013054812834224598, + "loss": 0.2043, + "step": 1044 + }, + { + "epoch": 1.0452939889343877, + "grad_norm": 0.1158304512500763, + "learning_rate": 0.0001304812834224599, + "loss": 0.2215, + "step": 1045 + }, + { + "epoch": 1.0462942702635116, + "grad_norm": 0.10438673943281174, + "learning_rate": 0.00013041443850267382, + "loss": 0.2196, + "step": 1046 + }, + { + "epoch": 1.0472945515926355, + "grad_norm": 0.1166171059012413, + "learning_rate": 0.0001303475935828877, + "loss": 0.2087, + "step": 1047 + }, + { + "epoch": 1.0482948329217592, + "grad_norm": 0.1188088208436966, + "learning_rate": 0.0001302807486631016, + "loss": 0.2177, + "step": 1048 + }, + { + "epoch": 1.049295114250883, + "grad_norm": 0.10798342525959015, + "learning_rate": 0.00013021390374331552, + "loss": 0.2219, + "step": 1049 + }, + { + "epoch": 1.050295395580007, + "grad_norm": 0.11776283383369446, + "learning_rate": 0.00013014705882352943, + "loss": 0.216, + "step": 1050 + }, + { + "epoch": 1.0512956769091306, + "grad_norm": 0.11055237799882889, + "learning_rate": 0.00013008021390374333, + "loss": 0.2161, + "step": 1051 + }, + { + "epoch": 1.0522959582382545, + "grad_norm": 0.11248302459716797, + "learning_rate": 0.00013001336898395722, + "loss": 0.2116, + "step": 1052 + }, + { + "epoch": 1.0532962395673784, + "grad_norm": 0.10515984892845154, + "learning_rate": 0.00012994652406417113, + "loss": 0.2104, + "step": 1053 + }, + { + "epoch": 1.054296520896502, + "grad_norm": 0.13536454737186432, + "learning_rate": 0.00012987967914438502, + "loss": 0.2329, + "step": 1054 + }, + { + "epoch": 1.055296802225626, + "grad_norm": 0.12209105491638184, + "learning_rate": 0.00012981283422459894, + "loss": 0.2208, + "step": 1055 + }, + { + "epoch": 1.0562970835547498, + "grad_norm": 0.09903733432292938, + "learning_rate": 0.00012974598930481283, + "loss": 0.2051, + "step": 1056 + }, + { + "epoch": 1.0572973648838735, + "grad_norm": 0.12912599742412567, + "learning_rate": 0.00012967914438502675, + "loss": 0.2211, + "step": 1057 + }, + { + "epoch": 1.0582976462129974, + "grad_norm": 0.1307183802127838, + "learning_rate": 0.00012961229946524064, + "loss": 0.2291, + "step": 1058 + }, + { + "epoch": 1.0592979275421213, + "grad_norm": 0.1040973886847496, + "learning_rate": 0.00012954545454545456, + "loss": 0.2067, + "step": 1059 + }, + { + "epoch": 1.0602982088712452, + "grad_norm": 0.11457894742488861, + "learning_rate": 0.00012947860962566845, + "loss": 0.2138, + "step": 1060 + }, + { + "epoch": 1.0612984902003688, + "grad_norm": 0.1062072291970253, + "learning_rate": 0.00012941176470588237, + "loss": 0.2164, + "step": 1061 + }, + { + "epoch": 1.0622987715294927, + "grad_norm": 0.10179958492517471, + "learning_rate": 0.00012934491978609626, + "loss": 0.2162, + "step": 1062 + }, + { + "epoch": 1.0632990528586166, + "grad_norm": 0.10438190400600433, + "learning_rate": 0.00012927807486631015, + "loss": 0.2197, + "step": 1063 + }, + { + "epoch": 1.0642993341877403, + "grad_norm": 0.10387393832206726, + "learning_rate": 0.00012921122994652407, + "loss": 0.2185, + "step": 1064 + }, + { + "epoch": 1.0652996155168641, + "grad_norm": 0.10165292024612427, + "learning_rate": 0.00012914438502673799, + "loss": 0.2057, + "step": 1065 + }, + { + "epoch": 1.066299896845988, + "grad_norm": 0.11831758916378021, + "learning_rate": 0.00012907754010695188, + "loss": 0.2258, + "step": 1066 + }, + { + "epoch": 1.0673001781751117, + "grad_norm": 0.11722327023744583, + "learning_rate": 0.00012901069518716577, + "loss": 0.2289, + "step": 1067 + }, + { + "epoch": 1.0683004595042356, + "grad_norm": 0.1045883446931839, + "learning_rate": 0.00012894385026737969, + "loss": 0.2124, + "step": 1068 + }, + { + "epoch": 1.0693007408333595, + "grad_norm": 0.09917068481445312, + "learning_rate": 0.0001288770053475936, + "loss": 0.214, + "step": 1069 + }, + { + "epoch": 1.0703010221624831, + "grad_norm": 0.11233076453208923, + "learning_rate": 0.0001288101604278075, + "loss": 0.2201, + "step": 1070 + }, + { + "epoch": 1.071301303491607, + "grad_norm": 0.1170390248298645, + "learning_rate": 0.00012874331550802139, + "loss": 0.216, + "step": 1071 + }, + { + "epoch": 1.072301584820731, + "grad_norm": 0.1092529222369194, + "learning_rate": 0.0001286764705882353, + "loss": 0.2187, + "step": 1072 + }, + { + "epoch": 1.0733018661498546, + "grad_norm": 0.10764916986227036, + "learning_rate": 0.0001286096256684492, + "loss": 0.1993, + "step": 1073 + }, + { + "epoch": 1.0743021474789785, + "grad_norm": 0.11166828125715256, + "learning_rate": 0.0001285427807486631, + "loss": 0.2159, + "step": 1074 + }, + { + "epoch": 1.0753024288081023, + "grad_norm": 0.110441192984581, + "learning_rate": 0.000128475935828877, + "loss": 0.2097, + "step": 1075 + }, + { + "epoch": 1.076302710137226, + "grad_norm": 0.10855505615472794, + "learning_rate": 0.00012840909090909092, + "loss": 0.2286, + "step": 1076 + }, + { + "epoch": 1.07730299146635, + "grad_norm": 0.11431234329938889, + "learning_rate": 0.0001283422459893048, + "loss": 0.221, + "step": 1077 + }, + { + "epoch": 1.0783032727954738, + "grad_norm": 0.10840080678462982, + "learning_rate": 0.00012827540106951873, + "loss": 0.2102, + "step": 1078 + }, + { + "epoch": 1.0793035541245974, + "grad_norm": 0.10828956961631775, + "learning_rate": 0.00012820855614973262, + "loss": 0.2105, + "step": 1079 + }, + { + "epoch": 1.0803038354537213, + "grad_norm": 0.10958611965179443, + "learning_rate": 0.00012814171122994654, + "loss": 0.2311, + "step": 1080 + }, + { + "epoch": 1.0813041167828452, + "grad_norm": 0.10427571088075638, + "learning_rate": 0.00012807486631016043, + "loss": 0.2134, + "step": 1081 + }, + { + "epoch": 1.0823043981119689, + "grad_norm": 0.10889621078968048, + "learning_rate": 0.00012800802139037432, + "loss": 0.2278, + "step": 1082 + }, + { + "epoch": 1.0833046794410928, + "grad_norm": 0.10879098623991013, + "learning_rate": 0.00012794117647058824, + "loss": 0.2093, + "step": 1083 + }, + { + "epoch": 1.0843049607702167, + "grad_norm": 0.10152193158864975, + "learning_rate": 0.00012787433155080216, + "loss": 0.2056, + "step": 1084 + }, + { + "epoch": 1.0853052420993405, + "grad_norm": 0.1010005995631218, + "learning_rate": 0.00012780748663101605, + "loss": 0.2124, + "step": 1085 + }, + { + "epoch": 1.0863055234284642, + "grad_norm": 0.10704009979963303, + "learning_rate": 0.00012774064171122994, + "loss": 0.207, + "step": 1086 + }, + { + "epoch": 1.087305804757588, + "grad_norm": 0.1215653344988823, + "learning_rate": 0.00012767379679144386, + "loss": 0.229, + "step": 1087 + }, + { + "epoch": 1.088306086086712, + "grad_norm": 0.1207938939332962, + "learning_rate": 0.00012760695187165777, + "loss": 0.205, + "step": 1088 + }, + { + "epoch": 1.0893063674158356, + "grad_norm": 0.10882195830345154, + "learning_rate": 0.00012754010695187166, + "loss": 0.2442, + "step": 1089 + }, + { + "epoch": 1.0903066487449595, + "grad_norm": 0.104824960231781, + "learning_rate": 0.00012747326203208555, + "loss": 0.2174, + "step": 1090 + }, + { + "epoch": 1.0913069300740834, + "grad_norm": 0.12660905718803406, + "learning_rate": 0.00012740641711229947, + "loss": 0.218, + "step": 1091 + }, + { + "epoch": 1.092307211403207, + "grad_norm": 0.12053773552179337, + "learning_rate": 0.0001273395721925134, + "loss": 0.2277, + "step": 1092 + }, + { + "epoch": 1.093307492732331, + "grad_norm": 0.1109408438205719, + "learning_rate": 0.00012727272727272728, + "loss": 0.2318, + "step": 1093 + }, + { + "epoch": 1.0943077740614549, + "grad_norm": 0.10368873178958893, + "learning_rate": 0.00012720588235294117, + "loss": 0.2103, + "step": 1094 + }, + { + "epoch": 1.0953080553905785, + "grad_norm": 0.10493618249893188, + "learning_rate": 0.0001271390374331551, + "loss": 0.2214, + "step": 1095 + }, + { + "epoch": 1.0963083367197024, + "grad_norm": 0.10940803587436676, + "learning_rate": 0.00012707219251336898, + "loss": 0.2192, + "step": 1096 + }, + { + "epoch": 1.0973086180488263, + "grad_norm": 0.10427404195070267, + "learning_rate": 0.0001270053475935829, + "loss": 0.2141, + "step": 1097 + }, + { + "epoch": 1.09830889937795, + "grad_norm": 0.10865108668804169, + "learning_rate": 0.0001269385026737968, + "loss": 0.2164, + "step": 1098 + }, + { + "epoch": 1.0993091807070738, + "grad_norm": 0.10540167987346649, + "learning_rate": 0.0001268716577540107, + "loss": 0.2067, + "step": 1099 + }, + { + "epoch": 1.1003094620361977, + "grad_norm": 0.1088499203324318, + "learning_rate": 0.0001268048128342246, + "loss": 0.2275, + "step": 1100 + }, + { + "epoch": 1.1013097433653214, + "grad_norm": 0.11432886868715286, + "learning_rate": 0.00012673796791443852, + "loss": 0.2263, + "step": 1101 + }, + { + "epoch": 1.1023100246944453, + "grad_norm": 0.10245134681463242, + "learning_rate": 0.0001266711229946524, + "loss": 0.2102, + "step": 1102 + }, + { + "epoch": 1.1033103060235692, + "grad_norm": 0.11726287007331848, + "learning_rate": 0.00012660427807486632, + "loss": 0.2065, + "step": 1103 + }, + { + "epoch": 1.104310587352693, + "grad_norm": 0.10728625953197479, + "learning_rate": 0.00012653743315508022, + "loss": 0.2219, + "step": 1104 + }, + { + "epoch": 1.1053108686818167, + "grad_norm": 0.11919640004634857, + "learning_rate": 0.0001264705882352941, + "loss": 0.2326, + "step": 1105 + }, + { + "epoch": 1.1063111500109406, + "grad_norm": 0.10696747899055481, + "learning_rate": 0.00012640374331550802, + "loss": 0.2297, + "step": 1106 + }, + { + "epoch": 1.1073114313400645, + "grad_norm": 0.1080828309059143, + "learning_rate": 0.00012633689839572194, + "loss": 0.2258, + "step": 1107 + }, + { + "epoch": 1.1083117126691882, + "grad_norm": 0.1082526221871376, + "learning_rate": 0.00012627005347593583, + "loss": 0.2286, + "step": 1108 + }, + { + "epoch": 1.109311993998312, + "grad_norm": 0.11087387800216675, + "learning_rate": 0.00012620320855614972, + "loss": 0.2165, + "step": 1109 + }, + { + "epoch": 1.110312275327436, + "grad_norm": 0.10974619537591934, + "learning_rate": 0.00012613636363636364, + "loss": 0.2325, + "step": 1110 + }, + { + "epoch": 1.1113125566565596, + "grad_norm": 0.11098938435316086, + "learning_rate": 0.00012606951871657756, + "loss": 0.2221, + "step": 1111 + }, + { + "epoch": 1.1123128379856835, + "grad_norm": 0.11327999085187912, + "learning_rate": 0.00012600267379679145, + "loss": 0.2236, + "step": 1112 + }, + { + "epoch": 1.1133131193148074, + "grad_norm": 0.10366078466176987, + "learning_rate": 0.00012593582887700534, + "loss": 0.2116, + "step": 1113 + }, + { + "epoch": 1.114313400643931, + "grad_norm": 0.1121227964758873, + "learning_rate": 0.00012586898395721926, + "loss": 0.2128, + "step": 1114 + }, + { + "epoch": 1.115313681973055, + "grad_norm": 0.11292851716279984, + "learning_rate": 0.00012580213903743318, + "loss": 0.2265, + "step": 1115 + }, + { + "epoch": 1.1163139633021788, + "grad_norm": 0.11685734987258911, + "learning_rate": 0.00012573529411764707, + "loss": 0.2203, + "step": 1116 + }, + { + "epoch": 1.1173142446313025, + "grad_norm": 0.11855416744947433, + "learning_rate": 0.00012566844919786096, + "loss": 0.2276, + "step": 1117 + }, + { + "epoch": 1.1183145259604264, + "grad_norm": 0.10524473339319229, + "learning_rate": 0.00012560160427807488, + "loss": 0.2164, + "step": 1118 + }, + { + "epoch": 1.1193148072895502, + "grad_norm": 0.10147272795438766, + "learning_rate": 0.00012553475935828877, + "loss": 0.2175, + "step": 1119 + }, + { + "epoch": 1.120315088618674, + "grad_norm": 0.10443752259016037, + "learning_rate": 0.00012546791443850269, + "loss": 0.2142, + "step": 1120 + }, + { + "epoch": 1.1213153699477978, + "grad_norm": 0.09947992861270905, + "learning_rate": 0.00012540106951871658, + "loss": 0.2209, + "step": 1121 + }, + { + "epoch": 1.1223156512769217, + "grad_norm": 0.11207697540521622, + "learning_rate": 0.0001253342245989305, + "loss": 0.2355, + "step": 1122 + }, + { + "epoch": 1.1233159326060456, + "grad_norm": 0.11477234959602356, + "learning_rate": 0.00012526737967914439, + "loss": 0.229, + "step": 1123 + }, + { + "epoch": 1.1243162139351692, + "grad_norm": 0.11826887726783752, + "learning_rate": 0.0001252005347593583, + "loss": 0.2284, + "step": 1124 + }, + { + "epoch": 1.1253164952642931, + "grad_norm": 0.10183574259281158, + "learning_rate": 0.0001251336898395722, + "loss": 0.2053, + "step": 1125 + }, + { + "epoch": 1.1263167765934168, + "grad_norm": 0.1009439155459404, + "learning_rate": 0.0001250668449197861, + "loss": 0.2224, + "step": 1126 + }, + { + "epoch": 1.1273170579225407, + "grad_norm": 0.10415584594011307, + "learning_rate": 0.000125, + "loss": 0.2281, + "step": 1127 + }, + { + "epoch": 1.1283173392516646, + "grad_norm": 0.11737174540758133, + "learning_rate": 0.0001249331550802139, + "loss": 0.2218, + "step": 1128 + }, + { + "epoch": 1.1293176205807884, + "grad_norm": 0.1162266656756401, + "learning_rate": 0.0001248663101604278, + "loss": 0.2508, + "step": 1129 + }, + { + "epoch": 1.130317901909912, + "grad_norm": 0.11246795207262039, + "learning_rate": 0.00012479946524064173, + "loss": 0.2175, + "step": 1130 + }, + { + "epoch": 1.131318183239036, + "grad_norm": 0.09997264295816422, + "learning_rate": 0.00012473262032085562, + "loss": 0.2147, + "step": 1131 + }, + { + "epoch": 1.1323184645681599, + "grad_norm": 0.10060079395771027, + "learning_rate": 0.0001246657754010695, + "loss": 0.2195, + "step": 1132 + }, + { + "epoch": 1.1333187458972835, + "grad_norm": 0.11128344386816025, + "learning_rate": 0.00012459893048128343, + "loss": 0.2192, + "step": 1133 + }, + { + "epoch": 1.1343190272264074, + "grad_norm": 0.11528977751731873, + "learning_rate": 0.00012453208556149735, + "loss": 0.2195, + "step": 1134 + }, + { + "epoch": 1.1353193085555313, + "grad_norm": 0.10216296464204788, + "learning_rate": 0.00012446524064171124, + "loss": 0.2136, + "step": 1135 + }, + { + "epoch": 1.136319589884655, + "grad_norm": 0.10256506502628326, + "learning_rate": 0.00012439839572192513, + "loss": 0.2137, + "step": 1136 + }, + { + "epoch": 1.1373198712137789, + "grad_norm": 0.10524075478315353, + "learning_rate": 0.00012433155080213905, + "loss": 0.2049, + "step": 1137 + }, + { + "epoch": 1.1383201525429028, + "grad_norm": 0.11876089870929718, + "learning_rate": 0.00012426470588235294, + "loss": 0.2296, + "step": 1138 + }, + { + "epoch": 1.1393204338720264, + "grad_norm": 0.12065361440181732, + "learning_rate": 0.00012419786096256685, + "loss": 0.2292, + "step": 1139 + }, + { + "epoch": 1.1403207152011503, + "grad_norm": 0.10976608097553253, + "learning_rate": 0.00012413101604278075, + "loss": 0.2217, + "step": 1140 + }, + { + "epoch": 1.1413209965302742, + "grad_norm": 0.11010421812534332, + "learning_rate": 0.00012406417112299466, + "loss": 0.2258, + "step": 1141 + }, + { + "epoch": 1.142321277859398, + "grad_norm": 0.12316484749317169, + "learning_rate": 0.00012399732620320855, + "loss": 0.2374, + "step": 1142 + }, + { + "epoch": 1.1433215591885217, + "grad_norm": 0.10903903096914291, + "learning_rate": 0.00012393048128342247, + "loss": 0.2178, + "step": 1143 + }, + { + "epoch": 1.1443218405176456, + "grad_norm": 0.11023163050413132, + "learning_rate": 0.00012386363636363636, + "loss": 0.2168, + "step": 1144 + }, + { + "epoch": 1.1453221218467693, + "grad_norm": 0.10460875183343887, + "learning_rate": 0.00012379679144385028, + "loss": 0.2082, + "step": 1145 + }, + { + "epoch": 1.1463224031758932, + "grad_norm": 0.1269470900297165, + "learning_rate": 0.00012372994652406417, + "loss": 0.2279, + "step": 1146 + }, + { + "epoch": 1.147322684505017, + "grad_norm": 0.09975128620862961, + "learning_rate": 0.00012366310160427806, + "loss": 0.2087, + "step": 1147 + }, + { + "epoch": 1.148322965834141, + "grad_norm": 0.10157779604196548, + "learning_rate": 0.000123596256684492, + "loss": 0.2105, + "step": 1148 + }, + { + "epoch": 1.1493232471632646, + "grad_norm": 0.11378750950098038, + "learning_rate": 0.0001235294117647059, + "loss": 0.2189, + "step": 1149 + }, + { + "epoch": 1.1503235284923885, + "grad_norm": 0.10354934632778168, + "learning_rate": 0.0001234625668449198, + "loss": 0.2088, + "step": 1150 + }, + { + "epoch": 1.1513238098215124, + "grad_norm": 0.10327152907848358, + "learning_rate": 0.00012339572192513368, + "loss": 0.2032, + "step": 1151 + }, + { + "epoch": 1.152324091150636, + "grad_norm": 0.10712700337171555, + "learning_rate": 0.0001233288770053476, + "loss": 0.2139, + "step": 1152 + }, + { + "epoch": 1.15332437247976, + "grad_norm": 0.10584112256765366, + "learning_rate": 0.00012326203208556152, + "loss": 0.209, + "step": 1153 + }, + { + "epoch": 1.1543246538088838, + "grad_norm": 0.10985728353261948, + "learning_rate": 0.0001231951871657754, + "loss": 0.2228, + "step": 1154 + }, + { + "epoch": 1.1553249351380075, + "grad_norm": 0.11061149090528488, + "learning_rate": 0.0001231283422459893, + "loss": 0.2149, + "step": 1155 + }, + { + "epoch": 1.1563252164671314, + "grad_norm": 0.11156737059354782, + "learning_rate": 0.00012306149732620322, + "loss": 0.2236, + "step": 1156 + }, + { + "epoch": 1.1573254977962553, + "grad_norm": 0.10013081878423691, + "learning_rate": 0.00012299465240641713, + "loss": 0.1968, + "step": 1157 + }, + { + "epoch": 1.158325779125379, + "grad_norm": 0.10728764533996582, + "learning_rate": 0.00012292780748663102, + "loss": 0.2053, + "step": 1158 + }, + { + "epoch": 1.1593260604545028, + "grad_norm": 0.10990236699581146, + "learning_rate": 0.00012286096256684492, + "loss": 0.2347, + "step": 1159 + }, + { + "epoch": 1.1603263417836267, + "grad_norm": 0.11535795032978058, + "learning_rate": 0.00012279411764705883, + "loss": 0.2188, + "step": 1160 + }, + { + "epoch": 1.1613266231127504, + "grad_norm": 0.11049818247556686, + "learning_rate": 0.00012272727272727272, + "loss": 0.2344, + "step": 1161 + }, + { + "epoch": 1.1623269044418743, + "grad_norm": 0.10996738076210022, + "learning_rate": 0.00012266042780748664, + "loss": 0.2139, + "step": 1162 + }, + { + "epoch": 1.1633271857709981, + "grad_norm": 0.11199158430099487, + "learning_rate": 0.00012259358288770053, + "loss": 0.2371, + "step": 1163 + }, + { + "epoch": 1.1643274671001218, + "grad_norm": 0.10963901877403259, + "learning_rate": 0.00012252673796791445, + "loss": 0.2184, + "step": 1164 + }, + { + "epoch": 1.1653277484292457, + "grad_norm": 0.11525794863700867, + "learning_rate": 0.00012245989304812834, + "loss": 0.2107, + "step": 1165 + }, + { + "epoch": 1.1663280297583696, + "grad_norm": 0.12791381776332855, + "learning_rate": 0.00012239304812834226, + "loss": 0.2287, + "step": 1166 + }, + { + "epoch": 1.1673283110874935, + "grad_norm": 0.10105018317699432, + "learning_rate": 0.00012232620320855615, + "loss": 0.2181, + "step": 1167 + }, + { + "epoch": 1.1683285924166171, + "grad_norm": 0.10252846777439117, + "learning_rate": 0.00012225935828877007, + "loss": 0.1962, + "step": 1168 + }, + { + "epoch": 1.169328873745741, + "grad_norm": 0.1163405030965805, + "learning_rate": 0.00012219251336898396, + "loss": 0.2134, + "step": 1169 + }, + { + "epoch": 1.170329155074865, + "grad_norm": 0.12065989524126053, + "learning_rate": 0.00012212566844919785, + "loss": 0.2285, + "step": 1170 + }, + { + "epoch": 1.1713294364039886, + "grad_norm": 0.11344461143016815, + "learning_rate": 0.00012205882352941178, + "loss": 0.2221, + "step": 1171 + }, + { + "epoch": 1.1723297177331125, + "grad_norm": 0.12093393504619598, + "learning_rate": 0.00012199197860962569, + "loss": 0.234, + "step": 1172 + }, + { + "epoch": 1.1733299990622363, + "grad_norm": 0.10900023579597473, + "learning_rate": 0.00012192513368983958, + "loss": 0.2156, + "step": 1173 + }, + { + "epoch": 1.17433028039136, + "grad_norm": 0.11394291371107101, + "learning_rate": 0.00012185828877005348, + "loss": 0.2187, + "step": 1174 + }, + { + "epoch": 1.175330561720484, + "grad_norm": 0.11865662783384323, + "learning_rate": 0.0001217914438502674, + "loss": 0.2275, + "step": 1175 + }, + { + "epoch": 1.1763308430496078, + "grad_norm": 0.10707909613847733, + "learning_rate": 0.00012172459893048129, + "loss": 0.2125, + "step": 1176 + }, + { + "epoch": 1.1773311243787314, + "grad_norm": 0.11197365075349808, + "learning_rate": 0.0001216577540106952, + "loss": 0.2273, + "step": 1177 + }, + { + "epoch": 1.1783314057078553, + "grad_norm": 0.10113795846700668, + "learning_rate": 0.00012159090909090908, + "loss": 0.2058, + "step": 1178 + }, + { + "epoch": 1.1793316870369792, + "grad_norm": 0.11658436805009842, + "learning_rate": 0.000121524064171123, + "loss": 0.2097, + "step": 1179 + }, + { + "epoch": 1.1803319683661029, + "grad_norm": 0.12685951590538025, + "learning_rate": 0.00012145721925133691, + "loss": 0.2271, + "step": 1180 + }, + { + "epoch": 1.1813322496952268, + "grad_norm": 0.10609004646539688, + "learning_rate": 0.00012139037433155081, + "loss": 0.2049, + "step": 1181 + }, + { + "epoch": 1.1823325310243507, + "grad_norm": 0.10939063876867294, + "learning_rate": 0.0001213235294117647, + "loss": 0.2298, + "step": 1182 + }, + { + "epoch": 1.1833328123534743, + "grad_norm": 0.10683566331863403, + "learning_rate": 0.00012125668449197862, + "loss": 0.2058, + "step": 1183 + }, + { + "epoch": 1.1843330936825982, + "grad_norm": 0.10511665046215057, + "learning_rate": 0.00012118983957219252, + "loss": 0.2098, + "step": 1184 + }, + { + "epoch": 1.185333375011722, + "grad_norm": 0.11079413443803787, + "learning_rate": 0.00012112299465240642, + "loss": 0.2221, + "step": 1185 + }, + { + "epoch": 1.186333656340846, + "grad_norm": 0.10557086765766144, + "learning_rate": 0.00012105614973262032, + "loss": 0.2133, + "step": 1186 + }, + { + "epoch": 1.1873339376699696, + "grad_norm": 0.10528892278671265, + "learning_rate": 0.00012098930481283424, + "loss": 0.2224, + "step": 1187 + }, + { + "epoch": 1.1883342189990935, + "grad_norm": 0.1264183223247528, + "learning_rate": 0.00012092245989304813, + "loss": 0.2281, + "step": 1188 + }, + { + "epoch": 1.1893345003282172, + "grad_norm": 0.10961788147687912, + "learning_rate": 0.00012085561497326203, + "loss": 0.2115, + "step": 1189 + }, + { + "epoch": 1.190334781657341, + "grad_norm": 0.10892324149608612, + "learning_rate": 0.00012078877005347595, + "loss": 0.2205, + "step": 1190 + }, + { + "epoch": 1.191335062986465, + "grad_norm": 0.10153544694185257, + "learning_rate": 0.00012072192513368985, + "loss": 0.2073, + "step": 1191 + }, + { + "epoch": 1.1923353443155889, + "grad_norm": 0.11882033944129944, + "learning_rate": 0.00012065508021390375, + "loss": 0.2163, + "step": 1192 + }, + { + "epoch": 1.1933356256447125, + "grad_norm": 0.11150426417589188, + "learning_rate": 0.00012058823529411765, + "loss": 0.2184, + "step": 1193 + }, + { + "epoch": 1.1943359069738364, + "grad_norm": 0.10831573605537415, + "learning_rate": 0.00012052139037433157, + "loss": 0.2094, + "step": 1194 + }, + { + "epoch": 1.1953361883029603, + "grad_norm": 0.10810606926679611, + "learning_rate": 0.00012045454545454546, + "loss": 0.2251, + "step": 1195 + }, + { + "epoch": 1.196336469632084, + "grad_norm": 0.09927061945199966, + "learning_rate": 0.00012038770053475936, + "loss": 0.2066, + "step": 1196 + }, + { + "epoch": 1.1973367509612078, + "grad_norm": 0.11117838323116302, + "learning_rate": 0.00012032085561497325, + "loss": 0.2198, + "step": 1197 + }, + { + "epoch": 1.1983370322903317, + "grad_norm": 0.10765928775072098, + "learning_rate": 0.00012025401069518719, + "loss": 0.2161, + "step": 1198 + }, + { + "epoch": 1.1993373136194554, + "grad_norm": 0.1030687466263771, + "learning_rate": 0.00012018716577540108, + "loss": 0.2026, + "step": 1199 + }, + { + "epoch": 1.2003375949485793, + "grad_norm": 0.11261389404535294, + "learning_rate": 0.00012012032085561498, + "loss": 0.2153, + "step": 1200 + }, + { + "epoch": 1.2013378762777032, + "grad_norm": 0.11929168552160263, + "learning_rate": 0.00012005347593582887, + "loss": 0.2251, + "step": 1201 + }, + { + "epoch": 1.2023381576068268, + "grad_norm": 0.11565817147493362, + "learning_rate": 0.00011998663101604279, + "loss": 0.2236, + "step": 1202 + }, + { + "epoch": 1.2033384389359507, + "grad_norm": 0.1021752879023552, + "learning_rate": 0.0001199197860962567, + "loss": 0.207, + "step": 1203 + }, + { + "epoch": 1.2043387202650746, + "grad_norm": 0.11508005857467651, + "learning_rate": 0.00011985294117647058, + "loss": 0.2195, + "step": 1204 + }, + { + "epoch": 1.2053390015941985, + "grad_norm": 0.12688252329826355, + "learning_rate": 0.00011978609625668449, + "loss": 0.2226, + "step": 1205 + }, + { + "epoch": 1.2063392829233222, + "grad_norm": 0.11833857744932175, + "learning_rate": 0.00011971925133689841, + "loss": 0.2246, + "step": 1206 + }, + { + "epoch": 1.207339564252446, + "grad_norm": 0.10420876741409302, + "learning_rate": 0.00011965240641711231, + "loss": 0.2123, + "step": 1207 + }, + { + "epoch": 1.2083398455815697, + "grad_norm": 0.11918733268976212, + "learning_rate": 0.0001195855614973262, + "loss": 0.2275, + "step": 1208 + }, + { + "epoch": 1.2093401269106936, + "grad_norm": 0.11753445118665695, + "learning_rate": 0.0001195187165775401, + "loss": 0.23, + "step": 1209 + }, + { + "epoch": 1.2103404082398175, + "grad_norm": 0.11419044435024261, + "learning_rate": 0.00011945187165775402, + "loss": 0.219, + "step": 1210 + }, + { + "epoch": 1.2113406895689414, + "grad_norm": 0.10295890271663666, + "learning_rate": 0.00011938502673796792, + "loss": 0.211, + "step": 1211 + }, + { + "epoch": 1.212340970898065, + "grad_norm": 0.10617538541555405, + "learning_rate": 0.00011931818181818182, + "loss": 0.2238, + "step": 1212 + }, + { + "epoch": 1.213341252227189, + "grad_norm": 0.11215183138847351, + "learning_rate": 0.00011925133689839574, + "loss": 0.2343, + "step": 1213 + }, + { + "epoch": 1.2143415335563128, + "grad_norm": 0.1216965913772583, + "learning_rate": 0.00011918449197860964, + "loss": 0.243, + "step": 1214 + }, + { + "epoch": 1.2153418148854365, + "grad_norm": 0.11013886332511902, + "learning_rate": 0.00011911764705882353, + "loss": 0.2188, + "step": 1215 + }, + { + "epoch": 1.2163420962145604, + "grad_norm": 0.09862317144870758, + "learning_rate": 0.00011905080213903744, + "loss": 0.2235, + "step": 1216 + }, + { + "epoch": 1.2173423775436842, + "grad_norm": 0.10906413942575455, + "learning_rate": 0.00011898395721925135, + "loss": 0.2215, + "step": 1217 + }, + { + "epoch": 1.218342658872808, + "grad_norm": 0.12304907292127609, + "learning_rate": 0.00011891711229946525, + "loss": 0.2116, + "step": 1218 + }, + { + "epoch": 1.2193429402019318, + "grad_norm": 0.10474281013011932, + "learning_rate": 0.00011885026737967915, + "loss": 0.2181, + "step": 1219 + }, + { + "epoch": 1.2203432215310557, + "grad_norm": 0.10716108977794647, + "learning_rate": 0.00011878342245989304, + "loss": 0.2168, + "step": 1220 + }, + { + "epoch": 1.2213435028601793, + "grad_norm": 0.1210319995880127, + "learning_rate": 0.00011871657754010697, + "loss": 0.2245, + "step": 1221 + }, + { + "epoch": 1.2223437841893032, + "grad_norm": 0.10937487334012985, + "learning_rate": 0.00011864973262032086, + "loss": 0.2197, + "step": 1222 + }, + { + "epoch": 1.2233440655184271, + "grad_norm": 0.1190948337316513, + "learning_rate": 0.00011858288770053477, + "loss": 0.2202, + "step": 1223 + }, + { + "epoch": 1.2243443468475508, + "grad_norm": 0.10808120667934418, + "learning_rate": 0.00011851604278074866, + "loss": 0.2279, + "step": 1224 + }, + { + "epoch": 1.2253446281766747, + "grad_norm": 0.11375033855438232, + "learning_rate": 0.00011844919786096258, + "loss": 0.2017, + "step": 1225 + }, + { + "epoch": 1.2263449095057986, + "grad_norm": 0.12218756228685379, + "learning_rate": 0.00011838235294117648, + "loss": 0.2217, + "step": 1226 + }, + { + "epoch": 1.2273451908349222, + "grad_norm": 0.10410333424806595, + "learning_rate": 0.00011831550802139037, + "loss": 0.2159, + "step": 1227 + }, + { + "epoch": 1.228345472164046, + "grad_norm": 0.11102598905563354, + "learning_rate": 0.00011824866310160428, + "loss": 0.2316, + "step": 1228 + }, + { + "epoch": 1.22934575349317, + "grad_norm": 0.10408420860767365, + "learning_rate": 0.0001181818181818182, + "loss": 0.2103, + "step": 1229 + }, + { + "epoch": 1.2303460348222939, + "grad_norm": 0.1024235188961029, + "learning_rate": 0.0001181149732620321, + "loss": 0.2037, + "step": 1230 + }, + { + "epoch": 1.2313463161514175, + "grad_norm": 0.10356253385543823, + "learning_rate": 0.00011804812834224599, + "loss": 0.2068, + "step": 1231 + }, + { + "epoch": 1.2323465974805414, + "grad_norm": 0.10571043193340302, + "learning_rate": 0.0001179812834224599, + "loss": 0.2193, + "step": 1232 + }, + { + "epoch": 1.2333468788096653, + "grad_norm": 0.10934340208768845, + "learning_rate": 0.00011791443850267381, + "loss": 0.2341, + "step": 1233 + }, + { + "epoch": 1.234347160138789, + "grad_norm": 0.1156490221619606, + "learning_rate": 0.0001178475935828877, + "loss": 0.2251, + "step": 1234 + }, + { + "epoch": 1.2353474414679129, + "grad_norm": 0.10883632302284241, + "learning_rate": 0.0001177807486631016, + "loss": 0.2282, + "step": 1235 + }, + { + "epoch": 1.2363477227970368, + "grad_norm": 0.10380252450704575, + "learning_rate": 0.00011771390374331552, + "loss": 0.2148, + "step": 1236 + }, + { + "epoch": 1.2373480041261604, + "grad_norm": 0.11076192557811737, + "learning_rate": 0.00011764705882352942, + "loss": 0.2257, + "step": 1237 + }, + { + "epoch": 1.2383482854552843, + "grad_norm": 0.10647089034318924, + "learning_rate": 0.00011758021390374332, + "loss": 0.2068, + "step": 1238 + }, + { + "epoch": 1.2393485667844082, + "grad_norm": 0.10665105283260345, + "learning_rate": 0.00011751336898395722, + "loss": 0.2205, + "step": 1239 + }, + { + "epoch": 1.2403488481135319, + "grad_norm": 0.10603989660739899, + "learning_rate": 0.00011744652406417114, + "loss": 0.218, + "step": 1240 + }, + { + "epoch": 1.2413491294426557, + "grad_norm": 0.12585783004760742, + "learning_rate": 0.00011737967914438503, + "loss": 0.2266, + "step": 1241 + }, + { + "epoch": 1.2423494107717796, + "grad_norm": 0.11196950078010559, + "learning_rate": 0.00011731283422459894, + "loss": 0.2179, + "step": 1242 + }, + { + "epoch": 1.2433496921009033, + "grad_norm": 0.12112867087125778, + "learning_rate": 0.00011724598930481283, + "loss": 0.209, + "step": 1243 + }, + { + "epoch": 1.2443499734300272, + "grad_norm": 0.11002309620380402, + "learning_rate": 0.00011717914438502675, + "loss": 0.2308, + "step": 1244 + }, + { + "epoch": 1.245350254759151, + "grad_norm": 0.11513447761535645, + "learning_rate": 0.00011711229946524065, + "loss": 0.2238, + "step": 1245 + }, + { + "epoch": 1.2463505360882747, + "grad_norm": 0.11998051404953003, + "learning_rate": 0.00011704545454545454, + "loss": 0.2255, + "step": 1246 + }, + { + "epoch": 1.2473508174173986, + "grad_norm": 0.11045163124799728, + "learning_rate": 0.00011697860962566845, + "loss": 0.2227, + "step": 1247 + }, + { + "epoch": 1.2483510987465225, + "grad_norm": 0.10709457099437714, + "learning_rate": 0.00011691176470588236, + "loss": 0.2147, + "step": 1248 + }, + { + "epoch": 1.2493513800756464, + "grad_norm": 0.10837852209806442, + "learning_rate": 0.00011684491978609627, + "loss": 0.2231, + "step": 1249 + }, + { + "epoch": 1.25035166140477, + "grad_norm": 0.10169129818677902, + "learning_rate": 0.00011677807486631016, + "loss": 0.1996, + "step": 1250 + }, + { + "epoch": 1.251351942733894, + "grad_norm": 0.09892980009317398, + "learning_rate": 0.00011671122994652406, + "loss": 0.2101, + "step": 1251 + }, + { + "epoch": 1.2523522240630176, + "grad_norm": 0.11100195348262787, + "learning_rate": 0.00011664438502673798, + "loss": 0.2229, + "step": 1252 + }, + { + "epoch": 1.2533525053921415, + "grad_norm": 0.10280987620353699, + "learning_rate": 0.00011657754010695187, + "loss": 0.2131, + "step": 1253 + }, + { + "epoch": 1.2543527867212654, + "grad_norm": 0.10166744887828827, + "learning_rate": 0.00011651069518716578, + "loss": 0.2104, + "step": 1254 + }, + { + "epoch": 1.2553530680503893, + "grad_norm": 0.10268554836511612, + "learning_rate": 0.0001164438502673797, + "loss": 0.2133, + "step": 1255 + }, + { + "epoch": 1.256353349379513, + "grad_norm": 0.10194896906614304, + "learning_rate": 0.0001163770053475936, + "loss": 0.2024, + "step": 1256 + }, + { + "epoch": 1.2573536307086368, + "grad_norm": 0.10676533728837967, + "learning_rate": 0.00011631016042780749, + "loss": 0.2083, + "step": 1257 + }, + { + "epoch": 1.2583539120377607, + "grad_norm": 0.10382092744112015, + "learning_rate": 0.00011624331550802139, + "loss": 0.1977, + "step": 1258 + }, + { + "epoch": 1.2593541933668844, + "grad_norm": 0.11634174734354019, + "learning_rate": 0.00011617647058823531, + "loss": 0.2175, + "step": 1259 + }, + { + "epoch": 1.2603544746960083, + "grad_norm": 0.11611123383045197, + "learning_rate": 0.0001161096256684492, + "loss": 0.2264, + "step": 1260 + }, + { + "epoch": 1.2613547560251321, + "grad_norm": 0.1250474900007248, + "learning_rate": 0.0001160427807486631, + "loss": 0.2099, + "step": 1261 + }, + { + "epoch": 1.2623550373542558, + "grad_norm": 0.12509945034980774, + "learning_rate": 0.000115975935828877, + "loss": 0.2247, + "step": 1262 + }, + { + "epoch": 1.2633553186833797, + "grad_norm": 0.11208090931177139, + "learning_rate": 0.00011590909090909093, + "loss": 0.2161, + "step": 1263 + }, + { + "epoch": 1.2643556000125036, + "grad_norm": 0.10543353855609894, + "learning_rate": 0.00011584224598930482, + "loss": 0.1996, + "step": 1264 + }, + { + "epoch": 1.2653558813416272, + "grad_norm": 0.1154838353395462, + "learning_rate": 0.00011577540106951872, + "loss": 0.2301, + "step": 1265 + }, + { + "epoch": 1.2663561626707511, + "grad_norm": 0.11428222805261612, + "learning_rate": 0.00011570855614973261, + "loss": 0.2231, + "step": 1266 + }, + { + "epoch": 1.267356443999875, + "grad_norm": 0.10909832268953323, + "learning_rate": 0.00011564171122994653, + "loss": 0.213, + "step": 1267 + }, + { + "epoch": 1.268356725328999, + "grad_norm": 0.11378253996372223, + "learning_rate": 0.00011557486631016044, + "loss": 0.2134, + "step": 1268 + }, + { + "epoch": 1.2693570066581226, + "grad_norm": 0.1124582514166832, + "learning_rate": 0.00011550802139037433, + "loss": 0.2187, + "step": 1269 + }, + { + "epoch": 1.2703572879872465, + "grad_norm": 0.10859386622905731, + "learning_rate": 0.00011544117647058823, + "loss": 0.2238, + "step": 1270 + }, + { + "epoch": 1.2713575693163701, + "grad_norm": 0.12885946035385132, + "learning_rate": 0.00011537433155080215, + "loss": 0.251, + "step": 1271 + }, + { + "epoch": 1.272357850645494, + "grad_norm": 0.1093195453286171, + "learning_rate": 0.00011530748663101605, + "loss": 0.2164, + "step": 1272 + }, + { + "epoch": 1.273358131974618, + "grad_norm": 0.11165391653776169, + "learning_rate": 0.00011524064171122995, + "loss": 0.23, + "step": 1273 + }, + { + "epoch": 1.2743584133037418, + "grad_norm": 0.1040864959359169, + "learning_rate": 0.00011517379679144386, + "loss": 0.2246, + "step": 1274 + }, + { + "epoch": 1.2753586946328654, + "grad_norm": 0.1049712672829628, + "learning_rate": 0.00011510695187165777, + "loss": 0.2137, + "step": 1275 + }, + { + "epoch": 1.2763589759619893, + "grad_norm": 0.11247608065605164, + "learning_rate": 0.00011504010695187166, + "loss": 0.2177, + "step": 1276 + }, + { + "epoch": 1.277359257291113, + "grad_norm": 0.11736306548118591, + "learning_rate": 0.00011497326203208556, + "loss": 0.2174, + "step": 1277 + }, + { + "epoch": 1.2783595386202369, + "grad_norm": 0.1128850132226944, + "learning_rate": 0.00011490641711229948, + "loss": 0.2157, + "step": 1278 + }, + { + "epoch": 1.2793598199493608, + "grad_norm": 0.10518353432416916, + "learning_rate": 0.00011483957219251338, + "loss": 0.2216, + "step": 1279 + }, + { + "epoch": 1.2803601012784847, + "grad_norm": 0.10513953119516373, + "learning_rate": 0.00011477272727272728, + "loss": 0.2121, + "step": 1280 + }, + { + "epoch": 1.2813603826076083, + "grad_norm": 0.12127160280942917, + "learning_rate": 0.00011470588235294118, + "loss": 0.2192, + "step": 1281 + }, + { + "epoch": 1.2823606639367322, + "grad_norm": 0.11220681667327881, + "learning_rate": 0.0001146390374331551, + "loss": 0.2137, + "step": 1282 + }, + { + "epoch": 1.283360945265856, + "grad_norm": 0.10587983578443527, + "learning_rate": 0.00011457219251336899, + "loss": 0.2093, + "step": 1283 + }, + { + "epoch": 1.2843612265949798, + "grad_norm": 0.10703995078802109, + "learning_rate": 0.00011450534759358289, + "loss": 0.2059, + "step": 1284 + }, + { + "epoch": 1.2853615079241036, + "grad_norm": 0.11333701014518738, + "learning_rate": 0.00011443850267379678, + "loss": 0.2237, + "step": 1285 + }, + { + "epoch": 1.2863617892532275, + "grad_norm": 0.10058709233999252, + "learning_rate": 0.0001143716577540107, + "loss": 0.2157, + "step": 1286 + }, + { + "epoch": 1.2873620705823514, + "grad_norm": 0.10317469388246536, + "learning_rate": 0.0001143048128342246, + "loss": 0.206, + "step": 1287 + }, + { + "epoch": 1.288362351911475, + "grad_norm": 0.10771293193101883, + "learning_rate": 0.00011423796791443851, + "loss": 0.2215, + "step": 1288 + }, + { + "epoch": 1.289362633240599, + "grad_norm": 0.10581102967262268, + "learning_rate": 0.0001141711229946524, + "loss": 0.2177, + "step": 1289 + }, + { + "epoch": 1.2903629145697226, + "grad_norm": 0.11833846569061279, + "learning_rate": 0.00011410427807486632, + "loss": 0.2151, + "step": 1290 + }, + { + "epoch": 1.2913631958988465, + "grad_norm": 0.10840478539466858, + "learning_rate": 0.00011403743315508022, + "loss": 0.2124, + "step": 1291 + }, + { + "epoch": 1.2923634772279704, + "grad_norm": 0.10112584382295609, + "learning_rate": 0.00011397058823529411, + "loss": 0.2084, + "step": 1292 + }, + { + "epoch": 1.2933637585570943, + "grad_norm": 0.11672486364841461, + "learning_rate": 0.00011390374331550802, + "loss": 0.2288, + "step": 1293 + }, + { + "epoch": 1.294364039886218, + "grad_norm": 0.11796103417873383, + "learning_rate": 0.00011383689839572194, + "loss": 0.2188, + "step": 1294 + }, + { + "epoch": 1.2953643212153418, + "grad_norm": 0.10772399604320526, + "learning_rate": 0.00011377005347593583, + "loss": 0.2148, + "step": 1295 + }, + { + "epoch": 1.2963646025444655, + "grad_norm": 0.1071363016963005, + "learning_rate": 0.00011370320855614973, + "loss": 0.2127, + "step": 1296 + }, + { + "epoch": 1.2973648838735894, + "grad_norm": 0.10482805222272873, + "learning_rate": 0.00011363636363636365, + "loss": 0.2306, + "step": 1297 + }, + { + "epoch": 1.2983651652027133, + "grad_norm": 0.11823964864015579, + "learning_rate": 0.00011356951871657755, + "loss": 0.2168, + "step": 1298 + }, + { + "epoch": 1.2993654465318372, + "grad_norm": 0.10563730448484421, + "learning_rate": 0.00011350267379679144, + "loss": 0.209, + "step": 1299 + }, + { + "epoch": 1.3003657278609608, + "grad_norm": 0.10991596430540085, + "learning_rate": 0.00011343582887700535, + "loss": 0.2199, + "step": 1300 + }, + { + "epoch": 1.3013660091900847, + "grad_norm": 0.10185655206441879, + "learning_rate": 0.00011336898395721927, + "loss": 0.2214, + "step": 1301 + }, + { + "epoch": 1.3023662905192086, + "grad_norm": 0.11055384576320648, + "learning_rate": 0.00011330213903743316, + "loss": 0.211, + "step": 1302 + }, + { + "epoch": 1.3033665718483323, + "grad_norm": 0.11013782769441605, + "learning_rate": 0.00011323529411764706, + "loss": 0.2124, + "step": 1303 + }, + { + "epoch": 1.3043668531774562, + "grad_norm": 0.11421462893486023, + "learning_rate": 0.00011316844919786095, + "loss": 0.2141, + "step": 1304 + }, + { + "epoch": 1.30536713450658, + "grad_norm": 0.11334969103336334, + "learning_rate": 0.00011310160427807488, + "loss": 0.2272, + "step": 1305 + }, + { + "epoch": 1.306367415835704, + "grad_norm": 0.10643415153026581, + "learning_rate": 0.00011303475935828878, + "loss": 0.2318, + "step": 1306 + }, + { + "epoch": 1.3073676971648276, + "grad_norm": 0.11799966543912888, + "learning_rate": 0.00011296791443850268, + "loss": 0.2198, + "step": 1307 + }, + { + "epoch": 1.3083679784939515, + "grad_norm": 0.10618843883275986, + "learning_rate": 0.00011290106951871657, + "loss": 0.2103, + "step": 1308 + }, + { + "epoch": 1.3093682598230751, + "grad_norm": 0.0983828529715538, + "learning_rate": 0.00011283422459893049, + "loss": 0.195, + "step": 1309 + }, + { + "epoch": 1.310368541152199, + "grad_norm": 0.10963460057973862, + "learning_rate": 0.00011276737967914439, + "loss": 0.2203, + "step": 1310 + }, + { + "epoch": 1.311368822481323, + "grad_norm": 0.11698295921087265, + "learning_rate": 0.00011270053475935828, + "loss": 0.2195, + "step": 1311 + }, + { + "epoch": 1.3123691038104468, + "grad_norm": 0.10051514953374863, + "learning_rate": 0.00011263368983957219, + "loss": 0.2097, + "step": 1312 + }, + { + "epoch": 1.3133693851395705, + "grad_norm": 0.1085805594921112, + "learning_rate": 0.0001125668449197861, + "loss": 0.2141, + "step": 1313 + }, + { + "epoch": 1.3143696664686944, + "grad_norm": 0.10205250978469849, + "learning_rate": 0.00011250000000000001, + "loss": 0.2111, + "step": 1314 + }, + { + "epoch": 1.315369947797818, + "grad_norm": 0.10300610214471817, + "learning_rate": 0.0001124331550802139, + "loss": 0.2066, + "step": 1315 + }, + { + "epoch": 1.316370229126942, + "grad_norm": 0.10185340791940689, + "learning_rate": 0.00011236631016042782, + "loss": 0.218, + "step": 1316 + }, + { + "epoch": 1.3173705104560658, + "grad_norm": 0.11197194457054138, + "learning_rate": 0.00011229946524064172, + "loss": 0.2226, + "step": 1317 + }, + { + "epoch": 1.3183707917851897, + "grad_norm": 0.11062564700841904, + "learning_rate": 0.00011223262032085561, + "loss": 0.2161, + "step": 1318 + }, + { + "epoch": 1.3193710731143133, + "grad_norm": 0.10071063786745071, + "learning_rate": 0.00011216577540106952, + "loss": 0.2086, + "step": 1319 + }, + { + "epoch": 1.3203713544434372, + "grad_norm": 0.11609186232089996, + "learning_rate": 0.00011209893048128344, + "loss": 0.2393, + "step": 1320 + }, + { + "epoch": 1.3213716357725611, + "grad_norm": 0.10754919052124023, + "learning_rate": 0.00011203208556149734, + "loss": 0.2154, + "step": 1321 + }, + { + "epoch": 1.3223719171016848, + "grad_norm": 0.11132339388132095, + "learning_rate": 0.00011196524064171123, + "loss": 0.2045, + "step": 1322 + }, + { + "epoch": 1.3233721984308087, + "grad_norm": 0.11251602321863174, + "learning_rate": 0.00011189839572192514, + "loss": 0.2259, + "step": 1323 + }, + { + "epoch": 1.3243724797599326, + "grad_norm": 0.11466008424758911, + "learning_rate": 0.00011183155080213905, + "loss": 0.2086, + "step": 1324 + }, + { + "epoch": 1.3253727610890562, + "grad_norm": 0.11656272411346436, + "learning_rate": 0.00011176470588235294, + "loss": 0.2197, + "step": 1325 + }, + { + "epoch": 1.32637304241818, + "grad_norm": 0.11661850661039352, + "learning_rate": 0.00011169786096256685, + "loss": 0.2126, + "step": 1326 + }, + { + "epoch": 1.327373323747304, + "grad_norm": 0.10672106593847275, + "learning_rate": 0.00011163101604278074, + "loss": 0.2123, + "step": 1327 + }, + { + "epoch": 1.3283736050764277, + "grad_norm": 0.10479258000850677, + "learning_rate": 0.00011156417112299467, + "loss": 0.2264, + "step": 1328 + }, + { + "epoch": 1.3293738864055515, + "grad_norm": 0.11438891291618347, + "learning_rate": 0.00011149732620320856, + "loss": 0.2261, + "step": 1329 + }, + { + "epoch": 1.3303741677346754, + "grad_norm": 0.10022715479135513, + "learning_rate": 0.00011143048128342247, + "loss": 0.2103, + "step": 1330 + }, + { + "epoch": 1.3313744490637993, + "grad_norm": 0.10805850476026535, + "learning_rate": 0.00011136363636363636, + "loss": 0.2066, + "step": 1331 + }, + { + "epoch": 1.332374730392923, + "grad_norm": 0.1080065593123436, + "learning_rate": 0.00011129679144385028, + "loss": 0.2234, + "step": 1332 + }, + { + "epoch": 1.3333750117220469, + "grad_norm": 0.11281702667474747, + "learning_rate": 0.00011122994652406418, + "loss": 0.2241, + "step": 1333 + }, + { + "epoch": 1.3343752930511705, + "grad_norm": 0.11385636776685715, + "learning_rate": 0.00011116310160427807, + "loss": 0.2184, + "step": 1334 + }, + { + "epoch": 1.3353755743802944, + "grad_norm": 0.1062302216887474, + "learning_rate": 0.00011109625668449199, + "loss": 0.2149, + "step": 1335 + }, + { + "epoch": 1.3363758557094183, + "grad_norm": 0.11073213815689087, + "learning_rate": 0.00011102941176470589, + "loss": 0.2212, + "step": 1336 + }, + { + "epoch": 1.3373761370385422, + "grad_norm": 0.1101662665605545, + "learning_rate": 0.0001109625668449198, + "loss": 0.2187, + "step": 1337 + }, + { + "epoch": 1.3383764183676659, + "grad_norm": 0.11318746954202652, + "learning_rate": 0.00011089572192513369, + "loss": 0.2308, + "step": 1338 + }, + { + "epoch": 1.3393766996967897, + "grad_norm": 0.10213375091552734, + "learning_rate": 0.0001108288770053476, + "loss": 0.2193, + "step": 1339 + }, + { + "epoch": 1.3403769810259134, + "grad_norm": 0.1051267758011818, + "learning_rate": 0.00011076203208556151, + "loss": 0.2207, + "step": 1340 + }, + { + "epoch": 1.3413772623550373, + "grad_norm": 0.11033023148775101, + "learning_rate": 0.0001106951871657754, + "loss": 0.2176, + "step": 1341 + }, + { + "epoch": 1.3423775436841612, + "grad_norm": 0.10777070373296738, + "learning_rate": 0.0001106283422459893, + "loss": 0.2358, + "step": 1342 + }, + { + "epoch": 1.343377825013285, + "grad_norm": 0.11919771134853363, + "learning_rate": 0.00011056149732620322, + "loss": 0.2177, + "step": 1343 + }, + { + "epoch": 1.3443781063424087, + "grad_norm": 0.11171173304319382, + "learning_rate": 0.00011049465240641711, + "loss": 0.2302, + "step": 1344 + }, + { + "epoch": 1.3453783876715326, + "grad_norm": 0.11045338958501816, + "learning_rate": 0.00011042780748663102, + "loss": 0.2285, + "step": 1345 + }, + { + "epoch": 1.3463786690006565, + "grad_norm": 0.10334324091672897, + "learning_rate": 0.00011036096256684492, + "loss": 0.2116, + "step": 1346 + }, + { + "epoch": 1.3473789503297802, + "grad_norm": 0.10447406768798828, + "learning_rate": 0.00011029411764705884, + "loss": 0.2123, + "step": 1347 + }, + { + "epoch": 1.348379231658904, + "grad_norm": 0.11587905883789062, + "learning_rate": 0.00011022727272727273, + "loss": 0.215, + "step": 1348 + }, + { + "epoch": 1.349379512988028, + "grad_norm": 0.10127585381269455, + "learning_rate": 0.00011016042780748664, + "loss": 0.221, + "step": 1349 + }, + { + "epoch": 1.3503797943171518, + "grad_norm": 0.10887004435062408, + "learning_rate": 0.00011009358288770053, + "loss": 0.2148, + "step": 1350 + }, + { + "epoch": 1.3513800756462755, + "grad_norm": 0.12198647111654282, + "learning_rate": 0.00011002673796791444, + "loss": 0.2283, + "step": 1351 + }, + { + "epoch": 1.3523803569753994, + "grad_norm": 0.11372344940900803, + "learning_rate": 0.00010995989304812835, + "loss": 0.2094, + "step": 1352 + }, + { + "epoch": 1.353380638304523, + "grad_norm": 0.10559667646884918, + "learning_rate": 0.00010989304812834224, + "loss": 0.2181, + "step": 1353 + }, + { + "epoch": 1.354380919633647, + "grad_norm": 0.1059967428445816, + "learning_rate": 0.00010982620320855614, + "loss": 0.2169, + "step": 1354 + }, + { + "epoch": 1.3553812009627708, + "grad_norm": 0.09842636436223984, + "learning_rate": 0.00010975935828877006, + "loss": 0.2121, + "step": 1355 + }, + { + "epoch": 1.3563814822918947, + "grad_norm": 0.09859322011470795, + "learning_rate": 0.00010969251336898397, + "loss": 0.2123, + "step": 1356 + }, + { + "epoch": 1.3573817636210184, + "grad_norm": 0.10696738213300705, + "learning_rate": 0.00010962566844919786, + "loss": 0.2096, + "step": 1357 + }, + { + "epoch": 1.3583820449501423, + "grad_norm": 0.11054214090108871, + "learning_rate": 0.00010955882352941178, + "loss": 0.2082, + "step": 1358 + }, + { + "epoch": 1.359382326279266, + "grad_norm": 0.1111525371670723, + "learning_rate": 0.00010949197860962568, + "loss": 0.2182, + "step": 1359 + }, + { + "epoch": 1.3603826076083898, + "grad_norm": 0.10844685882329941, + "learning_rate": 0.00010942513368983957, + "loss": 0.2171, + "step": 1360 + }, + { + "epoch": 1.3613828889375137, + "grad_norm": 0.1152367815375328, + "learning_rate": 0.00010935828877005347, + "loss": 0.2235, + "step": 1361 + }, + { + "epoch": 1.3623831702666376, + "grad_norm": 0.11865795403718948, + "learning_rate": 0.00010929144385026739, + "loss": 0.2095, + "step": 1362 + }, + { + "epoch": 1.3633834515957612, + "grad_norm": 0.10809239745140076, + "learning_rate": 0.0001092245989304813, + "loss": 0.2029, + "step": 1363 + }, + { + "epoch": 1.3643837329248851, + "grad_norm": 0.10157923400402069, + "learning_rate": 0.00010915775401069519, + "loss": 0.2061, + "step": 1364 + }, + { + "epoch": 1.365384014254009, + "grad_norm": 0.11636494845151901, + "learning_rate": 0.00010909090909090909, + "loss": 0.2091, + "step": 1365 + }, + { + "epoch": 1.3663842955831327, + "grad_norm": 0.11290279775857925, + "learning_rate": 0.00010902406417112301, + "loss": 0.222, + "step": 1366 + }, + { + "epoch": 1.3673845769122566, + "grad_norm": 0.11001241207122803, + "learning_rate": 0.0001089572192513369, + "loss": 0.2174, + "step": 1367 + }, + { + "epoch": 1.3683848582413805, + "grad_norm": 0.11491668224334717, + "learning_rate": 0.0001088903743315508, + "loss": 0.2318, + "step": 1368 + }, + { + "epoch": 1.3693851395705043, + "grad_norm": 0.11041668802499771, + "learning_rate": 0.0001088235294117647, + "loss": 0.2097, + "step": 1369 + }, + { + "epoch": 1.370385420899628, + "grad_norm": 0.11703222990036011, + "learning_rate": 0.00010875668449197863, + "loss": 0.2184, + "step": 1370 + }, + { + "epoch": 1.371385702228752, + "grad_norm": 0.11403464525938034, + "learning_rate": 0.00010868983957219252, + "loss": 0.2176, + "step": 1371 + }, + { + "epoch": 1.3723859835578756, + "grad_norm": 0.10866453498601913, + "learning_rate": 0.00010862299465240642, + "loss": 0.2201, + "step": 1372 + }, + { + "epoch": 1.3733862648869994, + "grad_norm": 0.1143585741519928, + "learning_rate": 0.00010855614973262031, + "loss": 0.2249, + "step": 1373 + }, + { + "epoch": 1.3743865462161233, + "grad_norm": 0.10900405049324036, + "learning_rate": 0.00010848930481283423, + "loss": 0.2298, + "step": 1374 + }, + { + "epoch": 1.3753868275452472, + "grad_norm": 0.10825719684362411, + "learning_rate": 0.00010842245989304814, + "loss": 0.2163, + "step": 1375 + }, + { + "epoch": 1.3763871088743709, + "grad_norm": 0.1157190203666687, + "learning_rate": 0.00010835561497326203, + "loss": 0.2335, + "step": 1376 + }, + { + "epoch": 1.3773873902034948, + "grad_norm": 0.10434065759181976, + "learning_rate": 0.00010828877005347596, + "loss": 0.2143, + "step": 1377 + }, + { + "epoch": 1.3783876715326184, + "grad_norm": 0.10802070796489716, + "learning_rate": 0.00010822192513368985, + "loss": 0.2157, + "step": 1378 + }, + { + "epoch": 1.3793879528617423, + "grad_norm": 0.10767026245594025, + "learning_rate": 0.00010815508021390375, + "loss": 0.2209, + "step": 1379 + }, + { + "epoch": 1.3803882341908662, + "grad_norm": 0.1139465942978859, + "learning_rate": 0.00010808823529411764, + "loss": 0.2283, + "step": 1380 + }, + { + "epoch": 1.38138851551999, + "grad_norm": 0.1004989892244339, + "learning_rate": 0.00010802139037433156, + "loss": 0.2191, + "step": 1381 + }, + { + "epoch": 1.3823887968491138, + "grad_norm": 0.10083423554897308, + "learning_rate": 0.00010795454545454547, + "loss": 0.2004, + "step": 1382 + }, + { + "epoch": 1.3833890781782376, + "grad_norm": 0.10467991977930069, + "learning_rate": 0.00010788770053475936, + "loss": 0.2032, + "step": 1383 + }, + { + "epoch": 1.3843893595073615, + "grad_norm": 0.10337726026773453, + "learning_rate": 0.00010782085561497326, + "loss": 0.2124, + "step": 1384 + }, + { + "epoch": 1.3853896408364852, + "grad_norm": 0.11278467625379562, + "learning_rate": 0.00010775401069518718, + "loss": 0.2275, + "step": 1385 + }, + { + "epoch": 1.386389922165609, + "grad_norm": 0.10587099939584732, + "learning_rate": 0.00010768716577540108, + "loss": 0.2036, + "step": 1386 + }, + { + "epoch": 1.387390203494733, + "grad_norm": 0.09782831370830536, + "learning_rate": 0.00010762032085561497, + "loss": 0.2049, + "step": 1387 + }, + { + "epoch": 1.3883904848238566, + "grad_norm": 0.10039117187261581, + "learning_rate": 0.00010755347593582888, + "loss": 0.2127, + "step": 1388 + }, + { + "epoch": 1.3893907661529805, + "grad_norm": 0.11189547181129456, + "learning_rate": 0.0001074866310160428, + "loss": 0.2138, + "step": 1389 + }, + { + "epoch": 1.3903910474821044, + "grad_norm": 0.10317673534154892, + "learning_rate": 0.00010741978609625669, + "loss": 0.1966, + "step": 1390 + }, + { + "epoch": 1.391391328811228, + "grad_norm": 0.10362192988395691, + "learning_rate": 0.00010735294117647059, + "loss": 0.2184, + "step": 1391 + }, + { + "epoch": 1.392391610140352, + "grad_norm": 0.10511241108179092, + "learning_rate": 0.00010728609625668448, + "loss": 0.2118, + "step": 1392 + }, + { + "epoch": 1.3933918914694758, + "grad_norm": 0.11975912004709244, + "learning_rate": 0.0001072192513368984, + "loss": 0.2205, + "step": 1393 + }, + { + "epoch": 1.3943921727985997, + "grad_norm": 0.1069861352443695, + "learning_rate": 0.0001071524064171123, + "loss": 0.2067, + "step": 1394 + }, + { + "epoch": 1.3953924541277234, + "grad_norm": 0.10979753732681274, + "learning_rate": 0.00010708556149732621, + "loss": 0.2223, + "step": 1395 + }, + { + "epoch": 1.3963927354568473, + "grad_norm": 0.12132672220468521, + "learning_rate": 0.0001070187165775401, + "loss": 0.2314, + "step": 1396 + }, + { + "epoch": 1.397393016785971, + "grad_norm": 0.1105416938662529, + "learning_rate": 0.00010695187165775402, + "loss": 0.2082, + "step": 1397 + }, + { + "epoch": 1.3983932981150948, + "grad_norm": 0.1076776310801506, + "learning_rate": 0.00010688502673796792, + "loss": 0.2285, + "step": 1398 + }, + { + "epoch": 1.3993935794442187, + "grad_norm": 0.10756873339414597, + "learning_rate": 0.00010681818181818181, + "loss": 0.2186, + "step": 1399 + }, + { + "epoch": 1.4003938607733426, + "grad_norm": 0.10076678544282913, + "learning_rate": 0.00010675133689839573, + "loss": 0.219, + "step": 1400 + }, + { + "epoch": 1.4013941421024663, + "grad_norm": 0.10140713304281235, + "learning_rate": 0.00010668449197860964, + "loss": 0.2208, + "step": 1401 + }, + { + "epoch": 1.4023944234315902, + "grad_norm": 0.11015689373016357, + "learning_rate": 0.00010661764705882353, + "loss": 0.2362, + "step": 1402 + }, + { + "epoch": 1.4033947047607138, + "grad_norm": 0.10933560132980347, + "learning_rate": 0.00010655080213903743, + "loss": 0.2284, + "step": 1403 + }, + { + "epoch": 1.4043949860898377, + "grad_norm": 0.0992734432220459, + "learning_rate": 0.00010648395721925135, + "loss": 0.2011, + "step": 1404 + }, + { + "epoch": 1.4053952674189616, + "grad_norm": 0.09986284375190735, + "learning_rate": 0.00010641711229946525, + "loss": 0.2127, + "step": 1405 + }, + { + "epoch": 1.4063955487480855, + "grad_norm": 0.10646960884332657, + "learning_rate": 0.00010635026737967914, + "loss": 0.2074, + "step": 1406 + }, + { + "epoch": 1.4073958300772091, + "grad_norm": 0.10831514745950699, + "learning_rate": 0.00010628342245989305, + "loss": 0.2162, + "step": 1407 + }, + { + "epoch": 1.408396111406333, + "grad_norm": 0.10946635156869888, + "learning_rate": 0.00010621657754010697, + "loss": 0.2165, + "step": 1408 + }, + { + "epoch": 1.409396392735457, + "grad_norm": 0.11144988983869553, + "learning_rate": 0.00010614973262032086, + "loss": 0.2136, + "step": 1409 + }, + { + "epoch": 1.4103966740645806, + "grad_norm": 0.11861346662044525, + "learning_rate": 0.00010608288770053476, + "loss": 0.2126, + "step": 1410 + }, + { + "epoch": 1.4113969553937045, + "grad_norm": 0.11905977874994278, + "learning_rate": 0.00010601604278074865, + "loss": 0.2327, + "step": 1411 + }, + { + "epoch": 1.4123972367228284, + "grad_norm": 0.11618269234895706, + "learning_rate": 0.00010594919786096258, + "loss": 0.2102, + "step": 1412 + }, + { + "epoch": 1.4133975180519522, + "grad_norm": 0.1202562004327774, + "learning_rate": 0.00010588235294117647, + "loss": 0.208, + "step": 1413 + }, + { + "epoch": 1.414397799381076, + "grad_norm": 0.11111394315958023, + "learning_rate": 0.00010581550802139038, + "loss": 0.2265, + "step": 1414 + }, + { + "epoch": 1.4153980807101998, + "grad_norm": 0.10783133655786514, + "learning_rate": 0.00010574866310160427, + "loss": 0.2183, + "step": 1415 + }, + { + "epoch": 1.4163983620393235, + "grad_norm": 0.10602202266454697, + "learning_rate": 0.00010568181818181819, + "loss": 0.2113, + "step": 1416 + }, + { + "epoch": 1.4173986433684473, + "grad_norm": 0.12160863727331161, + "learning_rate": 0.00010561497326203209, + "loss": 0.2267, + "step": 1417 + }, + { + "epoch": 1.4183989246975712, + "grad_norm": 0.1083454117178917, + "learning_rate": 0.00010554812834224598, + "loss": 0.2169, + "step": 1418 + }, + { + "epoch": 1.4193992060266951, + "grad_norm": 0.1062861829996109, + "learning_rate": 0.00010548128342245991, + "loss": 0.2177, + "step": 1419 + }, + { + "epoch": 1.4203994873558188, + "grad_norm": 0.10446483641862869, + "learning_rate": 0.0001054144385026738, + "loss": 0.2116, + "step": 1420 + }, + { + "epoch": 1.4213997686849427, + "grad_norm": 0.10060501843690872, + "learning_rate": 0.00010534759358288771, + "loss": 0.1956, + "step": 1421 + }, + { + "epoch": 1.4224000500140663, + "grad_norm": 0.11409254372119904, + "learning_rate": 0.0001052807486631016, + "loss": 0.2187, + "step": 1422 + }, + { + "epoch": 1.4234003313431902, + "grad_norm": 0.11280606687068939, + "learning_rate": 0.00010521390374331552, + "loss": 0.2156, + "step": 1423 + }, + { + "epoch": 1.424400612672314, + "grad_norm": 0.10444837808609009, + "learning_rate": 0.00010514705882352942, + "loss": 0.2195, + "step": 1424 + }, + { + "epoch": 1.425400894001438, + "grad_norm": 0.11143112182617188, + "learning_rate": 0.00010508021390374331, + "loss": 0.2086, + "step": 1425 + }, + { + "epoch": 1.4264011753305617, + "grad_norm": 0.11173329502344131, + "learning_rate": 0.00010501336898395722, + "loss": 0.2309, + "step": 1426 + }, + { + "epoch": 1.4274014566596855, + "grad_norm": 0.11144950240850449, + "learning_rate": 0.00010494652406417114, + "loss": 0.2197, + "step": 1427 + }, + { + "epoch": 1.4284017379888094, + "grad_norm": 0.11408905684947968, + "learning_rate": 0.00010487967914438504, + "loss": 0.2116, + "step": 1428 + }, + { + "epoch": 1.429402019317933, + "grad_norm": 0.10689879953861237, + "learning_rate": 0.00010481283422459893, + "loss": 0.2168, + "step": 1429 + }, + { + "epoch": 1.430402300647057, + "grad_norm": 0.10701993107795715, + "learning_rate": 0.00010474598930481284, + "loss": 0.215, + "step": 1430 + }, + { + "epoch": 1.4314025819761809, + "grad_norm": 0.10655166208744049, + "learning_rate": 0.00010467914438502675, + "loss": 0.216, + "step": 1431 + }, + { + "epoch": 1.4324028633053048, + "grad_norm": 0.10909377783536911, + "learning_rate": 0.00010461229946524064, + "loss": 0.2223, + "step": 1432 + }, + { + "epoch": 1.4334031446344284, + "grad_norm": 0.10624208301305771, + "learning_rate": 0.00010454545454545455, + "loss": 0.1979, + "step": 1433 + }, + { + "epoch": 1.4344034259635523, + "grad_norm": 0.1118341013789177, + "learning_rate": 0.00010447860962566844, + "loss": 0.2262, + "step": 1434 + }, + { + "epoch": 1.435403707292676, + "grad_norm": 0.10507479310035706, + "learning_rate": 0.00010441176470588237, + "loss": 0.215, + "step": 1435 + }, + { + "epoch": 1.4364039886217999, + "grad_norm": 0.11005060374736786, + "learning_rate": 0.00010434491978609626, + "loss": 0.2171, + "step": 1436 + }, + { + "epoch": 1.4374042699509237, + "grad_norm": 0.11328362673521042, + "learning_rate": 0.00010427807486631017, + "loss": 0.218, + "step": 1437 + }, + { + "epoch": 1.4384045512800476, + "grad_norm": 0.10928196460008621, + "learning_rate": 0.00010421122994652406, + "loss": 0.2113, + "step": 1438 + }, + { + "epoch": 1.4394048326091713, + "grad_norm": 0.11330752819776535, + "learning_rate": 0.00010414438502673797, + "loss": 0.2209, + "step": 1439 + }, + { + "epoch": 1.4404051139382952, + "grad_norm": 0.104282446205616, + "learning_rate": 0.00010407754010695188, + "loss": 0.2167, + "step": 1440 + }, + { + "epoch": 1.4414053952674188, + "grad_norm": 0.10631252825260162, + "learning_rate": 0.00010401069518716577, + "loss": 0.2179, + "step": 1441 + }, + { + "epoch": 1.4424056765965427, + "grad_norm": 0.11476306617259979, + "learning_rate": 0.00010394385026737969, + "loss": 0.2168, + "step": 1442 + }, + { + "epoch": 1.4434059579256666, + "grad_norm": 0.11525887250900269, + "learning_rate": 0.00010387700534759359, + "loss": 0.2198, + "step": 1443 + }, + { + "epoch": 1.4444062392547905, + "grad_norm": 0.1013231948018074, + "learning_rate": 0.0001038101604278075, + "loss": 0.2293, + "step": 1444 + }, + { + "epoch": 1.4454065205839142, + "grad_norm": 0.09863581508398056, + "learning_rate": 0.00010374331550802139, + "loss": 0.2039, + "step": 1445 + }, + { + "epoch": 1.446406801913038, + "grad_norm": 0.09742731600999832, + "learning_rate": 0.0001036764705882353, + "loss": 0.2074, + "step": 1446 + }, + { + "epoch": 1.447407083242162, + "grad_norm": 0.11330757290124893, + "learning_rate": 0.00010360962566844921, + "loss": 0.2159, + "step": 1447 + }, + { + "epoch": 1.4484073645712856, + "grad_norm": 0.10752193629741669, + "learning_rate": 0.0001035427807486631, + "loss": 0.2105, + "step": 1448 + }, + { + "epoch": 1.4494076459004095, + "grad_norm": 0.10365419089794159, + "learning_rate": 0.000103475935828877, + "loss": 0.2208, + "step": 1449 + }, + { + "epoch": 1.4504079272295334, + "grad_norm": 0.11370562016963959, + "learning_rate": 0.00010340909090909092, + "loss": 0.2193, + "step": 1450 + }, + { + "epoch": 1.451408208558657, + "grad_norm": 0.10761107504367828, + "learning_rate": 0.00010334224598930481, + "loss": 0.228, + "step": 1451 + }, + { + "epoch": 1.452408489887781, + "grad_norm": 0.10586556792259216, + "learning_rate": 0.00010327540106951872, + "loss": 0.2188, + "step": 1452 + }, + { + "epoch": 1.4534087712169048, + "grad_norm": 0.10861437022686005, + "learning_rate": 0.00010320855614973262, + "loss": 0.2175, + "step": 1453 + }, + { + "epoch": 1.4544090525460285, + "grad_norm": 0.11651885509490967, + "learning_rate": 0.00010314171122994654, + "loss": 0.2146, + "step": 1454 + }, + { + "epoch": 1.4554093338751524, + "grad_norm": 0.09943918138742447, + "learning_rate": 0.00010307486631016043, + "loss": 0.2059, + "step": 1455 + }, + { + "epoch": 1.4564096152042763, + "grad_norm": 0.10095467418432236, + "learning_rate": 0.00010300802139037434, + "loss": 0.2102, + "step": 1456 + }, + { + "epoch": 1.4574098965334001, + "grad_norm": 0.10872470587491989, + "learning_rate": 0.00010294117647058823, + "loss": 0.2171, + "step": 1457 + }, + { + "epoch": 1.4584101778625238, + "grad_norm": 0.10536850988864899, + "learning_rate": 0.00010287433155080214, + "loss": 0.2151, + "step": 1458 + }, + { + "epoch": 1.4594104591916477, + "grad_norm": 0.10445639491081238, + "learning_rate": 0.00010280748663101605, + "loss": 0.2135, + "step": 1459 + }, + { + "epoch": 1.4604107405207714, + "grad_norm": 0.10472364723682404, + "learning_rate": 0.00010274064171122994, + "loss": 0.2147, + "step": 1460 + }, + { + "epoch": 1.4614110218498952, + "grad_norm": 0.10720168799161911, + "learning_rate": 0.00010267379679144387, + "loss": 0.2123, + "step": 1461 + }, + { + "epoch": 1.4624113031790191, + "grad_norm": 0.10865159332752228, + "learning_rate": 0.00010260695187165776, + "loss": 0.2137, + "step": 1462 + }, + { + "epoch": 1.463411584508143, + "grad_norm": 0.10492442548274994, + "learning_rate": 0.00010254010695187167, + "loss": 0.2129, + "step": 1463 + }, + { + "epoch": 1.4644118658372667, + "grad_norm": 0.11085029691457748, + "learning_rate": 0.00010247326203208556, + "loss": 0.2118, + "step": 1464 + }, + { + "epoch": 1.4654121471663906, + "grad_norm": 0.10799961537122726, + "learning_rate": 0.00010240641711229947, + "loss": 0.2151, + "step": 1465 + }, + { + "epoch": 1.4664124284955142, + "grad_norm": 0.1062743067741394, + "learning_rate": 0.00010233957219251338, + "loss": 0.2144, + "step": 1466 + }, + { + "epoch": 1.4674127098246381, + "grad_norm": 0.10964664816856384, + "learning_rate": 0.00010227272727272727, + "loss": 0.213, + "step": 1467 + }, + { + "epoch": 1.468412991153762, + "grad_norm": 0.11286632716655731, + "learning_rate": 0.00010220588235294117, + "loss": 0.2292, + "step": 1468 + }, + { + "epoch": 1.469413272482886, + "grad_norm": 0.10352053493261337, + "learning_rate": 0.00010213903743315509, + "loss": 0.2163, + "step": 1469 + }, + { + "epoch": 1.4704135538120096, + "grad_norm": 0.11184529960155487, + "learning_rate": 0.000102072192513369, + "loss": 0.2226, + "step": 1470 + }, + { + "epoch": 1.4714138351411334, + "grad_norm": 0.11039632558822632, + "learning_rate": 0.00010200534759358289, + "loss": 0.2223, + "step": 1471 + }, + { + "epoch": 1.4724141164702573, + "grad_norm": 0.11355464905500412, + "learning_rate": 0.00010193850267379679, + "loss": 0.216, + "step": 1472 + }, + { + "epoch": 1.473414397799381, + "grad_norm": 0.11336928606033325, + "learning_rate": 0.00010187165775401071, + "loss": 0.2241, + "step": 1473 + }, + { + "epoch": 1.4744146791285049, + "grad_norm": 0.10922425240278244, + "learning_rate": 0.0001018048128342246, + "loss": 0.2091, + "step": 1474 + }, + { + "epoch": 1.4754149604576288, + "grad_norm": 0.11782589554786682, + "learning_rate": 0.0001017379679144385, + "loss": 0.2213, + "step": 1475 + }, + { + "epoch": 1.4764152417867527, + "grad_norm": 0.11659085750579834, + "learning_rate": 0.0001016711229946524, + "loss": 0.2173, + "step": 1476 + }, + { + "epoch": 1.4774155231158763, + "grad_norm": 0.10084234178066254, + "learning_rate": 0.00010160427807486633, + "loss": 0.2246, + "step": 1477 + }, + { + "epoch": 1.4784158044450002, + "grad_norm": 0.1274043172597885, + "learning_rate": 0.00010153743315508022, + "loss": 0.212, + "step": 1478 + }, + { + "epoch": 1.4794160857741239, + "grad_norm": 0.1067751869559288, + "learning_rate": 0.00010147058823529412, + "loss": 0.2024, + "step": 1479 + }, + { + "epoch": 1.4804163671032478, + "grad_norm": 0.12341886758804321, + "learning_rate": 0.00010140374331550801, + "loss": 0.2331, + "step": 1480 + }, + { + "epoch": 1.4814166484323716, + "grad_norm": 0.10220299661159515, + "learning_rate": 0.00010133689839572193, + "loss": 0.2118, + "step": 1481 + }, + { + "epoch": 1.4824169297614955, + "grad_norm": 0.11613665521144867, + "learning_rate": 0.00010127005347593584, + "loss": 0.2065, + "step": 1482 + }, + { + "epoch": 1.4834172110906192, + "grad_norm": 0.10381147265434265, + "learning_rate": 0.00010120320855614973, + "loss": 0.2045, + "step": 1483 + }, + { + "epoch": 1.484417492419743, + "grad_norm": 0.10971830040216446, + "learning_rate": 0.00010113636363636366, + "loss": 0.2186, + "step": 1484 + }, + { + "epoch": 1.4854177737488667, + "grad_norm": 0.10020864754915237, + "learning_rate": 0.00010106951871657755, + "loss": 0.2046, + "step": 1485 + }, + { + "epoch": 1.4864180550779906, + "grad_norm": 0.11323567479848862, + "learning_rate": 0.00010100267379679145, + "loss": 0.2191, + "step": 1486 + }, + { + "epoch": 1.4874183364071145, + "grad_norm": 0.11057696491479874, + "learning_rate": 0.00010093582887700534, + "loss": 0.2194, + "step": 1487 + }, + { + "epoch": 1.4884186177362384, + "grad_norm": 0.1122126653790474, + "learning_rate": 0.00010086898395721926, + "loss": 0.2166, + "step": 1488 + }, + { + "epoch": 1.489418899065362, + "grad_norm": 0.12285346537828445, + "learning_rate": 0.00010080213903743317, + "loss": 0.2306, + "step": 1489 + }, + { + "epoch": 1.490419180394486, + "grad_norm": 0.1010991558432579, + "learning_rate": 0.00010073529411764706, + "loss": 0.2132, + "step": 1490 + }, + { + "epoch": 1.4914194617236098, + "grad_norm": 0.11029645800590515, + "learning_rate": 0.00010066844919786096, + "loss": 0.2167, + "step": 1491 + }, + { + "epoch": 1.4924197430527335, + "grad_norm": 0.1131289079785347, + "learning_rate": 0.00010060160427807488, + "loss": 0.2119, + "step": 1492 + }, + { + "epoch": 1.4934200243818574, + "grad_norm": 0.10018480569124222, + "learning_rate": 0.00010053475935828878, + "loss": 0.1967, + "step": 1493 + }, + { + "epoch": 1.4944203057109813, + "grad_norm": 0.11729665845632553, + "learning_rate": 0.00010046791443850267, + "loss": 0.2268, + "step": 1494 + }, + { + "epoch": 1.4954205870401052, + "grad_norm": 0.10619371384382248, + "learning_rate": 0.00010040106951871658, + "loss": 0.2154, + "step": 1495 + }, + { + "epoch": 1.4964208683692288, + "grad_norm": 0.10219235718250275, + "learning_rate": 0.0001003342245989305, + "loss": 0.2145, + "step": 1496 + }, + { + "epoch": 1.4974211496983527, + "grad_norm": 0.10226699709892273, + "learning_rate": 0.00010026737967914439, + "loss": 0.2063, + "step": 1497 + }, + { + "epoch": 1.4984214310274764, + "grad_norm": 0.10711207985877991, + "learning_rate": 0.00010020053475935829, + "loss": 0.2016, + "step": 1498 + }, + { + "epoch": 1.4994217123566003, + "grad_norm": 0.10331369191408157, + "learning_rate": 0.00010013368983957218, + "loss": 0.2156, + "step": 1499 + }, + { + "epoch": 1.5004219936857242, + "grad_norm": 0.11426566541194916, + "learning_rate": 0.0001000668449197861, + "loss": 0.2188, + "step": 1500 + }, + { + "epoch": 1.501422275014848, + "grad_norm": 0.11031512916088104, + "learning_rate": 0.0001, + "loss": 0.2187, + "step": 1501 + }, + { + "epoch": 1.5024225563439717, + "grad_norm": 0.1046115979552269, + "learning_rate": 9.993315508021391e-05, + "loss": 0.2156, + "step": 1502 + }, + { + "epoch": 1.5034228376730956, + "grad_norm": 0.11534816026687622, + "learning_rate": 9.986631016042781e-05, + "loss": 0.2154, + "step": 1503 + }, + { + "epoch": 1.5044231190022193, + "grad_norm": 0.11485407501459122, + "learning_rate": 9.979946524064172e-05, + "loss": 0.2142, + "step": 1504 + }, + { + "epoch": 1.5054234003313431, + "grad_norm": 0.09823386371135712, + "learning_rate": 9.973262032085562e-05, + "loss": 0.2045, + "step": 1505 + }, + { + "epoch": 1.506423681660467, + "grad_norm": 0.1149950623512268, + "learning_rate": 9.966577540106953e-05, + "loss": 0.2267, + "step": 1506 + }, + { + "epoch": 1.507423962989591, + "grad_norm": 0.10567053407430649, + "learning_rate": 9.959893048128343e-05, + "loss": 0.2139, + "step": 1507 + }, + { + "epoch": 1.5084242443187148, + "grad_norm": 0.11233936995267868, + "learning_rate": 9.953208556149734e-05, + "loss": 0.2273, + "step": 1508 + }, + { + "epoch": 1.5094245256478385, + "grad_norm": 0.10278680920600891, + "learning_rate": 9.946524064171123e-05, + "loss": 0.2126, + "step": 1509 + }, + { + "epoch": 1.5104248069769621, + "grad_norm": 0.10968869924545288, + "learning_rate": 9.939839572192514e-05, + "loss": 0.2288, + "step": 1510 + }, + { + "epoch": 1.511425088306086, + "grad_norm": 0.09708631783723831, + "learning_rate": 9.933155080213903e-05, + "loss": 0.2098, + "step": 1511 + }, + { + "epoch": 1.51242536963521, + "grad_norm": 0.10178802162408829, + "learning_rate": 9.926470588235295e-05, + "loss": 0.198, + "step": 1512 + }, + { + "epoch": 1.5134256509643338, + "grad_norm": 0.11479008197784424, + "learning_rate": 9.919786096256684e-05, + "loss": 0.2156, + "step": 1513 + }, + { + "epoch": 1.5144259322934577, + "grad_norm": 0.1110190749168396, + "learning_rate": 9.913101604278076e-05, + "loss": 0.2091, + "step": 1514 + }, + { + "epoch": 1.5154262136225813, + "grad_norm": 0.11105778068304062, + "learning_rate": 9.906417112299465e-05, + "loss": 0.225, + "step": 1515 + }, + { + "epoch": 1.516426494951705, + "grad_norm": 0.10996527969837189, + "learning_rate": 9.899732620320856e-05, + "loss": 0.2166, + "step": 1516 + }, + { + "epoch": 1.517426776280829, + "grad_norm": 0.10849972814321518, + "learning_rate": 9.893048128342246e-05, + "loss": 0.213, + "step": 1517 + }, + { + "epoch": 1.5184270576099528, + "grad_norm": 0.11246906220912933, + "learning_rate": 9.886363636363637e-05, + "loss": 0.2086, + "step": 1518 + }, + { + "epoch": 1.5194273389390767, + "grad_norm": 0.11339261382818222, + "learning_rate": 9.879679144385027e-05, + "loss": 0.2227, + "step": 1519 + }, + { + "epoch": 1.5204276202682006, + "grad_norm": 0.11148970574140549, + "learning_rate": 9.872994652406417e-05, + "loss": 0.2154, + "step": 1520 + }, + { + "epoch": 1.5214279015973242, + "grad_norm": 0.10175742208957672, + "learning_rate": 9.866310160427808e-05, + "loss": 0.2196, + "step": 1521 + }, + { + "epoch": 1.522428182926448, + "grad_norm": 0.1137404665350914, + "learning_rate": 9.859625668449198e-05, + "loss": 0.2184, + "step": 1522 + }, + { + "epoch": 1.5234284642555718, + "grad_norm": 0.11410399526357651, + "learning_rate": 9.852941176470589e-05, + "loss": 0.2141, + "step": 1523 + }, + { + "epoch": 1.5244287455846957, + "grad_norm": 0.11105949431657791, + "learning_rate": 9.846256684491979e-05, + "loss": 0.2087, + "step": 1524 + }, + { + "epoch": 1.5254290269138195, + "grad_norm": 0.11356593668460846, + "learning_rate": 9.83957219251337e-05, + "loss": 0.2182, + "step": 1525 + }, + { + "epoch": 1.5264293082429434, + "grad_norm": 0.10612566769123077, + "learning_rate": 9.83288770053476e-05, + "loss": 0.2279, + "step": 1526 + }, + { + "epoch": 1.527429589572067, + "grad_norm": 0.11434666812419891, + "learning_rate": 9.82620320855615e-05, + "loss": 0.2249, + "step": 1527 + }, + { + "epoch": 1.528429870901191, + "grad_norm": 0.1005062609910965, + "learning_rate": 9.819518716577541e-05, + "loss": 0.2157, + "step": 1528 + }, + { + "epoch": 1.5294301522303146, + "grad_norm": 0.10602647811174393, + "learning_rate": 9.812834224598931e-05, + "loss": 0.205, + "step": 1529 + }, + { + "epoch": 1.5304304335594385, + "grad_norm": 0.10368629544973373, + "learning_rate": 9.806149732620322e-05, + "loss": 0.2206, + "step": 1530 + }, + { + "epoch": 1.5314307148885624, + "grad_norm": 0.10456206649541855, + "learning_rate": 9.799465240641712e-05, + "loss": 0.2102, + "step": 1531 + }, + { + "epoch": 1.5324309962176863, + "grad_norm": 0.10952454805374146, + "learning_rate": 9.792780748663101e-05, + "loss": 0.2029, + "step": 1532 + }, + { + "epoch": 1.5334312775468102, + "grad_norm": 0.10971931368112564, + "learning_rate": 9.786096256684493e-05, + "loss": 0.2179, + "step": 1533 + }, + { + "epoch": 1.5344315588759339, + "grad_norm": 0.11332719027996063, + "learning_rate": 9.779411764705882e-05, + "loss": 0.2045, + "step": 1534 + }, + { + "epoch": 1.5354318402050575, + "grad_norm": 0.11003556847572327, + "learning_rate": 9.772727272727274e-05, + "loss": 0.212, + "step": 1535 + }, + { + "epoch": 1.5364321215341814, + "grad_norm": 0.10037890076637268, + "learning_rate": 9.766042780748663e-05, + "loss": 0.2185, + "step": 1536 + }, + { + "epoch": 1.5374324028633053, + "grad_norm": 0.10277759283781052, + "learning_rate": 9.759358288770053e-05, + "loss": 0.2213, + "step": 1537 + }, + { + "epoch": 1.5384326841924292, + "grad_norm": 0.10957152396440506, + "learning_rate": 9.752673796791444e-05, + "loss": 0.2183, + "step": 1538 + }, + { + "epoch": 1.539432965521553, + "grad_norm": 0.11526452004909515, + "learning_rate": 9.745989304812834e-05, + "loss": 0.211, + "step": 1539 + }, + { + "epoch": 1.5404332468506767, + "grad_norm": 0.11079391837120056, + "learning_rate": 9.739304812834225e-05, + "loss": 0.2067, + "step": 1540 + }, + { + "epoch": 1.5414335281798006, + "grad_norm": 0.10437075048685074, + "learning_rate": 9.732620320855615e-05, + "loss": 0.2066, + "step": 1541 + }, + { + "epoch": 1.5424338095089243, + "grad_norm": 0.12434093654155731, + "learning_rate": 9.725935828877006e-05, + "loss": 0.2287, + "step": 1542 + }, + { + "epoch": 1.5434340908380482, + "grad_norm": 0.12912294268608093, + "learning_rate": 9.719251336898396e-05, + "loss": 0.222, + "step": 1543 + }, + { + "epoch": 1.544434372167172, + "grad_norm": 0.11466380953788757, + "learning_rate": 9.712566844919787e-05, + "loss": 0.2146, + "step": 1544 + }, + { + "epoch": 1.545434653496296, + "grad_norm": 0.10675331205129623, + "learning_rate": 9.705882352941177e-05, + "loss": 0.2083, + "step": 1545 + }, + { + "epoch": 1.5464349348254196, + "grad_norm": 0.10839785635471344, + "learning_rate": 9.699197860962567e-05, + "loss": 0.215, + "step": 1546 + }, + { + "epoch": 1.5474352161545435, + "grad_norm": 0.10307509452104568, + "learning_rate": 9.692513368983958e-05, + "loss": 0.215, + "step": 1547 + }, + { + "epoch": 1.5484354974836672, + "grad_norm": 0.10201635211706161, + "learning_rate": 9.685828877005348e-05, + "loss": 0.2117, + "step": 1548 + }, + { + "epoch": 1.549435778812791, + "grad_norm": 0.09652825444936752, + "learning_rate": 9.679144385026739e-05, + "loss": 0.2103, + "step": 1549 + }, + { + "epoch": 1.550436060141915, + "grad_norm": 0.09914563596248627, + "learning_rate": 9.672459893048129e-05, + "loss": 0.2146, + "step": 1550 + }, + { + "epoch": 1.5514363414710388, + "grad_norm": 0.09930212050676346, + "learning_rate": 9.66577540106952e-05, + "loss": 0.2005, + "step": 1551 + }, + { + "epoch": 1.5524366228001627, + "grad_norm": 0.10766632854938507, + "learning_rate": 9.65909090909091e-05, + "loss": 0.2033, + "step": 1552 + }, + { + "epoch": 1.5534369041292864, + "grad_norm": 0.11091277748346329, + "learning_rate": 9.652406417112299e-05, + "loss": 0.2269, + "step": 1553 + }, + { + "epoch": 1.55443718545841, + "grad_norm": 0.1077127754688263, + "learning_rate": 9.645721925133691e-05, + "loss": 0.2055, + "step": 1554 + }, + { + "epoch": 1.555437466787534, + "grad_norm": 0.10168565809726715, + "learning_rate": 9.63903743315508e-05, + "loss": 0.2089, + "step": 1555 + }, + { + "epoch": 1.5564377481166578, + "grad_norm": 0.11611717939376831, + "learning_rate": 9.632352941176472e-05, + "loss": 0.2251, + "step": 1556 + }, + { + "epoch": 1.5574380294457817, + "grad_norm": 0.10269337892532349, + "learning_rate": 9.625668449197861e-05, + "loss": 0.2113, + "step": 1557 + }, + { + "epoch": 1.5584383107749056, + "grad_norm": 0.1092357262969017, + "learning_rate": 9.618983957219251e-05, + "loss": 0.2224, + "step": 1558 + }, + { + "epoch": 1.5594385921040292, + "grad_norm": 0.10343734174966812, + "learning_rate": 9.612299465240642e-05, + "loss": 0.2169, + "step": 1559 + }, + { + "epoch": 1.5604388734331531, + "grad_norm": 0.1017196774482727, + "learning_rate": 9.605614973262032e-05, + "loss": 0.2181, + "step": 1560 + }, + { + "epoch": 1.5614391547622768, + "grad_norm": 0.09910575300455093, + "learning_rate": 9.598930481283423e-05, + "loss": 0.214, + "step": 1561 + }, + { + "epoch": 1.5624394360914007, + "grad_norm": 0.10937146842479706, + "learning_rate": 9.592245989304813e-05, + "loss": 0.2208, + "step": 1562 + }, + { + "epoch": 1.5634397174205246, + "grad_norm": 0.10042283684015274, + "learning_rate": 9.585561497326203e-05, + "loss": 0.2043, + "step": 1563 + }, + { + "epoch": 1.5644399987496485, + "grad_norm": 0.10303983092308044, + "learning_rate": 9.578877005347594e-05, + "loss": 0.2024, + "step": 1564 + }, + { + "epoch": 1.5654402800787721, + "grad_norm": 0.11757756024599075, + "learning_rate": 9.572192513368984e-05, + "loss": 0.2178, + "step": 1565 + }, + { + "epoch": 1.566440561407896, + "grad_norm": 0.11159020662307739, + "learning_rate": 9.565508021390375e-05, + "loss": 0.2126, + "step": 1566 + }, + { + "epoch": 1.5674408427370197, + "grad_norm": 0.10356171429157257, + "learning_rate": 9.558823529411765e-05, + "loss": 0.2036, + "step": 1567 + }, + { + "epoch": 1.5684411240661436, + "grad_norm": 0.10913465172052383, + "learning_rate": 9.552139037433156e-05, + "loss": 0.2151, + "step": 1568 + }, + { + "epoch": 1.5694414053952674, + "grad_norm": 0.11238259822130203, + "learning_rate": 9.545454545454546e-05, + "loss": 0.2041, + "step": 1569 + }, + { + "epoch": 1.5704416867243913, + "grad_norm": 0.10680039972066879, + "learning_rate": 9.538770053475937e-05, + "loss": 0.2108, + "step": 1570 + }, + { + "epoch": 1.5714419680535152, + "grad_norm": 0.10895457863807678, + "learning_rate": 9.532085561497327e-05, + "loss": 0.2098, + "step": 1571 + }, + { + "epoch": 1.5724422493826389, + "grad_norm": 0.10954467952251434, + "learning_rate": 9.525401069518717e-05, + "loss": 0.207, + "step": 1572 + }, + { + "epoch": 1.5734425307117625, + "grad_norm": 0.1125275045633316, + "learning_rate": 9.518716577540108e-05, + "loss": 0.2126, + "step": 1573 + }, + { + "epoch": 1.5744428120408864, + "grad_norm": 0.10717690736055374, + "learning_rate": 9.512032085561497e-05, + "loss": 0.1977, + "step": 1574 + }, + { + "epoch": 1.5754430933700103, + "grad_norm": 0.10968661308288574, + "learning_rate": 9.505347593582889e-05, + "loss": 0.2229, + "step": 1575 + }, + { + "epoch": 1.5764433746991342, + "grad_norm": 0.10862824320793152, + "learning_rate": 9.498663101604278e-05, + "loss": 0.2263, + "step": 1576 + }, + { + "epoch": 1.577443656028258, + "grad_norm": 0.10937079042196274, + "learning_rate": 9.49197860962567e-05, + "loss": 0.2047, + "step": 1577 + }, + { + "epoch": 1.5784439373573818, + "grad_norm": 0.12127456814050674, + "learning_rate": 9.485294117647059e-05, + "loss": 0.2113, + "step": 1578 + }, + { + "epoch": 1.5794442186865054, + "grad_norm": 0.10156920552253723, + "learning_rate": 9.47860962566845e-05, + "loss": 0.2134, + "step": 1579 + }, + { + "epoch": 1.5804445000156293, + "grad_norm": 0.09869392216205597, + "learning_rate": 9.47192513368984e-05, + "loss": 0.2013, + "step": 1580 + }, + { + "epoch": 1.5814447813447532, + "grad_norm": 0.12013694643974304, + "learning_rate": 9.46524064171123e-05, + "loss": 0.216, + "step": 1581 + }, + { + "epoch": 1.582445062673877, + "grad_norm": 0.10934696346521378, + "learning_rate": 9.45855614973262e-05, + "loss": 0.2098, + "step": 1582 + }, + { + "epoch": 1.583445344003001, + "grad_norm": 0.10526914894580841, + "learning_rate": 9.451871657754011e-05, + "loss": 0.2087, + "step": 1583 + }, + { + "epoch": 1.5844456253321246, + "grad_norm": 0.10731624066829681, + "learning_rate": 9.445187165775401e-05, + "loss": 0.2142, + "step": 1584 + }, + { + "epoch": 1.5854459066612485, + "grad_norm": 0.11411863565444946, + "learning_rate": 9.438502673796792e-05, + "loss": 0.2175, + "step": 1585 + }, + { + "epoch": 1.5864461879903722, + "grad_norm": 0.10996554046869278, + "learning_rate": 9.431818181818182e-05, + "loss": 0.2219, + "step": 1586 + }, + { + "epoch": 1.587446469319496, + "grad_norm": 0.10800634324550629, + "learning_rate": 9.425133689839573e-05, + "loss": 0.2197, + "step": 1587 + }, + { + "epoch": 1.58844675064862, + "grad_norm": 0.10492725670337677, + "learning_rate": 9.418449197860963e-05, + "loss": 0.2124, + "step": 1588 + }, + { + "epoch": 1.5894470319777438, + "grad_norm": 0.10127617418766022, + "learning_rate": 9.411764705882353e-05, + "loss": 0.2152, + "step": 1589 + }, + { + "epoch": 1.5904473133068675, + "grad_norm": 0.109656922519207, + "learning_rate": 9.405080213903744e-05, + "loss": 0.2201, + "step": 1590 + }, + { + "epoch": 1.5914475946359914, + "grad_norm": 0.11269284039735794, + "learning_rate": 9.398395721925134e-05, + "loss": 0.2244, + "step": 1591 + }, + { + "epoch": 1.592447875965115, + "grad_norm": 0.11160044372081757, + "learning_rate": 9.391711229946525e-05, + "loss": 0.2222, + "step": 1592 + }, + { + "epoch": 1.593448157294239, + "grad_norm": 0.10462243109941483, + "learning_rate": 9.385026737967915e-05, + "loss": 0.2001, + "step": 1593 + }, + { + "epoch": 1.5944484386233628, + "grad_norm": 0.10957028716802597, + "learning_rate": 9.378342245989306e-05, + "loss": 0.2099, + "step": 1594 + }, + { + "epoch": 1.5954487199524867, + "grad_norm": 0.1106254830956459, + "learning_rate": 9.371657754010695e-05, + "loss": 0.2098, + "step": 1595 + }, + { + "epoch": 1.5964490012816106, + "grad_norm": 0.10405244678258896, + "learning_rate": 9.364973262032087e-05, + "loss": 0.214, + "step": 1596 + }, + { + "epoch": 1.5974492826107343, + "grad_norm": 0.10569903254508972, + "learning_rate": 9.358288770053476e-05, + "loss": 0.2085, + "step": 1597 + }, + { + "epoch": 1.598449563939858, + "grad_norm": 0.11747591197490692, + "learning_rate": 9.351604278074867e-05, + "loss": 0.2144, + "step": 1598 + }, + { + "epoch": 1.5994498452689818, + "grad_norm": 0.10684362053871155, + "learning_rate": 9.344919786096256e-05, + "loss": 0.2168, + "step": 1599 + }, + { + "epoch": 1.6004501265981057, + "grad_norm": 0.10157062113285065, + "learning_rate": 9.338235294117648e-05, + "loss": 0.2, + "step": 1600 + }, + { + "epoch": 1.6014504079272296, + "grad_norm": 0.10556425154209137, + "learning_rate": 9.331550802139037e-05, + "loss": 0.2057, + "step": 1601 + }, + { + "epoch": 1.6024506892563535, + "grad_norm": 0.10767196863889694, + "learning_rate": 9.324866310160428e-05, + "loss": 0.2278, + "step": 1602 + }, + { + "epoch": 1.6034509705854771, + "grad_norm": 0.1117590144276619, + "learning_rate": 9.318181818181818e-05, + "loss": 0.2112, + "step": 1603 + }, + { + "epoch": 1.604451251914601, + "grad_norm": 0.11272083967924118, + "learning_rate": 9.311497326203209e-05, + "loss": 0.2226, + "step": 1604 + }, + { + "epoch": 1.6054515332437247, + "grad_norm": 0.10809974372386932, + "learning_rate": 9.3048128342246e-05, + "loss": 0.2117, + "step": 1605 + }, + { + "epoch": 1.6064518145728486, + "grad_norm": 0.12140113115310669, + "learning_rate": 9.29812834224599e-05, + "loss": 0.2254, + "step": 1606 + }, + { + "epoch": 1.6074520959019725, + "grad_norm": 0.11645787209272385, + "learning_rate": 9.29144385026738e-05, + "loss": 0.2121, + "step": 1607 + }, + { + "epoch": 1.6084523772310964, + "grad_norm": 0.10584229230880737, + "learning_rate": 9.28475935828877e-05, + "loss": 0.2182, + "step": 1608 + }, + { + "epoch": 1.60945265856022, + "grad_norm": 0.10512715578079224, + "learning_rate": 9.278074866310161e-05, + "loss": 0.2062, + "step": 1609 + }, + { + "epoch": 1.610452939889344, + "grad_norm": 0.1002088412642479, + "learning_rate": 9.271390374331551e-05, + "loss": 0.2072, + "step": 1610 + }, + { + "epoch": 1.6114532212184676, + "grad_norm": 0.10402756184339523, + "learning_rate": 9.264705882352942e-05, + "loss": 0.2046, + "step": 1611 + }, + { + "epoch": 1.6124535025475915, + "grad_norm": 0.1045699417591095, + "learning_rate": 9.258021390374332e-05, + "loss": 0.2025, + "step": 1612 + }, + { + "epoch": 1.6134537838767153, + "grad_norm": 0.11762456595897675, + "learning_rate": 9.251336898395723e-05, + "loss": 0.2197, + "step": 1613 + }, + { + "epoch": 1.6144540652058392, + "grad_norm": 0.09863623231649399, + "learning_rate": 9.244652406417113e-05, + "loss": 0.207, + "step": 1614 + }, + { + "epoch": 1.6154543465349631, + "grad_norm": 0.10165554285049438, + "learning_rate": 9.237967914438503e-05, + "loss": 0.2128, + "step": 1615 + }, + { + "epoch": 1.6164546278640868, + "grad_norm": 0.11495894193649292, + "learning_rate": 9.231283422459893e-05, + "loss": 0.2229, + "step": 1616 + }, + { + "epoch": 1.6174549091932104, + "grad_norm": 0.10019998997449875, + "learning_rate": 9.224598930481284e-05, + "loss": 0.2127, + "step": 1617 + }, + { + "epoch": 1.6184551905223343, + "grad_norm": 0.10050822049379349, + "learning_rate": 9.217914438502673e-05, + "loss": 0.204, + "step": 1618 + }, + { + "epoch": 1.6194554718514582, + "grad_norm": 0.11399617046117783, + "learning_rate": 9.211229946524065e-05, + "loss": 0.2287, + "step": 1619 + }, + { + "epoch": 1.620455753180582, + "grad_norm": 0.10086178034543991, + "learning_rate": 9.204545454545454e-05, + "loss": 0.2079, + "step": 1620 + }, + { + "epoch": 1.621456034509706, + "grad_norm": 0.10343603789806366, + "learning_rate": 9.197860962566846e-05, + "loss": 0.2077, + "step": 1621 + }, + { + "epoch": 1.6224563158388297, + "grad_norm": 0.1042860895395279, + "learning_rate": 9.191176470588235e-05, + "loss": 0.2124, + "step": 1622 + }, + { + "epoch": 1.6234565971679535, + "grad_norm": 0.10938288271427155, + "learning_rate": 9.184491978609626e-05, + "loss": 0.2193, + "step": 1623 + }, + { + "epoch": 1.6244568784970772, + "grad_norm": 0.10304956883192062, + "learning_rate": 9.177807486631016e-05, + "loss": 0.2068, + "step": 1624 + }, + { + "epoch": 1.625457159826201, + "grad_norm": 0.10855516791343689, + "learning_rate": 9.171122994652406e-05, + "loss": 0.2277, + "step": 1625 + }, + { + "epoch": 1.626457441155325, + "grad_norm": 0.11576005071401596, + "learning_rate": 9.164438502673798e-05, + "loss": 0.23, + "step": 1626 + }, + { + "epoch": 1.6274577224844489, + "grad_norm": 0.10778895765542984, + "learning_rate": 9.157754010695187e-05, + "loss": 0.2097, + "step": 1627 + }, + { + "epoch": 1.6284580038135725, + "grad_norm": 0.11882217973470688, + "learning_rate": 9.151069518716579e-05, + "loss": 0.2187, + "step": 1628 + }, + { + "epoch": 1.6294582851426964, + "grad_norm": 0.1104099303483963, + "learning_rate": 9.144385026737968e-05, + "loss": 0.2214, + "step": 1629 + }, + { + "epoch": 1.63045856647182, + "grad_norm": 0.11682116240262985, + "learning_rate": 9.137700534759359e-05, + "loss": 0.2211, + "step": 1630 + }, + { + "epoch": 1.631458847800944, + "grad_norm": 0.1024133712053299, + "learning_rate": 9.131016042780749e-05, + "loss": 0.2011, + "step": 1631 + }, + { + "epoch": 1.6324591291300679, + "grad_norm": 0.10174795985221863, + "learning_rate": 9.12433155080214e-05, + "loss": 0.2034, + "step": 1632 + }, + { + "epoch": 1.6334594104591917, + "grad_norm": 0.11420445889234543, + "learning_rate": 9.11764705882353e-05, + "loss": 0.2215, + "step": 1633 + }, + { + "epoch": 1.6344596917883156, + "grad_norm": 0.10743626952171326, + "learning_rate": 9.11096256684492e-05, + "loss": 0.2208, + "step": 1634 + }, + { + "epoch": 1.6354599731174393, + "grad_norm": 0.10999636352062225, + "learning_rate": 9.104278074866311e-05, + "loss": 0.2167, + "step": 1635 + }, + { + "epoch": 1.636460254446563, + "grad_norm": 0.11092141270637512, + "learning_rate": 9.097593582887701e-05, + "loss": 0.2262, + "step": 1636 + }, + { + "epoch": 1.6374605357756868, + "grad_norm": 0.10303587466478348, + "learning_rate": 9.090909090909092e-05, + "loss": 0.2121, + "step": 1637 + }, + { + "epoch": 1.6384608171048107, + "grad_norm": 0.10024894773960114, + "learning_rate": 9.084224598930482e-05, + "loss": 0.2075, + "step": 1638 + }, + { + "epoch": 1.6394610984339346, + "grad_norm": 0.11568810790777206, + "learning_rate": 9.077540106951871e-05, + "loss": 0.2139, + "step": 1639 + }, + { + "epoch": 1.6404613797630585, + "grad_norm": 0.10673253238201141, + "learning_rate": 9.070855614973263e-05, + "loss": 0.2133, + "step": 1640 + }, + { + "epoch": 1.6414616610921822, + "grad_norm": 0.12340815365314484, + "learning_rate": 9.064171122994652e-05, + "loss": 0.2404, + "step": 1641 + }, + { + "epoch": 1.6424619424213058, + "grad_norm": 0.11298071593046188, + "learning_rate": 9.057486631016044e-05, + "loss": 0.2278, + "step": 1642 + }, + { + "epoch": 1.6434622237504297, + "grad_norm": 0.11273998767137527, + "learning_rate": 9.050802139037433e-05, + "loss": 0.2239, + "step": 1643 + }, + { + "epoch": 1.6444625050795536, + "grad_norm": 0.10492943972349167, + "learning_rate": 9.044117647058823e-05, + "loss": 0.2149, + "step": 1644 + }, + { + "epoch": 1.6454627864086775, + "grad_norm": 0.1166037917137146, + "learning_rate": 9.037433155080214e-05, + "loss": 0.2141, + "step": 1645 + }, + { + "epoch": 1.6464630677378014, + "grad_norm": 0.10623501241207123, + "learning_rate": 9.030748663101604e-05, + "loss": 0.2168, + "step": 1646 + }, + { + "epoch": 1.647463349066925, + "grad_norm": 0.10060600191354752, + "learning_rate": 9.024064171122996e-05, + "loss": 0.2018, + "step": 1647 + }, + { + "epoch": 1.648463630396049, + "grad_norm": 0.10876615345478058, + "learning_rate": 9.017379679144385e-05, + "loss": 0.2062, + "step": 1648 + }, + { + "epoch": 1.6494639117251726, + "grad_norm": 0.0990198478102684, + "learning_rate": 9.010695187165777e-05, + "loss": 0.2174, + "step": 1649 + }, + { + "epoch": 1.6504641930542965, + "grad_norm": 0.10246013849973679, + "learning_rate": 9.004010695187166e-05, + "loss": 0.2098, + "step": 1650 + }, + { + "epoch": 1.6514644743834204, + "grad_norm": 0.10005972534418106, + "learning_rate": 8.997326203208556e-05, + "loss": 0.1979, + "step": 1651 + }, + { + "epoch": 1.6524647557125443, + "grad_norm": 0.1049349382519722, + "learning_rate": 8.990641711229947e-05, + "loss": 0.2195, + "step": 1652 + }, + { + "epoch": 1.653465037041668, + "grad_norm": 0.11116772890090942, + "learning_rate": 8.983957219251337e-05, + "loss": 0.2016, + "step": 1653 + }, + { + "epoch": 1.6544653183707918, + "grad_norm": 0.10333818197250366, + "learning_rate": 8.977272727272728e-05, + "loss": 0.2106, + "step": 1654 + }, + { + "epoch": 1.6554655996999155, + "grad_norm": 0.11180448532104492, + "learning_rate": 8.970588235294118e-05, + "loss": 0.2139, + "step": 1655 + }, + { + "epoch": 1.6564658810290394, + "grad_norm": 0.10754438489675522, + "learning_rate": 8.963903743315509e-05, + "loss": 0.2066, + "step": 1656 + }, + { + "epoch": 1.6574661623581632, + "grad_norm": 0.11590081453323364, + "learning_rate": 8.957219251336899e-05, + "loss": 0.2256, + "step": 1657 + }, + { + "epoch": 1.6584664436872871, + "grad_norm": 0.11364041268825531, + "learning_rate": 8.95053475935829e-05, + "loss": 0.2049, + "step": 1658 + }, + { + "epoch": 1.659466725016411, + "grad_norm": 0.12753775715827942, + "learning_rate": 8.94385026737968e-05, + "loss": 0.231, + "step": 1659 + }, + { + "epoch": 1.6604670063455347, + "grad_norm": 0.11407069116830826, + "learning_rate": 8.937165775401069e-05, + "loss": 0.2206, + "step": 1660 + }, + { + "epoch": 1.6614672876746583, + "grad_norm": 0.10583723336458206, + "learning_rate": 8.930481283422461e-05, + "loss": 0.2146, + "step": 1661 + }, + { + "epoch": 1.6624675690037822, + "grad_norm": 0.09623390436172485, + "learning_rate": 8.92379679144385e-05, + "loss": 0.2054, + "step": 1662 + }, + { + "epoch": 1.6634678503329061, + "grad_norm": 0.11790817230939865, + "learning_rate": 8.917112299465242e-05, + "loss": 0.2152, + "step": 1663 + }, + { + "epoch": 1.66446813166203, + "grad_norm": 0.1058674305677414, + "learning_rate": 8.910427807486631e-05, + "loss": 0.2043, + "step": 1664 + }, + { + "epoch": 1.665468412991154, + "grad_norm": 0.10389213263988495, + "learning_rate": 8.903743315508021e-05, + "loss": 0.1942, + "step": 1665 + }, + { + "epoch": 1.6664686943202776, + "grad_norm": 0.10507882386445999, + "learning_rate": 8.897058823529412e-05, + "loss": 0.2076, + "step": 1666 + }, + { + "epoch": 1.6674689756494014, + "grad_norm": 0.09810091555118561, + "learning_rate": 8.890374331550802e-05, + "loss": 0.2061, + "step": 1667 + }, + { + "epoch": 1.668469256978525, + "grad_norm": 0.09805666655302048, + "learning_rate": 8.883689839572194e-05, + "loss": 0.2298, + "step": 1668 + }, + { + "epoch": 1.669469538307649, + "grad_norm": 0.10857600718736649, + "learning_rate": 8.877005347593583e-05, + "loss": 0.2156, + "step": 1669 + }, + { + "epoch": 1.6704698196367729, + "grad_norm": 0.10095620155334473, + "learning_rate": 8.870320855614975e-05, + "loss": 0.2077, + "step": 1670 + }, + { + "epoch": 1.6714701009658968, + "grad_norm": 0.10749758034944534, + "learning_rate": 8.863636363636364e-05, + "loss": 0.2123, + "step": 1671 + }, + { + "epoch": 1.6724703822950204, + "grad_norm": 0.10441987216472626, + "learning_rate": 8.856951871657754e-05, + "loss": 0.2139, + "step": 1672 + }, + { + "epoch": 1.6734706636241443, + "grad_norm": 0.11343187093734741, + "learning_rate": 8.850267379679145e-05, + "loss": 0.2328, + "step": 1673 + }, + { + "epoch": 1.674470944953268, + "grad_norm": 0.11036334186792374, + "learning_rate": 8.843582887700535e-05, + "loss": 0.2247, + "step": 1674 + }, + { + "epoch": 1.6754712262823919, + "grad_norm": 0.10940060019493103, + "learning_rate": 8.836898395721926e-05, + "loss": 0.228, + "step": 1675 + }, + { + "epoch": 1.6764715076115158, + "grad_norm": 0.11064553260803223, + "learning_rate": 8.830213903743316e-05, + "loss": 0.2084, + "step": 1676 + }, + { + "epoch": 1.6774717889406396, + "grad_norm": 0.10345584154129028, + "learning_rate": 8.823529411764706e-05, + "loss": 0.1961, + "step": 1677 + }, + { + "epoch": 1.6784720702697635, + "grad_norm": 0.11640360951423645, + "learning_rate": 8.816844919786097e-05, + "loss": 0.22, + "step": 1678 + }, + { + "epoch": 1.6794723515988872, + "grad_norm": 0.11305073648691177, + "learning_rate": 8.810160427807487e-05, + "loss": 0.2057, + "step": 1679 + }, + { + "epoch": 1.6804726329280109, + "grad_norm": 0.10483142733573914, + "learning_rate": 8.803475935828878e-05, + "loss": 0.2069, + "step": 1680 + }, + { + "epoch": 1.6814729142571347, + "grad_norm": 0.10991936177015305, + "learning_rate": 8.796791443850267e-05, + "loss": 0.2074, + "step": 1681 + }, + { + "epoch": 1.6824731955862586, + "grad_norm": 0.11477912962436676, + "learning_rate": 8.790106951871659e-05, + "loss": 0.2165, + "step": 1682 + }, + { + "epoch": 1.6834734769153825, + "grad_norm": 0.0983867421746254, + "learning_rate": 8.783422459893048e-05, + "loss": 0.2062, + "step": 1683 + }, + { + "epoch": 1.6844737582445064, + "grad_norm": 0.10866306722164154, + "learning_rate": 8.77673796791444e-05, + "loss": 0.2117, + "step": 1684 + }, + { + "epoch": 1.68547403957363, + "grad_norm": 0.11202388256788254, + "learning_rate": 8.770053475935829e-05, + "loss": 0.2127, + "step": 1685 + }, + { + "epoch": 1.686474320902754, + "grad_norm": 0.11255204677581787, + "learning_rate": 8.76336898395722e-05, + "loss": 0.2176, + "step": 1686 + }, + { + "epoch": 1.6874746022318776, + "grad_norm": 0.11509371548891068, + "learning_rate": 8.75668449197861e-05, + "loss": 0.2141, + "step": 1687 + }, + { + "epoch": 1.6884748835610015, + "grad_norm": 0.10615269839763641, + "learning_rate": 8.75e-05, + "loss": 0.2127, + "step": 1688 + }, + { + "epoch": 1.6894751648901254, + "grad_norm": 0.1074584349989891, + "learning_rate": 8.743315508021392e-05, + "loss": 0.213, + "step": 1689 + }, + { + "epoch": 1.6904754462192493, + "grad_norm": 0.1094815731048584, + "learning_rate": 8.736631016042781e-05, + "loss": 0.2226, + "step": 1690 + }, + { + "epoch": 1.691475727548373, + "grad_norm": 0.097354955971241, + "learning_rate": 8.729946524064173e-05, + "loss": 0.2066, + "step": 1691 + }, + { + "epoch": 1.6924760088774968, + "grad_norm": 0.10424072295427322, + "learning_rate": 8.723262032085562e-05, + "loss": 0.2102, + "step": 1692 + }, + { + "epoch": 1.6934762902066205, + "grad_norm": 0.10481325536966324, + "learning_rate": 8.716577540106952e-05, + "loss": 0.2153, + "step": 1693 + }, + { + "epoch": 1.6944765715357444, + "grad_norm": 0.10457316786050797, + "learning_rate": 8.709893048128343e-05, + "loss": 0.2199, + "step": 1694 + }, + { + "epoch": 1.6954768528648683, + "grad_norm": 0.10526327043771744, + "learning_rate": 8.703208556149733e-05, + "loss": 0.2106, + "step": 1695 + }, + { + "epoch": 1.6964771341939922, + "grad_norm": 0.11397548019886017, + "learning_rate": 8.696524064171123e-05, + "loss": 0.2178, + "step": 1696 + }, + { + "epoch": 1.697477415523116, + "grad_norm": 0.09890329092741013, + "learning_rate": 8.689839572192514e-05, + "loss": 0.2115, + "step": 1697 + }, + { + "epoch": 1.6984776968522397, + "grad_norm": 0.10453622788190842, + "learning_rate": 8.683155080213904e-05, + "loss": 0.211, + "step": 1698 + }, + { + "epoch": 1.6994779781813634, + "grad_norm": 0.10167501121759415, + "learning_rate": 8.676470588235295e-05, + "loss": 0.2128, + "step": 1699 + }, + { + "epoch": 1.7004782595104873, + "grad_norm": 0.1141904890537262, + "learning_rate": 8.669786096256685e-05, + "loss": 0.213, + "step": 1700 + }, + { + "epoch": 1.7014785408396111, + "grad_norm": 0.11739058047533035, + "learning_rate": 8.663101604278076e-05, + "loss": 0.2049, + "step": 1701 + }, + { + "epoch": 1.702478822168735, + "grad_norm": 0.10483638942241669, + "learning_rate": 8.656417112299465e-05, + "loss": 0.2187, + "step": 1702 + }, + { + "epoch": 1.703479103497859, + "grad_norm": 0.10907094925642014, + "learning_rate": 8.649732620320856e-05, + "loss": 0.2168, + "step": 1703 + }, + { + "epoch": 1.7044793848269826, + "grad_norm": 0.10213258117437363, + "learning_rate": 8.643048128342246e-05, + "loss": 0.1984, + "step": 1704 + }, + { + "epoch": 1.7054796661561062, + "grad_norm": 0.11457552760839462, + "learning_rate": 8.636363636363637e-05, + "loss": 0.2091, + "step": 1705 + }, + { + "epoch": 1.7064799474852301, + "grad_norm": 0.12065617740154266, + "learning_rate": 8.629679144385026e-05, + "loss": 0.2245, + "step": 1706 + }, + { + "epoch": 1.707480228814354, + "grad_norm": 0.10916194319725037, + "learning_rate": 8.622994652406418e-05, + "loss": 0.2089, + "step": 1707 + }, + { + "epoch": 1.708480510143478, + "grad_norm": 0.11573344469070435, + "learning_rate": 8.616310160427807e-05, + "loss": 0.2234, + "step": 1708 + }, + { + "epoch": 1.7094807914726018, + "grad_norm": 0.11099280416965485, + "learning_rate": 8.609625668449198e-05, + "loss": 0.2195, + "step": 1709 + }, + { + "epoch": 1.7104810728017255, + "grad_norm": 0.10968057811260223, + "learning_rate": 8.60294117647059e-05, + "loss": 0.2204, + "step": 1710 + }, + { + "epoch": 1.7114813541308493, + "grad_norm": 0.11191478371620178, + "learning_rate": 8.596256684491979e-05, + "loss": 0.2135, + "step": 1711 + }, + { + "epoch": 1.712481635459973, + "grad_norm": 0.10917894542217255, + "learning_rate": 8.58957219251337e-05, + "loss": 0.2167, + "step": 1712 + }, + { + "epoch": 1.713481916789097, + "grad_norm": 0.11488395184278488, + "learning_rate": 8.58288770053476e-05, + "loss": 0.2179, + "step": 1713 + }, + { + "epoch": 1.7144821981182208, + "grad_norm": 0.10785124450922012, + "learning_rate": 8.57620320855615e-05, + "loss": 0.2145, + "step": 1714 + }, + { + "epoch": 1.7154824794473447, + "grad_norm": 0.11953595280647278, + "learning_rate": 8.56951871657754e-05, + "loss": 0.221, + "step": 1715 + }, + { + "epoch": 1.7164827607764683, + "grad_norm": 0.10059189796447754, + "learning_rate": 8.562834224598931e-05, + "loss": 0.2123, + "step": 1716 + }, + { + "epoch": 1.7174830421055922, + "grad_norm": 0.10600530356168747, + "learning_rate": 8.556149732620321e-05, + "loss": 0.2156, + "step": 1717 + }, + { + "epoch": 1.7184833234347159, + "grad_norm": 0.10740195214748383, + "learning_rate": 8.549465240641712e-05, + "loss": 0.2171, + "step": 1718 + }, + { + "epoch": 1.7194836047638398, + "grad_norm": 0.10259955376386642, + "learning_rate": 8.542780748663102e-05, + "loss": 0.2019, + "step": 1719 + }, + { + "epoch": 1.7204838860929637, + "grad_norm": 0.11497773975133896, + "learning_rate": 8.536096256684493e-05, + "loss": 0.2231, + "step": 1720 + }, + { + "epoch": 1.7214841674220875, + "grad_norm": 0.10886845737695694, + "learning_rate": 8.529411764705883e-05, + "loss": 0.2155, + "step": 1721 + }, + { + "epoch": 1.7224844487512114, + "grad_norm": 0.107436403632164, + "learning_rate": 8.522727272727273e-05, + "loss": 0.2135, + "step": 1722 + }, + { + "epoch": 1.723484730080335, + "grad_norm": 0.10516581684350967, + "learning_rate": 8.516042780748662e-05, + "loss": 0.2048, + "step": 1723 + }, + { + "epoch": 1.7244850114094588, + "grad_norm": 0.11618438363075256, + "learning_rate": 8.509358288770054e-05, + "loss": 0.2204, + "step": 1724 + }, + { + "epoch": 1.7254852927385826, + "grad_norm": 0.1045473963022232, + "learning_rate": 8.502673796791443e-05, + "loss": 0.2048, + "step": 1725 + }, + { + "epoch": 1.7264855740677065, + "grad_norm": 0.10112176090478897, + "learning_rate": 8.495989304812835e-05, + "loss": 0.2023, + "step": 1726 + }, + { + "epoch": 1.7274858553968304, + "grad_norm": 0.10988399386405945, + "learning_rate": 8.489304812834224e-05, + "loss": 0.2076, + "step": 1727 + }, + { + "epoch": 1.7284861367259543, + "grad_norm": 0.10352276265621185, + "learning_rate": 8.482620320855616e-05, + "loss": 0.2055, + "step": 1728 + }, + { + "epoch": 1.729486418055078, + "grad_norm": 0.1032441258430481, + "learning_rate": 8.475935828877005e-05, + "loss": 0.2073, + "step": 1729 + }, + { + "epoch": 1.7304866993842019, + "grad_norm": 0.11377128958702087, + "learning_rate": 8.469251336898396e-05, + "loss": 0.2054, + "step": 1730 + }, + { + "epoch": 1.7314869807133255, + "grad_norm": 0.11277464777231216, + "learning_rate": 8.462566844919787e-05, + "loss": 0.212, + "step": 1731 + }, + { + "epoch": 1.7324872620424494, + "grad_norm": 0.1109778955578804, + "learning_rate": 8.455882352941176e-05, + "loss": 0.2197, + "step": 1732 + }, + { + "epoch": 1.7334875433715733, + "grad_norm": 0.10970083624124527, + "learning_rate": 8.449197860962568e-05, + "loss": 0.2168, + "step": 1733 + }, + { + "epoch": 1.7344878247006972, + "grad_norm": 0.11743413656949997, + "learning_rate": 8.442513368983957e-05, + "loss": 0.2249, + "step": 1734 + }, + { + "epoch": 1.7354881060298208, + "grad_norm": 0.10438866168260574, + "learning_rate": 8.435828877005349e-05, + "loss": 0.1998, + "step": 1735 + }, + { + "epoch": 1.7364883873589447, + "grad_norm": 0.11464299261569977, + "learning_rate": 8.429144385026738e-05, + "loss": 0.2215, + "step": 1736 + }, + { + "epoch": 1.7374886686880684, + "grad_norm": 0.10639890283346176, + "learning_rate": 8.422459893048129e-05, + "loss": 0.226, + "step": 1737 + }, + { + "epoch": 1.7384889500171923, + "grad_norm": 0.10950041562318802, + "learning_rate": 8.415775401069519e-05, + "loss": 0.2135, + "step": 1738 + }, + { + "epoch": 1.7394892313463162, + "grad_norm": 0.10413890331983566, + "learning_rate": 8.40909090909091e-05, + "loss": 0.2112, + "step": 1739 + }, + { + "epoch": 1.74048951267544, + "grad_norm": 0.11211550235748291, + "learning_rate": 8.4024064171123e-05, + "loss": 0.2063, + "step": 1740 + }, + { + "epoch": 1.741489794004564, + "grad_norm": 0.1060446947813034, + "learning_rate": 8.39572192513369e-05, + "loss": 0.2189, + "step": 1741 + }, + { + "epoch": 1.7424900753336876, + "grad_norm": 0.10858438163995743, + "learning_rate": 8.389037433155081e-05, + "loss": 0.2204, + "step": 1742 + }, + { + "epoch": 1.7434903566628113, + "grad_norm": 0.10869374126195908, + "learning_rate": 8.382352941176471e-05, + "loss": 0.2141, + "step": 1743 + }, + { + "epoch": 1.7444906379919352, + "grad_norm": 0.10977107286453247, + "learning_rate": 8.375668449197862e-05, + "loss": 0.2064, + "step": 1744 + }, + { + "epoch": 1.745490919321059, + "grad_norm": 0.09883537888526917, + "learning_rate": 8.368983957219252e-05, + "loss": 0.1978, + "step": 1745 + }, + { + "epoch": 1.746491200650183, + "grad_norm": 0.10449741780757904, + "learning_rate": 8.362299465240641e-05, + "loss": 0.1989, + "step": 1746 + }, + { + "epoch": 1.7474914819793068, + "grad_norm": 0.10172144323587418, + "learning_rate": 8.355614973262033e-05, + "loss": 0.2069, + "step": 1747 + }, + { + "epoch": 1.7484917633084305, + "grad_norm": 0.11277353763580322, + "learning_rate": 8.348930481283422e-05, + "loss": 0.2216, + "step": 1748 + }, + { + "epoch": 1.7494920446375544, + "grad_norm": 0.11038568615913391, + "learning_rate": 8.342245989304814e-05, + "loss": 0.2226, + "step": 1749 + }, + { + "epoch": 1.750492325966678, + "grad_norm": 0.11066082119941711, + "learning_rate": 8.335561497326203e-05, + "loss": 0.2287, + "step": 1750 + }, + { + "epoch": 1.751492607295802, + "grad_norm": 0.09979293495416641, + "learning_rate": 8.328877005347593e-05, + "loss": 0.2129, + "step": 1751 + }, + { + "epoch": 1.7524928886249258, + "grad_norm": 0.1008995845913887, + "learning_rate": 8.322192513368985e-05, + "loss": 0.2009, + "step": 1752 + }, + { + "epoch": 1.7534931699540497, + "grad_norm": 0.10486951470375061, + "learning_rate": 8.315508021390374e-05, + "loss": 0.2149, + "step": 1753 + }, + { + "epoch": 1.7544934512831734, + "grad_norm": 0.10083181411027908, + "learning_rate": 8.308823529411766e-05, + "loss": 0.2083, + "step": 1754 + }, + { + "epoch": 1.7554937326122972, + "grad_norm": 0.10758165270090103, + "learning_rate": 8.302139037433155e-05, + "loss": 0.2126, + "step": 1755 + }, + { + "epoch": 1.756494013941421, + "grad_norm": 0.11825454980134964, + "learning_rate": 8.295454545454547e-05, + "loss": 0.217, + "step": 1756 + }, + { + "epoch": 1.7574942952705448, + "grad_norm": 0.11239483207464218, + "learning_rate": 8.288770053475936e-05, + "loss": 0.2197, + "step": 1757 + }, + { + "epoch": 1.7584945765996687, + "grad_norm": 0.11691700667142868, + "learning_rate": 8.282085561497326e-05, + "loss": 0.2183, + "step": 1758 + }, + { + "epoch": 1.7594948579287926, + "grad_norm": 0.11044658720493317, + "learning_rate": 8.275401069518717e-05, + "loss": 0.2058, + "step": 1759 + }, + { + "epoch": 1.7604951392579165, + "grad_norm": 0.11619790643453598, + "learning_rate": 8.268716577540107e-05, + "loss": 0.2156, + "step": 1760 + }, + { + "epoch": 1.7614954205870401, + "grad_norm": 0.10028060525655746, + "learning_rate": 8.262032085561498e-05, + "loss": 0.2023, + "step": 1761 + }, + { + "epoch": 1.7624957019161638, + "grad_norm": 0.10769213736057281, + "learning_rate": 8.255347593582888e-05, + "loss": 0.2066, + "step": 1762 + }, + { + "epoch": 1.7634959832452877, + "grad_norm": 0.11169063299894333, + "learning_rate": 8.248663101604279e-05, + "loss": 0.2219, + "step": 1763 + }, + { + "epoch": 1.7644962645744116, + "grad_norm": 0.10832317173480988, + "learning_rate": 8.241978609625669e-05, + "loss": 0.2167, + "step": 1764 + }, + { + "epoch": 1.7654965459035354, + "grad_norm": 0.11215530335903168, + "learning_rate": 8.23529411764706e-05, + "loss": 0.2123, + "step": 1765 + }, + { + "epoch": 1.7664968272326593, + "grad_norm": 0.10337299108505249, + "learning_rate": 8.22860962566845e-05, + "loss": 0.2207, + "step": 1766 + }, + { + "epoch": 1.767497108561783, + "grad_norm": 0.10118553787469864, + "learning_rate": 8.221925133689839e-05, + "loss": 0.2089, + "step": 1767 + }, + { + "epoch": 1.7684973898909067, + "grad_norm": 0.10753005743026733, + "learning_rate": 8.215240641711231e-05, + "loss": 0.2147, + "step": 1768 + }, + { + "epoch": 1.7694976712200305, + "grad_norm": 0.10666035115718842, + "learning_rate": 8.20855614973262e-05, + "loss": 0.2016, + "step": 1769 + }, + { + "epoch": 1.7704979525491544, + "grad_norm": 0.10878204554319382, + "learning_rate": 8.201871657754012e-05, + "loss": 0.2145, + "step": 1770 + }, + { + "epoch": 1.7714982338782783, + "grad_norm": 0.10458698868751526, + "learning_rate": 8.195187165775401e-05, + "loss": 0.199, + "step": 1771 + }, + { + "epoch": 1.7724985152074022, + "grad_norm": 0.11779174208641052, + "learning_rate": 8.188502673796791e-05, + "loss": 0.2204, + "step": 1772 + }, + { + "epoch": 1.7734987965365259, + "grad_norm": 0.10379134118556976, + "learning_rate": 8.181818181818183e-05, + "loss": 0.2198, + "step": 1773 + }, + { + "epoch": 1.7744990778656498, + "grad_norm": 0.1094084307551384, + "learning_rate": 8.175133689839572e-05, + "loss": 0.2048, + "step": 1774 + }, + { + "epoch": 1.7754993591947734, + "grad_norm": 0.11885455250740051, + "learning_rate": 8.168449197860964e-05, + "loss": 0.2413, + "step": 1775 + }, + { + "epoch": 1.7764996405238973, + "grad_norm": 0.10773204267024994, + "learning_rate": 8.161764705882353e-05, + "loss": 0.2027, + "step": 1776 + }, + { + "epoch": 1.7774999218530212, + "grad_norm": 0.10064983367919922, + "learning_rate": 8.155080213903745e-05, + "loss": 0.2106, + "step": 1777 + }, + { + "epoch": 1.778500203182145, + "grad_norm": 0.10450369119644165, + "learning_rate": 8.148395721925134e-05, + "loss": 0.2247, + "step": 1778 + }, + { + "epoch": 1.7795004845112687, + "grad_norm": 0.10685044527053833, + "learning_rate": 8.141711229946524e-05, + "loss": 0.2069, + "step": 1779 + }, + { + "epoch": 1.7805007658403926, + "grad_norm": 0.12208949774503708, + "learning_rate": 8.135026737967915e-05, + "loss": 0.2179, + "step": 1780 + }, + { + "epoch": 1.7815010471695163, + "grad_norm": 0.11099454760551453, + "learning_rate": 8.128342245989305e-05, + "loss": 0.2126, + "step": 1781 + }, + { + "epoch": 1.7825013284986402, + "grad_norm": 0.11030252277851105, + "learning_rate": 8.121657754010696e-05, + "loss": 0.2185, + "step": 1782 + }, + { + "epoch": 1.783501609827764, + "grad_norm": 0.1002948135137558, + "learning_rate": 8.114973262032086e-05, + "loss": 0.1936, + "step": 1783 + }, + { + "epoch": 1.784501891156888, + "grad_norm": 0.10328270494937897, + "learning_rate": 8.108288770053476e-05, + "loss": 0.1972, + "step": 1784 + }, + { + "epoch": 1.7855021724860118, + "grad_norm": 0.10595179349184036, + "learning_rate": 8.101604278074867e-05, + "loss": 0.2114, + "step": 1785 + }, + { + "epoch": 1.7865024538151355, + "grad_norm": 0.10791525989770889, + "learning_rate": 8.094919786096257e-05, + "loss": 0.2101, + "step": 1786 + }, + { + "epoch": 1.7875027351442592, + "grad_norm": 0.10468003153800964, + "learning_rate": 8.088235294117648e-05, + "loss": 0.2185, + "step": 1787 + }, + { + "epoch": 1.788503016473383, + "grad_norm": 0.11155645549297333, + "learning_rate": 8.081550802139037e-05, + "loss": 0.2178, + "step": 1788 + }, + { + "epoch": 1.789503297802507, + "grad_norm": 0.10646934807300568, + "learning_rate": 8.074866310160429e-05, + "loss": 0.2058, + "step": 1789 + }, + { + "epoch": 1.7905035791316308, + "grad_norm": 0.11313287913799286, + "learning_rate": 8.068181818181818e-05, + "loss": 0.2083, + "step": 1790 + }, + { + "epoch": 1.7915038604607547, + "grad_norm": 0.10804971307516098, + "learning_rate": 8.06149732620321e-05, + "loss": 0.2225, + "step": 1791 + }, + { + "epoch": 1.7925041417898784, + "grad_norm": 0.10363703221082687, + "learning_rate": 8.0548128342246e-05, + "loss": 0.2071, + "step": 1792 + }, + { + "epoch": 1.7935044231190023, + "grad_norm": 0.11234106868505478, + "learning_rate": 8.04812834224599e-05, + "loss": 0.2068, + "step": 1793 + }, + { + "epoch": 1.794504704448126, + "grad_norm": 0.11610477417707443, + "learning_rate": 8.041443850267381e-05, + "loss": 0.21, + "step": 1794 + }, + { + "epoch": 1.7955049857772498, + "grad_norm": 0.10835031419992447, + "learning_rate": 8.03475935828877e-05, + "loss": 0.2113, + "step": 1795 + }, + { + "epoch": 1.7965052671063737, + "grad_norm": 0.09929194301366806, + "learning_rate": 8.028074866310162e-05, + "loss": 0.2095, + "step": 1796 + }, + { + "epoch": 1.7975055484354976, + "grad_norm": 0.0975765734910965, + "learning_rate": 8.021390374331551e-05, + "loss": 0.205, + "step": 1797 + }, + { + "epoch": 1.7985058297646213, + "grad_norm": 0.1147439181804657, + "learning_rate": 8.014705882352943e-05, + "loss": 0.2063, + "step": 1798 + }, + { + "epoch": 1.7995061110937451, + "grad_norm": 0.11163458228111267, + "learning_rate": 8.008021390374332e-05, + "loss": 0.2005, + "step": 1799 + }, + { + "epoch": 1.8005063924228688, + "grad_norm": 0.0958055704832077, + "learning_rate": 8.001336898395722e-05, + "loss": 0.2052, + "step": 1800 + }, + { + "epoch": 1.8015066737519927, + "grad_norm": 0.1088767871260643, + "learning_rate": 7.994652406417112e-05, + "loss": 0.211, + "step": 1801 + }, + { + "epoch": 1.8025069550811166, + "grad_norm": 0.10337083786725998, + "learning_rate": 7.987967914438503e-05, + "loss": 0.2105, + "step": 1802 + }, + { + "epoch": 1.8035072364102405, + "grad_norm": 0.1079808846116066, + "learning_rate": 7.981283422459893e-05, + "loss": 0.2157, + "step": 1803 + }, + { + "epoch": 1.8045075177393644, + "grad_norm": 0.10174074023962021, + "learning_rate": 7.974598930481284e-05, + "loss": 0.2121, + "step": 1804 + }, + { + "epoch": 1.805507799068488, + "grad_norm": 0.0995740070939064, + "learning_rate": 7.967914438502674e-05, + "loss": 0.1993, + "step": 1805 + }, + { + "epoch": 1.8065080803976117, + "grad_norm": 0.09516981244087219, + "learning_rate": 7.961229946524065e-05, + "loss": 0.1945, + "step": 1806 + }, + { + "epoch": 1.8075083617267356, + "grad_norm": 0.11642731726169586, + "learning_rate": 7.954545454545455e-05, + "loss": 0.2052, + "step": 1807 + }, + { + "epoch": 1.8085086430558595, + "grad_norm": 0.1034930869936943, + "learning_rate": 7.947860962566846e-05, + "loss": 0.1996, + "step": 1808 + }, + { + "epoch": 1.8095089243849833, + "grad_norm": 0.10340449213981628, + "learning_rate": 7.941176470588235e-05, + "loss": 0.2117, + "step": 1809 + }, + { + "epoch": 1.8105092057141072, + "grad_norm": 0.11522252857685089, + "learning_rate": 7.934491978609626e-05, + "loss": 0.2198, + "step": 1810 + }, + { + "epoch": 1.811509487043231, + "grad_norm": 0.11083301901817322, + "learning_rate": 7.927807486631015e-05, + "loss": 0.2176, + "step": 1811 + }, + { + "epoch": 1.8125097683723548, + "grad_norm": 0.12410628795623779, + "learning_rate": 7.921122994652407e-05, + "loss": 0.2286, + "step": 1812 + }, + { + "epoch": 1.8135100497014784, + "grad_norm": 0.11115820705890656, + "learning_rate": 7.914438502673798e-05, + "loss": 0.211, + "step": 1813 + }, + { + "epoch": 1.8145103310306023, + "grad_norm": 0.1103898361325264, + "learning_rate": 7.907754010695188e-05, + "loss": 0.2012, + "step": 1814 + }, + { + "epoch": 1.8155106123597262, + "grad_norm": 0.12027064710855484, + "learning_rate": 7.901069518716579e-05, + "loss": 0.2337, + "step": 1815 + }, + { + "epoch": 1.81651089368885, + "grad_norm": 0.11153366416692734, + "learning_rate": 7.894385026737968e-05, + "loss": 0.2118, + "step": 1816 + }, + { + "epoch": 1.8175111750179738, + "grad_norm": 0.10536403208971024, + "learning_rate": 7.88770053475936e-05, + "loss": 0.2137, + "step": 1817 + }, + { + "epoch": 1.8185114563470977, + "grad_norm": 0.11436942219734192, + "learning_rate": 7.881016042780749e-05, + "loss": 0.2171, + "step": 1818 + }, + { + "epoch": 1.8195117376762213, + "grad_norm": 0.10944034159183502, + "learning_rate": 7.87433155080214e-05, + "loss": 0.2087, + "step": 1819 + }, + { + "epoch": 1.8205120190053452, + "grad_norm": 0.11831872165203094, + "learning_rate": 7.86764705882353e-05, + "loss": 0.2114, + "step": 1820 + }, + { + "epoch": 1.821512300334469, + "grad_norm": 0.10910452902317047, + "learning_rate": 7.86096256684492e-05, + "loss": 0.2121, + "step": 1821 + }, + { + "epoch": 1.822512581663593, + "grad_norm": 0.11008214205503464, + "learning_rate": 7.85427807486631e-05, + "loss": 0.2121, + "step": 1822 + }, + { + "epoch": 1.8235128629927169, + "grad_norm": 0.10772056877613068, + "learning_rate": 7.847593582887701e-05, + "loss": 0.2178, + "step": 1823 + }, + { + "epoch": 1.8245131443218405, + "grad_norm": 0.10806655883789062, + "learning_rate": 7.840909090909091e-05, + "loss": 0.2074, + "step": 1824 + }, + { + "epoch": 1.8255134256509642, + "grad_norm": 0.11554738879203796, + "learning_rate": 7.834224598930482e-05, + "loss": 0.2093, + "step": 1825 + }, + { + "epoch": 1.826513706980088, + "grad_norm": 0.10009465366601944, + "learning_rate": 7.827540106951872e-05, + "loss": 0.1989, + "step": 1826 + }, + { + "epoch": 1.827513988309212, + "grad_norm": 0.10841847211122513, + "learning_rate": 7.820855614973262e-05, + "loss": 0.2158, + "step": 1827 + }, + { + "epoch": 1.8285142696383359, + "grad_norm": 0.10409947484731674, + "learning_rate": 7.814171122994653e-05, + "loss": 0.208, + "step": 1828 + }, + { + "epoch": 1.8295145509674597, + "grad_norm": 0.11051104962825775, + "learning_rate": 7.807486631016043e-05, + "loss": 0.2182, + "step": 1829 + }, + { + "epoch": 1.8305148322965834, + "grad_norm": 0.11080829054117203, + "learning_rate": 7.800802139037432e-05, + "loss": 0.2185, + "step": 1830 + }, + { + "epoch": 1.831515113625707, + "grad_norm": 0.10957919806241989, + "learning_rate": 7.794117647058824e-05, + "loss": 0.2137, + "step": 1831 + }, + { + "epoch": 1.832515394954831, + "grad_norm": 0.12292095273733139, + "learning_rate": 7.787433155080213e-05, + "loss": 0.234, + "step": 1832 + }, + { + "epoch": 1.8335156762839548, + "grad_norm": 0.10798388719558716, + "learning_rate": 7.780748663101605e-05, + "loss": 0.201, + "step": 1833 + }, + { + "epoch": 1.8345159576130787, + "grad_norm": 0.10908353328704834, + "learning_rate": 7.774064171122996e-05, + "loss": 0.1964, + "step": 1834 + }, + { + "epoch": 1.8355162389422026, + "grad_norm": 0.1159132719039917, + "learning_rate": 7.767379679144386e-05, + "loss": 0.221, + "step": 1835 + }, + { + "epoch": 1.8365165202713263, + "grad_norm": 0.09885982424020767, + "learning_rate": 7.760695187165776e-05, + "loss": 0.1991, + "step": 1836 + }, + { + "epoch": 1.8375168016004502, + "grad_norm": 0.09949743747711182, + "learning_rate": 7.754010695187165e-05, + "loss": 0.1935, + "step": 1837 + }, + { + "epoch": 1.8385170829295738, + "grad_norm": 0.11341581493616104, + "learning_rate": 7.747326203208557e-05, + "loss": 0.2207, + "step": 1838 + }, + { + "epoch": 1.8395173642586977, + "grad_norm": 0.10507316887378693, + "learning_rate": 7.740641711229946e-05, + "loss": 0.1964, + "step": 1839 + }, + { + "epoch": 1.8405176455878216, + "grad_norm": 0.11824993044137955, + "learning_rate": 7.733957219251338e-05, + "loss": 0.2234, + "step": 1840 + }, + { + "epoch": 1.8415179269169455, + "grad_norm": 0.11139892041683197, + "learning_rate": 7.727272727272727e-05, + "loss": 0.2161, + "step": 1841 + }, + { + "epoch": 1.8425182082460692, + "grad_norm": 0.10503961890935898, + "learning_rate": 7.720588235294119e-05, + "loss": 0.2068, + "step": 1842 + }, + { + "epoch": 1.843518489575193, + "grad_norm": 0.11308833956718445, + "learning_rate": 7.713903743315508e-05, + "loss": 0.2164, + "step": 1843 + }, + { + "epoch": 1.8445187709043167, + "grad_norm": 0.10557015985250473, + "learning_rate": 7.707219251336899e-05, + "loss": 0.214, + "step": 1844 + }, + { + "epoch": 1.8455190522334406, + "grad_norm": 0.10252027213573456, + "learning_rate": 7.700534759358289e-05, + "loss": 0.206, + "step": 1845 + }, + { + "epoch": 1.8465193335625645, + "grad_norm": 0.12165025621652603, + "learning_rate": 7.69385026737968e-05, + "loss": 0.2099, + "step": 1846 + }, + { + "epoch": 1.8475196148916884, + "grad_norm": 0.10591330379247665, + "learning_rate": 7.68716577540107e-05, + "loss": 0.2055, + "step": 1847 + }, + { + "epoch": 1.8485198962208123, + "grad_norm": 0.10396409779787064, + "learning_rate": 7.68048128342246e-05, + "loss": 0.2162, + "step": 1848 + }, + { + "epoch": 1.849520177549936, + "grad_norm": 0.10257303714752197, + "learning_rate": 7.673796791443851e-05, + "loss": 0.2029, + "step": 1849 + }, + { + "epoch": 1.8505204588790596, + "grad_norm": 0.1056220754981041, + "learning_rate": 7.667112299465241e-05, + "loss": 0.2154, + "step": 1850 + }, + { + "epoch": 1.8515207402081835, + "grad_norm": 0.1047137901186943, + "learning_rate": 7.660427807486632e-05, + "loss": 0.2083, + "step": 1851 + }, + { + "epoch": 1.8525210215373074, + "grad_norm": 0.12248492240905762, + "learning_rate": 7.653743315508022e-05, + "loss": 0.2234, + "step": 1852 + }, + { + "epoch": 1.8535213028664312, + "grad_norm": 0.10575231164693832, + "learning_rate": 7.647058823529411e-05, + "loss": 0.2092, + "step": 1853 + }, + { + "epoch": 1.8545215841955551, + "grad_norm": 0.10592091083526611, + "learning_rate": 7.640374331550803e-05, + "loss": 0.2054, + "step": 1854 + }, + { + "epoch": 1.8555218655246788, + "grad_norm": 0.10453582555055618, + "learning_rate": 7.633689839572193e-05, + "loss": 0.2049, + "step": 1855 + }, + { + "epoch": 1.8565221468538027, + "grad_norm": 0.1122579425573349, + "learning_rate": 7.627005347593584e-05, + "loss": 0.2038, + "step": 1856 + }, + { + "epoch": 1.8575224281829263, + "grad_norm": 0.10616695880889893, + "learning_rate": 7.620320855614974e-05, + "loss": 0.2038, + "step": 1857 + }, + { + "epoch": 1.8585227095120502, + "grad_norm": 0.1032104343175888, + "learning_rate": 7.613636363636363e-05, + "loss": 0.2252, + "step": 1858 + }, + { + "epoch": 1.8595229908411741, + "grad_norm": 0.0954289585351944, + "learning_rate": 7.606951871657755e-05, + "loss": 0.2115, + "step": 1859 + }, + { + "epoch": 1.860523272170298, + "grad_norm": 0.10593399405479431, + "learning_rate": 7.600267379679144e-05, + "loss": 0.2175, + "step": 1860 + }, + { + "epoch": 1.8615235534994217, + "grad_norm": 0.10119510442018509, + "learning_rate": 7.593582887700536e-05, + "loss": 0.1976, + "step": 1861 + }, + { + "epoch": 1.8625238348285456, + "grad_norm": 0.11174940317869186, + "learning_rate": 7.586898395721925e-05, + "loss": 0.2203, + "step": 1862 + }, + { + "epoch": 1.8635241161576692, + "grad_norm": 0.10503729432821274, + "learning_rate": 7.580213903743317e-05, + "loss": 0.2091, + "step": 1863 + }, + { + "epoch": 1.864524397486793, + "grad_norm": 0.11219252645969391, + "learning_rate": 7.573529411764706e-05, + "loss": 0.2088, + "step": 1864 + }, + { + "epoch": 1.865524678815917, + "grad_norm": 0.10351797938346863, + "learning_rate": 7.566844919786096e-05, + "loss": 0.2004, + "step": 1865 + }, + { + "epoch": 1.8665249601450409, + "grad_norm": 0.10859484225511551, + "learning_rate": 7.560160427807487e-05, + "loss": 0.2037, + "step": 1866 + }, + { + "epoch": 1.8675252414741648, + "grad_norm": 0.11929026246070862, + "learning_rate": 7.553475935828877e-05, + "loss": 0.2116, + "step": 1867 + }, + { + "epoch": 1.8685255228032884, + "grad_norm": 0.11226033419370651, + "learning_rate": 7.546791443850268e-05, + "loss": 0.2177, + "step": 1868 + }, + { + "epoch": 1.869525804132412, + "grad_norm": 0.10874029994010925, + "learning_rate": 7.540106951871658e-05, + "loss": 0.2109, + "step": 1869 + }, + { + "epoch": 1.870526085461536, + "grad_norm": 0.1041000708937645, + "learning_rate": 7.533422459893049e-05, + "loss": 0.2124, + "step": 1870 + }, + { + "epoch": 1.8715263667906599, + "grad_norm": 0.11132880300283432, + "learning_rate": 7.526737967914439e-05, + "loss": 0.2197, + "step": 1871 + }, + { + "epoch": 1.8725266481197838, + "grad_norm": 0.12619589269161224, + "learning_rate": 7.52005347593583e-05, + "loss": 0.2164, + "step": 1872 + }, + { + "epoch": 1.8735269294489076, + "grad_norm": 0.11134762316942215, + "learning_rate": 7.51336898395722e-05, + "loss": 0.2128, + "step": 1873 + }, + { + "epoch": 1.8745272107780313, + "grad_norm": 0.11566010117530823, + "learning_rate": 7.506684491978609e-05, + "loss": 0.2344, + "step": 1874 + }, + { + "epoch": 1.8755274921071552, + "grad_norm": 0.10966410487890244, + "learning_rate": 7.500000000000001e-05, + "loss": 0.2097, + "step": 1875 + }, + { + "epoch": 1.8765277734362789, + "grad_norm": 0.10297410935163498, + "learning_rate": 7.493315508021391e-05, + "loss": 0.2082, + "step": 1876 + }, + { + "epoch": 1.8775280547654027, + "grad_norm": 0.12511096894741058, + "learning_rate": 7.486631016042782e-05, + "loss": 0.25, + "step": 1877 + }, + { + "epoch": 1.8785283360945266, + "grad_norm": 0.10232366621494293, + "learning_rate": 7.479946524064172e-05, + "loss": 0.2001, + "step": 1878 + }, + { + "epoch": 1.8795286174236505, + "grad_norm": 0.10492394864559174, + "learning_rate": 7.473262032085561e-05, + "loss": 0.1947, + "step": 1879 + }, + { + "epoch": 1.8805288987527742, + "grad_norm": 0.09889867156744003, + "learning_rate": 7.466577540106953e-05, + "loss": 0.1974, + "step": 1880 + }, + { + "epoch": 1.881529180081898, + "grad_norm": 0.0997120589017868, + "learning_rate": 7.459893048128342e-05, + "loss": 0.204, + "step": 1881 + }, + { + "epoch": 1.8825294614110217, + "grad_norm": 0.10496684908866882, + "learning_rate": 7.453208556149734e-05, + "loss": 0.2112, + "step": 1882 + }, + { + "epoch": 1.8835297427401456, + "grad_norm": 0.10077778249979019, + "learning_rate": 7.446524064171123e-05, + "loss": 0.2025, + "step": 1883 + }, + { + "epoch": 1.8845300240692695, + "grad_norm": 0.10133393108844757, + "learning_rate": 7.439839572192515e-05, + "loss": 0.2041, + "step": 1884 + }, + { + "epoch": 1.8855303053983934, + "grad_norm": 0.11074411869049072, + "learning_rate": 7.433155080213904e-05, + "loss": 0.2247, + "step": 1885 + }, + { + "epoch": 1.8865305867275173, + "grad_norm": 0.0992373675107956, + "learning_rate": 7.426470588235294e-05, + "loss": 0.1968, + "step": 1886 + }, + { + "epoch": 1.887530868056641, + "grad_norm": 0.10569492727518082, + "learning_rate": 7.419786096256685e-05, + "loss": 0.2213, + "step": 1887 + }, + { + "epoch": 1.8885311493857646, + "grad_norm": 0.10399606823921204, + "learning_rate": 7.413101604278075e-05, + "loss": 0.211, + "step": 1888 + }, + { + "epoch": 1.8895314307148885, + "grad_norm": 0.11129829287528992, + "learning_rate": 7.406417112299465e-05, + "loss": 0.2063, + "step": 1889 + }, + { + "epoch": 1.8905317120440124, + "grad_norm": 0.1060144305229187, + "learning_rate": 7.399732620320856e-05, + "loss": 0.2233, + "step": 1890 + }, + { + "epoch": 1.8915319933731363, + "grad_norm": 0.1103084608912468, + "learning_rate": 7.393048128342246e-05, + "loss": 0.2015, + "step": 1891 + }, + { + "epoch": 1.8925322747022602, + "grad_norm": 0.11073312163352966, + "learning_rate": 7.386363636363637e-05, + "loss": 0.2173, + "step": 1892 + }, + { + "epoch": 1.8935325560313838, + "grad_norm": 0.10715494304895401, + "learning_rate": 7.379679144385027e-05, + "loss": 0.2058, + "step": 1893 + }, + { + "epoch": 1.8945328373605075, + "grad_norm": 0.10859578102827072, + "learning_rate": 7.372994652406418e-05, + "loss": 0.2142, + "step": 1894 + }, + { + "epoch": 1.8955331186896314, + "grad_norm": 0.1130855605006218, + "learning_rate": 7.366310160427807e-05, + "loss": 0.2316, + "step": 1895 + }, + { + "epoch": 1.8965334000187553, + "grad_norm": 0.11098507046699524, + "learning_rate": 7.359625668449199e-05, + "loss": 0.2149, + "step": 1896 + }, + { + "epoch": 1.8975336813478791, + "grad_norm": 0.10511645674705505, + "learning_rate": 7.352941176470589e-05, + "loss": 0.2019, + "step": 1897 + }, + { + "epoch": 1.898533962677003, + "grad_norm": 0.10731218755245209, + "learning_rate": 7.34625668449198e-05, + "loss": 0.2159, + "step": 1898 + }, + { + "epoch": 1.8995342440061267, + "grad_norm": 0.0991966724395752, + "learning_rate": 7.33957219251337e-05, + "loss": 0.2076, + "step": 1899 + }, + { + "epoch": 1.9005345253352506, + "grad_norm": 0.10622479766607285, + "learning_rate": 7.33288770053476e-05, + "loss": 0.208, + "step": 1900 + }, + { + "epoch": 1.9015348066643742, + "grad_norm": 0.1042945384979248, + "learning_rate": 7.326203208556151e-05, + "loss": 0.2118, + "step": 1901 + }, + { + "epoch": 1.9025350879934981, + "grad_norm": 0.10532768815755844, + "learning_rate": 7.31951871657754e-05, + "loss": 0.1905, + "step": 1902 + }, + { + "epoch": 1.903535369322622, + "grad_norm": 0.10074347257614136, + "learning_rate": 7.312834224598932e-05, + "loss": 0.2102, + "step": 1903 + }, + { + "epoch": 1.904535650651746, + "grad_norm": 0.11233386397361755, + "learning_rate": 7.30614973262032e-05, + "loss": 0.2233, + "step": 1904 + }, + { + "epoch": 1.9055359319808696, + "grad_norm": 0.10498382896184921, + "learning_rate": 7.299465240641712e-05, + "loss": 0.2092, + "step": 1905 + }, + { + "epoch": 1.9065362133099935, + "grad_norm": 0.09968066215515137, + "learning_rate": 7.292780748663102e-05, + "loss": 0.2045, + "step": 1906 + }, + { + "epoch": 1.9075364946391171, + "grad_norm": 0.10847284644842148, + "learning_rate": 7.286096256684492e-05, + "loss": 0.2088, + "step": 1907 + }, + { + "epoch": 1.908536775968241, + "grad_norm": 0.09969563782215118, + "learning_rate": 7.279411764705882e-05, + "loss": 0.206, + "step": 1908 + }, + { + "epoch": 1.909537057297365, + "grad_norm": 0.11192680895328522, + "learning_rate": 7.272727272727273e-05, + "loss": 0.2195, + "step": 1909 + }, + { + "epoch": 1.9105373386264888, + "grad_norm": 0.11669693887233734, + "learning_rate": 7.266042780748663e-05, + "loss": 0.2159, + "step": 1910 + }, + { + "epoch": 1.9115376199556127, + "grad_norm": 0.10719669610261917, + "learning_rate": 7.259358288770054e-05, + "loss": 0.2043, + "step": 1911 + }, + { + "epoch": 1.9125379012847363, + "grad_norm": 0.10686182975769043, + "learning_rate": 7.252673796791444e-05, + "loss": 0.2187, + "step": 1912 + }, + { + "epoch": 1.91353818261386, + "grad_norm": 0.10967845469713211, + "learning_rate": 7.245989304812835e-05, + "loss": 0.1958, + "step": 1913 + }, + { + "epoch": 1.9145384639429839, + "grad_norm": 0.11067681014537811, + "learning_rate": 7.239304812834225e-05, + "loss": 0.2117, + "step": 1914 + }, + { + "epoch": 1.9155387452721078, + "grad_norm": 0.09932591021060944, + "learning_rate": 7.232620320855615e-05, + "loss": 0.1922, + "step": 1915 + }, + { + "epoch": 1.9165390266012317, + "grad_norm": 0.10270752757787704, + "learning_rate": 7.225935828877005e-05, + "loss": 0.2055, + "step": 1916 + }, + { + "epoch": 1.9175393079303555, + "grad_norm": 0.10826575756072998, + "learning_rate": 7.219251336898396e-05, + "loss": 0.2041, + "step": 1917 + }, + { + "epoch": 1.9185395892594792, + "grad_norm": 0.11257299035787582, + "learning_rate": 7.212566844919787e-05, + "loss": 0.2011, + "step": 1918 + }, + { + "epoch": 1.919539870588603, + "grad_norm": 0.11007867753505707, + "learning_rate": 7.205882352941177e-05, + "loss": 0.2082, + "step": 1919 + }, + { + "epoch": 1.9205401519177268, + "grad_norm": 0.11282702535390854, + "learning_rate": 7.199197860962568e-05, + "loss": 0.2172, + "step": 1920 + }, + { + "epoch": 1.9215404332468506, + "grad_norm": 0.10107994824647903, + "learning_rate": 7.192513368983958e-05, + "loss": 0.2131, + "step": 1921 + }, + { + "epoch": 1.9225407145759745, + "grad_norm": 0.1053793728351593, + "learning_rate": 7.185828877005349e-05, + "loss": 0.214, + "step": 1922 + }, + { + "epoch": 1.9235409959050984, + "grad_norm": 0.11204124242067337, + "learning_rate": 7.179144385026738e-05, + "loss": 0.2008, + "step": 1923 + }, + { + "epoch": 1.924541277234222, + "grad_norm": 0.1145598441362381, + "learning_rate": 7.17245989304813e-05, + "loss": 0.2105, + "step": 1924 + }, + { + "epoch": 1.925541558563346, + "grad_norm": 0.11473219841718674, + "learning_rate": 7.165775401069518e-05, + "loss": 0.2131, + "step": 1925 + }, + { + "epoch": 1.9265418398924696, + "grad_norm": 0.10151656717061996, + "learning_rate": 7.15909090909091e-05, + "loss": 0.1961, + "step": 1926 + }, + { + "epoch": 1.9275421212215935, + "grad_norm": 0.11718515306711197, + "learning_rate": 7.1524064171123e-05, + "loss": 0.2119, + "step": 1927 + }, + { + "epoch": 1.9285424025507174, + "grad_norm": 0.11715052276849747, + "learning_rate": 7.14572192513369e-05, + "loss": 0.2163, + "step": 1928 + }, + { + "epoch": 1.9295426838798413, + "grad_norm": 0.10801360011100769, + "learning_rate": 7.13903743315508e-05, + "loss": 0.2058, + "step": 1929 + }, + { + "epoch": 1.9305429652089652, + "grad_norm": 0.10730385035276413, + "learning_rate": 7.13235294117647e-05, + "loss": 0.2169, + "step": 1930 + }, + { + "epoch": 1.9315432465380888, + "grad_norm": 0.11009212583303452, + "learning_rate": 7.125668449197861e-05, + "loss": 0.2032, + "step": 1931 + }, + { + "epoch": 1.9325435278672125, + "grad_norm": 0.10993629693984985, + "learning_rate": 7.118983957219252e-05, + "loss": 0.2092, + "step": 1932 + }, + { + "epoch": 1.9335438091963364, + "grad_norm": 0.11734863370656967, + "learning_rate": 7.112299465240642e-05, + "loss": 0.2147, + "step": 1933 + }, + { + "epoch": 1.9345440905254603, + "grad_norm": 0.10309990495443344, + "learning_rate": 7.105614973262032e-05, + "loss": 0.2133, + "step": 1934 + }, + { + "epoch": 1.9355443718545842, + "grad_norm": 0.10092102736234665, + "learning_rate": 7.098930481283423e-05, + "loss": 0.207, + "step": 1935 + }, + { + "epoch": 1.936544653183708, + "grad_norm": 0.09628443419933319, + "learning_rate": 7.092245989304813e-05, + "loss": 0.1949, + "step": 1936 + }, + { + "epoch": 1.9375449345128317, + "grad_norm": 0.1127362996339798, + "learning_rate": 7.085561497326202e-05, + "loss": 0.205, + "step": 1937 + }, + { + "epoch": 1.9385452158419556, + "grad_norm": 0.10608649998903275, + "learning_rate": 7.078877005347594e-05, + "loss": 0.2131, + "step": 1938 + }, + { + "epoch": 1.9395454971710793, + "grad_norm": 0.10919354856014252, + "learning_rate": 7.072192513368985e-05, + "loss": 0.207, + "step": 1939 + }, + { + "epoch": 1.9405457785002032, + "grad_norm": 0.10192367434501648, + "learning_rate": 7.065508021390375e-05, + "loss": 0.2001, + "step": 1940 + }, + { + "epoch": 1.941546059829327, + "grad_norm": 0.10631133615970612, + "learning_rate": 7.058823529411765e-05, + "loss": 0.2024, + "step": 1941 + }, + { + "epoch": 1.942546341158451, + "grad_norm": 0.10917597264051437, + "learning_rate": 7.052139037433156e-05, + "loss": 0.1957, + "step": 1942 + }, + { + "epoch": 1.9435466224875746, + "grad_norm": 0.10295380651950836, + "learning_rate": 7.045454545454546e-05, + "loss": 0.2057, + "step": 1943 + }, + { + "epoch": 1.9445469038166985, + "grad_norm": 0.10071974992752075, + "learning_rate": 7.038770053475935e-05, + "loss": 0.2022, + "step": 1944 + }, + { + "epoch": 1.9455471851458221, + "grad_norm": 0.09994477778673172, + "learning_rate": 7.032085561497327e-05, + "loss": 0.201, + "step": 1945 + }, + { + "epoch": 1.946547466474946, + "grad_norm": 0.11314476281404495, + "learning_rate": 7.025401069518716e-05, + "loss": 0.215, + "step": 1946 + }, + { + "epoch": 1.94754774780407, + "grad_norm": 0.10418917238712311, + "learning_rate": 7.018716577540108e-05, + "loss": 0.2008, + "step": 1947 + }, + { + "epoch": 1.9485480291331938, + "grad_norm": 0.09869718551635742, + "learning_rate": 7.012032085561497e-05, + "loss": 0.1975, + "step": 1948 + }, + { + "epoch": 1.9495483104623177, + "grad_norm": 0.10971954464912415, + "learning_rate": 7.005347593582889e-05, + "loss": 0.2186, + "step": 1949 + }, + { + "epoch": 1.9505485917914414, + "grad_norm": 0.1083192229270935, + "learning_rate": 6.998663101604278e-05, + "loss": 0.2091, + "step": 1950 + }, + { + "epoch": 1.951548873120565, + "grad_norm": 0.09883376955986023, + "learning_rate": 6.991978609625668e-05, + "loss": 0.2063, + "step": 1951 + }, + { + "epoch": 1.952549154449689, + "grad_norm": 0.1221231073141098, + "learning_rate": 6.985294117647059e-05, + "loss": 0.2084, + "step": 1952 + }, + { + "epoch": 1.9535494357788128, + "grad_norm": 0.10982025414705276, + "learning_rate": 6.97860962566845e-05, + "loss": 0.208, + "step": 1953 + }, + { + "epoch": 1.9545497171079367, + "grad_norm": 0.1132119670510292, + "learning_rate": 6.97192513368984e-05, + "loss": 0.2062, + "step": 1954 + }, + { + "epoch": 1.9555499984370606, + "grad_norm": 0.11278846859931946, + "learning_rate": 6.96524064171123e-05, + "loss": 0.2064, + "step": 1955 + }, + { + "epoch": 1.9565502797661842, + "grad_norm": 0.11189232766628265, + "learning_rate": 6.95855614973262e-05, + "loss": 0.2313, + "step": 1956 + }, + { + "epoch": 1.957550561095308, + "grad_norm": 0.11401370912790298, + "learning_rate": 6.951871657754011e-05, + "loss": 0.2189, + "step": 1957 + }, + { + "epoch": 1.9585508424244318, + "grad_norm": 0.10936152935028076, + "learning_rate": 6.945187165775402e-05, + "loss": 0.2042, + "step": 1958 + }, + { + "epoch": 1.9595511237535557, + "grad_norm": 0.10642879456281662, + "learning_rate": 6.938502673796792e-05, + "loss": 0.2052, + "step": 1959 + }, + { + "epoch": 1.9605514050826796, + "grad_norm": 0.10062849521636963, + "learning_rate": 6.931818181818182e-05, + "loss": 0.2164, + "step": 1960 + }, + { + "epoch": 1.9615516864118034, + "grad_norm": 0.12028147280216217, + "learning_rate": 6.925133689839573e-05, + "loss": 0.2212, + "step": 1961 + }, + { + "epoch": 1.962551967740927, + "grad_norm": 0.1064947247505188, + "learning_rate": 6.918449197860963e-05, + "loss": 0.1968, + "step": 1962 + }, + { + "epoch": 1.963552249070051, + "grad_norm": 0.11135193705558777, + "learning_rate": 6.911764705882354e-05, + "loss": 0.2034, + "step": 1963 + }, + { + "epoch": 1.9645525303991747, + "grad_norm": 0.10464580357074738, + "learning_rate": 6.905080213903744e-05, + "loss": 0.2038, + "step": 1964 + }, + { + "epoch": 1.9655528117282985, + "grad_norm": 0.10277816653251648, + "learning_rate": 6.898395721925133e-05, + "loss": 0.2083, + "step": 1965 + }, + { + "epoch": 1.9665530930574224, + "grad_norm": 0.10462097823619843, + "learning_rate": 6.891711229946525e-05, + "loss": 0.212, + "step": 1966 + }, + { + "epoch": 1.9675533743865463, + "grad_norm": 0.1041736975312233, + "learning_rate": 6.885026737967914e-05, + "loss": 0.2171, + "step": 1967 + }, + { + "epoch": 1.96855365571567, + "grad_norm": 0.09999759495258331, + "learning_rate": 6.878342245989306e-05, + "loss": 0.2085, + "step": 1968 + }, + { + "epoch": 1.9695539370447939, + "grad_norm": 0.10501585155725479, + "learning_rate": 6.871657754010695e-05, + "loss": 0.205, + "step": 1969 + }, + { + "epoch": 1.9705542183739175, + "grad_norm": 0.11044589430093765, + "learning_rate": 6.864973262032087e-05, + "loss": 0.2238, + "step": 1970 + }, + { + "epoch": 1.9715544997030414, + "grad_norm": 0.10147418081760406, + "learning_rate": 6.858288770053476e-05, + "loss": 0.2086, + "step": 1971 + }, + { + "epoch": 1.9725547810321653, + "grad_norm": 0.10847562551498413, + "learning_rate": 6.851604278074866e-05, + "loss": 0.208, + "step": 1972 + }, + { + "epoch": 1.9735550623612892, + "grad_norm": 0.1056070625782013, + "learning_rate": 6.844919786096257e-05, + "loss": 0.1978, + "step": 1973 + }, + { + "epoch": 1.974555343690413, + "grad_norm": 0.0980309471487999, + "learning_rate": 6.838235294117647e-05, + "loss": 0.1924, + "step": 1974 + }, + { + "epoch": 1.9755556250195367, + "grad_norm": 0.11507856100797653, + "learning_rate": 6.831550802139038e-05, + "loss": 0.2157, + "step": 1975 + }, + { + "epoch": 1.9765559063486604, + "grad_norm": 0.10719770938158035, + "learning_rate": 6.824866310160428e-05, + "loss": 0.2037, + "step": 1976 + }, + { + "epoch": 1.9775561876777843, + "grad_norm": 0.10877916216850281, + "learning_rate": 6.818181818181818e-05, + "loss": 0.2128, + "step": 1977 + }, + { + "epoch": 1.9785564690069082, + "grad_norm": 0.10973481088876724, + "learning_rate": 6.811497326203209e-05, + "loss": 0.2176, + "step": 1978 + }, + { + "epoch": 1.979556750336032, + "grad_norm": 0.10916756838560104, + "learning_rate": 6.804812834224599e-05, + "loss": 0.2078, + "step": 1979 + }, + { + "epoch": 1.980557031665156, + "grad_norm": 0.10991036891937256, + "learning_rate": 6.79812834224599e-05, + "loss": 0.2035, + "step": 1980 + }, + { + "epoch": 1.9815573129942796, + "grad_norm": 0.10958372056484222, + "learning_rate": 6.79144385026738e-05, + "loss": 0.2102, + "step": 1981 + }, + { + "epoch": 1.9825575943234035, + "grad_norm": 0.11009581387042999, + "learning_rate": 6.78475935828877e-05, + "loss": 0.2074, + "step": 1982 + }, + { + "epoch": 1.9835578756525272, + "grad_norm": 0.10819099098443985, + "learning_rate": 6.778074866310161e-05, + "loss": 0.2147, + "step": 1983 + }, + { + "epoch": 1.984558156981651, + "grad_norm": 0.10459486395120621, + "learning_rate": 6.771390374331552e-05, + "loss": 0.2085, + "step": 1984 + }, + { + "epoch": 1.985558438310775, + "grad_norm": 0.10157530009746552, + "learning_rate": 6.764705882352942e-05, + "loss": 0.213, + "step": 1985 + }, + { + "epoch": 1.9865587196398988, + "grad_norm": 0.10728179663419724, + "learning_rate": 6.758021390374331e-05, + "loss": 0.2288, + "step": 1986 + }, + { + "epoch": 1.9875590009690225, + "grad_norm": 0.11445151269435883, + "learning_rate": 6.751336898395723e-05, + "loss": 0.2079, + "step": 1987 + }, + { + "epoch": 1.9885592822981464, + "grad_norm": 0.11336226761341095, + "learning_rate": 6.744652406417112e-05, + "loss": 0.2247, + "step": 1988 + }, + { + "epoch": 1.98955956362727, + "grad_norm": 0.10331938415765762, + "learning_rate": 6.737967914438504e-05, + "loss": 0.2271, + "step": 1989 + }, + { + "epoch": 1.990559844956394, + "grad_norm": 0.09976281970739365, + "learning_rate": 6.731283422459893e-05, + "loss": 0.2077, + "step": 1990 + }, + { + "epoch": 1.9915601262855178, + "grad_norm": 0.10254520922899246, + "learning_rate": 6.724598930481285e-05, + "loss": 0.2204, + "step": 1991 + }, + { + "epoch": 1.9925604076146417, + "grad_norm": 0.10867884010076523, + "learning_rate": 6.717914438502674e-05, + "loss": 0.2119, + "step": 1992 + }, + { + "epoch": 1.9935606889437656, + "grad_norm": 0.10568103194236755, + "learning_rate": 6.711229946524064e-05, + "loss": 0.2234, + "step": 1993 + }, + { + "epoch": 1.9945609702728893, + "grad_norm": 0.09934551268815994, + "learning_rate": 6.704545454545455e-05, + "loss": 0.2033, + "step": 1994 + }, + { + "epoch": 1.995561251602013, + "grad_norm": 0.0999583750963211, + "learning_rate": 6.697860962566845e-05, + "loss": 0.2074, + "step": 1995 + }, + { + "epoch": 1.9965615329311368, + "grad_norm": 0.11354430764913559, + "learning_rate": 6.691176470588235e-05, + "loss": 0.2307, + "step": 1996 + }, + { + "epoch": 1.9975618142602607, + "grad_norm": 0.09576801210641861, + "learning_rate": 6.684491978609626e-05, + "loss": 0.2015, + "step": 1997 + }, + { + "epoch": 1.9985620955893846, + "grad_norm": 0.1140224039554596, + "learning_rate": 6.677807486631016e-05, + "loss": 0.211, + "step": 1998 + }, + { + "epoch": 1.9995623769185085, + "grad_norm": 0.09992174804210663, + "learning_rate": 6.671122994652407e-05, + "loss": 0.2037, + "step": 1999 + }, + { + "epoch": 2.0005626582476324, + "grad_norm": 0.09511711448431015, + "learning_rate": 6.664438502673797e-05, + "loss": 0.1938, + "step": 2000 + }, + { + "epoch": 2.001562939576756, + "grad_norm": 0.09209245443344116, + "learning_rate": 6.657754010695188e-05, + "loss": 0.177, + "step": 2001 + }, + { + "epoch": 2.0025632209058797, + "grad_norm": 0.09760528802871704, + "learning_rate": 6.651069518716578e-05, + "loss": 0.1916, + "step": 2002 + }, + { + "epoch": 2.0035635022350036, + "grad_norm": 0.0948125422000885, + "learning_rate": 6.644385026737968e-05, + "loss": 0.1817, + "step": 2003 + }, + { + "epoch": 2.0045637835641275, + "grad_norm": 0.09869059175252914, + "learning_rate": 6.637700534759359e-05, + "loss": 0.1978, + "step": 2004 + }, + { + "epoch": 2.0055640648932513, + "grad_norm": 0.10199198871850967, + "learning_rate": 6.631016042780749e-05, + "loss": 0.1826, + "step": 2005 + }, + { + "epoch": 2.0065643462223752, + "grad_norm": 0.10061972588300705, + "learning_rate": 6.62433155080214e-05, + "loss": 0.2004, + "step": 2006 + }, + { + "epoch": 2.0075646275514987, + "grad_norm": 0.10280162841081619, + "learning_rate": 6.61764705882353e-05, + "loss": 0.19, + "step": 2007 + }, + { + "epoch": 2.0085649088806226, + "grad_norm": 0.10061551630496979, + "learning_rate": 6.61096256684492e-05, + "loss": 0.1881, + "step": 2008 + }, + { + "epoch": 2.0095651902097464, + "grad_norm": 0.10982374846935272, + "learning_rate": 6.60427807486631e-05, + "loss": 0.1863, + "step": 2009 + }, + { + "epoch": 2.0105654715388703, + "grad_norm": 0.10312630981206894, + "learning_rate": 6.597593582887701e-05, + "loss": 0.1855, + "step": 2010 + }, + { + "epoch": 2.011565752867994, + "grad_norm": 0.10569493472576141, + "learning_rate": 6.59090909090909e-05, + "loss": 0.1985, + "step": 2011 + }, + { + "epoch": 2.012566034197118, + "grad_norm": 0.10891608148813248, + "learning_rate": 6.584224598930482e-05, + "loss": 0.1861, + "step": 2012 + }, + { + "epoch": 2.0135663155262415, + "grad_norm": 0.11321307718753815, + "learning_rate": 6.577540106951871e-05, + "loss": 0.2018, + "step": 2013 + }, + { + "epoch": 2.0145665968553654, + "grad_norm": 0.11450721323490143, + "learning_rate": 6.570855614973262e-05, + "loss": 0.1928, + "step": 2014 + }, + { + "epoch": 2.0155668781844893, + "grad_norm": 0.10977186262607574, + "learning_rate": 6.564171122994652e-05, + "loss": 0.1749, + "step": 2015 + }, + { + "epoch": 2.016567159513613, + "grad_norm": 0.10844075679779053, + "learning_rate": 6.557486631016043e-05, + "loss": 0.1923, + "step": 2016 + }, + { + "epoch": 2.017567440842737, + "grad_norm": 0.10274696350097656, + "learning_rate": 6.550802139037433e-05, + "loss": 0.1833, + "step": 2017 + }, + { + "epoch": 2.018567722171861, + "grad_norm": 0.11026908457279205, + "learning_rate": 6.544117647058824e-05, + "loss": 0.1964, + "step": 2018 + }, + { + "epoch": 2.019568003500985, + "grad_norm": 0.10636306554079056, + "learning_rate": 6.537433155080214e-05, + "loss": 0.1944, + "step": 2019 + }, + { + "epoch": 2.0205682848301083, + "grad_norm": 0.1100206971168518, + "learning_rate": 6.530748663101605e-05, + "loss": 0.1838, + "step": 2020 + }, + { + "epoch": 2.021568566159232, + "grad_norm": 0.10723818838596344, + "learning_rate": 6.524064171122995e-05, + "loss": 0.1797, + "step": 2021 + }, + { + "epoch": 2.022568847488356, + "grad_norm": 0.11446322500705719, + "learning_rate": 6.517379679144385e-05, + "loss": 0.204, + "step": 2022 + }, + { + "epoch": 2.02356912881748, + "grad_norm": 0.10304328054189682, + "learning_rate": 6.510695187165776e-05, + "loss": 0.1947, + "step": 2023 + }, + { + "epoch": 2.024569410146604, + "grad_norm": 0.10630247741937637, + "learning_rate": 6.504010695187166e-05, + "loss": 0.1808, + "step": 2024 + }, + { + "epoch": 2.0255696914757277, + "grad_norm": 0.10085494816303253, + "learning_rate": 6.497326203208557e-05, + "loss": 0.1847, + "step": 2025 + }, + { + "epoch": 2.026569972804851, + "grad_norm": 0.10419415682554245, + "learning_rate": 6.490641711229947e-05, + "loss": 0.1863, + "step": 2026 + }, + { + "epoch": 2.027570254133975, + "grad_norm": 0.10746020823717117, + "learning_rate": 6.483957219251338e-05, + "loss": 0.1871, + "step": 2027 + }, + { + "epoch": 2.028570535463099, + "grad_norm": 0.10058224201202393, + "learning_rate": 6.477272727272728e-05, + "loss": 0.1801, + "step": 2028 + }, + { + "epoch": 2.029570816792223, + "grad_norm": 0.1132676899433136, + "learning_rate": 6.470588235294118e-05, + "loss": 0.1954, + "step": 2029 + }, + { + "epoch": 2.0305710981213467, + "grad_norm": 0.10064389556646347, + "learning_rate": 6.463903743315508e-05, + "loss": 0.1871, + "step": 2030 + }, + { + "epoch": 2.0315713794504706, + "grad_norm": 0.10540352761745453, + "learning_rate": 6.457219251336899e-05, + "loss": 0.1852, + "step": 2031 + }, + { + "epoch": 2.032571660779594, + "grad_norm": 0.10502824932336807, + "learning_rate": 6.450534759358288e-05, + "loss": 0.1779, + "step": 2032 + }, + { + "epoch": 2.033571942108718, + "grad_norm": 0.11056734621524811, + "learning_rate": 6.44385026737968e-05, + "loss": 0.1889, + "step": 2033 + }, + { + "epoch": 2.034572223437842, + "grad_norm": 0.10051354020833969, + "learning_rate": 6.437165775401069e-05, + "loss": 0.1914, + "step": 2034 + }, + { + "epoch": 2.0355725047669657, + "grad_norm": 0.11463151127099991, + "learning_rate": 6.43048128342246e-05, + "loss": 0.1918, + "step": 2035 + }, + { + "epoch": 2.0365727860960896, + "grad_norm": 0.11973311752080917, + "learning_rate": 6.42379679144385e-05, + "loss": 0.1893, + "step": 2036 + }, + { + "epoch": 2.0375730674252135, + "grad_norm": 0.11207733303308487, + "learning_rate": 6.41711229946524e-05, + "loss": 0.1889, + "step": 2037 + }, + { + "epoch": 2.038573348754337, + "grad_norm": 0.10689655691385269, + "learning_rate": 6.410427807486631e-05, + "loss": 0.1927, + "step": 2038 + }, + { + "epoch": 2.039573630083461, + "grad_norm": 0.10392475873231888, + "learning_rate": 6.403743315508021e-05, + "loss": 0.201, + "step": 2039 + }, + { + "epoch": 2.0405739114125847, + "grad_norm": 0.10292202979326248, + "learning_rate": 6.397058823529412e-05, + "loss": 0.1841, + "step": 2040 + }, + { + "epoch": 2.0415741927417086, + "grad_norm": 0.10683207958936691, + "learning_rate": 6.390374331550802e-05, + "loss": 0.1944, + "step": 2041 + }, + { + "epoch": 2.0425744740708325, + "grad_norm": 0.11113958805799484, + "learning_rate": 6.383689839572193e-05, + "loss": 0.1793, + "step": 2042 + }, + { + "epoch": 2.0435747553999564, + "grad_norm": 0.10987640172243118, + "learning_rate": 6.377005347593583e-05, + "loss": 0.192, + "step": 2043 + }, + { + "epoch": 2.0445750367290803, + "grad_norm": 0.11180393397808075, + "learning_rate": 6.370320855614974e-05, + "loss": 0.1861, + "step": 2044 + }, + { + "epoch": 2.0455753180582037, + "grad_norm": 0.10225307941436768, + "learning_rate": 6.363636363636364e-05, + "loss": 0.1785, + "step": 2045 + }, + { + "epoch": 2.0465755993873276, + "grad_norm": 0.10819721221923828, + "learning_rate": 6.356951871657755e-05, + "loss": 0.1862, + "step": 2046 + }, + { + "epoch": 2.0475758807164515, + "grad_norm": 0.11129755526781082, + "learning_rate": 6.350267379679145e-05, + "loss": 0.1842, + "step": 2047 + }, + { + "epoch": 2.0485761620455754, + "grad_norm": 0.10544439405202866, + "learning_rate": 6.343582887700535e-05, + "loss": 0.1829, + "step": 2048 + }, + { + "epoch": 2.0495764433746992, + "grad_norm": 0.11544738709926605, + "learning_rate": 6.336898395721926e-05, + "loss": 0.1829, + "step": 2049 + }, + { + "epoch": 2.050576724703823, + "grad_norm": 0.1023552343249321, + "learning_rate": 6.330213903743316e-05, + "loss": 0.1811, + "step": 2050 + }, + { + "epoch": 2.0515770060329466, + "grad_norm": 0.10466215759515762, + "learning_rate": 6.323529411764705e-05, + "loss": 0.1926, + "step": 2051 + }, + { + "epoch": 2.0525772873620705, + "grad_norm": 0.10610965639352798, + "learning_rate": 6.316844919786097e-05, + "loss": 0.1846, + "step": 2052 + }, + { + "epoch": 2.0535775686911943, + "grad_norm": 0.10881342738866806, + "learning_rate": 6.310160427807486e-05, + "loss": 0.1876, + "step": 2053 + }, + { + "epoch": 2.0545778500203182, + "grad_norm": 0.11460968852043152, + "learning_rate": 6.303475935828878e-05, + "loss": 0.2016, + "step": 2054 + }, + { + "epoch": 2.055578131349442, + "grad_norm": 0.11115551739931107, + "learning_rate": 6.296791443850267e-05, + "loss": 0.188, + "step": 2055 + }, + { + "epoch": 2.056578412678566, + "grad_norm": 0.11366444826126099, + "learning_rate": 6.290106951871659e-05, + "loss": 0.1999, + "step": 2056 + }, + { + "epoch": 2.0575786940076894, + "grad_norm": 0.11990818381309509, + "learning_rate": 6.283422459893048e-05, + "loss": 0.2054, + "step": 2057 + }, + { + "epoch": 2.0585789753368133, + "grad_norm": 0.1032952293753624, + "learning_rate": 6.276737967914438e-05, + "loss": 0.1859, + "step": 2058 + }, + { + "epoch": 2.059579256665937, + "grad_norm": 0.10626443475484848, + "learning_rate": 6.270053475935829e-05, + "loss": 0.1833, + "step": 2059 + }, + { + "epoch": 2.060579537995061, + "grad_norm": 0.1057368740439415, + "learning_rate": 6.263368983957219e-05, + "loss": 0.1903, + "step": 2060 + }, + { + "epoch": 2.061579819324185, + "grad_norm": 0.11752955615520477, + "learning_rate": 6.25668449197861e-05, + "loss": 0.1895, + "step": 2061 + }, + { + "epoch": 2.062580100653309, + "grad_norm": 0.11860334873199463, + "learning_rate": 6.25e-05, + "loss": 0.1857, + "step": 2062 + }, + { + "epoch": 2.0635803819824328, + "grad_norm": 0.11060094833374023, + "learning_rate": 6.24331550802139e-05, + "loss": 0.1842, + "step": 2063 + }, + { + "epoch": 2.064580663311556, + "grad_norm": 0.11680097877979279, + "learning_rate": 6.236631016042781e-05, + "loss": 0.1936, + "step": 2064 + }, + { + "epoch": 2.06558094464068, + "grad_norm": 0.11144144833087921, + "learning_rate": 6.229946524064171e-05, + "loss": 0.1877, + "step": 2065 + }, + { + "epoch": 2.066581225969804, + "grad_norm": 0.10163434594869614, + "learning_rate": 6.223262032085562e-05, + "loss": 0.1808, + "step": 2066 + }, + { + "epoch": 2.067581507298928, + "grad_norm": 0.10844971984624863, + "learning_rate": 6.216577540106952e-05, + "loss": 0.1886, + "step": 2067 + }, + { + "epoch": 2.0685817886280518, + "grad_norm": 0.10463793575763702, + "learning_rate": 6.209893048128343e-05, + "loss": 0.1792, + "step": 2068 + }, + { + "epoch": 2.0695820699571756, + "grad_norm": 0.10424848645925522, + "learning_rate": 6.203208556149733e-05, + "loss": 0.1818, + "step": 2069 + }, + { + "epoch": 2.070582351286299, + "grad_norm": 0.09933369606733322, + "learning_rate": 6.196524064171124e-05, + "loss": 0.187, + "step": 2070 + }, + { + "epoch": 2.071582632615423, + "grad_norm": 0.1079430803656578, + "learning_rate": 6.189839572192514e-05, + "loss": 0.1945, + "step": 2071 + }, + { + "epoch": 2.072582913944547, + "grad_norm": 0.11242985725402832, + "learning_rate": 6.183155080213903e-05, + "loss": 0.1911, + "step": 2072 + }, + { + "epoch": 2.0735831952736707, + "grad_norm": 0.11914853006601334, + "learning_rate": 6.176470588235295e-05, + "loss": 0.2125, + "step": 2073 + }, + { + "epoch": 2.0745834766027946, + "grad_norm": 0.11119861155748367, + "learning_rate": 6.169786096256684e-05, + "loss": 0.2029, + "step": 2074 + }, + { + "epoch": 2.0755837579319185, + "grad_norm": 0.1044730693101883, + "learning_rate": 6.163101604278076e-05, + "loss": 0.1875, + "step": 2075 + }, + { + "epoch": 2.076584039261042, + "grad_norm": 0.10686076432466507, + "learning_rate": 6.156417112299465e-05, + "loss": 0.1816, + "step": 2076 + }, + { + "epoch": 2.077584320590166, + "grad_norm": 0.11088142544031143, + "learning_rate": 6.149732620320857e-05, + "loss": 0.1892, + "step": 2077 + }, + { + "epoch": 2.0785846019192897, + "grad_norm": 0.10599001497030258, + "learning_rate": 6.143048128342246e-05, + "loss": 0.1799, + "step": 2078 + }, + { + "epoch": 2.0795848832484136, + "grad_norm": 0.11646291613578796, + "learning_rate": 6.136363636363636e-05, + "loss": 0.192, + "step": 2079 + }, + { + "epoch": 2.0805851645775375, + "grad_norm": 0.10922300070524216, + "learning_rate": 6.129679144385027e-05, + "loss": 0.1926, + "step": 2080 + }, + { + "epoch": 2.0815854459066614, + "grad_norm": 0.1162891536951065, + "learning_rate": 6.122994652406417e-05, + "loss": 0.1906, + "step": 2081 + }, + { + "epoch": 2.0825857272357853, + "grad_norm": 0.11783435940742493, + "learning_rate": 6.116310160427808e-05, + "loss": 0.1918, + "step": 2082 + }, + { + "epoch": 2.0835860085649087, + "grad_norm": 0.11803604662418365, + "learning_rate": 6.109625668449198e-05, + "loss": 0.1859, + "step": 2083 + }, + { + "epoch": 2.0845862898940326, + "grad_norm": 0.11169655621051788, + "learning_rate": 6.102941176470589e-05, + "loss": 0.1892, + "step": 2084 + }, + { + "epoch": 2.0855865712231565, + "grad_norm": 0.11597800999879837, + "learning_rate": 6.096256684491979e-05, + "loss": 0.1812, + "step": 2085 + }, + { + "epoch": 2.0865868525522804, + "grad_norm": 0.10951055586338043, + "learning_rate": 6.08957219251337e-05, + "loss": 0.1681, + "step": 2086 + }, + { + "epoch": 2.0875871338814043, + "grad_norm": 0.11477774381637573, + "learning_rate": 6.08288770053476e-05, + "loss": 0.1798, + "step": 2087 + }, + { + "epoch": 2.088587415210528, + "grad_norm": 0.11752168089151382, + "learning_rate": 6.07620320855615e-05, + "loss": 0.1842, + "step": 2088 + }, + { + "epoch": 2.0895876965396516, + "grad_norm": 0.10977164655923843, + "learning_rate": 6.0695187165775406e-05, + "loss": 0.184, + "step": 2089 + }, + { + "epoch": 2.0905879778687755, + "grad_norm": 0.1119425892829895, + "learning_rate": 6.062834224598931e-05, + "loss": 0.1841, + "step": 2090 + }, + { + "epoch": 2.0915882591978994, + "grad_norm": 0.11903315037488937, + "learning_rate": 6.056149732620321e-05, + "loss": 0.1905, + "step": 2091 + }, + { + "epoch": 2.0925885405270233, + "grad_norm": 0.12525925040245056, + "learning_rate": 6.049465240641712e-05, + "loss": 0.1948, + "step": 2092 + }, + { + "epoch": 2.093588821856147, + "grad_norm": 0.11736462265253067, + "learning_rate": 6.0427807486631016e-05, + "loss": 0.1937, + "step": 2093 + }, + { + "epoch": 2.094589103185271, + "grad_norm": 0.11103154718875885, + "learning_rate": 6.036096256684493e-05, + "loss": 0.182, + "step": 2094 + }, + { + "epoch": 2.0955893845143945, + "grad_norm": 0.11489589512348175, + "learning_rate": 6.0294117647058825e-05, + "loss": 0.1843, + "step": 2095 + }, + { + "epoch": 2.0965896658435184, + "grad_norm": 0.11018618196249008, + "learning_rate": 6.022727272727273e-05, + "loss": 0.1814, + "step": 2096 + }, + { + "epoch": 2.0975899471726422, + "grad_norm": 0.11812925338745117, + "learning_rate": 6.016042780748663e-05, + "loss": 0.1942, + "step": 2097 + }, + { + "epoch": 2.098590228501766, + "grad_norm": 0.11074060201644897, + "learning_rate": 6.009358288770054e-05, + "loss": 0.1794, + "step": 2098 + }, + { + "epoch": 2.09959050983089, + "grad_norm": 0.11756374686956406, + "learning_rate": 6.0026737967914436e-05, + "loss": 0.1797, + "step": 2099 + }, + { + "epoch": 2.100590791160014, + "grad_norm": 0.1115092858672142, + "learning_rate": 5.995989304812835e-05, + "loss": 0.192, + "step": 2100 + }, + { + "epoch": 2.1015910724891373, + "grad_norm": 0.11741135269403458, + "learning_rate": 5.9893048128342244e-05, + "loss": 0.1945, + "step": 2101 + }, + { + "epoch": 2.1025913538182612, + "grad_norm": 0.12050450593233109, + "learning_rate": 5.9826203208556156e-05, + "loss": 0.2025, + "step": 2102 + }, + { + "epoch": 2.103591635147385, + "grad_norm": 0.11961017549037933, + "learning_rate": 5.975935828877005e-05, + "loss": 0.1928, + "step": 2103 + }, + { + "epoch": 2.104591916476509, + "grad_norm": 0.12092681229114532, + "learning_rate": 5.969251336898396e-05, + "loss": 0.1999, + "step": 2104 + }, + { + "epoch": 2.105592197805633, + "grad_norm": 0.112901471555233, + "learning_rate": 5.962566844919787e-05, + "loss": 0.1997, + "step": 2105 + }, + { + "epoch": 2.106592479134757, + "grad_norm": 0.1087542474269867, + "learning_rate": 5.9558823529411766e-05, + "loss": 0.1799, + "step": 2106 + }, + { + "epoch": 2.1075927604638807, + "grad_norm": 0.11269745975732803, + "learning_rate": 5.949197860962568e-05, + "loss": 0.1892, + "step": 2107 + }, + { + "epoch": 2.108593041793004, + "grad_norm": 0.1053754985332489, + "learning_rate": 5.9425133689839575e-05, + "loss": 0.1751, + "step": 2108 + }, + { + "epoch": 2.109593323122128, + "grad_norm": 0.11903897672891617, + "learning_rate": 5.9358288770053486e-05, + "loss": 0.1909, + "step": 2109 + }, + { + "epoch": 2.110593604451252, + "grad_norm": 0.110353022813797, + "learning_rate": 5.9291443850267384e-05, + "loss": 0.1907, + "step": 2110 + }, + { + "epoch": 2.1115938857803758, + "grad_norm": 0.10226050764322281, + "learning_rate": 5.922459893048129e-05, + "loss": 0.1822, + "step": 2111 + }, + { + "epoch": 2.1125941671094997, + "grad_norm": 0.12173362076282501, + "learning_rate": 5.9157754010695186e-05, + "loss": 0.2016, + "step": 2112 + }, + { + "epoch": 2.1135944484386235, + "grad_norm": 0.10548361390829086, + "learning_rate": 5.90909090909091e-05, + "loss": 0.1835, + "step": 2113 + }, + { + "epoch": 2.114594729767747, + "grad_norm": 0.11204254627227783, + "learning_rate": 5.9024064171122994e-05, + "loss": 0.1809, + "step": 2114 + }, + { + "epoch": 2.115595011096871, + "grad_norm": 0.12098168581724167, + "learning_rate": 5.8957219251336906e-05, + "loss": 0.1982, + "step": 2115 + }, + { + "epoch": 2.1165952924259948, + "grad_norm": 0.12292486429214478, + "learning_rate": 5.88903743315508e-05, + "loss": 0.1964, + "step": 2116 + }, + { + "epoch": 2.1175955737551186, + "grad_norm": 0.10708366334438324, + "learning_rate": 5.882352941176471e-05, + "loss": 0.1721, + "step": 2117 + }, + { + "epoch": 2.1185958550842425, + "grad_norm": 0.11189883947372437, + "learning_rate": 5.875668449197861e-05, + "loss": 0.1835, + "step": 2118 + }, + { + "epoch": 2.1195961364133664, + "grad_norm": 0.11659733951091766, + "learning_rate": 5.8689839572192516e-05, + "loss": 0.1871, + "step": 2119 + }, + { + "epoch": 2.1205964177424903, + "grad_norm": 0.1093573272228241, + "learning_rate": 5.8622994652406414e-05, + "loss": 0.1755, + "step": 2120 + }, + { + "epoch": 2.1215966990716137, + "grad_norm": 0.10529494285583496, + "learning_rate": 5.8556149732620325e-05, + "loss": 0.1776, + "step": 2121 + }, + { + "epoch": 2.1225969804007376, + "grad_norm": 0.1113007441163063, + "learning_rate": 5.848930481283422e-05, + "loss": 0.1809, + "step": 2122 + }, + { + "epoch": 2.1235972617298615, + "grad_norm": 0.11573533713817596, + "learning_rate": 5.8422459893048134e-05, + "loss": 0.1782, + "step": 2123 + }, + { + "epoch": 2.1245975430589854, + "grad_norm": 0.1093338131904602, + "learning_rate": 5.835561497326203e-05, + "loss": 0.1874, + "step": 2124 + }, + { + "epoch": 2.1255978243881093, + "grad_norm": 0.10674653202295303, + "learning_rate": 5.8288770053475936e-05, + "loss": 0.1856, + "step": 2125 + }, + { + "epoch": 2.126598105717233, + "grad_norm": 0.1141589879989624, + "learning_rate": 5.822192513368985e-05, + "loss": 0.1923, + "step": 2126 + }, + { + "epoch": 2.1275983870463566, + "grad_norm": 0.116493359208107, + "learning_rate": 5.8155080213903744e-05, + "loss": 0.1949, + "step": 2127 + }, + { + "epoch": 2.1285986683754805, + "grad_norm": 0.10836219042539597, + "learning_rate": 5.8088235294117656e-05, + "loss": 0.1777, + "step": 2128 + }, + { + "epoch": 2.1295989497046044, + "grad_norm": 0.11083490401506424, + "learning_rate": 5.802139037433155e-05, + "loss": 0.1863, + "step": 2129 + }, + { + "epoch": 2.1305992310337283, + "grad_norm": 0.11411403119564056, + "learning_rate": 5.7954545454545464e-05, + "loss": 0.1838, + "step": 2130 + }, + { + "epoch": 2.131599512362852, + "grad_norm": 0.10097968578338623, + "learning_rate": 5.788770053475936e-05, + "loss": 0.1881, + "step": 2131 + }, + { + "epoch": 2.132599793691976, + "grad_norm": 0.11204564571380615, + "learning_rate": 5.7820855614973266e-05, + "loss": 0.1747, + "step": 2132 + }, + { + "epoch": 2.1336000750210995, + "grad_norm": 0.12858900427818298, + "learning_rate": 5.7754010695187164e-05, + "loss": 0.2081, + "step": 2133 + }, + { + "epoch": 2.1346003563502234, + "grad_norm": 0.11226882040500641, + "learning_rate": 5.7687165775401075e-05, + "loss": 0.1822, + "step": 2134 + }, + { + "epoch": 2.1356006376793473, + "grad_norm": 0.10801079869270325, + "learning_rate": 5.762032085561497e-05, + "loss": 0.1842, + "step": 2135 + }, + { + "epoch": 2.136600919008471, + "grad_norm": 0.11271747201681137, + "learning_rate": 5.7553475935828884e-05, + "loss": 0.194, + "step": 2136 + }, + { + "epoch": 2.137601200337595, + "grad_norm": 0.11369425803422928, + "learning_rate": 5.748663101604278e-05, + "loss": 0.1928, + "step": 2137 + }, + { + "epoch": 2.138601481666719, + "grad_norm": 0.11754394322633743, + "learning_rate": 5.741978609625669e-05, + "loss": 0.191, + "step": 2138 + }, + { + "epoch": 2.1396017629958424, + "grad_norm": 0.11535073816776276, + "learning_rate": 5.735294117647059e-05, + "loss": 0.1866, + "step": 2139 + }, + { + "epoch": 2.1406020443249663, + "grad_norm": 0.11725279688835144, + "learning_rate": 5.7286096256684494e-05, + "loss": 0.1836, + "step": 2140 + }, + { + "epoch": 2.14160232565409, + "grad_norm": 0.11564541608095169, + "learning_rate": 5.721925133689839e-05, + "loss": 0.1968, + "step": 2141 + }, + { + "epoch": 2.142602606983214, + "grad_norm": 0.11071733385324478, + "learning_rate": 5.71524064171123e-05, + "loss": 0.182, + "step": 2142 + }, + { + "epoch": 2.143602888312338, + "grad_norm": 0.10674645751714706, + "learning_rate": 5.70855614973262e-05, + "loss": 0.1769, + "step": 2143 + }, + { + "epoch": 2.144603169641462, + "grad_norm": 0.10954803228378296, + "learning_rate": 5.701871657754011e-05, + "loss": 0.1818, + "step": 2144 + }, + { + "epoch": 2.1456034509705857, + "grad_norm": 0.1106463298201561, + "learning_rate": 5.695187165775401e-05, + "loss": 0.1841, + "step": 2145 + }, + { + "epoch": 2.146603732299709, + "grad_norm": 0.11861899495124817, + "learning_rate": 5.6885026737967914e-05, + "loss": 0.1885, + "step": 2146 + }, + { + "epoch": 2.147604013628833, + "grad_norm": 0.11067533493041992, + "learning_rate": 5.6818181818181825e-05, + "loss": 0.1977, + "step": 2147 + }, + { + "epoch": 2.148604294957957, + "grad_norm": 0.11268913000822067, + "learning_rate": 5.675133689839572e-05, + "loss": 0.1749, + "step": 2148 + }, + { + "epoch": 2.149604576287081, + "grad_norm": 0.10840161889791489, + "learning_rate": 5.6684491978609634e-05, + "loss": 0.1884, + "step": 2149 + }, + { + "epoch": 2.1506048576162047, + "grad_norm": 0.11316945403814316, + "learning_rate": 5.661764705882353e-05, + "loss": 0.1767, + "step": 2150 + }, + { + "epoch": 2.1516051389453286, + "grad_norm": 0.11524755507707596, + "learning_rate": 5.655080213903744e-05, + "loss": 0.1812, + "step": 2151 + }, + { + "epoch": 2.152605420274452, + "grad_norm": 0.11191259324550629, + "learning_rate": 5.648395721925134e-05, + "loss": 0.1886, + "step": 2152 + }, + { + "epoch": 2.153605701603576, + "grad_norm": 0.11462168395519257, + "learning_rate": 5.6417112299465244e-05, + "loss": 0.1837, + "step": 2153 + }, + { + "epoch": 2.1546059829327, + "grad_norm": 0.12059546262025833, + "learning_rate": 5.635026737967914e-05, + "loss": 0.2045, + "step": 2154 + }, + { + "epoch": 2.1556062642618237, + "grad_norm": 0.11421585083007812, + "learning_rate": 5.628342245989305e-05, + "loss": 0.1821, + "step": 2155 + }, + { + "epoch": 2.1566065455909476, + "grad_norm": 0.11610234528779984, + "learning_rate": 5.621657754010695e-05, + "loss": 0.1897, + "step": 2156 + }, + { + "epoch": 2.1576068269200714, + "grad_norm": 0.1134008839726448, + "learning_rate": 5.614973262032086e-05, + "loss": 0.195, + "step": 2157 + }, + { + "epoch": 2.158607108249195, + "grad_norm": 0.10600201040506363, + "learning_rate": 5.608288770053476e-05, + "loss": 0.1867, + "step": 2158 + }, + { + "epoch": 2.1596073895783188, + "grad_norm": 0.11804906278848648, + "learning_rate": 5.601604278074867e-05, + "loss": 0.1785, + "step": 2159 + }, + { + "epoch": 2.1606076709074427, + "grad_norm": 0.11915843188762665, + "learning_rate": 5.594919786096257e-05, + "loss": 0.1847, + "step": 2160 + }, + { + "epoch": 2.1616079522365665, + "grad_norm": 0.11580421775579453, + "learning_rate": 5.588235294117647e-05, + "loss": 0.1793, + "step": 2161 + }, + { + "epoch": 2.1626082335656904, + "grad_norm": 0.12210778146982193, + "learning_rate": 5.581550802139037e-05, + "loss": 0.1938, + "step": 2162 + }, + { + "epoch": 2.1636085148948143, + "grad_norm": 0.11953824013471603, + "learning_rate": 5.574866310160428e-05, + "loss": 0.1974, + "step": 2163 + }, + { + "epoch": 2.1646087962239378, + "grad_norm": 0.11176320910453796, + "learning_rate": 5.568181818181818e-05, + "loss": 0.186, + "step": 2164 + }, + { + "epoch": 2.1656090775530616, + "grad_norm": 0.11296023428440094, + "learning_rate": 5.561497326203209e-05, + "loss": 0.1822, + "step": 2165 + }, + { + "epoch": 2.1666093588821855, + "grad_norm": 0.11028660088777542, + "learning_rate": 5.5548128342245994e-05, + "loss": 0.18, + "step": 2166 + }, + { + "epoch": 2.1676096402113094, + "grad_norm": 0.11178257316350937, + "learning_rate": 5.54812834224599e-05, + "loss": 0.1846, + "step": 2167 + }, + { + "epoch": 2.1686099215404333, + "grad_norm": 0.12787844240665436, + "learning_rate": 5.54144385026738e-05, + "loss": 0.1947, + "step": 2168 + }, + { + "epoch": 2.169610202869557, + "grad_norm": 0.12101306021213531, + "learning_rate": 5.53475935828877e-05, + "loss": 0.2101, + "step": 2169 + }, + { + "epoch": 2.170610484198681, + "grad_norm": 0.11210435628890991, + "learning_rate": 5.528074866310161e-05, + "loss": 0.1879, + "step": 2170 + }, + { + "epoch": 2.1716107655278045, + "grad_norm": 0.11329761892557144, + "learning_rate": 5.521390374331551e-05, + "loss": 0.1877, + "step": 2171 + }, + { + "epoch": 2.1726110468569284, + "grad_norm": 0.11082441359758377, + "learning_rate": 5.514705882352942e-05, + "loss": 0.1874, + "step": 2172 + }, + { + "epoch": 2.1736113281860523, + "grad_norm": 0.11098521947860718, + "learning_rate": 5.508021390374332e-05, + "loss": 0.1792, + "step": 2173 + }, + { + "epoch": 2.174611609515176, + "grad_norm": 0.11709199100732803, + "learning_rate": 5.501336898395722e-05, + "loss": 0.1902, + "step": 2174 + }, + { + "epoch": 2.1756118908443, + "grad_norm": 0.10912490636110306, + "learning_rate": 5.494652406417112e-05, + "loss": 0.178, + "step": 2175 + }, + { + "epoch": 2.176612172173424, + "grad_norm": 0.11525019258260727, + "learning_rate": 5.487967914438503e-05, + "loss": 0.1866, + "step": 2176 + }, + { + "epoch": 2.1776124535025474, + "grad_norm": 0.1099248006939888, + "learning_rate": 5.481283422459893e-05, + "loss": 0.1829, + "step": 2177 + }, + { + "epoch": 2.1786127348316713, + "grad_norm": 0.10990653187036514, + "learning_rate": 5.474598930481284e-05, + "loss": 0.1908, + "step": 2178 + }, + { + "epoch": 2.179613016160795, + "grad_norm": 0.11586243659257889, + "learning_rate": 5.467914438502674e-05, + "loss": 0.1965, + "step": 2179 + }, + { + "epoch": 2.180613297489919, + "grad_norm": 0.11226756870746613, + "learning_rate": 5.461229946524065e-05, + "loss": 0.1881, + "step": 2180 + }, + { + "epoch": 2.181613578819043, + "grad_norm": 0.11199352145195007, + "learning_rate": 5.4545454545454546e-05, + "loss": 0.1844, + "step": 2181 + }, + { + "epoch": 2.182613860148167, + "grad_norm": 0.1089581623673439, + "learning_rate": 5.447860962566845e-05, + "loss": 0.1812, + "step": 2182 + }, + { + "epoch": 2.1836141414772907, + "grad_norm": 0.11387810111045837, + "learning_rate": 5.441176470588235e-05, + "loss": 0.1705, + "step": 2183 + }, + { + "epoch": 2.184614422806414, + "grad_norm": 0.12142767012119293, + "learning_rate": 5.434491978609626e-05, + "loss": 0.2016, + "step": 2184 + }, + { + "epoch": 2.185614704135538, + "grad_norm": 0.11274643242359161, + "learning_rate": 5.427807486631016e-05, + "loss": 0.1761, + "step": 2185 + }, + { + "epoch": 2.186614985464662, + "grad_norm": 0.11214972287416458, + "learning_rate": 5.421122994652407e-05, + "loss": 0.1719, + "step": 2186 + }, + { + "epoch": 2.187615266793786, + "grad_norm": 0.11063642054796219, + "learning_rate": 5.414438502673798e-05, + "loss": 0.1862, + "step": 2187 + }, + { + "epoch": 2.1886155481229097, + "grad_norm": 0.11395307630300522, + "learning_rate": 5.407754010695188e-05, + "loss": 0.1774, + "step": 2188 + }, + { + "epoch": 2.1896158294520336, + "grad_norm": 0.1106448546051979, + "learning_rate": 5.401069518716578e-05, + "loss": 0.204, + "step": 2189 + }, + { + "epoch": 2.190616110781157, + "grad_norm": 0.12402638792991638, + "learning_rate": 5.394385026737968e-05, + "loss": 0.1903, + "step": 2190 + }, + { + "epoch": 2.191616392110281, + "grad_norm": 0.11451542377471924, + "learning_rate": 5.387700534759359e-05, + "loss": 0.1793, + "step": 2191 + }, + { + "epoch": 2.192616673439405, + "grad_norm": 0.11183454841375351, + "learning_rate": 5.381016042780749e-05, + "loss": 0.1854, + "step": 2192 + }, + { + "epoch": 2.1936169547685287, + "grad_norm": 0.10850141942501068, + "learning_rate": 5.37433155080214e-05, + "loss": 0.1805, + "step": 2193 + }, + { + "epoch": 2.1946172360976526, + "grad_norm": 0.1163683533668518, + "learning_rate": 5.3676470588235296e-05, + "loss": 0.1846, + "step": 2194 + }, + { + "epoch": 2.1956175174267765, + "grad_norm": 0.11309783160686493, + "learning_rate": 5.36096256684492e-05, + "loss": 0.1906, + "step": 2195 + }, + { + "epoch": 2.1966177987559, + "grad_norm": 0.11485867202281952, + "learning_rate": 5.3542780748663105e-05, + "loss": 0.1879, + "step": 2196 + }, + { + "epoch": 2.197618080085024, + "grad_norm": 0.1104653850197792, + "learning_rate": 5.347593582887701e-05, + "loss": 0.1827, + "step": 2197 + }, + { + "epoch": 2.1986183614141477, + "grad_norm": 0.11980270594358444, + "learning_rate": 5.340909090909091e-05, + "loss": 0.1933, + "step": 2198 + }, + { + "epoch": 2.1996186427432716, + "grad_norm": 0.11870072782039642, + "learning_rate": 5.334224598930482e-05, + "loss": 0.1842, + "step": 2199 + }, + { + "epoch": 2.2006189240723955, + "grad_norm": 0.1231977567076683, + "learning_rate": 5.3275401069518716e-05, + "loss": 0.1978, + "step": 2200 + }, + { + "epoch": 2.2016192054015193, + "grad_norm": 0.11404883116483688, + "learning_rate": 5.320855614973263e-05, + "loss": 0.1957, + "step": 2201 + }, + { + "epoch": 2.202619486730643, + "grad_norm": 0.1224348396062851, + "learning_rate": 5.3141711229946524e-05, + "loss": 0.1899, + "step": 2202 + }, + { + "epoch": 2.2036197680597667, + "grad_norm": 0.11751773208379745, + "learning_rate": 5.307486631016043e-05, + "loss": 0.1885, + "step": 2203 + }, + { + "epoch": 2.2046200493888906, + "grad_norm": 0.11732390522956848, + "learning_rate": 5.3008021390374326e-05, + "loss": 0.1829, + "step": 2204 + }, + { + "epoch": 2.2056203307180144, + "grad_norm": 0.12854495644569397, + "learning_rate": 5.294117647058824e-05, + "loss": 0.189, + "step": 2205 + }, + { + "epoch": 2.2066206120471383, + "grad_norm": 0.11028435826301575, + "learning_rate": 5.2874331550802135e-05, + "loss": 0.181, + "step": 2206 + }, + { + "epoch": 2.207620893376262, + "grad_norm": 0.11900980025529861, + "learning_rate": 5.2807486631016046e-05, + "loss": 0.1742, + "step": 2207 + }, + { + "epoch": 2.208621174705386, + "grad_norm": 0.11395134776830673, + "learning_rate": 5.274064171122996e-05, + "loss": 0.1928, + "step": 2208 + }, + { + "epoch": 2.2096214560345095, + "grad_norm": 0.12317701429128647, + "learning_rate": 5.2673796791443855e-05, + "loss": 0.1985, + "step": 2209 + }, + { + "epoch": 2.2106217373636334, + "grad_norm": 0.10991912335157394, + "learning_rate": 5.260695187165776e-05, + "loss": 0.1776, + "step": 2210 + }, + { + "epoch": 2.2116220186927573, + "grad_norm": 0.12468933314085007, + "learning_rate": 5.254010695187166e-05, + "loss": 0.2039, + "step": 2211 + }, + { + "epoch": 2.212622300021881, + "grad_norm": 0.11135058104991913, + "learning_rate": 5.247326203208557e-05, + "loss": 0.1843, + "step": 2212 + }, + { + "epoch": 2.213622581351005, + "grad_norm": 0.11962714791297913, + "learning_rate": 5.2406417112299466e-05, + "loss": 0.1996, + "step": 2213 + }, + { + "epoch": 2.214622862680129, + "grad_norm": 0.11343930661678314, + "learning_rate": 5.233957219251338e-05, + "loss": 0.1778, + "step": 2214 + }, + { + "epoch": 2.2156231440092524, + "grad_norm": 0.11276109516620636, + "learning_rate": 5.2272727272727274e-05, + "loss": 0.1851, + "step": 2215 + }, + { + "epoch": 2.2166234253383763, + "grad_norm": 0.11701574921607971, + "learning_rate": 5.2205882352941185e-05, + "loss": 0.1813, + "step": 2216 + }, + { + "epoch": 2.2176237066675, + "grad_norm": 0.10809197276830673, + "learning_rate": 5.213903743315508e-05, + "loss": 0.1826, + "step": 2217 + }, + { + "epoch": 2.218623987996624, + "grad_norm": 0.12051920592784882, + "learning_rate": 5.207219251336899e-05, + "loss": 0.2003, + "step": 2218 + }, + { + "epoch": 2.219624269325748, + "grad_norm": 0.1184920072555542, + "learning_rate": 5.2005347593582885e-05, + "loss": 0.1924, + "step": 2219 + }, + { + "epoch": 2.220624550654872, + "grad_norm": 0.10980666428804398, + "learning_rate": 5.1938502673796796e-05, + "loss": 0.1717, + "step": 2220 + }, + { + "epoch": 2.2216248319839953, + "grad_norm": 0.1160164624452591, + "learning_rate": 5.1871657754010694e-05, + "loss": 0.2025, + "step": 2221 + }, + { + "epoch": 2.222625113313119, + "grad_norm": 0.11981448531150818, + "learning_rate": 5.1804812834224605e-05, + "loss": 0.1788, + "step": 2222 + }, + { + "epoch": 2.223625394642243, + "grad_norm": 0.11175105720758438, + "learning_rate": 5.17379679144385e-05, + "loss": 0.1926, + "step": 2223 + }, + { + "epoch": 2.224625675971367, + "grad_norm": 0.11789514124393463, + "learning_rate": 5.167112299465241e-05, + "loss": 0.1916, + "step": 2224 + }, + { + "epoch": 2.225625957300491, + "grad_norm": 0.12483401596546173, + "learning_rate": 5.160427807486631e-05, + "loss": 0.1815, + "step": 2225 + }, + { + "epoch": 2.2266262386296147, + "grad_norm": 0.11274877190589905, + "learning_rate": 5.1537433155080216e-05, + "loss": 0.1817, + "step": 2226 + }, + { + "epoch": 2.227626519958738, + "grad_norm": 0.12207219749689102, + "learning_rate": 5.147058823529411e-05, + "loss": 0.1913, + "step": 2227 + }, + { + "epoch": 2.228626801287862, + "grad_norm": 0.11443532258272171, + "learning_rate": 5.1403743315508024e-05, + "loss": 0.1913, + "step": 2228 + }, + { + "epoch": 2.229627082616986, + "grad_norm": 0.11030254513025284, + "learning_rate": 5.1336898395721935e-05, + "loss": 0.1733, + "step": 2229 + }, + { + "epoch": 2.23062736394611, + "grad_norm": 0.11382201313972473, + "learning_rate": 5.127005347593583e-05, + "loss": 0.1872, + "step": 2230 + }, + { + "epoch": 2.2316276452752337, + "grad_norm": 0.11711333692073822, + "learning_rate": 5.120320855614974e-05, + "loss": 0.2028, + "step": 2231 + }, + { + "epoch": 2.2326279266043576, + "grad_norm": 0.10720774531364441, + "learning_rate": 5.1136363636363635e-05, + "loss": 0.1893, + "step": 2232 + }, + { + "epoch": 2.2336282079334815, + "grad_norm": 0.10950113832950592, + "learning_rate": 5.1069518716577546e-05, + "loss": 0.1855, + "step": 2233 + }, + { + "epoch": 2.234628489262605, + "grad_norm": 0.11133814603090286, + "learning_rate": 5.1002673796791444e-05, + "loss": 0.1856, + "step": 2234 + }, + { + "epoch": 2.235628770591729, + "grad_norm": 0.12179043143987656, + "learning_rate": 5.0935828877005355e-05, + "loss": 0.1936, + "step": 2235 + }, + { + "epoch": 2.2366290519208527, + "grad_norm": 0.11695338785648346, + "learning_rate": 5.086898395721925e-05, + "loss": 0.2037, + "step": 2236 + }, + { + "epoch": 2.2376293332499766, + "grad_norm": 0.10655202716588974, + "learning_rate": 5.0802139037433164e-05, + "loss": 0.1918, + "step": 2237 + }, + { + "epoch": 2.2386296145791005, + "grad_norm": 0.1147528663277626, + "learning_rate": 5.073529411764706e-05, + "loss": 0.1839, + "step": 2238 + }, + { + "epoch": 2.2396298959082244, + "grad_norm": 0.11131749302148819, + "learning_rate": 5.0668449197860966e-05, + "loss": 0.1823, + "step": 2239 + }, + { + "epoch": 2.240630177237348, + "grad_norm": 0.11273054033517838, + "learning_rate": 5.060160427807486e-05, + "loss": 0.184, + "step": 2240 + }, + { + "epoch": 2.2416304585664717, + "grad_norm": 0.11535310745239258, + "learning_rate": 5.0534759358288774e-05, + "loss": 0.1917, + "step": 2241 + }, + { + "epoch": 2.2426307398955956, + "grad_norm": 0.12009578943252563, + "learning_rate": 5.046791443850267e-05, + "loss": 0.1934, + "step": 2242 + }, + { + "epoch": 2.2436310212247195, + "grad_norm": 0.11545684933662415, + "learning_rate": 5.040106951871658e-05, + "loss": 0.1905, + "step": 2243 + }, + { + "epoch": 2.2446313025538434, + "grad_norm": 0.11626545339822769, + "learning_rate": 5.033422459893048e-05, + "loss": 0.1912, + "step": 2244 + }, + { + "epoch": 2.2456315838829672, + "grad_norm": 0.11583666503429413, + "learning_rate": 5.026737967914439e-05, + "loss": 0.1923, + "step": 2245 + }, + { + "epoch": 2.246631865212091, + "grad_norm": 0.11767042428255081, + "learning_rate": 5.020053475935829e-05, + "loss": 0.1917, + "step": 2246 + }, + { + "epoch": 2.2476321465412146, + "grad_norm": 0.10912401974201202, + "learning_rate": 5.0133689839572194e-05, + "loss": 0.1806, + "step": 2247 + }, + { + "epoch": 2.2486324278703385, + "grad_norm": 0.10802789032459259, + "learning_rate": 5.006684491978609e-05, + "loss": 0.1827, + "step": 2248 + }, + { + "epoch": 2.2496327091994623, + "grad_norm": 0.11802612245082855, + "learning_rate": 5e-05, + "loss": 0.1908, + "step": 2249 + }, + { + "epoch": 2.2506329905285862, + "grad_norm": 0.11066144704818726, + "learning_rate": 4.993315508021391e-05, + "loss": 0.1807, + "step": 2250 + }, + { + "epoch": 2.25163327185771, + "grad_norm": 0.11843036115169525, + "learning_rate": 4.986631016042781e-05, + "loss": 0.1828, + "step": 2251 + }, + { + "epoch": 2.2526335531868336, + "grad_norm": 0.11480563133955002, + "learning_rate": 4.9799465240641715e-05, + "loss": 0.1853, + "step": 2252 + }, + { + "epoch": 2.2536338345159574, + "grad_norm": 0.11702784895896912, + "learning_rate": 4.973262032085561e-05, + "loss": 0.1927, + "step": 2253 + }, + { + "epoch": 2.2546341158450813, + "grad_norm": 0.11699986457824707, + "learning_rate": 4.966577540106952e-05, + "loss": 0.1864, + "step": 2254 + }, + { + "epoch": 2.255634397174205, + "grad_norm": 0.1117796078324318, + "learning_rate": 4.959893048128342e-05, + "loss": 0.1924, + "step": 2255 + }, + { + "epoch": 2.256634678503329, + "grad_norm": 0.1060035303235054, + "learning_rate": 4.9532085561497326e-05, + "loss": 0.1839, + "step": 2256 + }, + { + "epoch": 2.257634959832453, + "grad_norm": 0.10654386878013611, + "learning_rate": 4.946524064171123e-05, + "loss": 0.181, + "step": 2257 + }, + { + "epoch": 2.258635241161577, + "grad_norm": 0.11589439958333969, + "learning_rate": 4.9398395721925135e-05, + "loss": 0.1986, + "step": 2258 + }, + { + "epoch": 2.2596355224907003, + "grad_norm": 0.10987628251314163, + "learning_rate": 4.933155080213904e-05, + "loss": 0.1762, + "step": 2259 + }, + { + "epoch": 2.260635803819824, + "grad_norm": 0.12153096497058868, + "learning_rate": 4.9264705882352944e-05, + "loss": 0.1858, + "step": 2260 + }, + { + "epoch": 2.261636085148948, + "grad_norm": 0.11020601540803909, + "learning_rate": 4.919786096256685e-05, + "loss": 0.1796, + "step": 2261 + }, + { + "epoch": 2.262636366478072, + "grad_norm": 0.11172652989625931, + "learning_rate": 4.913101604278075e-05, + "loss": 0.1828, + "step": 2262 + }, + { + "epoch": 2.263636647807196, + "grad_norm": 0.1174544170498848, + "learning_rate": 4.906417112299466e-05, + "loss": 0.19, + "step": 2263 + }, + { + "epoch": 2.2646369291363198, + "grad_norm": 0.11813440918922424, + "learning_rate": 4.899732620320856e-05, + "loss": 0.1824, + "step": 2264 + }, + { + "epoch": 2.265637210465443, + "grad_norm": 0.11889355629682541, + "learning_rate": 4.8930481283422465e-05, + "loss": 0.1994, + "step": 2265 + }, + { + "epoch": 2.266637491794567, + "grad_norm": 0.1216985285282135, + "learning_rate": 4.886363636363637e-05, + "loss": 0.1843, + "step": 2266 + }, + { + "epoch": 2.267637773123691, + "grad_norm": 0.11741454154253006, + "learning_rate": 4.879679144385027e-05, + "loss": 0.1865, + "step": 2267 + }, + { + "epoch": 2.268638054452815, + "grad_norm": 0.11740723997354507, + "learning_rate": 4.872994652406417e-05, + "loss": 0.1846, + "step": 2268 + }, + { + "epoch": 2.2696383357819387, + "grad_norm": 0.11540872603654861, + "learning_rate": 4.8663101604278076e-05, + "loss": 0.1909, + "step": 2269 + }, + { + "epoch": 2.2706386171110626, + "grad_norm": 0.11549116671085358, + "learning_rate": 4.859625668449198e-05, + "loss": 0.1819, + "step": 2270 + }, + { + "epoch": 2.2716388984401865, + "grad_norm": 0.1092991754412651, + "learning_rate": 4.8529411764705885e-05, + "loss": 0.1838, + "step": 2271 + }, + { + "epoch": 2.27263917976931, + "grad_norm": 0.11281930655241013, + "learning_rate": 4.846256684491979e-05, + "loss": 0.1754, + "step": 2272 + }, + { + "epoch": 2.273639461098434, + "grad_norm": 0.12035032361745834, + "learning_rate": 4.8395721925133694e-05, + "loss": 0.1895, + "step": 2273 + }, + { + "epoch": 2.2746397424275577, + "grad_norm": 0.11022516340017319, + "learning_rate": 4.83288770053476e-05, + "loss": 0.1893, + "step": 2274 + }, + { + "epoch": 2.2756400237566816, + "grad_norm": 0.10957834124565125, + "learning_rate": 4.8262032085561496e-05, + "loss": 0.1859, + "step": 2275 + }, + { + "epoch": 2.2766403050858055, + "grad_norm": 0.11057478189468384, + "learning_rate": 4.81951871657754e-05, + "loss": 0.1789, + "step": 2276 + }, + { + "epoch": 2.2776405864149294, + "grad_norm": 0.11292680352926254, + "learning_rate": 4.8128342245989304e-05, + "loss": 0.1829, + "step": 2277 + }, + { + "epoch": 2.278640867744053, + "grad_norm": 0.11414257436990738, + "learning_rate": 4.806149732620321e-05, + "loss": 0.1915, + "step": 2278 + }, + { + "epoch": 2.2796411490731767, + "grad_norm": 0.10744684934616089, + "learning_rate": 4.799465240641711e-05, + "loss": 0.1735, + "step": 2279 + }, + { + "epoch": 2.2806414304023006, + "grad_norm": 0.11022327095270157, + "learning_rate": 4.792780748663102e-05, + "loss": 0.1808, + "step": 2280 + }, + { + "epoch": 2.2816417117314245, + "grad_norm": 0.11543437838554382, + "learning_rate": 4.786096256684492e-05, + "loss": 0.1941, + "step": 2281 + }, + { + "epoch": 2.2826419930605484, + "grad_norm": 0.11070392280817032, + "learning_rate": 4.7794117647058826e-05, + "loss": 0.1849, + "step": 2282 + }, + { + "epoch": 2.2836422743896723, + "grad_norm": 0.12021474540233612, + "learning_rate": 4.772727272727273e-05, + "loss": 0.1866, + "step": 2283 + }, + { + "epoch": 2.284642555718796, + "grad_norm": 0.1137891560792923, + "learning_rate": 4.7660427807486635e-05, + "loss": 0.1927, + "step": 2284 + }, + { + "epoch": 2.2856428370479196, + "grad_norm": 0.11389897763729095, + "learning_rate": 4.759358288770054e-05, + "loss": 0.1994, + "step": 2285 + }, + { + "epoch": 2.2866431183770435, + "grad_norm": 0.11621644347906113, + "learning_rate": 4.7526737967914444e-05, + "loss": 0.189, + "step": 2286 + }, + { + "epoch": 2.2876433997061674, + "grad_norm": 0.11656080931425095, + "learning_rate": 4.745989304812835e-05, + "loss": 0.1911, + "step": 2287 + }, + { + "epoch": 2.2886436810352913, + "grad_norm": 0.11101100593805313, + "learning_rate": 4.739304812834225e-05, + "loss": 0.1726, + "step": 2288 + }, + { + "epoch": 2.289643962364415, + "grad_norm": 0.1374567300081253, + "learning_rate": 4.732620320855615e-05, + "loss": 0.1933, + "step": 2289 + }, + { + "epoch": 2.2906442436935386, + "grad_norm": 0.11960419267416, + "learning_rate": 4.7259358288770054e-05, + "loss": 0.1902, + "step": 2290 + }, + { + "epoch": 2.2916445250226625, + "grad_norm": 0.11164494603872299, + "learning_rate": 4.719251336898396e-05, + "loss": 0.1758, + "step": 2291 + }, + { + "epoch": 2.2926448063517864, + "grad_norm": 0.11720620840787888, + "learning_rate": 4.712566844919786e-05, + "loss": 0.1832, + "step": 2292 + }, + { + "epoch": 2.2936450876809102, + "grad_norm": 0.12942937016487122, + "learning_rate": 4.705882352941177e-05, + "loss": 0.1953, + "step": 2293 + }, + { + "epoch": 2.294645369010034, + "grad_norm": 0.11260172724723816, + "learning_rate": 4.699197860962567e-05, + "loss": 0.1871, + "step": 2294 + }, + { + "epoch": 2.295645650339158, + "grad_norm": 0.11481954902410507, + "learning_rate": 4.6925133689839576e-05, + "loss": 0.1933, + "step": 2295 + }, + { + "epoch": 2.296645931668282, + "grad_norm": 0.10846007615327835, + "learning_rate": 4.6858288770053474e-05, + "loss": 0.176, + "step": 2296 + }, + { + "epoch": 2.2976462129974053, + "grad_norm": 0.11981305480003357, + "learning_rate": 4.679144385026738e-05, + "loss": 0.1924, + "step": 2297 + }, + { + "epoch": 2.2986464943265292, + "grad_norm": 0.11099913716316223, + "learning_rate": 4.672459893048128e-05, + "loss": 0.1823, + "step": 2298 + }, + { + "epoch": 2.299646775655653, + "grad_norm": 0.12039750814437866, + "learning_rate": 4.665775401069519e-05, + "loss": 0.1895, + "step": 2299 + }, + { + "epoch": 2.300647056984777, + "grad_norm": 0.11250042170286179, + "learning_rate": 4.659090909090909e-05, + "loss": 0.1898, + "step": 2300 + }, + { + "epoch": 2.301647338313901, + "grad_norm": 0.11518436670303345, + "learning_rate": 4.6524064171123e-05, + "loss": 0.1863, + "step": 2301 + }, + { + "epoch": 2.302647619643025, + "grad_norm": 0.11186159402132034, + "learning_rate": 4.64572192513369e-05, + "loss": 0.1765, + "step": 2302 + }, + { + "epoch": 2.303647900972148, + "grad_norm": 0.11237137764692307, + "learning_rate": 4.6390374331550804e-05, + "loss": 0.195, + "step": 2303 + }, + { + "epoch": 2.304648182301272, + "grad_norm": 0.12056280672550201, + "learning_rate": 4.632352941176471e-05, + "loss": 0.1906, + "step": 2304 + }, + { + "epoch": 2.305648463630396, + "grad_norm": 0.11730792373418808, + "learning_rate": 4.625668449197861e-05, + "loss": 0.1886, + "step": 2305 + }, + { + "epoch": 2.30664874495952, + "grad_norm": 0.11265797168016434, + "learning_rate": 4.618983957219252e-05, + "loss": 0.18, + "step": 2306 + }, + { + "epoch": 2.3076490262886438, + "grad_norm": 0.11889608949422836, + "learning_rate": 4.612299465240642e-05, + "loss": 0.1796, + "step": 2307 + }, + { + "epoch": 2.3086493076177677, + "grad_norm": 0.12113308161497116, + "learning_rate": 4.6056149732620326e-05, + "loss": 0.2023, + "step": 2308 + }, + { + "epoch": 2.3096495889468915, + "grad_norm": 0.10867419093847275, + "learning_rate": 4.598930481283423e-05, + "loss": 0.1793, + "step": 2309 + }, + { + "epoch": 2.310649870276015, + "grad_norm": 0.11322946101427078, + "learning_rate": 4.592245989304813e-05, + "loss": 0.1771, + "step": 2310 + }, + { + "epoch": 2.311650151605139, + "grad_norm": 0.11752918362617493, + "learning_rate": 4.585561497326203e-05, + "loss": 0.1971, + "step": 2311 + }, + { + "epoch": 2.3126504329342628, + "grad_norm": 0.1141502633690834, + "learning_rate": 4.578877005347594e-05, + "loss": 0.1884, + "step": 2312 + }, + { + "epoch": 2.3136507142633866, + "grad_norm": 0.1181328296661377, + "learning_rate": 4.572192513368984e-05, + "loss": 0.1834, + "step": 2313 + }, + { + "epoch": 2.3146509955925105, + "grad_norm": 0.11048746854066849, + "learning_rate": 4.5655080213903745e-05, + "loss": 0.1774, + "step": 2314 + }, + { + "epoch": 2.315651276921634, + "grad_norm": 0.11803404241800308, + "learning_rate": 4.558823529411765e-05, + "loss": 0.1906, + "step": 2315 + }, + { + "epoch": 2.316651558250758, + "grad_norm": 0.12568484246730804, + "learning_rate": 4.5521390374331554e-05, + "loss": 0.1878, + "step": 2316 + }, + { + "epoch": 2.3176518395798817, + "grad_norm": 0.11656241118907928, + "learning_rate": 4.545454545454546e-05, + "loss": 0.1927, + "step": 2317 + }, + { + "epoch": 2.3186521209090056, + "grad_norm": 0.12032818049192429, + "learning_rate": 4.5387700534759356e-05, + "loss": 0.1809, + "step": 2318 + }, + { + "epoch": 2.3196524022381295, + "grad_norm": 0.12214060872793198, + "learning_rate": 4.532085561497326e-05, + "loss": 0.1935, + "step": 2319 + }, + { + "epoch": 2.3206526835672534, + "grad_norm": 0.11161692440509796, + "learning_rate": 4.5254010695187165e-05, + "loss": 0.1812, + "step": 2320 + }, + { + "epoch": 2.3216529648963773, + "grad_norm": 0.11367062479257584, + "learning_rate": 4.518716577540107e-05, + "loss": 0.1791, + "step": 2321 + }, + { + "epoch": 2.3226532462255007, + "grad_norm": 0.11451861262321472, + "learning_rate": 4.512032085561498e-05, + "loss": 0.182, + "step": 2322 + }, + { + "epoch": 2.3236535275546246, + "grad_norm": 0.11706706881523132, + "learning_rate": 4.5053475935828885e-05, + "loss": 0.1881, + "step": 2323 + }, + { + "epoch": 2.3246538088837485, + "grad_norm": 0.12343233078718185, + "learning_rate": 4.498663101604278e-05, + "loss": 0.1945, + "step": 2324 + }, + { + "epoch": 2.3256540902128724, + "grad_norm": 0.11111657321453094, + "learning_rate": 4.491978609625669e-05, + "loss": 0.1796, + "step": 2325 + }, + { + "epoch": 2.3266543715419963, + "grad_norm": 0.11170869320631027, + "learning_rate": 4.485294117647059e-05, + "loss": 0.1837, + "step": 2326 + }, + { + "epoch": 2.32765465287112, + "grad_norm": 0.11757306754589081, + "learning_rate": 4.4786096256684495e-05, + "loss": 0.1775, + "step": 2327 + }, + { + "epoch": 2.3286549342002436, + "grad_norm": 0.11097419261932373, + "learning_rate": 4.47192513368984e-05, + "loss": 0.187, + "step": 2328 + }, + { + "epoch": 2.3296552155293675, + "grad_norm": 0.12024274468421936, + "learning_rate": 4.4652406417112304e-05, + "loss": 0.1895, + "step": 2329 + }, + { + "epoch": 2.3306554968584914, + "grad_norm": 0.10873235017061234, + "learning_rate": 4.458556149732621e-05, + "loss": 0.1775, + "step": 2330 + }, + { + "epoch": 2.3316557781876153, + "grad_norm": 0.10483104735612869, + "learning_rate": 4.4518716577540106e-05, + "loss": 0.1758, + "step": 2331 + }, + { + "epoch": 2.332656059516739, + "grad_norm": 0.11706613004207611, + "learning_rate": 4.445187165775401e-05, + "loss": 0.1834, + "step": 2332 + }, + { + "epoch": 2.333656340845863, + "grad_norm": 0.11985024064779282, + "learning_rate": 4.4385026737967915e-05, + "loss": 0.1863, + "step": 2333 + }, + { + "epoch": 2.334656622174987, + "grad_norm": 0.11843805760145187, + "learning_rate": 4.431818181818182e-05, + "loss": 0.1808, + "step": 2334 + }, + { + "epoch": 2.3356569035041104, + "grad_norm": 0.11651023477315903, + "learning_rate": 4.4251336898395724e-05, + "loss": 0.1898, + "step": 2335 + }, + { + "epoch": 2.3366571848332343, + "grad_norm": 0.11633055657148361, + "learning_rate": 4.418449197860963e-05, + "loss": 0.1864, + "step": 2336 + }, + { + "epoch": 2.337657466162358, + "grad_norm": 0.11121684312820435, + "learning_rate": 4.411764705882353e-05, + "loss": 0.185, + "step": 2337 + }, + { + "epoch": 2.338657747491482, + "grad_norm": 0.12212249636650085, + "learning_rate": 4.405080213903744e-05, + "loss": 0.186, + "step": 2338 + }, + { + "epoch": 2.339658028820606, + "grad_norm": 0.1182757019996643, + "learning_rate": 4.3983957219251334e-05, + "loss": 0.1846, + "step": 2339 + }, + { + "epoch": 2.34065831014973, + "grad_norm": 0.1163279116153717, + "learning_rate": 4.391711229946524e-05, + "loss": 0.1854, + "step": 2340 + }, + { + "epoch": 2.3416585914788532, + "grad_norm": 0.10840893536806107, + "learning_rate": 4.385026737967914e-05, + "loss": 0.1858, + "step": 2341 + }, + { + "epoch": 2.342658872807977, + "grad_norm": 0.11753135174512863, + "learning_rate": 4.378342245989305e-05, + "loss": 0.1791, + "step": 2342 + }, + { + "epoch": 2.343659154137101, + "grad_norm": 0.11415723711252213, + "learning_rate": 4.371657754010696e-05, + "loss": 0.1946, + "step": 2343 + }, + { + "epoch": 2.344659435466225, + "grad_norm": 0.11853901296854019, + "learning_rate": 4.364973262032086e-05, + "loss": 0.1818, + "step": 2344 + }, + { + "epoch": 2.345659716795349, + "grad_norm": 0.12028326094150543, + "learning_rate": 4.358288770053476e-05, + "loss": 0.194, + "step": 2345 + }, + { + "epoch": 2.3466599981244727, + "grad_norm": 0.12150484323501587, + "learning_rate": 4.3516042780748665e-05, + "loss": 0.1881, + "step": 2346 + }, + { + "epoch": 2.3476602794535966, + "grad_norm": 0.12614937126636505, + "learning_rate": 4.344919786096257e-05, + "loss": 0.192, + "step": 2347 + }, + { + "epoch": 2.34866056078272, + "grad_norm": 0.11329622566699982, + "learning_rate": 4.3382352941176474e-05, + "loss": 0.1868, + "step": 2348 + }, + { + "epoch": 2.349660842111844, + "grad_norm": 0.11910104006528854, + "learning_rate": 4.331550802139038e-05, + "loss": 0.1812, + "step": 2349 + }, + { + "epoch": 2.350661123440968, + "grad_norm": 0.11528147011995316, + "learning_rate": 4.324866310160428e-05, + "loss": 0.1828, + "step": 2350 + }, + { + "epoch": 2.3516614047700917, + "grad_norm": 0.12987194955348969, + "learning_rate": 4.318181818181819e-05, + "loss": 0.2013, + "step": 2351 + }, + { + "epoch": 2.3526616860992156, + "grad_norm": 0.12168455123901367, + "learning_rate": 4.311497326203209e-05, + "loss": 0.1987, + "step": 2352 + }, + { + "epoch": 2.353661967428339, + "grad_norm": 0.1206904798746109, + "learning_rate": 4.304812834224599e-05, + "loss": 0.1826, + "step": 2353 + }, + { + "epoch": 2.354662248757463, + "grad_norm": 0.11357525736093521, + "learning_rate": 4.298128342245989e-05, + "loss": 0.1763, + "step": 2354 + }, + { + "epoch": 2.3556625300865868, + "grad_norm": 0.1240660697221756, + "learning_rate": 4.29144385026738e-05, + "loss": 0.1909, + "step": 2355 + }, + { + "epoch": 2.3566628114157107, + "grad_norm": 0.11602874845266342, + "learning_rate": 4.28475935828877e-05, + "loss": 0.1785, + "step": 2356 + }, + { + "epoch": 2.3576630927448345, + "grad_norm": 0.12321159243583679, + "learning_rate": 4.2780748663101606e-05, + "loss": 0.188, + "step": 2357 + }, + { + "epoch": 2.3586633740739584, + "grad_norm": 0.11633746325969696, + "learning_rate": 4.271390374331551e-05, + "loss": 0.1801, + "step": 2358 + }, + { + "epoch": 2.3596636554030823, + "grad_norm": 0.11106952279806137, + "learning_rate": 4.2647058823529415e-05, + "loss": 0.184, + "step": 2359 + }, + { + "epoch": 2.3606639367322058, + "grad_norm": 0.11286384612321854, + "learning_rate": 4.258021390374331e-05, + "loss": 0.1844, + "step": 2360 + }, + { + "epoch": 2.3616642180613296, + "grad_norm": 0.11054420471191406, + "learning_rate": 4.251336898395722e-05, + "loss": 0.1738, + "step": 2361 + }, + { + "epoch": 2.3626644993904535, + "grad_norm": 0.12266630679368973, + "learning_rate": 4.244652406417112e-05, + "loss": 0.1937, + "step": 2362 + }, + { + "epoch": 2.3636647807195774, + "grad_norm": 0.11773712188005447, + "learning_rate": 4.2379679144385025e-05, + "loss": 0.1783, + "step": 2363 + }, + { + "epoch": 2.3646650620487013, + "grad_norm": 0.10771642625331879, + "learning_rate": 4.231283422459894e-05, + "loss": 0.1789, + "step": 2364 + }, + { + "epoch": 2.365665343377825, + "grad_norm": 0.12225739657878876, + "learning_rate": 4.224598930481284e-05, + "loss": 0.1924, + "step": 2365 + }, + { + "epoch": 2.3666656247069486, + "grad_norm": 0.11678391695022583, + "learning_rate": 4.2179144385026745e-05, + "loss": 0.1835, + "step": 2366 + }, + { + "epoch": 2.3676659060360725, + "grad_norm": 0.11920485645532608, + "learning_rate": 4.211229946524064e-05, + "loss": 0.1777, + "step": 2367 + }, + { + "epoch": 2.3686661873651964, + "grad_norm": 0.11549738049507141, + "learning_rate": 4.204545454545455e-05, + "loss": 0.1872, + "step": 2368 + }, + { + "epoch": 2.3696664686943203, + "grad_norm": 0.12811139225959778, + "learning_rate": 4.197860962566845e-05, + "loss": 0.1898, + "step": 2369 + }, + { + "epoch": 2.370666750023444, + "grad_norm": 0.1255708634853363, + "learning_rate": 4.1911764705882356e-05, + "loss": 0.1916, + "step": 2370 + }, + { + "epoch": 2.371667031352568, + "grad_norm": 0.11459209769964218, + "learning_rate": 4.184491978609626e-05, + "loss": 0.1772, + "step": 2371 + }, + { + "epoch": 2.372667312681692, + "grad_norm": 0.11659294366836548, + "learning_rate": 4.1778074866310165e-05, + "loss": 0.1926, + "step": 2372 + }, + { + "epoch": 2.3736675940108154, + "grad_norm": 0.11713515222072601, + "learning_rate": 4.171122994652407e-05, + "loss": 0.1827, + "step": 2373 + }, + { + "epoch": 2.3746678753399393, + "grad_norm": 0.11409571766853333, + "learning_rate": 4.164438502673797e-05, + "loss": 0.1842, + "step": 2374 + }, + { + "epoch": 2.375668156669063, + "grad_norm": 0.12399493157863617, + "learning_rate": 4.157754010695187e-05, + "loss": 0.1872, + "step": 2375 + }, + { + "epoch": 2.376668437998187, + "grad_norm": 0.1219099834561348, + "learning_rate": 4.1510695187165775e-05, + "loss": 0.1945, + "step": 2376 + }, + { + "epoch": 2.377668719327311, + "grad_norm": 0.12482144683599472, + "learning_rate": 4.144385026737968e-05, + "loss": 0.1956, + "step": 2377 + }, + { + "epoch": 2.3786690006564344, + "grad_norm": 0.1202552318572998, + "learning_rate": 4.1377005347593584e-05, + "loss": 0.1855, + "step": 2378 + }, + { + "epoch": 2.3796692819855583, + "grad_norm": 0.12018322199583054, + "learning_rate": 4.131016042780749e-05, + "loss": 0.1889, + "step": 2379 + }, + { + "epoch": 2.380669563314682, + "grad_norm": 0.12173015624284744, + "learning_rate": 4.124331550802139e-05, + "loss": 0.1828, + "step": 2380 + }, + { + "epoch": 2.381669844643806, + "grad_norm": 0.11855415254831314, + "learning_rate": 4.11764705882353e-05, + "loss": 0.1845, + "step": 2381 + }, + { + "epoch": 2.38267012597293, + "grad_norm": 0.12529151141643524, + "learning_rate": 4.1109625668449195e-05, + "loss": 0.195, + "step": 2382 + }, + { + "epoch": 2.383670407302054, + "grad_norm": 0.1116953119635582, + "learning_rate": 4.10427807486631e-05, + "loss": 0.1814, + "step": 2383 + }, + { + "epoch": 2.3846706886311777, + "grad_norm": 0.12129536271095276, + "learning_rate": 4.0975935828877004e-05, + "loss": 0.1913, + "step": 2384 + }, + { + "epoch": 2.385670969960301, + "grad_norm": 0.11808820813894272, + "learning_rate": 4.0909090909090915e-05, + "loss": 0.2003, + "step": 2385 + }, + { + "epoch": 2.386671251289425, + "grad_norm": 0.11070773005485535, + "learning_rate": 4.084224598930482e-05, + "loss": 0.174, + "step": 2386 + }, + { + "epoch": 2.387671532618549, + "grad_norm": 0.11057881265878677, + "learning_rate": 4.0775401069518723e-05, + "loss": 0.1749, + "step": 2387 + }, + { + "epoch": 2.388671813947673, + "grad_norm": 0.1222819983959198, + "learning_rate": 4.070855614973262e-05, + "loss": 0.1944, + "step": 2388 + }, + { + "epoch": 2.3896720952767967, + "grad_norm": 0.1125517413020134, + "learning_rate": 4.0641711229946525e-05, + "loss": 0.1743, + "step": 2389 + }, + { + "epoch": 2.3906723766059206, + "grad_norm": 0.12441930174827576, + "learning_rate": 4.057486631016043e-05, + "loss": 0.1811, + "step": 2390 + }, + { + "epoch": 2.391672657935044, + "grad_norm": 0.12603558599948883, + "learning_rate": 4.0508021390374334e-05, + "loss": 0.1905, + "step": 2391 + }, + { + "epoch": 2.392672939264168, + "grad_norm": 0.11271943897008896, + "learning_rate": 4.044117647058824e-05, + "loss": 0.1821, + "step": 2392 + }, + { + "epoch": 2.393673220593292, + "grad_norm": 0.11729811131954193, + "learning_rate": 4.037433155080214e-05, + "loss": 0.1799, + "step": 2393 + }, + { + "epoch": 2.3946735019224157, + "grad_norm": 0.11467090249061584, + "learning_rate": 4.030748663101605e-05, + "loss": 0.1861, + "step": 2394 + }, + { + "epoch": 2.3956737832515396, + "grad_norm": 0.11983012408018112, + "learning_rate": 4.024064171122995e-05, + "loss": 0.1838, + "step": 2395 + }, + { + "epoch": 2.3966740645806635, + "grad_norm": 0.12707597017288208, + "learning_rate": 4.017379679144385e-05, + "loss": 0.1965, + "step": 2396 + }, + { + "epoch": 2.3976743459097873, + "grad_norm": 0.12787874042987823, + "learning_rate": 4.0106951871657754e-05, + "loss": 0.1771, + "step": 2397 + }, + { + "epoch": 2.398674627238911, + "grad_norm": 0.12668639421463013, + "learning_rate": 4.004010695187166e-05, + "loss": 0.1909, + "step": 2398 + }, + { + "epoch": 2.3996749085680347, + "grad_norm": 0.12270843982696533, + "learning_rate": 3.997326203208556e-05, + "loss": 0.1971, + "step": 2399 + }, + { + "epoch": 2.4006751898971586, + "grad_norm": 0.11067384481430054, + "learning_rate": 3.990641711229947e-05, + "loss": 0.1787, + "step": 2400 + }, + { + "epoch": 2.4016754712262824, + "grad_norm": 0.11204174160957336, + "learning_rate": 3.983957219251337e-05, + "loss": 0.1838, + "step": 2401 + }, + { + "epoch": 2.4026757525554063, + "grad_norm": 0.11892908811569214, + "learning_rate": 3.9772727272727275e-05, + "loss": 0.1872, + "step": 2402 + }, + { + "epoch": 2.40367603388453, + "grad_norm": 0.11761950701475143, + "learning_rate": 3.970588235294117e-05, + "loss": 0.1868, + "step": 2403 + }, + { + "epoch": 2.4046763152136537, + "grad_norm": 0.11995532363653183, + "learning_rate": 3.963903743315508e-05, + "loss": 0.1908, + "step": 2404 + }, + { + "epoch": 2.4056765965427775, + "grad_norm": 0.12083796411752701, + "learning_rate": 3.957219251336899e-05, + "loss": 0.2023, + "step": 2405 + }, + { + "epoch": 2.4066768778719014, + "grad_norm": 0.11097664386034012, + "learning_rate": 3.950534759358289e-05, + "loss": 0.1886, + "step": 2406 + }, + { + "epoch": 2.4076771592010253, + "grad_norm": 0.11539682745933533, + "learning_rate": 3.94385026737968e-05, + "loss": 0.1876, + "step": 2407 + }, + { + "epoch": 2.408677440530149, + "grad_norm": 0.11644270271062851, + "learning_rate": 3.93716577540107e-05, + "loss": 0.1851, + "step": 2408 + }, + { + "epoch": 2.409677721859273, + "grad_norm": 0.11812936514616013, + "learning_rate": 3.93048128342246e-05, + "loss": 0.2049, + "step": 2409 + }, + { + "epoch": 2.410678003188397, + "grad_norm": 0.11457454413175583, + "learning_rate": 3.9237967914438504e-05, + "loss": 0.1778, + "step": 2410 + }, + { + "epoch": 2.4116782845175204, + "grad_norm": 0.11274266988039017, + "learning_rate": 3.917112299465241e-05, + "loss": 0.1734, + "step": 2411 + }, + { + "epoch": 2.4126785658466443, + "grad_norm": 0.12072449922561646, + "learning_rate": 3.910427807486631e-05, + "loss": 0.1986, + "step": 2412 + }, + { + "epoch": 2.413678847175768, + "grad_norm": 0.11160512268543243, + "learning_rate": 3.903743315508022e-05, + "loss": 0.177, + "step": 2413 + }, + { + "epoch": 2.414679128504892, + "grad_norm": 0.11605464667081833, + "learning_rate": 3.897058823529412e-05, + "loss": 0.1876, + "step": 2414 + }, + { + "epoch": 2.415679409834016, + "grad_norm": 0.10968615114688873, + "learning_rate": 3.8903743315508025e-05, + "loss": 0.1783, + "step": 2415 + }, + { + "epoch": 2.4166796911631394, + "grad_norm": 0.11751143634319305, + "learning_rate": 3.883689839572193e-05, + "loss": 0.1924, + "step": 2416 + }, + { + "epoch": 2.4176799724922633, + "grad_norm": 0.12495880573987961, + "learning_rate": 3.877005347593583e-05, + "loss": 0.1888, + "step": 2417 + }, + { + "epoch": 2.418680253821387, + "grad_norm": 0.11140533536672592, + "learning_rate": 3.870320855614973e-05, + "loss": 0.1948, + "step": 2418 + }, + { + "epoch": 2.419680535150511, + "grad_norm": 0.11281417310237885, + "learning_rate": 3.8636363636363636e-05, + "loss": 0.1863, + "step": 2419 + }, + { + "epoch": 2.420680816479635, + "grad_norm": 0.11595267057418823, + "learning_rate": 3.856951871657754e-05, + "loss": 0.1874, + "step": 2420 + }, + { + "epoch": 2.421681097808759, + "grad_norm": 0.11251599341630936, + "learning_rate": 3.8502673796791445e-05, + "loss": 0.1846, + "step": 2421 + }, + { + "epoch": 2.4226813791378827, + "grad_norm": 0.1064729318022728, + "learning_rate": 3.843582887700535e-05, + "loss": 0.1787, + "step": 2422 + }, + { + "epoch": 2.423681660467006, + "grad_norm": 0.11546791344881058, + "learning_rate": 3.8368983957219254e-05, + "loss": 0.184, + "step": 2423 + }, + { + "epoch": 2.42468194179613, + "grad_norm": 0.1130218431353569, + "learning_rate": 3.830213903743316e-05, + "loss": 0.1907, + "step": 2424 + }, + { + "epoch": 2.425682223125254, + "grad_norm": 0.11970975995063782, + "learning_rate": 3.8235294117647055e-05, + "loss": 0.1766, + "step": 2425 + }, + { + "epoch": 2.426682504454378, + "grad_norm": 0.11060624569654465, + "learning_rate": 3.816844919786097e-05, + "loss": 0.1883, + "step": 2426 + }, + { + "epoch": 2.4276827857835017, + "grad_norm": 0.12293516099452972, + "learning_rate": 3.810160427807487e-05, + "loss": 0.197, + "step": 2427 + }, + { + "epoch": 2.4286830671126256, + "grad_norm": 0.11492732912302017, + "learning_rate": 3.8034759358288775e-05, + "loss": 0.1861, + "step": 2428 + }, + { + "epoch": 2.429683348441749, + "grad_norm": 0.11447480320930481, + "learning_rate": 3.796791443850268e-05, + "loss": 0.1855, + "step": 2429 + }, + { + "epoch": 2.430683629770873, + "grad_norm": 0.10875410586595535, + "learning_rate": 3.7901069518716584e-05, + "loss": 0.193, + "step": 2430 + }, + { + "epoch": 2.431683911099997, + "grad_norm": 0.1235925629734993, + "learning_rate": 3.783422459893048e-05, + "loss": 0.18, + "step": 2431 + }, + { + "epoch": 2.4326841924291207, + "grad_norm": 0.12194931507110596, + "learning_rate": 3.7767379679144386e-05, + "loss": 0.1933, + "step": 2432 + }, + { + "epoch": 2.4336844737582446, + "grad_norm": 0.12830987572669983, + "learning_rate": 3.770053475935829e-05, + "loss": 0.2021, + "step": 2433 + }, + { + "epoch": 2.4346847550873685, + "grad_norm": 0.1108524426817894, + "learning_rate": 3.7633689839572195e-05, + "loss": 0.1806, + "step": 2434 + }, + { + "epoch": 2.4356850364164924, + "grad_norm": 0.1251210868358612, + "learning_rate": 3.75668449197861e-05, + "loss": 0.1844, + "step": 2435 + }, + { + "epoch": 2.436685317745616, + "grad_norm": 0.11926569789648056, + "learning_rate": 3.7500000000000003e-05, + "loss": 0.1845, + "step": 2436 + }, + { + "epoch": 2.4376855990747397, + "grad_norm": 0.12355847656726837, + "learning_rate": 3.743315508021391e-05, + "loss": 0.1951, + "step": 2437 + }, + { + "epoch": 2.4386858804038636, + "grad_norm": 0.11597749590873718, + "learning_rate": 3.7366310160427805e-05, + "loss": 0.1924, + "step": 2438 + }, + { + "epoch": 2.4396861617329875, + "grad_norm": 0.11872819811105728, + "learning_rate": 3.729946524064171e-05, + "loss": 0.1856, + "step": 2439 + }, + { + "epoch": 2.4406864430621114, + "grad_norm": 0.11575272679328918, + "learning_rate": 3.7232620320855614e-05, + "loss": 0.1895, + "step": 2440 + }, + { + "epoch": 2.441686724391235, + "grad_norm": 0.12192070484161377, + "learning_rate": 3.716577540106952e-05, + "loss": 0.1896, + "step": 2441 + }, + { + "epoch": 2.4426870057203587, + "grad_norm": 0.11034922301769257, + "learning_rate": 3.709893048128342e-05, + "loss": 0.1855, + "step": 2442 + }, + { + "epoch": 2.4436872870494826, + "grad_norm": 0.11821943521499634, + "learning_rate": 3.703208556149733e-05, + "loss": 0.1889, + "step": 2443 + }, + { + "epoch": 2.4446875683786065, + "grad_norm": 0.12413725256919861, + "learning_rate": 3.696524064171123e-05, + "loss": 0.1991, + "step": 2444 + }, + { + "epoch": 2.4456878497077303, + "grad_norm": 0.11876273155212402, + "learning_rate": 3.6898395721925136e-05, + "loss": 0.1849, + "step": 2445 + }, + { + "epoch": 2.4466881310368542, + "grad_norm": 0.12016553431749344, + "learning_rate": 3.6831550802139034e-05, + "loss": 0.1857, + "step": 2446 + }, + { + "epoch": 2.447688412365978, + "grad_norm": 0.11945883929729462, + "learning_rate": 3.6764705882352945e-05, + "loss": 0.1825, + "step": 2447 + }, + { + "epoch": 2.4486886936951016, + "grad_norm": 0.12088269740343094, + "learning_rate": 3.669786096256685e-05, + "loss": 0.1834, + "step": 2448 + }, + { + "epoch": 2.4496889750242254, + "grad_norm": 0.12082383781671524, + "learning_rate": 3.6631016042780753e-05, + "loss": 0.1896, + "step": 2449 + }, + { + "epoch": 2.4506892563533493, + "grad_norm": 0.11892774701118469, + "learning_rate": 3.656417112299466e-05, + "loss": 0.1909, + "step": 2450 + }, + { + "epoch": 2.451689537682473, + "grad_norm": 0.10890307277441025, + "learning_rate": 3.649732620320856e-05, + "loss": 0.1743, + "step": 2451 + }, + { + "epoch": 2.452689819011597, + "grad_norm": 0.11104000359773636, + "learning_rate": 3.643048128342246e-05, + "loss": 0.1887, + "step": 2452 + }, + { + "epoch": 2.453690100340721, + "grad_norm": 0.12348121404647827, + "learning_rate": 3.6363636363636364e-05, + "loss": 0.1876, + "step": 2453 + }, + { + "epoch": 2.4546903816698444, + "grad_norm": 0.11569233238697052, + "learning_rate": 3.629679144385027e-05, + "loss": 0.1802, + "step": 2454 + }, + { + "epoch": 2.4556906629989683, + "grad_norm": 0.11577185243368149, + "learning_rate": 3.622994652406417e-05, + "loss": 0.1796, + "step": 2455 + }, + { + "epoch": 2.456690944328092, + "grad_norm": 0.11597847938537598, + "learning_rate": 3.616310160427808e-05, + "loss": 0.1732, + "step": 2456 + }, + { + "epoch": 2.457691225657216, + "grad_norm": 0.12893490493297577, + "learning_rate": 3.609625668449198e-05, + "loss": 0.1807, + "step": 2457 + }, + { + "epoch": 2.45869150698634, + "grad_norm": 0.11901641637086868, + "learning_rate": 3.6029411764705886e-05, + "loss": 0.1924, + "step": 2458 + }, + { + "epoch": 2.459691788315464, + "grad_norm": 0.12525837123394012, + "learning_rate": 3.596256684491979e-05, + "loss": 0.193, + "step": 2459 + }, + { + "epoch": 2.4606920696445878, + "grad_norm": 0.1220000684261322, + "learning_rate": 3.589572192513369e-05, + "loss": 0.183, + "step": 2460 + }, + { + "epoch": 2.461692350973711, + "grad_norm": 0.11522224545478821, + "learning_rate": 3.582887700534759e-05, + "loss": 0.1899, + "step": 2461 + }, + { + "epoch": 2.462692632302835, + "grad_norm": 0.11516457051038742, + "learning_rate": 3.57620320855615e-05, + "loss": 0.1733, + "step": 2462 + }, + { + "epoch": 2.463692913631959, + "grad_norm": 0.11374516785144806, + "learning_rate": 3.56951871657754e-05, + "loss": 0.1835, + "step": 2463 + }, + { + "epoch": 2.464693194961083, + "grad_norm": 0.12314411252737045, + "learning_rate": 3.5628342245989305e-05, + "loss": 0.1832, + "step": 2464 + }, + { + "epoch": 2.4656934762902067, + "grad_norm": 0.1215660348534584, + "learning_rate": 3.556149732620321e-05, + "loss": 0.1945, + "step": 2465 + }, + { + "epoch": 2.4666937576193306, + "grad_norm": 0.1125190258026123, + "learning_rate": 3.5494652406417114e-05, + "loss": 0.1758, + "step": 2466 + }, + { + "epoch": 2.467694038948454, + "grad_norm": 0.12295130640268326, + "learning_rate": 3.542780748663101e-05, + "loss": 0.1822, + "step": 2467 + }, + { + "epoch": 2.468694320277578, + "grad_norm": 0.12160823494195938, + "learning_rate": 3.536096256684492e-05, + "loss": 0.1853, + "step": 2468 + }, + { + "epoch": 2.469694601606702, + "grad_norm": 0.11709440499544144, + "learning_rate": 3.529411764705883e-05, + "loss": 0.175, + "step": 2469 + }, + { + "epoch": 2.4706948829358257, + "grad_norm": 0.12441233545541763, + "learning_rate": 3.522727272727273e-05, + "loss": 0.1928, + "step": 2470 + }, + { + "epoch": 2.4716951642649496, + "grad_norm": 0.11695539951324463, + "learning_rate": 3.5160427807486636e-05, + "loss": 0.1944, + "step": 2471 + }, + { + "epoch": 2.4726954455940735, + "grad_norm": 0.12155629694461823, + "learning_rate": 3.509358288770054e-05, + "loss": 0.1893, + "step": 2472 + }, + { + "epoch": 2.4736957269231974, + "grad_norm": 0.12062817066907883, + "learning_rate": 3.5026737967914445e-05, + "loss": 0.1805, + "step": 2473 + }, + { + "epoch": 2.474696008252321, + "grad_norm": 0.12057454884052277, + "learning_rate": 3.495989304812834e-05, + "loss": 0.188, + "step": 2474 + }, + { + "epoch": 2.4756962895814447, + "grad_norm": 0.1165895164012909, + "learning_rate": 3.489304812834225e-05, + "loss": 0.1937, + "step": 2475 + }, + { + "epoch": 2.4766965709105686, + "grad_norm": 0.1261078268289566, + "learning_rate": 3.482620320855615e-05, + "loss": 0.1936, + "step": 2476 + }, + { + "epoch": 2.4776968522396925, + "grad_norm": 0.1178465411067009, + "learning_rate": 3.4759358288770055e-05, + "loss": 0.187, + "step": 2477 + }, + { + "epoch": 2.4786971335688164, + "grad_norm": 0.11524347215890884, + "learning_rate": 3.469251336898396e-05, + "loss": 0.1765, + "step": 2478 + }, + { + "epoch": 2.47969741489794, + "grad_norm": 0.1144275888800621, + "learning_rate": 3.4625668449197864e-05, + "loss": 0.1856, + "step": 2479 + }, + { + "epoch": 2.4806976962270637, + "grad_norm": 0.11870783567428589, + "learning_rate": 3.455882352941177e-05, + "loss": 0.1791, + "step": 2480 + }, + { + "epoch": 2.4816979775561876, + "grad_norm": 0.11471819877624512, + "learning_rate": 3.4491978609625666e-05, + "loss": 0.1814, + "step": 2481 + }, + { + "epoch": 2.4826982588853115, + "grad_norm": 0.1113131046295166, + "learning_rate": 3.442513368983957e-05, + "loss": 0.1836, + "step": 2482 + }, + { + "epoch": 2.4836985402144354, + "grad_norm": 0.1167987808585167, + "learning_rate": 3.4358288770053475e-05, + "loss": 0.1858, + "step": 2483 + }, + { + "epoch": 2.4846988215435593, + "grad_norm": 0.12235665321350098, + "learning_rate": 3.429144385026738e-05, + "loss": 0.1878, + "step": 2484 + }, + { + "epoch": 2.485699102872683, + "grad_norm": 0.12216199189424515, + "learning_rate": 3.4224598930481284e-05, + "loss": 0.1964, + "step": 2485 + }, + { + "epoch": 2.4866993842018066, + "grad_norm": 0.12777546048164368, + "learning_rate": 3.415775401069519e-05, + "loss": 0.1816, + "step": 2486 + }, + { + "epoch": 2.4876996655309305, + "grad_norm": 0.10907205939292908, + "learning_rate": 3.409090909090909e-05, + "loss": 0.1832, + "step": 2487 + }, + { + "epoch": 2.4886999468600544, + "grad_norm": 0.12209538370370865, + "learning_rate": 3.4024064171122997e-05, + "loss": 0.191, + "step": 2488 + }, + { + "epoch": 2.4897002281891782, + "grad_norm": 0.12112060189247131, + "learning_rate": 3.39572192513369e-05, + "loss": 0.1812, + "step": 2489 + }, + { + "epoch": 2.490700509518302, + "grad_norm": 0.11776310205459595, + "learning_rate": 3.3890374331550805e-05, + "loss": 0.1879, + "step": 2490 + }, + { + "epoch": 2.491700790847426, + "grad_norm": 0.11678838729858398, + "learning_rate": 3.382352941176471e-05, + "loss": 0.1866, + "step": 2491 + }, + { + "epoch": 2.4927010721765495, + "grad_norm": 0.12153811752796173, + "learning_rate": 3.3756684491978614e-05, + "loss": 0.1912, + "step": 2492 + }, + { + "epoch": 2.4937013535056733, + "grad_norm": 0.11978083103895187, + "learning_rate": 3.368983957219252e-05, + "loss": 0.1904, + "step": 2493 + }, + { + "epoch": 2.4947016348347972, + "grad_norm": 0.11907623708248138, + "learning_rate": 3.362299465240642e-05, + "loss": 0.1909, + "step": 2494 + }, + { + "epoch": 2.495701916163921, + "grad_norm": 0.12018263339996338, + "learning_rate": 3.355614973262032e-05, + "loss": 0.1876, + "step": 2495 + }, + { + "epoch": 2.496702197493045, + "grad_norm": 0.11943220347166061, + "learning_rate": 3.3489304812834225e-05, + "loss": 0.1849, + "step": 2496 + }, + { + "epoch": 2.497702478822169, + "grad_norm": 0.1211661547422409, + "learning_rate": 3.342245989304813e-05, + "loss": 0.196, + "step": 2497 + }, + { + "epoch": 2.498702760151293, + "grad_norm": 0.11045736074447632, + "learning_rate": 3.3355614973262033e-05, + "loss": 0.1877, + "step": 2498 + }, + { + "epoch": 2.499703041480416, + "grad_norm": 0.12105974555015564, + "learning_rate": 3.328877005347594e-05, + "loss": 0.1853, + "step": 2499 + }, + { + "epoch": 2.50070332280954, + "grad_norm": 0.12437592446804047, + "learning_rate": 3.322192513368984e-05, + "loss": 0.1934, + "step": 2500 + } + ], + "logging_steps": 1, + "max_steps": 2997, + "num_input_tokens_seen": 0, + "num_train_epochs": 3, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 5.299592979025306e+18, + "train_batch_size": 2, + "trial_name": null, + "trial_params": null +}